Isoform-specific modulation of the chemical sensitivity of conserved TRPA1 channel in the major honeybee ectoparasitic mite, Tropilaelaps mercedesae

We identified and characterized the TRPA1 channel of Tropilaelaps mercedesae (TmTRPA1), one of two major species of honeybee ectoparasitic mite. Three TmTRPA1 isoforms with unique N-terminal sequences were activated by heat, and the isoform highly expressed in the mite's front legs, TmTRPA1b, was also activated by 27 plant-derived compounds including electrophiles. This suggests that the heat- and electrophile-dependent gating mechanisms as nocisensitive TRPA1 channel are well conserved between arthropod species. Intriguingly, one TmTRPA1 isoform, TmTRPA1a, was activated by only six compounds compared with two other isoforms, demonstrating that the N-terminal sequences are critical determinants for the chemical sensitivity. This is the first example of isoform-specific modulation of chemical sensitivity of TRPA1 channel in one species. α-terpineol showed repellent activity towards T. mercedesae in a laboratory assay and repressed T. mercedesae entry for reproduction into the brood cells with fifth instar larvae in hives. Thus, α-terpineol could be used as the potential compound to control two major honeybee ectoparasitic mites, T. mercedesae and Varroa destructor, in the apiculture industry.


Introduction
The transient receptor potential (TRP) channel family shares six common transmembrane segments that form sensor and pore domains, and confer cation permeability. However, TRP channels are unique among various ion channels by having diverse cation selectivities and activation mechanisms [1]. TRP channels play major roles for various sensory perceptions such as vision, thermosensation, olfaction, hearing, taste sensation and mechanosensation by functioning as primary signal integrators that allow animals to perceive the external stimuli [2]. TRP channels also enable individual cells to detect changes in temperature, osmolarity and fluid flow in their local environment [1,3].
The metazoan TRP family is classified into eight subfamilies-TRPA, TRPC, TRPM, TRPML, TRPN, TRPP, TRPV and TRPVL-based on the phylogenetic tree constructed by their amino acid sequences of transmembrane segments [4]. Among them, TRPA1 specifically contains 15 sensitivity to various stimuli [5,6]. Molecular structure of TRPA1 is similar to that of TRPV1 by forming a homotetramer to contain the pore with two gates [6][7][8]. The physiological functions of TRPA1 have been characterized using model organisms (mouse, zebra fish, fruit fly and nematode). TRPA1 is activated by nociceptive thermal (either heat or cold) and chemical stimuli, demonstrating that it plays major roles in nociception and inflammatory pain [5]. In addition, the roles in temperature entrainment of the circadian clock, promoting longevity at cold temperatures and induction of embryonic diapause in progeny have been recently reported using Drosophila melanogaster, Caenorhabtitis elegans and Bombyx mori, respectively [9][10][11]. However, TRP channels have been characterized in a limited number of species, and we have recently identified and characterized TRPA1 channel of Varroa destructor (VdTRPA1), the major ectoparasitic mite of honeybee as the first parasite TRP channel [12].
The losses of managed honeybee (Apis mellifera) colonies have considerably increased in Europe and North America during recent years, and various pathogens and parasites associated with honeybees are considered the major causes [13]. Among the parasites, the ectoparasitic mite V. destructor has affected honeybees most severely. Varroa mites are the primary pests of honeybees in all beekeeping continents, except Australia, causing direct impacts on honeybee health (largescale death of larvae and pupae by feeding on hemolymph) as well as indirect effects by vectoring viruses and other honeybee disease agents [13,14]. Furthermore, in many Asian countries, A. mellifera colonies are also infested with another ectoparasitic mite, Tropilaelaps mercedesae, and both mites are usually present in the single colony [15,16]. Although there are some differences in their habitats, such as shorter life cycle and phoretic phase for Tropilaelaps mite, they share many characteristics as new emerging parasites of A. mellifera. Their reproductive strategies are quite similar [16,17] and both mites were reported to vector DWV [18,19]. Thus, the negative impact of Tropilaelaps mite infestation on an A. mellifera colony is almost equal to that of the Varroa mite. Although the current distribution of Tropilaelaps mites is still limited to Asia [16], they could be spread and established in Europe and North America via global trade. The best method to control Tropilaelaps mites is to remove all brood for several days from the hive [16]. However, this is not practical for large-scale commercial beekeepers that manage thousands of honeybee colonies. Thus, controlling Tropilaelaps mites becomes as difficult as managing Varroa mites. Sustainable control strategies for Tropilaelaps mites based on the use of natural compounds would be highly desirable for beekeeping and related industries. In this study, we identified and characterized T. mercedesae TRPA1 (TmTRPA1) and compared it with VdTRPA1. We discuss the evolution and physiological functions of TmTRPA1, the isoform-specific modulation of chemical sensitivity and the potential use of TmTRPA1-activating compounds to control honeybee ectoparasitic mites in the apiculture industry.

