Oral Delivery of Double-Stranded RNAs and siRNAs Induces RNAi Effects in the Potato/Tomato Psyllid, Bactericerca cockerelli

The potato/tomato psyllid, Bactericerca cockerelli (B. cockerelli), and the Asian citrus psyllid, Diaphorina citri (D. citri), are very important plant pests, but they are also vectors of phloem-limited bacteria that are associated with two devastating plant diseases. B. cockerelli is the vector of Candidatus Liberibacter psyllaurous (solanacearum), which is associated with zebra chip disease of potatoes, and D. citri is the vector of Ca. Liberibacter asiaticus, which is associated with the Huanglongbing (citrus greening) disease that currently threatens the entire Florida citrus industry. Here we used EST sequence information from D. citri to identify potential targets for RNA interference in B. cockerelli. We targeted ubiquitously expressed and gut-abundant mRNAs via injection and oral acquisition of double-stranded RNAs and siRNAs and were able to induce mortality in recipient psyllids. We also showed knockdown of target mRNAs, and that oral acquisition resulted primarily in mRNA knockdown in the psyllid gut. Concurrent with gene knockdown was the accumulation of target specific ∼ 21 nucleotide siRNAs for an abundant mRNA for BC-Actin. These results showed that RNAi can be a powerful tool for gene function studies in psyllids, and give support for continued efforts for investigating RNAi approaches as possible tools for psyllid and plant disease control.


Introduction
The potato/tomato psyllid, Bactericerca cockerelli (B. cockerelli), is a serious pest of potato and tomato in several regions of the world (http://www.eppo.org/QUARANTINE/Alert_List/bacteria/ Liberibacter_psyllaurous.htm), and is associated with plant diseases including psyllid yellows of tomatoes [1] and zebra chip disease of potatoes [2]. Psyllid yellows is likely associated with feeding damage and toxin production by B. cockerelli, but this psyllid is also a vector of the phloem-limited bacterium, Candidatus Liberibacter psyllaurous (solanacearum), which is believed to be the causal agent of zebra chip disease. Zebra chip is rapidly emerging as a widespread and important disease of potatoes in many potato-growing regions worldwide, and in many ways it is very similar to another plant disease, Huanglongbing (HLB), or Citrus greening. HLB is believed to be caused by a phloem-limited bacterium, Ca. Liberibacter asiaticus which is transmitted from plant-to-plant by the Asian citrus psyllid, Diaphorina citri (D. citri) [3]. HLB is presently the world's most important citrus disease, causing serious losses wherever it occurs. While HLB is relatively new to the U. S., it is now widespread in Florida posing a serious threat to the entire Florida citrus industry [4]. New strategies to control D. citri, B. cockerelli and/or HLB and zebra chip are greatly needed. The biological similarities of HLB to zebra chip, and the phylogenetic conservation of the causal agents of the diseases and their psyllid vectors suggest that success with one insect species, B. cockerelli, will have potential application to the other, D. citri.
RNA interference (RNAi) is a natural gene regulation and antiviral defense system of eukaryotic cells. RNAi functions as a ''gene silencing'' mechanism by targeting specific RNA sequences for degradation (or possibly translational inhibition) resulting in eliminating or reducing expression of the RNA target [5,6]. RNAi has rapidly become an intensely studied research area with great potential for both fundamental and practical biology. RNAi-based strategies for conferring plant resistance to bacterial, nematode and virus-induced plant diseases have been demonstrated [7][8][9], and recently have been studied for a number of insect systems [10], even leading to the suggestion of developing ''insect-proof plants'' [11,12]. Whether or not RNAi strategies could be useful for controlling psyllids (and their associated plant diseases) is currently unknown, but based on results with other insects it is an important question to assess. Furthermore, while B. cockerelli is an important psyllid, it may also serve as a more tractable model for the very important citrus-associated D. citri because the former has a wide host range among solanaceous herbaceous plants (tomato, tobacco and potato).
In the present study, we demonstrated for the first time RNAi effects in B. cockerelli, as manifested by the specific downregulation of endogenous mRNA and the production of ,21 nucleotide siRNAs, which are specific products of the RNAi pathway. Delivery of gene-specific dsRNAs or siRNAs (targeting Actin, ATPase, Hsp70 or CLIC) by microinjection or by a newly developed oral delivery protocol induced RNAi effects in teneral adult B. cockerelli and decreased survival of recipient psyllids. Our oral delivery system will allow the efficient screening of psyllid candidate gene targets for RNAi as more sequence information (for both B. cockerelli and D. citri) becomes available.

