INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Mosquito and black fly populations, which transmit diseases such as malaria, filariasis, and onchocerciasis, have primarily been controlled with chemical pesticides. The emergence of resistant insect populations has led to increased interest in biological control, including utilization of naturally occurring entomopathogenic bacterial strains. Two bacterial strains, Bacillus thuringiensis subsp. israelensis and Bacillus sphaericus, have been used safely and efficiently to control black flies and Culex larvae, respectively. The entomopathogenic properties of these bacteria are due to their synthesis of protoxin crystals during sporulation. B. thuringiensis subsp. israelensis and B. sphaericus inclusions differ in toxin composition, activity spectra, and mode of action (reviewed in references 25-27).

The insecticidal activity of B. thuringiensis subsp. israelensis results from the synergistic action of four major proteins, of 125, 135, 68, and 27 kDa (6, 24), which are now referred to as Cry4Aa1, Cry4Ba1, Cry11Aa1, and Cyt1Aa1 (7), respectively. In this paper, the corresponding genes will be referred to as cry4A, cry4B, cry11A, and cytA, respectively. Two kinds of toxin (crystal toxins and Mtx toxins), differing in both composition and time of synthesis, seem to be responsible for the larvicidal activity of B. sphaericus. The crystal proteins (a binary toxin consisting of equimolar quantities of the 41.9-kDa and 51.4-kDa proteins) are present in all highly active strains and are produced during sporulation. The Mtx toxins, responsible for the activity of most strains with low activity, seem to be synthesized only during the vegetative phase (reviewed in references 3 and 5).

A number of other B. thuringiensis strains with mosquitocidal activity have been identified (28). Bacillus thuringiensis subsp. jegathesan and Bacillus thuringiensis subsp. medellin are the most promising of these, because both synthesize proteins, Cry11Ba1 and Cry11Bb1, respectively, that are 10 times more toxic to mosquitoes than the most active toxin of B. thuringiensis subsp. israelensis (9, 20). For simplicity, we will refer to the crystal toxins from B. thuringiensis subsp. jegathesan and B. thuringiensis subsp. medellin as Cry11Ba and Cry11Bb, respectively.

The simultaneous production of several larvicidal toxins in a single organism may broaden the host range of the strain and increase its toxicity. Combinations of toxins with overlapping targets but different modes of action might also delay the appearance of resistance in treated mosquito populations. B. sphaericus persists for longer in mosquito breeding sites than B. thuringiensis subsp. israelensis and can recycle in the larvae (4, 11, 21). It therefore appears to be the best host strain for combining multiple toxin genes.

Several mosquitocidal toxin genes have been transferred into B. sphaericus (2, 23, 31, 33) by using shuttle vectors with antibiotic resistance genes which are undesirable in biopesticides. A method based on in vivo recombination, allowing the integration of heterologous genes into the chromosome of B. sphaericus, has been developed (22). Genes transferred in this way are stably maintained in the absence of antibiotic selective pressure and other foreign DNA. In this case, the expression of a single gene, cry11A, conferred activity against organisms of the genus Aedes on B. sphaericus and decreased the resistance of a resistant laboratory population of Culex quinquefasciatus.

We further increased toxicity to Aedes and resistant organisms of the genus Culex by introducing the highly mosquitocidal Cry11Ba from B. thuringiensis subsp. jegathesan, alone or in combination with Cry11A, into the chromosome of B. sphaericus 2297. We constructed a novel integrative vector to achieve this: the gene encoding the 36-kDa protease gene that may be involved in the cleavage of the Mtx protein (34) was used as the integration site for heterologous genes. The larvicidal activities of recombinant strains were determined by using Aedes aegypti and Culex pipiens larvae. We also report the activities of the recombinant strains against Culex populations resistant to the binary toxin.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, media, and plasmids. Escherichia coli TG1 [K-12(lac-proAB) supE thi hsdD F'(traD36 proA⁺ proB⁺ lacI^q lacZM15)] (14), used in cloning experiments, was grown in Luria broth (LB) containing ampicillin (100 µg/ml) or kanamycin (25 µg/ml). B. sphaericus 2297 from the IEBC Collection of the Laboratoire des Bactéries et Champignons Entomopathogènes (Institut Pasteur, Paris, France) and a recombinant strain of B. sphaericus 2297 containing the cry11A gene (B. sphaericus 2297bin::cry11A) (22) were used as recipient strains and were transformed by electroporation as previously described (23), except that the cells were grown in LB. Only unmethylated DNA isolated from E. coli ET12567 (17) was used for the transformation of B. sphaericus. Transformants were selected on erythromycin (20 µg/ml) or kanamycin (5 µg/ml). The plasmid vectors used, pRN5101, pHT560 (22), and pUC19 (35), have been described elsewhere.

