In plants, RNAi is a gene silencing phenomenon that involves sequence-specific suppression of genes, which can be induced by dsRNAs precursor that may vary in length and origin (Waterhouse et al., 2001). In plant cellular system, dsRNAs are mainly processed into three categories: short interfering RNAs (siRNAs), micro RNAs (miRNAs), and piwi interacting RNAs (piRNAs). Generally, siRNAs and miRNAs are collectively known as small RNAs (sRNAs) (Carthew and Sontheimer, 2009). The possible cellular mechanism of dsRNA induced RNAi in plants involves the following steps (Meister and Tuschl, 2004; Dalakouras et al., 2020): (i) Upon cellular uptake of dsRNAs, DICER-LIKE (DCL) endonucleases rapidly cleavages them into 20 to 25-nucleotide siRNAs with 2-nt 3´ overhangs at both ends; (ii) One strand of siRNAs is incorporated into an ARGONAUTE (AGO) protein to form an RNA-induced silencing complex (RISC); (iii) Finally, the siRNA molecules guide the RISC to scan the cytoplasm for recognition and cleavage/degradation of the complementary transcripts, thus resulting in post-transcriptional gene silencing (PTGS).

Successful cellular uptake and subsequent initiation of target gene silencing by exogenous dsRNAs are subject to the influences of length and/or concentrations, application methods, and delivery techniques, as well as the sensitivity of plant organs to dsRNA applications (Numata et al., 2014; Dalakouras et al., 2016; Mitter et al., 2017; Dubrovina and Kiselev, 2019). To date, very little is known regarding how these factors influence the exogenous dsRNAs-induced gene silencing in plant systems. Therefore, it is necessary to find out the optimal physical conditions along with the exogenous application method of external dsRNAs to develop eco-friendly approaches for plant protection and improvement of crop productivity ( Figure 1 ).

In general, exogenous application methods include spraying, infiltration, injection, spreading, mechanical inoculation, and root/seed soaking, and they all have been widely used to apply dsRNAs/siRNAs/hpRNAs onto plants for target gene silencing ( Table 1 ). When high-pressure spraying was used for the exogenous application of siRNAs, it was successful in inducing local and systemic silencing of the green fluorescent protein (GFP) transgene in Nicotiana benthamiana (Dalakouras et al., 2016). According to this study, high-pressure spraying was more efficient compared to wiping, infiltration, and gene gun methods. In contrast, another study reported that direct exogenous application of dsRNA by spreading without using any additional techniques induced efficient suppression of enhanced green fluorescent protein (EGFP) and neomycin phosphotransferase–II (NPTII) transgenes in Arabidopsis (Dubrovina et al., 2019). The authors of this study also analyzed the effects of different dsRNA concentrations (0.1, 0.35 and 1.0 µg/µl) and the results indicated that optimum concentration (0.35 µg/µl) had a higher significant influence on transgene-silencing efficiency (Dubrovina et al., 2019). The effects of different lengths of dsRNAs (315, 596, and 977-bp) targeting different virus genes were also investigated in N. tabacum leaves and results indicated that shorter dsRNAs showed reduced antiviral activity, indicating that dsRNA length could influence on its efficacy (Tenllado and Diaz‐Ruiz, 2001).

