Silencing of Aphid Genes by dsRNA Feeding from Plants

Background RNA interference (RNAi) is a valuable reverse genetics tool to study gene function in various organisms, including hemipteran insects such as aphids. Previous work has shown that RNAi-mediated knockdown of pea aphid (Acyrthosiphon pisum) genes can be achieved through direct injection of double-stranded RNA (dsRNA) or small-interfering RNAs (siRNA) into the pea aphid hemolymph or by feeding these insects on artificial diets containing the small RNAs. Methodology/Principal Findings In this study, we have developed the plant-mediated RNAi technology for aphids to allow for gene silencing in the aphid natural environment and minimize handling of these insects during experiments. The green peach aphid M. persicae was selected because it has a broad plant host range that includes the model plants Nicotiana benthamiana and Arabidopsis thaliana for which transgenic materials can relatively quickly be generated. We targeted M. persicae Rack1, which is predominantly expressed in the gut, and M. persicae C002 (MpC002), which is predominantly expressed in the salivary glands. The aphids were fed on N. benthamiana leaf disks transiently producing dsRNA corresponding to these genes and on A. thaliana plants stably producing the dsRNAs. MpC002 and Rack-1 expression were knocked down by up to 60% on transgenic N. benthamiana and A. thaliana. Moreover, silenced M. persicae produced less progeny consistent with these genes having essential functions. Conclusions/Significance Similar levels of gene silencing were achieved in our plant-mediated RNAi approach and published silencing methods for aphids. Furthermore, the N. benthamiana leaf disk assay can be developed into a screen to assess which genes are essential for aphid survival on plants. Our results also demonstrate the feasibility of the plant-mediated RNAi approach for aphid control.


Introduction
RNA interference (RNAi) is a valuable reverse genetics tool to study gene function in various organisms [1]. The process of RNAi was described as 'post-transcriptional gene silencing' (PTGS) in plant systems [2] and is a technique well established in numerous eukaryotic systems across kingdoms, e.g. Caenorhabditis elegans [3], Arabidopsis thaliana [4] and Drosophila melanogaster [5].
With the RNAi method, double-stranded RNA (dsRNA) can specifically lower the transcript abundance of a target gene when injected into an organism or introduced into cultured cells [3]. RNAi involves the cleavage of dsRNA precursors into smallinterfering RNA (siRNA) of approximately 21 to 23 nucleotides by the enzyme Dicer [6]. These siRNAs are then incorporated into a RNA-induced silencing complex (RISC). Argonaute proteins, the catalytic components of RISC, use the siRNA as a template to recognize and degrade the complementary messenger RNA (mRNA) [6]. RNAi can therefore be exploited to suppress gene expression through highly specific depletion of target transcripts.
Aphids are sap-sucking insects of the order Hemiptera and are important crop pests in terms of direct feeding damage and also transmission of plant viruses [7]. RNAi has been successfully used to investigate gene function in the pea aphid Acyrthosiphon pisum, a relatively large aphid that can be injected with dsRNA. Nonetheless, the A. pisum host range is predominantly restricted to leguminous species. On the other hand, the green peach aphid Myzus persicae can feed on over 40 different plant families [8] and is capable of efficiently transmitting over 100 types of plant viruses [9]. Hence, M. persicae is one of the most important aphid pests in agricultural crops. However, RNAi has not previously been documented in this species.
RNAi-mediated gene knockdown can be achieved in aphids through direct injection of dsRNA or small-interfering RNAs (siRNA) into aphid hemolymph [10,11]. This approach was used to silence C002, a gene strongly expressed in the salivary glands of A. pisum [10]. Silencing the gene resulted in lethality of the aphids on plants, but not on artificial diet, indicating that C002 has a function in aphid interaction with the plant host [10,12]. We identified the homologue of C002 from M. persicae and named it MpC002 [13]. MpC002 is predominantly expressed in the M. persicae salivary glands and transient over-expression of MpC002 in Nicotiana benthamiana improved M. persicae fecundity [13]. Microinjection of long dsRNA into A. pisum also leads to silencing of genes encoding calreticulin and cathepsin by 30-40% [11]. Calreticulin is a calcium-binding protein that is produced in most aphid tissues, while cathepsin is specifically produced in the pea aphid gut. Thus, gene silencing appears to occur in different aphid tissues [11].