DNA manipulations. Restriction enzymes, T4 DNA ligase, and the Klenow fragment of DNA polymerase I were used as recommended by the manufacturers. DNA polymerase, isolated from Pyrococcus furiosus, was used for PCR as recommended by the supplier (Stratagene, Cambridge, United Kingdom). Plasmid DNA was extracted and purified from E. coli with the Qiagen (Düsseldorf, Germany) plasmid kit. The procedure used for the extraction of genomic DNA from B. sphaericus has been described elsewhere (8). DNA was analyzed by electrophoresis in 0.8% agarose gels. Southern blot analysis and colony hybridization were carried out as described by Sambrook et al. (29) on Hybond N⁺ filters (Amersham, Little Chalfont, Buckinghamshire, United Kingdom). DNA probes were labeled by using [-³²P]dATP and a nick translation kit (Amersham). The nucleotide sequences of both strands were determined by using the Thermosequenase core sequencing kit with 7-deaza-dGTP (Pharmacia, Uppsala, Sweden) and a Vistra DNA sequencer 725.

Plasmid construction. (i) Disruption of the protease gene. Two internal, 550-bp fragments of the protease gene were amplified from B. sphaericus 2297 chromosomal DNA by PCR by using oligonucleotide primers PS304 and -305 (5'ATGTCGACGTTGGGAATTAACAGAGAACGG3' and 5'ATGGATCCTGCTGCATGTCGAATAGCAGC3') or PS306 and -307 (5'ATGGATCCTAGATATCAAGCAAGCAACAGCTACTGGTACCAA 3'-5'ATAAGCTTATTG AACGCGAGCAAATCCAAATC 3') based on the sequence of the gene encoding the subtilisin-like serine protease from B. sphaericus, SSII-1 (34). The 550-bp fragments were amplified and cloned into pUC19 BamHI/SalI sites for the 5' end of the protease gene to yield pPS408 and into pUC19 BamHI/HindIII sites for the 3' end to yield pPS409. The coding sequences of the protease gene from B. sphaericus 2297 and SSII-1 strains differ at few positions. Introducing three enzyme restriction sites (BamHI, EcoRV, and KpnI) into oligonucleotides PS306 and -307 facilitated further plasmid construction. The integrative plasmid, pPS411, was constructed by insertion of the two 550-bp internal fragments (a SalI/BamHI fragment from pPS408 and a BamHI/HindIII fragment from pPS409) into pRN5101 cut with SalI/HindIII. The pPS424 plasmid was used to disrupt the protease gene by two crossover events. It was constructed by inserting the BglII/KpnI fragment carrying the kanamycin resistance gene aphA3 from Enterococcus faecalis (32) into pPS411 cut with BamHI/KpnI.

(ii) Cloning of the cry11A and cry11Ba genes into pPS411. The plasmid pHT643 (10), which carries the p19, cry11A, and p20 genes from B. thuringiensis subsp. israelensis was used as a source of the cry11A gene. The plasmid pPS414 was obtained by inserting the 4.1-kb BamHI/KpnI fragment from pHT643 between the BamHI and KpnI sites of pPS411.