Identification of three TmTRPA1 mRNA isoforms
We amplified a partial TmTRPA1 cDNA at 5 0 end by nested RT-PCR with degenerate primers designed based on VdTRPA1 and TRPA1 sequences of two mite/tick species, Metaseiulus occidentalis (XM_003748080.1) and Ixodes scapularis (XM_002405489.1). We obtained the RT-PCR product with an expected size of approximately 500 bp and the sequence had high similarity to TRPA1 sequences of other species. We then designed the primer sets for 5 0 and 3 0 RACE using this cDNA fragment, and 1.3 kb 5 0 RACE and 2.3 kb 3 0 RACE products were obtained and sequenced. Although only a single sequence represented the 3 0 RACE product, we identified three different sequences in the 5 0 RACE product, demonstrating that TmTRPA1 mRNA has at least three isoforms. In fact, three full-length TmTRPA1 cDNAs (TmTRPA1a, TmTRPA1b and TmTRPA1c) which differ at 5 0 ends were finally isolated (figure 1a). According to the assembled genomic sequence of T. mercedesae, we found that these three mRNAs are encoded by 23 (TmTRPA1a) and 26 (TmTRPA1b and TmTRPA1c) exons, and 22 exons are shared among them. They are likely to be generated by transcription from the different initiation sites (figure 1a). Because the translational start codon is located in the unique 5 0 end exon of each isoform, three TmTRPA1 variants contain the different N-terminal sequences. TmTRPA1b and TmTRPA1c share the most N-terminal sequence and TmTRPA1a has the shortest N-terminal sequence (figure 1b). Nevertheless, all isoforms contain 15 ARs and six transmembrane segments (S1-S6) forming the ion-transport domain (figure 1b).

Expression profiles of three TmTRPA1 mRNA isoforms in Tropilaelaps mercedesae
We examined the expression of TmTRPA1a, TmTRPA1b and TmTRPA1c mRNAs in the first legs, second to fourth legs and the whole body (excluding the mouthparts and legs) of T. mercedesae using RT-PCR. The first legs are much longer and thinner than the other legs (figure 1c) and not used for mite movement, and are instead always raised in the air. This demonstrates that the first legs correspond to the insect antennae, as suggested for many mite species [20]. TmTRPA1a and TmTRPA1b mRNAs were present in all of the above body parts; however, higher levels of TmTRPA1a and TmTRPA1b mRNAs were detected in the whole body and the first legs, respectively (figure 1d). TmTRPA1c mRNA was only detected in the whole body but not in any legs by our assay (figure 1d). These results were reproducible with another RNA preparation. Because of the preferential expression of TmTRPA1b in the first legs of T. mercedesae, this isoform is likely to have major roles for sensory perception by the sensory pit organ in the first legs of mites.

