INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacillus thuringiensis is a gram-positive bacterium which, during sporulation, produces parasporal crystalline inclusions consisting of one or several delta-endotoxins, or Cry proteins, which have insecticidal properties. The cry gene family is a large, still-growing family of homologous genes, and each gene encodes a protein with activity against only one or a few insect species. To be toxic, B. thuringiensis crystals have to be ingested by the insect. In the insect midgut, the crystals are dissolved, releasing their constituent Cry proteins as protoxins with molecular masses of 70 to 130 kDa. Midgut proteases subsequently trim the protoxins, resulting in a truncated, N-terminal fragment with a mass of approximately 65 kDa, which is the activated toxin. The activated toxin then binds to specific receptors on the surface of the midgut epithelial cells and penetrates the cell membranes, forming pores and killing the epithelial cells by colloid osmotic lysis (reviewed in reference 13). Much, although not all, of the observed specificity of the Cry proteins is determined by the interaction of the toxin with specific receptors, if present, on the midgut epithelial cell membranes (14).

Progress has been made both in determining the three-dimensional structure of the toxin molecule and in identifying the primary sequences involved in specificity and receptor binding, allowing the study of structure-function relationships. The published structures of two delta-endotoxins show a three-domain structure (8, 10). The N-terminal domain I consists of seven alpha-helices and is thought to be responsible for insertions into the insect cell membrane and to be involved in pore formation. The more variable domain II contains the primary sequences that have been shown to be involved in insect specificity and in high-affinity binding (13). Domain II is therefore assumed to be involved in actual interactions with receptors and, to a large extent, in determining specificity through that process. The function of the C-terminal domain III at the molecular level was unknown until recently, although a variety of mutagenesis and recombination experiments have shown that it can be involved in specificity. Most notably, an exchange of domain III (by in vivo recombination) between toxins with differing specificities can change the specificity and even create new hybrid toxins with increased activity (2, 5, 7, 11). Our group, as well as others, has shown that a substitution of domain III alters the binding of toxins to putative receptors on ligand blots of brush border membrane vesicle proteins separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (1, 5, 6, 9). Recently, using various different binding assays, we have shown that domain III of Cry1Ac is involved in N-acetylgalactosamine (GalNAc)-inhibited binding to the putative receptor in Manduca sexta, aminopeptidase N (APN) (4). These recent studies make it increasingly likely that the role of domain III in determining insect specificity is probably correlated with its ability to bind, with or without domain II, to specific receptors on the surfaces of the gut epithelial cells of the target insect.

In contrast to the numerous studies on the role of domain II sequences in specificity and target membrane binding, few data on the role of domain III amino acids in these functions are currently available. A mutation of three amino acids in domain III of Cry1Ac was shown to strongly decrease GalNAc-inhibited binding by Cry1Ac to M. sexta membranes, implicating a role of these residues in a physical interaction between domain III and GalNAc or a similar sugar on APN (3). In yet another study, a mutation of two amino acids in domain III of Cry1Ac was shown to diminish its toxicity for M. sexta as well as for Heliothis virescens, while binding to putative receptors on ligand blots of brush border membrane proteins from those insects had decreased as well. In the latter study, the effects on toxicity for M. sexta were larger than those on toxicity for H. virescens (1).

Exchanging domains III of Cry1C, which is active against Spodoptera exigua, and Cry1E, which is not active, resulted in the S. exigua-active Cry1E-Cry1C hybrid G27 (2). These results indicated that domain III of Cry1C is involved in specificity for S. exigua. This was confirmed by our finding that domain III of Cry1C can also render the inactive toxin Cry1Ab active against S. exigua (5). In this study, we characterized intradomain hybrids between domains III of Cry1E and Cry1E-Cry1C hybrid protein G27 (homologue scanning) in order to more precisely determine which parts of domain III of Cry1C are involved in S. exigua-specific activity. In addition, we identified individual amino acids that are involved in this specificity by mutagenesis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. Cry1E expression plasmid pBD160, expression plasmid pBD151 containing XmnI (base 1853)-truncated cry1C, and cry1E-cry1C hybrids G27, H7, H8, and H17 were described earlier (2). For the recombination of the 3' end of the hybrid G27 gene with cry1E, tandem plasmid pHK4 was constructed (Fig. 1). For this purpose, the 3' part of the G27 gene, a BstBI-XhoI fragment (from base 1488 of cry1C to its end) was replaced by the corresponding fragment of pBD151, which contains the truncated 3' end of the active-Cry1C-encoding DNA (up to base 1835) followed by the polylinker of pBluescript SK(+). Subsequently, an EcoRV-XhoI fragment of pBD160 containing the 3' end of the Cry1E toxin-encoding DNA from base 1473 to the end of the gene was cloned into the vector fragment of SmaI-XhoI-digested pBD151.

