Article | Open | Published:

Using phage display technology to obtain Crybodies active against non-target insects

Scientific Reportsvolume 7, Article number: 14922 (2017) | Download Citation


The insecticidal Cry toxins produced by Bacillus thuringiensis (Bt) are increasingly important in the biological control of insect pests and vectors of human disease. Markets for Bt products and transgenic plants expressing their toxins are driven by their specificity, safety and the move away from chemical control agents. However, the high specificity of Cry toxins can also prove to be a limitation when there is no known Cry toxin active against a particular target. Novel activities can be discovered by screening natural Bt isolates or through modifications of the Cry proteins. Here we demonstrate the use of λ-phage displaying Cry1Aa13 toxin variants modified in domain II loop 2 (Crybodies) to select retargeted toxins. Through biopanning using gut tissue from larvae of the non-target insect Aedes aegypti, we isolated a number of phage for further testing. Two of the overexpressed Cry toxin variants showed significant activity against A. aegypti larvae while another induced mortality at the pupal stage. We present the first report of the use of phage display to identify novel activities toward insects from distant taxonomic Orders and establish this technology based on the use of Crybodies as a powerful tool for developing tailor-made insecticides against new target insects.


Bacillus thuringiensis (Bt), is characterized by its ability to produce proteins with insecticidal activity. Toxins produced by Bt can form crystalline inclusions during sporulation1 known as δ endotoxins (toxins Cry and Cyt)2,3 or can secrete toxins during vegetative growth (VIP and Sip toxins)4,5. Among these proteins, Cry toxins are the best characterized and exert their effect on the host by causing lysis of the midgut epithelial cells, which leads to gut paralysis, cessation of insect feeding, loss of membrane integrity, release of the cellular contents and eventual death3,6,7,8.

Currently, 302 holotypes of Cry toxins are described ( as a result of a huge international effort to isolate and characterize novel Bt strains. Each of these toxins is reported to be active against a limited number of targets (mostly insects but also nematodes, molluscs and, in a few cases, human cancer cells9) and this specificity is one of the most remarkable characteristics of Cry toxins. This specificity is a consequence of their complex and sophisticated mode of action, which, although not completely understood, is well-accepted to require the presence of receptors on the epithelial membrane of the target gut cells2,10,11,12.

Although Cry toxins may belong to a number of distinct structural families9,13, the major group within Cry proteins is the so-called 3-domain toxins. The three-dimensional structure of these toxins shows a remarkable similarity14. Domain II, formed by three antiparallel beta-sheets arranged in a beta-prism form15, has been associated with toxin specificity12,16. Domain II is one of the most variable of the three domains, especially in three apical loops in this part of the molecule. These loops differ not only in sequence and conformation, but also in length. This variability, together with domain II’s similarity to the complementarity-determining region of immunoglobulins15 led researchers to associate this domain with the role of receptor binding and hence with specificity. Exchange of the domain II regions among Cry toxins or modification of the sequences of the loops seems to be enough to alter the toxicity profile toward an insect17,18,19.

As mentioned above, the mechanism of action of 3-domain Cry toxins is not fully elucidated and currently three different models have been postulated for their function14,20. All proposed mechanisms involve binding to receptors present on the surface of target cells such as cadherin-like proteins, the amino peptidase N (APN), alkaline phosphatase (ALP), glycolipids21, the glycoconjugate BTR-27022, the P252 protein23 and the recently described transporter ABCC224,25. Although binding to the receptor is not the only requirement for toxicity, it is an essential requirement and it seems that natural evolution of Cry toxins has adapted them to recognize different proteins as receptors to be used in their mechanism of action.

The toxic effect of Cry proteins has been commercially exploited for more than 70 years to control insect populations that have an impact on forestry, agriculture and health2,26,27 and toxin genes have also been used in plant transgenesis28,29,30,31,32,33,34. The specificity of the Bt Cry toxins is a huge advantage; each toxin only affects a restricted number of insects species, leaving other, non-target organisms unaffected35. Unfortunately, in many cases, there is no known toxin active against the insect of interest that can be used for its control. The traditional approach to search for new activities against a specific insect has been the isolation of hundreds of new Bt strains, and bioassaying their activity against the target insect to probe toxicity36,37,38. This represents an extensive work programme and frequently no activity at all is found against the insect of interest.

An alternative strategy to obtain Cry toxins with novel specificities is the modification of existing toxins through molecular techniques in order to increase the activity toward certain insects39,40,41 and rationally redesign activity from one insect to another17,19,41,42.

Although these studies demonstrate that manipulation of toxins can render novel activities, this process is labour intensive and not very effective as many of the mutants rationally designed in this way may lack structural stability and function. Here, we demonstrate a high throughput screening methodology for the potential isolation of new toxins with novel activities against insect targets. The methodology is based on phage display technology, a potent molecular tool for in vitro selection of proteins with a specific binding profile. In this technique, a protein or a mutant library is fused to a phage protein so that it is displayed on the phage surface, ready to interact freely with other proteins. Several groups have described phage display of Bt toxins43,44,45,46,47,48,49 but successful isolation of variants active toward taxonomically distant non-target insects has never been reported.

Previously, we reported the display of a Cry1Ac toxin on the surface of a lambda phage that successfully interacted with its natural receptor50. This was followed by a report of the construction of several Cry1Aa mutant libraries by replacing loops 1, 2 and 3 from the Domain II of the toxin with the hypervariable region contained at the complementary determinant region 3 (CDR-H3) of a human antibody library51. The resulting cc2 library was displayed on the surface of a phage. The system combined the structure and scaffolding of the Cry toxins with the potential binding capability and specificity of a library of human antibodies (Crybodies). Here we present the screening of this library for the selection of toxins with novel binding profiles towards Aedes aegypti, an important vector of several flavivirus disease agents such as Dengue virus, Yellow fever virus and the currently emerging Zika virus. The selected Cry1Aa variant toxins were cloned in a heterologous expression system, overexpressed and bioassayed against A. aegypti, the non-canonical target insect chosen as a model. Cry toxin variants selected in vitro using the phage display showed significant toxicity to A. aegypti. Thus, we demonstrate the proof of principle that in vitro evolution of a Cry toxin coupled with in vitro selection using phage display is a powerful tool for selecting novel toxins redirected against new-target insects across insect orders.


In vitro selection of phage with affinity to A. aegypti guts

The λcc2 phage display library, containing a diversity of 2.2 × 107 unique clones of Cry toxins with the domain II loop 2 of the lepidopteran active Cry1Aa13 toxin replaced by CDR-H3 sequences obtained from a human antibody library52 was used in this study. A series of biopanning experiments was performed in order to select and enrich those phage from the library with enhanced affinity to targets present in the A. aegypti guts compared to the wild-type toxin. The pool of the phage obtained was considered as the sub-library 1. These phage were amplified and used for a new round of selection using fresh A. aegypti guts and this procedure was used in two further steps, each using the enriched phage from the previous round. The sub-library 4, resulting from the fourth round of selection, was used for further analysis. The number of phage recovered in each round of selection is shown in supplementary Table S1. As no significant increase of phage number in each round of selection was observed, further analysis was carried out in order to gain insight into any specific selection that may have occurred. To do this, the length of the loop 2 sequences present in a group of 15 randomly selected phage from sub-library 4 obtained by PCR with primers A2f and A2r, Supplementary Table 2S) was compared to the loop length from a group of 15 randomly selected clones from the original library λcc2 (Supplementary Fig. S1). The loop 2 size of phage in sub-library 4 seemed to be less variable than the original library, probably indicating that a selective process had successfully occurred.

