![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Microbiology and Molecular Biology Reviews, June 2007, p. 255-281, Vol. 71, No. 2
1092-2172/07/$08.00+0 doi:10.1128/MMBR.00034-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Department of Biochemistry, University of Cambridge, 80 Tennis Court Road, Cambridge CB2 1GA, United Kingdom
SUMMARY INTRODUCTION Modes of Action Cry Toxins as Biopesticides Toxin Diversity Toxin Structure and Function Domain I. Domain II. Domain III. IDENTIFICATION AND VALIDATION OF RECEPTORS APN APN as a Cry-binding protein. (i) Class 1. (ii) Class 2. (iii) Class 3. (iv) Class 4. (v) Class 5. (vi) Other APNs. (vii) Summary. APN as a mediator of Cry toxin susceptibility. (i) Permeability. (ii) In vitro toxicity. (iii) In vivo toxicity. (iv) Summary. Cadherin BT-R1 (Manduca sexta). BtR175 (Bombyx mori). HevCaLP (Heliothis virescens). Cadherin-like proteins in other species. Summary. ALP Glycolipid Other Receptors DETERMINANTS OF BINDING APN Receptor determinants. Toxin determinants. (i) Domain III. (ii) Domain II. (iii) Domain II/III interface. Cadherin Receptor determinants. Toxin determinants. TOXIN MOLECULAR MECHANISM OF ACTION The Bravo Model The Zhang Model The Jurat-Fuentes Model CONCLUDING REMARKS REFERENCES
| SUMMARY |
|---|
 Top NextReferences |
|---|
| INTRODUCTION |
|---|
|
TopPreviousNextReferences |
|---|
B. thuringiensis is pathogenic to insects and can be readily distinguished from other members of the B. cereus group by the production of large crystalline inclusions that consist of entomocidal protein protoxins. When activated upon ingestion, these toxins, in addition to other virulence factors, weaken or kill insects and allow B. thuringiensis spores to germinate in the insect. The type and number of different protoxins in the crystalline inclusions of B. thuringiensis determine a particular strain's toxicity profile. Cry proteins are highly diverse and primarily target insects in the orders Lepidoptera (butterflies and moths), Diptera (flies and mosquitoes), and Coleoptera (beetles and weevils) (152); however, some Cry toxins have been reported to kill hymenopterans (wasps and bees) (46) and nematodes (118, 186).
The extraordinary diversity of Cry toxins is believed to be due to a high degree of genetic plasticity. Many cry genes are associated with transposable elements that may facilitate gene amplification, leading to the evolution of new toxins (29). In addition, most cry genes are found on plasmids, and horizontal transfer by conjugation may result in the creation of new strains with a novel complement of cry genes (166, 167).
The large number of known Cry proteins has permitted comparative sequence analyses and has helped to elucidate elements important for both basic toxin function and insect specificity. In 1989, Höfte and Whiteley (70) carried out the first detailed analysis of Cry protein sequence. At that time, 13 holotype Cry proteins were known and assigned to one of four groups based on their insect specificity. Sequence alignment showed a high degree of diversity among the Cry proteins; however, five blocks of conserved amino acids were identified that were found in most sequences. The discovery of new Cry proteins prompted further analysis: first by Bravo in 1997 (17) and then by de Maagd et al. in 2001 (29). In the more recent work, the sequences of proteins from Cry1 to Cry31 were analyzed. Most toxins had some or all of the five conserved blocks identified by Höfte and Whiteley (70), suggesting that these regions may be important for some aspects of toxin stability or function. It was also evident that Cry toxins were generally one of two lengths: either 130 to 140 kDa or approximately 70 kDa. The conserved blocks were present in the N-terminal half of the longer toxins, whereas the C-terminal half constituted a protoxin domain not found in the smaller toxins.
Using domain information derived from the crystal structures of active Cry1Aa, Cry2Aa, and Cry3Aa (described in the next section), de Maagd et al. (29) aligned each of three toxin domains separately and created phylogenetic trees to assess the individual contribution of each domain to insect specificity. The different trees showed that in general, there was a correlation between sequence similarity and insect order specificity but that relatively unrelated clusters could sometimes have similar activities. This suggested that insect specificity may have developed along multiple evolutionary paths. A comparison of the different trees showed that domains I and II had relatively similar tree architectures, suggesting coevolution in many cases. In contrast, the topology of the domain III tree was quite different, and thus it was hypothesized that shuffling in this domain may have contributed to Cry toxin diversity.
|
|
|
The structure of domain II has been compared to those of other ß-prism proteins (28), including vitelline (158) and the plant lectins jacalin (151) and Maclura pomifera agglutinin (107). Other proteins with a ß-prism fold were identified in the Protein Data Bank (PDB) (http://www.rcsb.org/pdb/Welcome.do) and include Helianthus tuberosus lectin (15), artocarpin (141), Calystegia sepium lectin (14), and banana lectin (123). Vitelline is found in the vitelline membrane of hen eggs and although its biological function is unknown, it is believed to bind the carbohydrate N-acetylglucosamine pentasaccharide at its apex (157). The structure of vitelline is much more symmetrical than that of domain II; each four-ß-strand sheet is related by sequence, unlike the ß-sheets that comprise domain II of Cry toxins. The protein is also characterized by long, flexible loops at its apex (158), similar to what is observed for some Cry toxins. The plant lectins are part of the jacalin-related superfamily of lectins (139) and are either mannose or galactose specific. Several of these lectins have been cocrystallized with their ligands, and the binding site is invariably at the apex. Banana lectin is unique in that two carbohydrate binding sites have been identified at the apex (123) (Fig. 3). The structural similarity between domain II and lectin domains has led to speculation that domain II may bind to carbohydrates, but this has not yet been demonstrated.
