Gene silencing by RNA interference in the ectoparasitic mite, Psoroptes ovis

The presence of components of the RNA interference (RNAi) pathway in Psoroptes ovis, an ectoparasitic mite responsible for psoroptic mange, was investigated through interrogation of the P. ovis genome. Homologues of transcripts representing critical elements for achieving effective RNAi in the mite, Tetranychus urticae and the model organisms Caenorhabditis elegans and Drosophila melanogaster were identified and, following the development of a non-invasive immersion method of double stranded RNA delivery, gene silencing by RNAi was successfully demonstrated in P. ovis. Significant reductions in transcript levels were achieved for three target genes which encode the Group 2 allergen (Pso o 2), mu-class glutathione S-transferase (PoGST-mu1) and beta-tubulin (Poβtub). This is the first demonstration of RNAi in P. ovis and provides a mechanism for mining transcriptomic and genomic datasets for novel control targets against this economically important ectoparasite.


Introduction
Psoroptic mange, caused by infestation with the ectoparasitic mite Psoroptes ovis, Hering, 1838, is a disease of significant welfare and economic importance with a global presence. Affecting production animals such as sheep, cattle, bison, water buffalo, camelids and rabbits and wildlife including giraffe and bighorn sheep [1][2][3][4][5], Psoroptes sp. are responsible for diseases that are typically highly contagious and cause intense inflammatory responses accompanied by extreme pruritis, leading to serious welfare concerns [6]. Psoroptes ovis infestation in sheep leads to the disease known as sheep scab and results in economic losses as a result of, for example, costs accrued through prevention and treatment, damage to wool and leather quality, rejection of carcasses at slaughter and fatalities due to hypothermia and epileptiform convulsions [7][8][9][10][11][12][13][14]. Although dipping of sheep in organophosphate acaricides largely remains effective, it is potentially hazardous to the users, traumatic for sheep and can give rise to serious environmental concerns [15,16]. Resistance to some of the compounds used to treat and control P. ovis has been recorded, some of which are also relied upon for the control of gastrointestinal nematodes, thereby increasing the risk of selecting for further resistance [7]. More recently, resistance to the macrocyclic lactone compound, moxidectin, has been confirmed in populations of P. ovis in England and Wales, placing further pressure on the existing methods of control and highlighting the need for novel means of control, such as vaccination [17].
The first draft genome for P. ovis was recently published [18] providing a valuable resource for molecular studies into this ectoparasite. This new genome resource can be exploited for the identification of novel vaccine and diagnostic candidates and also for molecular genetic studies into the development of acaricide resistance. However, these studies will be further enhanced by an effective means of investigating gene function in P. ovis. We therefore interrogated the P. ovis genome to identify components of the RNAi machinery and then, based on the outcomes of that process and previous studies in other closely related astigmatid mites [19,20], we sought to develop a non-invasive means of RNAi in P. ovis. These new tools provide a crucial resource for the identification and validation of novel means of control for this economically important ectoparasite. Marr et al. Vet Res (2018) 49:112 Materials and methods

Identification of P. ovis RNAi pathway genes
The P. ovis genome (DDBJ/ENA/GenBank accession PQWQ00000000) is predicted to encode 12 041 proteincoding genes, of which annotation has been achieved at the gene ontology (GO) level for 7957 genes, with functional annotation of 5217 genes [18]. The genome was interrogated for homologues of known RNAi pathway components identified from the two-spotted spider mite, Tetranychus urticae and the model organisms Caenorhabditis elegans and Drosophila melanogaster. Predicted coding sequences of the major proteins involved in RNAi biosynthesis and processing in these other species were obtained from previously-published analyses [21] and were used to search the predicted coding sequences from the P. ovis genome (as a reference database). A tBLASTx analysis was performed using BLAST ® [22] with each sequence (target and reference) being translated into each of the 6 possible reading frames and a significance threshold of < 1e−10 −4 applied. Phylogenetic clustering of P. ovis argonaute proteins was performed in CLC Genomics Workbench 11.0 (Qiagen Ltd, UK).

