RNA interference as a gene silencing tool to control Tuta absoluta in tomato (Solanum lycopersicum)

RNA interference (RNAi), a gene-silencing mechanism that involves providing double-stranded RNA molecules that match a specific target gene sequence, is now widely used in functional genetic studies. The potential application of RNAi-mediated control of agricultural insect pests has rapidly become evident. The production of transgenic plants expressing dsRNA molecules that target essential insect genes could provide a means of specific gene silencing in larvae that feed on these plants, resulting in larval phenotypes that range from loss of appetite to death. In this report, we show that the tomato leafminer (Tuta absoluta), a major threat to commercial tomato production, can be targeted by RNAi. We selected two target genes (Vacuolar ATPase-A and Arginine kinase) based on the RNAi response reported for these genes in other pest species. In view of the lack of an artificial diet for T. absoluta, we used two approaches to deliver dsRNA into tomato leaflets. The first approach was based on the uptake of dsRNA by leaflets and the second was based on “in planta-induced transient gene silencing” (PITGS), a well-established method for silencing plant genes, used here for the first time to deliver in planta-transcribed dsRNA to target insect genes. Tuta absoluta larvae that fed on leaves containing dsRNA of the target genes showed an ∼60% reduction in target gene transcript accumulation, an increase in larval mortality and less leaf damage. We then generated transgenic ‘Micro-Tom’ tomato plants that expressed hairpin sequences for both genes and observed a reduction in foliar damage by T. absoluta in these plants. Our results demonstrate the feasibility of RNAi as an alternative method for controlling this critical tomato pest.


Introduction 44
The mechanism of RNA interference (RNAi), in which small RNAs can rapidly cause post-45 transcriptional specific gene silencing, has become a powerful tool for analysing gene function in a 46 variety of organisms (Hannon, 2002). To clone the target gene fragments in the binary silencing vector pK7GWIWG2(I) (Karimi,166 Inze, Depicker, 2002) primers were synthesized to amplify fragments flanked by the recombination 167 sequences attL1 and attL2 (Table S2)  The green fluorescent protein (GFP) gene was used as a negative control. The vector 197 pCAMBIA1302 was used as a template to amplify a negative control GFP gene fragment (276 bp)  198   with the specific primers GFP-F (5'-TAATACGACTCACTATAGGGCAGTGGAGAGGGTGAA)  199 and GFP-R (5'-TAATACGACTCACTATAGGGTTGACGAGGGTGTCTC), both containing 200 additional T7 sequences (underlined). Similar transcription in vitro was done with this template. 201 202 Labeling dsRNA to follow uptake by tomato leaflets and ingestion by T. absoluta 203 In vitro transcription of dsRNA was done as described above, except that 2 µL of Cy3-204 labelled riboCTP was added . Fluorescently-labelled dsRNA was purified using In subsequent feeding assays, tomato leaflets were infiltrated with Agrobacterium cells 233 carrying the V-ATPase or AK hairpin constructs or with Agrobacterium cells carrying the GFPi 234 construct as a negative control, in triplicate. The Agrobacterium cells were grown on LB medium 235 containing gentamycin (25 µg mL -1 ) and spectinomycin (100 µg mL -1 ) for 12 h, centrifuged at 236 3,000 g for 5 min and resuspended in water to an OD 600nm = 0.5. After 24 h, first instar T. absoluta 237 larvae were placed on the treated leaf areas and 24 h, 48 h and 72 h later treated larvae and their 238 respective controls were sampled for analysis of V-ATPase and AK expression by  Feeding assays were also done to estimate larval mortality. Detached 'Santa Clara' leaflets 240 had their petioles immersed in an aqueous solution containing 1 µg of dsRNA of each target gene 241 (V-ATPase or AK, plus GFP control), a procedure that was repeated daily for 10 days, with a total 242 of 10 µg being provided to each leaflet. A total of 10 first instar larvae were placed to feed on these 243 leaflets (in triplicate) and larvae mortality was estimated after 5, 7, 10 and 24 days of treatment. 244 Under the conditions used here, the feeding cycle of T. absoluta lasted 10-12 days from larval 245 emergence to pupae, and a total of ca. 20-24 days for adults to emerge. An additional assay was 246 done using 'Santa Clara' tomato leaflets with the petiole immersed in increasing amounts of dsRNA 247 (total: 500, 1000 or 5000 ng) of the target genes (V-ATPase or AK) or GFP negative control (in 248 11 duplicate). Five larvae were placed on each leaflet and leaflets were photographed for 11 days to 249 visually assess the extent of damage. days on 3 mL of LB medium with spectinomycin (100 mg L -1 ), rifampicin (50 mg L -1 ) and 274 gentamycin (25 mg L -1 ) was inoculated into 50 mL of LB medium with the same antibiotics and 275 incubated overnight at 120 rpm and 28 ºC. The suspensions were then centrifuged (1,000 g, 15 min, 276 20 ºC) and the pellet was resuspended in liquid MS medium containing sucrose (30 g L -1 ), with the 277 OD 600nm adjusted to 0.2-0.3. Acetosyringone (100 µM) was added to the suspensions 10 min before 278 co-cultivation, which was done on the same semi-solid MS medium for two days in the dark at 25 279 ºC. Explants were then transferred to fresh MS medium supplemented with B5 vitamins, sucrose 280 (30 g L -1 ), 5 µM benzylamino purine (BAP), kanamycin (100 mg L -1 ) and timetin (300 mg L -1 ) and 281 maintained on a 16 h photoperiod at 25 ºC for three weeks. Subsequently, adventitious shoots >5 282 mm long were transferred to identical medium until roots developed and the plantlets were 283 hardened (~two weeks), after which they were moved to a greenhouse. 284 285