|
Domain III has been compared to a number of different proteins (28), but its similarity to carbohydrate-binding modules (CBMs) found in microbial glycoside hydrolases, lyases, and esterases is particularly striking. These enzymes generally consist of a catalytic domain linked to one or more CBMs. The function of the CBM is to target the catalytic domain to its polysaccharide substrate (172). This enhances the enzyme's catalytic efficiency by increasing its effective concentration at the substrate surface. The structure of several CBMs in complex with carbohydrate ligands has now been solved, and two binding sites have been identified (140). One site (cleft A) is found at the loops connecting the two ß-sheets, and the other (cleft B) is located on the concave surface of one of the ß-sheets (Fig. 4). Aromatic residues are important components of each binding site, and in general they are the best-characterized mediators of carbohydrate-protein interactions in CBMs (13). Figure 4 shows an overlay of domain III from Cry1Aa and CmCBM6-2 (the family 6 CBM from Cellvibrio mixtus endoglucanase 5a) in complex with two cellotriose molecules (140). As shown, there is significant structural similarity between these domains, suggesting that some Cry toxins may also bind to carbohydrates in these regions. It should be noted, however, that the aromatic residues important for carbohydrate binding in CBMs are generally not conserved in Cry toxins.
|
| IDENTIFICATION AND VALIDATION OF RECEPTORS |
|---|
|
TopPreviousNextReferences |
|---|
After it was demonstrated that specific high-affinity toxin binding sites were present in the insect midgut, efforts to identify and clone toxin receptors were intensified. Many putative Cry toxin receptors have since been reported, of which the best characterized are the aminopeptidase N (APN) receptors (51, 93, 142, 150) and the cadherin-like receptors (44, 130, 131, 174, 175) identified in lepidopterans. In nematodes, glycolipids are believed to be an important class of Cry toxin receptors (60). Other putative receptors include alkaline phosphatases (ALPs) (38, 85, 86), a 270-kDa glycoconjugate (176), and a 252-kDa protein (73). In the following sections, each receptor class will be discussed with a particular focus on toxin-receptor binding interactions and the ability of receptors to confer toxin susceptibility.
In addition to being studied for their role in digestion, APNs have been extensively studied as putative Cry toxin receptors. Since it was first shown that Cry toxins can bind to APN (93, 150), many different forms have been isolated and characterized. Figure 5 shows the phylogenetic relationship between representative lepidopteran APNs and indicates those that have been reported to interact with Cry toxins. As shown, the APNs have been divided into five different classes (67). The average sequence identity within a class varies from 56% (class 5) to 67% (class 4). Among the different classes, class 2 is the least like the others, with an average sequence identity of only 25 to 26% relative to the other classes, whereas class 1 and class 3 are the most similar, with an average sequence identity of 38%. To date, all known APNs within a particular species have been found to cluster into different classes. In fact, some APNs share higher sequence identity with those in nonlepidopterans than with other APNs in the same species. For example, class 2 APN from M. sexta is more similar to APNs in chicken and frog (GenBank accession numbers NP_990192 and AAH85055, respectively) than to class 1 M. sexta APN, and yet both M. sexta APNs are reported to bind to Cry1Ab (31, 120).
|
|
|
(i) Class 1. Class 1 APNs have been identified in nine different lepidopterans. In addition to the features already discussed, class 1 APNs generally have a threonine-rich sequence downstream of the C-terminal GPI signal sequence. This region is believed to have extensive O-linked glycosylation and is thought to form a rigid stalk that elevates the active site of the enzyme above the membrane surface (92, 93). By use of the NetOGlyc 3.1 server (84) (http://www.cbs.dtu.dk/services/NetOGlyc), the number of predicted O-linked glycosylation sites in class 1 APNs has been shown to vary from 6 in Bombyx mori to 39in Helicoverpa armigera (Fig. 7). In species where native APN has been isolated from the midgut, the correlation between observed molecular mass and the number of predicted O-linked glycosylation sites is strong. For example, APNs from B. mori, M. sexta, and Heliothis virescens are reported to have observed molecular masses of 120 kDa (192), 120 kDa (93), and 170 kDa (116) and totals of 6, 10, and 36 predicted O-linked glycosylation sites, respectively. In M. sexta, all 10 O-glycosylation sites are predicted to reside in the C-terminal stalk and are believed to be rich in GalNAc; thus, Knight et al. (92) have proposed that this region is highly likely to be a Cry1Ac binding site.
Five of the nine known class 1 APNs have been tested for their ability to bind to Cry toxins, and the interaction between the 120-kDa M. sexta APN and Cry1Ac is perhaps the best studied. APN was initially shown to bind to Cry1Ac by ligand blot analysis (93, 150) and was subsequently purified using a Cry1Ac protoxin affinity column (93). The fact that GalNAc could be used to elute APN from the column and that purified APN could be detected with the GalNAc-specific lectin soybean agglutinin (SBA) suggested that this carbohydrate was involved in toxin binding (93). Using surface plasmon resonance (SPR) analysis, Masson et al. (120) showed that in addition to Cry1Ac, the closely related Cry1Aa and Cry1Ab could also bind to purified native M. sexta APN. Cry1Ac was found to bind to APN at two different sites, one of which it shared with Cry1Aa and Cry1Ab. The affinity constants for Cry1Aa, Cry1Ab and Cry1Ac at the common binding site were 28.4 nM, 42.8 nM, and 40.7 to 95.3 nM (depending on the source of toxin), respectively, whereas that at the second Cry1Ac binding site was reported to be 149.4 to 299.3 nM. Unlike Cry1Ac, the interaction between APN and Cry1Aa or Cry1Ab was not inhibited by GalNAc. The more distantly related toxin Cry1Ca was also tested for binding to purified APN, but an interaction could not be detected. A 106-kDa APN has since been reported to bind to Cry1Ca, but the gene encoding this putative receptor has yet to be identified (114).
