Introduction

Global agriculture faces the increasing pressure from biotic stresses and climate change, which threaten crop productivity and food security1. While approaches like chemical pesticides and breeding for genetic resistance have historically supported pest and disease management, emerging challenges such as pathogen adaptation and increased environmental safety concerns highlight the need for incorporating innovative, sustainable, and environmentally friendly strategies to complementarily address issues like chemical residue persistence in food systems and in the environment2.

Spray-Induced Gene Silencing (SIGS) has emerged as one such promising strategy that leverages the plant’s natural RNA interference (RNAi) mechanism to target specific viral or fungal pathogens by applying pathogen-specific exogenous double-stranded RNA (dsRNA)3. Among the different application methods of exogenous dsRNA, spray seems the most accessible, especially in large-scale usage which allows the farmers to apply the dsRNA in the form of particles4. Upon entry into the plant, the dsRNA is processed into small interfering RNAs (siRNAs) that guide the plant’s defense machinery to degrade the corresponding viral RNA, thus silencing the virus’s replication in a sequence-specific manner as a rapid method complementing conventional breeding and transgenic approaches that often require more time and effort5. Although the role of SIGS in plant virus control has been intensively explored, the delivery method of dsRNA remains a bottleneck6. Due to the instability of naked dsRNA, coupled with its variable and inefficient uptake by plant cells, without an efficient delivery system, the amount of dsRNA reaching the plant cells to trigger the RNAi pathway effectively might be insufficient7.

Nanotechnology offers a solution by developing delivery systems that protect dsRNA molecules and enhance their delivery to plant cells4. Nanoparticles (NPs) can shield nucleic acids from environmental degradation, enhance their cellular uptake, and enable controlled release, thereby overcoming key limitations of SIGS6,7,8. Carbon dots (CDs) are a class of carbon-based economically important nanomaterials with unique optical properties, high water solubility, and low toxicity9, making them suitable for agricultural applications such as seed priming, photosynthetic enhancers, plant stress ameliorators, sensors10, and as dsRNA nanocarriers that can replace chemical pesticides11. Mesoporous silica nanoparticles (MSNs), which feature tunable pore sizes and large surface areas, have been widely used for drug delivery in biomedical fields12, present low cytotoxicity13, and are now being adapted for agricultural purposes for pesticide14 or siRNA delivery in plants13,15.

In this study, we aimed to apply NP-enhanced SIGS for the control of two important plant viruses; the RNA Potyvirus: turnip mosaic virus (TuMV) and the DNA Curtovirus: beet curly top virus (BCTV), which cause substantial economic damages16. TuMV has a broad host range of over 300 plant species, including economically important vegetable crops, ornamentals, and weeds. Importantly, TuMV, unlike other potyviruses, is capable of infecting brassicas. The virus is transmitted by aphids in a non-persistent manner and exhibits high genetic diversity due to frequen recombination events, further complicating the development of resistant crop varieties17. BCTV poses a serious threat to sugar beet production, leading to stunted growth, leaf curling, and yield losses18. This DNA virus is transmitted by the beet leafhopper, and once inside the plant, it replicates in the phloem tissue19. Both the severe strain of BCTV (BCTV-Svr) and beet curly top Iran virus (BCTIV) are the causal agents of beet curly top disease in Iran where they are widespread20. Controlling these viruses is challenging due to their efficient insect transmission, which complicates traditional management strategies such as chemical insecticides application or the breeding of resistant cultivars21. It has been revealed that Nicotiana benthamiana and Arabidopsis transgenics expressing the coat protein (CP) gene of TuMV exhibited reduced or delayed symptom expression22,23. In addition, it has been reported that BCTV-infected N. benthamiana plants, carrying a transgenic subgenomic viral segment, showed symptom relief and a decrease in viral load24. Genome editing techniques also offer alternatives for seeking virus resistance to BCTV25. In a previous work, we reported that the control of cucumber green mottle mosaic virus (CGMMV), an RNA virus, was effective under greenhouse conditions, when dsRNAs were applied by an airbrush26. However, this treatment was ineffective against the DNA virus, tomato leaf curl New Delhi virus (ToLCNDV), under our experimental conditions. Notably, high doses of naked dsRNA delivered to plants in a way similar to mechanical plant virus inoculation have been recently reported to be efficient to protect zucchini against ToLCNDV27.