TmTRPA1 protein expression in HEK293 cells
Prior to conducting calcium imaging of HEK293 cells expressing one of three TmTRPA1 isoforms, we characterized the protein expression and cellular localization of V5-epitopetagged TmTRPA1 isoforms in HEK293 cells. The proteins with expected molecular weights (134, 143 and 144 kD for TmTRPA1a, TmTRPA1b and TmTRPA1c, respectively) were specifically detected using Western blot (figure 1e). We then tested the cellular localizations of TmTRPA1a, TmTRPA1b and TmTRPA1c by staining the cells expressing the channels with FITC-WGA (wheat germ agglutinin) and anti-V5-epitope antibody. As shown in figure 1f, the fractions of three isoform proteins co-localized with FITC-WGA, indicating that some of these proteins are present at the plasma   rsob.royalsocietypublishing.org Open Biol. 6: 160042 membrane. TmTRPA1b isoform is highly expressed in the first legs of Tropilaelaps mite and localized at the plasma membrane of the transfected HEK293 cells, and it is likely to play major roles in the sensory perception. We thus focused on characterizing TmTRPA1b in this study.

Heat activation of TmTRPA1
We used a calcium-imaging technique with Fura-2 to measure the relative changes of intracellular calcium levels of HEK293 cells expressing TmTRPA1 by temperature fluctuations. Activation of the channel is expected to increase the intracellular calcium levels by influx of extracellular calcium. As shown in figure 2, increased temperature elevated the relative intracellular calcium levels of cells expressing TmTRPA1a, TmTRPA1b or TmTRPA1c, but not mocktransfected cells. Low temperature did not increase the relative intracellular calcium levels of cells expressing TmTRPA1 (figure 2). These results may suggest that all of TmTRPA1 isoforms are heat-activated.

Differential activation of TmTRPA1 isoforms by chemical compounds
Previous reports have shown that mammalian TRPA1 and DmTRPA1 can be activated by a variety of compounds, including electrophilic compounds, which covalently modify the cysteine residues, and by other compounds (for example, menthol and nifedipine) which do not covalently bind with the channel [5]. Nevertheless, it is not known how far the above chemical activation profiles can be extended to TRPA1 in other species. We therefore tested the activation of three TmTRPA1 isoforms by various chemical compounds and particularly focused on plant-derived compounds with tick-repellent activity [21]. We tested 39   rsob.royalsocietypublishing.org Open Biol. 6: 160042 b-citronellol, nerol, methyl jasmonate, carvacrol and a-terpineol) as shown in figure 3. The activation profiles of TmTRPA1b by other compounds are shown in electronic supplementary material, figure S1a,b. As 1,8-cineole inhibits human TRPA1 activity [22], its effect on TmTRPA1b appears to be the opposite. The compound 2-undecanone is already used as a major ingredient of commercially available natural arthropod repellents [23]. The chemical structures of TmTRPA1b-activating compounds are diverse (figure 3; electronic supplementary material, figure S1a,b), suggesting  rsob.royalsocietypublishing.org Open Biol. 6: 160042 diverse activation mechanisms. These results suggest that at least some of the plant-derived tick/mite repellents activate the TRPA1 channels. Similar to the activation of TRPA1 of other species, such as DmTRPA1, electrophilic compounds such as allyl isothiocyanate (AITC), cinnamaldehyde (CA) and diallyl disulfide also activate TmTRPA1b ( figure 4). Thus, the chemical activation profile of TmTRPA1b is the same as that of VdTRPA1 L [12]. Although the chemical activation profile of TmTRPA1c was identical to that of TmTRPA1b (electronic supplementary material, table S1 and figure S2a-d), we found that only six of the compounds activated TmTRPA1a. These are nerol, 2-undecanone, carvacrol, geranylacetone, eugenol and terpinen-4-ol (figure 5; electronic supplementary material, figure S3). However, many other TmTRPA1b/ c-activating compounds such as 1,8-cineole, methyl jasmonate, a-terpineol and AITC failed to activate TmTRPA1a (figure 5; electronic supplementary material, table S1).

