Artificial trans-kingdom RNAi of FolRDR1 is a potential strategy to control tomato wilt disease

Tomato is cultivated worldwide as a nutrient-rich vegetable crop. Tomato wilt disease caused by Fusarium oxysporum f.sp. Lycopersici (Fol) is one of the most serious fungal diseases posing threats to tomato production. Recently, the development of Spray-Induced Gene Silencing (SIGS) directs a novel plant disease management by generating an efficient and environmental friendly biocontrol agent. Here, we characterized that FolRDR1 (RNA-dependent RNA polymerase 1) mediated the pathogen invasion to the host plant tomato, and played as an essential regulator in pathogen development and pathogenicity. Our fluorescence tracing data further presented that effective uptakes of FolRDR1-dsRNAs were observed in both Fol and tomato tissues. Subsequently, exogenous application of FolRDR1-dsRNAs on pre-Fol-infected tomato leaves resulted in significant alleviation of tomato wilt disease symptoms. Particularly, FolRDR1-RNAi was highly specific without sequence off-target in related plants. Our results of pathogen gene-targeting RNAi have provided a new strategy for tomato wilt disease management by developing an environmentally-friendly biocontrol agent.

To assess the potential regulated gene by impaired FolRDR1, the miRNA levels were evaluated by sRNA-seq using the KO-strains FolRDR1-KO-36 (named as FolRDR1-1 in library), FolRDR1-KO-126 (named as FolRDR1-2 in library) and wild type strain (named as Fol-WT in library). The DEGs of miRNAs were listed in S2 Table. Well correlation was showed between  Table), and the results showed that knockouting of FolRDR1 mainly affected the metabolic pathway in both KO strains (S4 Fig). With above results, we concluded that the levels of miRNA were correlated with FolRDR1 in Fol.
We further checked the growth rate of wild type (WT) Fol and FolRDR1-KO strains cultured on PDA plates, respectively. Statistic data showed that no significant difference was observed between Fol and FolRDR1-KO strains (S5 Fig). Further, no significant difference in colony morphology was observed either under salt, alkali and osmotic pressure stress at different concentrations (S6 Fig). The above results indicated that the absence of FolRDR1 did not change the response to abiotic stresses in Fol. However, we found that the colony edge of both FolRDR1-KO strains were more loose than Fol. Furthermore, the mycelia of FolRDR1-KO strains presented abnormal growth such as mycelia ablation and increased sclerotia (Fig 1A). To filamentous fungi, sporulation ability is an important physiological index to measure the pathogenicity. Knocking out FolRDR1 also leaded to lower sporulation but larger size of conidia compare to WT strain (Fig 1B-1D). Intriguingly, the growth and penetrability of Fol on PDA plate covered with cellophane were nearly unchanged, contrarily, both FolRDR1-KO strains showed dramatic decreased penetrability ( Fig 1E). There results indicated that FolRDR1 was essential to the vegetative growth and conidiogenesis in Fol as well as penetrability.

FolRDR1 is required for pathogenicity in Fol
To investigate the response of FolRDR1upto Fol infection in tomato roots, two-week susceptible cultivar Moneymaker seedlings were infected by Fol. Fusarium wilt symptoms were developed at early stage (7 day post infection, 7dpi) (Fig 2A). Total RNA was extracted from infected tomato roots which included both tomato and pathogen RNA. The transcript level of FolRDR1was further valuated by Northern blot, and the results indicated that FolRDR1was constantly induced during the pathogen infection ( Fig 2B).
To further evaluate the pathogenicity of FolRDR1, tomato seedlings were inoculated with Fol and two FolRDR1-KO strains, respectively. We observed impaired infection of Moneymaker supported by alleviated Fusarium wilt symptoms, less presence of the fungus within the plant stem and fungal mycelium regeneration compared to Fol-treated Moneymaker, while no Fusarium wilt symptoms were observed in resistant cultivar Motelle under infection with all

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease three individual strains (Fig 2C-2E). Based on above results, we concluded that FolRDR1 was a critical pathogenic factor in Fol.

