RNA-Spray-Mediated Silencing of Fusarium graminearum AGO and DCL Genes Improve Barley Disease Resistance

Over the last decade, several studies have revealed the enormous potential of RNA-silencing strategies as a potential alternative to conventional pesticides for plant protection. We have previously shown that targeted gene silencing mediated by an in planta expression of non-coding inhibitory double-stranded RNAs (dsRNAs) can protect host plants against various diseases with unprecedented efficiency. In addition to the generation of RNA-silencing (RNAi) signals in planta, plants can be protected from pathogens, and pests by spray-applied RNA-based biopesticides. Despite the striking efficiency of RNA-silencing-based technologies holds for agriculture, the molecular mechanisms underlying spray-induced gene silencing (SIGS) strategies are virtually unresolved, a requirement for successful future application in the field. Based on our previous work, we predict that the molecular mechanism of SIGS is controlled by the fungal-silencing machinery. In this study, we used SIGS to compare the silencing efficiencies of computationally-designed vs. manually-designed dsRNA constructs targeting ARGONAUTE and DICER genes of Fusarium graminearum (Fg). We found that targeting key components of the fungal RNAi machinery via SIGS could protect barley leaves from Fg infection and that the manual design of dsRNAs resulted in higher gene-silencing efficiencies than the tool-based design. Moreover, our results indicate the possibility of cross-kingdom RNA silencing in the Fg-barley interaction, a phenomenon in which sRNAs operate as effector molecules to induce gene silencing between species from different kingdoms, such as a plant host and their interacting pathogens.

RNAi constructs differ also in the length of the dsRNA molecules that were applied by the 105 topical spray of barley leaves.  The construction of pGEMT plasmids comprised of the tool-and manually-designed target 120 sequences was performed using restriction enzyme-cloning strategies. The first step in 121 constructing pGEMT plasmids containing manually-designed double targets was to amplify 122 target sequences of AGO1, AGO2, DCL1 and DCL2 from the confirmed plasmids with primers 123 containing restriction sites (Table S1). The manually designed dsRNA targeting FgAGO1 and FgAGO2 had a length of 1529 bp and was therefore named ago1/ago2_1529nt. According to 125 this scheme the other manually-designed dsRNAs were named ago1/dcl1_1570nt, 126 ago1/dcl2_1528nt, ago2/dcl1_1783nt, ago2/dcl2_1741nt and dcl1/dcl2_1782nt. Briefly, an 127 AGO2 PCR fragment was inserted between NotI and NdeI restriction sites of pGEMT plasmids 128 containing AGO1 or DCL1 target sequences to generate ago1/ago2_1529nt and 129 ago1/dcl2_1528nt constructs. The PCR fragment of AGO1 was inserted between NotI and NdeI 130 restriction sites of pGEMT plasmids containing the DCL1 target sequence to construct 131 ago1/dcl1_1570nt target plasmid. The other manually designed constructs (ago1/dcl2_1528nt, 132 ago2/dcl2_1741nt and dcl1/dcl2_1782nt) were generated following the same procedure as 133 described above: DCL2 PCR fragments were inserted in the AGO1 background (using NotI 134 and NdeI), in AGO2 (using NotI and BstXI) and in DCL1 (using NotI and SalI). To construct 135 pGEMT plasmids containing tool-designed target sequences (ago1/ago2_365nt, 136 ago1/dcl1_355nt, ago2/dcl1_374nt, ago1/dcl2_366nt), the single targets were amplified using 137 primers containing a restriction site (Table S1), as described above. A tool-designed sequence 138 of DCL1 was inserted between NotI and SalI restriction sites of the pGEMT plasmid containing 139 AGO1 and AGO2 targets to generate ago1/dcl1_355nt and ago2/dcl1_374nt constructs, 140 respectively. The DCL2 fragment was inserted between the NotI and SalI restriction sites of the 141 pGEMT plasmid containing the AGO1 sequence to construct ago1/dcl2_366nt. Finally, AGO2 142 was inserted between the NotI and SalI restriction sites of the pGEMT plasmid containing the 143 AGO1 target sequence to generate an ago1/ago2_365nt construct.