View larger version (7K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the G27 gene-cry1E tandem plasmid pHK4. Open and dashed boxes represent Cry1E- and Cry1C-encoding sequences, respectively. Locations of restriction enzyme sites used for cloning are indicated. Dotted vertical lines indicate the borders of the domain III-encoding sequences. For clarity, the overlapping regions of G27 and Cry1E are aligned vertically, and the polylinker between the two genes is shown as cut by NotI and SpeI.

Mutagenesis. All mutations were made with the QuickChange kit (Stratagene) and complementary mutagenic oligonucleotides. Mutant plasmids pNS1, pNS20, pNS22, and pB16 were derived from cry1E. pNS2, pNS3, pNS21, and pNS23 were derived from H8. All mutations were checked by DNA sequencing.

In vivo recombination. E. coli JM101 (recA⁺) was transformed with pHK4, and then plasmid DNA was isolated. To select for recombinant plasmids, DNA was digested with NotI and SpeI, which have unique sites in the polylinker region between the G27- and Cry1E-encoding parts (Fig. 1). Digested DNA was transferred to E. coli XL-1 by transformation, and transformants were screened for recombination events by restriction analysis of isolated DNA.

Toxin production, purification, and bioassays. All protoxins were produced in E. coli XL-1, and trypsin-activated toxins were purified by fast protein liquid chromatography as described earlier (2). The toxicities of the proteins were tested by spreading toxin dilutions on artificial diet. Neonate larvae of S. exigua were used, and their mortality was scored after 6 days at 28°C. For M. sexta bioassays, 1-day-old larvae were used and their mortality was scored after 6 days at 28°C. The concentrations with 50% lethality (LC₅₀s) and 95% fiducial limits were estimated by a Probit analysis of results from three or more independent experiments, using the PoloPC computer program (12).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Substitution of N-terminal domain III sequences. Our original Cry1E-Cry1C hybrid protein G27 consisted of domains I and II, as well as the most N-terminal part of domain III of Cry1E, while the rest of the protein, comprising the major C-terminal part of domain III as well as the protoxin-specific fragment, was derived from Cry1C (Fig. 2 and 3). In contrast to Cry1E, Cry1C and G27 are toxic to S. exigua, indicating that the Cry1C-derived part of G27 contains sequences that are specifically required for activity against this insect. Besides G27, the original set of Cry1E-Cry1C hybrids contained a number of hybrids with smaller contributions of domain III from Cry1C (Fig. 2 and 3, hybrids H8, H17, and H7) (2). The locations of the crossover points define blocks (A through F) of the domain III sequence which were replaced in the various hybrids produced in that study (Fig. 2 and 3). Whereas G27 has high activity against S. exigua, the activities of the other hybrids decreased rapidly as the crossover point moved further into domain III (i.e., less of domain III from Cry1C) (Fig. 3). A replacement in G27 of only three amino acids (block B) by their corresponding amino acids from Cry1E (H8) decreased activity approximately 10-fold. A further replacement of two amino acids in block C (H17) decreased activity again fivefold. Whereas both H8 and H17 were still distinctively more toxic than Cry1E, a further replacement of block D (H7) left no detectable activity against S. exigua. At the same time, G27, H8, and H17 were equally active against M. sexta (Fig. 3), suggesting that blocks B and C are involved in specific activity against S. exigua. H7 also had no activity against M. sexta, suggesting that a simultaneous substitution of blocks B to D destroys biological activity in a less specific way.

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Amino acid alignment of domains III of Cry1E and Cry1C. Identical amino acids of Cry1C are depicted by dots. Arrowheads indicate approximate locations of crossover sites in Cry1E-Cry1C hybrids (above) and in G27-Cry1E-hybrids (below). Because of the homology in these regions, exact locations of the crossover sites could not be determined. On top of the alignment, the relative positions of amino acid blocks A through F are depicted by dashes. I1 and I2 indicate the positions of two insertions in Cry1C. Amino acid numbering follows that of G27.

