Use of Recombinant Tobacco Mosaic Virus To Achieve RNA Interference in Plants against the Citrus Mealybug, Planococcus citri (Hemiptera: Pseudococcidae)

The citrus mealybug, Planococcus citri, is an important plant pest with a very broad plant host range. P. citri is a phloem feeder and loss of plant vigor and stunting are characteristic symptoms induced on a range of host plants, but P. citri also reduces fruit quality and causes fruit drop leading to significant yield reductions. Better strategies for managing this pest are greatly needed. RNA interference (RNAi) is an emerging tool for functional genomics studies and is being investigated as a practical tool for highly targeted insect control. Here we investigated whether RNAi effects can be induced in P. citri and whether candidate mRNAs could be identified as possible targets for RNAi-based P. citri control. RNAi effects were induced in P. citri, as demonstrated by specific target reductions of P. citri actin, chitin synthase 1 and V-ATPase mRNAs after injection of the corresponding specific double-stranded RNA inducers. We also used recombinant Tobacco mosaic virus (TMV) to express these RNAi effectors in Nicotiana benthamiana plants. We found that P. citri showed lower fecundity and pronounced death of crawlers after feeding on recombinant TMV-infected plants. Taken together, our data show that actin, chitin synthase 1 and V-ATPase mRNAs are potential targets for RNAi against P. citri, and that recombinant TMV is an effective tool for evaluating candidate RNAi effectors in plants.