Environmental dsRNA-FolRDR1 is effectively taken up by Fol
To validate whether SIGS of FolRDR1 control Fusarium wilt disease, we generated To trace external FolRDR1-dsRNA, we synthesized fluorescein-labeled FolRDR1-dsRNA using Fluorescein-12-UTP in vitro. Fluorescein-12-UTP and water treatment were used as negative controls. WT conidia were cultured with fluorescein-labeled FolRDR1-dsRNA for 24 hours on PDA plate followed by detecting fluorescence signals. Fluorescence signals of FolRDR1-dsRNAs and GFP-dsRNA were observed in conidia (Fig 3A, left). To further confirm whether the dsRNAs enter Fol cells, the Fol mycelium were cultured with the fluoresceinlabelled FolRDR1-dsRNAs and GFP-dsRNA respectively in liquid culture for 48 hours. Subsequently, conidia were collected for protoplast preparation. Similarly, fluorescence signals of FolRDR1-dsRNAs and GFP-dsRNA were observed clearly in protoplast (Fig 3A, middle). Further, we also observed the fluorescence signals in 48-hour hyphae (Fig 3A, right). Above results A the mycelia of FolRDR1-KO strains presented abnormal growth such as mycelia ablation and increased sclerotia (indicated by red arrow). All three strains were cultured on PDA plates, and images were taken at fourth days. B-D Knocking out FolRDR1 leaded to lower sporulation but larger size of conidia compare to WT strain. * indicates significant difference when compared to WT at P < 0.05, chi-square test, Error bars indicate the Standard Deviation of three replicates. 40 x scale bars, 50 μm, 100 x scale bars, 20 μm. E Knocking out FolRDR1 resulted in dramatic decreased penetrability in Fol. All three strains were cultured on the center of PDA plates covered with half cellophane, and images were taken at fourth days. Front, images were taken from the front of plate. Back, images were taken from the back of plate. Three biological replicates were used in each experiment. https://doi.org/10.1371/journal.ppat.1011463.g001

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease indicated that Fol cells took up dsRNAs from the environment, and this action is not likely to be selective.
To determine whether dsRNA can be effectively transported in Fol mycelium, we applied fluorescein-labeled dsRNA on the center of PDA plate inoculated with Fol. Two glass slides (1 cm x 2 cm) were inserted into the medium about 2 cm far away from the inoculation site to minimize the dissociative dsRNA in the plate (Fig 3B). After 3 days of mycelium expansion, we detected the fluorescence signals in the fungal mycelia climbing on the glass slides ( Fig 3C). The corresponding dsRNA fragments in the fungal mycelia were further detected ( Fig 3D). Fol efficiently uptakes fluorescein-labelled dsRNA. A Conidia were cultured in Vogel's minimal medium at a concentration of 10 5 spores/mL with 150 ng/mL dsRNA for 24 hours (Left, scale bars, 10 μm), and conidia protoplast were made subsequently (Middle, scale bars, 10 μm). Hyphae were collected after 48 hours (Right, scale bars, 1 mm). Samples were treated with micrococcal nuclease (MNase) 30 min before images were taken using the confocal microscopy laser scanner (CMLS). B Two vertical glass slides (1 cm x 2 cm) were inserted into the PDA medium about 2 cm far away from the inoculation site of Fol to minimize the dissociative fluorescein-labeled dsRNA in the plate. Front, images were taken from the front of plate. Back, images were taken from the back of plate. C After 3 days of mycelium growth and expansion, the fluorescence signals in the mycelia climbing the glass slides were checked using CMLS. Scale bars, 1 mm. Three biological replicates were used in each experiment. D The corresponding dsRNA fragments in the marginal mycelia were further detected by RT-PCR. E FolRDR1-dsRNA-derived siRNAs were enriched in both FolRDR1-dsRNA treated fungal mycelia resulting in repressed the transcript levels of FolRDR1. Total ten 30-nt DNA fragments (detailed in S5 Table), which uniformly distributed in the predicted regions in the CDS of FolRDR1, were synthesized and mixed as a pool followed by labeling with [γ-32P]ATP as probes for to detect the enrichment of FolRDR1-dsRNA-derived siRNAs (Up panel). [γ-32P] ATP-labeled specific nucleotide probe sequences of FolRDR1 were used for hybridization to detect the transcript levels of FolRDR1 (Bottom panel). Sly-18s rRNA was used as a loading control in both Northern blots, respectively. https://doi.org/10.1371/journal.ppat.1011463.g003