Insects and plants
The B. cockerelli used in this study were obtained from Dr. T. D. Paine's laboratory, University of California, Riverside [1]. Psyllids were reared on tomato plants (Solanum lycopersicum Early Pak 7) in mesh cages within growth chambers at 25uC and 70% humidity under 14:10 (light: dark) photoperiod. Teneral adult psyllids were used for all of the experiments.

Psyllid RNA isolation, cDNA synthesis and Northern blot analysis
Total RNAs were extracted from whole B. cockerelli or specific tissues using the RNeasy Mini Kit (Qiagen). To isolate B. cockerelli gut RNAs, adult psyllids were immobilized at 220uC for 10 min, quickly transferred onto an ice box, and dissected with forceps while examining with a dissecting microscope. Approximately 100 individuals were dissected for each experiment and total gut RNA was isolated from pooled guts. Simultaneously, other tissues including head, abdomen, thorax, and reproductive organs were collected for total RNA extraction. cDNA was synthesized from 1 mg total RNAs using SuperScriptH III First-Strand Synthesis System from Invitrogen (catalog NO. 11904-018).
Northern blot hybridization analysis was performed using 1.5 mg of total RNAs of B. cockerelli and D. citri (D. citri total RNAs were kindly provided by Dr. W. O. Dawson, University of Florida). The total RNAs were denatured using glyoxal and separated on 1% agarose gels in HEPES-EDTA buffer, then transferred to nylon membranes (Amersham Hybond TM -NM) using the methods described previously [25]. Hybridization was performed using gene-specific probes in PerfectHyb TM Plus Hybridization Buffer (Sigma, catalog NO. H7033). The pGEMH-T Easy plasmids containing BC-Actin, BC-ATPase, BC-Hsp70 or BC-CLIC sequences were digested with restriction endonucleases PstI, NcoI, SphI or SphI, respectively, and the linearized plasmids were used as templates for the labeling reaction described below. 32 P-UTP-labeled negative strand BC-Actin, BC-ATPase, BC-Hsp70 or BC-CLIC RNA transcripts were generated in vitro using T7, SP6, SP6 or SP6 RNA Polymerase (Ambion, catalog NO. AM1320) respectively and used as probes for hybridization. Hybridizations were performed at 65uC overnight. Blots were washed by shaking in low stringency wash solution [2 X SSC, 0.1% (w/v) SDS] at 25uC for 15 min, followed by medium stringency wash solution [0.5 X SSC, 0.1%(w/v) SDS] at 25uC for 15 min, and finally in high stringency wash solution [0.1 X SSC, 0.1% (w/v) SDS] at 65uC for 15 min. Blots were removed, drained, wrapped in SaranWrap and exposed to X-ray film at 280uC. The exposure times for BC-Actin, BC-ATPase, BC-Hsp70 and BC-CLIC were 15 h, 2 h, 38 h and 38 h, respectively.

dsRNA and siRNA preparation
We used both dsRNAs and small interfering RNAs (siRNAs) for RNAi assays. The dsRNAs were synthesized via in vitro transcription using the MEGAscript RNAi kit (Ambion, catalog NO. AM1626) using PCR products as templates for each target. The T7 primer sequence was placed in front of both the forward and reverse primers which were used for PCR amplification of the template for dsRNA synthesis (Table S1). The dsRNA quality was monitored by agarose gel electrophoresis and the concentration was determined by spectrophotometry. In some cases, dsRNAs were digested with ShortCutH RNase III (New England Biolabs, catalog NO. M0245S) to prepare ,21nt fragments (siRNAs) according to the manufacturer's protocol. Briefly, 10 mg dsRNAs were digested with ShortCut RNase III in a 100 mL reaction at 37uC for 30 min. Ten mL 10 X EDTA (0.5 M) was added to stop the reaction, and the small RNAs were precipitated with ethanol and dissolved in ddH2O.

dsRNA injection
DsRNA injection assays were performed using a micro-manipulator and an insect injection system (http://www.tritechresearch. com/MINJ-PD.html) fitted with a glass capillary needle. Psyllids were immobilized at 220uC for 10 min, transferred into a box that was placed on ice prior to injection with the dsRNAs. Each individual was injected with 200 nL of 100 ng/mL dsRNAs between the 2 nd and 3 rd abdominal segments. After injection, the psyllids were placed onto the leaves of tomato seedlings, caged and maintained in a growth chamber as described above. Mortality was scored daily, and total RNAs from individual live psyllids were extracted at two days post injection and used for cDNA preparation and quantitative real-time PCR.

