Nicotiana benthamiana and tomato plants were maintained in a growth chamber at 24°C with a light/dark cycle of 16/8 h. Plants with 4–5 fully expanded leaves were used for the bioassays. Plants were inoculated with a TSWV pepper isolate [P105, PLAVIT collection, IPSP-CNR Torino, Italy, World Data Center for Microorganism (WDCM) no. 1057, http://www.wfcc.info/ccinfo/collection/col_by_country/i/39/].

To produce dsRNA molecules, a two-step PCR approach followed by in vitro transcription was used (Voloudakis et al., 2015). In the case of N- and NSm-targeting dsRNAs, total RNA from TSWV-infected N. benthamiana plants was extracted using Trizol (Life Technologies, United States) according to manufacturer’s instructions. 1 μg of total RNA extracted from TSWV-infected N. benthamiana plants was used as template for cDNA synthesis, using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, United States), according to manufacturer’s instructions. Viral fragments were amplified with specific primers designed using Primer3 software¹. The T7 RNA polymerase promoter/binding site sequence was added at the 5 ì end of both the forward and the reverse primers (Supplementary Table 1). PCR was carried out in a final volume of 50 μl, containing 10X reaction buffer, 1 μl cDNA template (diluted 1:5), 200 μM dNTPs, 0.2 μM of each primer, 1.5 mM MgCl₂, and 2.0 units of Platinum Taq DNA polymerase (Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States). After an initial denaturation at 95°C for 10 min, the first 10 cycles were performed as follows: 95°C for 30 s, 40°C for 45 s, 72°C for 1 min, followed by 35 cycles each of 95°C for 30 s, 55°C for 45 s, and 72°C for 1 min. After a final extension at 72°C for 7 min, PCR fragments were loaded on 1% agarose gel and purified by the DNA Clean and Concentrator kit (Zymo Research, Irvine, United States). To in vitro synthesize sense and antisense viral derived ssRNAs, 1 μg of purified DNA template was used in 50 μl transcription reaction containing T7 reaction buffer, 500 μM rNTPs, 5 mM DTT, 50 units of T7 RNA polymerase (Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States), conducted at 37°C for 2 h. After removing DNA template by TURBO RNase-free DNAse (Ambion, Thermo Fisher Scientific, Waltham, MA, United States), dsRNAs were obtained by mixing the specific sense and antisense ssRNAs, then incubating at 95°C for 3 min, and at 37°C for 30 min. The formation of dsRNAs was confirmed by treatment with Mung Bean Nuclease (New England Biolabs, MA, United States).

As negative control, a dsRNA targeting the replication initiator protein (Rep) coding region of the geminivirus tomato yellow leaf curl Sardinia virus (TYLCSV) was used. In this case, DNA was extracted from TYLCSV-infected N. benthamiana plants using the TLES method described in Maffei et al. (2014). The Rep-targeting dsRNA was synthetized as described above, using specific primers reported in Supplementary Table 1.

In order to ensure the uniformity of the TSWV inoculum in all the experiments, a TSWV inoculum stock was prepared from systemically infected N. benthamiana leaves. For this, symptomatic leaf tissue was cut in small slices, split into 500 mg aliquots and stored in liquid nitrogen until use. Inoculation was performed by homogenizing each aliquot in 50 ml of inoculation buffer (5 mM diethyldithiocarbamic acid, 1 mM ethylenediaminetetraacetic acid, and 5 mM sodium sulfite), and applying a 50 μl-dose to the upper side of two different leaves of each plant previously dusted with carborundum. When required, 10 μg of dsRNAs were added to the inoculum, just before inoculation.