Sequence analysis of loop 2 variants from sub-library 4

A2f-A2r PCR fragments obtained from a further 16 phage from sub-library 4 (different from the pool analysed in Fig. S1 and named from A1 to A16) were cloned in the pGEM-T vector and the resulting plasmids were sequenced and the amino acid sequence present at the loop 2 of the selected Cry mutants was determined (Fig. 1). The sequence analysis revealed that five of the phage (A3, A4, A5, A14 and A15) might display a truncated version of Cry toxins as a stop codon was found in their sequences (shown as “*”). The existence of these truncated proteins was probably due to errors in the assembly of the library or DNA amplification that rendered a shift in the reading frame as previously reported51. Phage A3 and A4 showed the same sequence and the stop codon was found at the hypervariable region. Phage A14 and A15 show complex insertion patterns in which Gly-Ala-Arg- sequences introduced during library production, occur in the middle of the inserts rather than at the end. Phage A6 showed the same sequence as A7, corresponding to the wild type loop of Cry1Aa13. The rest of the phage A1, A2, A8, A9, A10, A11, A12, A13 and A16 (Fig. 1, black arrows) showed a sequence completely different to each other and different from the wild type with no stop codons introduced.

Figure 1
Figure 1

Alignment of the loop 2 sequences present in the in vitro selected Cry phage and the wild type Cry1Aa13 toxin. Arrowheads indicate novel loop variants in sequences that do not contain premature stop codons. Bold letters represent the predicted hypervariable region present in each mutant after loop 2 replacement.

The number of aa present at the hypervariable region in these 9 selected phage ranged from 13 to 19 aa. Compared to the original λcc2 library (7 to 25 aa reported by51), the variability in sizes decreased significantly, probably due to the selection process.

The 9 loop-variant mutants that did not contain stop codons (A1, A2, A8, A9, A10, A11, A12, A13 and A16), began with either Asp, Arg or Gly (after the Gly-Ala-Arg- sequence introduced during library production). Seven of the 9 ended with Cys (the other two ending with Trp), Tyr was the penultimate residue in 5 of these variants, preceded by Asp in 8 of the variants, preceded by an aromatic Phe (5 variants) or Tyr (2 variants). Whether this reflects a positive selection for these residues during biopanning or a feature of the original library is not known at present.

Reconstituting Crybodies from the selected phage

Phage in the library contain translational fusions between the active part of Cry1Aa13 toxin (containing loop variants) and the phage coat protein gpD to allow toxin to be displayed on the phage surface50,51. In order to assay the toxin variants selected with A. aegypti guts, loop variants were recloned into the pCP10-Δloop2 plasmid (following the strategy described in the Materials and Methods section and summarized in Supplementary Fig. S2). The resulting constructs were PCR screened with A2f and TD2 primers (Supplementary Table S2) in order to select those clones with the correct orientation of the fragment. Plasmids pCP10-A1, pCP10-A2, pCP10-A5, pCP10-A8, pCP10-A9, pCP10-A10, pCP10-A11, pCP10-A12, pCP10-A13 and pCP10-A16 (named collectively as pCP10-Ai) were produced and contained mutant cry1Aa13 protoxin genes from phage A1, A2, A5, A8, A9, A10, A11, A12, A13 and A16 respectively.

Expression profile of the cloned Crybodies

E. coli clones containing pCP10-Ai plasmids were grown in liquid media with IPTG and the accumulation of the mutant protoxins was analysed by SDS-PAGE (Fig. 2). In the pellet fraction of an E. coli pCP10 culture, a protein of 130 kDa corresponding the wild type Cry1Aa13 protoxin was observed (Fig. 2, lane 1). No such protein was detected in the clone bearing pCP10-Δloop2, as it contains a deleted version of the cry1Aa13 gene (Fig. 2, lane 2) and hence a truncated Cry protein is produced (that is probably unstable and does not accumulate). Clones pCP10-A8, -A10, -A11 and -A12 showed a strong band of 130 kDa in size corresponding to Crybodies Cry1Aa13-A8, -A10, -A11, and -A12 respectively. Clone pCP10-A1 (Fig. 2, lane 3) produced a minor band at this size. The rest of the clones did not render any visible protoxin.

Figure 2
Figure 2

Production of variant protoxins. Coomassie blue stained 12% acrylamide gel showing the insoluble fractions of IPTG induced cultures of E. coli with pCP10 (lane wt) as positive control, and pCP10-Δ loop2 (lane Δ) as negative control. Clones bearing plasmids pCP10-A1 (lane A1), pCP10-A2 (lane A2), pCP10-A5 (lane A5), pCP10-A13 (lane A13), pCP10-A16 (lane A16), and pCP10-A9 (lane A9), did not show over-expression of the Cry protoxin. Clones bearing plasmids pCP10-A8 (lane A8), pCP10-A10 (lane A10), pCP10-A11 (lane A11) and pCP10-A12 (lane A12) showed a strong Cry protoxin production similar to the wild type Cry protoxin.

Those clones with no visible accumulation of the full Cry protoxin were analysed by western blot (Supplementary Fig. S3) and expression of the full protoxin was only observed in the clone pCP10-A1, although at a very low level, consistent with the results from Coomassie staining. In the rest of the clones, smaller proteins were detected by immunoanalysis, indicating that protoxin was expressed but the protein was not stable enough to render the full Cry toxin, probably because of the loop replacement.

Solubilisation and trypsin activation of Cry1Aa13-A8, -A10, -A11, and -A12 Crybodies

The four Crybodies showing a good level of expression and stability (Cry1Aa13-A8, -A10, -A11 and -A12) were solubilized and activated with trypsin under standard in vitro conditions51. As shown in Fig. 3, all four Cry1Aa13-Ai Crybodies were successfully solubilized under the same conditions as the Cry1Aa13 wild type (Fig. 3, lane 1) and rendered a 130 kDa protein. Solubilized Cry1Aa13-A8, -A10, and -A12 Crybodies treated with trypsin produced a band around 65 kDa corresponding to the theoretical molecular weight of the active Cry toxin52. However, Cry1Aa13-A11 showed a faint band around 65 kDa and more intense protein fragments of 37 and 28 kDa than the other mutants and the wild type toxin (probably as a result of digestion of the 65 kDa protein), indicating a lower stability of this mutant compared to the rest. As the stability of Cry1Aa13-A11 was compromised, this variant was not used in further experiments.

Figure 3
Figure 3

Solubilization and trypsin treatment of protoxin variants. Coomassie blue stained 12% acrylamide gel showing solubilized protein treated (+) and untreated (−) with trypsin. Cry1Aa13 wild type was used as positive control for solubilisation (lane wt −) and trypsin activation (lane wt +). Lanes A8, A10, A11 and A12 show the solubilized Cry1Aa13-A8, Cry1Aa13-A10, Cry1Aa13-A11 and Cry1Aa13-A12 respectively. Trypsin-treated toxins exhibited a protein of 65 kDa corresponding to the core of the toxin or activated toxin. Cry1Aa13-A11 showed the same band but less intense than the rest of the mutants. Arrows A and B indicate the molecular weight of the approximately 37 kDa and 28 kDa fragments observed during trypsin digestion.

Evaluating the toxicity of selected Crybodies against A. aegypti

The susceptibility of A. aegypti larvae to mutant toxins Cry1Aa13-A8, -A10 and -A12 was determined. Figure 4 shows the larval mortality obtained in the bioassays with each toxin. The mixture of Cry4Aa, Cry4Ba, Cry11Aa and Cyt1Aa obtained from Bti showed 100% mortality 24 h after the beginning of the bioassay as expected. The lepidopteran-active Cry1Aa13 toxin did not produce any mortality as A. aegypti is not susceptible to this protein. However, two mutant toxins, Cry1Aa13-A8 and -A12, showed a significant percentage of mortality after 120 h (around 90% mortality).

Figure 4
Figure 4

Percentage mortality of A. aegypti larvae on treatment with soluble toxins. Results obtained with Cry1Aa13-wt toxin (dark blue), Cry1Aa13-A8 (red), Cry1Aa13-A10 (green) and Cry1Aa13-A12 (purple) and Bti (cyan) treatments.