We hypothesized that dsRNA loaded on CDs- or Polyaminated MSNs (PMSNs) nanoparticles as nanocomposites would improve its stability, uptake, and silencing efficiency. First, the dsRNA loading on NPs was optimized on N. benthamiana and cucumber and then these nanocomposites were evaluated for controlling TuMV and BCTV infections in N. benthamiana.

Results

Characterization of the unloaded nanoparticles

Two types of nanoparticles were obtained in this study: (1) carbon dot-based nanoparticles (CDs), prepared and characterized as previously reported with some modifications28, and (2) mesoporous silica nanoparticles (PMSNs), synthesized as described in the Methods section. An average size of 5 nm and a ζ potential of + 11.94 mV were observed in the CDs (Fig. S1). TEM observation of a population of about 400 PMSNs determined an average diameter of 89.07 ± 14.71 nm (Fig. 1A,B). The size distribution histogram indicated that 86% of the nanoparticles had diameters ranging between 70 and 115 nm, fitting well to a normal distribution pattern (Fig. 1C). Dynamic light scattering (DLS) measurements were performed to determine the hydrodynamic diameter and surface charge of the nanoparticles at each stage of functionalization (Table 1). The DLS analysis revealed remarkable changes in nanoparticles’ surface charge, confirming successful functionalization. Bare mesoporous silica nanoparticles (BMSNs) exhibited a negative surface charge (-20.3 mV), which shifted to a positive value (+ 2.95 mV) following amin-functionalization (AMSNs). Further functionalization with carboxyl groups resulted in a return to a negative surface charge (-31.5 mV) for the carboxyl-modified silica nanoparticles (CMSNs). The final step, involving the loading of PEI polymeric groups onto CMSNs, resulted in a surface charge increase to + 22.55 mV, indicating successful formation of PMSNs, necessitated for the loading of the nucleic acids. The resulting PMSNs also offered a good polydispersity index (PdI) (Table 1). A steady increase in hydrodynamic diameter was observed with each functionalization step. The hydrodynamic diameter of PMSNs measured by DLS was 309.1 nm, which was 3.5-fold larger than the diameter obtained from TEM images (89.07 nm) (Table 1).

Fig. 1
Fig. 1
Full size image

Transmission electron microscope (TEM) images of PMSNs at different magnifications (A,B) and normal distribution histogram of PMSNs nanoparticle diameter size (C).

Table 1 Hydrodynamic size, polydispersity index (PdI) and surface charge of the different MSN-derived nanoparticles.
Full size table

The FT-IR spectra of the MSNs confirmed the successful incorporation of functional groups at each stage of the synthesis process (Fig. 2). BMSNs displayed peaks corresponding to Si-OH (970 cm−1), Si-O-Si (1113 cm−1), H2O (1650 cm−1), and O-H (3478 cm−1)29,30. Upon functionalization with amino groups, the FT-IR spectrum of AMSNs exhibited peaks corresponding to N-H bonding (1580–1650 cm−1), C-N-N (1659 cm−1), C-H (2835–2953 cm−1), and N-H stretch (3300–3500 cm−1). After further modification with carboxyl groups, peaks related to COOH absorption (1413 cm−1), C = O vibrations (1750–1700 cm−1), and amide C = O strain (1690–1630 cm−1) were observed in CMSNs. Finally, for PMSNs, the FT-IR spectrum exhibited prominent -N-H and -C-H stretching bands at 3300–3500 cm−1 and 2835–2953 cm−1, respectively, confirming the successful loading of PEI. The FT-IR spectra of dsRNA: PMSNs nanocomposites is presented in Fig. S2.

Fig. 2
Fig. 2
Full size image

FT-IR spectra of PMSNs and its precursors. Bare mesoporous silica nanoparticles (BMSNs), amine-functionalized MSNs (AMSNs), carboxy-functionalized MSNs (CMSNs), and PEI-functionalized MSNs (PMSNs).