Nicotiana benthamiana plants were used to evaluate the effect of dsRNAs on local lesions. Systemic infection induced by TSWV was evaluated in N. benthamiana and tomato plants, at different time points from the inoculation. Plants were inoculated with TSWV together with dsRNAs (N and NSm). The positive control group was inoculated with TSWV only and water. As negative controls, one group of plants received TSWV and the Rep-targeting dsRNAs originated from TYLCSV, while another one received only water. A total of thirteen N. benthamiana and seven tomato plants per treatment were analyzed in three and one different experiments, respectively. Symptoms development was monitored until 40 days post inoculation (dpi) in the case of N. benthamiana and 63 dpi in the case of tomato. Presence of viral RNA was checked by PCR in newly emerged leaves that did not receive the inoculum.

In order to evaluate the persistence/movement of dsRNAs in the plants that were not inoculated with TSWV, 10 μg of dsRNAs were mixed with 100 μl of sterile water and mechanically inoculated on two carborundum-dusted fully expanded leaves (50 μl per leaf) by gentle rubbing. Two leaf disks from each treated (local) and first expanded leaf from the apex (systemic) were collected at 1, 3, 6, and 9 dpi. Three plants were used for each time point. Just before sampling, leaves were washed with Triton X-100 (0.05%) and water to eliminate residual dsRNAs present on the leaf surface. Total RNA was extracted by Trizol reagent (Life Technologies, United States). cDNA synthesis was performed with random primers using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, United States), according to manufacturer’s instructions.

Tomato spotted wilt virus infection was evaluated by RT-PCR with a primers pair (TSWV_L_fw/TSWV_L_rv, amplicon size 646 bp; Supplementary Table 1) designed on the L RNA, a genomic component unrelated to the N and NSm genes used to design the dsRNAs. In the dsRNAs mobility experiment, the presence of N- and NSm-targeting dsRNAs was estimated by PCR using primers TSWV_N_fw/TSWV_N_rv and TSWV_NSm_fw/TSWV_NSm_rv, respectively (Supplementary Table 1). PCR was performed in a final volume of 25 μl, containing 2.5 μl reaction buffer, 1 μl cDNA template, 200 μM each dNTPs, 0.4 μM each primer, 2 mM MgCl2, and 1 unit Platinum Taq DNA polymerase (Invitrogen). After an initial denaturation step for 4 min at 95°C, 30 cycles consisting each of 30 s at 95°C, 30 s at 58°C, and 30 s at 72°C were performed, followed by a final extension step for 7 min at 72°C.

Quantitative RT-PCR (qRT-PCR) was carried out using iCycler iQTM Real-Time PCR Detection System (BioRad Laboratories, Hercules, CA, United States), with the following cycling parameters: 1 cycle at 95°C for 5 min, 45 cycles, each consisting of 15 s at 95°C and 1 min at 60°C. A melting curve was recorded at the end of each run to assess the specificity of amplification. All reactions were performed with three technical replicates. RT-PCR efficiency was calculated using standard curves constructed with serial dilutions of cDNA extracted from infected plants. Data acquisition and analysis were handled by the BioRad iCycler software (version 3.06070) that calculates Ct values and standard curves. qRT_TSWV_N_fw/qRT_TSWV_N_rv (Mason et al., 2002) and qRT_TSWV_NSm_fw/qRT_TSWV_NSm_rv primers pairs (Supplementary Table 1) were used to amplify exogenous dsRNAs, while the primer pair qRT_NbCOX_fw/qRT_NbCOX_rv (Supplementary Table 1) was used for the amplification of the N. benthamiana gene Niben101Scf02399 coding for the cytochrome c oxidase (NbCOX), used as reference gene (Nerva et al., 2017). The relative dsRNAs amount was estimated for each sample using the ΔCt method, where ΔCt is |Ct_dsRNA – Ct_cox|.

Small RNA populations originating from N- and NSm-targeting dsRNAs were analyzed in samples collected for the dsRNA movement investigation at 1 dpi. Small RNA libraries preparation and sequencing were performed by Novogene (United Kingdom) Company Limited (United Kingdom). After adapter removal with fastp (Chen et al., 2018) and low-quality filtering and artifact removal with fastx-toolkit², clean reads were mapped to the TSWV genome segment targeted by the applied exogenous dsRNA (S segment, Acc. Num. DQ376178.1; M segment, Acc. Num. KJ575621.1) using bowtie v1-3-0 (Langmead et al., 2009), with 0 mismatches. Mapping results were visualized using Misis (Seguin et al., 2014). Sequence data have been submitted to the Sequence Read Archive (SRA) with the BioProject ID PRJNA672300.