Chemical activation of the N-terminal deletion mutants of TmTRPA1b
The above results demonstrate that chemosensitivities of TmTRPA1a and TmTRPA1b isoforms with the unique N-terminal sequences are different. This suggests that the N-terminal sequence is the critical determinant for the chemosensitivity of TmTRPA1 channel. Because TmTRPA1a contains a shorter N-terminal sequence than TmTRPA1b (figure 1b), we constructed serial N-terminal deletion mutants (D21-134, D49-134, D77-134 and D105-134) as well as D84-92 mutant lacking a stretch of nine aspartic acids of TmTRPA1b (figure 6a). Protein expression levels and plasma membrane localizations of the five deletion mutants were comparable to those of wild-type by Western blot and immunofluorescence as shown in electronic supplementary material, S5. This indicates that the N-terminal sequence of TmTRPA1b does not significantly affect the protein stability and intracellular localization in HEK293 cells. We then tested the responses of the wild-type and the above five deletion mutants to a-terpineol and carvacrol by calcium imaging, and found that the wild-type was activated by both compounds, but all of the deletion mutants were activated by carvacrol but not a-terpineol (figure 6b). These results demonstrate that a stretch of nine aspartic acids (84-92) as well as other amino acids in the N-terminal sequence of TmTRPA1b are necessary for the activation by a-terpineol.

a-terpineol repels Tropilaelaps mercedesae
Although most of the plant-derived TmTRPA1b-activating compounds also stimulated AmHsTRPA, a honeybee nocisensitive TRPA channel [24], a few compounds such as a-terpineol and carvacrol described above were inactive to open AmHsTRPA channel [12]. Thus, they may strongly repel Tropilaelaps mites rather than honeybees. Therefore, we tested the repellent activity of a-terpineol on Tropilaelaps mites in the laboratory. We tested the behaviour of Tropilaelaps mites placed on two adjacent filter papers soaked with either 2% DMSO or a-terpineol in 2% DMSO covered with a nylon mesh. In the control assay, both filter papers were soaked with 2% DMSO (figure 7a). As shown in figure 7b, a-terpineol significantly repelled Tropilaelaps mites at concentrations of more than 0.25 mM.

Activation of TmTRPA1b is sufficient to induce gustatory avoidance and attraction behaviours in Drosophila melanogaster
To test whether the activation of TmTRPA1b by a-terpineol is sufficient to elicit any avoidance behaviour, we characterized the gustatory responses of D. melanogaster trpA1 1 mutants [25] expressing TmTRPA1b under Gr33a-Gal4 to sucrose solution containing a-terpineol by proboscis extension response (PER) [12,24]. Drosophila melanogaster Gr33a is one of the gustatory receptor genes widely expressed in aversive taste neurons [26]. The PER frequency of trpA1 1 flies did not significantly rsob.royalsocietypublishing.org Open Biol. 6: 160042 change even in the presence of 120 mM a-terpineol; however, that of trpA1 1 mutants expressing TmTRPA1b but not TmTRPA1a (a-terpineol insensitive isoform, figure 5) was significantly reduced at concentrations more than 80 mM (figure 7c). The results demonstrate that in D. melanogaster, direct chemical activation of TmTRPA1b could induce gustatory avoidance behaviour when expressed in aversive taste neurons where DmTRPA1 is also present [27]. Similarly, because a-terpineol is a neutral compound for D. melanogaster, we tested whether stimulation of TmTRPA1b in the sugar taste neurons by a-terpineol was sufficient to induce gustatory attraction behaviour. We expressed both TmTRPA1b and a calcium-sensor protein, GCaMP6 [28], in the Gr64f-positive sugar taste neurons [29], and GCaMP6 fluorescence was detected in several discrete neurons in the distal segments of foreleg [12]. As shown in figure 7d, applying 10 mM a-terpineol to the distal segments of foreleg increased GCaMP6 fluorescence in the Gr64f-expressing sugar taste neurons associated with multiple sensilla including 5D1, 5V1 and 5V2 sensilla [30]. We detected the increase of intracellular calcium level (DF/F 0 ) in the sugar taste neuron associated with 5D1 sensilla expressing both TmTRPA1b and GCaMP6, but not expressing either GCaMP6 alone or both TmTRPA1a and GCaMP6 (figure 7e). TmTRPA1b stimulation by a-terpineol was able to activate the sugar taste neurons, and as a result, the application of 10 mM a-terpineol to the forelegs was sufficient to induce significant PER in the fruit flies expressing both TmTRPA1b and GCaMP6, but not either GCaMP6 alone or both TmTRPA1a and GCaMP6 (figure 7f). The results demonstrate that manipulating the activity of taste neurons by the introduction of TmTRPA1b and by the application of a-terpineol is sufficient to modify the gustatory behaviours of fruit flies. rsob.royalsocietypublishing.org Open Biol. 6: 160042 2.9. Application of a-terpineol to Tropilaelaps mercedesae-infested colonies For reproduction of T. mercedesae, the female mite needs to enter into brood cell containing the fifth instar larvae prior to capping (reproductive phase) [16]. Subsequently, the mother mite starts to lay eggs within the sealed brood cell [16,17]. As a-terpineol is capable of repelling Tropilaelaps mites as shown above, we tested whether it could repress female mites from entering brood cells containing the fifth instar larvae. Brood cells treated with a-terpineol (1 ml of 0.3 M solution) showed 29.5% significantly lesser infestation than that in the control cells treated using DMSO (infestation in control cells ¼ 47.9%, infestation in cells treated with a-terpineol ¼ 33.8%; p , 0.018, Mantel-Haenszel test). The results demonstrate that the appropriate concentration of a-terpineol could repress the infestation of T. mercedesae females to brood cells with the fifth instar worker larvae in the colony.