The XPS survey spectra of CMSNs and PMSNs shown in Fig. 3 revealed the presence of characteristic peaks corresponding to the expected elements present on the nanoparticles’ surfaces. The intense C 1s peak, observed at around 284 eV in the PMSNs compared with the CMSNs, should correspond to PEI carbon atoms. The asymmetric deconvoluted C 1s in the CMSNs was resolved into three peaks corresponding to C-C and C-N bonds of the functionalization. Two peaks were resolved in the deconvolution of C 1s of the PMSNs (Fig. 3F). One peak at 283.24 eV was predominant (96.3% of the area) corresponded plausibly to C-C bonds of PEI and the other peak, originating from C-N bonds was observed at 285.67 eV showed clear differences in the CMSNs and PMSNs (Fig. 3B,F). The contribution of the amine associated oxygen that could be expected in PMSNs was negligible. The N 1s peak at approximately 400 eV is a key indicator of nitrogen atoms from the amine groups present in PEI. The corresponding peak was almost negligible in the CMSNs (Fig. 3A,C) but evident in the PMSNs (Fig. 3E,G). The deconvolution of the N 1s showed that most of the peak (73%) at 397.21 eV corresponded to primary C-N bonds and the secondary amines were at 398.76 eV (Fig. 3G). This confirmed that the PEI coating contains nitrogen-rich functional groups. The dominant O 1s peak at approximately 532 eV is attributed to oxygen atoms, which likely originate from the silica core. Deconvolution of O 1s showed that Si-OH bonds represented 97.9% and 95.45% for CMSNs and PMSNs, respectively (Fig. 3D,H). The contribution from Si-O-Si bonds resulted, respectively, 2.11% and 4.55% for CMSNs and PMSNs. Additionally, peaks corresponding to silicon, including Si 2p (~ 100 eV) and Si 2s (~ 150 eV), were observed, indicating the presence of the silica core beneath the polymer layer.

Fig. 3
Fig. 3
Full size image

XPS spectra of the CMSNs and PMSNs. High resolution deconvoluted spectra of the C 1s, N 1s and O 1s regions are shown on the right of the respective survey spectra.

The nitrogen adsorption/desorption isotherms of the BMSNs and PMSNs exhibited in both cases Type IV profiles with distinct hysteresis loops, which is a characteristic of mesoporous materials (Fig. 4A,D). At low relative pressures (p/p0 < 0.3), linear regions were observed in the BET plots, indicating monolayer adsorption on the surface of those materials. From this region, a specific surface area of approximately 639.4 m2/g was calculated using the BET equation for the BMSNs, distributed as follows: the micropore area was 393.6 cm2/g and the external area 245.8 cm2/g. The BJH pore size distribution plot revealed that the material has an adsorption average pore size of 3.06 nm, as indicated by a sharp peak in dAp/dlog(w) similar to the estimation from the BET analysis that resulted in 3.6 nm. This result confirmed that the BMSNs were predominantly mesoporous. Beyond this range, minimal contributions from larger mesopores and macropores (rp > 10 nm) were observed, as evidenced by a flattening of the curve at higher pore radii (Fig. 4B). The adsorption cumulative pore volume derived from BJH analysis resulted in 0.47 cm3/g that aligned well with the total adsorption volume obtained from BET analysis (0.32 cm3/g), further validating the mesoporous nature of the BMSNs (Fig. 4C). For the PMSNs adsorption/desorption isotherms (Fig. 4D), the BET surface area was 499.66 m2/g, distributed as 171.78 m2/g and 327.88 m2/g for the micropore and external surface areas, respectively. The BJH adsorption average pore width resulted in 3.2 nm (Fig. 4E) and the BJH adsorption cumulative volume of pores was 0.38 cm3/g, compared with the adsorption volume of 0.13 cm3/g from the BET analysis (Fig. 4F). As for the BMSNs, the PMSNs showed minimal contribution from pores larger than 10 nm, indicating their primarily mesoporous character. All parameters measured in BET and BJH analysis are presented in Table S1.

Fig. 4
Fig. 4
Full size image

Nitrogen adsorption-desorption analysis of BMSNs and PMSNs. (A) N2 adsorption-desorption isotherm of BMSNs exhibiting Type IV isotherm with H1 hysteresis loop characteristic of ordered mesoporous materials. BMSNs TEM image is embedded. (B) BJH pore size distribution of BMSNs showing cumulative surface area (blue line, left axis) and differential surface area distribution (red line, right axis). (C) BJH pore volume analysis of BMSNs displaying cumulative pore volume (blue line, left axis) and differential pore volume distribution (red line, right axis). (D) N2 adsorption-desorption isotherm of PMSNs displaying Type IV isotherm with enhanced H2-type hysteresis loop indicating complex pore structure after functionalization. PMSNs TEM image is embedded. (E) BJH pore size distribution of PMSNs showing reduced surface area and shifted pore size distribution compared to BMSNs. (F) BJH pore volume analysis of PMSNs revealing decreased pore volume and modified pore architecture resulting from PEI functionalization.

