Loss of the membrane anchor of the target receptor is a mechanism of bioinsecticide resistance

Abstract

The mosquitocidal activity of Bacillus sphaericus is because of a binary toxin (Bin), which binds to Culex pipiens maltase 1 (Cpm1), an α-glucosidase present in the midgut of Culex pipiens larvae. In this work, we studied the molecular basis of the resistance to Bin developed by a strain (GEO) of C. pipiens. Immunohistochemical and in situ hybridization experiments showed that Cpm1 was undetectable in the midgut of GEO larvae, although the gene was correctly transcribed. The sequence of the cpm1_GEO cDNA differs from the sequence we previously reported for a susceptible strain (cpm1_IP) by seven mutations: six missense mutations and a mutation leading to the premature termination of translation. When produced in insect cells, Cpm1_IP was attached to the membrane by a glycosylphosphatidylinositol (GPI). In contrast, the premature termination of translation of Cpm1_GEO resulted in the targeting of the protein to the extracellular compartment because of truncation of the GPI-anchoring site. The interaction between Bin and Cpm1_GEO and the enzyme activity of the receptor were not affected. Thus, Bin is not toxic to GEO larvae because it cannot interact with the midgut cell membrane, even though its receptor site is unaffected. This mechanism contrasts with other known resistance mechanisms in which point mutations decrease the affinity of binding between the receptor and the toxin.

Environmentally safe toxins produced by Bacillus thuringiensis and/or Bacillus sphaericus have been integrated in management programs to control crop pests such as Heliothis virescens and Plutella xylostella, and disease vectors such as the mosquitoes Anopheles gambiae and Culex pipiens (1, 2). However, the potential benefits of these biopesticides may be rapidly lost because of the proliferation of highly resistant insect populations (3–6). Control strategies to delay or prevent the development of resistance have been developed, based on several assumptions. The most important of these assumptions are that the resistance gene is recessive and that the rate of mutation to generate resistance alleles is low. Currently, it is difficult to evaluate the success of these strategies because we lack adequate methods for monitoring resistance alleles because of our very restricted knowledge of the mechanisms of resistance to bioinsecticides. Bacillus sphaericus is toxic to mosquitoes, mainly because it produces a binary toxin (Bin) in crystals during sporulation. Following the ingestion and solubilization of crystals by larvae, the released toxin is activated and interacts with the brush-border membrane of the midgut epithelium. In a previous study, we reported the partial purification of a Bin-binding protein from IP, a susceptible strain of C. pipiens. This receptor displayed sequence similarity to α-glucosidases and other maltase-like proteins, and was thus named Cpm1, for Culex pipiens maltase 1 (7). We recently isolated the cDNA encoding Cpm1 from IP larvae (cpm1_IP) and showed that Cpm1 has α-glucosidase activity when produced in bacteria (8). On the basis of sequence analysis and biochemical data, it has been suggested that Cpm1 is anchored to the midgut epithelial membrane by a glycosylphosphatidylinositol (GPI) moiety (7, 8). Resistance to B. sphaericus has been described in laboratory-selected strains and in several field populations of Culex isolated from the U.S., France, Brazil, India, Tunisia, and China (4, 6, 9–13). The high level of resistance (about 100,000 times greater than that of IP) developed by GEO, a Californian laboratory-selected strain, is inherited as a single recessive gene (9, 10). Biochemical studies have demonstrated that Bin does not bind to brush-border membrane fractions (BBMF) prepared from the midguts of GEO larvae, whereas a single class of receptor has been identified in susceptible mosquito larvae (10). In this study, we show that a single point mutation (generating a premature stop codon) in the cpm1_GEO sequence results in the production of a secreted form of the receptor that has lost its membrane anchor. The other six mutations detected in this strain have no effect on Bin binding or α-glucosidase activity. Thus, this mutation blocks the toxicity of Bin by preventing the toxin from damaging the membrane, so the insect is able to survive.

Materials and Methods

Mosquito Strains.

Culex pipiens strains IP (susceptible) and GEO [resistant to B. sphaericus (10)], were maintained at the Unit of the Entomopathogenic Bacteria Laboratory at Institut Pasteur, Paris, France, under standard conditions.

In Situ Hybridization.