The interaction between Cry1Ac and exogenously expressed class 1 APN has also been studied. Using ligand blot analysis, Gill and Ellar (50) showed that Cry1Ac could bind to APN expressed in a transgenic line of Drosophila melanogaster. In contrast, Luo et al. (115) expressed the same APN in Sf21 cells, and despite its enzymatic activity, glycosylation, and membrane localization, binding to Cry1Ac could not be demonstrated. The authors suggested that posttranslational modification mechanisms in Sf21 cells may be different from those in epithelial cells in the M. sexta midgut (115).
A 170-kDa class 1 APN from H. virescens has also been identified as a Cry1A-binding protein (116). Like M. sexta APN, the first H. virescens APN was isolated by Cry1Ac affinity chromatography using GalNAc to elute the protein. By use of SPR analysis, Cry1Aa, Cry1Ab, and Cry1Ac, but not Cry1Ca or Cry1Ea, were shown to interact with the purified APN, and only Cry1Ac binding to APN could be inhibited with GalNAc. Unlike the observation with M. sexta, all Cry1A proteins were thought to bind to H. virescens APN at two sites, as determined by the observed 2:1 molar ratio of bound toxin to receptor, and by the good fit of experimental data to a two-binding-site model based on kinetics (116). In a later study (8), Cry1Fa was shown to bind to this APN by ligand blot analysis, and thus it appeared that class 1 H. virescens APN was not exclusively a Cry1A-binding protein.
Class 1 APN from B. mori has also been shown to interact with Cry toxins and is best characterized as a Cry1Aa-binding protein. The 120-kDa APN was released from BBMV with phosphatidylinositol-specific phospholipase C (PI-PLC) and then purified by ion-exchange chromatography (192). This preparation was shown to interact with Cry1Aa by both dot blot and ligand blot analysis. Cry1Aa binding to denatured APN expressed as a glutathione S-transferase (GST) fusion protein in Escherichia coli has also been demonstrated, providing evidence that Cry1Aa can interact with APN independently of glycosylation (194). By ligand blot analysis, Cry1Ab has also been reported to bind to E. coli-expressed APN (132), whereas Cry1Ac bound only weakly to purified native receptor (193). In contrast, neither Cry1Ab nor Cry1Ac were reported to bind to purified native B. mori APN by use of SPR analysis (80), and thus it not clear whether the interaction between APN and these toxins is biologically relevant. Of the three toxins, Cry1Aa is the most toxic towards B. mori larvae, and only limited toxicity has been observed with Cry1Ac (49, 99, 104).
Two other species have been reported to produce class 1 APN capable of interacting with Cry toxins: H. armigera (142) and Plutella xylostella (132). In both cases, binding was studied using exogenously expressed protein. Rajagopal et al. (142) expressed H. armigera APN in Trichoplusia ni cells by use of a baculovirus expression vector. The expressed protein was found to be membrane associated, catalytically active, and glycosylated and by ligand blot analysis could bind to Cry1Aa, Cry1Ab, and Cry1Ac. To study the interaction of Cry toxins with P. xylostella APN, a truncation mutant was expressed in E. coli as a GST fusion protein (132). This mutant was based on a previously identified toxin binding region found in a homologous region of B. mori APN (193). By ligand blot analysis, both Cry1Aa and Cry1Ab were reported to bind to this truncated mutant (132).
(ii) Class 2. Lepidopteran APNs in class 2 share the least sequence identity with the other classes (Fig. 5). Each member is predicted to be N glycosylated, but in stark contrast to class 1, there are no O-linked glycosylation sites predicted for any of the APNs, and the threonine-rich C-terminal stalk region reported for class 1 APN is completely absent. Interestingly, Cry1Ac has not been reported to bind to any member of class 2, supporting the theory that Cry1Ac binds to APN at the O-glycosylated C-terminal stalk (92). Cry1Aa and Cry1Ab have been reported to bind to class 2 APNs under certain conditions (31, 132), but it remains to be seen whether these interactions are biologically relevant.
Cry1Ab was first shown to bind to class 2 M. sexta APN by Denolf et al. (31). This interaction was demonstrated when APN was partially purified using a Cry1Ab affinity column. Using a high-pH carbonate buffer, proteins of many different molecular weights were eluted, in contrast to the Cry1Ac protoxin affinity purification of class 1 APN from M. sexta in which a single band was observed following GalNAc elution (93). Nevertheless, by ligand blot analysis Cry1Ab was shown to bind to a 120-kDa band in the purified fraction, and internal amino acid sequence data facilitated cloning of the encoding gene (31). Attempts to express the protein in Sf9 cells were unsuccessful, and thus it was not possible to confirm that the cloned gene actually encoded a Cry1Ab-binding protein.
Using the sequence information derived from class 2 M. sexta APN, Denolf et al. (31) were also able to study the corresponding APN in P. xylostella. In this case, expression in Sf9 cells was possible, and a 105-kDa glycoprotein with enzymatic activity was produced. Homologous competition binding assays using Cry1Ab, Cry1Ac, Cry1Ba, Cry1Ca, or Cry9Ca and intact Sf9 cells expressing the APN did not reveal any specific toxin binding, and similar results were obtained with cell-derived membrane preparations. In addition, Cry1Ab binding to APN could not be demonstrated by ligand blot analysis. Since Cry1Ab had previously been shown by ligand blot analysis to bind to a 120-kDa protein in P. xylostella BBMV, it was not clear whether the cloned APN was a different protein or whether differences in glycosylation were responsible for the lack of binding. A study by Nakanishi et al. (132) further complicated the matter. This group expressed a putative Cry toxin binding region from class 2 P. xylostella APN as a GST fusion protein in E. coli and by ligand blotting showed binding to both Cry1Aa and Cry1Ab. In addition, they showed that Cry1Aa and Cry1Ab could bind to the same region in a class 2 APN from B. mori. Thus, it seems that further work must be carried out to clarify whether class 2 APNs are genuine Cry1A receptors, and if so, to what extent glycosylation plays a role in toxin binding.
