J.-C. Pernollet, INRA, Domaine de Vilvert, Biochimie et Structure des Protéines, Bâtiment 526, F-78352 Jouy-en-Josas Cedex, France. Tel.: + 33 1 34 65 27 50, Fax: + 33 1 34 65 27 65, E-mail: firstname.lastname@example.org
In insects, the transport of airborne, hydrophobic odorants and pheromones through the sensillum lymph is generally thought to be accomplished by odorant-binding proteins (OBPs). We report the structural and functional properties of a honeybee OBP called ASP2, heterologously expressed by the yeast Pichia pastoris. ASP2 disulfide bonds were assigned after classic trypsinolysis followed by ion-spray mass spectrometry combined with microsequencing. The pairing [Cys(I)-Cys(III), Cys(II)-Cys(V), Cys(IV)-Cys(VI)] was found to be identical to that of Bombyx mori OBP, suggesting that this pattern occurs commonly throughout the highly divergent insect OBPs. CD measurements revealed that ASP2 is mainly constituted of α helices, like other insect OBPs, but different from lipocalin-like vertebrate OBPs. Gel filtration analysis showed that ASP2 is homodimeric at neutral pH, but monomerizes upon acidification or addition of a chaotropic agent. A general volatile-odorant binding assay allowed us to examine the uptake of some odorants and pheromones by ASP2. Recombinant ASP2 bound all tested molecules, except β-ionone, which could not interact with it at all. The affinity constants of ASP2 for these ligands, determined at neutral pH by isothermal titration calorimetry, are in the micromolar range, as observed for vertebrate OBP. These results suggest that odorants occupy three binding sites per dimer, probably one in the core of each monomer and another whose location and biological role are questionable. At acidic pH, no binding was observed, in correlation with monomerization and a local conformational change supported by CD experiments.
reversed phase high performance liquid chromatography
volatile-odorant binding assay
During the pericellular events of olfaction, the transport of airborne, hydrophobic odorant molecules through the sensillum lymph or nasal mucus is generally thought to be accomplished by odorant-binding proteins (OBPs), both in insects and vertebrates [1–5]. In insects, the OBPs especially involved in pheromone detection have been distinguished and called pheromone-binding proteins (PBPs). They have been particularly studied in Lepidopterans [6–9] but only few data concern Hymenopterans . PBPs have been extensively studied in the moth specialist system for sex-pheromone binding [9,11,12].
Insect OBPs are small acidic soluble proteins (13–16 kDa), highly concentrated in the sensilla lymph. The presence of six cysteines and their interval spacing are the most striking features shared by proteins belonging to this family . Recently, disulfide bond arrangements have been established for PBP and OBP from the silkworm moth, Bombyx mori[14,15]. In contrast to vertebrate OBPs, the insect OBP sequences do not show homology with any member of the lipocalin family [16,17]. This was recently supported by the report of the first PBP three-dimensional structure, that of B. mori, which exhibits a novel protein fold made of six α helices. Several OBPs have generally been found in the same species [2,19–21], specially in honeybee , suggesting that, in addition to their carrier function, they may play a role in olfactory coding . Although the physiological function of OBPs is not yet well understood, their essential role in eliciting behavioral response has been demonstrated in the fruit fly . Another class of chemosensory proteins (CSPs) has been described in insects, which have not yet been demonstrated to play an olfactory role in spite of their tissue location. Such proteins are named Os-D in Drosophila melanogaster[19,20], CSP-sg in the desert locust Schistocerca gregaria and Lmig OS-D in Locusta migratoria.