A 218-bp fragment (nucleotides 1 to 218) of the cpm1 cDNA was used to synthesize digoxigenin-labeled single-stranded DNA probes by PCR. Fourth-instar Culex larvae were cut into 12-μm sections, which were treated as previously described (14). Hybridization was performed at 42°C for 16 h. Sections were washed and incubated for 2 h at 22°C with alkaline phosphatase-conjugated Fab fragments of sheep anti-digoxigenin IgG (Roche Diagnostics). The reaction was developed with a nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) solution supplemented with 5 mM levamisole and 0.1% Tween-20. Sections were mounted in Permount (Fisher Scientific).

Isolation and Sequencing of the cpm1_GEO cDNA.

cDNA was synthesized from poly(A)⁺ RNA isolated from the midgut of GEO fourth-instar larvae by reverse transcription (RT) with Superscript II (GIBCO/BRL) as previously described (8). The 5′ and 3′ ends of the cpm1_GEO cDNA were obtained by using the Marathon cDNA Amplification Kit (CLONTECH). The full-length coding sequence was amplified by PCR with CPC-K (5′-CGGGGTACCCCGATGCGACCGCTGGGAGC-3′, (nucleotides 1 to 17) as the forward primer, and CPT-X (5′-CTAGTCTAGATTCACGAAGATATACCTGGC-3′, (nucleotides 1723 to 1740) as the reverse primer. Additional restriction sites (underlined) were incorporated into each of the two primers: KpnI in CPC-K and XbaI in CPT-X. PCR was performed with the Advantage cDNA polymerase mix (CLONTECH), and the amplification product was subcloned into the pBAD-TOPO vector (Invitrogen). Five clones were analyzed by nucleotide sequencing.

Plasmid Constructs.

We obtained the Sf9-IP construct by amplifying the complete coding sequence of cpm1_IP, using cpm1_IP cDNA as the template and primers CPC-K and CPT-X. The Sf9-GEO and Sf9-IPMut constructs were generated by PCR using the cpm1_GEO and cpm1_IP cDNAs_, respectively, as templates. The primers used were CPC-K and Leu-X (5′-CTAGTCTAGACCAATCGAAAGGTTGATAGC-3′, nucleotides 1684 to 1703), which contains a XbaI site (underlined). Sf9-GEOMut was obtained by site-directed mutagenesis (GeneEditor System, Promega). The stop codon present in cpm1_GEO at position 1705–1707 was changed to a leucine codon by using the primer 5′-TCGATTGGATTGCTGCTAGCG-3′ (the point mutation is underlined). The resulting cpm1_GEOMut cDNA was subsequently amplified by PCR using primers CPC-K and CPT-X. All PCR products were digested with KpnI and XbaI and inserted between the KpnI and XbaI sites of pIZ vector (Invitrogen). We transformed Escherichia coli One Shot cells (Invitrogen) with the ligation mixtures and the cloned PCR products were verified by DNA sequencing.

Cell Culture and Transfection.

Sf9 cells were maintained at 25°C in TNM-FH medium (Invitrogen) supplemented with 10% heat-inactivated FBS and 10 μg/ml gentamycin. Transfection was carried out as recommended by the supplier except that 7.5 μg of construct was used in each experiment.

Generation of Polyclonal Rat Anti-Cpm1 Antibody.

E. coli DH5α bacteria were transformed with pGEX-4T2 (Amersham Pharmacia Biotech) into which the SalI fragment (nucleotides 1254 to 1612) of cpm1_IP cDNA had been inserted. After induction, the recombinant protein was purified on glutathione-Sepharose (Amersham Biosciences). The glutathione S-transferase (GST) moiety was subsequently removed by digestion with thrombin. Rats were immunized with the purified Cpm1 polypeptide at Eurogentec, Brussels, Belgium.

Immunohistochemistry.

Frozen sections were incubated with undiluted normal goat serum for 1 h at 22°C, washed with PBS, and incubated for 16 h with anti-Cpm1 antibody (1:400 dilution in PBS/0.1% saponin). The binding of anti-Cpm1 antibody was detected with a Cy3-conjugated goat anti-rat IgG (Jackson ImmunoResearch). Sections were mounted in Gel/Mount (Biomeda, Foster City, CA). Slides were viewed with a TRITC filter under epifluorescence illumination on a Zeiss Axioplan II microscope.

Membrane Preparations.

BBMF were prepared from the midgut of fourth-instar larvae, as previously described (15). Sf9 cell membrane-enriched fractions were prepared by homogenizing transfected cell pellets in PBS supplemented with Complete protease inhibitor mixture (Roche Diagnostics). Six cycles of freezing/thawing were performed to lyse the cells. Membranes were recovered by centrifugation at 13,000 × g for 20 min and resuspended in cold PBS/Complete.