(iii) Class 3. Class 3 is made up of the largest group of known lepidopteran APNs, with members from 11 different species. This class is most closely related to class 1 (Fig. 5), and similarly has a threonine-rich C terminus predicted to be highly glycosylated by the NetOGlyc 3.1 server (84) (http://www.cbs.dtu.dk/services/NetOGlyc/). Class 3 APNs generally have fewer predicted O-linked glycosylation sites than class 1 APNs (Fig. 7) and, of those isolated from BBMV, all have had a molecular mass near 120 kDa. Within this class, binding to Cry1Aa, Cry1Ab, Cry1Ac, Cry1B,a and Cry1Fa has been reported, as will be discussed.
APN from L. dispar is perhaps the best-studied member of class 3. It was cloned by Garner et al. (48), and based on sequence identity was believed to be the APN1 previously described by Valaitis et al. (177). In these earlier experiments, APN was released from BBMV by use of PI-PLC and was subsequently purified using a series of chromatographic steps. By ligand blot analysis, Cry1Ac was shown to bind to purified denatured APN. Binding was also tested by SPR analysis and under these conditions Cry1Ac, but not Cry1Aa or Cry1Ab, bound to native APN and Cry1Ac binding could be completely blocked with competing GalNAc (177). The binding of Cry1Ac to APN was reported to occur in a 1:1 ratio, in contrast to the 2:1 ratio reported for the interaction between Cry1Ac and class 1 M. sexta APN (120) and class 1 H. virescens APN (116). Cloned, exogenously expressed L. dispar APN has also been studied (48). Sf9 cells were transformed with a baculovirus vector encoding this APN, and the expressed protein was recognized by an APN-specific antibody; however, ligand blot analysis revealed only weak binding to Cry1Ac. As suggested in other studies, differences in posttranslational modification may account for the differences observed in toxin binding to native and recombinant forms of APN.
The binding of Cry1Ac to class 3 APN purified from H. virescens has also been reported (51). In this case, Cry1Ac was shown to bind to purified protein by ligand blot analysis, and the interaction could be blocked with competing GalNAc. It was also reported that Cry1Ac failed to bind to APN prepared by in vitro translation, providing additional evidence that glycosylation, or at least some form of posttranslational modification, was important for binding. Additional characterization was reported by Banks et al. (8), who partially purified a 120-kDa APN believed to be the same as that reported by Gill et al. (51) but whose identity was not confirmed by amino acid sequencing. The study showed that APN could bind to Cry1Ac and Cry1Fa by both affinity chromatography and ligand blotting (8). APN was also shown to react with SBA, a lectin specific for GalNAc, and chemical deglycosylation with periodate eliminated toxin binding.
Another APN from class 3 was identified in H. armigera (142). In this study, toxin binding to APN expressed exogenously in T. ni cells was studied. The expressed protein was 120 kDa in size, glycosylated, and enzymatically active and was shown by ligand blot analysis to react with Cry1Ac but not with Cry1Ab or Cry1Aa. Although the involvement of GalNAc in binding was not reported, a study by Wang et al. (183) suggests that glycosylation may not be required for this interaction. This group expressed APN in E. coli and showed binding to Cry1Ac with ligand blot analysis.
Class 3 APN from Epiphyas postvittana has been studied in both its native and exogenously expressed forms (160). The native protein was purified from detergent-solubilized BBMV proteins by use of a combination of gel filtration and ion-exchange chromatography. APN expressed in Sf9 cells was purified using a similar method. Both Cry1Ac and Cry1Ba could bind to either form of APN by ligand blotting, but in competitive binding assays, neither toxin bound specifically to Sf9 cells expressing APN. To ensure that APN was being expressed on the cell surface, the researchers measured APN activity in cells before and after lysis and found the values to be comparable. Thus, concerns about the relevance of binding demonstrated by ligand blotting were raised (160).
Finally, there is some evidence that class 3 APNs from B. mori and P. xylostella can bind to Cry1Aa and Cry1Ab, based on reports by Nakanishi et al. (132), where binding to toxin binding regions expressed as GST fusion proteins in E. coli was shown by ligand blot analysis.
(iv) Class 4. Like class 2, class 4 lacks the C-terminal threonine-rich tract found in class 1 and class 3. There are currently nine members in this class, and three have been reported to interact with Cry toxins (2, 8, 132).
Class 4 APN from H. virescens has been reported to be a Cry1Ac-binding protein. This has been demonstrated using several different methods and was first shown by affinity chromatography, where CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate}-solubilized BBMV proteins were passed over a Cry1Ac affinity column and bound proteins eluted with 2 M NaSCN (8). Under these conditions, three major binding proteins were eluted: a 170-kDa protein (class 1 APN), a 120-kDa protein (class 3 APN), and a 110-kDa protein (class 4 APN). Similar results were obtained with Cry1Fa, but no proteins were isolated using Cry1Ea, a toxin previously shown not to bind to H. virescens BBMV. That Cry1Ac could bind to class 4 APN following periodate treatment was evidence that glycosylation was not important for binding (8). In addition, SBA did not bind to this APN, suggesting that GalNAc moieties were not present in its glycans. It thus appeared that the interaction between Cry1Ac and class 4 APN differed from the interaction of Cry1Ac with class 1 and class 3 APN, where GalNAc was believed to be an important determinant of binding (93, 116, 177). Cry toxin binding to exogenously expressed class 4 H. virescens APN was also reported (7) and here it was shown by fluorescence microscopy that Cry1Ac could bind to S2 cells expressing APN but not to control cells. Attempts were also made to express APN in E. coli, but recombinant protein was not produced and thus the importance of glycosylation could not be tested by this method.