The honeybee (Apis mellifera L.) is able to discriminate a wide range of odorants [27,28]. Honeybee OBPs, which are evolutionary divergent from the Lepidopteran OBPs , were classified in three subclasses of antennal-specific proteins (ASPs), namely ASP1, ASP2 and ASP3 [22,30]. ASP1 has been shown to be associated with queen pheromone detection because of its higher abundance in drones, its location in sensilla placodea and ability to bind 9-keto-2(E)-decenoic acid and 9-hydroxy-2(E)-decenoic acid , the most active components of the queen pheromone blend [31,32]. Based on sequence similarity and tissue-specificity, ASP2, which does not bind any of these queen pheromone components , was assigned to be a member of the insect OBP family . In contrast, the ASP3 subclass  was classified as a member of the CSP protein family due to sequence homology [19,20,26,33]. ASP2 was shown to be a protein of 13 695.2 ± 1.6 Da, deprived of post-translational modifications other than peptide signal removal and formation of three disulfide bridges .
General odorant-binding proteins (GOBPs) involved in the detection of odorants by the insect generalist system have mainly been studied through biochemistry and only few reports deal with odorant binding [6,34,35], but without affinity constant determination.
In the present work we report several structural features of ASP2, such as its disulfide bridge pattern, and its secondary and quaternary structure modifications upon pH variation. We also studied its odorant-binding properties using a global volatile odorant binding assay and microcalorimetry. This report relates the first affinity constant determination of general odorant binding by an insect GOBP.
Materials and methods
Materials and chemicals
ASP2 was expressed by Pichia pastoris and purified as described by Briand et al. . Modified trypsin was obtained from Promega as sequencing grade proteinase. Except where otherwise indicated, chemicals were of reagent grade.
Two nanomoles of lyophilized ASP2 were dissolved in 15 µL of 0.2 m ammonium hydrogen carbonate buffer, pH 8.0, containing 0.5 mm CaCl2 and treated with trypsin at an enzyme/protein ratio of 1 : 50 (w/w) at 37 °C. After 17 h, the reaction was stopped by adding 0.5 µL of formic acid.
HPLC separations and mass spectrometry analysis of tryptic peptides
Peptides were separated by RPLC on-line coupled with ion-spray mass spectrometry (IS-MS; PerkinElmer Sciex API100). RPLC was run using a Perkin Elmer Biosystems device composed of a 140D pump and a 785 UV detector with U shaped fused silica tubing (7 mm path length) using a C18 RP 300 capillary LC column (0.5 × 150 mm, 300 Å) at controlled temperature (40 °C). The gradient was made by mixing solvent A [0.1% (v/v) formic acid, 4 mm ammonium acetate (Fluka)] in Merck HPLC water with solvent B [90% PE-Biosystems acetonitrile, 0.1% (v/v) formic acid, 4 mm ammonium acetate] at a flow rate of 5 µL·min−1. The acetonitrile gradient began at 4.5%, linearly increased to 9% in 15 min, then to 51% in 100 min, then to 85.5% in 10 min, remained constant for 20 min, was lowered down to 45% in 10 min and ended with a continuous wash at this concentration. After being monitored for absorbance at 215 nm, the flow was split between the microion-spray source (0.4 µL·min−1) and a fraction collector device (PE-Biosystems Microblotter 173 A), which allowed peptide blotting onto a Problott polyvinylidene difluoride (PVDF) membrane. IS-MS experiments were controlled with the Sample Control 1.3 software using a positive mode from 200–2000 amu with 0.25 amu steps and a 0.4 ms dwell time. The microion-spray voltage was +5000 V and the orifice plate voltage +40 V. The PVDF membrane containing the peptide fractions was cut out into pieces with a scalpel, which were treated with Biobrene prior to sequencing. Mass spectrometry data were analyzed with the PerkinElmer Sciex biomultiview 1.3 software. The average molar masses were calculated from the sequence using the PerkinElmer sciex peptide map 2.2 software.
Automated Edman sequencing was performed using a Procise 494-HT sequencer with reagents and methods according to the manufacturer (PE-Biosystems).