In order to select the genomic regions of TSWV most suitable as target by dsRNAs, we first considered the vsRNAs profile of TSWV available in the literature (Mitter et al., 2013; Margaria et al., 2015), starting from the hypothesis that the genomic regions characterized by a high number of mapping vsRNAs could be more subjected to RNAi-mediated degradation. According to Mitter et al. (2013), most vsRNAs detected in TSWV-infected N. benthamiana plants mapped to the M segment, followed by the S segment, while the L RNA had the least number of mapping vsRNAs. Therefore, we focused our attention on the M and S segments. Among the two open reading frames (ORFs) of the M segment, GP and NSm, the GP ORF has the higher number of mapping vsRNAs (Mitter et al., 2013). However, we decided to select the NSm coding region as dsRNA target, since the movement protein has an important effect on viral systemic spread while the glycoproteins are not needed for in planta replication and intercellular spread of the virus; indeed, amiRNAs targeting the GP fragment did not show any antiviral protective effect (Carbonell et al., 2019). In the case of the S segment containing the N and NSs ORFs, we focused on the N coding region, since it codes for the nucleocapsid protein, a structural protein essential for virions formation. In addition, artificial miRNAs targeting the NSs ORF were not effective in protecting plants against TSWV infection (Mitter et al., 2016; Carbonell et al., 2019).

Since previous work showed that the exogenous dsRNAs must be longer than 300 bp to effectively interfere with virus infection (Tenllado and Díaz-Ruíz, 2001), we designed to produce dsRNAs longer than this threshold, i.e., 761 bp for the N-targeting dsRNA and 603 bp for the NSm-targeting dsRNA.

Finally, to better evaluate the dsRNAs protective effect, we decided to include a dsRNA homologous to a portion of the gene encoding the Rep pf TYLCSV as a negative control. The length of the Rep-targeting dsRNA (598 bp) was similar to the N- and NSm-targeting dsRNAs, but, due to the lack of homology with the TSWV genome, this dsRNA was not expected to show any protective effect against TSWV.

The method employed to produce dsRNAs, based on two sequential PCR reactions, coupled with in vitro transcription (Voloudakis et al., 2015) typically yielded 50–80 μg of dsRNAs, starting from 1 μg of DNA template. The double-stranded nature of the molecules obtained was confirmed by incubating the in vitro-transcribed RNAs before and after the annealing step with Mung Bean Nuclease, an enzyme that specifically degrades single-stranded nucleic acids (Figure 1).

At 7 dpi, typical chlorotic and necrotic local lesions occurred on leaves of N. benthamiana plants inoculated with TSWV alone, with no significant difference (p-value < 0.05) in their average number compared to plants inoculated with TSWV + NSm-targeting dsRNAs and with TSWV + Rep-targeting dsRNA. At the same time point, plants inoculated with TSWV + N-targeting dsRNAs showed drastic reduction in the number of local lesions (Figure 2), highlighting a protective effect of the N-targeting dsRNAs.

All the N. benthamiana plants inoculated with TSWV alone showed typical systemic symptoms of chlorosis, necrosis, stunting, and wilting at 11 dpi. At the same time point, similar symptoms were observed on 92% of plants treated with TSWV + NSm-targeting dsRNAs and 75% of plants treated with TSWV + Rep-targeting dsRNAs; conversely, only 15% of plants treated with TSWV + N-targeting dsRNAs appeared symptomatic. At later observation (25 dpi), all plants treated with TSWV + NSm-targeting dsRNAs or with TSWV + Rep-targeting dsRNAs showed the systemic TSWV symptoms. In the case of plants treated with TSWV + N-targeting dsRNAs, the rate of symptomatic plants reached 46% at 17 dpi and remained unchanged until the end of the experiment (40 dpi; Figure 3). The RT-PCR performed on a subset of N. benthamiana plants confirmed visual symptoms evaluation (not shown).