Class 4 also includes an APN isolated from Spodoptera litura. This species has been reported to be susceptible to Cry1Ca but resistant to Cry1Ac (2). Agrawal et al. (2) examined the binding of Cry1Ca and Cry1Ac to this APN expressed in Sf21 cells and showed that the protein was glycosylated, enzymatically active, and present on the cell surface. Ligand blot analysis showed that Cry1Ca but not Cry1Ac could bind to the denatured form of the protein. Toxin binding to APN was also studied under nondenaturing conditions (2). CHAPS-solubilized APN was incubated with Cry1Ca and immunoprecipitated with anti-Cry1Ca antibodies attached to protein A-Sepharose beads. Using negative control Sf21 cells or omitting Cry1Ca failed to precipitate the APN. A binding assay was also carried out with intact cells, and immunofluorescence showed that Cry1Ca but not Cry1Ac could bind to cells expressing APN. Attempts to express this protein in E. coli were unsuccessful, however, and thus the importance of glycosylation in toxin binding was not determined.
There is also some evidence that class 4 B. mori APN can bind to Cry1Aa and Cry1Ab, in particular to the proposed toxin binding region described by Nakanishi et al. (132).
(v) Class 5. Class 5 makes up the smallest group of APNs and consists of only two members identified in P. xylostella (132) and H. armigera. These APNs have an altered form of the highly conserved GAMEN motif, where methionine has been changed to threonine. The effect of mutations in this motif on enzymatic activity have been studied in the related protein insulin-regulated aminopeptidase (101). Here it was shown that mutating methionine to isoleucine, lysine, or glutamic acid decreased activity by 16-fold, decreased activity by 30-fold, or completely abolished activity, respectively. Although the enzymatic activity of the class 5 APNs has not been reported, it seems likely that at least some decrease in activity would be expected.
A comparison of the amino acid sequences of the P. xylostella and H. armigera APNs shows that there are marked differences in the threonine-rich C-terminal region. H. armigera APN has the longest reported open reading frame of any of the lepidopteran APNs and has many threonine residues at the C terminal, 32 of which are predicted to be O glycosylated according to the NetOGlyc 3.1 server (84) (http://www.cbs.dtu.dk/services/NetOGlyc/). In contrast, class 5 APN from P. xylostella completely lacks this threonine-rich region and has no predicted O-linked glycosylation sites.
Studies on the binding of class 5 APNs to Cry toxins are limited. The only report is from Nakanishi et al. (132), who showed by ligand blot analysis that a region of P. xylostella APN expressed in E. coli as a GST fusion protein could bind to Cry1Aa and Cry1Ab. Additional research is needed to determine the significance of these interactions and the general importance of class 5 APNs in mediating Cry toxin susceptibility.
(vi) Other APNs. In addition to the APNs already discussed, three other variants have been reported to bind to Cry toxins but have yet to be cloned and sequenced: a 106-kDa protein from M. sexta (114), a 100-kDa protein from the dipteran Anopheles quadrimaculatus (1), and a 96-kDa protein from B. mori (159). Based on immunoprecipitation experiments, the 106-kDa protein from M. sexta appeared to be a Cry1Ca-binding protein, although weaker binding to Cry1Ac was also detected. The N-terminal sequence was nearly identical to that of class 1 M. sexta APN (93, 113, 150), whereas an internal sequence—as it was later discovered—was identical to a region of class 2 M. sexta APN (31). Whether the preparation of 106-kDa APN used for sequencing contained a single novel fusion of class 1 and class 2 APN or two separate APNs derived from each class was never reported. The 100-kDa protein isolated from A. quadrimaculatus was purified from solubilized BBMV and tested for binding to mosquitocidal Cry toxins. By SPR analysis (1), Cry11Ba, but not Cry2Aa, Cry4Ba, or Cry11Aa, was found to bind to the purified protein. Database searches with the N-terminal sequence of the 100-kDa protein led to its classification as an APN. The 96-kDa protein from B. mori was shown to bind to Cry1Ac by ligand blot analysis, and this interaction could be blocked with competing GalNAc (159). The protein was recognized by an antibody with specificity for class 3 B. mori APN, but by peptide mass fingerprinting, only 54% of the peptides could be matched. It was thus proposed that the 96-kDa protein was a novel isoform of class 3 APN.
(vii) Summary. The data on Cry toxin binding to APNs are complex, and several factors make this interaction difficult to study. First, several different APNs are believed to be simultaneously expressed in the larval gut. These proteins share sequence identity and can have similarities in properties such as molecular weight, enzymatic activity, and glycosylation. This can make it difficult to purify a particular APN to homogeneity and equally difficult to prove that the protein is pure. Indeed, mass spectrometry analysis of purified class 1 APN from M. sexta revealed the presence of contaminating class 3 and class 4 APNs (163). Although exogenous expression of cloned APNs is a possible solution, this can sometimes be difficult (7, 31, 48) and even when successful, tissue- or organism-specific differences in posttranslational modification may eliminate the toxin binding site (48, 115). To further complicate matters, different methods of studying toxin-APN interactions can sometimes give conflicting results (31, 132, 160). This was studied in detail by Daniel et al. (26), who showed that denaturing M. sexta APN or Cry1A toxins exposes binding epitopes hidden under nondenaturing conditions. Studies on Cry toxin binding to APN may also be complicated by the presence of cadherins, a second class of Cry toxin receptor particularly sensitive to proteolytic degradation (23, 119, 174). Cadherins have been shown to form approximately 120-kDa degradation products that could possibly be misinterpreted as APN in ligand blot assays (119).