(iii) Cloning of a cry11A fragment. An internal fragment of the cry11A gene was amplified by PCR by using oligonucleotides PS308 and -309 (5'ATGGATCCGAACCTACTATTGCGCCAGC3' and 5'TAGCATGCGTATATAGGATGGACGCCACG3') and inserted between the BamHI and SphI sites of pUC19 to yield pPS401. This fragment was used as a probe in Southern blot analysis.

In vivo recombination in B. sphaericus. B. sphaericus transformants were plated on selective medium containing erythromycin or kanamycin at a permissive temperature (30°C), for the replication of the pRN5101-derived plasmid. One transformant was grown in nonselective medium for about 20 generations at 37°C (nonpermissive temperature). At this temperature, the pRN5101-derived plasmid does not replicate in gram-positive bacteria. The cultures were plated on erythromycin plates: only cells resulting from integration via a single crossover event between the resident protease gene and the homologous fragment carried by the plasmid were able to grow. One clone obtained by a single-crossover event was grown at 30°C in LB without antibiotic selective pressure for 60 to 100 generations and was plated on LB plates without antibiotic (except for pPS424 transformants, which were plated on kanamycin). A second crossover event eliminated the vector carrying the erythromycin resistance gene. We selected a clone with a cry gene obtained via two crossover events, by colony hybridization with a specific probe using Erm^s cells.

Protein analysis. Wild-type and recombinant B. sphaericus strains were grown in MBS medium (15) at 30°C until cell lysis. Spore-crystal mixtures were then washed twice and suspended in 100 ml of ice-cold deionized water. Aliquots (10 ml) were frozen for bioassay experiments. The protein concentration of alkali-solubilized (30 min at 37°C in 0.05 N NaOH) bacterial suspensions was determined by Bradford assay (Bio-Rad, Munich, Germany). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed as previously described (16). The gel was stained with Coomassie brilliant blue, or proteins were transferred to Hybond-C super membrane (Amersham) and screened with the Amersham ECL Western blotting kit, according to the manufacturer's recommendations. Rabbit antisera directed against either total solubilized crystals from B. thuringiensis subsp. israelensis or purified Cry11Ba inclusions were used as primary antibodies.

Mosquitocidal activity assays. Bioassays were performed on fourth-instar larvae of C. pipiens populations either susceptible (strain Montpellier) or resistant to the binary toxin (strain Montpellier SPHAE), A. aegypti (Bora-Bora) and C. quinquefasciatus (GeoR) larvae resistant to the binary toxin (18, 19). Mosquitoes were reared in the laboratory at 26°C and 80% relative humidity with a 14-h day/10-h night photoperiod. Larvae were reared in dechlorinated water and were fed with commercial cat food. The crystal-spore mixtures of B. sphaericus strains were mixed with 10 ml of deionized water in petri dishes (diameter, 5.5 cm) and were tested in duplicate against 20 larvae each. Each bioassay was repeated at least three times. Larval mortality was recorded after 48 h, and 50% lethal concentrations (LC₅₀) were determined by Probit analysis (12) with a program designed by E. Frachon (Laboratoire des Bactéries et Champignons Entomopathogènes). In general, differences between log dose-Probit mortality lines were considered significant if there was no overlapping of fiducial limits of the lines at LC₅₀.