Tropilaelaps mercedesae expresses three TmTRPA1 isoforms
Tropilaelaps mercedesae expresses three TmTRPA1 mRNA isoforms, TmTRPA1a, TmTRPA1b and TmTRPA1c, by transcription from the different initiation sites (figure 1a). All isoforms contain the same number of ARs but unique N-terminal amino acid sequences. Similar isoforms are also present for DmTRPA1, such as TrpA1-RG and TrpA1-RI (FlyBase, http://flybase.org). Although DmTRPA1 isoform with short N-terminal sequence is sensitive to both heat and electrophile stimulations, the isoform with long N-terminal sequence is only activated by chemical stimulation and is preferentially expressed in the chemosensory neurons. Thus, this could be a mechanism to discriminate heat and chemical stimuli by using a single TRPA1 gene in D. melanogaster [31,32]. Similarly, in V. destructor, two TRPA1 isoforms (VdTRPA1 L and VdTRPA1S) with the unique N-terminal sequences and different numbers of ARs are present. In contrast with VdTRPA1 L, VdTRPA1S containing fewer ARs is not activated by chemical stimulation when expressed in HEK293 cells and D. melanogaster [12]. Only VdTRPA1 L appears to be a direct sensor involved in chemo-reception with its exclusive expression in the front legs of V. destructor. VdTRPA1S was proposed to be a downstream component of signalling pathways activated by certain sensory stimuli [12]. Three TmTRPA1 isoforms are different from the above DmTRPA1 and VdTRPA1 isoforms because all of them can be activated by heat ( figure 2) with the same number of ARs. However, as shown in figures 3-5, and electronic supplementary material, table S1 and figures S1 -S3, TmTRPA1a and TmTRPA1b/c have the different chemical sensitivities for activation. The long TmTRPA1b/c is activated by more compounds than the short TmTRPA1a. These results demonstrate that the N-terminal amino acid sequence of TmTRPA1 facing the cytosol is the critical determinant for the chemical activation. In fact, partially deleting N-terminal amino acid sequence of TmTRPA1b resulted in the loss of reactivity to a-terpineol but not carvacrol ( figure 6). Both nine aspartic acids (84 -92) and 30 amino acids (105-134) were independently necessary for TmTRPA1b activation by a-terpineol (figure 6), suggesting that many amino acids in the N-terminus would be required. Alternatively, whole N-terminus of TmTRPA1b may operate as the functional module to support the chemical reactivity. Amino acid substitutions in the N-terminal amino acid sequence of TmTRPA1b by random mutagenesis followed by testing the activities of mutant channels may help to delineate the specific amino acids responsible for the chemical activation. Furthermore, this is the first example to show the evolutionary plasticity of TRP channels by mutations in the cis-regulatory elements (changing the transcriptional initiation sites) to generate the isoforms with different chemical sensitivities. TmTRPA1c is exclusively present in the body but not in any legs of the Tropilaelaps mite by our method, suggesting that it may be specifically expressed in the brain and/or internal organs of body ( figure 1c,d). TmTRPA1a and TmTRPA1b are present throughout the whole body but at a higher level in the body and the first legs, respectively (figure 1c,d). This may suggest that TmTRPA1a and TmTRPA1b are expressed in the different types of sensory neurons enriched in the body and the first legs, respectively. It is difficult to predict how many TmTRPA1 mRNA isoforms are present from the genomic sequence. The deep transcriptome sequencing may help to identify all isoforms present in the Tropilaelaps mite. The first legs of the Tropilaelaps mite morphologically differentiate from the other legs (figure 1c) and are always held upright, demonstrating that they function as sensory organs like insect antennae [20]. TmTRPA1b is likely to be a direct sensor involved in heat-and chemo-reception by the major sensory organ (the front legs) of the Tropilaelaps mite.