Aphids can be fed on artificial diet, which is sandwiched between thin parafilm membranes. A. pisum fed on an artificial diet containing dsRNA corresponding to the aquaporin transcript lead to downregulation by more than 2-fold within 24 hours [14]. Since aquaporin is involved in osmoregulation, this resulted in elevated osmotic pressure in the hemolymph [14]. Feeding of dsRNA targeting vATPase transcripts from an artificial diet achieved a 30% decrease in transcript levels in A. pisum and a significant increase in aphid mortality [15].
Both micro-injection and artificial diets are valuable methods for achieving RNAi in aphids. However, dsRNA/siRNA has to be synthesized in both cases and neither treatment is natural for aphids. As RNAi in aphids is indeed feasible, it has the potential to be expanded to include gene knockdown via the delivery of dsRNA from plants (plant-mediated RNAi). This method could allow for studying aphid gene function in the aphid natural habitat and may be useful for control aphid pests in crop production. The plant-mediated RNAi method effectively silences genes of lepidopteran and coleopteran insect species [16,17] and the brown planthopper, a hemipteran species [18]. However, these insects are larger than aphids and hence consume more plant tissue/sap while feeding. Our goal was to determine if the plantmediated RNAi approach also silences aphid genes. The green peach aphid M. persicae was selected because it has a broad plant host range, including the model plants N. benthamiana and Arabidopsis thaliana for which transgenic materials can relatively quickly be generated. Furthermore, transgenes can be rapidly expressed in N. benthamiana leaves using Agrobacterium-mediated transient expression providing the possibility to develop a highthroughput system to assess which genes in the aphid genome are essential for survival of aphids on plant hosts. To test the plantmediated RNAi approach, we selected two M. persicae genes, MpC002 and Receptor of Activated Kinase C (Rack-1) as targets. As discussed above, MpC002 is predominantly expressed in the aphid salivary gland. In contrast, Rack-1 is predominantly expressed in the aphid gut.
Rack1 is an intracellular receptor that binds activated protein kinase C (PKC), an enzyme primarily involved in signal transduction cascades [19]. Rack-1 is conserved amongst plants and animals and is an essential multifunctional scaffold protein which physically connects diverse signal transduction components into stable complexes [20]. Rack-1 binds to integrins [21], has a function in actin organisation [22] and is an integral component of the mammalian circadian clock [23]. Rack-1 from M. persicae was identified as a luteovirus-binding protein [19] as it was found to bind in vitro to purified wild type or mutant particles of Beet Mild Yellows Virus (BMYV). Rack-1 is a good candidate for RNAi in aphids as Rack-1 knockdown has been demonstrated in the nematodes Caenorhabditis elegans [24,25] and Heterorhabditis bacteriophora, [26]. Knockdown of Rack-1 resulted in developmentally defective phenotypes in C. elegans including slow growth, embryonic lethality, egg laying defectiveness and sluggishness [24,25] as well as sterility and abnormal gonad development [26].
Rack-1 in Drosophila functions during oogenesis [27] and is required in early oocyte polarity [28].
We found that the expression of both MpC002 and Rack-1 is knocked down when M. persicae are fed from transgenic plants that transiently (N. benthamiana) and stably (A. thaliana) express dsRNA corresponding to MpC002 and Rack-1. Moreover, silenced aphids have reduced progeny production. Thus, plant-mediated RNAi is feasible, and is a useful tool for studying aphid gene function.

Results
Expression profiles of RNAi target genes C002 and MpC002 are predominantly expressed in the salivary glands of A. pisum and M. persicae [10,12,13], and Rack-1 in aphid gut tissues [19]. To verify this in our colony of M. persicae, RT-PCR was performed on total RNA extracted from different aphid tissues. MpC002 transcripts were detected in M. persicae heads and salivary glands, at relatively low abundance in whole aphids but not in dissected aphid guts ( Figure 1). Conversely, Rack-1 transcripts were found in all aphid body parts and at highest abundance in the gut ( Figure 1). These results confirmed previous findings and provided RNAi targets predominantly expressed in the aphid salivary glands and gut.