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease To verify the production of FolRDR1-dsRNA-derived siRNAs, firstly, the efficient siRNAs generated in different regions of FolRDR1 were analyzed using siRNA-Finder (Si-Fi) (http:// www.wheatgenome.info/) (S9 Fig) [19]. Total ten 30-nt DNA fragments (detailed in S5 Table), which uniformly distributed in the predicted regions, were synthesized and mixed as a pool followed by labeling with [γ-32P]ATP as probes for Northern blot. The data clearly shown that FolRDR1-dsRNA-derived siRNAs were enriched in both FolRDR1-dsRNA1 and FolRDR1-dsRNA2 treated fungal mycelia, respectively ( Fig 3E, up panel). Furthermore, the transcript level of FolRDR1 in both FolRDR1-dsRNAs treated fungal mycelia were repressed significantly compared to Fol or water treatment samples (Fig 3E, bottom panel). These results indicated that external dsRNAs were effectively transported in Fol growing-mycelium and repressed the transcript level of FolRDR1.

Application of external FolRDR1-dsRNA destructs the biological functions of FolRDR1 in Fol
Previously, we have shown that FolRDR1 was essential to the vegetative growth and conidiogenesis in Fol. We further explore whether external FolRDR1-dsRNA impair the biological functions of FolRDR1 in Fol. We applied FolRDR1-dsRNA on the center of the plate colony and observed the colony growth at 5 dpi. The data showed that the growth rate of the mycelium were unchanged under the treatments of FolRDR1-dsRNAs and GFP-dsRNA ( Fig 4A  and 4B). To the conidiogenesis, however, both FolRDR1-dsRNA1 and FolRDR1-dsRNA2, but not GFP-dsRNA, significantly suppressed the production of conidia ( Fig 4C). Moreover, the transcript level of FolRDR1 was repressed accordingly at optimal treatment with the concentration of 150 ng/mL and 24 hours in liquid PDA medium (Fig 4D and 4E). We also noticed that the RNAi-based silencing efficiency of two dsRNAs in interference processing was about 90% for FolRDR1-dsRNA1 and 81% for FolRDR1-dsRNA2, respectively ( Fig 4D and 4E), which was partially addressed by the analysis using siRNA-Finder (Si-Fi) (S9 Fig). Above results indicated that FolRDR1-dsRNA1 generated efficient siRNAs more than FolRDR1-dsRNA2, and application of external FolRDR1-dsRNAs effectively destructed the biological functions of FolRDR1 in Fol.

Host plant takes up and transfers environmental FolRDR1-dsRNAs
To evaluate the residual period of the external dsRNA on the host leaves, FolRDR1-dsRNAs and GFP-dsRNA were daubed on the 2-week tomato seedling leaves. The treated leaves, stems and roots were collected at different time points for detecting dsRNA by RT-PCR. All three dsRNAs were detected until 7 dps (day post spray) which indicated that the environmental dsRNA could remain on leaves, stems and roots for at least 7 days without dsRNA selectivity ( Fig 5A).
To verify whether external dsRNA applied on tomato leaf may be effectively transferred in plant tissues, one side leaf was daubed with fluorescein-labeled dsRNA followed by checking the fluoresce signals in undaubed leaf, root and stem. No fluoresce signal was detected in water treated leaf which indicated no background excitation fluorescence in nature tomato leaf. On the other hand, strong fluoresce signals were presented in daubed leaf ( Fig 5B). Then, visible fluoresce signals were detected in undaubed leaf 2 days after treating for all three dsRNAs ( Fig  5C). Intriguingly, relatively strong fluoresce signals were shown in stem and root compared to leaf. In stem, fluoresce signals were obviously emerged in vascular bundle (Fig 5D), and distributed in the entire root, especially in root hair (Fig 5E). Above results demonstrated that tomato plant effectively took up the external FolRDR1-dsRNAs and transferred to the different tissues.