Plant cell contains complex cellular structures, such as the rigid cell wall that acts as a physical barrier to provide tensile strength and protection against several stresses (Islam et al., 2019). Therefore, the delivery of exogenous dsRNAs/siRNAs/hpRNAs into plant cells is considered the most crucial step in initiating RNAi machinery. Nanotechnology-based delivery and surfactants-based delivery methods were extensively used along with naked-dsRNAs application in plants (reviewed in Dubrovina and Kiselev, 2019; Dalakouras et al., 2020). The major limitation of exogenous applications of naked-dsRNAs is their short-term stability. Some studies reported that nanoparticle-based delivery could enhance the stability and efficacy of exogenously applied dsRNAs when compared to naked-dsRNA delivery (Numata et al., 2014; Mitter et al., 2017). According to Numata et al. (2014), naked-dsRNA and ionic dsRNA-peptide complexes were infiltrated into Arabidopsis thaliana leaves to induce RNAi of the yellow fluorescent protein (YEP) transgene and chalcone synthase (CHS) endogenous gene. The peptide complex-based delivery of dsRNAs was found to downregulate the YEP expression within 12 h after infiltration, and the silencing effect was partly maintained for at least 36 h; whereas, the effect of naked-dsRNA on target genes was not significant. The dsRNA-peptide complexes were also found to be effective in downregulating the target YFP transgene in Populus tremula plant (Numata et al., 2014). In a subsequent study, the stability of exogenously applied naked-dsRNAs was compared with layered double hydroxide (LDH) clay nanosheets-based delivery (Mitter et al., 2017). Confocal microscopic analyses of Cy3 fluorophore labeled naked-dsRNA and LDH-loaded dsRNA targeting CMV2b found that most of the naked-dsRNA was washed away, whereas LDH-loaded dsRNA largely remained on the A. thaliana leaves. This study also reported that LDH-loaded dsRNA showed sustained release and were detectable after 30 days of application on sprayed N. tabacum leaves, but the naked-dsRNA was nearly undetectable after 20 days. Therefore, it was suggested that LDH nanosheets-based delivery can significantly improve the stability of exogenous dsRNA (Mitter et al., 2017). When combined with cationic fluorescent nanoparticles, dsRNAs also exhibited more dramatic suppression of target genes than naked-dsRNAs (Jiang et al., 2014). However, there are some reports where naked-dsRNAs proved effective. For instance, a study by Li et al. (2015) reported the naked-dsRNA induced suppression of target genes when exogenously applied to Arabidopsis and rice roots. In this study, naked-dsRNAs targeting MOB Kinase Activator 1A (Mob1A), and Actin genes were applied by root soaking in Arabidopsis and rice, respectively. In Arabidopsis, Mob1A regulates the root growth by controlling appropriate cell number and size and Actin plays a role as a cytoskeletal protein and also regulates root growth. The absorption of dsRNAs by Arabidopsis roots resulted in the suppression of root lengths and numbers as well as inhibited bolting and flowering. Whereas, Actin targeting dsRNAs absorbed by rice roots significantly suppressed root growth (Li et al., 2015). In another study, mechanical inoculation of naked-dsRNAs targeting the MYB1 gene in the hybrid orchid plant was found to prominently reduce orchid flower buds (Lau et al., 2015). The MYB genes play an important role in the development of orchid varieties by regulating the pigmentation and morphogenesis of flowers.

Based on the above-mentioned studies, nanoparticle based-delivery techniques could facilitate the delivery of exogenous dsRNAs by increasing their stability and uptake (Numata et al., 2014; Jiang et al., 2014; Mitter et al., 2017). However, it is worth noting that nanotechnology is quite an expensive technology and also sensitive to the encapsulation process. Some other delivery methods like high-pressure spraying (Dalakouras et al., 2016) or direct exogenous application of naked-dsRNAs may also efficiently induce plant RNAi (Li et al., 2015; Lau et al., 2015; Dubrovina et al., 2019). Nevertheless, the efficiency of exogenously applied dsRNAs on plant RNAi are affected not only by application methods and delivery techniques but also the concentration and length of dsRNAs might play a crucial role (Tenllado and Diaz‐Ruiz, 2001; Carbonell et al., 2008; Dubrovina et al., 2019). Future studies in this direction could be beneficial for developing new technologies for the applications and delivery of dsRNAs towards inducing RNAi for the desired traits of the plants.