CD spectra were recorded using a Jasco J-810 spectropolarimeter and analyzed as previously described . ASP2 concentration was determined using UV spectroscopy employing the extinction coefficients of 10 435.8 m−1·cm−1 at 277.3 nm, calculated according to Pace et al. . Protein solutions (about 1 mg·mL−1 in 50 mm sodium phosphate buffer, pH 7.0, or in 50 mm sodium citrate buffer, pH 3.0) were placed in a 0.01 or 1 cm path length cell. Baseline was recorded with appropriate buffer. Estimation of secondary structure was made using the algorithm developed by Deleage & Geourjon .
The molecular mass of the recombinant protein was evaluated by exclusion-diffusion chromatography on a 24 mL (1 × 30 cm) bed volume Superose 12 column (Pharmacia). The column was calibrated with bovine albumin (67 kDa), chicken egg ovalbumin (43 kDa), bovine β-lactoglobulin (36 kDa as a dimer), bovine carbonic anhydrase (30 kDa), soybean trypsin inhibitor (21.5 kDa), bovine α-lactalbumin (14.3 kDa) and bovine ribonuclease A (13.7 kDa) purchased from Sigma. The elution was carried out at a flow rate of 0.2 mL·min−1 with 20 mm ammonium acetate, 150 mm NaCl, pH 7.0, and the elution profiles were obtained from on-line UV detection at 280 nm. A 100 µL sample of purified ASP2 was dissolved in elution buffer with or without 1 m GdmCl (1 h incubation at room temperature) and loaded at 0.5 mg·mL−1 onto the Superose column. For the pH dependent monomerization experiments, ASP2 or bovine β-lactoglobulin were dissolved in elution buffer adjusted to pH 3.0 with acetic acid, but elution was performed at pH 7.0.
Volatile odorant binding assay
A volatile-odorant binding assay (VOBA) was performed to study the binding of airborne odorants onto ASP2 solutions as described . Purified ASP2 was dissolved in 40 µL of 500 mm potassium phosphate buffer, pH 7.5, to a final concentration of 1.5 mm. Negative controls were obtained with both the buffer alone and a 1.5 mmα-lactalbumin solution, a protein known not to be a lipid carrier. Twelve 500 µL glass tubes containing the protein solutions and the control buffer were incubated overnight at 25 °C in a 2 L sealed glass chamber with a pure undiluted odorant (10 µL in the chamber) that evaporated freely. Concentrations in the chamber air were 29.8 µm for IBMP, 33.6 µm for isoamyl acetate, 35.8 µm for 2-heptanone, 29.9 µm for 1,8-cineol and 24.6 µm for β-ionone. Experiments were performed in quadruplicate. The proteins and the control buffer were then extracted at room temperature with 50 µL of chloroform and analyzed by gas chromatography using a GC 8000 Series 8180 Fisons Instrument (Thermoquest) equipped with an on-column injector and FID detector (300 °C). The analytical column used was a DB-1 column (30 m × 0.32 mm, i.d. 0.25 µm, J & W Scientific, Interchim, France) with a deactivated precolumn. The oven temperature gradient was applied from 60–200 °C at 10 °C·min−1 and then raised to 290 °C at 20 °C·min−1. The carrier gas was helium at 8.7 mL·min−1. Odorants directly diluted in chloroform were used for calibration.
Isothermal titration calorimetry
Isothermal titration experiments were carried out at 30 °C with a MCS System (MicroCal) microcalorimeter. ASP2 concentration (20–30 µm in 50 mm sodium phosphate buffer, pH 7.0, or in 50 mm sodium citrate buffer, pH 3.0) in the cell (1.36 mL) was determined using UV spectroscopy as described for CD measurements. Ligand solutions were freshly prepared in the same buffer at 200–500 µm added with 0.2–0.5% methanol and injected in 30 successive 5 µL aliquots at 4 min intervals. The heat of dilution was determined by injecting the same ligand solution into the buffer without protein. The raw data were processed with the Microcal origin software and fitted by using a model of multiple binding sites as described in the supplier’s manual. The titration data were fitted allowing variation of binding enthalpy and dissociation constants for the three binding sites.