The statistical analysis in a pair-wise comparison revealed significant differences between the toxicity of the proteins assayed. Each toxin was included in one of the 3 significantly different groups; Cry1Aa13-wt and Cry1Aa13-A10 were included in group A, Cry1Aa13-A8 and -A12 in group B, and Bti was included in group C. The Kaplan-Meier estimator (α = 0.05) was determined for Cry1Aa13-A8 and -A12 rendering values of 82 ± 0.87 h and 64.4 ± 0.88 h respectively. These results demonstrate that the two in vitro selected mutants Cry1Aa13-A8 and -A12 showed toxicity against A. aegypti, presumably mediated through affinity towards proteins present in the insect gut. LC50 values of these two variants were calculated by performing bioassays against 4th instar larvae, using toxin concentrations ranging from 20 μg/ml to 1.25 μg/ml. Using Probit analysis it was estimated that LC50 values for Cry1Aa13-A8 and Cry1Aa13-A12 were 10.4 μg/ml and 9.1 μg/ml respectively (Table 1). For comparative purposes, Table 1 also details the LC50 of the naturally mosquitocidal toxins Cry4Aa, Cry4Ba, Cry11Aa and CytAa toxins reported in the bibliography19,53,54,55,56,57,58 and reviewed by Otieno-Ayayo et al.59. Taking the highest LC50 values for each natural toxin we can observe that Cry1Aa13-A8 showed a LC50 approximately 4.2 times higher than Cry4Aa, 10 times higher than Cry4Ba and Cyt1Aa and 35 times higher than Cry11Aa. In the case of Cry1Aa13-A12, the LC50 was 3.75, 10, 31 and 9 times higher than Cry4Aa, Cry4Ba, Cry11Aa and Cyt1Aa respectively.

Table 1 LC50 values of Cry1Aa13 variants obtained in A. aegypti bioassays and comparison with the reported in the literature for individual toxins from Bti.

Evaluating the toxicity of selected Crybodies against Bombyx mori

In order to determine how loop 2 replacement influenced the activity of the selected Crybodies against a canonical insect target of the wild type, parental toxin, variants Cry1Aa13-A8, Cry1Aa13-A10 and Cry1Aa13-A12 were bioassayed against B. mori together with the wild type toxin Cry1Aa13 and a mixture of non-active toxins (Cry4, Cry11 and Cyt) from Bti. The results showed (Table 2) that after 96 h the Cry1Aa13-A10 variant (which is not active against A. aegypti larvae (Fig. 4)) showed 100% mortality in B. mori bioassays, as did the Cry1Aa13 wild-type toxin. In contrast, mortality produced by Cry1Aa13-A8 and Cry1Aa13-A12 variants was lower (45 ± 7% and 50 ± 14% respectively) than the wild-type, suggesting that loop 2 replacement negatively influenced in the activity against its natural lepidopteran target. As expected, low toxicity was recorded (20%) when Bti toxins were assayed against B. mori.

Table 2 B. mori mortality recorded after 96 h in contaminated mulberry leaf-bioassays.

Unexpected activities observed in the selected Crybodies

In addition to these results, another interesting observation was made in the bioassay performed with the Cry1Aa13-A10 mutant. Although the mortality of the larvae was 0% at the end of the bioassay (Fig. 4) as described above, and all larvae became pupae, it was observed that all of these insects died within 24 h after pupation. This was an unexpected activity that is worth noting as, to our knowledge, pupal mortality as a result of exposure to Bt 3-domain Cry toxins has not been reported previously. This result also demonstrates that completely new activities can be obtained by using the phage display methodology that we have employed. Further investigations of this novel activity will be published separately.


It has been demonstrated extensively that the specificity of the 3-domain Cry toxins can be mediated by domain II of the protein, since this region has a role in toxin binding to natural receptors60,61,62,63,64. Mutations within the loops of domain II can induce an increase of activity against target insects39,40,41, and deletions, substitutions or replacements can even redirect toxicity toward related species19,65. In addition, in a program of rational redesign, the lepidopteran-active Cry1Aa was rendered toxic to the dipteran Culex pipiens by specific replacements in loops 1 and 2 of domain II42. As these studies involved rational redesign, similar experiments are only feasible when assessing the effects of chimeric proteins made from toxins with known activities and not for the development of toxicity against species that are not targets of currently available toxins. These reports demonstrate that an important determinant of 3-domain Cry toxin specificity resides in the domain II loops and open the possibility of creating new toxin activities by manipulation of these loops. Construction of high-diversity Cry toxin mutant libraries through combinatorial techniques is relatively straightforward, the challenge in this artificial evolution is selecting and identifying those mutants from the library that may show interesting activity against a specific target. Individually bioassaying each artificially-generated mutant from a combinatorial library to determine activity is impossible from a practical point of view and phage display offers a potential method for bio-panning such libraries to enrich for specific binding activities against any selected insect target. Of course, not all the variants selected would be likely to show activity, but as binding to insect guts is a requirement for Cry toxicity, binding was considered to be an appropriate filter to allow selection of a subset of variants for detailed analysis.

The display of Cry toxins on the surface of phage has been attempted using several systems, which have sometimes encountered limitations. In the first attempt to display a Cry toxin on the surface of a phage (Cry1Aa on M13 phage), the toxin was not properly displayed, resulting in deletions of the fusion protein45. Later, a complete Cry1Ac was displayed on the surface of the M13 phage, although the toxin did not show binding to its natural APN receptor in vitro, suggesting structural constraints of the displayed toxin46. Display systems based on λ and T7 phage proved to be more suitable alternatives for displaying Cry toxins, as they assemble in the E. coli cytoplasm and are released by cell lysis instead of secreting their components through the bacterial membrane as in M1366. Cry1Ac was fused to the capsid protein gpD of λ phage and successfully displayed on the phage surface50 with the Cry1Ac1::gpD fusion showing similar toxicity to the wild type and interacting with its natural receptors. Pacheco et al.44 reported the fusion of Cry1Ac toxin to the capsid protein 10B of T7 phage, successfully displaying a M. sexta active toxin able to bind M. sexta BBMV44. This T7 display system was the basis for the construction of a mutant library containing 5 × 105 variants bearing a 7-8 aa-long random sequence at loop 2 of the Cry1Aa1 toxin47. Screening of this mutant library against Bombyx mori, a natural target of the parental Cry1Aa protein, identified a variant four times more potent than the wild type Cry1Aa1 but screening of the library against other insects was not reported. Recently, Craveiro et al. 2010 reported the selection of 4 variants of the lepidopteran-active Cry1Ia12 obtained by DNA shuffling and phage display technology. The Cry1Ia12 variants showed between 2 and 3 times more mortality than the wild type Cry1Ia12 toxin against the non-target lepidopteran sugarcane giant borer Telchin licus licus. This report showed that phage display can be used for selecting novel activities against new, taxonomically-related targets, but the selection of variants active against more distinct orders of insects was not reported. Thus, nearly two decades after the first publication of Cry toxin phage display, the production through this technology of toxins retargeted toward taxonomically distant targets has not been described.

The library used in our work, exploited our previous λ gpD-fusion phage display system50 to express loop 2 domain II variants of Cry1Aa1351. Compared to the display system developed by Ishikawa et al.47, this λcc2 library contained a higher number of mutants (2.2 × 107 vs 5 × 105), the variants contained loops with a wider range of mutant sequences (7–25 aa vs 7–8 aa) and our library was constructed by replacing loop 2 by an antibody CDR-H3 library, which may prove an inherently better source of binding motifs than artificially-generated random sequences. All these factors may underlie the ability of the system developed in the present work to provide the breakthrough to yield variants with completely novel activities.

After 4 cycles of biopanning with A. aegypti guts, the derived pool of phage showed a hypervariable region in loop 2 ranging from 13 to 19 aa as opposed to the 7–25 aa present in the original library51, a reduction in diversity consistent with a selection/enrichment process. The analysis of the sequences of the selected phage showed that some displayed a truncated version of the mutant toxin, probably due to assembly problems in the construction of the λcc2 library. The truncated Cry toxins are not expressed beyond domain II loop 2 and it is not clear whether their presence in the enriched phage pool is due to unexpected specific binding or whether they are part of the inevitable background binding. The presence of such phage in the library and their retention represents a decrease in the recovery efficiency of the system that we are currently working to circumvent.