Physiological roles of chemosensitive TmTRPA1 for
Tropilaelaps mercedesae We found that TmTRPA1 is activated by a variety of plantderived compounds including electrophiles (figures 3-5; electronic supplementary material, table S1 and figures S1-S3). Many active compounds appear to contain either a long carbon chain or ring structure, suggesting that they may affect the integrity and/or fluidity of plasma membrane. It could indirectly activate TmTRPA1 channels; however, the lack of activation by octanoic acid, decanoic acid and linoleic acid (electronic supplementary material, table S1) suggests that this is unlikely to occur. These relatively hydrophobic TmTRPA1-activating compounds may directly interact with the transmembrane segments of channel to open the gates. As honeybees collect nectar, pollen and resin from various plant species and bring them to the hive, Tropilaelaps mites would have chances to be exposed to the above TmTRPA1-activating compounds, which are potentially hazardous. Some components of plant essential oils are neurotoxic to insects and affect the activities of biogenic amine (tyramine and octopamine) receptors [33], GABA A receptor ion channels [34] and acetylcholine esterase [35], suggesting that they could have similar neurotoxic effects on Tropilaelaps mites. Thus, Tropilaelaps mites, using TmTRPA1, may avoid these compounds, as shown in figure 7a,b. We do not have direct evidence to support this hypothesis because the mock-treated mites for RNAi-knockdown are incapable of moving; however, replacing DmTRPA1 with TmTRPA1b in D. melanogaster was sufficient to induce gustatory avoidance to TmTRPA1b (but not DmTRPA1) activating compound, a-terpineol ( figure 7c). This demonstrates that rsob.royalsocietypublishing.org Open Biol. 6: 160042 chemical activation of TmTRPA1b in aversive taste neurons is sufficient to induce avoidance behaviours by the activating compounds. Meanwhile, the activation of TmTRPA1b in the sugar taste neurons by a-terpineol could induce PER (figure 7f), showing that the behavioural outcome of TmTRPA1b activation is context-dependent in D. melanogaster.

Potential use of TmTRPA1-activating compounds to control both Tropilaelaps mercedesae and Varroa destructor in apiculture industry
In many Asian counties, managed A. mellifera colonies are often infested by both Varroa and Tropilaelaps mites. Thus, miticides and other chemical compounds such as sulfur, formic acid and thymol have been used to control both mite species [16]. However, the efficacy of the above treatments would be different between two mite species, and Tropilaelaps mites may also develop resistance to pyrethrids (tau-fluvalinate and flumethrin) due to the mutations in the voltage-gated Na þ channel as demonstrated in Varroa mites [36,37]. Thus, developing effective control strategies for both mite species is crucial for the apiculture industry worldwide. Given the repellent activity of a-terpineol towards Tropilaelaps mites in a laboratory assay (figure 7a,b), we found that applying 0.3 M a-terpineol into brood cells of hives with fifth instar larvae repelled female mites from entering. However, a-terpineol failed to repel the mites at 0.1 M, suggesting that the activation of TmTRPA1 by a-terpineol may become ineffective inside a honeybee hive where lots of other odours and tastants are present at high concentrations. Because a-terpineol was also effective for Varroa mites by the same assay [12], it can be used as the potential compound to control both Tropilaelaps and Varroa mites. Although a-terpineol did not activate AmHsTRPA [12], its short-term effects on honeybees and its long-term effects on the entire colony are yet to be characterized. Nevertheless, the appropriate use of TmTRPA1-activating natural compounds derived from plants may serve as an alternative method to control Tropilaelaps mites in the apiculture industry.