144
MEGAscript Kit High Yield Transcription Kit (Ambion) was used for dsRNA synthesis by 145 following the manufacturers' instructions using primers containing a T7 promoter sequence at 146 the 5′ end of both forward and reverse primers (Table S1).   (Table S1). After an initial activation step at  The constructs ago1/dcl1_1570nt and ago1/dcl2_1528nt reduced FgAGO1 transcripts by only 215 17% and 29%, respectively ( Figure 2B). Analysing the transcript levels of FgAGO2 revealed 216 that (1) the silencing efficiencies of ago2/dcl1_1783nt and ago2/dcl2_1741nt were higher than 217 FgAGO1 target silencing and (2) targeting both FgAGO genes with the ago1/ago2_1529nt 218 construct resulted in 50% reduction for FgAGO1 and 62% for FgAGO2. This, therefore, 219 showed the highest overall FgAGOs gene silencing ( Figure 2B).

220
Interestingly and consistent with the tool-designed target gene silencing results, we detected the 221 strongest reduction of >70% for FgDCL1 ( Figure 2B). For example, ago2/dcl1_1783nt-dsRNA 222 10 provoked a 79% reduction of FgDCL1 transcripts. Target gene silencing for FgDCL2 was also 223 highly efficient, as use of all three constructs, ago1/dcl2_1528nt, ago2/dcl2_1741nt and 224 dcl1/dcl2_1782nt, resulted in an approximately 60% silencing efficiency ( Figure 2B). The most 225 efficient construct in terms of overall target gene silencing was dcl1/dcl2_1782nt, which 226 reduced the transcripts of FgDCL1 and FgDCL2 by 78% and 58%, respectively, compared to 227 control. Overall, these results suggest that silencing conferred by AGO-and DCL-dsRNAs  (Table 1). To assess whether tool-designed dsRNA is more efficient than manually designed constructs, 234 we directly compared target gene-silencing efficiencies of both design approaches (Figure 3). 235 We observed that target gene silencing of manually-designed constructs was superior to tool-  (Table 2).

242
For the manually-designed dsRNA, we calculated siRNAs that were 4-to 10-fold more efficient 243 compared to the tool-designed constructs (Table 2), thus underlining that the dsRNA precursor 244 length probably plays a role in determining the number of derived siRNAs. For example, we 245 predicted 49 efficient siRNAs were derived out of the 912-nt manually-designed dsRNA, which 246 targets FgDCL1, which is 10-fold greater than the 5 siRNA hits derived from the 182-nt tool-247 designed FgDCL1-dsRNA (Table 2). Notably, these differences resulted in only an overall 10% 248 11 silencing efficiency decrease of the tool-designed dsRNA compared to the manually-designed 249 constructs targeting FgDCL1 (Table 2). Together, these data suggest that longer dsRNAs result 250 in a higher number of efficient siRNAs, but there is no stringent correlation that reflects the 10-251 fold higher number of siRNA resulting in a 10-fold increase in target gene silencing (Table 2).  Based on these findings, the dsRNAs tested in this study were designed to target FgAGO and field test conditions. However, RNAi-based plant protection technologies are limited by the 300 uptake of RNAi-inducing trigger molecules, either siRNAs and/or dsRNAs, whereas both RNA 301 types have been shown to confer plant disease resistance independent of how they were 302 applied/delivered (i.e. endogenously or exogenously).

303
Previously, we discovered that longer dsRNAs of 400-800 nt exhibited a higher gene-silencing  (Table 2). Notably, we found that the number of efficient siRNAs derived from 313 the longer, manually-designed dsRNAs was 4-to 5-fold higher for the constructs that target FgDCL2 resulted in 10-fold more efficient siRNAs than the tool-designed versions (       Asterisks indicate statistical significance (**p<0,01; ***p< 0,001; students t-test).   with three 20µl droplets of Fg (50.000 spores/ml). The pictures were taken 5dpi.