Figure 8 presents a summary of the reported binding between Cry toxins and exogenous or endogenous nondenatured or denatured APN. While it is clear from this figure that several toxin-APN combinations have yet to be explored, some general conclusions can be drawn. Cry1Aa and Cry1Ab are best characterized as class 1 APN-binding proteins. Binding to both endogenous and exogenous forms of APN has been observed for several species and, as shown in Fig. 8, a lack of binding to any class 1 member has been reported in one case only (80). Binding to other APN classes under nondenaturing conditions either has not been reported or has not been observed. Cry1Ac has specificity broader than those of Cry1Aa and Cry1Ab and seems to be primarily a class 1 and class 3 APN-binding protein. Binding to class 1 APNs appears to occur at two sites, one of which it shares with Cry1Aa and Cry1Ab (120). Binding to the other site seems to be GalNAc dependent, and given that both class 1 and class 3 have a threonine-rich region predicted to be highly glycosylated, it is tempting to speculate that Cry1Ac mediates contact to both APNs in this region. As for the remaining toxins, it is difficult to make generalizations about their binding specificity based on the limited data available, and additional studies must be carried out to better characterize these proteins.
|
(i) Permeability. The insecticidal nature of Cry toxins is generally believed to be due to their ability to form pores in the midgut of susceptible organisms (96), and assays have been developed to assess whether putative toxin receptors can enhance pore formation. The 86Rb+ efflux assay has been used for this purpose, where pore formation is indicated by the release of 86Rb+ from phospholipid vesicles containing putative receptor. Sangadala et al. (150) used this technique to demonstrate that a mixture of class 1 APN and phosphatase from M. sexta could enhance Cry1Ac pore formation. When reconstituted into phospholipid vesicles, these proteins were reported to increase toxin binding by 35% and to enhance toxin induced 86Rb+ release 1,000-fold relative to protein-free vesicles. Similar results were obtained by Luo et al. (116), who showed that class 1 APN purified from H. virescens could enhance Cry1Aa-, Cry1Ab-, or Cry1Ac-induced release of 86Rb+ but had no effect on Cry1Ca-induced release; Cry1a is a toxin shown not to bind to this class of APN (116). Pore formation has also been studied by measuring toxin channel activity in planar lipid bilayers. Schwartz et al. (153) showed that the inclusion of a purified M. sexta receptor complex in phospholipid bilayers caused Cry1Aa, Cry1Ac, and Cry1Ca to form channels at concentrations much lower than that in receptor free membranes. Analysis of this receptor complex by ligand blotting suggested that class 1 APN was the major Cry-binding protein.
(ii) In vitro toxicity. While assays that measure membrane permeability are good indicators of pore formation, they do not necessarily predict whether a receptor will confer toxin susceptibility to an organism. A more direct approach is to test whether Cry toxin-resistant cell lines can be made susceptible by expressing putative toxin receptors. So far, testing APNs by this method has been relatively unsuccessful. Garner et al. (48) expressed class 3 APN from L. dispar in Sf9 cells but did not observe cytotoxicity at Cry1Ac concentrations between 0.2 and 50 µg/ml. Because the Sf9 cells expressed APN with a binding affinity for Cry1Ac much lower than that of the native protein, the experiment was somewhat inconclusive. Class 4 APN from H. virescens was also tested for its ability to confer toxin susceptibility and was expressed in S2 cells (7). While it was demonstrated that Cry1Ac could bind to APN on the surface of intact cells, cytotoxicity was not observed at a toxin concentration of 30 µg/ml. The limited number of in vitro cytotoxicity-based studies is likely due to difficulties in correctly expressing APN, and obtaining proper glycosylation seems to be the main obstacle (48, 115). If these problems could be resolved, this method of receptor validation may become more useful.
(iii) In vivo toxicity. In vivo methods have also been used to test APN receptors for functionality. Gill and Ellar (50) fed Cry1Ac to transgenic Drosophila larvae expressing class 1 APN from M. sexta and showed 100% toxicity at a toxin concentration of 50 µg/ml. In comparison, control larvae were resistant to Cry1Ac at concentrations up to 1 mg/ml, the highest concentration tested. The expression of M. sexta APN was confirmed by ligand blotting with Cry1Ac, and although expression levels were low, the receptor binding determinants were apparently intact. These results suggested that APN expressed in the Drosophila midgut may be properly glycosylated and that in vivo expression systems may be more suitable for evaluating toxin-receptor interactions than those based on cell lines.
Gene silencing has also been used to determine whether APN can confer toxin susceptibility. Rajagopal et al. (143) injected S. litura larvae with double-stranded RNA corresponding to a region of the class 4 APN gene and showed a 95% reduction in transcript levels over what was shown for control larvae. In addition, an 80% reduction in APN expression was observed, as determined by immunoblot analysis of BBMV proteins. When treated with Cry1Ca, a 75% reduction in mortality was observed in larvae previously injected with double-stranded RNA. These results suggest that class 4 APN can confer Cry1Ca susceptibility to S. litura and demonstrate that gene silencing may be an effective way to study the biological significance of toxin-receptor interactions.
(iv) Summary. The biological relevance of the Cry toxin-APN interaction has yet to be studied extensively. To date, 17 different APNs have been reported to bind to Cry toxins, and yet only 2 have been shown to mediate toxin susceptibility. Twelve of the 17 APNs have not been studied for functionality by any method. In vivo methods of testing APN functionality have shown considerable promise, and using these methods to study the remaining APNs may lead to a better understanding of the overall importance of this class of Cry toxin receptor.
In 1993, a novel cadherin-like protein was isolated from the midgut epithelium of M. sexta by virtue of its binding affinity for Cry1Ab (174). The protein was cloned in 1995 and sequence analysis predicted a signal peptide, 12 cadherin repeats, a membrane proximal extracellular domain, a transmembrane domain, and a small cytoplasmic domain (33, 175). Since then, additional lepidopteran cadherins have been identified, and all have been shown to have a similar domain organization (40, 44, 126, 130, 184). In M. sexta cadherin, additional features, such as the cell adhesion sequence HAV (10) and the integrin-binding sequences RGD (149) and LDV (98, 170), have been identified in the ectodomain; however, the functional role of these sequences has not yet been confirmed (33). In contrast, an analysis of the cytoplasmic domain did not reveal sequences known to interact with intracellular proteins such as catenins (33). While classical cadherins are located primarily within adherens junctions involved in cell-cell adhesion (4), lepidopteran cadherin-like proteins have been identified on the apical membrane of midgut columnar epithelial cells (3, 24, 66, 124), the target site of Cry toxins (16, 19, 20, 24). The expression of cadherin has been shown to vary with developmental stage and increases progressively from the first to the fifth instar in M. sexta larvae (124). In eggs and adults, however, cadherin expression has not been detected. Although the exact physiological function of midgut cadherins is not known, the tight control of cadherin levels during larval development has been proposed to indicate their importance in maintaining midgut epithelial organization (124).