Introduction
The citrus mealybug, Planococcus citri (Hemiptera: Pseudococcidae), is a serious pest of citrus, nursery plants, grapes and is occasionally considered a serious pest of greenhouse crops [1]. On citrus, P. citri damages plants by feeding on roots, bark, foliage and fruit [2]. During feeding P. citri injects toxic saliva into plants and extracts phloem sap, and induces symptoms including defoliation, fruit discoloration and fruit drop [2]. In addition, P. citri secretes copious amounts of honeydew leading to the development of sooty mold which also leads to reduced fruit quality and lowered plant vigor through the loss of photosynthetic capacity.
RNA interference (RNAi) is a natural gene regulation and antiviral defense found in eukaryotes. RNAi has not only been used to study gene function [3][4][5], but it has shown as a potential strategy for controlling insect pests [6][7][8][9]. Double-stranded (ds) and shortinterfering RNAs (siRNAs) both can induce RNAi effects in insects [10], resulting in degradation of the target mRNA. If the target mRNA is critical for insect development, severe phenotypic effects or even death may result [10,11]. RNAi effects have been achieved for insects by introducing specific ds-and siRNAs into the insects body by microinjection [12,13], by feeding [14][15][16][17][18][19][20] or even by exogenous applications [21]. For plant-feeding insects, using transgenic plants engineered to express dsRNAs corresponding to insect target mRNAs has also shown potential [7,17]. Thus, for practical RNAi application it is critical to identify targets which are vital for the insect and which could successfully be suppressed through RNAi.
RNAi effects have been demonstrated in many insects including fruit flies (Drosophila melanogaster), flour beetles (T. castaneum), pea aphids (Acyrthosiphon pisum), and tobacco hornworms (Manduca sexta) that were all selectively killed when fed species-specific dsRNAs targeting V-ATPase mRNAs [10,18]. Moreover, other studies have shown that using a transient plant-mediated RNAi delivery system for pea aphid (A. pisum) against the rak1 gene that is predominantly expressed in the gut, results in great reduction of the rak1 mRNA levels. This approach was further extended to the C002 gene that is expressed in the salivary glands and has also shown reduction in the mRNA levels of C002 [16]. In another study on potato/tomato psyllids (Bactericerca cockerelli) gut-specific mRNAs were successfully targeted through microinjection and oral acquisition of interfering RNAs [10]. Recently, RNAi effects were induced in the Glassywinged sharpshooter (Homalodisca vitripennis) by microinjection of interfering RNAs corresponding to actin mRNAs, thus reducing the mRNA levels and causing mortality in the injected H. vitripennis [11]. Furthermore, RNAi has been used successfully in transgenic corn plants against Diabrotica virgifera virgifera [7]. These transgenic plants showed a reduction in damage caused by D. virgifera virgifera suggesting that the RNAi pathway can be exploited to control insect pests via in planta expression of interfering RNAs. Earlier studies also showed reductions of CYP6AE14 transcripts in the D. virgifera virgifera midgut after feeding the larvae on plant material expressing dsRNAs specific to CYP6AE14, and this also resulted in a severe retardation of larval growth [17].
By contrast RNAi effects were not successfully achieved for several other insects. In Bombyx mori, injection of dsRNA targeting the ecdysone receptor mRNA or other targets in the epidermis did not result in detectable RNAi effects [22]. Similarly, a series of unsuccessful attempts through feeding dsRNAs were reported for Helicoverpa armigera and Spodoptera frugiperda [22], while two other independent studies showed efficient targeting of mRNAs expressed in the S. frugiperda midgut and brain after feeding dsRNAs [23,24]. Interestingly, while feeding long dsRNAs to H. armigera larvae was not very successful [22], good RNAi effects were achieved using custom-designed siRNAs [25], suggesting different efficiencies for short siRNAs vs. long dsRNAs, or at least their specific sequences, in these species. In general, there seems to be a positive correlation between amount of dsRNA and the degree of RNAi effects [22]. Interestingly, for Bicyclus anynana, Chrysodeixis includes and Spodoptera littoralis [22,26,27] even higher amounts of dsRNA (.1 mg per mg of tissue) did not result in detectable RNAi effects. This variation in responses to RNAi by different species may be due to variation in delivery methods, dose of effector RNAs introduced, RNAi targets or by differences in species with respect to their susceptibility to RNAi [22].
The above studies suggest that RNAi approaches could potentially be used as a tailor-made insecticide for at least some insects, but the correct target, effector RNA, and means of effector delivery must be identified in order to achieve desired RNAi effects. It is also likely that in insects, a continuous supply of effector RNAs may be required. Most insects appear to lack RdRp (RNA dependent RNA Polymerase) [6,[28][29][30] which in plants serves to amplify the RNAi effects thereby producing a continuous supply of effector RNAi molecules [31]. Thus, transgenic plants engineered to express a continuous supply of RNAi effectors are desirable. However, all mRNAs are not equal targets, and assessing potential effector RNAs via transgenic plants is a tedious, time consuming and expensive effort [32]. Recombinant plant virus vectors engineered to express the desired insect effector RNAs in plants provide a potential alternative approach that circumvents many of the above problems seen for transgenic plants [32]. One such virus is Tobacco mosaic virus (TMV) [33]. TMV has a wide plant host range, it is easily engineered, mechanically transmissible from plant to plant, and plants respond quickly to virus infection by producing TMV-specific siRNAs. Then if TMV can be engineered to contain sequences of the desired mRNA target, siRNAs towards this target will be generated. Furthermore, TMV replicates through dsRNA intermediates, and thus plants infected with recombinant TMVs produce at least two forms of RNAi effectors, specific siRNAs, and replicative form dsRNAs.
There is no information available on RNAi activity in P. citri. In this study we first determined if RNAi effects could be induced in P. citri and then we attempted to identify potential mRNA targets that might offer opportunities for P. citri control using RNAi based approaches. We used in vitro synthesized dsRNAs and recombinant-TMV infected plants and demonstrated the effectiveness of P. citri RNAi targets via both of these assays.