Analysis of full-length Cry protoxins, reconstituted from selected phage showed that not all the Cry mutants retained during enrichment were suitable for heterologous over-expression. Attempts to express mutant toxins such as Cry1Aa13-A1, -A2, -A5, -A9, -A13, and -A16 resulted in no significant accumulation of the protoxin. As all toxins were expressed from the same promoter (P tac promoter), the failure to accumulate these toxins is most likely due to the low stability of the mutants. This was confirmed in western blots, which detected small fragments of Cry toxins, implying that these mutant proteins were expressed but were subject to degradation. The lack of stability can only be attributed to the change in the sequence of the loop 2 as the rest of the molecule was unchanged. Low stability and increased proteolytic sensitivity observed for some mutants may arise from conformational changes elsewhere in the protein, induced in the molecule by the novel loop 2 variants (Supplementary Fig. S4). Alternatively, the introduction of random sequences may provide direct proteolytic target sites within loop 2. Mutants Cry1Aa13-A9, -A13 and -A16, which are particularly unstable and the more stable Cry1Aa13-A11 mutant may illustrate proteolytic sensitivity at the mutated loop as they contain Lys and or Arg residues in this region that would introduce potential susceptibility to cleavage by trypsin-like enzymes. Cry1Aa13-A11 is stable enough to be produced in the pellet fraction of recombinant E. coli but can be subsequently degraded by trypsin digestion, possibly at the Arg residue in its loop 2. In contrast, while the stable mutant Cry1Aa13-A10 also has an Arg residue in the loop, it appears to be activated normally by trypsin. However, the Arg in this variant is followed by a Pro residue and endoproteolytic cleavage N-terminal to a Pro residue is very unusual in nature67. Three mutants, Cry1Aa13-A8, -A10, and -A12 showed high levels of expression and normal trypsin-activation profiles, so they were selected for bioassays against A. aegypti. Cry1Aa13-A10 showed an unexpected activity, with no apparent effect on A. aegypti larvae but exhibiting delayed mortality at the pupal stage. Blocking the development of adults through pupal mortality may represent an important new means to disrupt disease transmission. The activity of this mutant needs to be characterized further but this variant demonstrates the capacity of the strategy developed here to expand selection of toxins beyond the larval stage of the insect, and widen the possibility of targeting other stages; a feature that could represent an advantage in the control of many natural insect populations.

Variants Cry1Aa13-A8 and -A12 were able to produce a high mortality in larval populations (90%). Although the amount of toxin used in the bioassays was quite high (20 µg/ml) the wild type Cry1Aa13 showed no toxicity at this concentration, (LC50 for the Cry1Aa13 is higher than 100 µg/ml according our experience). The mutant activity is clearly lower than the toxicity of Bti crystals used as the positive control, but these contain at least three different Cry toxins (Cry4Aa, Cry4Ba, and Cry11Aa) and one Cyt protein (Cyt1Aa) with possible production of Cyt2Ba and Cry1Ca and Cry10Aa68,69 active against A. aegypti with different toxicity levels70,71. The activity of each individual toxin produced by Bti against A. aegypti has been determined by several research groups and compiled by Otieno-Ayayo et al.59. These authors found a high variability of the LC50 reported in the literature for Cry4Aa (0.56–2.39 μg/ml), Cry4Ba (0.06–0.94 μg/ml), Cry11Aa (0.01–0.28 μg/ml) and Cyt1Aa (0.8–1.0 μg/ml) toxins against A. aegypti as a consequence of factors such as where the toxins are produced, how the toxins are purified and processed or the conditions of the bioassays. Although comparison of the LC50 obtained in different bioassays is not completely accurate, it provides an opportunity to demonstrate that the activity of the Cry1Aa13 variants obtained by phage display may be within an order of magnitude of the toxicity of toxins naturally active against the model insect. If we compare the highest reported LC50 value for each individual Bti toxin with the Cry1Aa13-A12 variant for example, we can roughly estimate that toxicity of the latter is 3.79–32 times lower than individual mosquitocidal toxins. Bti toxins have evolved to target nematoceran Diptera for millions of years, allowing evolutionary optimisation of the toxins for their target. Thus, we have demonstrated significant activity for Cry1Aa13-A8 and Cry1Aa13-A12 against A. aegypti (within one order of magnitude of the natural mosquitocidal toxins). This is a good level of toxicity for proof of principle for our system, considering that variations were introduced without rational design, in only a single loop and the derived toxins have not undergone further optimisation. The level of toxicity shown by variants Cry1Aa13-A8 and Cry1Aa13-A12 was similar to the activity gained in C. pipiens in the rational design previously reported by Liu and Dean, (2006). The mutant 1AaMosq, containing a Loop 1 substitution (311RG312 by YQDL) and a Loop 2 deletion (365LYRRIIL371) plus a substitution (376NNQ378 was replaced by a G) showed a LC50 value for 2nd instar larvae of 45.73 μg/ml while LC50 for A. aegypti 4th instar larvae for Cry1Aa13-A8 and Cry1Aa13-A12 was 10.4 and 9.1 μg/ml respectively. In the report of Liu and Dean42, loop 2 in the 1AaMosq was shortened to mimic the mosquito-active Cry4B loop 2. In contrast, our selected variants Cry1Aa13-A8 and Cry1Aa13-A12 showed a loop length similar to the Cry1Aa13 parental toxin (13 and 18 aa respectively) but very different in sequence. In addition, the loop2 sequences present in mosquito-active Cry1Aa13 variants have no significant similarity to loop 2 regions from other 3-Domain Cry toxins, including the mosquitocidal toxins Cry4Aa, Cry4Ba and Cry11.

A correlation is observed in our results between the development of toxicity of Cry1Aa13-A8 and Cry1Aa13-A12 toward A. aegypti concomitant with decreases in toxicity toward B. mori. It is interesting to note that Cry1Aa13-A10, which did not show any activity against A. aegypti larvae, retained its activity against B. mori. It will be interesting to investigate the molecular targets of these new variant toxins to determine whether they are associating with known receptor types (eg cadherins, amino peptidases, alkaline phosphatases) or new classes of binding ligands. Binding studies of the three variants with A. aegypti and B. mori BBMVs are in progress.

It should be noted that the two larvicidal and one pupicidal variants described were identified from a detailed analysis initiated from only 16 randomly selected phage from our biopanning, further refined to just 9 candidates after initial sequence analysis. This represents an excellent level of efficiency and clearly incentivises the screening of further variants. In the longer term, efficiency may be enhanced by improving library quality (with fewer truncated variants) and may be aided by the introduction of an increased number of biopanning rounds prior to clone testing. Increasing the biopanning stringency between rounds would be desirable but there are problems with changing stringency when there is little or no control over the number or concentration of target molecules72 (as in our screening with no defined target in tissue samples and the ability of phage to select the targets to which they might bind). Further development of higher toxicity variants may also be aided by the separate screening of libraries with diversity in variable loops 8, 1 and 3 with the subsequent combinatorial mixing (by simple combination or, for example, through DNA shuffling techniques) of mutants showing activity.

Very recently, a report on Cry toxin evolution using PACE (phage-assisted continuous evolution) has been described73. In this paper authors designed a complex and outstanding in vitro evolution system that enables novel Cry variants to be obtained that are active against resistant insects. This work, like ours, supports the idea that in vitro evolution is the strongest tool for obtaining tailor-made insecticides that the market demands. However, the evolved toxins obtained by PACE showed activity against closely-related insects species and maintained a similar insecticidal spectrum as the parental Bt toxin. As a result, there may be only relatively minor changes in receptor molecules to which toxins must be adapted. In contrast, we have demonstrated that Crybody phage display can go even further to obtain novel variant with activities against distinct orders of insects. The work presented here has provided the proof of principle that phage display, our library and the selection strategy developed are suitable molecular approaches for selecting toxins with activity redirected to new target insects even across taxonomic orders. This technology can now be exploited to tackle new pests for which no Bt toxins are currently available and also as a strategy to overcome resistance that has arisen in the field.