Recombinant ASP2 was secreted at high levels from the methylotrophic yeast P. pastoris with its natural signal peptide, allowing physico-chemical and functional studies. Its native-like state was confirmed by N-terminal sequencing and IS-MS . The yeast secretion machinery was observed to properly process this protein, allowing formation of three disulfide bridges.
Disulfide bridge assignment
The recombinant ASP2 protein was subjected to trypsin digestion, which was theoretically sufficient to cleave the polypeptide chain between all cysteines. Fig. 1(A) shows the complete amino acid sequence of ASP2 with the predicted tryptic cleavage pattern. The tryptic peptide mixture was separated by RPLC (Fig. 1B) and analyzed by IS-MS coupled with N-terminal sequencing. The calculated and experimentally determined peptide masses are listed in Table 1. All the expected peptides were identified by mass spectrometry in the chromatogram and, in each case, the measured mass was in perfect agreement with the calculated value. The peptide T2 of mass 3426.9 was found to be linked to the peptide T4 of mass 550.7, resulting in a peptide of 3975.7 ± 0.5 (theoretical mass 3975.7), thus indicating the existence of a disulfide bond between Cys21 and Cys53. Similarly, peptides T3 and T9 appeared as a linked peptide of measured mass 2413.6 ± 0.1 (theoretical mass 2413.7), revealing the disulfide bridge between Cys49 and Cys107. Finally, we observed the peptide resulting from a disulfide link between T8 and T10 with a total mass of 2193.7 ± 0.2, in agreement with a disulfide bond between Cys96 and Cys116 (theoretical mass 2193.5). In all cases, the mass spectrometry results were confirmed by the observation of two corresponding sequences, analyzed in approximately equimolar amounts by automated Edman sequencing. Sequencing and mass spectrometry allowed the direct assigment of the ASP2 disulfide pairs indicated in Fig. 1A as Cys(I)-Cys(III), Cys(II)-Cys(V) and Cys(IV)-Cys(VI).
Table 1. Molecular masses of ASP2 measured and theoretical tryptic peptides determined by IS-MS. Vertical bars between adjacent sequences indicate disulfide bonds. Peptides are numbered according to Fig. 1.
Measured mass deprived of standard deviation due to a monocharged peptide.
The far-UV CD spectrum of ASP2 at neutral pH (Fig. 2A) displayed a positive peak centered at 191 nm and two negative peaks at 208 nm and 222 nm. This clearly showed the presence of abundant α helices. The deconvolution of the CD spectrum indeed revealed that ASP2 was composed of approximately 50% α helix and 5% β-sheet. A spectrum with well pronouced negative peaks at 262.5 and 288 nm was recorded in the aromatic region, indicating a highly asymetric environment of tyrosine residues in the folded structure of ASP2 (Fig. 2B). We studied the effect of pH acidification on the structure of the protein. Whereas the secondary structure (Fig. 2A) is only slightly affected by acidification, a dramatic change of the near-UV spectrum (Fig. 2B) suggested a significant local conformational transition between pH 7.0 and 3.0, leading to a flexible side chain packing.
As shown in Fig. 3A, calibrated exclusion-diffusion chromatography of purified ASP2 at 0.5 mg·mL−1 exhibited an apparent molecular mass of 25.8 kDa at the sensillar lymph pH, which is approximately twice the value obtained from mass spectrometry (13 695.2 ± 1.6 Da), demonstrating dimerization of the recombinant protein at pH 7.0. Upon acidification at pH 3.0 of the sample solution (Fig. 3B), or after treatment with 1.0 m GdmCl before injection on the chromatography column (Fig. 3C), we observed a second peak eluting at an apparent molecular mass of approximately 4100 Da, corresponding to the monomeric form of this protein. Gel filtration was conducted in parallel with bovine β-lactoglobulin at pH 3.0, which is known to undergo a transition from dimer to monomer at acidic pH . The apparent molecular weight for the β-lactoglobulin monomer was approximately 4500 Da, in discrepancy with the real monomer molecular mass (18 kDa). This result can be explained by affinity for the exclusion-diffusion column, probably due to changes in the protein conformation as revealed by CD or to nonspecific hydrophobic interactions with the Superose 12 column. Gel filtration revealed that ASP2 is therefore a dimer at neutral pH at approximately 37 µm and undergoes a pH-dependent transition corresponding to the change of the dimer–monomer equilibrium at pH 3.0. The simultaneous observation of both monomeric and dimeric forms suggests that the equilibrium interconversion rate was slow in our conditions.