Insect rearing
Insects were collected from naturally infested Vitis vinifera plants on the U. C. Davis campus with the permission of greenhouse manager Mr. Steven Silva. P. citri were reared on Nicotiana tabacum and Solanum lycopersicum plants under greenhouse conditions at 24-28uC. Second instar P. citri nymphs were used in feeding and 3 rd instars for injection bioassay studies to evaluate the RNAi effects.
First strand cDNA was synthesized using the superscript cDNA synthesis kit and with an oligo (dT)-primer using 500 ng RNA and following manufacturer's instruction (Invitrogen, Cat. No. 11904-018). Briefly RNA was incubated with oligo (dT) primer at 65uC for 5 min. A 10 ml reaction mixture (10X RT buffer 2 ml, 25 mM MgCl 2 4 ml, 0.1 M DTT 2 ml, RNAse out (40 U/ml) 1 ml and SuperScript II 1 ml) was added and samples were incubated at 42uC for 50 min followed by 10 min incubation at 70uC. The cDNA (200 ng/reaction) was used as template for amplification through PCR using GoTaq Flexi DNA polymerase (Promega Corp., Madison, WI, USA Cat. No. M8291) and universal primers at the PCR conditions given (Table 1). PCR products were analyzed by agarose gel electrophoresis, purified using the ZymoClean PCR cleaning kit (Zymo Corp., Irvine, CA, USA Cat. No. D4006), and subsequently cloned into pGEM-T Easy (Promega Corp., Madison, WI, USA Cat. No. A1360). Sequencing was done by the DNA sequencing facility at UC Davis using M13 forward and reverse primers. The resulting sequences were confirmed by protein translations and BLAST searches (http://www.ncbi.nlm.nih.gov) and genespecific primers were designed. A second round of PCRs was performed to amplify the cDNAs to be used to generate dsRNAs ( Table 2). The PCR products were sequenced and sequences were submitted to GenBank (NCBI) with accession numbers JX443530 and JX443529 of CHS1 and V-ATPase gene fragments, respectively. The actin sequence was submitted to European Nucleotide Archive (EMBL) with accession number HF952554.

DsRNA synthesis and microinjection of P. citri
The RNAi target regions of CHS1 and V-ATPase mRNAs used here were selected using the SnapDragon dsRNA design software (http://www.flyrnai.org/cgi-bin/RNAi_find_primers.pl). Regions with sufficient nucleotide divergence from those of other insect species were selected to avoid potential off-target effects. The T7 promoter sequence was added to the 59 ends of both forward and reverse primers (Table 3), PCR products were amplified from cDNA and purified using the ZymoClean PCR cleaning kit (Zymo Corp., Irvine, CA, USA Cat. No: D4006). Purified PCR products were used as templates to synthesize dsRNA using T7 RNA polymerase and the MEGAscript RNAi Kit as per manufacturer's instructions (Ambion, Austin, TX, USA Cat. No. AM1626). The dsRNAs were treated with DNAse and RNAse at 37uC for 30 min following the manufacturer's protocol, and electrophoresed on an agarose gel to assess dsRNA quality.
To assess if RNAi effects could be induced in P. citri, individual mealybug third instar nymphs were injected with 200 nl of dsRNA by using a micromanipulator and microinjector (http:// www.tritechresearch.com/MINJ-PD.html). For microinjection, insects were placed individually on a 2% agarose gel to immobilize them, and were injected with 200 nl (500 ng/mL) of dsRNA using fine glass needles and the micromanipulator/ miroinjector. Injected insects were moved back to fresh tobacco leaves at 24-26 o C for post-injection observations and procedures. RNA was extracted from injected P. citri at 48 h, 72 h, 96 h and 150 h post-injection and qRT-PCR was performed to assess target mRNA amounts.

Cloning in TMV
Cloned sequences for the mRNA targets of interest (actin, CHS1 and V-ATPase) were excised from pGEM-T Easy by digesting with NotI, and ligated into the NotI-digested TMV-based vector, pJL36 [33]. Positive clones were screened first by restriction digestion and then by nucleotide sequence analysis to confirm the insert sequence orientation. Confirmed clones were transformed into Agrobacterium tumefaciens (GV3101 strain) for inoculation to N. benthamiana plants. Plants were inoculated by agro infiltration with the recombinant TMV-based vector following the protocol as described previously [34]. RT-PCR was performed using insect gene specific primers to check the presence of insect target mRNA in the inoculated plants. The pJL24 TMV-based vector has been engineered to express green fluorescent protein (GFP) was used as a control. Reverse transcriptase polymerase chain reaction (RT-PCR) was used to verify retention of the inserted sequence by the recombinant TMVs in infected plants. RNA was extracted from upper symptomatic leaves of inoculated plants, and primers from the TMV vector sequence flanking the insert were used for RT-PCR (Table 1).