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease

External application of FolRDR1-dsRNA alleviates the development of tomato wilt disease
To assess whether SIGS attenuate Fol infection, we sprayed the Fol pre-treated 2-week tomato seedling with FolRDR1-dsRNAs (200 ng/mL) at 24 hours after infecting with Fol and scaled the development of Fusarium wilt symptoms. At 15 dpi without spraying treatment, susceptible cultivar Moneymakers showed the initial symptoms of tomato wilt disease with cotyledon chlorosis and wilting. Meanwhile, Moneymakers treated with FolRDR1-dsRNA1 and FolRDR1-dsRNA2, respectively, developed severer Fusarium wilt symptoms with euphylla chlorosis and wilting. However, at 25 dpi, Moneymakers treated with or without GFP-dsRNA as negative controls were gradually died, showing severe symptoms of Fusarium wilt. However, the wilt disease symptoms of Moneymakers treated with FolRDR1-dsRNA1 and FolRDR1-dsRNA2 were significantly alleviated (Fig 6A). By staining for the presence of the fungus within the plant stem and fungal mycelium regeneration, we further observed alleviated infection in Moneymaker treated with FolRDR1-dsRNA1 and FolRDR1-dsRNA2, respectively,

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease Fluorescein-labeled dsRNAs were daubed on one side leaf followed by checking the fluoresce signals using CMLS after 24 hours. Scale bars, 1 mm. C The fluoresce signals on undaubed leaf were detected using CMLS 3 days after treatment described above. Scale bars, 1 mm. D The fluoresce signals in stem were detected using CMLS 3 days after treatment described above. Vascular bundles were pointed by white arrows. Scale bars, 1 mm. E The fluoresce signals in root were detected using CMLS 3 days after treatment described above. Vascular bundles were pointed by white arrows. Scale bars, 1 mm. https://doi.org/10.1371/journal.ppat.1011463.g005

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease as well as less presence of the fungus within the plant stem and fungal mycelium regeneration compared to Fol-treated Moneymaker, while no Fusarium wilt symptoms were observed in resistant cultivar Motelle under infection with all three individual strains (Fig 6B). The results of relative biomass of Fol in different sample further supported above impaired infection between different treatments which correlated with the symptoms of Fusarium wilt (Fig 6C  and 6D).
Actually, we performed both pre-treatment ( Fig 6) and after-treatment (S11 Fig) with FolRDR1-dsRNAs experiments. The data indicated that no significant difference was shown between these two treatments. Due to the unpredictable development of fusarium wilt disease in field, we think that application in the early stage of disease might reduce the cost of disease control. Based on these results, we concluded that SIGS of FolRDR1-dsRNAs attenuated tomato Fusarium wilt disease under lab conditions.
To validate the potential off-target of FolRDR1-dsRNAi, the off-targets were searched and predicted in different species using si-Fi algorithm with splitting FolRDR1-dsRNAi trigger sequence (the complement to the target sequence of the corresponding RNA) into all possible MERs [20]. Our data showed that no off-target of FolRDR1-RNAi in Fusarium oxysporum was predicted, and no off-target was predicted in Solanum lycopersicum (tomato), either (S4 Table). We further generated RNA-seq libraries using water (Mock), GFP-dsRNA (negative control), FolRDR1-dsRNA1 treated tomato seedlings to analyze the transcriptom of tomato. The data indicated that no significant changes were found between SIGS-FolRDR1 and two controls (S10 Fig, and all DEGs of mRNA were listed in S6 Table) (The raw sequence data for

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease this study are available in the National Genomics Data Center with accession no. CRA011174, https://bigd.big.ac.cn/gsa/browse/CRA011174). Taken together, our results supported that FolRDR1 could be considered as a suitable candidate for developing biological agent to control tomato wilt disease.

