Identification of a Bacillus thuringiensis Cry11Ba toxin-binding aminopeptidase from the mosquito, Anopheles quadrimaculatus

Background Aminopeptidase N (APN) type proteins isolated from several species of lepidopteran insects have been implicated as Bacillus thuringiensis (Bt) toxin-binding proteins (receptors) for Cry toxins. We examined brush border membrane vesicle (BBMV) proteins from the mosquito Anopheles quadrimaculatus to determine if APNs from this organism would bind mosquitocidal Cry toxins that are active to it. Results A 100-kDa protein with APN activity (APNAnq 100) was isolated from the brush border membrane of Anopheles quadrimaculatus. Native state binding analysis by surface plasmon resonance shows that APNAnq 100 forms tight binding to a mosquitocidal Bt toxin, Cry11Ba, but not to Cry2Aa, Cry4Ba or Cry11Aa. Conclusion An aminopeptidase from Anopheles quadrimaculatus mosquitoes is a specific binding protein for Bacillus thuringiensis Cry11Ba.


Background
The main African vectors of malaria are in the Anopheles gambiae complex mosquitoes [1]. In general, all species of Anopheles have been found to be susceptible to a certain extent to infection by some strain of human plasmodia [2]. Studies on lepidopteran insects revealed several types of Bt toxin-binding proteins (receptors): aminopeptidase N (APN) -like proteins [3,4]; cadherin-like proteins [5,6]; a glycoconjugate [7] and glycolipids [8]. In mosquitoes, two types of receptors were discovered: a protein with maltase activity from Culex. pipiens that binds the Bin toxin of Bacillus sphaericus [9], and a 65 kDa protein of unknown function (lacking aminopeptidase activity) from Aedes aegypti that binds Cry4Ba and Cry11Aa [10].
Two APNs have been identified in Ae. aegypti but not associated with binding Cry proteins [11] APNs (EC 3.4.11.2) are exopeptidases that cleave single amino acids from the N-terminus of a polypeptide. APNs are expressed in many tissues including the brain, the lung, blood vessels, primary cultures of fibroblasts [12], and have the highest levels in intestinal and kidney brushborder membranes [13]. APNs belong to the M 1 family of zinc metallopeptidases [14], which includes related enzymes like aminopeptidase A [15], aminopeptidase B [16,17], and leukotriene A4 hydrolase [18]. APNs have also been implicated as cellular receptors for human, canine, and feline coronaviruses [19].
In this study, intestinal APN from An. quadrimaculatus larvae was isolated and tested for binding ability to different mosquitocidal Cry toxins (Cry2Aa, Cry4Ba, Cry11Aa, and Cry11Ba). Membrane proteins were extracted from An. quadrimaculatus brush border membrane vesicles (BBMV) and separated by anion-exchange chromatography. Fractions containing APN activity were pooled and purified by size-exclusion chromatography. A 100-kDa protein with APN activity was isolated from the BBMV and its N-terminal sequence was determined to be AQLEDYRLND-DVRPTAYRIE. This protein was used to screen different mosquitocidal Cry toxins binding via Biacore analysis. From the screening, it was discovered that only Cry11Ba was able to bind the APN. A protein BLAST search limited to the arthropod database revealed three highly homologous An. gambiae APNs based on the N-terminal sequence.

Results
Purification of An. quadrimaculatus aminopeptidase N SPR analysis requires purified ligands and analytes to be used. Solubilized An. quadrimaculatus BBMV proteins were separated by anion-exchange chromatography and all elution fractions were tested for APN activity. Fractions 19-21 and 24-34 showed APN activity. Fractions 19-21 were made up of a single peak, and fractions 24-34 were made up of at least two peaks (Fig. 1). Fractions 19-21 were pooled, concentrated, and purified further by size-exclusion chromatography. A single peak was eluted at around 75 ml of run volume that correspond to a protein size of about 100 kDa ( Fig. 2A). This peak was collected and was determined to hold APN activity. SDS-PAGE analysis of the protein also indicated a size of 100 kDa (Fig. 2B) and the 100 kDa protein was highly purified. The 100 kDa protein was named APN Anq 100.

Determination of binding affinity by SPR analysis
Initially, APN Anq 100 was evaluated for binding by SPR analysis to four Cry toxins (Cry2Aa, Cry4Ba, Cry11Aa and Cry11Ba), which were previously determined in this laboratory to have mosquitocidal activity towards An. quadrimaculatus (data not shown). Only Cry11Ba bound significantly to APN Anq 100. Further analysis of real-time binding kinetic of Cry11Ba to APN Anq 100 was performed at different analyte concentrations (Fig. 3), followed by global fitting of all the response curves. A 1:1 binding stoichiometry, including a drifting-baseline correction, produced the following apparent rate constants of the bimolecular interaction: k a = 184.0 M -1 s -1 (± 1.0) and k d = 1.03 × 10 -7 s -1 (± 4.01 × 10 -6 ), K D = 0.56 nM. More complex binding models, such as 2-site independent binding (A + B1 ↔ AB1; A + B2 ↔ AB2), and 2-site sequential binding (A + B ↔ AB ↔ AB*) also gave as good fitting as the simple 1:1 binding (A + B ↔ AB) with χ 2 = 0.112 (data not shown).