Detection of MpC002 and Rack-1 siRNAs in N. benthamiana leaves First, we investigated if dsRNAs corresponding to M. persicae MpC002 (dsMpC002) and Rack-1 (dsRack-1) were produced and processed into siRNAs in N. benthamiana leaves. The entire MpC002 transcript without the region corresponding to the signal peptide (710 bp), a fragment corresponding to the 59 coding region of the Rack-1 transcript (309 bp) and a fragment corresponding to the majority of the open reading frame (537 bp) of the green fluorescent protein (GFP) were cloned into the pJawohl8-RNAi plasmid, which expresses the cloned fragments as inverted repeats under control of a double CaMV 35S promoter to produce dsRNAs (I.E. Sommsich, see acknowledgments). Double-stranded GFP (dsGFP) was used as a control for the dsRNA treatments as opposed to empty vector in order to assess whether the presence of dsRNA itself would induce some effect in plant response to aphids. The pJawohl8-RNAi constructs were transiently expressed by Agrobacterium-mediated infiltration (agro-infiltration) of N. benthamiana leaves. MpC002 and Rack-1 siRNAs were observed starting 2 days post agro-infiltration ( Figure 2). This indicated that the MpC002 and Rack-1 dsRNAs are being processed into 21 to 23 nucleotide siRNAs in N. benthamiana leaves. The agro-infiltrated leaves did not show obvious phenotypes such as chlorosis or leaf curling/crinkling upon agro-infiltration of the pJawohl8-RNAi constructs.
Silencing of M. persicae MpC002 and Rack-1 genes by feeding from transgenic N. benthamiana leaves Next we investigated if MpC002 and Rack-1 are down-regulated in M. persicae after feeding on N. benthamiana leaves transiently producing the MpC002 and Rack-1 RNAs. At one-day post agroinfiltration, 11-mm diameter leaf discs of the infiltrated leaves were placed on top of water agar in wells of 24-well titre plates and exposed to aphids as previously described [13]. Nymphs born on the leaf discs were transferred every 6 days to newly agroinfiltrated leaf discs to ensure continuous exposure of the aphids to the MpC002 and Rack-1 RNAs (Figure 2). At 17 days, the adult aphids were collected to assess MpC002 and Rack-1 expression levels by quantitative RT-PCR (qRT-PCR). Aphids fed for 17 days on N. benthamiana leaf discs infiltrated with dsGFP pJawohl8-RNAi constructs were used as controls. The expression levels of MpC002 and Rack-1 were reduced by an average 30-40% compared to the controls ( Figure 3A). This downregulation was consistent and highly significant among three biological replicates for MpC002 (Student's t-test, n = 3, p-value = 0.013) and Rack-1 (Student's t-test, n = 3, p-value = 0.012).

Silencing of aphid MpC002 and Rack-1 on stable transgenic Arabidopsis lines
We also investigated the downregulation of M. persicae genes MpC002 and Rack-1 upon feeding on stable transgenic A. thaliana plants. The transgenic lines were obtained by floral-dip transformation of Col-0 plants with the MpC002, Rack-1 and GFP pJawohl8-RNAi constructs used in the N. benthamiana transient assays. Three independent F3 homozygous dsMpC002 and dsRack-1 transgenic A. thaliana were generated. One F3 homozygous dsGFP transgenic Arabidopsis line was included as control. All lines contained the transgenes as confirmed by PCR and sequencing. Northern blot analysis of the transgenic Arabidopsis lines revealed the presence of siRNA for MpC002 and Rack-1 ( Figure 4). The siRNAs corresponding to M. persicae MpC002 were equally abundant in the three independent transgenic lines ( Figure 4A), while the siRNAs corresponding to Rack-1 were abundant in line 1, less abundant in line 3 and not detected in line 4 ( Figure 4B).