DsRNA-loaded characterization and in vitro adsorption studies.

After characterization of the nanoparticles, the interaction between CDs and PMSNs and dsRNA was investigated. Migration of dsRNA in electrophoresis was delayed when increasing the CDs ratio in the dsRNA: CDs composites (Fig. S1C). The hydrodynamic diameter of the dsRNA: PMSNs (1:10 ratio) composites showed an increase with respect to pristine PMSNs (Table 1), while keeping a closer value of the PdI. In addition, dsRNA loading capacity and efficiency were assessed by incubating at room temperature different ratios of PMSNs with a constant dsRNA amount (2.5 µg). The nanocomposites were run on agarose gels to observe the dsRNA loading behavior on PMSNs (Fig. 5A). While at dsRNA: PMSNs ratios 1:1 and 1:5 the dsRNA was not fully loaded, from ratios 1:10 and above the dsRNA was not visible in the gels, indicating the absence of free dsRNA molecules and that the dsRNA was stuck in the wells in the form of complex with the PMSNs. These assays allowed the calculation of the loading capacity (LC) and loading efficiency percent (LE%) of dsRNAs on PMSNs after the quantification of remnant RNAs in supernatants after precipitation of dsRNA: PMSNs nanocomposites (Fig. 5B). Given that the amounts of PMSNs were 30 and 40 times the amount of dsRNA, the LE% reached 65.7% and 68.9%, respectively, representing more electrostatically absorbed dsRNA due to the presence of the more positive-charged PMSNs. On the other hand, at the lower loading ratios (i.e. 1:1 and 1:5), the surface of the PMSNs was saturated with the dsRNA molecules, resulting in higher LC and lower LE%. The loading capacity difference between the 1:20 and 1:10 loading ratios was significant and it did not change dramatically at loading ratios of 1:1 and 1:5. Thus, the 1:10 ratio was selected as suitable ratio for loading dsRNA molecules on PMSNs. The isotherm derived from the binding assays was modeled using the Langmuir, Freundlich, and Langmuir-Freundlich models, with the latter model providing the best fit (R2 = 0.95) for the interaction between dsRNA and PMSNs (Fig. 5C) (Table S4).

Regarding the FT-IR analysis of dsRNA: PMSNs, the C = O stretching vibrations from nucleobases in dsRNA appear in the 1600–1750 cm−1 region. The data revealed higher transmittance values for the PMSNs than for the dsRNA: PMSNs composite at around 1650 cm−1. This decrease in transmittance (increase in absorption) suggests the loading of dsRNA into the nanoparticles. Base stacking vibrations characteristic of nucleic acid structures appeared in the 1500–1600 cm−1 region. Transmittance values here were approximately higher for the PMSNs than for the nanocomposite at around 1550 cm−1. Again, the lower transmittance in the composite might indicate the presence of dsRNA. A possible indicator of dsRNA incorporation was the phosphate group vibrations in the 1000–1200 cm−1 region. Although this overlaps with Si-O-Si stretching bands, examination revealed differences between the two spectra. Transmittance values at approximately 1150 cm[- [1 resulted slightly higher for the dsRNA: PMSN. The enhanced absorption in dsRNA: PMSNs could be attributed to the phosphodiester backbone of dsRNA (Fig. S2).

The hydrodynamic diameter as measured with DLS of the dsRNA: PMSNs was slightly higher than pristine PMSNs suggesting the presence of dsRNA molecules on the surface of PMSNs. Besides, we observed a remarkable decrease of surface charge according to ζ potential measurements, from + 22.55 (pristine PMSNs) to -9.91 mV (dsRNA: PMSNs) for 1:10 ratio confirming that the positive surface of PMSNs were loaded with the negative RNA molecules via electrostatic interactions (Table 1). The dsRNA: CDs composite (1:0.5 ratio) had a surface charge of -7.7 mV, whereas naked dsRNA resulted in -13.5 mV. SDS and EDTA treatments were used to study the dsRNA release from the PMSNs. SDS (0.1%), but not EDTA, successfully interfered with the electrostatic interactions among negative phosphate groups of dsRNA molecules and positive polyamine groups of PMSNs due to their amphipathic nature (Fig. 5D).