Exogenously applied dsRNAs on the plant are absorbed into plant tissues and cells (Koch et al., 2016; Mitter et al., 2017; Dalakouras et al., 2018; Dubrovina and Kiselev, 2019), but it can be utilized to induce RNAi machinery in both plants and their invading pathogens (reviewed in Dubrovina and Kiselev, 2019; Wang and Dean, 2020). Therefore, the uptake of exogenously applied dsRNAs by plant cells is the most critical step. However, the uptake mechanism of exogenously applied dsRNAs is still elusive to much extent. One of the factors that could affect the efficiency of exogenous dsRNAs is the absorption capacity of different plant organs, e.g., leaves, petioles, buds, roots, stems, and seeds. According to Dalakouras et al. (2018), exogenously applied siRNAs by high-pressure spraying onto plant leaves and buds triggered local and systemic RNAi, whereas, delivery of siRNAs by petiole absorption and hpRNA by trunk injection failed to induce RNAi. Interestingly, application of 22-nt siRNA targeting GFP by bud spraying was more efficient to induce RNAi than leaf spraying in N. benthamiana. Whereas, siRNAs and hpRNA that were delivered by petiole absorption were present only in xylem tissues, but not in the apoplast. Based on these observations and others, the authors concluded that: (i) As a delivery method, high-pressure spraying is an efficient approach to deliver exogenous siRNAs into plant cells to induce RNAi; and (ii) As far as organ-based absorption is concerned, leaf and bud spraying efficiently induce RNAi compared to petiole absorption or trunk injection. The authors also hypothesized that the retention of exogenous dsRNAs/siRNAs on plant surfaces could be useful to deliver intact (unprocessed by plant) dsRNAs. These intact dsRNAs if taken up by target insects/fungi, could result in pest and disease management (Dalakouras et al., 2018). In another study, Koch et al. (2016) analyzed the detection of spray-applied dsRNAs labeled with the green fluorescent dye (ATTO 488) in barley leaf. Using confocal laser microscopy, the authors detected the green fluorescent signal from fluorescing dsRNAs in the xylem, apoplast, symplast of phloem parenchyma cells, companion cells, and mesophyll cells, along with trichomes and stomata. Furthermore, topically applied naked- and bio-clay loaded dsRNAs labeled with Cy3 were observed in the xylem of Arabidopsis leaves (Mitter et al., 2017), but only bio-clay loaded dsRNAs showed larger uptake into the spongy mesophyll. Song et al. (2018) investigated the cellular uptake of spray-applied dsRNAs into wheat cells using healthy and wounded coleoptiles. The results indicated that dsRNA uptake was more efficient through the wounded surface than the healthy surface. Based on microscopic analyses, it was hypothesized that exogenous dsRNAs transferred via tracheary elements after entering into the damaged cells of the wounded coleoptiles.

The exact mechanisms underlying perception, recognition, and translocation of exogenously applied dsRNAs into the plant cell are still unknown. Plant-microbe interactions mediated extracellular DNA (eDNA) or RNA (eRNA) perception by plants crucially regulates self- and non-self-recognition and induces pattern-triggered immunity (PTI) (Niehl et al., 2016; Bhat and Ryu, 2016). It is proposed that the biomolecular markers of microbe- and pathogen-associated molecular patterns (MAMPs and PAMPs) in plants are perceived by cell-surface proteins called pattern-recognition receptors (PRRs). Upon recognition, signal transduction cascades are triggered by PRRs to induce the plant's innate immunity system, also called PTI (Niehl et al., 2016; Bhat and Ryu, 2016). Indeed, RNAs served as MAMPs and induced PTI responses when exogenously applied to Arabidopsis (Lee et al., 2016). According to this study, Arabidopsis leaves pre-infiltrated with total RNAs purified from Pseudomonas syringae pv. tomato DC3000 (Pto DC3000), elicited plant immune responses similar to those induced against Pto DC3000 bacterium, suggesting that total bacterial RNAs could trigger plant innate immunity responses. Yakushiji et al. (2009) investigated the elicitor activity of bacterial DNA in Arabidopsis and the results indicated that non-methylated CpG DNAs served as MAMPs and induced defense responses. Although the mechanisms underlying the recognition and cellular uptake of extracellular DNA by receptors have not been identified in plants, a study reported by Niehl et al. (2016) showed that purified dsRNAs from virus-infected plants and synthetic dsRNA analogs, both induced PTI responses in Arabidopsis. Exogenous dsRNA-induced PTI responses were dependent on the co-receptor SOMATIC EMBRYOGENESIS RECEPTOR KINASE 1 (SERK1) but were DCL- independent. Thus, it was proposed that membrane-bound SERK1could act as a potential dsRNA receptor. In this regard, global transcriptomics and proteomics analysis may help to screen trans-membrane marker proteins and genes for elucidating the receptor-mediated recognition, perception, and uptake of dsRNAs into plant cells.