Feeding assays
Feeding assays were performed using teneral adult psyllids which were allowed to feed on an artificial diet solution consisting of 15% (w:v) sucrose, contained between two layers of stretched parafilm. Feeding assays were performed at room temperature in the dark. In order to visually determine if psyllids fed on test solutions, green food coloring, fluorescent-labeled dsRNAs, or a GFP PCR product was added to the feeding solutions. Green food coloring (McCORMICK & CO) was supplied with 1/10 (v:v) dilution in artificial diet and green abdominal coloration was visualized in psyllids after 24 h feeding. Fluorescent-labeled GFP dsRNAs were prepared using a SilencerH siRNA Labeling Kit -Cy TM 3 (Ambion, catalog NO. AM1632) according to the manufacturer's protocol. Briefly, 10 mg dsRNA was labeled in 50 mL reaction at 37uC for 1 h, the Cy TM 3 -labeled dsRNA was ethanol precipitated and the free Cy TM 3 was removed in the supernatant. The resulting Cy TM 3 -labeled dsRNA was dissolved in 100 mL ddH2O. 50 ng/mL Cy TM 3-labeled dsRNA was supplied in artificial diet, and the fluorescence was visualized by fluorescence microscopy (Leica DM5000B) 48h after the start of feeding. The ability of the psyllids to uptake DNA was investigated by adding a 724 bp GFP PCR product to the artificial diet at a concentration of 50 ng/mL. The 724 bp PCR product corresponding to GFP sequence was amplified from plasmid pJL24 using the primers PacI-GFP left: 59-cggttaattaaatggctagcaaaggagaagaac-39 and JAL34 : 59-tttgcggccgcttatttgtagagctcatccatg-39. The presence of the 724 bp GFP DNA was determined by polymerase chain reaction with the same primers on DNA extracted from the psyllids after they were allowed to feed for 48h.
For the mortality bioassays, dsRNA concentrations were standardized and diluted in the artificial diet and used for feeding assays. Groups of 30 teneral adult psyllids were used for each treatment. Psyllid mortality was scored daily for 4 days. For the mRNA knockdown experiments, the dsRNAs were added to the artificial diet at a concentration of 700 ng/mL and siRNAs were added at a concentration of 100 ng/mL. After the teneral adult psyllids were allowed to feed for 72 h, total RNAs were isolated from individual live psyllids using TRIzol (Invitrogen, catalog NO. 15596-018) and used for cDNA preparation and quantitative real-time PCR. In some cases using BC-Actin and GFP dsRNAs for feeding, total RNAs were isolated from the guts of a pool of ,30 psyllids.

Quantitative real time PCR (qRT-PCR)
The effect of dsRNAs or siRNAs on target mRNA levels was quantified by quantitative real-time PCR using Fast SYBR Green Master Mix (Applied Biosystems). In order to avoid RT-PCR artifacts resulting from the input dsRNAs, primers for qRT-PCR were designed to detect target mRNAs by amplifying sequences that lay outside of the input dsRNA sequences. B. cockerelli rRNA (accession number GQ249868) was used as an internal control for normalization after validation. GFP dsRNA or GFP siRNAs was used as a control treatment. The relative gene expression data were analyzed using the relative 2 -DDCT method [26]. Briefly, qPCR experiments were performed for the target transcript and internal control rRNA in replicate using the primers described in Table S1. The reaction mixture contained 7.5 mL SYBR Green Master Mix, 0.5 mL 10 mM forward primer, 0.5 mL 10 mM reverse primer and 3 mL cDNA template in a final reaction volume of 18 mL. 40 cycles of reaction is performed for 3 s at 95uC and 30 s at 60uC. The DCT value for individual samples was obtained by subtracting average Ct (cycle threshold) value for rRNA from average Ct value for the target transcript. One of the samples in control groups, GFP dsRNA or siRNA-fed psyllids, was selected as a reference sample. The DDCT value for individual samples was obtained by subtracting DCT value for the reference sample from the DCT value of the test sample. The relative abundance of target transcript in comparison to rRNA level in each sample is represented by the 2 -DDCT value. The average value and standard error value (SE value) of the 2 -DDCT value in treatment and control groups were calculated separately, and the statistical analyses were performed between these two groups.

Statistical analysis
Experimental data from dsRNA feeding assays and qPCR results were analyzed by multiple tests including the Least Significant Difference (LSD) test, Ryan-Einot-Gabriel-Welsch (REGW) Multiple Range Test, Tukey's Studentized Range (HSD) Test, and the Bonferroni (Dunn) t Tests using SAS 9.1 software. We used the results from Bonferroni (Dunn) t Tests as it yielded the most conservative results and the statistical differences are shown as different letters or asterisks (*).