Harvesting of P. ovis mites
Mixed populations (mixture of all life stages) of P. ovis were harvested from skin sections taken from experimentally-infested donor sheep, under the legislation of a UK Home Office Project License (reference PPL 60/4238) in accordance with the Animals (Scientific Procedures) Act of 1986. Ethical approval for the maintenance of this model was obtained from the Moredun Research Institute local ethics committee [E03/17] and animals were monitored daily in accordance with UK Home Office guidelines.

Validation of non-invasive RNAi methodology in P. ovis mites
A non-invasive immersion technique for RNAi developed in Varroa destructor [23] was adapted for P. ovis using modifications [20]. Briefly, mites, or their eggs, were immersed in the detached cap of a 1.5 mL microcentrifuge tube in 15 µL of a solution containing either a fluorophore-labelled AllStars ™ Negative Control siRNA (Qiagen, UK) ("fluoro-siRNA") in 0.9% w/v NaCl solution, final fluoro-siRNA concentration 0.05 µg/µL, or 15 µL 0.9% w/v NaCl solution only. The mites were incubated at 4 °C overnight then washed three times in molecular biology grade water (Sigma, UK) before photographic images were captured through an Axiovert 25 CFL inverted fluorescent microscope (Zeiss, Germany) with 10× Achrostigmat magnification lens (Zeiss, Germany) using a D90 AF-5 DX NIKKOR digital camera with 18-105 mm f/3.5-5.6 G ED VR lens kit (Nikon, Japan).

Preparation of gene-specific dsRNA
Double stranded RNA (dsRNA) encoding 300 bp of a P. ovis major allergen (Pso o 2) (Gene ID: psovi17g08010), glutathione-S-transferase mu class 1 (PoGST-mu1) (Gene ID: psovi284g01800), 500 bp beta tubulin (Poβtub) (Gene ID: psovi22g01550) or 319 bp Escherichia coli strain K-12 sub-strain MG1655 lacZ (NC_000913.3) was produced using the dsRNA production vector pL4440 from the Fire Lab C. elegans Vector Kit 1999 (Addgene plasmid #1654). pL4440 vector containing the target gene sequence was produced by ligating the target gene into pL4440 using Sac1 and Sma1 restriction sites (Roche, UK). Primer sequences are available from the authors on request. Double stranded RNA (dsRNA) was synthesised from linearised plasmid using the T7 RiboMAX ™ Express RNAi System (Promega, UK) according to the manufacturer's protocol. The quality of dsRNA was assessed using gel electrophoresis and the concentration determined using an ND-1000 Nanodrop spectrophotometer (Thermo Scientific, UK).

Gene silencing by non-invasive RNAi in P. ovis mites
Twenty adult male P. ovis mites were immersed in 15 µL dsRNA encoding either Pso o 2, PoGST-mu1, Poβtub or lacZ, diluted in 0.9% w/v NaCl solution to a final concentration of 2.5 µg/µL in the detached cap of a 1.5 mL micro-centrifuge tube (Axygen, USA). Five replicates per treatment group were performed for each experiment. The mites were then either: (i) incubated at 4 °C for 24 h and then processed for RNA extraction or, (ii) incubated at 4 °C for 24 h, dried using filter paper (Whatman, UK) and then subjected to a second incubation step (without immersion in dsRNA) in an incubator for 48 h at 25 °C, 75% relative humidity. Following both protocols, RNA was extracted and the quality assessed as previously described [20].