SiRNA detection
Seven days post inoculation total RNAs were extracted from plant leaves using Trizol reagent following the manufacturer's instructions. The small RNA fraction was isolated from total RNA using 50% PEG 8000 precipitation. Small RNAs were electrophoresed on a 15% 8 M urea polyacrylamide gel with microRNA markers (New England BioLabs, Inc, Ipswich, MA, USA Cat. No. N2102S). RNAs were transferred to Hybond-NX Amersham membrane (GE Healthcare, Piscata way, NJ Cat. No. RPN203T) using a semi dry blotter as described [35]. A 32 P-UTP labeled probe was prepared by in vitro transcription using the Maxiscript T7 kit (Ambion, Austin, TX, USA Cat. No. AM1312) as described [11] and the resulting probe was fragmented by adding carbonation buffer (20 mM Na 2 CO 3 and 80mM NaHCO 3 ). Hybridization was done overnight in Ultrahyb-oligo hybridization buffer (Ambion, Austin, TX, USA Cat. No. AM8669) at 42uC followed by two washes in Northern Max Low Stringency Wash Solution (Ambion, Austin, TX, USA Cat. No. AM8673) for 15 minutes at 42uC. The blot was drained and then exposed to X-Ray film (F-BX810, Phenix) at 280uC for 72 hrs.

Insect bioassays and Quantitative Real-time PCR (qRT-PCR) analyses
At seven days post inoculation (dpi), ten 2 nd instar mealybugs were placed on inoculated and upper leaves of N. benthamiana plants. Data were recorded on mortality, insect growth, ovisac production, crawler emergence and their survival. RNA was extracted from individual mealybugs after feeding for 12 days on the respective TMV-infected plants, and four days post dsRNAinjection. cDNA was synthesized using 500 ng total RNA and oligo(dT) primer. Quantitative real time polymerase chain reaction (qRT-PCR) was performed using gene specific primers with the 18S rRNA as the internal control (Table 4)

Statistical Analysis
Experimental data were analyzed to test the difference among treatments at a significance level of 0.05 and 0.01. Analysis of variance (ANOVA) was performed using Statistix 8.1. The homogenous groups are shown with similar letters based on least significant difference (LSD) values. Mortality was corrected using Abbot's formula [36].

Sequence analysis
We successfully amplified and sequenced 507bp of CHS1, 1120bp of V-ATPase and 164bp of actin (ACTB) mRNAs from P. citri. To determine the sequence similarities of the three sequences from P. citri with those from other insect species, we  performed BLAST analysis. The CHS1 sequence showed a maximum (87%) nucleotide identity with that of P. solenopsis (unpublished data) and a minimum (74%) with that of T. castaneum (Table S1 in Tables S1). V-ATPase sequence showed a maximum (83%) identity with that of P. solenopsis (unpublished data) and a minimum (76%) with that of A. albopictus (Table S2in Tables S1). BLAST search results of actin showed a maximum identity of 88% with that of P. solenopsis (unpublished data) and a minimum (76%) with that of Liposcelis entomophila (Table S3 in Tables S1). The BLAST search results validated the fidelity of the sequences amplified in our study.

RNAi induction by microinjection
In our experiments there was a significant (.40%) reduction of the target mRNAs in P. citri injected with effector dsRNAs as compared to insects injected with elution buffer (EB) or the control GFP dsRNA (Fig. 1A, B). Statistical analysis showed significant differences between samples injected with EB (control) and V-ATPase dsRNA, or those injected with GFP dsRNA and V-ATPase dsRNA showing an LSD value of 0.08 (P,0.01). Similarly there were significant differences between samples injected with EB and CHS1 dsRNA, or GFP dsRNA and CHS1 dsRNA showing an LSD value of 0.10 (P,0.01). There was no significant difference between samples injected with EB or GFP dsRNA.