Discussion
Recently, studies illustrated that spraying dsRNAs/sRNAs targeting essential pathogen genes on plant surfaces afforded efficient crop protection, and RNAi-based SIGS strategy of disease control was potentially sustainable eco-friendly alternative to standard chemical pesticides for controlling agricultural losses caused by pests and pathogens [7,21]. Exogenous dsRNA triggering suppression of gene activity in a homology-dependent manner was firstly discovered in Caenorhabditis elegans [22]. Since then, SIGS was known as a powerful, fast, and environmentally friendly strategy to circumvent the problems in creating GMOs [7,13,23,24]. RDR1 (also termed as Rdp1) localizes to all known heterochromatic loci and is required for sense transgene-induced silencing to generate dsRNA molecules in fungi [25,26]. RDRs have mainly been described to be involved in amplification of RNAi in eukaryotes [18]. In the present study, we found that FolRDR1 mediated the invasion to the host plant tomato, and played as an essential regulator in pathogen development and pathogenicity in Fol. However, no similar biological functions of RDR1 were reported in other fungal so far. Even more intriguing, we found that abolishing of FolRDR1 leaded to mycelia ablation and abnormal sclerotia in Fol (Fig 1A), which highlights the importance to study underlying mechanisms.
Host Arabidopsis cells secreted exosome-like extracellular vesicles to deliver sRNAs into fungal pathogen B. cinerea, and transferred host sRNAs induced silencing of fungal genes critical for pathogenicity [27]. SIGS has been shown to be effective on controlling plant disease initialed by taking up external dsRNA. The uptake efficiency of dsRNA was significantly different among various fungi with exception such as Colletotrichum gloeosporioides [13]. Our present results showed that both fungal pathogen Fol and host plant could uptake FolRDR1/GFP-dsRNAs directly from environment without obvious selectivity (Fig 3), and dsRNAs were transferred in different tissues efficiently (Fig 4). These results promote us speculate that fungal pathogens take up external RNA unselectively but species dependently.
More than intriguing, the fluorescence signals of FolRDR1 /GFP-dsRNAs were dominantly localized in the host plant vascular bundles (Fig 5D and 5E). Germination of dormant spores in soil results in adherence and invasion of plant roots by Fol hypha, subsequently, move from the root cortex to the vascular bundles where microconidia spore are produced and disseminated. Using the vascular bundles as transport corridor, endogenetic hypha spreading to aboveground tissues is critical for disease progression for Fol. The characteristic wilt symptoms appear as a result of severe water stress, mainly due to vessel clogging [28]. Further questions need to be answered how these external dsRNAs are taken up by plant and fungal cells directly from environment traveling across the boundaries between organisms of different taxonomic kingdoms.
Previously, Arabidopsis and barley ectopically expressing a double-stranded RNA (dsRNA) targeting three fungal CYP51 genes significantly enhanced plant resistance to Fusarium graminearum species by disrupting fungal membrane integrity, subsequently, spraying detached barley leaves with a 791-nt long CYP3-dsRNAs that contains complementary sequences to CYP51family members prior to fungal infection could effectively inhibit disease and yield much smaller lesions [14,29]. Similarly, externally applying dsRNAs and small RNAs (sRNAs) targeting Dicer-like protein genes DCL1 and DCL2 of B. cinerea on vegetables, fruits, and flower petals could suppress grey mold disease effectively [12]. Here, applying FolRDR1-dsRNAs that target FolRDR1 on the surface of pre-infected tomato seedling leaves significantly inhibited the development of Fusarium wilt (Fig 6). To develop a successful SIGS-based crop protection strategy, several critical aspects must be considered. Firstly, a reasonable duration of efficacy is desired. Our data revealed that the FolRDR1/GFP-dsRNAs could be detected even at 7 dps of the local sprayed site, suggesting either external RNAs were stable for at least seven days on the surface of the leaves and/or remained stable in the plant cells (Fig 5A). Secondly, off-target is another considered factor for eco-friendly alternative to standard chemical pesticides. By bioinformatics prediction, our data showed that FolRDR1-RNAi resulted in no target-specific either in fungal pathogen or host plant (S2 Table, S8 Fig).
Taken together, eukaryotic pathogens, including fungi and oomycetes, cause vast worldwide economic losses in crop annually. Compared to traditional chemical pesticides, our collective data provided solid evidences that FolRDR1-RNAi-SIGS is an advantageous artificial trans-kingdom RNAi-based bio-pesticide to protect tomato from Fusarium wilt disease. Application strategies can be improved by encapsulating with chemical reagents to stabilize the dsRNAs and thus increase the strength and duration of plant protection. Such specific pathogen gene-targeting RNAs represents a new generation of environmentally-friendly fungicides for increasing safety and quality of crop yields to feed the growing population.
The pathogenic fungal strain is Fusarium oxysporum f. sp lycopersici (race 2, FGSC 9935, Fol). Fol was grown on potato dextrose agar medium (PDA) for 5 days at 28˚C with constant light. Spore suspensions were prepared by harvesting cultures in Vogel's minimal medium at a concentration of 10 7 spores/mL. Tomato seedlings were removed from soil, and rinsed with tap water roots were inoculated with Fol spores for 30 min. Water treatment was used as a mock control. All experiments were conducted using three biological replicates.
To assess the relative levels of Fol biomass in tomato tissues, genomic DNA was isolated from tomato tissues using CTAB [33]. The rDNA Intergenic Spacer Region (IGS) of Fol was amplified from genomic DNA using qPCR (Primers listed in S1 Table) as a marker to assess relative fungal biomass [2].