The principal role of dsRNAs is to trigger RNAi machinery in the plant system which involves several processing steps ( Figure 2 ). In plant cells, dsRNAs are first processed into small nucleotide primary siRNAs by DCL endonuclease. Four paralogues of DCL including DCL1, DCL2, DCL3, and DCL4 have been found in A. thaliana (Bologna and Voinnet, 2014). Among them, DCL2, DCL3, and DCL4 process dsRNAs into 22-, 24-, and 21- nucleotide siRNAs, respectively. These primary siRNAs are then incorporated onto AGO proteins to form RISC. Generally, the 21-nt siRNAs are loaded on AGO1, and target complementary mRNA transcripts for cleavage and degradation, resulting in “PTGS” or RNAi (Dalakouras et al., 2020). In Arabidopsis, among 10 identified AGO genes, AGO1 mainly found to initiate PTGS whereas, AGO4 mediates transcriptional gene silencing (TGS) (Bologna and Voinnet, 2014). On the other hand, when 22-nt siRNAs are loaded on AGO1, they recruit RNA DEPENDENT RNA POLYMERASE 6 (RDR6) to the target RNA transcript and transcribing it to dsRNAs. These newly synthesized dsRNAs lead to the generation of secondary siRNAs and the amplification of silencing signals, in a mechanism coined “transitive” silencing (Chen et al., 2010; Cuperus et al., 2010). A study conducted by Dalakouras et al. (2016) investigated the effects of size and structure of exogenous sRNAs (21-, 22-, and 24-nt sRNAs either as a perfect duplex or as sRNAs containing an asymmetric bulge) on local, transitive, and systemic RNAi in GFP-expressing N. benthamiana plant. The results indicated that all sRNAs were able to initiate local RNAi, whereas, only 22-nt sRNAs, in both forms, were able to induce systemic RNAi. DCL2 is responsible for processing dsRNAs into 22-nt siRNAs (Henderson et al., 2006), which then recruit RDR6 to initiate transitivity (Chen et al., 2010; Cuperus et al., 2010). In a recent study, transcript levels of NPTII and EGFP transgenes were significantly reduced after treating the transgenic Arabidopsis plants with the corresponding dsRNAs of these genes. The induction of the RNAi machinery in treated plants has become evident after the detection of EGFP-derived siRNA in the treated plants. However, what was surprising is that EGFP-derived siRNA was also detected in the plants treated with NPTII-dsRNA indicating that NPTII silencing could also be mediated by silencing transitivity. Indeed, the analysis of read-through transcripts of both NPTII and EGFP in treated and untreated transgenic Arabidopsis indicated that silencing transitivity has probably taken place even before treatments by dsRNAs, but was enhanced after dsRNA treatments. Supporting this assumption is the findings that NPTII transcript levels were downregulated in plants treated with EGFP-dsRNA and vice versa (Dubrovina and Kiselev, 2019).

DsRNAs can also be processed by DCL3 to produce 24-nt siRNAs. When 24-nt siRNAs are loaded on AGO4, they recognize cognate DNAs and recruit Pol V (a DNA dependent RNA polymerase) to form “RNA-directed DNA methylation” (RdDM) (Chen et al., 2010). Actually, RdDM is necessary for the de novo cytosine methylation primarily within the region of RNA-DNA sequence identity. In plants, DNA methylation can occur in all three sequence contexts (symmetric CG, CHG, and asymmetric CHH), and the RdDM pathway was found to methylate all sequence contexts (Singroha and Sharma, 2019). DsRNA induced RNAi machinery stimulates not only RdRM but also histone modifications which play an important role in epigenetic transcriptional gene silencing (Wassenegger, 2005). According to the study conducted by Dubrovina and Kiselev (2019), the exogenously applied dsRNAs targeting EGFP and NPTII in transgenic Arabidopsis considerably increased the cytosine DNA methylation at three different contexts of CG, CHG, and CHH. This study demonstrated that DNA methylation at the coding sequences of the EGFP and NPTII transgene could reflect the influence of 24-nt siRNAs. There are possibilities that exo-dsRNAs treatment initiates inductions and spreading of DNA methylation at the T-DNA regions bearing both EGFP and NPTII transgene, which could promote the transgene mRNA degradation as well as affect transcription of the transgene and/or heterochromatin formation (Dubrovina and Kiselev, 2019).