Functional conservation of TRPA1 channels between two major species of honeybee ectoparasitic mite
TmTRPA1 is the sixth arthropod TRPA1 to be characterized in detail besides DmTRPA1, Anopheles gambiae TRPA1, B. mori TRPA1, Helicoverpa armigera TRPA1 and VdTRPA1. It is the second TRPA1 characterized in Acari, which includes mites and ticks, representing the majority of ectoparasites of various animals and plants. As we previously reported, VdTRPA1 and DmTRPA1 share some channel properties (for example, electrophile sensitivity); however, they are activated by the different compounds [12]. This is consistent with the speciesdependent temperature/chemical activation of vertebrate TRPA1 channels [38][39][40][41]. These results led us to hypothesize that amino acid substitutions in TRPA1 would be often driven by adaptive evolution to adjust the physiological functions in association with different specific habitats and life histories of the respective animal species [42]. Amino acid sequences of TmTRPA1b and VdTRPA1 L share 87% identity and 92% similarity, and there are no differences in their channel properties as far as we tested. Both are activated by the same compounds. This is intriguing because the minimal amino acid substitutions are sufficient to change the properties of TRPA1 channel, such as the activation temperature directionality [43]. The important roles of TRPA1 to mediate honeybee-mite interactions as well as to adapt to the hive environment may keep the channel properties of VdTRPA1 and TmTRPA1 similar to each other. In fact, we found six positively selected amino acids in the ancestor of VdTRPA1 and TmTRPA1. These are 239S, 334I, 435N, 523Y, 677N and 1133R (the numbers refer to the positions in TmTRPA1b), and the first five amino acids are located in the AR3, AR5, AR8, AR10 and AR14 of TmTRPA1b, respectively. This may suggest the importance of ARs to modulate the gating property of the TRPA1 channel. Nevertheless, as shown in figures 1 and 3-5, generation of the isoforms and their expression profiles appear to be different and generate the diversity of TRPA1 channels in two species of honeybee ectoparasitic mite. Very little is known about the sensory physiology of T. mercedesae despite its critical roles in host finding and reproduction, and thus, further study on the TRP channels would aid in developing novel control methods to fight this major pest of the apiculture industry in Asia.

Construction of TmTRPA1-expressing vector for mammalian cells
We isolated full-length TmTRPA1a, TmTRPA1b and TmTRPAc cDNAs by nested RT-PCR with T. mercedesae total RNA and two primers for the first PCR, either 5 0 -AAAATCGTCGAA GGCTACTGCCTC-3 0 (for TmTRPA1a), 5 0 -AATCCGCGTTC TACCCTTGACCGT-3 0 (for TmTRPA1b) or 5 0 -GAGCAGA rsob.royalsocietypublishing.org Open Biol. 6: 160042 TCTCCATCATCCAGTCG-3 0 (for TmTRPA1c) and 5 0 -GTCC GAAGCCACCAAGGACGTAATAGG-3 0 . The second PCR was then carried out using the first PCR product as a template and the following two primers: either 5 0 -AATTTGCGGCCG CACCATGGCGCGAAG GCTCATGAGAGATGCTTC-3 0 (for TmTRPA1a), 5 0 -AATTT GCGGCCGCACCATGCGTCAACAA ACCGTGTCTTGG-3 0 (for TmTRPA1b) or 5 0 -AATTTGC GGCCGCACCATGGACCACGGAACGGTTCACACG-3 0 (for TmTRPA1c) and 5 0 -TTTCTAGACTCTTGACATCCGTTTGTC CATCTAGCTGTTC-3 0 . The second PCR products were digested with Not I and Xba I, and then cloned in pAc5.1/V5-His B vector (Life Technologies) in which D. melanogaster actin 5C promoter was replaced with CMV promoter. The TmTRPA1 protein expressed by this construct was tagged with a V5-epitope at the C-terminus and this was used for verifying the expression and the cellular localization in HEK 293 cells by Western blot and immunofluorescence with rabbit anti-V5-epitope antibody (Sigma-Aldrich), respectively. The staining patterns of V5-epitope-tagged TmTRPA1 channels were compared with FITC-WGA, which specifically labels the plasma membrane [44]. A construct expressing untagged TmTRPA1 protein was then prepared using the above DNA construct as a template, the above primer with the initiation codon and the primer 5 0 -TTTCTAGACTCTACTTGACAT CCGTTTGTCCATC-3 0 . This DNA construct was used for all of the experiments described in the text. The TmTRPA1b deletion mutants were generated by overlap extension PCR [45].