Fig. 5
Fig. 5
Full size image

Loading of dsRNA in PMSNs. (A) Retardation gel of dsRNA with increasing amounts of PMSNs on a 2% agarose gel. Lane 1: dsRNA: PMSNs pool: 1:1; lane 2: 1:5; lane 3: 1:10; lane 4: 1:20; lane 5: 1:30; lane 6: 1:40; lane 7: only dsRNA; lane 8: 2.5 µg PMSNs; lane 9: 25 µg PMSNs; and M: molecular weight marker (Ladder V NZY Tech). dsRNA is indicated by a red arrow. (B) Loading capacity and loading efficiency of dsRNA on PMSNs. (C) Isotherm plot of dsRNA adsorption on PMSNs and fitting it to the Langmuir, Freundlich and Langmuir-Freundlich models for the experimental data of dsRNA adsorption on PMSNs, being qe the quantity of adsorbed material per adsorbent mass. (D) Electrophoresis of dsRNA release test from PMSNs on a 2% agarose gel. Lane 1: dsRNA control (1.5 µg); lane 2: dsRNA: PMSNs; lane 3: dsRNA: PMSNs + 0.1% SDS; lane 4: dsRNA: PMSNs + 0.2% SDS; lane 5: dsRNA: PMSNs + 0.3% SDS; lane 6: dsRNA: PMSNs + 5 mM EDTA; lane 7: dsRNA: PMSNs + 10 mM EDTA; and M: molecular weight marker (Ladder V). The dsRNA band is indicated by a red arrow. All dsRNA loading ratios on PMSNs in this test were 1:10. The lumpy bands seen under the dsRNA bands in lanes 3, 4, and 5 are the SDS molecules that traveled on the gel. Images A and D have been cropped for clarity and conciseness and their original gels are presented in Supplementary Figs. S6 and S7, respectively.

Nanoparticle carriers improve dsRNA delivery to the plant and enhance its movement and stability

Delivery of dsRNA either naked or as nanocomposites was assessed by RT-qPCR when a steady amount of 60 µg (ds)RNA bacterial extract was sprayed on leaves at the loading ratio of 1:0.5 for dsRNA: CDs and varying dsRNA: NPs loading ratios to optimize the best ratio for subsequent sprays of dsRNA. In cucumber plants, sprayed with dsNbCHE, the dsRNA: NPs composites showed significant increases in dsRNA delivery compared to naked dsRNA (Fig. 6A). Average Cq values were 17.8, 20.2, 18.9, 20.24 and 22.5 for the proximal areas of the leaves sprayed with dsNbCHE: CDs, dsNbCHE: PMSNs (1:1), dsNbCHE: PMSNs (1:10), dsNbCHE: PMSNs (1:20) and naked dsNbCHE, respectively. These values resulted in 5.1- and 4.3-fold delivery increases for the dsNbCHE: CDs and dsNbCHE: PMSNs (1:10) with respect to naked dsNbCHE, respectively. Remarkably, dsRNA could be detected in the distal half of the leaf (unsprayed and covered) only when the formulation dsNbCHE: CDs was used (Fig. 6B). In the second experiment, a similar trend was observed when N. benthamiana plants were sprayed with naked dsBCTV and the respective nanocomposites. In this case, the average Cq values were 18.3, 18.9, 21.2 and 21.3 for dsBCTV: CDs, dsBCTV: PMSNs (1:10), dsBCTV: PMSNs (1:20) and naked dsBCTV, respectively. The respective fold increases showed significant differences (P < 0.05) with respect to the naked dsBCTV as 2.43- and 3.52-fold for the dsBCTV: CDs and dsBCTV: PMSNs (1:10), while the dsRNA delivery decreased 0.45-fold compared to naked dsRNA in dsBCTV: PMSNs (1:30) treatments (Fig. 6C). From these experiments, it was shown that increases in dsRNA: PMSNs loading ratios over 1:10 did not result in improved dsRNA uptake, but rather decreased it, indicating that this ratio was optimal and chosen for the subsequent experiments. Next, we further investigated dsRNA uptake after 7 dpt in cucumber. When compared with naked dsRNAs, CDs-assisted dsRNA delivery was 2.02-fold and the dsNbCHE: PMSNs (1:10) 1.96-fold, respectively (Fig. 6D). In all experiments, no dsRNA was detected in plants when sprayed with either DEPC-treated water or non-specific dsRNAs.