SDS/PAGE and Immunoblotting.

Proteins were separated by SDS/PAGE, then transferred to Immobilon P (Millipore). The blots were blocked in TBT buffer (10 mM Tris⋅HCl, 150 mM NaCl, pH 7.5 2% Tween-20 10% nonfat milk powder) for 1 h at 24°C. The membranes were probed with the anti-Cpm1 serum (1:5000) for 16 h at 4°C. They were then washed and incubated with a horseradish peroxidase-conjugated goat anti-rat IgG (Jackson ImmunoResearch). Reactive bands were detected with the ECL system (Amersham Biosciences).

Phosphatidylinositol-specific Phospholipase C (PI-PLC) Treatment.

PI-PLC treatment was performed either on BBMF from IP larval midgut, or on Sf9 cell membrane-enriched fractions. PI-PLC (0.6 unit/ml) was added to 25 μg of membrane extract and incubated for 16 h at 30°C. Samples were clarified by centrifugation at 13,000 × g for 20 min, and both pellets and supernatants were analyzed by Western blotting.

Binding Assays.

Bin toxin was produced, purified, and radiolabeled as previously described (15). Homologous competition was assessed by incubating 10 nM ¹²⁵I-Bin and 1 to 1000 nM unlabeled toxin with 25 μg of cell membrane proteins at 25°C for 16 h. In direct binding experiments, 25 μg of Sf9-IP or Sf9-GEOMut proteins was incubated with 2 to 150 nM ¹²⁵I-Bin. Nonspecific binding was assessed in parallel assays in the presence of 1 μM unlabeled Bin. Unbound toxin was removed by centrifugation, and the membrane-bound radioactivity was measured in a liquid-scintillation counter. Values of binding parameters (K_d and B_max) were determined with PRISM 3.0 software.

α-Glucosidase Assays.

Sf9-GEO and Sf9-IPMut were recovered from culture media and affinity-purified on a HiTrap column (Amersham Biosciences). α-Glucosidase activity was assayed by using p-nitrophenyl α-d-glucopyranoside as previously described (8), but at 37°C instead of 30°C. Assays were performed in duplicate for each construct. One enzyme unit was defined as one μmol of p-nitrophenol released per min.

Results and Discussion

Distribution of Cpm1 Transcript and Protein in Culex Midgut.

We have previously shown that Bin toxin does not bind to membranes prepared from GEO larval midgut, BBMF_GEO (10). We first checked for the presence and integrity of Cpm1 in the GEO strain, by using a polyclonal antiserum directed against a C-terminal portion of the protein. In Western blot analysis (Fig. 1), the anti-Cpm1 antibody recognized a 67-kDa protein in BBMF_IP. No signal was detected for BBMF_GEO, suggesting that in GEO mosquitoes, the Cpm1 protein is no longer associated with the midgut epithelial cell membranes. Immunohistochemical staining of cryosections of IP larvae revealed that Cpm1 was strongly and specifically produced in the brush border membranes of the gastric caeca, and the posterior stomach cells (Fig. 2 A and B, arrowheads), the principal site of nutrient digestion and absorption (16). These data are consistent with previous reports indicating that fluorescently labeled Bin binds to these regions (17). In contrast, no signal was detected in any part of the GEO larval midgut (Fig. 2 A and B). We then compared the transcription pattern of cpm1 in IP and GEO cryosections by in situ hybridization with an antisense single-stranded DNA probe (Fig. 2C). The pattern of accumulation of cpm1 was similar in the cardia cells, the gastric caeca, and the posterior midgut of larvae of both strains. Moreover, in Northern blot analysis, similar amounts of the 2.0-kb cpm1 mRNA were detected in the midguts of susceptible and resistant larvae (data not shown). Therefore, the lack of detection of Cpm1 protein in BBMF_GEO does not result from changes in the pattern of transcription, but, instead, from changes in the cpm1 coding sequence.

Figure 1.

Immunoblot analysis of Cpm1 in BBMF prepared from the midgut of mosquito larvae. A polyclonal anti-Cpm1 antibody detected the 67-kDa Cpm1 protein in IP larvae but not in GEO larvae.

Figure 2.