Ligand binding properties
The binding properties of recombinant ASP2 were first evaluated with airborne odorants using VOBA , a novel test somewhat different from that described by Ferrari et al. . This test mimics the in vivo uptake of airborne molecules by OBPs in conditions close to those of the perireceptor physiology. ASP2 solution and the two negative controls (α-lactalbumin solution and buffer alone) were tested. Gas chromatography allowed the quantification of the binding onto ASP2 of five odorants added at a micromolar amount [2-heptanone, isoamyl acetate, 1,8-cineol, 2-isobutyl-3-methoxypyrazine (IBMP) and β-ionone](Fig. 4A). These molecules, presented in Fig. 4(B), are of different chemical structures and odors. Some of them are known to be perceived by the honeybee as nonsexual pheromones, namely 2-heptanone and isoamyl acetate [28,31], while 1,8-cineol, IBMP and β-ionone are components of floral scents . IBMP was also chosen because of its use as an odorant reference in numerous studies of binding with vertebrate OBP [2,3,44,45]. In these conditions, the buffer only slightly solubilized these odorants, with a greater efficiency for 2-heptanone. The bovine α-lactalbumin solution did not solubilize any tested odorant significantly more efficiently than the buffer alone. With the exception of β-ionone, ASP2 bound all these molecules, with a weaker efficiency for 1,8-cineol. Under presumably saturating conditions, molar ratios (odorant : ASP2 dimer) were 2.73 ± 1.16 for IBMP, 3.57 ± 0.56 for 2-heptanone, 3.83 ± 0.7 for isoamyl acetate and 2.3 ± 0.34 for 1,8-cineol.
Affinity constants and number of binding sites for diverse odorants
Binding experiments were further conducted using ITC, a direct physical method for assessing ligand–proteins interactions. Four odorants, IBMP, isoamyl acetate, 2-heptanone and 1,8-cineol were studied. Titration calorimetric curves obtained with IBMP (Fig. 5A) clearly show that the binding sites are saturable and that two sites are apparently observed, as indicated by the abscissa of the inflection point. When analyzing the data with a polynomial least square fitting according to a widespread method , a third site was revealed. Whatever the tested odorant, the best fitting was obtained with a model of three binding sites per ASP2 dimer. Association constants and enthalpy are presented in Table 2. For each odorant, two sites (named 1 and 2) exhibited close association constants, generally not significantly different (except for isoamyl acetate), but distinct from a third one (named 3). The mean affinity value of sites 1 and 2 showed that the IBMP affinity constant (4.5 × 106m−1) was 10 times greater than that of the other tested odorants, which all exhibited affinities in the same order of magnitude (from 0.13–0.37 × 106m−1). As regards site 3, IBMP exhibited the weakest Kb3 constant (4.3 × 103m−1), while that of isoamyl acetate was slightly greater (8.5 × 104m−1), an order of magnitude lower than those of 1,8-cineol and 2-heptanone, 5.3 × 105m−1 and 2.2 × 105m−1, respectively. Whereas Kb3 was smaller than Kb1 and Kb2 for IBMP (1000-fold lower) and also for isoamyl acetate (four-fold lower), it was in the same range for the other odorants. In all cases, except for site 3 of 1,8-cineol, the enthalpies were negative. It is tempting to postulate that sites 1 and 2 could be located each within a monomer and site 3 at their interface. However, the weak affinity of site 3 does not support a significant biological role for this site. Upon acidification of the protein solution at pH 3.0, which resulted in monomerization, IBMP binding was abolished, as illustrated in Fig. 5, an observation also made with isoamyl acetate, while other odorants were not tested under acidic conditions.