In the case of tomato plants, most of the plants (71%) treated with TSWV + NSm-targeting dsRNAs and 100% of the plants inoculated only with TSWV were already infected at 21 dpi, but none of the plants inoculated with TSWV and treated with N-targeting dsRNAs became infected for the entire duration of the experiment, at 64 dpi (Figure 4A).

In both species, we didn’t observe any difference in the reproductive stage; all the asymptomatic plants, including those treated with dsRNAs, exhibited a normal flowering, while no flowering was observed in symptomatic plants, due to the heavy symptomatology.

The virus was detected in young leaves of all the symptomatic plants while it was absent in all asymptomatic plants treated with TSWV + N-targeting dsRNAs (Figure 4B), confirming visual inspection (Figure 4C).

Taken together, these results show that exogenous application of dsRNAs homologous to the N gene can completely suppress or robustly contrast viral replication and protects plants from TSWV infection, both in N. benthamiana and tomato, by inhibiting or strongly delaying symptom development, while dsRNAs targeting the NSm gene have almost no protective effect.

In order to investigate the persistence and the systemic movement of N and NSm-targeting dsRNAs applied on the leaves, in the absence of virus infection, we analyzed by RT-PCR the presence of both dsRNAs at 1, 3, 6, and 9 dpi in both treated (local persistence) or young untreated leaves (systemic movement) of N. benthamiana plants (Figure 5A). Both dsRNAs could be detected until 9 dpi in the treated leaves indicating that they persist there for several days after application. Moreover, we detected both N and NSm-targeting dsRNAs also in untreated leaves at all time points, although with a lower intensity compared to treated tissues.

When the quantitative evaluation of the dsRNAs was performed by qRT-PCR, we observed that most of the applied N- or NSm-targeting dsRNAs remained in the treated leaves and only a limited amount of them was transported systemically to the young untreated leaves. Interestingly, the dsRNAs systemic movement, even if very limited, could be observed at 1 dpi and remained steadily low until 9 dpi (Figure 5B).

These results indicate that both dsRNAs considered are able to move systemically within the plant, but that the majority of them remains in treated leaves. No significant differences in persistence or systemic movement capacity were observed between the N- and NSm-targeting dsRNAs.

In order to evaluate if N- and NSm-targeting dsRNAs were processed by Dicer and were able to originate the silencing effector molecules (siRNAs), we sequenced the sRNAs populations of treated (local), and newly emerged non-treated (systemic) N. benthamiana leaves, 1 day post exogenous dsRNAs application. After adapter removal and quality filtering, 18 to 28 million of reads were obtained (Supplementary Tables 2, 3). Sequences with length ranging from 20 to 25 nt were selected for further analyses. In the case of leaves treated with dsRNAs (local), the percentage of reads mapping to the N- and NSm-targeting dsRNAs regions was 0.39 and 0.50, respectively (Figure 6); when newly emerged non-treated leaves were considered, only few reads mapped to the corresponding viral region, i.e., 53 reads for the N-targeting dsRNA and 89 reads for NSm-targeting dsRNA. These results indicate that both dsRNAs are effectively introduced into the plants and processed, thus originating siRNAs that can guide the cleavage of cognate sequences. However, in agreement with the very limited amount of dsRNAs that move systemically into the plant, only few dsRNA-derived sRNAs were found in the newly emerged non-treated leaves.

The size distribution of sRNAs originating from the dsRNAs was basically uniform; in both cases, even if the amount of 21- and 22-nt reads was slightly higher than the number of other reads, no well-defined peaks related to a particular read size was observed (Figure 6). This may suggest that the inoculated dsRNA was not processed into sRNAs and the protective effect was possibly dsRNA- and not RNAi-mediated.