Quantitative PCR (qPCR) assessment of gene silencing in P. ovis
Gene-specific primers were designed for the amplification of the transcripts encoding Pso o 2, PoGST-mu1, Poβtub and E. coli lacZ using Primer3 and were sited outside of the gene region used for the construction of the dsRNA [24]. Primers for Poβactin were designed as described in Burgess et al. [25]. qPCR reactions were performed in triplicate on cDNA samples (diluted 1:3 with molecular grade water (Sigma, UK)) using an ABI Prism 7500 real-time thermal cycler (Applied Biosystems, UK); qPCR cycling times were 50 °C for 2 min then 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s, 57 °C for 40 s, and 72 °C for 40 s. Melt curve analysis cycle times were 95 °C for 15 s, 60 °C for 60 s and 95 °C for 15 s. A single 25 μL PCR reaction consisted of 1 μL cDNA, 12.5 μL SYBR ® GreenER ™ qPCR SuperMix for ABI PRISM ® (Life Technologies, UK), 0.5 μL of each primer (2 μM) and 10.5 μL of molecular biology grade water (Sigma, UK). Standard curves (10 8 -10 2 copies/μL) were prepared from 10 9 copies/μL stock of each plasmid DNA and amplified in triplicate. Correlation co-efficients of the standard curves were used to calculate PCR efficiencies, which averaged > 90%. The average number of copies/μL of cDNA was calculated, and the results were normalised to the reference gene, Poβactin. Melt curve analysis showed single amplicons in all cases and these were validated by Sanger sequencing.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 6 (GraphPad Software, UK). A Student's t-test was used to determine the statistical difference between the means of the two treatment groups in each experiment. Data were assessed for normal distribution using Minitab ® version 17 (Minitab Ltd, UK) and log transformations of raw data were applied as appropriate, p-values < 0.05 were considered significant.

Identification of RNAi pathway genes in the P. ovis genome
Homologous sequences to the major RNAi pathway components (identified from T. urticae, C. elegans and D. melanogaster) were identified in the P. ovis genome, including sequences encoding genes involved in RNA induced silencing complex (RISC) formation, small RNA biosynthesis and secondary amplification of small RNAs (Table 1). Notably absent were the dsRNA uptake and spreading components sid-1, sid-2, rsd-2 and rsd-6, which are known to mediate the uptake of extracellular dsRNA and also the intracellular spreading of dsRNA [26]. In particular, sid-1 has been shown in a number of species, including C. elegans and Bombyx mori, to play a critical role in the demonstration of robust RNAi by mediating cellular uptake of dsRNA in somatic and germ-line cells [27][28][29]. Piwi-clade argonautes were also absent from the P. ovis genome, an observation recently highlighted in the context of the relatively closely related Psoroptidae, the American house dust mite Dermatophagoides farinae [30] (Figure 1). Furthermore, as previously reported for D. farinae, we also identified a similar divergent clade of ago proteins in P. ovis, with five members (PsoAgo8-PsoAgo12, Figure 1). These divergent ago proteins share an unusual DEDD catalytic motif, compared to the more common DEDH slicer motif found in metazoan Ago and Piwi proteins [31]. Orthologs with the same DEDD motif have been identified in S. scabiei, C. elegans and the social spider, Stegodyphus mimosarum [32][33][34]. In addition, a further divergent clade of potential P. ovis-specific ago proteins with six members was identified (PsoAgo1-PsoAgo6, Figure 1) of which, PsoAgo 1, 3, 4 and 6 contain an atypical slicer motif (S[E/K]HE).

Validation of non-invasive RNAi methodology in P. ovis mites
The results of the fluoro-siRNA uptake experiment demonstrated uptake of siRNA in adult female P. ovis mites (Figure 2A). The fluorescence in the fluoro-siRNA immersed mites was most intense at the anterior end of the hysterosoma where the foregut is located (Figure 2A). Limited auto-fluorescence was also detected in the 0.9% w/v NaCl solution-immersed control mites, although this was substantially less intense compared to the fluoro-siRNA immersed mites and more generalised across the idiosoma in its entirety, with no localisation to the gut structure ( Figure 2B). Uptake was visible as increased fluorescence compared to controls from 1 h, increasing until 9 h, after which there was no discernible difference in fluorescence between 9 and 24 h (data not shown). Uptake of siRNA was not achieved by immersion of P. ovis eggs in fluoro-siRNA as there was no difference observed between eggs immersed in fluoro-siRNA (0.05 µg/µL) and control eggs immersed in 0.9% w/v NaCl solution (not shown). In male mites immersed in fluoro-siRNA, fluorescence was also most intense at the anterior-most end of the hysterosoma where the foregut/ anterior midgut is located ( Figure 2C) with limited autofluorescence detected in control male mites ( Figure 2D).