To investigate down-regulation of M. persicae MpC002 and Rack-1 on the stable transgenic lines, nymphs born on the transgenic plants were kept on these plants for 16 days at which time the adult aphids were collected for RNA extraction and qRT-PCRs. The aphids reared on three independent dsMpC002 lines showed an approximate 60% decrease in MpC002 expression compared to aphids reared on dsGFP ( Figure 5A). Furthermore, downregulation of Rack-1 by approximately 50% was demonstrated for aphids reared on dsRack-1 line 1 compared to dsGFP but not for aphids fed on dsRack-1 lines 3 and 4 ( Figure 5A). MpC002 down-regulation on the three independent lines was consistent in three replicates (Student's t-test, n = 3, p,0.05). Rack-1 was also consistently down-regulated on dsRack-1 line 1 among three replicates (Student's t-test, n = 3, p = 0.023), while Rack-1 was not significantly down-regulated on dsRack-1 lines 3 and 4 (Student's t-test, n = 3, p.0.05). These results are in agreement with the dsMpC002 and dsRack-1 expression levels in the transgenic lines in which the expression of the aphid Rack-1 gene was not downregulated on transgenic lines that have low levels of siRNAs corresponding to Rack-1 ( Figure 4B).
Silencing of MpC002 and Rack-1 reduces M. persicae fecundity It was previously shown that silencing of C002 by injection of dsRNAs in the pea aphid increased the lethality of these aphids on plants [10,12]. Hence, we assessed if silencing of MpC002 also affected survival of M. persicae feeding directly on N. benthamiana and A. thaliana. Nymphs exposed to the N. benthamiana leaf discs for 17 days became adults and started to produce their own nymphs after approximately 10 days. The overall survival of the aphids and the production of nymphs on leaf discs transiently producing dsMpC002 were not affected compared to aphids on leaf discs producing dsGFP ( Figure S1A). However, the nymph production by these aphids was significantly lower in six biological replicates (ANOVA, n = 4-6, p,0.05) ( Figure 3B). Similarly, on transgenic Arabidopsis plants the MpC002-silenced aphids survived equally well, but produced less nymphs in three biological replicates (GLM, n = 4, p,0.05) (Figure S1B, Figure 5B).

Discussion
We have shown that it is possible to down-regulate M. persicae gene expression by feeding the aphids dsRNA from plants. As far as we are aware, this is the first example of RNAi in an aphid system from direct plant feeding. We also show that RNAi is possible in M. persicae, as RNAi was shown previously in A. pisum only.
We measured a 30-60% decrease in gene expression, similar to that observed in microinjection and artificial feeding of small Northern blots were hybridized with probes prepared from MpC002 (A) or Rack-1 (B) PCR products. Total RNAs isolated from leaves 6 dpi with GFP pJawohl8-RNAi constructs were included to control for specific hybridization of the MpC002 and Rack-1 probes (lanes indicated with dsGFP). To control for equal RNA loading, blots were stripped and then hybridized with an snRNA probe corresponding to U6, which is constitutively produced in plants [45]. doi:10.1371/journal.pone.0025709.g002 RNAs to aphids. The reduction is also similar to that measured in other insects such as Schistocerca americana (injection) [29] and Rhodnius prolixus (injection and ingestion) [30] but overall lower than the levels found in Spodoptera litura (injection) [31] or in Drosophila melanogaster (injection) [32]. Our method allows the study of gene function during interactions of aphids with plants, which is not possible by feeding of dsRNA and siRNA from diets [14,15].
Previous studies have demonstrated the silencing signal to be mobile in plants [33], where expressed small RNAs to move within the phloem to where aphids feed. The CaMV 35S promoter enables constitutive expression of dsRNA in transgenic plants tissue, including the leaf phloem [34]. The CaMV 35S promoter also allows for transient expression and movement of dsRNAs in N. benthamiana phloem [35]. Our results demonstrate that siRNAs can travel from the plant phloem through the aphid stylet and reach the aphid intestinal tissues triggering the silencing of aphid target genes. Given that MpC002 expression is knocked down by up to 60% and is predominantly expressed in the salivary glands, the silencing signal appears to spread through the aphid. This is consistent with the finding that small RNA pathways that are highly conserved in animals are also present in aphids [36,37,38] Knockdown of Rack-1 and MpC002 reduced aphid fecundity ( Figure 3B, Figure 5B) but not survival ( Figure S1). This contrasts with the results obtained by dsRNA injection of A. pisum in which survival was reduced by silencing C002. It is possible that the lower A. pisum survival is caused by faster downregulation of the target gene as a result of the sudden higher presence of the injected dsRNA in the hemolymph. Alternatively, stress caused by the injection could exacerbate the negative impact of C002 downregulation. M. persicae are smaller than pea aphids and hence more difficult to inject without affecting aphid survival rates. Delivery by plant feeding therefore provides a gentle, natural method for studying gene function that is less likely to have indirect effects on aphid behaviour. Our method is therefore suited to investigating the effects of gene silencing on aphid/plant interactions, and for virus-transmission studies.