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Schematic representation of domains III of wild-type and hybrid toxins used in this study. Vertical lines represent the amino acid residues derived from Cry1C, which are different in Cry1E. At the top, the relative positions of amino acid blocks A through F are depicted by horizontal lines. I1 and I2 indicate the positions of two insertions. Toxicity for S. exigua and M. sexta is represented as LC50, in nanograms per square centimeter. FDL, fiducial limits.

Substitution of C-terminal domain III sequences. To test the contributions of C-terminal amino acids in domain III and in the protoxin of Cry1C to the activity of G27, a new tandem plasmid for in vivo recombination, pHK4, was constructed. This plasmid contains the gene encoding G27, which was truncated at the end of the domain III-encoding part, followed by a polylinker and the parts of the cry1E gene encoding domain III and its protoxin tail to the 3' end (Fig. 1). This allowed in vivo recombination between the domain III-encoding parts of the G27 gene and cry1E, in effect replacing 3' sequences of the G27 gene with those of cry1E. New recombinant toxins were screened at the DNA level by restriction analysis and sequencing to obtain crossover events throughout domain III without prior selection for soluble-protoxin-encoding genes of hybrid toxins. Hybrids BS21, NS6, NS8, and BS22 had increasing numbers of their C-terminal amino acids replaced by the corresponding Cry1E residues (Fig. 3). Only hybrids BS21 and NS6 produced soluble protoxins, and they also produced, upon trypsin treatment, a stable activated toxin. Hybrid BS21 had a crossover site near the C-terminal end of domain III, making it a protoxin that consisted mainly of Cry1E sequences with the exception of a part of domain III which was derived from Cry1C (Fig. 3). Trypsin-activated BS21 protein retained the activity of G27 against S. exigua, indicating that specificity for S. exigua is conferred by just this part (comprising blocks B, C, D, and E) of domain III of Cry1C (Fig. 3). A further replacement of domain III C-terminal amino acids (block E), as seen with hybrid NS6, decreased the activity of the resulting toxin against both S. exigua and M. sexta (Fig. 3). Still further replacement by Cry1E sequences (blocks D and C, respectively), as seen with hybrids NS8 and BS22, resulted in hybrid genes for which no soluble protoxin could be recovered. An analysis of whole-cell lysates showed that both genes did express proteins of the correct, expected size, although in amounts smaller (approximately 10%) than that of G27 (results not shown). Nonetheless, no protein was recovered during attempts to solubilize these protoxins. Therefore, they were not tested further for toxicity.

Mutagenesis of block B amino acids. A comparison of the toxicities of G27 and H8 showed that a replacement of only three amino acids (block B) by their corresponding Cry1E residues severely and specifically affected activity against S. exigua, but not activity against M. sexta (see above). To further study the contribution of each of these three amino acids to its specificity for S. exigua, we made a number of mutants of Cry1E in which one or two of these amino acids were replaced by their equivalents from Cry1C (Table 1). Our results with hybrid BS22 (above) had already shown that a simultaneous replacement of all three would result in the inability to recover a soluble protoxin. Replacements of neighboring Leu-475 and Gly-476 (mutant NS1) also resulted in no soluble protoxins. Individual replacements Leu475Val, Gly476Trp, and Lys483Thr to produce mutants NS22, NS20, and PB16, respectively, did result in the recovery of soluble protoxins. Only one mutant, NS20, was significantly more active against S. exigua than was Cry1E (Table 1), although only at a low level. This indicated that although one or more of these Cry1C amino acids was essential for specific activity, no single one was sufficient to raise the activity to a level comparable to that of G27. As the presence of other specific Cry1C amino acids was apparently required, we repeated the mutagenesis described above in the background of hybrid H8, which contained the rest of domain III from Cry1C. The simultaneous replacements of Leu475Val and Gly476Trp, as well as the replacement of only one of each of the three amino acids (NS2, NS23, NS21, and NS3), resulted in the production of soluble protoxins. However, only NS2 and NS21 (replacements Leu475Val and Gly476Trp, and Gly476Trp alone, respectively) were as toxic to S. exigua as was G27, while NS23 and NS3 were only as toxic as H8 at best (Table 1). This shows that of the three amino acids in block B, only Trp-476 is required, and sufficient, for optimal activity against S. exigua.