Gene silencing by RNAi in P. ovis mites
Immersion of adult male P. ovis mites for 24 h in dsRNA representing Pso o 2, PoGST-mu1 and Poβtub at 4 °C followed by removal of the dsRNA and a subsequent 48 h period at 25 °C, 75% relative humidity resulted in statistically significantly reduced mRNA transcript levels (p = 0.003, p = 0.0272, p = 0.0002 respectively), compared with mites immersed in control lacZ dsRNA (Figure 3). Mean reductions in transcript levels relative to the lacZ control were 67.7%, 40.5% and 84.2% respectively for Pso o 2, PoGST-mu1 and Poβtub. Immersion in gene-specific dsRNA at 4 °C for 24 h alone was not sufficient to produce significant decreases in transcripts representing Pso o 2 or PoGST-mu1, however, immersion at 4 °C for 24 h alone was sufficient to achieve a significant reduction in Poβtub expression compared to the lacZ control, averaging 61.6% (p = 0.0094). Log transformation of raw data was required for the 24 h immersion of Pso o 2 versus lacZ dsRNA ( Figure 3A) due to unequal variance between treatment groups. No adverse phenotypic effects following gene knockdown were observed in any of the experiments.
and it appears that P. ovis might also express RdRPs, with T. urticae RdRPs blast hits at the amino acid level being highly comparable to the P. ovis RdRP matches (Table 1). This may explain the relatively rapid onset of silencing in P. ovis, with at least one of the targets tested here, with significant gene silencing demonstrable within 24 h, suggesting that spreading or systemic RNAi may occur. Although no research exists on the viral infections challenging P. ovis mites, research in the model fruit fly, Drosophila melanogaster has demonstrated the necessity of dsRNA uptake pathways in order to mount protective immune responses to viral challenge; this may provide an explanation as to why many elements of the RNAi pathway were detected in the P. ovis genome [48]. Although there were some notable absences, for example sid-1 within the category of dsRNA uptake and spreading, rsd-3 was detected in the P. ovis genome (Table 1) and may play a role in systemic RNAi through intracellular vesicular transport of ds/siRNA suggesting P. ovis might be capable of systemic RNAi [26]. A recent study highlighted the absence of piwi clade argonautes and their associated piRNAs within the American house dust mite D. farinae [30]. A homology-based search of the P. ovis and S. scabiei genomes indicated that this loss may be at least partially conserved amongst Psoroptidae mites ( Figure 1). Furthermore, this loss may account for the expansion of species-specific argonautes revealed through phylogenetic evaluation of available argonaute sequences within P. ovis, D. farinae and S. scabiei when compared to T. urticae (Figure 1). Uptake of dsRNA by immersion was demonstrated in both male and female P. ovis (Figure 2). Based on the absence of RNA uptake in eggs, we decided to avoid the use of gravid females for gene silencing experiments, as their eggs could act as an RNAi-inaccessible reservoir for transcripts. Uptake of dsRNA was most probably via ingestion on the basis that P. ovis is an astigmatid mite; thereby having no stigmata and thus relying on ingestion to maintain water homeostasis. Topical application of dsRNA solution to the cuticle of the mesostigmatid mite, Varroa destructor did not result in gene silencing, suggesting that the cuticle is impermeable to dsRNA solution and further supporting the theory of ingestion resulting in uptake of dsRNA. Lipofectamine reagent is commonly used to achieve RNAi in parasitic nematodes such as Haemonchus contortus where it has been demonstrated to improve gene silencing, possibly by increasing uptake of dsRNA within the gut cells due to the positive charge of the liposomes encapsulating the negatively charged dsRNA [49]. The use of lipofectamine did not result in enhancement of the silencing effect in P. ovis across the parameters assessed (data not shown). A similar lack of enhancement was demonstrated recently in D. pteronyssinus [20] although whether this applies to the two step incubator protocol in the present study and/or to further targets remains to be determined. Similarly, the use of phagostimulants was considered during the development of the protocol however existing experimental data [50] indicates common phagostimulants such as adenosine triphosphate (ATP) did not enhance P. ovis feeding, although their effect in RNAi immersion experiments has yet to be determined and could form the basis of future optimisation experiments. The benefits of phagostimulants have been investigated in the context of tick feeding in Ornithodoros tholozani, with maximal intake of food achieved with the use of ATP, reduced nicotinamide adenine dinucleotide (NADH) or (S)-2-aminopentanedioic acid (l-glutatmate) [51]. Glutathione was also shown to moderately increase feeding in O. tholozani but the converse, reduced feeding, was seen in O. moubata [52].