M. persicae produces more progeny on N. benthamiana leaves that transiently express MpC002 [13]. Thus, the presence of more (in planta overexpression) and less (RNAi in aphids) MpC002 leads to, respectively, increased and reduced M. persicae performance on plants. In addition, silencing of A. pisum C002 decreases survival  Total RNA was isolated from two-week old seedlings of F3 homozygous stable dsMpC002 (A) and dsRack-1 (B) transgenic lines. Total RNA isolated from two-week old seedlings of a F3 homozygous stable dsGFP line was included to control for specific hybridization (lanes indicated with dsGFP). Each lane contains 15-20 mg of total RNA. Northern blots were hybridized with probes prepared from MpC002 (A) or Rack-1 (B) PCR products. To verify equal RNA loading, blots were stripped and then hybridized with an snRNA probe corresponding to U6, which is constitutively produced in plants [45]. doi:10.1371/journal.pone.0025709.g004 of this aphid on plants but not on diet and the C002 protein was detected in plants upon A. pisum feeding [10]. Finally, C002 was found in the saliva proteomes of M. persicae [39] and A. pisum [40]. Altogether, this indicates that the C002 genes of both M. persicae and A. pisum have essential functions in aphid-plant interactions.
Our finding that silencing of Rack-1 in M. persicae leads to decreased progeny production by this aphid is also in agreement with other findings. Indeed, Rack-1 is a scaffold protein that is involved in the regulation of cell proliferation, growth and movement in animals [20,21,22]. Silencing of Rack-1 in two species of nematodes, C. elegans and H. bacteriophora, reduces growth of these animals [24,25,26]. M. persicae Rack-1 also interacts with integrins and luteoviruses [19], which invade aphid gut cells [41], suggesting a role in endocytosis processes, such as nutrient/peptide uptake from the gut lumen. Given that Rack-1 is expressed in multiple tissues of the aphid and particularly in the gut, silencing this gene may affect aphid progeny reproduction indirectly, perhaps by reducing the growth of gut cells leading to decreased nutrient uptake. Alternatively, silencing may directly reduce the growth of embryo cells.
The M. persicae genome is being sequenced, but the functions of the majority of aphid genes are still unknown. Moreover, it is not fully understood how aphids modulate host defenses and mediate the transmission of plant viruses. The N. benthamiana leaf disc assay can be developed into a functional genomics screen to assess which aphid genes are essential for aphid survival on plants in the absence or presence of specific plant metabolites or synthetic pesticides. It is also possible to further investigate the role of aphid candidate effector proteins in plant infestation [13]. Finally, we can use plant-mediated RNAi to identify aphid proteins involved in the non-persistent and persistent transmission of plant viruses.

Insect rearing
The aphid lineage used in this study is Myzus persicae, lineage of RRes (genotype O) [13]. M. persicae were reared on Nicotiana tabacum plants for Nicotiana benthamiana leaf disc assays and on Chinese cabbage (Brassica rapa) for the fecundity assays on Arabidopsis thaliana. The insects were maintained in custom-built acrylic cages located in controlled environment conditions at 18uC under 16 hours of light.

Cloning
Total RNA was extracted using the TRIzol Reagent (Invitrogen, Paisley, UK) and the synthesis of cDNA was performed with poly-T primers using the M-MLV reverse transcriptase system (Promega, Southampton, UK) according to the manufacturer's instructions. MpC002 and Rack-1 coding sequences were amplified from M. persicae cDNA by PCR with specific primers containing additional attb1 and attb2 linkers (Table S1) for cloning with gateway system (Invitrogen). The Myzus persicae EST dataset was mined for the transcript sequences of both target genes [42]. A 710-bp MpC002 fragment corresponding to the entire mature MpC002 protein without the signal peptide, a 309-bp Rack-1 fragment starting at nucleotide position +49 (GGGTTAC) and ending at nucleotide position +358 (CGTCAAA) of the Rack1 transcript sequence, and a 537-bp GFP fragment starting at nucleotide position +29 (GAGTGG) and ending at nucleotide position +566 (…TTAGCAG) of the GFP open reading frame were introduced into pDONR TM 207 (Invitrogen) plasmid using Gateway BP reaction and transformed into DH5a. Subsequent clones were sequenced to verify correct size and sequence of inserts. Subsequently, the inserts were introduced into the pJawohl8-RNAi binary silencing vector (kindly provided by I.E. Somssich, Max Planck Institute for Plant Breeding Research, Germany) using Gateway LB reaction generating plasmids pJMpC002, pJRack-1 and pJGFP, which were introduced into A. tumefaciens strain GV3101 containing pMP90RK plasmid and used for transient assays in N. benthamiana leaves and transformation of A. thaliana.