Mutagenesis of block D amino acids. The initial experiments, which compared hybrids H17 and H7 (above), indicated that block D, which is of a different origin in these hybrids, plays a role in activity against S. exigua. However, because H17 already has low (but significant) activity, and because a simultaneous substitution of blocks B to D (H7) apparently destroys biological activity, the exact level of contribution of block D to toxicity for S. exigua is unclear. Interestingly, a major part of the amino acid differences between Cry1C and Cry1E domains III is concentrated in this block. To study the role of this block more closely, we produced a mosaic of hybrids NS8 and H7, called NS10, which effectively resulted in a replacement of block D amino acids in G27 by their equivalents from Cry1E (Fig. 3). In contrast to hybrid NS8, which yielded no soluble protoxin, hybrid NS10 yielded a soluble protoxin and a stable activated toxin upon trypsin treatment. This toxin had no detectable activity against S. exigua, but it was active against M. sexta (Fig. 3), indicating a major contribution of this block to S. exigua-specific activity. Curiously, NS10 was approximately seven times more toxic to M. sexta than was G27.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Cry1C is the most toxic, natural B. thuringiensis delta-endotoxin for S. exigua described so far. Domain III of Cry1C is an important factor in this toxicity, as a transfer of Cry1C domain III can change the inactive toxins Cry1E and Cry1Ab into S. exigua-active hybrid toxins (2, 5). In this study, we made use of the relatively high homology between the domain III-encoding parts of cry1C and cry1E to create a variety of intradomain hybrids by in vivo recombination in order to more precisely identify the parts of the Cry1C domain III which are responsible for specificity for S. exigua. Study of these intradomain hybrids not only allows conclusions on the requirements for activity against S. exigua but also provides information about interactions between different parts of the primary sequence of domain III.

For the convenience of the discussion we divided the amino acid differences between the domains III of Cry1C and of Cry1E into six blocks, each containing between 1 and 19 residue differences, which are defined by the crossover sites of the hybrid toxins used in this study (Fig. 2 and 3). A comparison of Cry1C, Cry1E, and BS21 showed that only blocks B through E of Cry1C, or parts thereof, are essential for a high level of activity against S. exigua (Fig. 3). As we have found no soluble, trypsin-stable hybrid toxin that contains all of domain III of Cry1C, including block A, we can draw no conclusions on the role of this block. The replacement of the C-terminal (protoxin-specific) part of G27 by a Cry1E sequence (hybrid BS21) had no effect on toxicity. This shows that the difference between G27 and Cry1E toxicities is not likely to be caused by differences in processing by trypsin, which could result from differences in the protoxin-encoding parts of these genes. Furthermore, the conservative substitution of block F (Ile609Leu) had no effect on toxicity.

Replacements of block B were shown to have a strong negative effect on the activity of G27 against S. exigua (Fig. 3, H8 and G27). Further study of the role of the three individual residue differences in this block revealed that the replacement of only a single amino acid, Trp-476, could account entirely for this effect (Table 1). However, this residue could exert maximal positive effect on the activity against S. exigua only in conjunction with other Cry1C sequences from blocks C through E (Table 1, NS20 and NS21). Interestingly, the simultaneous placement of the neighboring Val-475 and Trp-476 into a Cry1E background, as well as of the whole block B (BS22) or blocks B and C (NS8), yielded no soluble protoxin. This suggests that these two-amino-acid substitutions may interfere with proper folding in the absence of other complementary Cry1C domain III sequences. This may have led to degradation in E. coli by intracellular proteases, resulting in the inability to recover any soluble protoxin. The presence of either Cry1C block D or block E apparently restored proper folding, as hybrids NS6 and NS10 are soluble and stable upon trypsin treatment.

Either one or both of the amino acids (Phe-498 and Glu-500) of block C are involved in activity against S. exigua, as their replacement further decreased toxicity (Fig. 3, H17). An alignment of domains III of Cry1Ca, Cry1Ac, and Cry1Aa (for which the three-dimensional structure has been determined) (8) (Fig. 4) placed these two amino acids in approximately the same location as two Cry1Ac amino acids (serines 503 and 504) that were shown to play a role in toxicity for M. sexta and H. virescens (1) and close to the location of the three Cry1Ac amino acids involved in GalNAc-inhibited binding to M. sexta APN (3).

View larger version (12K): [in this window] [in a new window]   FIG. 4.   Alignment of domain III amino acid sequences encompassing block C of G27. Cry1Aa amino acids in -sheets 15 and 16 are in italics. Amino acids identical to those of Cry1Aa are depicted by dots. Cry1Ac amino acids implicated in toxicity or specificity and G27 amino acids of block C (see Discussion) are bold and underlined.