Materials and Methods

Bacterial and phage strains

Escherichia coli strain DH5α was used for cloning and expression studies. E. coli strain Y1088 was used for phage amplification. E. coli strain BLR containing the pCP10 plasmid was used for obtaining Cry1Aa13 toxin52. B. thuringiensis var israelensis 4Q5 was used to obtain Cry4Aa, Cry4Ba, Cry11Aa and Cyt1Aa toxins. All phage used were derived from the λEMBL3 vector (Promega)74. The λEMBL3-pTI11 contained the plasmid pTI11 as test insert plasmid supplied in the Promega kit and used for controlling packaging efficiency. λCP2 phage displayed the wild type Cry1Aa13 toxin and the λcc2 mutant library contained approximately 107 unique variants of Cry toxins (kindly provided by Prof. D. J. Ellar from the University of Cambridge)52.

Phage amplification and concentration

E. coli Y1088 was grown overnight at 37 °C (240 rpm) in 5 ml of NZCYM medium (5 g/l NaCl, 2 g/l MgSO4·7H2O, 5 g/l yeast extract, 10 g/l casein enzymatic hydrolysate, 1 g/l casein acid hydrolysate) supplemented with ampicillin (100 μg/ml). After determining the A600 nm of the culture, an appropriate volume, containing 1010 cells (assuming that 1 U of A600 nm corresponds to 109 cells/ml) was centrifuged and the pellet resuspended in 300 μl of SM (100 mM NaCl, 10 mM MgSO4, 50 mM Tris-HCl, 0.01% gelatin, pH 7.5), and mixed with a phage suspension containing 107–108 pfu. After 20 min incubation at 37 °C the mixture was added to 50 ml of fresh NZCYM and incubated (37 °C, 150 rpm) in a rotary shaker until lysis of the culture was completed (A600 nm 0.1–0.2). Then, 1 ml of chloroform was added, incubated for a further 20 min and treated with pancreatic DNase (Sigma) and RNase (1 μg/ml, Sigma) at RT for 30 min. Solid NaCl was added to a final concentration of 1 M and incubated on ice for 1 h before being centrifuged (12000 × g) for 20 min. The supernatant was mixed with PEG 8000 (10% (w/v) final concentration), incubated for 1 h on ice and centrifuged at 12000 × g for 20 min and the pellet resuspended in 800 μl of SM. An equal volume of chloroform was added and the mixture vortexed for 30 seconds. The upper aqueous phase containing the phage suspension was recovered and stored at 4 °C until use.

Phage quantification and phage propagation on plates

E. coli Y1088 was cultured in 50 ml of LB (plus maltose 0.2% (w/v), 10 mM MgSO4 and ampicillin 100 μg/ml) at 37 °C and 240 rpm until a A600 nm of 0.5–0.6. Culture was pelleted, resuspended in 25 ml of 10 mM MgSO4 and kept at 4 °C and used within a month when needed. For phage titration, 200 μl of this E. coli Y1088 suspension was mixed with 200 μl of a phage suspension, incubated for 20 min at 37 °C, mixed with 3–4 ml of molten top agarose (50 °C) (per liter: 10 g tryptone, 5 g yeast extract, 10 g NaCl, 6 g agarose) and poured onto a LB agar plate. Plates were incubated at 37 °C overnight for plaque counting. Plaques were propagated and amplified in solid media with the use of a sterile toothpick and a LB agar plate containing top agarose with E. coli Y1088.

Aedes aegypti rearing and management

A. aegypti eggs were incubated in a plastic container with dechlorinated water and ground artificial diet (commercial dry cat food) at 25 °C ± 2 °C, 70% humidity and a photoperiod (12 h light/12 h darkness). Eggs hatched within two days. Fourth instar larvae were used for biopanning experiments and bioassays.

In vitro biopanning

Biopanning was performed with ten guts extracted from Virkon®-treated 4th instar A. aegypti larvae. Guts were obtained by pulling from the neck and the anal segment of the larva with 2 ultra fine point tweezers. Guts were placed in cold PBS for peritrophic membrane removal and then placed in 200 µl of SM buffer supplemented with protease inhibitors (complete mini protease inhibitor cocktail, Roche®) in an Eppendorf tube. Insect tissue was disrupted with a micropestle (Sigma) and mixed with the appropriate volume of the phage suspension of λEMBL3-pTI11, λCP2 or λcc2 to give a final concentration of 109 PFU/ml. Phage and insect gut homogenates were incubated for 30 min at RT and then centrifuged at 16000 × g for 10 min. Pellets were resuspended in 500 μl of washing buffer (1.5 M NaCl, 0.1% Tween and protease inhibitor) and centrifuged again under the same conditions. Washing was repeated 5 more times and the final pellet was resuspended in 200 μl of SM buffer. Phage attached to material present in the pellet fractions were recovered by incubation (20 min at 37 °C) with 200 μl of E. coli Y1088 prepared as indicated above. The mixture was placed in 50 ml of NZCYM medium for phage amplification. When the number of the selected phage (sub-library) reached more than 1010 PFU/ml, a new round of biopanning was started to enrich phage able to bind to the gut (and deplete the relative proportion of phage present due to non-specific binding). Four rounds of selection were performed in total and phage present in the final sub-library were kept for further experiments and analysis.

Loop 2 cloning and sequence determination

Loop 2 mutants present in selected phage were PCR amplified using A2f-A2r primers (Supplementary Table S2), purified and ligated into pGEM®-T vector (Promega) following the manufacturer’s instructions. Plasmids were sequenced using primer M13f.

Constructing pCP10-Δloop2 plasmid

To facilitate recovery and overexpression of protoxin forms of cry gene variants identified through phage screening, we constructed the plasmid pCP10-Δloop2 containing a 500 bp deletion in the cry1Aa13 gene that included the region encoding loop 2 of the protein. This provided a toxin scaffold ready for the insertion of any of the loop 2 mutants from the selected phage and avoided the relatively inefficient amplification and cloning of ~ 2 kb encoding the entire toxin region. SalI restriction sites were introduced to facilitate the cloning strategy as the introduction of this restriction site did not change the aa composition of the protein. The pCP10-Δloop2 plasmid (Fig. S1) was constructed by PCR (95 °C/5 min, 30 cycles at 95 °C/45 sec, 60 °C/45 sec, 72 °C/7 min) using pCP10 plasmid as template and the divergent primers TD12 and TD13 (Table S2). The PCR amplicon was purified (Qiagen), digested with SalI (NEB), heated for enzyme inactivation and self-ligated with T4 DNA ligase (Promega). The circularized plasmid produced, pCP10-Δloop2, was then ready to accept any mutant loop 2 in the newly introduced SalI site.

Construction of new variant toxin plasmids

The region that includes the loop 2 of domain II of the different selected phage was amplified using the primers TD10 and TD11 (Table S2) by PCR (1 cycle at 95 °C/5 min, 30 cycles at 95 °C/30 sec, 63 °C/30 sec, 72 °C/30 sec and a final extension at 72 °C/5 min). PCR fragments were purified, SalI digested and ligated to the SalI-dephosphorylated pCP10-Δloop2 for transformation of E. coli DH5α. Successful constructs were screened by colony PCR using A2f and TD2 primers (Table S2).

Protein expression, solubilisation and trypsin activation conditions

Each toxin variant was cultured in 50 ml of LB (plus 1 mM IPTG and 100 μg/ml ampicillin) at 37 °C, 240 rpm for 16 h. The culture was pelleted and resuspended in 1 ml of lysis buffer (50 mM Tris HCl pH 8, 500 mM NaCl, 10 mM EDTA, 5 mM β-mercaptoethanol, 0.35 mg/ml lysozyme and 5% v/v triton) and incubated for 30 min at RT. Then, 10 μg/ml DNase (Sigma), 10 μg/ml RNase (Sigma) and 6 mM MgCl2 were added and incubated for a further 20 min. The cell suspension was sonicated on ice (Sonifier® SLP) for 6 min with cycles (10 s on/10 s off) at full power. The homogenate was centrifuged at 12000 × g for 20 min and the supernatant discarded. The pellets were resuspended in 500 µl of 50 mM Na2CO3 pH 9.6 and 10 mM DTT, incubated for 1 h at 37 °C, centrifuged and supernatant saved. For activation, solubilized protein was digested with TPCK-treated trypsin from bovine pancreas (Sigma) at a ratio of 1:1 (w/w) at 37 °C for 1 h. Protein samples were analysed by SDS-PAGE (12%, Coomassie blue stained). Total protein content was determined by the Bradford method. Single protein content was estimated using a ChemiDoc equipment (Bio-Rad) on SDS-PAGE gels using BSA as standards.