Small RNA analysis
RNA hybridization analysis was used to determine if sequencespecific siRNAs were present in psyllids following dsRNA acquisition. Small RNAs were extracted from psyllids after a 72 h feeding period, by using the mirVana PARIS small RNA isolation kit (Ambion, catalog NO. AM1556) according to the manufacturer's protocol. As a positive control, the Tobacco mosaic virus (TMV) vector pJL36 [27], containing BC-Actin, BC-ATPase or GFP sequence was agro-infiltrated into N. benthamiana plants and small RNAs were isolated 2 weeks post-infiltration. Small RNA hybridization was performed according to the method described previously [28]. Briefly, 1 mg small RNAs were separated on denaturing 8M urea 15% polyacrylamide gels and transferred to nylon membranes (Amersham Hybond TM -NX). The pGEMH-T Easy plasmids containing GFP, BC-Actin or BC-ATPase sequences were digested with restriction endonucleasesAvrII, PstI or NcoI, respectively, and the linearized plasmids were used as templates for the labeling reaction described below. 32 P-UTP-labeled negative strand GFP, BC-Actin or BC-ATPase RNA transcripts were generated in vitro using SP6, T7 or SP6 RNA Polymerase (Ambion, catalog NO. AM1320), respectively. Transcripts were fragmented to ,300nt in 120mM Na 2 CO 3 : 80mM NaHCO 3 solution at 65uC for 30 min and used as probes for hybridization. Hybridization was performed in ULTRAhybH-Oligo hybridization buffer (Ambion; catalog NO. AM8663) at 42uC overnight and the membrane was washed twice in NorthernMaxH Low Stringency Wash Buffer (Ambion, catalog NO. AM8673) at 42uC for 15 min. Blots were removed, drained, wrapped in SaranWrap and exposed to X-ray film at 280uC. The exposure times for BC-Actin, BC-ATPase and GFP were 15 h, 19 h and 8 h, respectively. MicroRNA Marker (New England Biolabs; catalog NO. N2102S) was used on the same gel to estimate RNA sizes.

Identification and tissue-specific transcription profile of partial gene sequences of B. cockerelli
When this study was initiated, only 40 nucleotide sequences were available for B. cockerelli. In contrast there were 19,605 EST sequences available for D. citri, including 6,201 sequences from midgut ESTs in the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). We took advantage of these D. citri EST sequences in order to amplify homologous sequences from B. cockerelli. To do this, the D. citri ESTs were aligned and analyzed by using the ArthropodEST EST analyses pipeline (http://bioinformatics.ksu.edu/ArthropodEST/) resulting in the retrieval of 1,904 contigs. 179 contigs from the midgut EST library were annotated and homologous to known protein sequences, among them 99 contigs were ribosomal or mitochondrial cDNAs and the other 80 sequences were further selected as targets of interest. We focused on the sequences that are highly expressed or specifically expressed in the gut, as they are likely to be good targets for orally-acquired anti-psyllid interfering RNAs because the midgut is one of the first barriers that the ingested RNAs will encounter following feeding, and some reports for other insects show that midgut-expressed mRNAs are susceptible to the effects of orally-delivered dsRNAs [10,18,23,29].
The 80 D. citri contig sequences of interest were compared by nucleotide BLAST using the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov) database, and consensus regions for homologous sequences were selected for designing primers (Table S1) for use in cloning the sequences from B. cockerelli. Twenty eight primers were successful yielding amplicons ranging from 150 bp to 700 bp. These cDNAs were cloned into pGEMH-T Easy vector (Promega) and sequenced (UC Davis sequencing facility). The resulting sequences showed 70-90% homology to their D. citri homologs, supporting our hypothesis that these two psyllid species are closely related (data not shown). We also verified expression for 23 of these sequences in B. cockerelli gut tissues by conventional RT-PCR (data not shown). Based on our preliminary results, putative homologues of Actin, ATPase, Hsp70 and CLIC (Chloride Intracellular Channel) were further analyzed. We named these sequences as BC-Actin, BC-ATPase, BC-Hsp70 and BC-CLIC, respectively. The sequences of BC-Actin, BC-ATPase, BC-Hsp70 and BC-CLIC showed 91.7%, 79.8%, 83.4% and 81.5% sequence identity with their corresponding homologs in D. citri (Fig. S1).
To verify the presence of mRNAs for the selected sequences above, Northern hybridization analysis was performed using B. cockerelli sequence-specific probes and total RNAs from both B. cockerelli and D. citri. All probes showed positive hybridization signals for the B. cockerelli RNAs (Fig. 1). Interestingly, three of the four probes also gave hybridization signals for D. citri RNAs, and these were very similar in size to those for B. cockerelli (Fig. 1, panels A, B and D). Hybridization analysis for BC-Actin mRNAs gave a ,1.5 kb RNA for both B. cockerelli and D. citri, and a larger fragment of ,3 kb (Fig. 1A). In the pea aphid (Acyrthosiphon pisum), another phloem-feeding hemipteran insect, the reported actin (GenBank Accession NO. FP925933) mRNA size is 1554 bp, very close to the size of the most abundant RNA detected by us, further suggesting that this in fact represents the actin mRNA. The BC-ATPase probe recognized a RNA of 2-3 kb in B. cockerelli and a slightly larger RNA in D. citri (Fig. 1B). For BC-Hsp70, an 1.5-2 kb RNA was detected in B. cockerelli, but in D. citri the hybridization signal was very faint (Fig. 1C). The BC-CLIC probe hybridized with RNAs from both B. cockerelli and D. citri, but the signal for the latter was less intense (Fig. 1D). The above results clearly indicate the presence of mRNAs of selected sequences in B. cockerelli, and homology to three of the four corresponding D. citri mRNA. We next used semi-quantitative RT-PCR to approximate target mRNA abundance in psyllid tissues. rRNA level was used as internal control. BC-Actin mRNA was abundant in guts and all other tissues tested while BC-ATPase, BC-Hsp70 and BC-CLIC mRNAs were relatively more abundant in psyllid guts than in other tissues including reproductive organs, abdomen, head and thorax (Fig. 2).