Lepidopteran cadherin-like proteins have been extensively studied as Cry1A receptors, and there is good evidence to suggest they play a critical role in mediating toxin susceptibility. The following sections describe the best-characterized cadherin-like proteins and their interactions with toxins of the Cry1A family.
BT-R1 (Manduca sexta). The first cadherin-like protein shown to interact with Cry toxins, BT-R1, was a 210-kDa glycoprotein identified in M. sexta BBMV (174). The protein was purified by immunoprecipitation with Cry1Ab followed by two-dimensional gel electrophoresis. Partial sequence information derived from the purified receptor facilitated cloning, and the identified gene was 30 to 60% similar and 20 to 40% identical to other members of the cadherin superfamily (175).
To confirm that BT-R1 was a genuine Cry1Ab receptor, it was expressed in cultured cells. The protein was first expressed in mammalian COS-7 and HEK-293 cells and by ligand blot analysis was detected as a 195-kDa band by probing with Cry1Ab (175). Cry1Ab binding to intact cells expressing BT-R1 was also demonstrated, and the measured dissociation constant of 1 nM was similar to that of the native receptor. BT-R1 was subsequently expressed in insect-derived Sf21 cells and was shown to bind to Cry1Aa and Cry1Ac but not to Cry3Aa and Cry11Aa, which are not toxic to M. sexta (91). In addition, competition binding studies showed that Cry1Aa and Cry1Ac could block Cry1Ab binding to membranes prepared from Sf21 cells expressing BT-R1, suggesting that the toxins bind to a common epitope. Taken together, these results showed that Cry1A toxins could bind to endogenously or exogenously expressed BT-R1 under both denaturing and nondenaturing conditions.
BT-R1 was also tested for its ability to confer toxin susceptibility. Initially, HEK-293, COS-7, or Sf21 cells transfected with BT-R1 failed to show any phenotypic changes when exposed to activated Cry1Ab, even at concentrations as high as 100 µg/ml (91). This unexpected result was explained when a frameshift mutation was discovered in the original cDNA clone (33). A revised analysis of the protein sequence showed that the frameshift mutation occurred upstream of the transmembrane domain, thus explaining why the protein was not embedded in the cytoplasmic membrane in earlier experiments and why the observed molecular weight of exogenously expressed BT-R1 was less than that of the native protein. Subsequently, the error-free protein localized to the cell surface and rendered COS-7 cells sensitive to Cry1Ab at 0.6 µg/ml (33). S2 cells expressing BT-R1 were also susceptible to Cry1A toxins, and 12 to 14% of cells were killed by Cry1Aa, Cry1Ab, or Cry1Ac at 20 µg/ml (75). The toxicity of Cry1Ab towards H5 cells expressing BT-R1 was determined at a range of concentrations, and the 50% lethal concentration was reported to be 65 nM (about 4 µg/ml) (195). These results strongly suggest that BT-R1 is an important determinant of Cry1A toxin specificity.
BtR175 (Bombyx mori). A cadherin-like protein was also identified as a Cry toxin receptor in B. mori. In this case, a 175-kDa glycoprotein, BtR175, was identified as a Cry1Aa receptor by immunoprecipitation (130, 131). Partial N-terminal sequencing of the purified receptor led to the cloning of a gene that shared significant homology to the cadherin superfamily of proteins and 69.5% identity to M. sexta BT-R1. The predicted molecular mass of the encoded protein (193.3 kDa) was larger than that of the 175-kDa natural protein, and it was believed that BtR175 was expressed as a proprotein. This was confirmed when the gene was expressed in Sf9 cells and a 175-kDa band comigrated with BtR175 isolated from BBMV. It was postulated that the sequence 288RPPRWV292 may be an endoproteolytic cleavage signal that when cut gives rise to a mature BtR175 with only nine cadherin repeats (130). Interestingly, the proposed cleavage signal is also present in M. sexta BT-R1, where there is no apparent cleavage at this site.
A second group has independently purified and partially sequenced a Cry1Aa receptor with a reported molecular mass of 180 kDa (77). The 103-amino-acid sequence obtained by this group was identical to a region within the sequence of BtR175 previously reported by Nagamatsu et al. (130). A later publication by the same group reported the sequence of three allelic BtR175 variants that differed from BtR175 by one, five, or six amino acids (78). All three receptors bound to Cry1Aa with a similar binding affinity (3.6 to 6.4 nM), although transient expression levels in COS7 cells varied considerably.
To further demonstrate the importance of BtR175 as a toxin receptor, various cell types expressing the gene were tested for cytotoxicity. Nagamatsu et al. (129) showed that exposure to 8 µg/ml Cry1Aa caused BtR175-expressing Sf9 cells, but not control cells, to swell within 15 min, and the number of swollen cells increased for 45 min after the addition of the toxin. These changes were quite similar to those of midgut columnar cells in B. mori fed with Cry1Aa and to those of epithelial cells isolated from the midgut and treated with toxin ex vivo (66). In another study, Cry1Aa caused cell swelling and cytotoxicity in mammalian cells expressing BtR175b (an allelic variant of BtR175) (171). Cry1Ab and Cry1Ac had similar albeit weaker effects, in correlation with their lower binding affinities for BtR175 (80, 171). This work demonstrated that BtR175 could confer Cry1A susceptibility outside of an insect system and made it possible to rule out the requirement for other insect specific factors in cytotoxicity.