N. benthamiana leaf infiltration and leaf disc assays
Single Agrobacterium colonies harboring pJMpC002, pJRack-1 or pJGFP were inoculated into Luria Broth (LB) containing 25 mg/l Kanamicin, 25 mg/l Gentamicin, 50 mg/l Rifampicin and 25 mg/l Carbenicillin and grown (28uC at 225 rpm) until an Optical Density (OD 600 nm ) of 0.3 was reached (EppendorfH BioPhotometer TM , Eppendorf, Cambridge, UK). Cultures were resuspended in infiltration medium (10 mM MgCl 2 , 10 mM MES 2-(N-morpholino)ethanesulfonic acid, pH 5.6) with 150 mM Acetosyringone to initiate expression. Each construct was infiltrated into the youngest fully expanded leaves of 4-6-week old N. benthamiana plants. The plants were grown in a growth chamber with daily temperatures ranging between 22u-25uC under a short day regime. One day after infiltration, leaves were harvested and used in leaf disc assays. The leaf discs were cut from the infiltrated areas using an 11 mm diameter borer and placed in single wells of a 24-well plate on top of a plug consisting of 1 ml solidified 1% distilled water agar (DWA). Four 1 st instar nymphs (1-2 days old) reared on N. tabacum were places onto the leaf discs for a total of 6 leaves per construct. The wells were individually sealed with mesh and put upside down in controlled environment conditions at temperature 18uC under 16 hours of light. The 24-well plate was replaced with freshly infiltrated (one day post infiltration) leaf discs after 6 and 12 days. Aphid survival by counting was assessed at 6, 12, 14 and 17 days after the day of transfer of aphids to the first 24-well plate and the numbers of nymphs produced by these aphids at 12, 14 and 17 days were also counted. The nymphs were removed after counting. This experiment was repeated 6 times to generate 6 independent biological replicates each containing 6 leaf discs per construct.

Generation of transgenic plants
The pJMpC002, pJRack-1 or pJGFP constructs were transformed into A. thaliana ecotype Col-0 using the floral dip method (Bechtold et al., 1993). Seeds were sown and seedlings were sprayed with phosphinothricin (BASTA) to select for transformants. F2 seeds were germinated on Murashige and Skoog (MS) medium supplemented with 20 mg ml BASTA for selection. Plant ratio of 3:1 dead/alive (evidence of single insertion) segregation, were taken forward to the F3. Seed from F3 were sown on MS+BASTA and lines with 100% survival ratio (homozygous) were selected. The presence of MpC002/Rack-1/GFP inserts was confirmed by PCR and sequencing. Three independent lines were chosen for dsMpC002/dsRack-1 and one for dsGFP.
M. persicae survival and fecundity assay on Arabidopsis transgenic lines F3 seed were sown and seedlings were transferred to single pots (10 cm diameter) and transferred to an environmental growth room at temperature 18uC day/16uC night under 8 hours of light. Five M. persicae adults were confined to single four-week-old Arabidopsis lines in sealed experimental cages containing the entire plant. Two days later adults were removed and five nymphs remained on the plants. The number of offspring produced on the 10th, 14th, 16th day of the experiment were counted and removed. This experiment was repeated three times to create data from three independent biological replicates with four plants per line per replicate.

Northern blot analysis
To assess siRNA accumulation levels by northern blot analyses, N. benthamiana leaves were harvested each day for 6 days after agroinfiltration with the pJawohl8-RNAi constructs and whole twoweek-old A. thaliana F3 transgenic seedlings were used.