Injection of BC-Actin dsRNA caused mortality in B. cockerelli
Direct injection of dsRNAs has proven to be an effective way to demonstrate RNAi effects in numerous insect species [13][14][15][16]. We tested B. cockerelli nymphs and adults, and found the best survival (, 50% 1 day post injection) and ease of injection when using teneral adults and injecting 200 nL dsRNAs between the 2 nd and 3 rd abdominal segments. In two experiments comparing the survival of teneral adult psyllids that were injected with 200 nL of BC-Actin and control GFP dsRNAs, psyllids injected with BC-Actin dsRNA had higher mortality compared with psyllids injected with GFP dsRNA, beginning at two days post injection, and the mortality increased until the end of the experiments. When experiments were concluded six days post injection, psyllids injected with BC-Actin dsRNA showed 8% and 18% survival, whereas the survival was 42% and 57% for those psyllids injected with GFP dsRNA (Table 1).

Feeding assays for B. cockerelli
Delivery of dsRNAs by injection was tedious and caused high mortality. These experiments, however, demonstrated that RNAi effects could be induced in recipient psyllids. In order to utilize an easier and more practical dsRNA delivery approach, we developed an oral dsRNA delivery system for psyllids. The oral delivery system was based on a liquid artificial diet containing 15% (w:v) sucrose sandwiched between two layers of Parafilm. The Parafilm was stretched over a clear plastic vial (25 mm 6 45 mm) in which the psyllids were housed and served as a feeding chamber. The psyllids readily congregated on the parafilm and imbibed on the artificial diet (Fig. 3A). In order to confirm that the psyllids were indeed feeding, three different markers were used. First, a 724 bp GFP PCR product was supplied in the artificial diet. After 48h . Second, when the artificial diet contained green food coloring, green psyllids were easily seen after 24 h of feeding (Fig. 3C, top row). Third, fluorescent-labeled GFP dsRNAs were supplied in the sucrose solution. After 48h of feeding, the Cy TM 3 label (fluorescence) was easily observed in the dsRNA-fed psyllids (Fig. 3D, bottom), and the localization appeared to be restricted in the abdomen.
4. Sequence-specific and dose-dependent B. cockerelli mortality induced by dsRNA feeding We used this feeding system to evaluate dsRNAs of different B. cockerelli sequences for their abilities to induce mortality in B. cockerelli. The dsRNAs were supplied at the concentration of 1000 ng/mL in the artificial diet and mortality was monitored daily for a 4 day feeding period. GFP dsRNA at the same concentration, and feeding solutions of 15% sucrose, or water only were used as controls for each experiment. Psyllids rapidly died when fed only on water, showing 35% mortality after 48 h and up to 97% mortality after 96 h, whereas less than 10% mortality was seen for the diet of only 15% sucrose after 96 h, demonstrating again the efficacy of our feeding system (Fig. 4A). Among the 10 dsRNAs evaluated here, six did not show differences compared to psyllids fed on the GFP dsRNA (Fig. S2, Table S2), whereas four dsRNAs gave higher mortality (Fig. 4). These were dsRNAs for BC-Actin, BC-ATPase, BC-Hsp70 and BC-CLIC.
We also compared dsRNA concentration effects on psyllid mortality for BC-Actin and BC-ATPase. GFP dsRNAs at the same concentrations as the test dsRNAs were used as controls. Significantly more mortality was induced by the BC-Actin and BC-ATPase dsRNAs at 1000 ng/mL and 500 ng/mL compared with GFP dsRNA, but differences were not significant at the lowest concentration tested, 100 ng/mL (Fig. 4B). Furthermore, significantly greater and earlier mortality was seen at 1000 ng/mL than at 500ng/mL for BC-Actin and BC-ATPase dsRNAs, but not for GFP dsRNA (Fig. 4B). It is interesting to note that high concentrations (1000 ng/mL) of the GFP dsRNA control also caused higher mortality compared to the 15% sucrose control (Fig. 4A). Perhaps high concentrations of dsRNAs are able to initiate off-target, or RNAi pathway saturation effects deleterious to psyllids.