There were target-specific variations in the degree of silencing amongst the three transcripts targeted here and also in the time at which silencing appears to take effect, or at least become significantly reduced compared to the control. These differences may pertain to localisation of the target mRNA transcripts and/or their relative level of expression and rate of transcription. There were no detrimental phenotypic consequences evident following significant gene silencing in any of the three  genes targeted. This result was not entirely unexpected, given that observations were limited to the experimental period (maximum 72 h). Furthermore, the genes targeted (with the possible exception of Poβtub) had no predicted lethal outcome following silencing given the roles they play within the mite in vivo: no specific functional role has yet been assigned, or demonstrated, for Pso o 2, but it is hypothesised to play a key role in establishing infestation by acting as a functional mimic of the toll-like receptor-4 (TLR4) accessory protein, MD-2; as demonstrated for the D. pteronyssinus orthologue, Der p 2 [53]. The phenotypic implications of Pso o 2 gene silencing are therefore likely only to be seen if the mites were returned to the host following successful RNAi. The results relating to PoGST-mu1 in P. ovis are consistent with those detailed previously in D. pteronyssinus, S. scabiei and V. destructor, where no detrimental phenotypic effects were evident following significant gene silencing of Dp/VdGST-mu1 by RNAi [19,20,23]. GSTs are widely implicated in the development of drug resistance and act by conjugating glutathione to toxic endo-and xenobiotics, rendering them more hydrophilic and thus more amenable to excretion [54]. The use of abiotic stressors that might result in an increased need for GST detoxification activity in RNAi experiments were discussed in Marr et al. [20] and are also relevant for P. ovis. Similar approaches have been adopted for other arthropod pests such as the cotton aphid, Aphis gossypii, where RNAi gene silencing of carboxylesterases resulted in reduced resistance to organophosphate-based insecticides [55]. GSTs are a multi-class family of enzymes and several classes have been identified within P. ovis, including delta, mu, alpha and epsilon class GSTs [56]. It is possible that there could be functional redundancy among the P. ovis GSTs, thus avoiding any potential detrimental phenotypic outcome following the successful silencing of one class member.
The use of RNAi in vitro for the identification of promising vaccine candidates or other novel control mechanisms has both potential and limitations (as discussed [43]). For example, here we demonstrated that silencing of the genes encoding Pso o 2 and PoGST-mu1 has no immediate phenotypic effect but the proteins encoded by both of these genes are components of a promising recombinant cocktail vaccine against P. ovis [57]. The future use of RNAi in P. ovis to identify promising vaccine candidates and/or other novel interventions may require in vitro silencing of genes followed by subsequent challenge infestation of host animals to determine the presence of any detrimental phenotypic effects, particularly when genes involved in host:parasite interactions are the targets. Such in vivo trials would require information pertaining to the longevity of gene silencing for each gene assessed and should focus on those genes that are likely to play a role in the initial onset of infestation such as the proteolytic allergen, Pso o 1 and the putative MD-2 mimic, Pso o 2. In these cases, the ability of mites to establish infestation following RNAi gene silencing could be assessed to further characterise a candidate gene's therapeutic potential.
This is the first demonstration of gene silencing by RNAi in P. ovis and paves the way for further RNAi studies in this important ectoparasite. Scaling up and adaptation of the RNAi protocol demonstrated here in order to perform genome-wide, in vitro and in vivo screens is an exciting possibility for future development. Genome-wide RNAi screens have demonstrated great potential in highlighting essential genes [58,59] and also highlighting the relationships with organisms symbionts [60]. This approach could prove to be highly beneficial for discovering novel candidates for control of this important ectoparasitic mite.