Mite collection and husbandry
Varroa destructor mites were collected from capped brood cells of Apis mellifera from hives in York, England that had purposefully been untreated for Varroa control. Prior to harvesting mites, the bee frames were maintained in Aberdeen at 27ºC and 40% relative humidity with a 15.5 h : 8.5 h, light:dark regime. Mites were either removed from newly emerged adult bees or from pupae following uncapping of the brood cell and extraction of the pupae. Mites were collected, bisected and placed in RNA ZR extraction buffer (Zymo Research, Orange, California, USA) and stored at -80 ºC prior to initial total RNA isolation.
RT-PCR of GSTs from V. destructor
Six adult (pharate and mobile) and 2 deutonymph mites were homogenized in 600 μl extraction buffer and total RNA prepared using a mini RNA Isolation II Kit (Zymo Research), as per manufacturer's instructions. Eluted RNA was co-precipitated with glycogen in 95% ethanol and resuspended in 10 μl of DEPC-treated water. After isolation, 2 μg total RNA was DNase treated with 2 μl (2U) RQ1-DNase (Promega, Southampton, UK) and 2 μl RQ1 buffer and incubated at 37°C for 30 min. DNase-treated total RNA (2 μg) was incubated at 70°C with 0.5 μg of oligo d(T)15 (Promega) in a total volume of 10 μl for 5 min. Material was snap-chilled on ice for 5 min prior to the addition of 5 μl 5×RT buffer, 1 μl dNTPs (25 mM each), 0.5 μl Bioscript-reverse transcriptase and DEPC water to 25 μl. The reaction was incubated at 42°C for 60 min prior to arrest by heating to 70°C for 5 min.
A V. destructor database from an ongoing pyrosequencing project (Dr. Jay Evans, pers. comm.) was mined for putative V. destructor GST sequences by BLASTx analysis using GST sequences from the deer tick, Ixodes scapularis. Of 14 putative V. destructor GST partial sequences identified, we chose three to evaluate further: two mu-class GSTs (Vd GST-mu1 and Vd GST-mu2) and one kappa class (Vd GST-kap). These Varroa transcripts were isolated by RT-PCR using primers based on these sequences. PCR reactions consisted of 1 μl cDNA template, 5 μl 10 × reaction buffer, 2 μl 50 mM MgCl2, 1 μl dNTPs (25 mM each), 1 μl each primer (10 mM each), 0.5 μl (1.25 U) Taq (Bioline, London,UK) and DEPC-treated water to give a 50 μl total volume. Primer pairs used were Vd GST-mu-F1 (CAAGTTAAATCTTGCGTTTCC) and Vd GST-mu1-R1 (CTGTCCGTCTTCATGTATGC), Vd GST-mu2-F1 (ACGATTTATCCGTTTTGACG) and Vd GST-mu2-R2 (CCAACTGATGTGTCGTTCC), Vd GST-Kap-F1 (TACTGGTGGTCGTTTCAGG) and Vd GST-Kap-R1 (CTTCATAGGCCAAGAGATGC), generating products of 494, 396 and 343 bp, respectively. PCR cycling conditions were as follows: 1 cycle of 5 min at 94°C, followed by 35 cycles of 1 min at 94°C, 1 min at 53°C and 45 s at 72°C followed by a final extension time of 15 min at 72°C. Products were visualised on an agarose gel and specific bands excised and cloned into pCR4-TOPO TA vector (Invitrogen). Purified plasmids were submitted for sequencing to Eurofins MWG (Ebersberg, Germany) from flanking T7 and T3 promoter regions to confirm the original sequence data.
The GST class of the different Varroa sequences was determined by database searches and phylogenetic analysis. For Vd GST-mu1, protein sequences from various acari were extracted from both non-redundant and EST databases at GenBank by tBLASTx. Additionally, the most similar GST sequences in four model insect species for which complete genomes are available were also obtained. Protein sequences were aligned in CLUSTALW2 . These sequences were used to estimate phylogeny with neighbour-joining, minimum evolution and maximum parsimony methods using MEGA version 4.1 . Phylogenetic trees were constructed with 10,000 bootstrap replicates.
Expression of GST transcripts in tissues of V. destructor
V. destructor mites, were removed from brood cells, as described above, and dissected under ice-cold dissection buffer (20 mM TRIS, 5 mM EDTA, 0.9% NaCl, pH 7.4). Malpighian tubules, gut and synganglia and were dissected from 20, 25 and 98 mites respectively, washed in fresh ice-cold dissection buffer and stored in 50 μl RNAlater (Sigma, Poole, UK) at -80 °C. Prior to RNA extraction, an additional 450 μl dissection buffer was added to sample tubes and centrifuged at 10,000 g for 15 min at 4°C. Supernatant was removed and tissue washed with fresh dissection buffer before a final spin at 10,000 g for 15 min at 4°C. Supernatant was discarded and 600 μl RNA ZR extraction buffer added to each tissue sample. Total RNA extraction, DNase treatment and reverse transcription were as described above. Vd GST-mu1, Vd GST-mu2 and Vd GST-kap transcripts in tissue-specific cDNA were assayed by PCR. PCR protocol and cycling conditions were carried out using the Vd GST-mu1, Vd GST-mu2 and Vd GST-kap primers, as described above over 30 cycles. Gel loading was normalized to V. destructor actin (Vd ActinF, CATCACCATTGGTAACGAG and Vd ActinR, CGATCCAGACGGAATACTT) generating a fragment of approximately 195 bp.