Nucleotide sequence accession number. The DNA sequence of the 36-kDa protease gene has been deposited in the EMBL Nucleotide Sequence Database under accession no. AJ238598.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Disruption of the protease gene by a double-crossover event. Poncet et al. (22) described an approach for introducing foreign DNA into the chromosome of B. sphaericus based on in vivo recombination. The cry11A gene of B. thuringiensis was integrated upstream from the B. sphaericus binary toxin operon by using the thermosensitive replicative plasmid pRN5101. A combination of heterologous cry genes was introduced by using a novel target locus of the B. sphaericus chromosome: the gene encoding the 36-kDa protease that may be responsible for the cleavage of the Mtx protein (34). The disruption of this gene may increase the production of the Mtx toxin during sporulation. Indeed, the product of the protease gene may be responsible for, or contribute to, the proteolysis of Mtx in B. sphaericus (34). We therefore cloned the protease gene from strain 2297 by PCR by using oligonucleotide primers derived from the known sequence of the SSII-1 protease gene. The DNA sequence was determined, and the amino acid sequence was deduced. The two proteases were 98% identical in amino acid sequence (data not shown). pPS424, which contains the kanamycin resistance gene (aphA3) introduced into the protease gene, was used to estimate double-crossover efficiency in B. sphaericus and to test the effect of disrupting the protease gene on larvicidal toxicity. In vivo recombination was performed as described in Materials and Methods. A second crossover event was obtained, eliminating the vector region, by culturing one clone resulting from a single-crossover event at 30°C without selective pressure for 60 to 100 generations. Only 3 to 6% of the cells were Erm^s. Of the Erm^s colonies that had lost the pRN5101 DNA by a second crossover event, 50 to 75% were Km^r. Similar values of double-crossover events were obtained with the cry genes (see next paragraph). The integration of aphA3 into the protease locus of strain 2297 to yield the B. sphaericus 2297pro::aphA3 strain was confirmed by Southern blot analysis (Fig. 1 and 2).

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Construction of the recombinant B. sphaericus strains by homologous recombination. The genes were introduced either upstream from the binary toxin (cry11A) or at the protease locus (cry11A, cry11Ba, and aphA3). E1, EcoRI. Probes used in Southern blot analysis were the 5' end of the binary toxin (BamHI/BglII fragment from pHT560) (), the 5' end of the protease gene (BamHI/SalI fragment from pPS408) (), the internal part of cry11Ba (PstI/HpaI fragment from pJEG80.2) (), and the 5' end of cry11A (BamHI/SphI fragment from pPS401) ().

View larger version (65K): [in this window] [in a new window]   FIG. 2.   Southern blot analysis of total DNA from the wild-type strain and recombinant B. sphaericus strains. Total DNA from strains 2297bin::cry11A pro::cry11Ba (lanes 1); 2297 (lanes 2); 2297pro::cry11Ba (lanes A3 and B3); 2297pro::cry11A (lanes A4 and D3); and 2297pro::aphA3 (lane A5). DNA was digested with EcoRI and subjected to electrophoresis in a 0.7% agarose gel. DNA fragments were transferred onto Hybond N+ membranes and hybridized with [-32P]dATP-labeled probes corresponding to the 5' end of the protease gene (A), the internal part of cry11Ba (B), the 5' end of the binary toxin operon (C), and the 5' end of cry11A (D). The size of the reactive fragments was as expected. In part C, the existence of two copies of the binary toxin in strain 2297 led to the presence of the three bands observed in lane 1.

Integration of the cry11A and cry11Ba toxin genes into the chromosome. The strategy described above was used to integrate the cry11A and cry11Ba genes into the chromosomal DNA of B. sphaericus 2297. The cry11A gene of B. thuringiensis subsp. israelensis was chosen because it encodes a polypeptide active against A. aegypti, C. pipiens, and Anopheles stephensi larvae (23). In addition, a recombinant strain of B. sphaericus 2297bin::cry11A (22), producing Cry11A, was available, making it possible to construct a recombinant strain containing both the cry11A and cry11Ba genes. The integration of the cry11A gene upstream from the toxin operon did not interfere with the sporulation process and did not abolish the synthesis of the binary toxin (see Fig. 3) (22). Plasmids pPS414 and pPS416 were constructed by insertion of the cry11A operon (encoding the P19 protein, Cry11A, and P20 polypeptide) and the cry11Ba gene, respectively, into pPS411. The cry11Ba gene was introduced into the protease loci of the wild-type strain B. sphaericus 2297 and B. sphaericus bin::cry11A to yield B. sphaericus pro::cry11Ba and B. sphaericus bin::cry11A pro::cry11Ba, respectively. The cry11A gene was also introduced into the protease gene of the parental strain, B. sphaericus 2297, to give the B. sphaericus 2297pro::cry11A recombinant strain. The various recombinant strains obtained were checked by Southern blot analysis by using four different probes (Fig. 1 and 2).