A. aegypti bioassays

Ten early fourth-instar A. aegypti larvae were placed in 10 ml dechlorinated water containing 20 µg/ml of each Cry1Aa13 selected mutant, the lepidopteran active Cry1Aa13-wt or a mixture of solubilised Cry4Aa, Cry4Ba, Cry11Aa and Cyt1Aa, obtained from a sporulated culture of B. thuringiensis subsp. israelensis strain 4Q5 (grown in T3 medium for 72 h, centrifuged at 4000 g for 20 min, and solubilized in Na2CO3 50 mM pH 9.6 and DTT 10 mM for 1 h at 37 °C). Apart from solubilized toxins, ground dry cat food was added in each assay. Assays were performed at 26 °C. Mortality was analyzed every 24 h until all larvae died or pupated. Each toxin was assayed in duplicate and the bioassays were performed twice. The results were analyzed statistically using the XLSTAT program.

For LC50 determination, bioassays were performed under the same conditions using 5 concentrations (20, 10, 5, 2.5 and 1.25 μg/ml) of Cry1Aa13-A8 and Cry1Aa13-A12. Each toxin was bioassayed in duplicate and each assay repeated twice. Mortality was recorded after 6 days and LC50 values were calculated by Probit analysis75.

B. mori bioassays

Toxicity of Cry variants against B. mori was determined using diet contamination bioassays in neonate larvae. A mulberry leaf cut in a square shape (size 3 × 3 cm) was placed in a plastic Petri dish. On the leaf, 100 μl of a toxin solution containing 3.2 μg/ml of toxin was spread with the use of a sterile L-shaped spreader, rendering a toxin concentration of 36 ng of toxin per cm2 of diet, (3 times higher than the LC50 for Cry1Aa reported in the bibliography against B. mori 25. Apart from the toxin variants selected by phage display, Cry1Aa13 and a mixture of Cry toxins obtained from Bti were used as positive and negative controls respectively. Ten neonates of B. mori were placed in each Petri dish with the help of a fine brush. Plates were kept at 25 °C and 50% humidity in an insect room and mortality was recorded daily for 96 h.

Modelling method

The structures of Cry1Aa13 variants were modelled on the structure of Cry1Aa76 submitted under accession number 1CIY in the pdb using the Swiss model program77. The template was believed to be valid due to the fact that it contains only 2 amino acid changes compared to the Cry1Aa13 protein.

Additional information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


  1. 1.

    Höfte, H. & Whiteley, H. R. Insecticidal crystal proteins of Bacillus thuringiensis. Microbiol. Rev. 53, 242–255 (1989).

  2. 2.

    Schnepf, E. et al. Bacillus thuringiensis and its pesticidal crystal proteins. Microbiol. Mol. Biol. R. 62, 775–806 (1998).

  3. 3.

    Bravo, A., Gill, S. S. & Soberón, M. Mode of action of Bacillus thuringiensis Cry and Cyt toxins and their potential for insect control. Toxicon 49, 423–435 (2007).

  4. 4.

    Donovan, W. P., Donovan, J. C. & Engleman, J. T. Gene knockout demonstrates that vip3A contributes to the pathogenesis of Bacillus thuringiensis toward Agrotis ipsilon and Spodoptera exigua. J. Invertebr. Pathol. 78, 45–51 (2001).

  5. 5.

    Donovan, W. P. et al. Discovery and characterization of Sip1A: A novel secreted protein from Bacillus thuringiensis with activity against coleopteran larvae. Appl. Microbiol. Biot. 72, 713–719 (2006).

  6. 6.

    Oestergaard, J., Ehlers, R.-U., Martínez-Ramírez, A. C. & Real, M. D. Binding of Cyt1Aa and Cry11Aa toxins of Bacillus thuringiensis serovar israelensis to brush border membrane vesicles of Tipula paludosa (Diptera: Nematocera) and subsequent pore formation. AppL. Environ. Microb. 73, 3623–3629 (2007).

  7. 7.

    Knaak, N., Franz, A. R., Santos, G. F. & Fiuza, L. M. Histopathology and the lethal effect of Cry proteins and strains of Bacillus thuringiensis Berliner in Spodoptera frugiperda JE Smith Caterpillars (Lepidoptera, Noctuidae). Braz. J. Biol. 70, 677–684 (2010).

  8. 8.

    Sousa, M. E. C. et al. Histopathology and ultrastructure of midgut of Alabama argillacea (Hübner)(Lepidoptera: Noctuidae) fed Bt-cotton. J. Insect Physiol. 56, 1913–1919 (2010).

  9. 9.

    Palma, L., Muñoz, D., Berry, C., Murillo, J. & Caballero, P. Bacillus thuringiensis Toxins: An Overview of Their Biocidal Activity. Toxins 6, 3296–3325 (2014).

  10. 10.

    Bravo, A. et al. Evolution of Bacillus thuringiensis Cry toxins insecticidal activity. Microb. Biotec. 6, 17–26 (2013).

  11. 11.

    Jurat-Fuentes, J. L. & Adang, M. J. Cry toxin mode of action in susceptible and resistant Heliothis virescens larvae. J. Invertebr. Pathol. 92, 166–171 (2006).

  12. 12.

    Pardo-Lopez, L., Soberon, M. & Bravo, A. Bacillus thuringiensis insecticidal three-domain Cry toxins: mode of action, insect resistance and consequences for crop protection. FEMS Microbiol. Rev. 37, 3–22 (2013).

  13. 13.

    Berry, C. & Crickmore, N. Structural classification of insecticidal proteins - Towards an in silico characterisation of novel toxins. J. Invertebr. Pathol. doi:10.1016/j.jip.2016.07.015 (2016).

  14. 14.

    Pigott, C. R. & Ellar, D. J. Role of receptors in Bacillus thuringiensis crystal toxin activity. Microbiol. Mol. Biol. R. 71, 255–281 (2007).

  15. 15.

    Li, J., Carroll, J. & Ellar, D. J. Crystal structure of insecticidal δ-endotoxin from Bacillus thuringiensis at 2.5 Å resolution. Nature 353, 815–821 (1991).

  16. 16.

    de Maagd, R. A., Bravo, A., Berry, C., Crickmore, N. & Schnepf, H. E. Structure, diversity, and evolution of protein toxins from spore-forming entomopathogenic bacteria. Annu. Rev. Genet. 37, 409–433 (2003).

  17. 17.

    Caramori, T., Albertini, A. M. & Galizzi, A. In vivo generation of hybrids between two Bacillus thuringiensis insect-toxin-encoding genes. Gene 98, 37–44 (1991).

  18. 18.

    Ge, A. Z., Rivers, D., Milne, R. & Dean, D. H. Functional domains of Bacillus thuringiensis insecticidal crystal proteins. Refinement of Heliothis virescens and Trichoplusia ni specificity domains on CryIA (c). J. Biol. Chem. 266, 17954–17958 (1991).

  19. 19.

    Abdullah, M. A. F. et al. Introduction of Culex toxicity into Bacillus thuringiensis Cry4Ba by protein engineering. AppL. Environ. Microb. 69, 5343–5353 (2003).

  20. 20.

    Soberon, M., Gill, S. S. & Bravo, A. Signaling versus punching hole: How do Bacillus thuringiensis toxins kill insect midgut cells? Cell. Mol. Life Sci. 66, 1337–1349 (2009).