Oral feeding of dsRNAs and siRNAs decreased endogenous mRNA levels in B. cockerelli.
In order to determine if the orally-acquired dsRNAs were inducing RNAi effects, we used quantitative real-time PCR to assess target mRNA levels in psyllids. Psyllids were collected 72h after feeding on 700ng/mL dsRNAs and total RNAs were extracted from individual live psyllids. GFP dsRNA-fed psyllids were used as controls. Quantitative real time PCR results were normalized by quantifying the abundance of B. cockerelli rRNA (accession number GQ249868). These analyses revealed that the mRNA abundance of the BC-ATPase mRNA was decreased by 37%-65% after feeding on BC-ATPase dsRNA (Fig. 5A, Table 2). BC-Hsp70 and BC-CLIC mRNAs also showed significant decreases in abundance when whole psyllids were analyzed ( Table 2). These results demonstrated that mRNA knockdown is achievable for guthighly-expressed mRNAs, and can be detected even when analyzing total psyllid RNAs. By contrast, analysis of total psyllid RNAs from the psyllids that fed on BC-Actin dsRNA did not show a reduction for the BC-Actin mRNA (Fig. 5B, Table 2). However, because we had observed ubiquitous expression of BC-Actin mRNA in all tissues we tested (Fig. 2), we asked if the BC-Actin mRNA knockdown in the guts could be masked by the total BC-Actin mRNA expressed in the whole body. Thus, we next compared BC-Actin mRNA abundance in the guts dissected from the psyllids that were fed on BC-Actin and control GFP dsRNAs. These analyses showed that BC-Actin mRNA was decreased by ,70% in gut tissues of BC-Actin dsRNA-fed psyllids (Fig. 5B, Table 2). These results, together with the results from Fig. 3D, suggest that the psyllid gut is accessible to the ingested dsRNA, and therefore gut-expressed or gut-specific mRNAs could be more susceptible RNAi targets.
In plants, dsRNAs are processed into siRNAs by the RNAi machinery [30]. Since our long term goals include evaluating psyllid-targeted RNAi in planta, we next tested whether we could induce RNAi effects by delivering siRNAs. ,21nt siRNAs for BC-ATPase and GFP sequences were generated in vitro (Fig. 6A), and added to the sucrose diet at final concentrations of 100ng/mL siRNA. The psyllids were collected at 72h after feeding and tested as above. Similar to our results for dsRNAs, BC-ATPase siRNAs also induced significant reductions in the target mRNA levels (Fig. 6B, Table 2).
6. Sequence specific siRNA was detected in the psyllids after oral acquisition of BC-Actin dsRNA An important hallmark of RNAi activity is the generation of target-specific ,21nt siRNAs by the RNA silencing machinery within the affected organism [6]. Therefore, we used RNA hybridization to investigate whether target-specific siRNAs were induced in dsRNA-fed psyllids. SiRNA accumulation was analyzed in B. cockerelli fed with sucrose solution containing BC-Actin, BC-ATPase, and GFP dsRNAs at a final concentration of 700 ng/mL. As plants respond systemically to virus infections by producing abundant siRNAs [31], as positive controls, we analyzed N. benthamiana plants infected with Tobacco mosaic virus (pJL36) containing the BC-Actin, BC-ATPase and GFP sequences. When sequence-specific probes were used, ,21nt siRNAs were detected from these virus-infected plants (Fig. 7A, B, C, lanes 1). When dsRNA-fed psyllids were similarly analyzed, resolution of specific siRNAs was not as straightforward as for virus-infected plants. Hybridization signals appeared as a smear for the dsRNA-  fed psyllids (Fig. 7A, B, lanes 2 and 3; 7C, lane 2), but not for the psyllids that fed only on 15% sucrose (Fig. 7A, B, lanes 4; 7C, lane 5). The smears in GFP dsRNA-fed psyllids were eliminated when the psyllids were transferred to tomato seedlings for 2 or more days immediately after feeding on dsRNA solutions (Fig. 7C, Lane 3 and 4), demonstrating that the smears represent sequence nonspecific hybridization with degraded products of ingested dsRNAs, which are thus excreted after feeding on the seedlings. Interestingly, our RNA hybiridization analyses using the probes for BC-Actin did show weak but distinct siRNAs of ,21nt in BC-Actin dsRNA-fed psyllids (Fig. 7A, lane 2). RNA samples from N. benthamiana plants infected with pJL36-GFP did not show small RNAs in the blot using BC-Actin probe, further confirming the sequence specificity of the probes for small RNA targets (Fig. S3).
In contrast to BC-Actin dsRNA-fed psyllids, specific siRNAs were not detected in the psyllids fed on BC-ATPase and GFP dsRNAs (Fig. 7B, lane 2; 7C, lane 2) when the probes for BC-ATPase and GFP, respectively, were used. We also used qRT-PCR to show that the BC-ATPase mRNA is ,10 times less abundant than BC-Actin mRNA even in the gut tissues (Fig. S4). These data ssuggest that the accumulation or stability of siRNAs in dsRNA-fed psyllids may be influenced by the abundance of endogenous mRNA.