Table 2. Thermodynamic parameters of ligand binding by ASP2 dimer measured at 30 °C in sodium phosphate buffer (50 mm, pH 7.0).Kb are expressed in m−1 × 10−6, Hb in kcal·mol−1. Error values originate from experimental binding isotherm fitting with a model of three binding sites.
4.18 ± 0.74
0.45 ± 0.05
0.29 ± 0.06
0.14 ± 0.004
−10.9 ± 0.3
−4.4 ± 0.1
−5.7 ± 0.3
−5.4 ± 0.5
4.79 ± 0.71
0.28 ± 0.03
0.32 ± 0.06
0.12 ± 0.005
−7.1 ± 0.3
−1.6 ± 0.2
−4.8 ± 0.6
−2.8 ± 0.3
0.0043 ± 0.001
0.085 ± 0.01
0.53 ± 0.06
0.22 ± 0.007
−98.2 ± 19
−6.7 ± 0.8
1.4 ± 0.4
−5.8 ± 0.1
Disulfide bridge arrangement
In spite of a low sequence homology, ASP2 and two OBPs from B. mori[14,15] exhibited the same cystein pairing (Fig. 1). The low homology between ASP2 and the highly divergent Lepidopteran OBPs enables us to predict that the disulfide bond pairings Cys(I)-Cys(III), Cys(II)-Cys(V) and Cys(IV)-Cys(VI) would be shared by all the members of this insect protein family. Conversely, a chemosensory protein (CSP-sg) of the desert locust S. gregaria, not assigned as an OBP, but supposed to be involved in chemical detection , has only little sequence homology with ASP2 and its disulfide bond pairing was quite different from what we found. These observations suggest that the recombinant ASP2 was properly folded since we found that cysteines were not randomly associated and exhibited a single arrangement similar to those already described for other OBPs [14,15].
CD corroborated that the recombinant honeybee ASP2 was well folded since a large amount of α helices was observed. The ASP2 CD spectrum and the secondary structure proportions obtained by its deconvolution are similar to those of other insect recombinant OBPs, such as Manduca sexta GOBP2 , B. mori PBP , Mamestra brassicae MbraPBP1 and MbraGOBP2 and Antheraea polyphemus ApolPBP1 . This suggests a general similar global fold for insect OBPs, which would be represented by the recently determined three-dimensional structure of B. mori PBP . These structural properties are completely different from those reported for lipocalins, a vast superfamily of odorant and hydrophobic molecule carriers made of an eight-strand β barrel [16,17] and strongly indicate that insect OBPs constitute a new class of lipid carrier proteins. Due to their size, high α helix content and number of disulfide bridges, insect OBPs might be compared to other small lipid carrier proteins such as plant LTP  and elicitins . Like many insect and vertebrate OBPs [2,52], honeybee ASP2 was demonstrated to be a homodimer by gel filtration at sensillar lymph pH.