  21. 21.

    Griffitts, J. S. et al. Glycolipids as receptors for Bacillus thuringiensis crystal toxin. Science 307, 922–925 (2005).

  22. 22.

    Valaitis, A. P., Jenkins, J. L., Lee, M. K., Dean, D. H. & Garner, K. J. Isolation and partial characterization of gypsy moth BTR‐270, an anionic brush border membrane glycoconjugate that binds Bacillus thuringiensis Cry1A toxins with high affinity. Arch. Insect Biochem. Physiol. 46, 186–200 (2001).

  23. 23.

    Hossain, D. M. et al. Characterization of a novel plasma membrane protein, expressed in the midgut epithelia of Bombyx mori, that binds to Cry1A toxins. AppL. Environ. Microb. 70, 4604–4612 (2004).

  24. 24.

    Gahan, L. J., Pauchet, Y., Vogel, H. & Heckel, D. G. An ABC transporter mutation is correlated with insect resistance to Bacillus thuringiensis Cry1Ac toxin. PLoS genetics 6, e1001248, doi:10.1371/journal.pgen.1001248 (2010).

  25. 25.

    Atsumi, S. et al. Single amino acid mutation in an ATP-binding cassette transporter gene causes resistance to Bt toxin Cry1Ab in the silkworm, Bombyx mori. P. Natl. Acad. Sci. USA 109, E1591–1598, doi:10.1073/pnas.1120698109 (2012).

  26. 26.

    Federici, B. A. Insecticidal bacteria: an overwhelming success for invertebrate pathology. J. Invertebr. Pathol. 89, 30–38 (2005).

  27. 27.

    Rosas-García, N. M. Avances en el desarrollo de formulaciones insecticidas a base de Bacillus thuringiensis. Rev. Colomb. Biotec. 10, 49–63 (2008).

  28. 28.

    Barton, K. A., Whiteley, H. R. & Yang, N.-S. Bacillus thuringiensis § -Endotoxin Expressed in Transgenic Nicotiana tabacum Provides Resistance to Lepidopteran Insects. Plant Physiol. 85, 1103–1109 (1987).

  29. 29.

    Fujimoto, H., Itoh, K., Yamamoto, M., Kyozuka, J. & Shimamoto, K. Insect resistant rice generated by introduction of a modified δ-endotoxin gene of Bacillus thuringiensis. Nat. Biotechnol. 11, 1151–1155 (1993).

  30. 30.

    Perlak, F. J. et al. Insect resistant cotton plants. Bio/technology 8, 939–943 (1990).

  31. 31.

    Pinto, L. M. N. et al. Indica Rice Cultivar IRGA 424, Transformed with cry Genes of B. thuringiensis, Provided High Resistance Against Spodoptera frugiperda (Lepidoptera: Noctuidae). J. Econ. Entomol. 106, 2587–2594 (2013).

  32. 32.

    Van Rie, J. Bacillus thuringiensis and its use in transgenic insect control technologies. Int. J. Med. Microbiol. 290, 463–469 (2000).

  33. 33.

    Wu, K., Guo, Y., Lv, N., Greenplate, J. T. & Deaton, R. Efficacy of transgenic cotton containing a cry1Ac gene from Bacillus thuringiensis against Helicoverpa armigera (Lepidoptera: Noctuidae) in northern China. J. Econ. Entomol. 96, 1322–1328 (2003).

  34. 34.

    Zhao, J.-Z. et al. Transgenic plants expressing two Bacillus thuringiensis toxins delay insect resistance evolution. Nat. Biotechnol. 21, 1493–1497 (2003).

  35. 35.

    Siegel, J. P. The mammalian safety of Bacillus thuringiensis-based insecticides. J. Invertebr. Pathol. 77, 13–21 (2001).

  36. 36.

    Choi, J. Y. et al. Isolation and characterization of strain of Bacillus thuringiensis subsp. kenyae containing two novel cry1-type toxin genes. J. Microbiol. Biotechnol. 17, 1498 (2007).

  37. 37.

    Higuchi, K., Saitoh, H., Mizuki, E., Hwang, S.-H. & Ohba, M. A novel isolate of Bacillus thuringiensis serovar leesis that specifically exhibits larvicidal activity against the moth-fly, Telmatoscopus albipunctatus. Syst. Appl. Microbiol. 21, 144–150 (1998).

  38. 38.

    Ibarra, J. E. et al. Diversity of Bacillus thuringiensis strains from Latin America with insecticidal activity against different mosquito species. Appl. Environ. Microb. 69, 5269–5274 (2003).

  39. 39.

    De Maagd, R. A. et al. Domain III substitution in Bacillus thuringiensis delta-endotoxin CryIA (b) results in superior toxicity for Spodoptera exigua and altered membrane protein recognition. Appl. Environ. Microb. 62, 1537–1543 (1996).

  40. 40.

    Rajamohan, F., Alzate, O., Cotrill, J. A., Curtiss, A. & Dean, D. H. Protein engineering of Bacillus thuringiensis δ-endotoxin: Mutations at domain II of CryIAb enhance receptor affinity and toxicity toward gypsy moth larvae. P. Natl. Acad. Sci. USA 93, 14338–14343 (1996).

  41. 41.

    Wu, S.-J., Koller, C. N., Miller, D. L., Bauer, L. S. & Dean, D. H. Enhanced toxicity of Bacillus thuringiensis Cry3A δ-endotoxin in coleopterans by mutagenesis in a receptor binding loop. FEBS lett. 473, 227–232 (2000).

  42. 42.

    Liu, X. S. & Dean, D. H. Redesigning Bacillus thuringiensis Cry1Aa toxin into a mosquito toxin. Protein Eng. Design Select. 19, 107–111 (2006).

  43. 43.

    Pacheco, S. et al. Improvement and efficient display of Bacillus thuringiensis toxins on M13 phages and ribosomes. AMB Express 5, 1–10 (2015).

  44. 44.

    Pacheco, S., Gómez, I., Sato, R., Bravo, A. & Soberón, M. Functional display of Bacillus thuringiensis Cry1Ac toxin on T7 phage. J. Invertebr. Pathol. 92, 45–49 (2006).

  45. 45.

    Marzari, R. et al. Phage display of Bacillus thuringiensis CryIA (a) insecticidal toxin. FEBS lett. 411, 27–31 (1997).

  46. 46.

    Kasman, L. M. et al. Phage display of a biologically active Bacillus thuringiensis toxin. Appl. Environ. Microb. 64, 2995–3003 (1998).

  47. 47.

    Ishikawa, H. et al. A system for the directed evolution of the insecticidal protein from Bacillus thuringiensis. Mol. Biotechnol. 36, 90–101 (2007).

  48. 48.

    Craveiro, K. I. et al. Variant Cry1Ia toxins generated by DNA shuffling are active against sugarcane giant borer. J. Biotechnol. 145, 215–221 (2010).

  49. 49.

    Oliveira, G. R. et al. Improving Cry8Ka toxin activity towards the cotton boll weevil (Anthonomus grandis). BMC Biotechnol. 11, 85 (2011).

  50. 50.

    Vílchez, S., Jacoby, J. & Ellar, D. J. Display of biologically functional insecticidal toxin on the surface of λ phage. Appl. Environ. Microb. 70, 6587–6594 (2004).

  51. 51.

    Pigott, C. R., King, M. S. & Ellar, D. J. Investigating the properties of Bacillus thuringiensis Cry proteins with novel loop replacements created using combinatorial molecular biology. Appl Environ Microbiol 74, 3497–3511, doi:10.1128/AEM.02844-07 (2008).

  52. 52.

    Pigott, C. R. Loop Replacement as a Strategy to Generate Bacillus thuringiensis Cry Protein Toxins with Novel Specificities, University of Cambridge. England (2006).

  53. 53.

    Poncet, S., Delécluse, A., Klier, A. & Rapoport, G. Evaluation of Synergistic Interactions among the CryIVA, CryIVB, and CryIVD Toxic Components of B. thuringiensis subsp. israelensis Crystals. J. Invertebr. Pathol. 66, 5 (1995).