Recombinant TMV inoculated plants contain P. citri RNAs
In order to elucidate whether P. citri effector RNAs were produced in plants after inoculation using the recombinant TMVs, we first confirmed the presence of the actin, CHS1 and V-ATPase RNAs by RT-PCR. These plants tested positive by RT-PCR ( Fig. 2A shows results for CHS1 and V-ATPase), by symptom development and by GFP fluorescence under UV light (for JL36-infected plants). These analyses identified recombinant TMVs containing actin RNA in the sense orientation (TMV-Actin), CHS1 RNA in sense (TMV-CHS1-S) and antisense (TMV-CHS1-AS) forms, and V-ATPase RNA in sense (TMV-V-ATPase-S) and antisense (TMV-V-ATPase-AS) forms. We further tested for siRNAs specific to the CHS1 RNA sequences by performing small RNA blot hybridization. RNAs from plants that were positive by RT-PCR also showed accumulation of the specific siRNAs (Fig. 2B).

Feeding on plants infected with recombinant TMVs induces RNAi in P. citri.
After 12 days of continuous feeding on plants infected with the recombinant TMVs, we collected P. citri for RNA extraction and qRT-PCR analyses. These analyses showed that there was ,10% reduction of the P. citri CHS1 mRNA (Fig. 3A) for P. citri fed on plants infected with the TMV-CHS1-S and a 50% reduction for P. citri fed on plants infected with TMV-CHS1-AS. There was a significant difference among P. citri fed on control plants infected with TMV-GFP, and those infected with TMV-CHS1-AS with an LSD value of 0.04 (P,0.01). Similarly ,30-40% reduction was seen for the V-ATPase mRNA (Fig. 3B) for P. citri fed on plants infected with both TMV-V-ATPase-S and TMV-V-ATPase-AS vs. those fed on plants infected with TMV-GFP, with an LSD value of 0.21 (P,0.01).   (186.6766.67). There was no significant difference among all three control groups with respect to adult mortality, ovisac production, or crawler emergence and mortality in crawlers.
Decrease in mRNA level of CHS1 and V-ATPase correlates with decrease in fecundity and increased mortality of P. citri crawlers We observed that P. citri fed on plants infected with recombinant TMVs expressing the CHS1 and V-ATPase RNAs reached adult stage. The number of P. citri crawlers emerging from ovisacs and their survival when adults fed on plants infected with these recombinant TMVs were greatly reduced and significantly different from that seen for controls (Figs. 6 & 7). We also found that the mortality among P. citri placed on plants at the beginning of the bioassays were significantly lower than the mortality in the crawlers which emerged from the ovisacs of these parents (Fig. 8). Significant differences (P,0.05, LSD: 1.29) were also observed in survival of adult P. citri fed on plants infected with TMV-Actin, TMV-CHS1-S, TMV-CHS1-AS, TMV-V-ATPase-S, and TMV-V-ATPase-AS compared to the control groups. For P. citri fed on plants infected with TMV expressing CHS1 and V-ATPase RNA fragments there was a significant difference (P,0.05, LSD: 1.40) between the control and treated groups for ovisac production  . Relative levels of CHS1 (A) and V-ATPase (B) mRNAs in P. citri after feeding on N. benthamiana plants inoculated with recombinant TMV expressing P. citri CHS1 and V-ATPase fragments. qRT-PCR results of CHS1 (A) and V-ATPase (B) mRNA levels from P. citri after feeding on N. benthamiana plants infected with recombinant TMVs. TMV-GFP (pJL24) was used as a control and sense and antisense inserts of P. citri CHS1/V-ATPase sequences were compared. Total RNA was extracted after 12 days feeding on plants and used for cDNA synthesis. qRT-PCR was performed using primers described in Table 3. The 18S ribosomal RNA was used as an endogenous control in all experiments. Calculation of mRNA levels between control and treated groups was carried out by the comparative CT or 2 -DD Ct method and analyzed by ANOVA using statistix 8.1 software. Numbers with the same letter indicate homogenous groups at p,0.01 (LSD 0.04 and 0.21) for CHS1 and V-ATPase respectively. All experiments were performed twice and data were used collectively. There was no significant difference between all three control groups, and similarly no significant difference between plants infected with TMV expressing sense and antisense V-ATPase sequences, but ovisac production was significantly lower in adults fed on plants infected with TMV-CHS1-S as compared to plants infected with TMV-CHS1-AS. A significant difference (P,0.05,