Genetic analysis of transgenic plants 286
Total DNA and RNA were extracted from putative transgenic plants using Trizol. 287 Confirmation of transgenesis was done by PCR using a 35S promoter sense primer 288 (GCACAATCCCACTATCCTTC) together with a target gene (ATPase and AK)-specific reverse 289 primer (Table S4) (Table S4)

Target gene isolation from T. absoluta and dsRNA transcription in vitro 328
Since little genomic information is available for T. absoluta, we conducted target gene 329 fragment cloning using degenerated primers for both target genes (estimated coding sequence of 330 ~1,850 bp for V-ATPase and ~1,065 bp for AK; unpublished data) by using nested PCR (Table S1). 331 The final amplification products were run on agarose gels and fragments for both genes were 332 purified and cloned. Three positive clones were sequenced in both directions for each target gene. 333 The consensus sequence assembled from the three clones contained 285 bp for V-ATPase 334 transcribed in vitro were provided in solution to detached tomato leaflets. The treated leaves and the 360 feeding larvae were imaged by confocal microscopy 6 h or 24 h after treatment ( Fig. 1 and Fig. S2). 361 Labelled dsRNA species were already strongly detected in the leaflet petiole and blade (mid-rib and 362 lateral veins) of the leaflets 6 h after treatment (Fig. 1Ab). After 24 h, Cy3-labeled RNA molecules 363 were detected throughout the leaf blade ( Fig. 1Ad and Fig. S2). With time, Cy3-labeled RNA 364 molecules accumulated at the leaf margin until saturation was reached in certain areas (Fig. 1Ad). 365 We then imaged larvae fed on treated or untreated leaflets using the 488 channel (green 366 fluorescence) to detect chlorophyll auto-fluorescence, indicative of plant tissue ingestion by the 367 larvae, and the 555 channel (red fluorescence) to detect Cy3 fluorescence (Fig. 1B). In both 368 treatments, green fluorescence was detected throughout the larval digestive tract (Fig. 1Bb), 369 indicating that the larvae fed normally under both circumstances. However, larvae fed on dsRNA-370 treated leaflets showed a strong Cy3 signal in the digestive tract, indicating the presence of leaflet-371 absorbed Cy3-labeled RNA molecules in the gastric caeca of the midgut (Fig. 1Bc). Confocal fluorescence microscopy showed that agro-infiltrated leaves with the eGFP line displayed 381 GFP (green) fluorescence as sparse cells on the leaf blade (Fig. S3a). When both Agrobacterium 382 strains (eGFP and GFPi) were co-infiltrated, there was a drastic reduction in the number and 383 intensity of cells with GFP fluorescence (Fig. S3b); this fluorescence was similar to that of leaf 384 regions without agro-infiltration (Fig. S3c). 385 Together, these results indicated that both approaches were suitable for delivering dsRNA 386 into tomato leaves. We next used both delivery methods to evaluate the effectiveness of RNAi in 387 silencing specific target genes in T. absoluta larvae. 388 389

Effect of RNAi on target gene expression 390
For both RNAi delivery methods, larvae were allowed to feed exclusively on RNAi-treated 391 (dsRNA uptake or agro-infiltration) leaflets and collected 24 h, 48 h and 72 h later. The relative 392 expression of V-ATPase and AK was quantified by RT-qPCR. For the PITGS dsRNA delivery 393 method, the target gene fragments were cloned into a binary vector as a hairpin-expressing cassette 394 and agro-infiltrated into 'Santa Clara' tomato leaves. The cloned fragments were amplified with 395 primers flanked by attL1 and attL2 sequences (Table S2) to enable direct recombination with the 396 binary vector pK7GWIWG2(I) (Karimi, Inze & Depicker, 2002). 397 Larvae fed on leaflets treated by the dsRNA uptake delivery method showed a significant 398 decrease in transcript accumulation for both genes 48 h and 72 h after treatment (~40% reduction at 399 72 h after treatment) ( Fig. 2A). Larvae fed on agro-infiltration leaflets showed a decrease in 400 transcript accumulation at all time points, with the highest decrease occurring 72 h after treatment, 401 (~35% reduction for V-ATPase and 40% reduction for AK) (Fig. 2B). 402 Considering that both dsRNA delivery approaches resulted in similar gene silencing effects, 403 subsequent experiments were done using only the leaf dsRNA uptake delivery method. 404 405