Statistics of spore production
All strains were inoculated on PDA medium for 4 days. 5 mm mycelium piece at the edge of the colony was cut and cultivated in 100 mL of Vogel's liquid medium at 28˚C, 200 rpm for 4 days. Spores were collected using three layers of sterile gauze. After spinning down, the number of spores were counted using hemocytometer under the optical microscope. These experiments were repeated three times and three biological replicates in each experiment.

Determination of fungal penetrability
In this experiment, cellophane was used to mock plant cell wall to test the fungal penetrability. The cellophane was cut to a semicircle piece with a radius of 4 cm and placed on the PDA medium plate. 5 mm mycelium piece at the edge of the colony was cut and cultivated in the center of the plate under light at 28˚C for 5 days. The colony morphology and mycelium morphology were recorded. These experiments were repeated three times and three biological replicates in each experiment.

Total RNA extraction, Northern blotting, quantitative real-time PCR (qRT-PCR)
Total RNA was extracted using the Trizol reagent (#15596026, Invitrogen, CA, USA). Purified RNA was treated with DNase I (Thermo Fisher Scientific, Waltham, Ma, USA). For small RNA gel blots, 40 μg of total RNA was separated on 7 M urea 15% denaturing polyacrylamide gels in Tris/Boric Acid/EDTA (1X TBE), followed by transferring to a nylon N + membrane. For high molecular weight RNA gel blots, 10 μg of total RNA was resolved by electrophoresis using urea polyacrylamide gel electrophoresis (PAGE) and transferred to anylon N + membrane. Gene-specific nucleotide probes (Primers listed in S1 Table, and DNA fragments listed in S5 Table). Gene-specific nucleotide probes (Primers listed in S1 Table) were end-labeled using [γ-32P]ATP (#M0201, New England Biolabs, Ipswich, MA; nucleotide probes were labeled according to the manufacturer's recommendations). Blots were stripped and re-probed using a Sly-18s rRNA nucleotide probe to provide a loading control. All blots were imaged using a PhosphorImager (Molecular Dynamics/GE Life Sciences, Pittsburgh, PA) [30].
For reverse transcriptase-polymerase chain reaction (RT-PCR), first-strand cDNA was synthesized from 1 μg of total RNA using the Superscript III First-Strand Synthesis System (#18080051, Thermo Fisher Scientific, Waltham, Ma, USA) according to the manufacturer's recommendations (Primers listed in S1 Table). Diluted cDNA was used as the template for quantitative RT-PCR (#1708880, Bio-Rad, Philadelphia, PA, USA), using Sly-18s rRNA as the internal control. Differential expression of genes was calculated using the 2 -ΔΔCt method [34].

Construction of FolRDR1 knockout strains
FolRDR1 knockout mutant strains were generated by using the split-marker approach previously described by our laboratory [35]. Briefly, for FolRDR1 knockout vector construction, the upstream flanking sequence, downstream flanking sequence of FolRDR1 and HPH cassette were amplified and purified, followed by transformation into protoplasts of the wild-type strain (Primers listed in S1 Table). The knockout construct was then transformed into protoplasts of Fol. Transformants with the desired genetic changes were identified using site-specific primer pairs (Primers listed in S1 Table).

Synthesis of dsRNA and uptake of fluorescein-labelled dsRNA in vitro
Synthesis of dsRNA in vitro was based on established protocols [13]. Briefly, selected fragments of FolRDR1 were amplified using gene-specific primers and inserted into the pL4440 vector containing double T7 promoter (Primers listed in S1 Table). FolRDR1/GFP-dsRNA was labeled using the fluorescein RNA Labeling Mix Kit following the manufacturer's instructions (#11685619910, Sigma, St. Louis, MO, USA). For confocal microscopy examination of fluorescein-labelled dsRNA uptake by fungal mycelium, 5 μL of 150 ng/μL fluorescent dsRNA was applied to the PDA medium or the microscope slides surface.