N-terminal sequence of APNAnq100
A twenty amino acid residue sequence (AQLEDYRLND-DVRPTAYRIE) was obtained from N-terminal sequencing of purified APN Anq 100. Data mining for similar sequences in the arthropod databases revealed high homology (80-85% identities) with 3 conceptual translated proteins from An. gambiae (Table 1). A BLAST search using the first protein's full amino acid sequence from An. gambiae (accession no. EAA08760.1) revealed homology with many aminopeptidases from organisms of other genera (data not shown). This would suggest that the three proteins from An. gambiae have aminopeptidase activity.
Analysis of the N-terminal region with the program Sig-nalP (http://www.cbs.dtu.dk/) predicted that the most probable cleavage site for the signal peptide sequence was between position 25 and 26 for EAA08760.1; between position 27 and 28 for EAA08763.1; and between position 28 and 29 for EAA08929.1. However, the sequences of the proteins shown in Table 2 start at positions further downstream from the predicted cleavage sites, which suggested that there might have been further processing of the N-terminal region of the An. quadrimaculatus APN.
Another protein BLAST search was performed using the sequence of a known conserved region for aminopepti-dases (MAAVPDFSAGAMENWGLL) [22], which yielded 16 homologous proteins from the An. gambiae genomic database ( Table 2). This indicated that there are a large number of aminopeptidase isomers in these mosquitoes.

Discussion and conclusion
An aminopeptidase N (APN) type protein has been implicated as a Cry toxin-binding protein in several lepidopteran species: Manduca sexta [4] [28], Trichoplusia ni [29], Helicoverpa armigera [30] and Spodoptera litura [31]. Recently the binding epitopes of Cry1Aa to an APN from B. mori have been mapped by monoclonal antibody inhibition [32]. Thus, targeting APN for analysis as a possible toxin-binding protein is a reasonable approach.
The surface plasmon resonance (SPR) method allows analysis of bimolecular interaction in the native state, without a potentially interfering label [33]. Thus, since the Cry11Ba and APN Anq 100 interaction detected in this study represents tight (ca. 1 nM K D ) native-state binding, we propose that APN Anq 100 is a putative receptor for Cry11Ba. APN Anq 100 did not bind to Cry2Aa, Cry4Ba or Cry11Aa even though the toxins have insecticidal activity against An. quadrimaculatus. The specific binding of Cry11Ba to APN Anq 100 suggests that its mode of action would be different from Cry2Aa, Cry4Ba, or Cry11Aa. The N-terminal sequence of APN Anq 100 showed high homology with three putative APNs from An. gambiae. One or more of these APNs could act as a binding protein for Cry11Ba.
Recently the binding epitopes of Cry1Aa to an APN from B. mori have been mapped by monoclonal antibody inhibition [32].

Preparation of mosquito brush border membrane vesicles (BBMV)
Fourth instars An. quadrimaculatus larvae were filtered with a nylon mesh, washed in distilled water, separated from large residual food particles, and dried briefly on a filter paper (Fisher) under vacuum suction. Harvested larvae were frozen at -70°C until needed. About 4-6 g of frozen larvae were homogenized in 8-12 ml of cold buffer A (300 mM mannitol, 5 mM EGTA, 17 mM Tris-HCl, pH 7.5). Larvae were homogenized by 40 strokes of Potter-Elvehjem PTFE pestle in glass tube at speed number 5 (~6000 rpm). BBMV were enriched through differential centrifugation by selective divalent-cation precipitations as described by Silva-Filha, et al [34]. The BBMV pellet was resuspended in 1 ml of ice-cold binding buffer (8 mM NaHPO 4 , 2 mM KH 2 PO 4 , 150 mM NaCl, pH 7.4) supplemented with COMPLETE™ (Roche) protease inhibitor and homogenized by10 extrusions using a small Teflon pestle.The protein concentration of the BBMV was measured with the Coomassie protein assay reagent (Pierce), using BSA as the standard. The BBMV was kept at -70°C until needed.

Purification of An. quadrimaculatus aminopeptidase N (APN) from BBMV
Approximately 20 mg of BBMV was solubilized overnight at 4°C in the binding buffer supplemented with 10 mg/ ml of CHAPS (Roche). Later, the solution was vortexed briefly and centrifuged at 15,000 rpm in a JA-17 rotor at 4°C for 10 min. The supernatant was treated with PIPLC for 1 hr at 37°C. The supernatant was separated by anionexchange chromatography (HiTrap 5 ml column, Phar-   [35].

Purification of Cry toxins
An E. coli clone of Cry2Aa (a grateful gift from Takashi Yamamoto) was used as a source of this gene. The cry2Aa gene was extracted by PCR and cloned into plasmid pHT600 and transformed into B. thuringiensis 4Q7, a plasmidless Cryderivitive. The genes cry4Aa, cry4Ba, cry11Aa and cry11Ba were received in the same plasmid vector and host B. thuringiensis strain (gratefully donated by Armelle Delécluse). Single Bt colonies were inoculated into a 5 ml LB medium supplemented with 10 μg/ml erythromycin and grown overnight at 30°C in an incubator-shaker at 250 rpm. These cultures were inoculated into a 500 ml SSM medium [36] also supplemented with erythromycin and incubated a further 4 days until sporulation and autolysis. Bt crystals in the autolysed-cells suspension were purified as described previously [37] for purification of Cry toxins expressed in E. coli, except that the sonication steps were omitted. The crystals were solubilized in carbonate buffer (30 mM Na 2 CO 3 , 20 mM NaHCO 3 , pH 10.0) supplemented with 10 mM dithiothreitol (Roche) at 37°C for 3 hours. Next, the solubilized toxin was incubated with 1/20 (v/v) 10 mg/ml trypsin (Sigma) at 37°C for 3 hours. The activated toxin was purified by FPLC using a Superdex 200 (Pharmacia) column in the carbonate buffer. Protein concentration was measured using the Coomassie protein assay reagent (Pierce) with bovine serum albumin as standard.

Authors' contributions
MAFA and DHD planned the study and wrote the initial draft of the manuscript. MAFA conducted all experiments, except the N-terminal amino acid analysis. APV conducted the N-terminal amino acid analysis. All authors were involved in revising the manuscript and giving final approval of the version to be published.