PER assay of Drosophila melanogaster
PER assay was basically performed as described in [12]. Fourto five-day-old fruit flies were starved overnight on wet Kim Wipe, anaesthetized on ice and affixed to a tooth pick. Fruit flies recovered in a humidified chamber for at least 2 h at room temperature before testing. Each subject was checked for intact PER before beginning the experiments. Flies that did not show the reflex were discarded. During the PER assay, the fruit fly was first satiated with water, then 100 mM sucrose solution containing none or the different concentrations of a-terpineol (WAKO) shown in figure 6c was touched to the forelegs with a pipette tip. If the proboscis was extended and contact with the sucrose solution was maintained for 3 s, the response was scored as 1. If the contact of proboscis was brief, a 0.5 was awarded. If the proboscis failed to contact the solution within 5 s of offering, a 0 was awarded. Each fruit fly was offered compounds seven times per experiment and between offering water was given to satiation. If a-terpineol was accepted on first offering, PER frequency was calculated for the second through to the seventh offerings (sum of six scores per fruit fly divided by six). Three groups of 7-8 flies per genotype were tested. For PER induction by a-terpineol, the forelegs of fruit flies prepared as above were touched with the 10 mM solution and then the extension of proboscis was scored followed by checking the intact PER by sucrose.

GCaMP6 imaging in Drosophila melanogaster
The female forelegs for GCaMP6 imaging were prepared as described previously [30], except that the distal segments were covered with 5 ml of water on a glass slide. Three to four images were obtained first (0 s), and then time-lapse recording (every 0.5 s for 30 s) by a CCD camera (RETIGA 2000-RV, Roper Scientific Photometrics) was started by adding 5 ml of 20 mM a-terpineol. Data were acquired and analysed by IMAGE-PRO PLUS software (Media Cybernetics, Inc.).
rsob.royalsocietypublishing.org Open Biol. 6: 160042 4.8. Repellent assay with Tropilaelaps mercedesae in a laboratory Two filter papers were soaked with either 2% DMSO or repellent in 2% DMSO and then covered with nylon mesh after joining them side by side. Tropilaelaps mites (approx. 20) were aligned at the boundary of two filter papers at the beginning of assay (0 min) and then allowed to move for 30 min in the testing arena (9 cm diameter circle). The number of mites in the area with either DMSO or repellent was counted after 30 min and the avoidance index was calculated. Two filter papers soaked with 2% DMSO were used as a control assay. The experiments were repeated three times for each concentration of a-terpineol.

4.9.
Bioassay of a-terpineol with Tropilaelaps mercedesae-infested Apis mellifera colonies One ml of 0.3 M a-terpineol was applied to each worker brood cell containing the fifth instar larvae in Tropilaelaps mite-infested honeybee colonies. An equal number of cells was also treated with 1 ml of DMSO and used as a control. After 2 days, the sealed cells were opened and inspected, and the number of mite-infested cells was counted. The experiment was repeated four times with the different colonies and 172 brood cells in total were tested for each experimental group.