Detection of Cpm1 in susceptible and Bin-resistant Culex larvae. (A and B) Immunohistofluorescence probing of Culex larval midgut. Parasagittal sections of whole fourth-instar larvae were incubated with the anti-Cpm1 antibody or with the preimmune serum as a control (not shown). L, lumen; i, intermediate stomach; p, posterior stomach. The broken line delimits the transition between the intermediate and posterior stomachs. (Bar, 40 μm in A and 50 μm in B.) (C) In situ hybridization of mosquito larvae shows that cpm1 transcripts are expressed in cardia cells (c), gastric caeca (cg), and posterior stomach (p). (Bar, 500 μm.)

A Nonsense Mutation in cpm1_GEO Leads to the Secretion of the Protein from Sf9 Cells.

We reverse-transcribed RNA from the midgut of GEO larvae and cloned the cpm1_GEO cDNA. Sequence analysis revealed that cpm1_GEO differed from cpm1_IP by seven mutations. Six of these mutations were missense mutations that led to amino acid substitutions: Ala-95 → Asp, Lys-115 → Met, Glu-178 → Thr, Asp-230 → His, Asn-265 → Asp, and Leu-486 → Met. The seventh was a T to A replacement at nucleotide 1706, which converted the Leu-569 codon into a stop codon (Fig. 3A). We investigated the relationships between cpm1 mutations and the resistance of the GEO strain to Bin by expressing four cpm1 constructs in Sf9 insect cells (Fig. 3B). Two of these constructs, Sf9-IP and Sf9-GEO, corresponded to the cpm1 cDNAs from IP and GEO mosquitoes, respectively. The third construct, Sf9-IPMut, corresponded to a cpm1_IP cDNA in which the translation was prematurely stopped at Leu-569. The recombinant protein produced was truncated at its C-terminal end, mimicking the nonsense mutation in the GEO sequence. In the fourth construct, Sf9-GEOMut, the wild-type Leu-569 codon was reintroduced into the cpm1_GEO sequence by site-directed mutagenesis. Culture media and membrane fractions from transfected and control cells were analyzed by Western blotting (Fig. 3C). A single band was detected in samples from transfected cells, but not in samples from untransfected Sf9 cells, or from cells transfected with the insert-less expression vector (Sf9-pIZ). The electrophoretic mobilities of the various recombinant proteins were as expected from their calculated molecular masses, indicating that the recombinant proteins were faithfully synthesized. Sf9-IP and Sf9-GEOMut proteins were detected in membrane fractions, whereas Sf9-GEO and Sf9-IPMut proteins were detected in the culture medium (Fig. 3C). Thus, the mutations in the cpm1_GEO cDNA impaired neither the translation, nor the stability of the protein. Nevertheless, one of these mutations, the Leu-569 → Stop substitution, altered the attachment of the receptor to the membrane. This mutation is located at the C-terminal end of Cpm1, in a region consisting of a few hydrophilic residues followed by 3 small amino acids and a stretch of 23 hydrophobic amino acids. This hydrophobic stretch, together with a signal peptide at the N-terminal end, may constitute a signature for the GPI-anchorage of Cpm1 to the epithelial cell membrane. We investigated whether Cpm1 was indeed anchored to the membrane in this way by treating membrane fractions prepared from cells producing Sf9-IP and Sf9-GEOMut with a PI-PLC. In both cases, Cpm1 was no longer detected in the membrane fraction after treatment, but was released into the medium, as shown by Western blotting (Fig. 3D). Similar results were obtained when the experiment was performed on BBMF_IP, demonstrating that GPI-anchoring of Cpm1 to the membrane occurs in vivo, as well as in transfected Sf9 cells. These data strongly suggest that, in GEO mosquitoes, the nonsense mutation results in the production of an unprocessed form of Cpm1, which is unable to anchor to the plasma membrane by a GPI, and is instead secreted by the midgut epithelial cells.

Figure 3.

Cpm1_GEO is not attached to plasma membranes because of the loss of a GPI anchor. (A) Schematic representation of Cpm1. The locations of the seven amino acid substitutions identified by analysis of the cpm1_GEO cDNA are indicated. (B) Diagram of the four Cpm1 proteins produced in Sf9 cells. Vertical black bars indicate the positions of the mutations. The signal peptide and the hydrophobic domain, putatively involved in the GPI-anchoring of the protein, are designated by a green and a blue box, respectively. The red bar indicates the three amino acids that putatively constitute the site of the cleavage required for anchoring. (C) Western blot of membrane-enriched fractions (M) or culture medium (C) collected from untransfected Sf9 cells (Sf9), or from cells transfected with the insert-less vector or the four constructs. Proteins were resolved by SDS/PAGE (5 μg per lane) and blotted onto poly(vinylidene difluoride) (PVDF) membranes, which were then probed with the anti-Cpm1 antibody. (D) PI-PLC treatment of membrane extracts from transfected Sf9 cells. Samples were prepared in serum-free medium and incubated at 30°C for 16 h with (+) or without (−) PI-PLC. Parallel experiments were performed on BBMF_IP. Both soluble (S) and membrane (M) fractions collected after treatment were analyzed.