sRNAs can move short and long-distance throughout the plant cells (Sarkies and Miska, 2014). Primary siRNAs are able to spread for short-range (10–15 cells), through the symplastic route, without producing secondary siRNAs (Kim, 2005). Symplastic movements of sRNAs require the passage through the plasmodesmata. The size exclusion limit of plasmodesmata is around 30 to 50 kDa which should not limit the passing of naked sRNAs, but it might affect the transport of sRNAs that are enclosed in vesicles and/or bound to RNA binding proteins (Wang and Dean, 2020). During the development of plant organs, plasmodesmata can change their size and selectivity that may allow the passage of sRNAs (Imlau et al., 1999). Long-distance movement or systemic silencing has been found to be phloem-mediated that requires amplification of the silencing signals by RDRPs (Chen et al., 2010; Cuperus et al., 2010; Liang et al., 2012; Dalakouras et al., 2018). The generation of secondary siRNAs is led by transitive silencing cascades and amplification of silencing. In Arabidopsis, DCL2 is known to play an essential role in the accumulation of secondary siRNAs and silencing transitivity (Chen et al., 2010; Cuperus et al., 2010). In vascular plants, phloem serves as a highway for mobile signals. Several lines of evidence indicated that the symplastic signaling follows the photoassimilate translocation route from source to sink tissues, which involves reaching the companion cells (specialized cells of the phloem tissue) through plasmodesmata. Then signal molecules could transfer from here to the other phloem‐specific types of cells called sieve elements. The sieve elements are known as end‐to‐end connected enlarged cells which create a channel for fast communication to distant organs (Mermigka et al., 2016). The phloem-based source to sink movement of the systemic sRNAs signal is also reported by some studies (Pant et al., 2008; Liang et al., 2012; Zhang et al., 2014; Patil and Fauquet, 2015). Different types of RNAs have been found in the phloem exudate including viral RNAs, siRNAs, miRNAs, transfer RNAs, and messenger RNAs (Kehr and Buhtz, 2008). For sRNAs movement throughout the plants, RNA binding proteins play important roles (Wang and Dean, 2020). However, the exact mechanism underlying the functions of sRNAs derived from exogenous dsRNAs in mobile signaling is still largely unknown.

Exogenous applications of dsRNAs have been reported to induce plant resistance against pathogens by activating the RNAi machinery (Tenllado and Diaz‐Ruiz, 2001; Numata et al., 2014; Kamthan et al., 2015; Wang et al., 2016; Guo et al., 2016; Mitter et al., 2017; Dubrovina and Kiselev, 2019; Dalakouras et al., 2020). In this section, inhibitory activities of exogenous dsRNAs on fungi and viruses are summarized.

To control fungal diseases in the plant, spray-induced gene silencing (SIGS) is currently considered as an innovative, eco-friendly, biological tool that involves an exogenous spray of dsRNAs or siRNAs onto plant surface. Exogenously applied dsRNAs and siRNAs have been reported to protect several plant species including barley, tomato, strawberry, grape, oilseed rape, wheat, onion, rose, lettuce, cucumber, soybean, and Arabidopsis against several fungi such as Fusarium graminearum, Botrytis cinerea, Sclerotinia sclerotiorum, and Fusarium asiaticum (Koch et al., 2016; Wang et al., 2016; McLoughlin et al., 2018; Song et al., 2018; Gu et al., 2019).