Protein analysis. Crystal proteins synthesized during sporulation by the parental strain 2297 and by the recombinant strains were compared. Cells were grown at 30°C in MBS medium with shaking until cell lysis (48 to 72 h). Aliquots of spore-crystal mixtures were analyzed by SDS-PAGE (Fig. 3). The parental strain 2297 and the recombinant strains produced the 42- and 51-kDa components of the binary toxin. The integration of a heterologous gene upstream from the binary toxin operon did not prevent the synthesis of the 42- and 51-kDa components. The recombinant strains also synthesized additional proteins: a 68-kDa component corresponding to the molecular weight of the Cry11A toxin from B. thuringiensis subsp. israelensis in strains 2297bin::cry11A, 2297pro::cry11A, and 2297bin::cry11A pro::cry11Ba and an 80-kDa protein corresponding to the molecular weight of the Cry11Ba toxin from B. thuringiensis subsp. jegathesan in strain 2297pro::cry11Ba. The presence of these proteins was confirmed by Western blot analysis, by using an antiserum directed against solubilized inclusions from B. thuringiensis subsp. israelensis or against Cry11Ba (Fig. 4). SDS-PAGE analysis showed that all recombinant strains except B. sphaericus pro::aphA3 synthesized about half as much binary toxin as the wild-type strain. Cry11A was produced in all recombinants harboring the corresponding gene, in large enough amounts to be detected by gel staining. In contrast, only small amounts of the Cry11Ba toxin were detected in strain 2297pro::cry11Ba by Coomassie blue staining.

View larger version (75K): [in this window] [in a new window]   FIG. 3.   Protein analysis of spore-crystal mixtures from wild-type and recombinant B. sphaericus strains producing the B. thuringiensis toxins. Spore-crystal suspensions (10 µg of protein) were subjected to SDS-PAGE (10% polyacrylamide) followed by staining with Coomassie blue. Lane 1, wild-type B. sphaericus 2297; lane 2, 2297pro::aphA3; lane 3, 2297bin::cry11A; lane 4, 2297bin::cry11A pro::cry11Ba; lane 5, 2297pro::cry11Ba; and lane 6, 2297pro::cry11A. Arrows indicate the positions of Cry11A (66 kDa) and Cry11Ba (80 kDa).

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Detection (indicated by arrows) of Cry11A (A) and Cry11Ba (B) in various B. sphaericus 2297 strains. Spore-crystal mixtures (1 µg of protein) were subjected to SDS-PAGE (10% polyacrylamide) and the proteins were transferred onto a nitrocellulose membrane. The membrane was incubated with antisera raised against either total solubilized crystals from B. thuringiensis subsp. israelensis (A) or solubilized Cry11Ba inclusions (B). Lanes 1, wild-type B. sphaericus 2297; lanes 2, B. sphaericus 2297bin::cry11A pro::cry11Ba; lane A3, B. sphaericus 2297bin::cry11A; lane B3, B. sphaericus 2297pro::cry11Ba; and lane A4, B. sphaericus 2297pro::cry11A. Immunoreactive polypeptides were detected with a peroxidase-conjugated secondary antibody.