Discussion
We first demonstrated RNAi effects in P. citri after injection of specific dsRNAs. This approach has been utilized by researchers [37,38] on different insects for screening candidate targets for RNAi, but there is often high variability in terms of insect mortality due to the injection procedures, target RNAs and the target insect. Furthermore, evaluating RNAi effects for practical potential often does not reflect reality as injection is an artificial delivery system. For phloem-feeding hemipterans, such as P. citri, practical delivery is likely to be via the plants upon which the target insects feed. In order to achieve in planta evaluations of potential RNAi effectors, we used the recombinant virus-based vector, TMV. We hypothesized that plant viruses can be effective tools for targeting plant-feeding hemipterans by RNAi as viruses are both targets and powerful inducers of RNAi activity in plants [39]. Recently, Tobacco rattle virus (TRV) expression of antisense fragments from a chewing insect, M. sexta, in Nicotiana attenuata plants specifically silenced three midgut-expressed MsCYPs RNAs when larvae fed on these plants [32], and recombinant TMV was used to induce RNAi effects in Bactericera cockerelli on tomatillo plants [40]. Our data and that of Wuriyanghan and Falk [40] suggest that RNAi effectors can be present in the phloem in sufficient amounts to induce RNAi activity in phloem feeders, and other analyses have shown that siRNAs can be recovered from phloem sap [16].
We used the plant, N. benthamiana, due to its susceptibility to both the TMV and P. citri. We first confirmed by RT-PCR that the P. citri mRNA sequences were retained by recombinant TMV in infected plants and we used small RNA hybridization to demonstrate the presence of specific P. citri siRNAs in recombinant virus-infected plants. When P. citri were fed on N. benthamiana plants infected with recombinant TMV, we were able to detect specific RNAi effects including target mRNA reductions, P. citri mortality and reduced fecundity and hampered growth. Interestingly, the RNAi effects were much more prominent in the emerging nymphs/crawlers than in the adults which were initially placed to the TMV-inoculated plants. Wuriyanghan and Falk [40] also noted phenotypic effects on nymphs of the phloem-feeding B. cockerelli. Our combined data suggest that these effects may be due to the fact that the developing nymphs feed more voraciously than the older adults, hence can take up greater amounts of RNAi inducers as compared to adults, or they may be better RNAi targets.
Our qRT-PCR data showed significant reductions of the target mRNAs. Both dsRNA injection experiments and when P. citri were fed recombinant TMV-infected plants showed reductions in the corresponding mRNA target. By contrast, injection of GFP dsRNAs, or when P. citri fed on TMV-GFP-infected plants target mRNA reductions were not reduced. We cannot be sure that some non-target mRNAs were not also reduced (off target effects), but as the mRNAs evaluated here were not affected by GFP effectors and as we designed our sequences carefully using ''SnapDragon'' software (http://www.flyrnai.org/cgi-bin/RNAi_find_primers.pl), our data suggest that target mRNA reductions seen here were due to RNAi.
We also observed high mortality in adult mealybugs on plants infected with TMV-Actin and significant mortality (P,0.05) also on plants infected with TMV-CHS1 and TMV-V-ATPase. Interestingly we observed high mortality of 71.2264.22 in the emerging P. citri crawlers on plants infected with TMV-V-ATPase. Crawler emergence and mortality were significantly different between samples fed on control group plants and TMV-V-ATPase infected plants. For P. citri fed on plants infected with TMV-CHS1, we found significant differences for plants infected with recombinant TMV having CHS1 inserts in the sense (TMV-CHS1-S) vs. antisense (TMV-CHS1-AS) orientations with respect to crawler emergence and mortality. Similar trends were also observed in qRT-PCR results where there was a significant difference in sense and antisense only for CHS1 constructs.
Previous studies have shown successful RNAi effects in insects through artificial feeding of siRNAs/dsRNAs, and through transient or stable transformation of plants expressing interfering RNAs [15][16][17]19,37,[41][42][43]. Our studies here provide another direction for rapid, yet functional in planta RNAi studies. TMV is easily manipulated, has a wide plant host range and gives fairly quick effects in plants. This approach provides a robust method to study the RNAi effects towards plant feeding insects, even phloem feeders, and the dramatic effects seen here by us particularly on nymphal/carwler mealybugs gives strong support for further efforts in RNAi-based insect control.