A replacement of block D in G27 (Fig. 3, NS10) resulted in the complete destruction of activity against S. exigua without negatively affecting activity against M. sexta, indicating a major role of this block, which contains the most residue differences, in specificity. These differences are concentrated in one area around a seven-amino-acid insertion in Cry1C (around residue 540), on which we focused in further experiments. As homology between Cry1C and Cry1E in this area is low, the validity of the alignment around that area, and hence the exact location of the insertion (if continuous) in the sequence, as shown in Fig. 2, is uncertain. Therefore, we chose to study, by deletion and alanine substitution, two consecutive stretches of four amino acids (Table 2). Both deletions completely destroyed toxicity for S. exigua, indicating that either these residues are directly involved in an interaction with the target insect or the deletion of these residues perturbs the tertiary structure of the region, which could be essential for that interaction. Alanine substitutions of the first two amino acids (Tyr534Ala and Gly535Ala) of the first four-amino-acid stretch had only a slight effect on toxicity. On the other hand, alanine substitutions of the second four-amino-acid stretch (Ser-538 to Val-541) also completely destroyed activity against S. exigua, indicating that the identity of one or more of these four residues is a major determinant of specificity for S. exigua.

Initially, the toxicities of the hybrids and mutants to M. sexta were considered controls for checking to what extent changes affected S. exigua-specific activity as opposed to more general, nonspecific changes, which could result from suboptimal protein function due to, for example, improper folding. Since Cry1C and Cry1E are approximately equally active against M. sexta, any functional hybrid toxin derived from a combination of these two would be expected to be as active as G27 against M. sexta, or less so in the case of incompatibility of the different parts leading to misfolding. Surprisingly, hybrid NS10, in which block D of G27 was replaced by that of Cry1E, was significantly more active against M. sexta (approximately sevenfold) than G27 (Fig. 3). Also, the mutations within this block, described in Table 2, all led to higher toxicity for M. sexta. This indicates not only that block D of G27, or a part thereof, is strongly involved in high activity against S. exigua but also that its replacement in G27 by the corresponding block from Cry1E increases the toxicity of the resulting hybrid for M. sexta. A similar effect is achieved by short deletions or, to a lesser extent, by alanine substitutions within that block.

A replacement of block E of G27 by the corresponding block of Cry1E (hybrid NS6) severely diminished activity against both S. exigua and M. sexta. At present, we have no explanation for this effect. Although the protoxin of NS6 was readily solubilized and appeared to yield a stable toxin upon trypsin treatment, the combination of Cry1C and Cry1E parts may form a product which is less stable in the gut of both insects, leading to lower toxicity regardless of the insect species. This makes it impossible to determine the specificity-determining function of this block as a whole, although it may still be possible in the future to identify individual amino acid substitutions that affect activity against S. exigua and not that against M. sexta.

In conclusion, we have identified Cry1C domain III amino acids in larger blocks (blocks D and E) as well as in small blocks or groups (block C and the insertion in block D) and a single amino acid (Trp-476), which have a strong positive effect on activity against S. exigua. Some blocks and residue groups simultaneously have a negative effect on activity against M. sexta (block D), whereas block E in G27 is required for high activity against both insects. One might speculate that at least some of the identified specificity-determining residues are involved in binding, which has been shown to be a function of another domain III, i.e., that of Cry1Ac (3, 4). It is interesting to speculate about the location of the S. exigua-specific residues in the three-dimensional structure of domain III of Cry1C. Although no X-ray-diffraction-derived structure is available for this toxin, sequence alignment with Cry1Aa allows the prediction of its structure, to some extent. This places Trp-476 at the C-terminal end of beta-strand 13b and, in close proximity, places Phe-498 and Glu-500 between beta strands 15 and 16. The Cry1C-specific insertion in block D which is involved in specificity for S. exigua may lead to an increase in the length of the loop between beta strands 18 and 19, although structural changes in this area may be more extensive. Interestingly, all of these S. exigua specificity-determining residues appear to be located on the slightly concave outer beta sheet of domain III (8). Residues in the same sheet were implicated in the GalNAc-binding capacity of domain III of Cry1Ac (3). Our preliminary results have shown that domain III of Cry1C in G27 does increase the affinity of binding to S. exigua membranes, compared with Cry1E (results not shown). Future detailed studies of binding properties of hybrids and mutants may further clarify the functions of the identified residues at the molecular level.