Larvicidal toxicity of the recombinant strains. Spore-crystal mixtures of the wild-type strain and of each B. sphaericus recombinant strain were assayed for mosquitocidal activity against larvae of A. aegypti and C. pipiens (Table 1). Disruption of the protease gene with the aphA3 gene did not significantly increase the larvicidal activity of the recombinant strain against A. aegypti or C. pipiens. The parental B. sphaericus strain and recombinants producing Cry11A were equally toxic to C. pipiens. In contrast, the strain producing Cry11Ba was twice as toxic as wild-type B. sphaericus 2297 to C. pipiens. Surprisingly, the presence of both the Cry11A and Cry11Ba toxins halved the toxicity to C. pipiens. Strains producing Cry11A and Cry11Ba, alone or in combination, with the binary toxin were toxic to A. aegypti larvae, whereas the parental strain 2297 was not. Strain 2297pro::aphA3 was not toxic to A. aegypti, demonstrating that the toxicity was due to Cry11 toxins. The two recombinant strains that produced Cry11A (2297pro::cry11A and 2297bin::cry11A) were equally toxic to A. aegypti. Cry11Ba conferred the strongest toxicity against A. aegypti, the strain containing this toxin being twice as toxic as those containing Cry11A. As for bioassays performed on Culex, the simultaneous production of Cry11A, Cry11Ba, and the binary toxin did not increase larvicidal toxicity to A. aegypti over that of recombinant strains producing only one heterologous toxin (Cry11A or Cry11Ba): the toxicity was similar to that for the strain expressing Cry11A alone.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We constructed recombinant B. sphaericus strains by developing a new integration vector, facilitating the replacement of a 36-kDa protease gene by a heterologous toxin gene. The percentage of double crossover with this system was lower (6%) than previously reported (63%) if the integration occurred upstream from the binary toxin gene (22), but the frequency of gene insertion versus gene excision was higher than previously reported (75%). The efficiency was similar if the DNA from a strain already containing an integration at the binary toxin locus was used. This system is therefore valuable for the construction of recombinants containing combinations of toxin genes.

By using this novel vector, recombinant B. sphaericus strains containing Cry11A and Cry11Ba, alone or in combination, were constructed. All recombinants were bioassayed against Aedes and both susceptible and resistant Culex populations. Despite the low level of cry11Ba expression in the recombinant producing Cry11Ba, the strain 2297pro::cry11Ba was the most toxic for all mosquito species tested. This is due to the activity of Cry11Ba, which has been shown to be the most active mosquitocidal toxin (9). However, none of the recombinants was as toxic as B. thuringiensis subsp. israelensis to this species: 10 to 20 ng of crystals was enough to provide the LC₅₀ under the same conditions. The overexpression of cry11Ba might increase activity against A. aegypti. However, it is more likely that the combination with Cry11Ba of other mosquitocidal polypeptides which act synergically would increase activity. The presence of several polypeptides in a single host would also delay the appearance of insect resistance, as previously demonstrated (13).

Surprisingly, the combination of Cry11A and Cry11Ba was less toxic than Cry11Ba alone. Although Cry11Ba and Cry11A are similar, they differ in many amino acids whose role in toxicity is not known. One explanation for these results is the possibility that the two toxins bind to similar receptors, resulting in competition. Another possible explanation concerns the relative amounts of each toxin contained in spore-crystal mixtures of the different strains. The production of Cry11Ba is lower in B. sphaericus 2297bin::cry11A pro::cry11Ba than in B. sphaericus 2297bin::cry11Ba. Consequently, the global activity in the strain containing both toxins will be lower than the global activity in the strain containing Cry11Ba alone.

In conclusion, our results demonstrated that in vivo recombination is a valuable tool for the construction of new B. sphaericus strains with enlarged activity spectra and higher toxicities. Such recombinants would be of value in terms of vector control and environmental-risk limitation, because they contain no antibiotic resistance genes or other foreign DNA, except the toxin genes of interest. Moreover, since the heterologous genes are integrated into the chromosome, the risk of transfer to other microorganisms is low, and no selective pressure is needed. It would be of interest to show that recombinant B. sphaericus strains are also able to delay the emergence of resistance among treated mosquito populations. The utilization of two sites of integration into the chromosomal DNA of B. sphaericus (the binary toxin locus and the 36-kDa protease gene) allows the creation of a combination of appropriate toxin genes in B. sphaericus. This technique permits not only the introduction of heterologous genes into the chromosome but also the disruption of specific genes. For instance, the major protease genes can be deleted in order to increase toxin stability as previously demonstrated by Thanabalu and Porter (30). It may be also noted that in vivo recombination could be used to engineer toxin genes by, for example, the introduction of strong promoters upstream of toxin genes.