Discussion
In this study, we demonstrated the induction of specific RNAi effects in the potato/tomato psyllid (B. cockerelli) by injection and oral acquisition of dsRNA and siRNA inducers. Injection of teneral adult B. cockerelli with dsRNA corresponding to BC-Actin, led to increased mortality relative to control psyllids injected with GFP dsRNA. Injection of dsRNA corresponding to BC-ATPase also resulted in , 30% reduction of the endogenous BC-ATPase mRNA level, verifying target specificity (Table S3). The injection procedure is very laborious and resulted in high mortality of recipient psyllids (,50%). Injection is a powerful but inefficient means to screen large numbers of target sequences for RNAi activity, especially for a small and delicate insect such as the potato/tomato psyllid, but it is commonly used for fundamental studies of RNAi in insects and we included it here as a standard approach.
To overcome the limitations of injection, an artificial feeding system was developed for B. cockerelli. The use of food coloring and Cy TM 3-labeled dsRNA allowed us to confirm that psyllids actively fed on test solutions. We also found that dsRNAs were stable in the 15% sucrose artificial diet (data not shown), as is consistent with   other studies [32]. Using this feeding system, we were able to direct RNAi activity against mRNAs for the homologues of B. cockerelli Actin, ATPase, Hsp70, and CLIC. In our feeding experiments, we could only detect significant RNAi effects at high dsRNA concentrations such as 500 ng/mL or 1000 ng/mL, as is consistent with the reports in other insects such as light brown apple moth (Epiphyas postvittana) and pea aphid (Acyrthosiphon pisum) [18,33]. If high dsRNA concentrations were required for RNAi effects in target insects, this might compromise the practical application of RNAi approaches for insect control. However,  significant knockdown of target mRNA was also detected by us after feeding in vitro-prepared siRNA at lower concentrations (100 ng/mL). It is also interesting to note that a statistically significant down-regulation for the BC-Actin mRNA, a globallyexpressed housekeeping gene, was only detected in psyllid gut tissues but not in whole psyllids. This may suggest that even though robust RNAi activity was induced in guts (presumably in cells where dsRNAs were absorbed), the RNAi effects were not detected in distal tissues, possibly due to the failure of systemic RNAi responses in B. cockerelli. In contrast, our data showed that BC-ATPase, BC-Hsp70 and BC-CLIC are predominantly expressed in psyllid guts, and RNAi effects against their mRNAs were apparent even when total RNAs from whole psyllids were analyzed. To confirm that the mRNA reductions seen in above experiments were due to RNAi, we were able to identify the presence of gene-specific small-interfering RNAs (siRNAs) for the abundant mRNA for BC-Actin. This is consistent with our previous data showing that actin-specific ,21nt siRNAs were induced by transfection of specific dsRNAs in cultured cells of H. vitripennis, another hemipteran insect [28]. For a less abundant mRNA like that for BC-ATPase, we did not detect specific siRNAs. This could be due to the fact that BC-ATPase mRNA is ,10 times less abundant than is the BC-Actin mRNA, even in the gut tissues (Fig.  