Total RNA was extracted from leaves/seedlings using TRIzol reagent (Invitrogen). 15 mg of total RNA was resolved on a 15% polyacrylamide gel (15% acrylamide-bisacrylamide solution 19:1/ 7 M urea/20 mM MOPS pH 7.0) and blotted to a Hybond-N membrane (Amersham, Little Chalfont, UK) by a Trans-blot TM (Biorad, Hempstead, UK) semi-dry transfer cell. Cross-linking of RNA was performed by incubating the membrane for two hours using a pH 8.0 solution of 0.2 M 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) (Sigma-Aldrich, Gillingham, UK) and 0.1 M 1-methlyimidazol (Sigma-Aldrich). DNA probes were labeled using Klenow fragment (Ambion, Lingley House, UK) with [a-32P] dCTP to generate highly specific probes. To control for equal loading of RNA amounts, blots were hybridized with a probe to U6 (snRNA 59-GCTAATCTTCTCTGTATCGTTCC-39) [43]. MicroRNA marker (NEB, Hitchin, UK) consisting of three synthetic single-stranded RNA oligonucleotides of 17, 21 and 25 residues was loaded in gels and hybridized on blots with corresponding microRNA probe to determine size of siRNA between 21-23 nucleotides. The signals were detected after 3-day exposure to phosphor storage plates (GE Healthcare, Little Chalfont, UK) scanned with a Typhoon TM 9200 scanner (GE Healthcare) and analyzed using ImageQuant TM (GE Healthcare).

Quantitative real-time PCR analysis
Total RNA was extracted from adult Myzus persicae after A. thaliana and N. benthamiana fecundity assays using TRIzol reagent. DNA contaminations were removed by treating RNA extraction with RNase-free DNase (QIAGEN, West Sussex, UK) and purified with QIAamp columns (QIAGEN). First-strand cDNA was synthesized at 37uC from total RNA using M-MLV (Invitrogen) reverse transcriptase according to the manufacturer's instructions.
Each reaction contained 1 ml of cDNA, 0.5 ml of each specific primers (10 pmol/ml) (Table S1), and 10 ml of 26 SYBR Green Super-mix reagent (Bio-Rad) in a final volume of 20 ml. The following PCR program was used for all PCR reactions: 90uC for 3 m, followed by 40 cycles of 95uC for 30 s, 60uC for 30 s, 72uC for 30 s followed by 10 m at 72uC at the end. Threshold cycle (CT) values were calculated using Bio-Rad CFX Manager TM software (Bio-Rad).
The CT values were normalized for difference in cDNA amount using ßTubulin and L27 CT values [10,14]. Fold changes were calculated by comparing the normalized transcript levels of MpC002 and Rack-1 of M. persicae fed on dsMpC002 and dsRack-1 transgenic plants to aphids fed on dsGFP transgenic plants.

Statistical analyses
All statistical analyses were conducted using GenStat 11 statistical package (VSNi Ltd, Hemel Hempstead, UK) ( Table  S2, Table S3). Data were checked for approximate normal distribution by visualising residuals. Classical linear regression analysis using a generalized linear model (GLM) with Poisson distributions was applied to analyse the M. persicae fecundity data on A. thaliana transgenic lines, with ''nymphs'' as a response variate. The aphid nymph production on 4 plants per treatment was used as independent data points in statistical analyses in which the biological replicate was used as a variable.
N. benthamiana leaf disc assay fecundity data were analyzed using an unbalanced one-way ANOVA design with ''construct'' as the treatment and ''repeat'' as the block. In the N. benthamiana leaf disc assay, aphid fecundity was monitored on individual leaf discs at 6 discs per treatment. Numbers of aphid nymph produced on each leaf disc were used as independent data points in statistical analyses in which the biological replicate was used as a variable. Leaf discs that dried up because of lack of humidity were excluded giving 4-6 leaf discs per construct for each biological replicate. The relative gene expression data were analyzed using 2 2DDC T method as previously described [44]. The results were analyzed for significant difference with Student's t-test. For replication, 'n = ?' refers to number of technical replicates used for each variable in each biological replicate i.e. n = 4 Arabidopsis plants per line per biological replicate, n = 4-6 N. benthamiana leaf discs per construct per biological replicate, n = 3 technical replicates per qRT-PCR biological replicate.