Comparison of odorant and pheromone binding
A previous functional study of another honeybee OBP, ASP1, revealed it to be a queen pheromone binding protein, while recombinant ASP2 was unable to bind the pheromone blend . In an attempt to study the binding specificities of ASP2, we tested odorant molecules of different chemical structures and odors known to be components of floral scents (1,8-cineol, IBMP and β-ionone) or perceived by honeybee as nonsexual pheromones (2-heptanone and isoamyl acetate). VOBA was performed with a 100-fold excess of airborne odorant in the chamber to saturate ASP2 specific binding sites. Compared to odorant amounts dissolved in the buffer alone, α-lactalbumin exhibited no ability to bind any odorant. In contrast, ASP2 increased the odorant solubility by a factor of from 3–17, illustrating the specific physiological role of this protein as an OBP, being able to bind floral scents at micromolar concentrations. VOBA indicated that ASP2 was not able to bind all tested odorants with the same efficiency. The small globular odorant 1,8-cineol interacted more weakly and a larger and cumbersome odorant (β-ionone) was not observed to be bound by ASP2. Microcalorimetry clearly showed that the binding sites were saturable. Three sites were observed per ASP2 dimer, that is, all odorants bound ASP2 with a 3 : 2 stoichiometry (ligand/monomer), in agreement with the approximate molar ratios of odorants bound to the protein determined by VOBA (from 2.3 ± 0.34 for 1,8-cineol to 3.8 ± 0.7 for isoamyl acetate), although the experimental conditions were totally different (vapor phase odorant delivery instead of addition in solution, millimolar protein concentrations instead of micromolar). Among these three sites, two exhibited close affinity constants and might each be located in a separate monomer, while the third one (site 3), which showed a different binding affinity, would be sited at their interface in the dimer. In the case of IBMP, and isoamyl acetate to a lesser extent, sites 1 and 2 are obviously the specific sites, with high affinities, whereas site 3 only exhibited a low affinity and would not have an important role in odorant transport. Nevertheless, with the two other odorants (1,8-cineol and 2-heptanone), site 3 exhibited a similar (even slightly higher) affinity constant than sites 1 and 2. The significance of the physiological role of site 3 and its location are questionable and may be better understood with the help of the three-dimensional structure of the protein when determined. This is opposite to the single affinity constant reported for pheromones binding moth PBP [9,12], which would support differences between Lepidopteran PBPs and Hymenopteran OBPs.
The affinity constant we found for IBMP (4.5 × 106m−1) is close to those reported for OBPs from diverse vertebrates, which range from 0.3 × 106M−1 for bovine OBP to 107m−1 for porcupine OBP, through 0.6 × 106m−1 for rat OBP  and 3.2 × 106m−1 for porcine OBP . As regards insects, the affinity constants of moth PBP for their corresponding mating pheromones, whose values were between 0.1 and 1.7 × 106m−1[9,12], are close to our data with respect to isoamyl acetate (0.4 × 106m−1), a honeybee alarm pheromone.
Binding alteration at acidic pH
Acidification abolished the binding of ligand onto ASP2, as already observed between B. mori PBP and sex pheromone bombykol . In the same way, although porcine OBP is capable of binding IBMP at acidic pH with high affinity, its binding stoichiometry is nearly halved . We also examined the effect of pH acidification on the protein structure by CD spectroscopy and gel filtration. As already observed with B. mori PBP , the CD spectrum of ASP2 at acidic pH demonstrated that the secondary structure was not largely affected, but local conformational changes involved the aromatic side chains, which became much more flexible. Monomerization of ASP2 was observed to occur at acidic pH or under denaturing conditions as already reported for B. mori PBP  and bovine OBP . It is tempting to postulate that dimer dissociation and conformational changes might be coupled. These results are consistent with a model in which a pheromone–PBP complex or odorant-bound OBP undergo a pH-dependent conformational change triggered by membranes leading to ligand release to the olfactory receptor [47,53].
This is the first report in which binding stoichiometry and affinity have been determined for an insect GOBP with several odorants. This is also an original demonstration of the binding of odorants and an alarm pheromone to a Hymenopteran OBP. In contrast to Lepidopteran OBP, which can be clearly distinguished from PBP, honeybee OBP raises the question of their precise specificity, since an alarm pheromone was observed to bind ASP2, whereas it was not able to bind the queen pheromone components . The ability of the PBP ASP1 to interact with odorants will now be investigated. In addition, we expect to locate the binding sites using site-directed mutagenesis and three-dimensional investigations, experiments made feasable by the large amount of recombinant protein produced, with the aim of clearly defining the relationships between the structure and function of honeybee OBP.
This work was supported by the French Institut National de la Recherche Agronomique and by the Association pour la Recherche sur le Cancer.
Enzyme: trypsin (EC.220.127.116.11).*Present address: FRE 2230, Centre National de la Recherche Scientifique, Université de Nantes, F-44322 Nantes, France.