To determine whether Cry1Aa-induced cell swelling was due to changes in ionic permeability, the membrane currents of Sf9 cells expressing BtR175 with or without the toxin binding region (discussed later) were compared (129). The toxin-induced currents of cells expressing the toxin binding region increased dramatically upon the addition of Cry1Aa, whereas no appreciable difference was observed with the control cells. These results suggested that pore formation leading to aberrations in osmoregulation was responsible for the observed morphological changes in the cells.
HevCaLP (Heliothis virescens). In H. virescens, genetics preceded biochemistry in identifying a cadherin-like Cry toxin receptor. This was accomplished by Gahan et al. (44), who studied a laboratory strain of H. virescens, YHD2, with a high level of recessive resistance to Cry1Ac (resistance ratio, 10,128x). Genetic studies revealed that a single major gene was responsible for 40 to 80% of resistance. With the knowledge that in some insects, resistance is accompanied by a loss in toxin binding, the researchers tested the genes of known Cry toxin-binding proteins to see whether they mapped to the region that conferred resistance. Two genes encoding APNs (class 1 and class 3) were tested for linkage, but they mapped to different regions of the genome. It was known that cadherin-like proteins bound to Cry toxins, but they had yet to be isolated from H. virescens. For this reason, the researchers searched for and found a BtR175 homologue in a susceptible strain. The gene was 70% identical to BtR175 and was named HevCaLP. Subsequently, the gene was mapped in resistant insects and found to reside in the resistance locus (44). The allele in resistant strains (r1) differed from the allele in susceptible strains (s1) by the presence of a 2.3-kb insert with hallmarks of a long terminal repeat-type retrotransposon. The insertion introduced a stop codon that truncated the encoded protein prior to the predicted transmembrane domain; thus, an explanation for why the mutation may have conferred resistance was provided.
The importance of HevCaLP as a toxin receptor was further studied by looking at the correlation between expression, binding, and toxin susceptibility in strains believed to have different mechanisms of Cry toxin resistance (89). Initially, it was confirmed that only s1 homozygotes or heterozygotes expressed full-length HevCaLP. It was subsequently shown that HevCaLP expression was necessary for Cry1Aa, but not Cry1Ab or Cry1Ac, binding to BBMV. This was in agreement with earlier studies showing that Cry1Ab and Cry1Ac bind to multiple sites on H. virescens BBMV, whereas Cry1Aa binds to a single site (88, 179). Confirmation that Cry1Ab and Cry1Ac could also bind to H. virescens cadherin was later provided by Xie et al., who expressed the receptor recombinantly in E. coli (190).
More recently, HevCaLP has been expressed in cell lines to determine whether the receptor can confer toxin susceptibility. Drosophila S2 cells expressing HevCaLP were sensitive to Cry1Aa, Cry1Ab, and Cry1Ac but, unexpectedly, not to Cry1Fa (87). Human embryonic kidney (HEK) cells expressing H. virescens cadherin were also treated with Cry1Ab or Cry1Ac, and although membrane blebbing was observed in some cells, cytotoxicity could not be demonstrated (3). Based on these results, it was proposed that other receptors, such as ALP or aminopeptidase, may be necessary for full toxicity (3, 87).
Cadherin-like proteins in other species. A link between cadherin-like proteins and Cry toxin susceptibility has been demonstrated for several other lepidopteran species. In 2005, two groups independently reported the sequence of a cadherin-like protein in Ostrinia nubilalis (25, 40). Flannagan et al. (40) cloned and expressed the putative Cry1Ab receptor in Sf9 cells and showed toxin susceptibility at concentrations as low as 0.1 µg/ml. Morin et al. (126) reported the sequence of a cadherin-like gene in Pectinophora gossypiella and identified three mutant alleles linked with resistance to Cry1Ac. In 2005, Xu et al. (191) published the sequence of a cadherin-like gene in H. armigera and found that disruption of the gene by a premature stop codon was linked to Cry1Ac resistance. Another group (184) demonstrated Cry1Ac binding to a recombinant H. armigera cadherin-like protein and using semiquantitative reverse transcription-PCR showed reduced gene expression in a strain resistant to Cry1Ac. A gene encoding a cadherin-like protein was also cloned from L. dispar, and the E. coli-expressed protein was reported to bind to Cry1A toxins (89a). In addition, insect cells expressing the gene were rendered toxin susceptible. Finally, the sequences of several other lepidopteran cadherin-like proteins have been deposited in GenBank (http://www.ncbi.nlm.nih.gov/GenBank/index.html), including those of P. xylostella, Chilo suppressalis, Helicoverpa zea, Agrotis ipsilon, and Spodoptera frugiperda. Figure 9 shows the phylogenetic relationship of reported lepidopteran cadherins.
|
Although cadherin-like proteins are clearly important mediators of Cry1A susceptibility, it seems unlikely that they are universal Cry toxin receptors. For instance, the H. virescens strain YHD2 expresses a truncated form of HevCaLP and is highly resistant to Cry1A toxins but shows little cross-resistance to Cry2Aa, Cry1Ca, or Cry1Ba (59). Whether APNs, glycolipids, ALPs, or a yet to be discovered class of receptor mediates specificity for these toxins remains to be investigated.
In H. virescens, ALP is a 68-kDa GPI-anchored membrane glycoprotein (85). Binding to Cry1Ac was demonstrated by ligand blot analysis of BBMV and appears to be dependent on the presence of an N-linked oligosaccharide containing a terminal GalNAc residue. Interestingly, ALP expression levels were reduced in a resistant strain of H. virescens, suggesting a functional role in toxicity. The presence of a GPI anchor and the importance of GalNAc in toxin binding shows clear parallels to APN and its interaction with Cry1Ac (93, 116, 177).
In M. sexta, a 65-kDa BBMV protein was identified as a Cry1Ac-binding protein by two-dimensional gel electrophoresis followed by ligand blot analysis (122). It was identified as ALP by database searches of peptide mass fingerprints and by detection with an ALP-specific antibody. The protein was predicted to be GPI anchored but was not present in the pool of proteins released from BBMV by PI