Effect of RNAi on larval mortality 406
To determine the effect of RNAi on larval mortality, T. absoluta larvae were allowed to 407 feed on single leaflets (n=3) that absorbed 10 µg of dsRNA from V-ATPase, AK or GFP. Larvae 408 were sampled after five, seven and ten days of treatment and an additional pupal sample was 409 collected after 24 days. Larval mortality was significantly higher in larvae fed on leaflets that 410 absorbed dsRNA of either target gene when compared to the GFP control at all time points, with an 411 additional increase over time (Fig. 3). By day 24, mortality had reached an average of 50% for V-412 ATPase and 43% for AK compared to 17% for the GFP control (Fig. 3). Independent experiments 413 using different total amounts of dsRNA in the leaflets yielded similar results (not shown). 414 Evaluation of larvae after 11 days of treatment (Fig. S4a-c) and at the pupal stage (Fig. S4d-f) 415 revealed that larvae fed on RNAi-treated leaflets displayed developmental delay, reduced body size, 416 external morphologies of the 3 rd instar stage (when 4 th instar was expected) (Fig. S4a-c), failure to 417 pupate ( Fig. S4d-f) and failure to emerge as adults (data not shown). 418 419

Effect of RNAi on tomato leaf damage 420
We next assessed whether the gene silencing and larval mortality observed for both RNAi 421 target genes resulted in less herbivory by T. absoluta on tomato leaves. After 11 days of T. absoluta 422 herbivory, leaflet blades treated with increasing amounts of dsRNA (total: 500, 1000 or 5000 ng) of 423 the target genes (V-ATPase or AK) (Fig. 4c-f) were visibly less damaged by larval herbivory when 424 compared to leaflets treated with GFP dsRNA (Fig. 4a,b), and the observed protective effect 425 appeared to be dose-dependent (Fig. 4c-f). A protective effect was seen even at lower doses of 426 dsRNA treatment, particularly for V-ATPase (Fig. 4d). increase in larval mortality that ranged from 30% in events ATPase1.1 and ATPase7 to 40% in 455 ATPase 9 and AK 1.1 (Fig. 5A). The effect of RNAi was also assessed by comparing larval weight 456 between treatments. Larvae fed on non-transgenic controls had a mean weight of 3.5 mg, while 457 those fed on leaves of the different RNAi transgenic plants had a mean weight of 1.7 to 2.4 mg (Fig.  458   5B). 459 Visual analysis of leaves from the different treatments revealed a clear protective effect 460 against larval herbivory in RNAi transgenic tomato plants based on an analysis of the leaflets before 461 ( Fig. 5Aa-e) and after (Fig. 5Af-j) larval feeding. Whereas non-transgenic leaves had almost no 462 undamaged leaf blade (Fig. 5Af), areas of apparently intact leaf blade were clearly seen in RNAi 463 transgenic leaves (Fig. 5Ag-j).   Table S5. Gene-specific and universal primer sequences used to detect the predicted small 793 interfering RNAs (siRNA) derived from the target genes V-ATPase (siRNA 794 AATACATGCGCGCTCTAGATGAC) and AK (siRNA AAGTATCGTCCACACTGTCTGGC) 795 and the control microRNA156 (UGACAGAAGAGAGUGAGCAC) in transgenic plants.

Figure 2(on next page)
Effect of the dsRNA uptake delivery method on the relative expression of target genes in T. absoluta larvae.
Relative expression of target genes V-ATPase and AK in larvae of T. absoluta fed on tomato leaflets that absorbed dsRNA solution (25 μg mL -1 ), sampled at 24, 48 and 72 h after initiation of treatments normalized to positive controls that were exposed to GFP dsRNA (n= 3). The Rpl 5 gene was used as gene reference (n= 3). The asterisk indicates significant differences (t-test; "*" at P<0.05; "**"at P<0.01).

Figure 3
RNAi effects on larval mortality.
Mortality of Tuta absoluta larvae (n=10) after feeding on tomato leaf treated with dsRNA from V-ATPase, AK or GFP control for 24 days. Tomato leaflets were provided with one μg of each dsRNA (V-ATPase or AK, plus GFP control) a day per leaflet for 10 days in a total 10 μg.

Figure 4
RNAi effects on leaf damage caused by larval herbivory.