Light microscopy studies
To track the fluorescein-labeled FolRDR1/GFP-dsRNA, plant tissues and fungal materials were collected after dsRNA treatment with the concentration of 200 ng/μL and 150 ng/mL, respectively. Images were taken using a Zeiss LSM 710 confocal microscope with a 63/1.2 NA

PLOS PATHOGENS
FolRDR1-RNAi is a potential strategy to control tomato wilt disease C-Apochromat oil immersion objective (Zeiss, Oberkochen, Germany). The relative fluorescent density was analyzed usingImage-pro Plus (Media Cybernetics Inc., Shanghai, China).

Construction of sRNA-seq and RNA-seq libraries and analysis
For sRNA-seq, total RNA of the KO-strains FolRDR1-KO-36 (named as FolRDR1-1 in library), FolRDR1-KO-126 (named as FolRDR1-2 in library) and wild type strain (named as Fol-WT in library) were extracted individually using the TRIzol reagent (#15596026; Life Technologies) according to the manufacturer's recommendations. For each Illumina library, 1 μg total RNA was used, according to the manufacturer's instructions. The libraries were subsequently sequenced using the Illumina HiSeq 2000 (Biomarker Technologies, Rohnert Park, CA, USA). For RNA-seq, two-week-old tomato seedlings were pre-infected with Fol followed by spraying FolRDR1/GFP-dsRNA with the concentration oxcf 200 ng/μL. Three biological replicates were used, with 5 seedlings for each treatment. The leaves were collected and then frozen immediately in liquid nitrogen. Total RNA was extracted described previously.
For individual Illumina library, raw reads were subjected to quality control (QC). After QC, raw reads were filtered into clean reads. All sequence reads were trimmed to remove the low-quality sequences. The sequence data were subsequently processed using in-house software tool SeqQC V2.2. House-keeping small RNAs including rRNAs, tRNAs, snRNAs and snoRNAs were removed by blasting against GenBank (http://www.ncbi.nih.gov/Genbank) servers. The trimmed reads were then aligned to the Fusarium oxysporum and Solanum lycopersicum reference genome respectively using TopHat v2.0.0 and Bowtie v0.12.5 [36] with default settings. The expression levels of miRNAs or mRNAs were normalized to the reads per million (rpm) value for each individual library.

Statistical analysis
Each result was presented as the mean ± standard deviation (SD) of at least three replicate measurements. Significant differences between treatments were statistically evaluated by SD and one-way analysis of variance (ANOVA) using SPSS 2.0 (Chicago, IL, USA). The data for two specific different treatments were compared statistically by ANOVA, followed by Student's T-test if the ANOVA result was significant at p < 0.01. has no effect on the growth of Fol. A All strains were culture on PDA plate, and photographed at different time points. B The growth curve was generated based on the colony diameter. Front, images were taken from the front of plate. Back, images were taken from the back of plate. Three biological replicates were used in each experiment. (TIF) S6 Fig. Knockouting FolRDR1 has no effect on the response to abiotic stress. A All strains were culture on PDA plate with different concentration of NaCl (Left). The growth of colony was scaled at different time points, and the growth curve was generated (Right). B All strains were culture on PDA plate with different pH (Left). The growth of colony was scaled at different time points, and the growth curve was generated (Right). C All strains were culture on PDA plate with different concentration of sorbital (Left). The growth of colony was scaled at different time points, and the growth curve was generated (Right). Front, images were taken from the front of plate. Back, images were taken from the back of plate. Three biological replicates were used in each experiment. a presents no significant differences (p>0.05). were sprayed with FolRDR1-dsRNAs on the leaves respectively, followed by infecting by WT Fol two days later as described previously. Wilt disease symptoms were photographed 2 weeks after inoculation. Front, images were taken from the front of plants. Top, images were taken from the top of plants. B Cotton blue staining results reflect the abundance of Fol in the stem of tomato plants. More intense cotton blue staining correlates with higher levels of Fol (Up panel). The outgrowth of fungi from tomato stems of plants inoculated with the indicated strains on PDA, and images were taken at 2 dpi, respectively (Middle panel). Diseased vascular bundles were checked in longitudinal splitting stem (Pointed by red arrows) (Bottom panel). (TIF) S1