  54. 54.

    Boonserm, P., Mo, M., Angsuthanasombat, C. & Lescar, J. Structure of the functional form of the mosquito larvicidal Cry4Aa toxin from Bacillus thuringiensis at a 2.8-angstrom resolution. J. Bacteriol. 188, 3391–3401, doi:10.1128/jb.188.9.3391-3401.2006 (2006).

  55. 55.

    Angsuthanasombat, C., Crickmore, N. & Ellar, D. J. Comparison of Bacillus thuringiensis subsp. israelensis CryIVA and CryIVB cloned toxins reveals synergism in vivo. FEMS Microbiol. Lett. 73, 63–68 (1992).

  56. 56.

    Revina, L. P. et al. Reconstruction of Bacillus thuringiensis ssp. israelensis Cry11A endotoxin from fragments corresponding to its N- and C-moieties restores its original biological activity. Biochemistry. Biokhimiia 69, 181–187 (2004).

  57. 57.

    Thiery, I., Delecluse, A., Tamayo, M. C. & Orduz, S. Identification of a gene for Cyt1A-like hemolysin from Bacillus thuringiensis subsp. medellin and expression in a crystal-negative B. thuringiensis strain. Appl. Environ. Microbiol. 63, 468–473 (1997).

  58. 58.

    Juarez-Perez, V., Guerchicoff, A., Rubinstein, C. & Delecluse, A. Characterization of Cyt2Bc toxin from Bacillus thuringiensis subsp. medellin. Appl. Environ. Microbiol. 68, 1228–1231 (2002).

  59. 59.

    Otieno-Ayayo, Z. N. et al. Variations in the mosquito larvicidal activities of toxins from Bacillus thuringiensis ssp. israelensis. Environ. Microbiol. 10, 2191–2199, doi:10.1111/j.1462-2920.2008.01696.x (2008).

  60. 60.

    Knight, P. J. K., Crickmore, N. & Ellar, D. J. The receptor for Bacillus thuringiensis CrylA (c) delta‐endotoxin in the brush border membrane of the lepidopteran Manduca sexta is aminopeptidase N. Mol. Microbiol. 11, 429–436 (1994).

  61. 61.

    Vadlamudi, R. K., Weber, E., Ji, I., Ji, T. H. & Bulla, L. A. Cloning and expression of a receptor for an insecticidal toxin of Bacillus thuringiensis. J. Biol. Chem. 270, 5490–5494 (1995).

  62. 62.

    Ma, G., Rahman, M. M., Grant, W., Schmidt, O. & Asgari, S. Insect tolerance to the crystal toxins Cry1Ac and Cry2Ab is mediated by the binding of monomeric toxin to lipophorin glycolipids causing oligomerization and sequestration reactions. Develop. Comp. Immunol. 37, 184–192 (2012).

  63. 63.

    Valaitis, A. P. Localization of Bacillus thuringiensis Cry1A toxin-binding molecules in gypsy moth larval gut sections using fluorescence microscopy. J. Invertebr. Pathol. 108, 69–75 (2011).

  64. 64.

    Jurat-Fuentes, J. L. & Adang, M. J. Importance of Cry1 δ-endotoxin domain II loops for binding specificity in Heliothis virescens (L.). AppL. Environ. Microb. 67, 323–329 (2001).

  65. 65.

    Abdullah, M. A. F. & Dean, D. H. Enhancement of Cry19Aa mosquitocidal activity against Aedes aegypti by mutations in the putative loop regions of domain II. Appl. Environ. Microb. 70, 3769–3771 (2004).

  66. 66.

    Russel, M. Filamentous phage assembly. Mol. Microbiol. 5, 1607–1613 (1991).

  67. 67.

    Roberts, N. A. & Redshaw, S. In The Search for Antiviral Drugs. 129–151 (Springer, 1993).

  68. 68.

    Berry, C. et al. Complete sequence and organization of pBtoxis, the toxin-coding plasmid of Bacillus thuringiensis subsp. israelensis. Appl. Environ. Microb. 68, 5082–5095 (2002).

  69. 69.

    Gammon, K. et al. Conjugal transfer of a toxin-coding megaplasmid from Bacillus thuringiensis subsp. israelensis to mosquitocidal strains of Bacillus sphaericus. Appl. Environ. Microb. 72, 1766–1770 (2006).

  70. 70.

    Hernández-Soto, A., Del Rincón-Castro, M. C., Espinoza, A. M. & Ibarra, J. E. Parasporal body formation via overexpression of the Cry10Aa toxin of Bacillus thuringiensis subsp. israelensis, and Cry10Aa-Cyt1Aa synergism. Appl. Environ. Microb. 75, 4661–4667 (2009).

  71. 71.

    Crickmore, N., Bone, E. J., Williams, J. A. & Ellar, D. J. Contribution of the individual components of the δ-endotoxin crystal to the mosquitocidal activity of Bacillus thuringiensis subsp. israelensis. FEMS Microbiol. Lett. 131, 249–254 (1995).

  72. 72.

    Liu, Y. et al. Controlling the selection stringency of phage display using a microfluidic device. Lab on a chip 9, 1033–1036, doi:10.1039/b820985e (2009).

  73. 73.

    Badran, A. H. et al. Continuous evolution of Bacillus thuringiensis toxins overcomes insect resistance. Nature 533, 58–63, doi:10.1038/nature17938 (2016).

  74. 74.

    Frischauf, A.-M., Lehrach, H., Poustka, A. & Murray, N. Lambda replacement vectors carrying polylinker sequences. J. Mol. Biol. 170, 827–842 (1983).

  75. 75.

    Finney, D. J. Probit analysis. 3rd ed., (Cambridge University Press, London, United Kingdom, 1971).

  76. 76.

    Grochulski, P. et al. Bacillus thuringiensis CryIA(a) insecticidal toxin: crystal structure and channel formation. J. Mo.l Biol. 254, 447–464 (1995).

  77. 77.

    Schwede, T., Kopp, J., Guex, N. & Peitsch, M. C. SWISS-MODEL: An automated protein homology-modeling server. Nucl. Acids Res. 31, 3381–3385 (2003).

Download references


We thank Prof. Ellar from the University of Cambridge for providing the λcc2 library used in this study. We thank the Ministry of Education and Science (MEC) for funding this project (CGL2008-02011) and for the mobility grant for Cardiff University. We thank the Consejería de Ciencia, Innovación y Empresa of Junta de Andalucía for supporting Tania Domínguez (P10-AGR-6409).

Author information


  1. Institute of Biotechnology, Campus Fuentenueva s/n, University of Granada, 18071, Granada, Spain

    • Tania Domínguez-Flores
    • , María Dolores Romero-Bosquet
    • , Diana Marcela Gantiva-Díaz
    • , María José Luque-Navas
    • , Antonio Osuna
    •  & Susana Vílchez
  2. Cardiff School of Biosciences, Cardiff University, Park Place, Cardiff, CF10 3AT, United Kingdom

    • Tania Domínguez-Flores
    •  & Colin Berry
  3. Department of Biochemistry and Molecular Biology I, Campus Fuentenueva s/n, University of Granada, 18071, Granada, Spain

    • Susana Vílchez


  1. Search for Tania Domínguez-Flores in:

  2. Search for María Dolores Romero-Bosquet in:

  3. Search for Diana Marcela Gantiva-Díaz in:

  4. Search for María José Luque-Navas in:

  5. Search for Colin Berry in:

  6. Search for Antonio Osuna in:

  7. Search for Susana Vílchez in:


T.D.F., M.D.R.B., D.M.G.D., and M.J.L.N. performed the experiments, T.D.F., C.B., and S.V. designed the experiments and wrote the manuscript, A.O., C.B., and S.V. applied for funding. All authors helped in reviewing the manuscript and accepted the final version.

Competing Interests

The authors declare that they have no competing interests.

Corresponding author

Correspondence to Susana Vílchez.

Electronic supplementary material

About this article

Publication history






By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.