Preparation of dsRNA
Vd GST-mu1-dsRNA was prepared using a BLOCK-iT RNAi TOPO transcription kit (Invitrogen), according to the manufacturer's instructions. LacZ-dsRNA was prepared and used as a negative control. Briefly, PCR was carried out as described above using adult female V. destructor cDNA in conjunction with Vd GST-mu1 specific primers (Vd GST-mu1 F1/R1), or with control LacZ-plasmid and LacZ specific primers (LacZ-F2, ACCAGAAGCGGTGCCGGAAA and LacZ-R2, CCACAGCGGTGGTTCGGAT). Products were resolved on an agarose gel, excised and purified using a Qiagen gel extraction kit (Qiagen, Crawley, UK). TOPO-T7 linker was ligated to Vd GST-mu1 and LacZ reactions before a secondary PCR was carried out to gain sense and antisense templates. T7-RNA polymerase was used in transcription reactions with target templates to generate sense and antisense RNA. Finally, RNA strands were annealed and the resultant dsRNA purified and quantified in a micro-spectrophotometer (Nanodrop Technology Ltd). dsRNA was ethanol precipitated and resuspended in DEPC-treated water to a working concentration of 2.5 μg/μl and stored at -80°C.
Protocol of dsRNA injection and soaking
Adult female V. destructor were removed from capped brood cells along with associated bee larvae. Microinjections were carried out using pulled glass capillary needles in conjunction with a Harvard micro-injector system. Mites were placed on double-sided tape ventral side up, and injected with 20 nl (2.5 μg/μl) of either Vd GST-mu1-dsRNA or LacZ-dsRNA in either the soft tissue proximal to the anal region and postcoxal plate, or in the coxa IV region, as indicated in Figure 7. Needles were left in each mite for 1 - 2 min to reduce the expulsion of fluid from the wound and withdrawn slowly. Mites were left for 1 - 2 min to allow the injection site to "seal" then returned to Petri dishes containing 1 bee larvae per 4 mites. Dead or unhealthy looking mites were removed after 1 hour and mortality was monitored over 72 h in LacZ-dsRNA, Vd GSTmu1-dsRNA and non-injected mites.
To assess non-invasive techniques for dsRNA delivery, mites were either completely immersed in dsRNA or were exposed to a droplet of dsRNA on their ventral carapace. For soaking experiments, adult mites were removed from capped brood cells and placed in 500 μl microfuge tubes containing 20 μl Vd GST-mu1-dsRNA or LacZ-dsRNA (2.5 μg/μl) supplemented with either nothing, 0.9% NaCl, 0.2% Triton-X100 or both. Mites were soaked at 4ºC overnight before being removed, dried and placed in Petri dishes at 27ºC, 95% relative humidity with bee larvae. Alternatively, a sample of mites was exposed to dsRNA by attaching them to double-sided tape and placing a 1 μl drop of Vd GST-mu1-dsRNA or LacZ-dsRNA (2.5 μl/μg) supplemented with either nothing, 0.9% NaCl, 0.2% Triton-X100 or both on the ventral carapace. Mortality was monitored for 48 h prior to collection and validation of knockdown.
Validation of RNAi
To validate RNAi in injected, soaked and droplet-exposed adults, the total RNA was extracted from individual mites 48 h post-treatment. In addition, persistence of the RNAi effect was measured in injected mites by harvesting at 18, 24, 48 and 72 h. Total RNA was extracted from mites using mini-RNA isolation kit II, prior to DNase-treatment and reverse transcription, as described previously. PCR was carried out using either Vd GST-mu1-dsRNA or LacZ-dsRNA treated sample cDNA in conjunction with primers specific for actin or Vd GST-mu1 using primers and cycling conditions, as described above. Products were visualized on an agarose gel normalized to actin loading. To assess persistence and approximate % knockdown, ImageJ software was used to carry out semi-quantitative densitometric analysis on gel images.
Glutathione S-transferase activity in RNAi silenced mites
An enzymatic assay was used to assess the effect of silencing Vd GST-mu1 transcript on GST enzyme activity in dsRNA-treated mites. GST activity was measured by using monochlorobimane (MCB) as the substrate, as previously described for scabies mites  and plant mites , but here we adapted it for cuvettes rather than microplates. MCB forms a stable fluorescent conjugate with reduced glutathione when catalyzed with GST. Mites injected with LacZ-dsRNA or Vd GST-mu1-dsRNA were harvested 48 h post-treatment and frozen at -80ºC. Mites were grouped into pairs and homogenized by milling 2 × 30 s, with a single 3 mm tungsten bead in 175 μl of ice cold 0.05 M Tris buffer pH7.5 in a mechanical tissue disruptor (TissueLyser, Qiagen). After milling, samples were disrupted further in a sonicating water bath for 3 × 20 s with chilling in between bursts then centrifuged at 10,000 g for 5 min at 4ºC. The supernatant was retained, subjected to three freeze thaw-cycles and sonicated for a further 2 × 20 s. Samples were then centrifuged at 10,000 g for 5 min at 4ºC and 150 μl of supernatant from each sample used in the GST assay. Reduced glutathione was dissolved in 0.05 M Tris pH 7.5 and MCB dissolved in methanol and added to each reaction to a final concentration of 3 mM and 0.3 mM, respectively, in a total reaction volume of 200 μl. Mite homogenate and substrate were incubated at 37ºC for 10 min with fluorescence of GSH-bimane adduct measured every 2 s using a F4500 fluorimeter (Hitachi) with emission wavelength at 465 nm and excitation wavelength at 390 nm. All results presented were corrected for non-enzymatic fluorescence obtained from samples substituting 150 μl 0.05 M Tris for the mite material.