Functional Analysis of Cpm1_GEO.

We investigated the possible involvement of the six missense mutations in the resistance of the GEO strain by determining whether Sf9-GEOMut could bind the Bin toxin. In competition experiments, ¹²⁵I-Bin bound to Sf9-GEOMut and Sf9-IP membrane-enriched fractions with a K_0.5 value of 10 nM for both constructs. No specific binding of Bin to Sf9-GEO, Sf9-IPMut, Sf9-pIZ, or Sf9 membranes was observed (Fig. 4A). We also observed direct binding of ¹²⁵I-Bin to Sf9-IP or Sf9-GEOMut membrane fractions (Fig. 4 B and C). A Scatchard plot of the specific binding component was linear for both constructs, demonstrating that the toxin binds to a single class of receptor. The dissociation constant (K_d) was 11.4 nM, which is close to the K_d value reported for BBMF_IP (7), and the maximal binding capacity (B_max) was 7.6 pmol/mg of protein. These results indicate that the missense mutations in the cpm1_GEO coding sequence did not affect the interaction between the toxin and its receptor. They also had no effect on α-glucosidase activity, as purified Sf9-GEO and Sf9-IPMut catalyzed the release from p-nitrophenol by p-nitrophenyl α-d-glucopyranoside at similar rates (1.02 unit/mg of protein for Sf9-GEO vs. 1.15 unit/mg of protein for Sf9-IPMut).

Figure 4.

Equilibrium binding of ¹²⁵I-Bin to recombinant Cpm1. (A) Homologous competition experiments performed on membranes from untransfected Sf9 (▵) or Sf9-pIZ (⋄), Sf9-IP (●), Sf9-GEO (■), Sf9-IPMut (○), or Sf9-GEOMut (□) constructs. (B and C, Left) Direct binding of ¹²⁵I-Bin to Sf9-IP, and to Sf9-GEOMut membranes, respectively. Specific binding (●) and nonspecific binding (○) are shown. (Right) Scatchard plots of specific binding indicate that Bin binds to a single class of receptor.

In summary, we have demonstrated that the receptor for the B. sphaericus binary toxin in C. pipiens, the α-glucosidase Cpm1, is anchored to the midgut epithelial cells by a GPI. In the resistant GEO strain, one mutation results in the shortening of the hydrophobic tail of Cpm1, preventing the processing of this protein and its attachment to the plasma membrane, but leaving the toxin binding site unaffected. This mechanism is unusual, as in most other cases of resistance resulting from point mutations in the target sequence, the affinity of the insecticide for its receptor is decreased by several orders of magnitude (18–20). Our findings demonstrate that a point mutation located in a discrete part of a receptor not involved in toxin binding can effectively overcome toxicity, enabling insects to survive. Enzyme assays indicated that Cpm1 retained its physiological activity after its release from epithelial cells. Thus, this type of resistance to the B. sphaericus binary toxin may have fewer pleiotropic effects than resistance to B. thuringiensis toxins, resulting from the loss of a cadherin-like protein (5), or of a β-1,3-galactosyltransferase (21). Consequently, little or no fitness cost would be associated with the cpm1_GEO allele, facilitating the rapid spread of this allele in field populations of C. pipiens. A diagnostic test would now make it possible to evaluate the frequency of this resistance allele in natural populations of Culex that are no longer controlled by B. sphaericus (4, 6, 10–13). Such PCR-based tests have proven to be reliable for estimating the relative importance of alleles of genes involved in resistance to chemical insecticides (22, 23). GPI-anchoring is a known structural feature of some B. thuringiensis toxin receptors (24–28). It would be useful to investigate whether similar resistance mechanisms occur in pests that develop resistance to these or other microbial insecticides.

Abbreviations

Bin

binary toxin

Cpm1

Culex pipiens maltase 1

GPI

glycosylphosphatidylinositol

BBMF

brush-border membrane fractions

PI-PLC

phosphatidylinositol-specific phospholipase C