S4), and the accumulation or the stability of the siRNAs derived from the dsRNA might be affected by the abundance of endogenous mRNA. It should also be noted that the accumulation level of the siRNAs seen here was very low, maybe owing to the observation that the RNA interference effects are restricted to the gut tissues. When the foreign sequence GFP dsRNAs were fed to psyllids, we did not see GFP-specific siRNAs, but a hybridization signal as a smear, likely indicating non-specific degradation of the acquired GFP dsRNA. The accumulation of siRNAs following the introduction of dsRNAs or from virus infection, has been shown in other insect orders including the Lepidopteran species Helicoverpa armigera [23,24], Spodoptera litura [34] and Bombyx mori [35], and in the Dipteran species Aedes aegypyti [36]. Next generation sequencing efforts also confirmed the presence of specific 21nt siRNAs in cultured cells of different mosquito species including Aedes aegypti [36], Culex quinquefasciatus [37] and Anopheles gambiae [38] after virus infections. The hemipteran species Acyrthosiphon pisum possesses homologues of Dicer 1 and Argonaute 2b, both of which are components of the RNA silencing complex in Drosophila, suggesting the existence of the RNAi pathway in hemipteran species [39], and thus our finding of siRNA production and observed RNAi effects are to be expected. The product of the V-ATPase gene transports protons across intracellular and plasma membranes and plays important cellular roles including the regulation of intracellular pH and maintenance of membrane potential in insects [40]. V-ATPase mRNAs have been shown to be susceptible to RNAi effects, even leading to mortality in insects including Acyrthosiphon pisum [21], Diabrotica virgifera virgifera LeConte [22], Leptinotarsa decemlineata [41] and Bemisia tabaci [42]. The target efficiency of the ATPase gene family for RNAi effects is further confirmed in the present study, as we show that both injection and oral delivery of BC-ATPase dsRNA significantly decreased the endogenous BC-ATPase mRNA level. SiRNAs are also reported to be effective inducers of RNAi activity in several insect species [43][44][45]. For example, in the whitefly, Bemisia tabaci, oral ingestion of V-ATPase siRNAs that were generated by RNase III digestions showed lethal effects [42]. In the present study, we show that oral introduction of in vitrogenerated siRNAs suppressed the accumulation of endogenous BC-ATPase mRNA, indicating that knockdown of insect mRNAs can be induced by oral delivery of siRNAs, further supporting the existence of RNAi-mediated silencing pathways in psyllids. Most of the sap-sucking insects including B. cockerelli feed on the plant phloem sap as it is transported within sieve tubes. Small RNAs, including siRNAs and microRNAs, have been reported to be present in the plant phloem [46,47]. Therefore, the efficacy and specificity of orally-delivered siRNAs for insect RNAi will be of a tremendous value for future applications of RNAi via transgenic plants, especially for the phloem feeders like B. cockerelli.
In conclusion, our results demonstrate the induction of RNAi effects in the potato/tomato psyllid (B. cockerelli). This is important and supports the possibility of potentially using RNAi-based strategies for controlling this psyllid, and perhaps even the most-serious disease of citrus, HLB, in which the pathogen (Candidatus Liberibacter asiaticus) is transmitted by the Asian citrus psyllid (D. citri).  Figure S4 Analysis of BC-Actin and BC-ATPase mRNA levels in psyllid guts and whole insect. Quantitative realtime PCR was performed using qRT-PCR primers for BC-Actin, BC-ATPase and rRNA using the cDNA of whole insects and dissected guts of B. cockerelli. pGEMH-T Easy plasmids containing fragments of BC-Actin, BC-ATPase and rRNA served as standards for copy number calculation. The plasmid concentration was converted into copy number and a dilution series of each plasmid with copy number from 10 7 to10 2 was used as the DNA standard for quantitative real-time PCR. Standard curves were drawn by plotting the threshold cycle (CT) against the natural log of the copy number of plasmid molecules. BC-Actin and BC-ATPase mRNA levels in each sample are indicated as copy number per 10,000 rRNA mRNA molecules. (TIF)  Figure S2.