According to Koch et al. (2016), spray applications of dsRNAs and siRNAs onto barley detached leaves attenuated fungal diseases by inhibiting fungal growth and suppressing three fungal cytochrome P450 genes, i.e., CYP51A, CYP51B, CYP51C, of F. graminearum. In another study, foliar applications of dsRNAs onto oilseed rape and Arabidopsis leaf surface exhibited antifungal potential against S. sclerotiorum and B. cinerea (McLoughlin et al., 2018). A recent study conducted by Song et al. (2018) also found that spraying of dsRNAs targeting the Myo5 gene of F. asiaticum resulted in reduced fungal virulence in wheat. However, the RNAi effect was maintained only when the continuous supply of dsRNA was provided. Based on sRNA deep sequencing analysis, it was revealed that F. asiaticum was not able to amplify secondary siRNAs, whereas, Myo5-dsRNA derived siRNAs were detected in plant cells. These results indicate that in fungi, RNAi is not maintained by RDRP amplification loop (Song et al., 2018). Exogenous applications of dsRNAs and siRNAs targeting DCL1 and DCL2 genes of B. cinerea was found to reduce fungal virulence in fruits (e.g., tomato, strawberry, and grape), vegetables (e.g., lettuce and onion), and flower petals (e.g., rose) (Wang et al., 2016). Exogenously applied dsRNAs and siRNAs can induce resistance against fungi, either indirectly through their uptake by host plant cells and then their introduction to fungal cells, or directly through fungal cells that uptake dsRNA/siRNA, leading to silencing of target genes (Koch et al., 2016; Wang et al., 2016; Song et al., 2018). Based on these abovementioned findings, it is reasonable to conclude that dsRNAs/siRNAs could be used as biofungicides to control phytopathogenic fungi, pre- and post-harvest. However, more investigations are still needed to find out suitable/stable application methods, delivery techniques, effective time periods, uptake regulatory factors, and mechanisms underlying the translocation of dsRNAs/siRNAs between plant and fungal cells.

Several studies reported that foliar applications of dsRNAs induce plant resistance against target viruses (Tenllado and Diaz‐Ruiz, 2001; Carbonell et al., 2008; Konakalla et al., 2016; Kaldis et al., 2018; Worrall et al., 2019; Vadlamudi et al., 2020). The effects of exogenously applied dsRNAs on conferring resistance against viruses have been reported in various host species, including tomato, tobacco, maize, papaya, cowpea, cucumber, watermelon, and squash against different viruses such as tobacco etch virus (TEV), tobacco mosaic virus (TMV), alfalfa mosaic virus (AMV), pepper mild mottle virus (PMMoV), potyvirus, bean common mosaic virus (BCMV), papaya ringspot virus (PRSV), and zucchini yellow mosaic virus (ZYMV) (reviewed in Dubrovina and Kiselev, 2019). These studies demonstrated that dsRNA-treated plants are capable of triggering RNAi-mediated processes to reduce or delay viral infection by silencing target viral genes.

Exogenously applied dsRNAs on plants for antiviral effects were first reported by Tenllado and Diaz‐Ruiz (2001). According to this study, dsRNAs targeting the replicase protein (RP) gene of PMMoV, TEV, and AMV attenuated viral infections when introduced in tobacco leaves along with the target virus. Other studies demonstrated that exogenously applied dsRNAs were effective in conferring protection against TMV in tobacco (Yin et al., 2009; Konakalla et al., 2016; Niehl et al., 2018); PMMoV in tobacco (Tenllado et al., +2003); ZYMV in cucumber, watermelon, and squash plants (Kaldis et al., 2018). A recent study conducted by Vadlamudi et al. (2020) demonstrated that the topical application of dsRNAs molecules derived from both CP and HC-Pro genes of the PRSV-Tirupati isolate conferred resistance to papaya plants against PRSV-Tirupati and PRSV-Delhi viral isolates. The results found that the dsRNA molecules conferred 100% resistance against PRSV-Tirupati infection and the same dsRNA molecules were highly effective against the PRSV-Delhi isolate conferring resistance of 94% and 81%, respectively. The great concern regarding the possible instability of naked-dsRNA applications could account for the short-term protection against viral pathogens. To address such a concern, a recent study demonstrated that a single application of LDH-loaded dsRNAs effectively provided RNAi-mediated virus protection and its effect lasted for at least 20 days in cowpea leaves (Mitter et al., 2017). According to this study, the exogenous spray of LDH-loaded dsRNAs in tobacco and cowpea was found to induce resistance against PMMoV and CMV by targeting RP and 2b suppressor genes, respectively (Mitter et al., 2017). In a subsequent study, spraying of LDH-loaded dsRNA on tobacco and cowpea was also found to provide resistance to BCMV infections (Worrall et al., 2019). Unfortunately, the efficacy of exogenously applied dsRNAs against viral pathogenes have largely been investigated under controlled experimental conditions. Therefore, it is still unclear if pretreating the plants with dsRNAs to prevent the subsequent viral infection or using dsRNAs to treat the existing viral infection could be commercially feasible under open field conditions.