Fig. 6
Fig. 6
Full size image

Comparisons of delivery of dsRNA to cucumber and N. benthamiana plants by RT-qPCR. (A) Delivery of dsRNA (NbCHE-derived) naked or as nanocomposites at different loading ratios in cucumber leaves (3 dpt). (B) Detection of dsRNAs in cucumber in a distal location from the application site. (C) Delivery of dsRNA (BCTV-derived) in N. benthamiana plants after spraying (3 dpt) with naked dsRNA or as dsRNA: NPs composites. (D) Delivery in cucumber leaves of dsRNA (NbCHE-derived) as naked dsRNA or dsRNA: NPs composites at different loading ratios at 7 dpt. Different letters on the bars refer to significant differences (P < 0.05). *P < 0.05, **P < 0.01 and ***P < 0.005 represent significance after one-way ANOVA test.

Evaluation of SIGS treatment against BCTV in N. benthamiana

In a preliminary experiment, the effect of naked dsRNA targeting BCTV was assessed in plants that were agroinoculated with the virus. Three days after application of naked dsRNA on all the leaves, two lower leaves from each plant were agroinoculated with BCTV and symptom development was monitored. In the control, inoculated plants that did not receive dsRNA treatment, yellowing of the leaves began 7 dpi. However, in the plants treated with naked dsRNA, symptom onset was delayed until 10 dpi (Fig. 7A). Despite this delay, by 14 dpi, all plants exhibited symptoms with comparable severity, indicating that while naked dsRNA treatment postponed disease onset, it did not ultimately prevent disease progression.

After the preliminary results, a more comprehensive experiment was conducted, including various treatments: naked dsBCTV, dsRNA-loaded nanoparticles (dsBCTV: CDs and dsBCTV: PMSNs), as well as negative (uninoculated) and positive (mock-treated) controls. By 6 dpi, plants in the positive control group began to show initial symptoms of BCTV infection. The area under the disease progress curve (AUDPC) revealed a statistically significant reduction in disease severity for plants treated with dsBCTV: PMSNs compared to untreated inoculated plants (P = 0.019) (Table S5, Fig. 7B). Regarding virus quantifications, no significant differences were observed between the mock and the naked dsRNA treatments at 9 dpi. However, differences in virus titer resulted when the dsRNAs were delivered with the NPs (Fig. 7C). By 16 dpi, the differences compared to the mock were highly significant in all the treatments that showed reduced BCTV titers ranging from 8- to 28-fold (Fig. 7C, Fig. S4).

Fig. 7
Fig. 7
Full size image

Effects of the dsRNA treatments on BCTV-infected plants. (A) Disease symptoms in BCTV-inoculated N. benthamiana plants treated with naked dsRNA and the nanocomposites at 14 dpi (days post virus inoculation) and 17 dpt (days post dsRNA treatment). (B) Disease progress curves for the BCTV-inoculated plants in the different conditions studied. Mock (treated with dsMP, as unspecific dsRNA for BCTV); dsRNA (treated with naked dsBCTV); dsRNA: CDs (treated with dsBCTV-loaded CDs); dsRNA: PMSNs (treated with the dsBCTV-loaded PMSNs). (C) BCTV accumulation among the different treatments determined at 9 and 16 dpi. Relative fold differences in BCTV quantitation with respect to the mock control are shown. *P < 0.05, **P < 0.01 and ***P < 0.005 represent significance after one-way ANOVA test. Bars represent mean ± SE.

Evaluation of SIGS treatment against TuMV in N. benthamiana

The two TuMV isolates initially considered in this study belong to different clades and pathotypes. Phylogenetic analyses of the nucleotide sequences of the P1 and CP genes allocated the TuMV-UK1 and TuMV-Svr to the world-B and basal-BR groups, respectively (Fig. S3). The nucleotide identities of the corresponding HC-Pro and CP regions between the two TuMV isolates resulted 81.71% and 90.13%, respectively. In preliminary assays, a mix of dsCP and dsHCPro was sprayed on N. benthamiana plants followed one hour later by inoculation with the TuMV UK1 isolate. After two weeks, plants that were only inoculated with the virus began to show chlorosis and mosaic symptoms; meanwhile, the dsRNA-sprayed plants did not show clear symptoms of mosaic disease until the end of the assay at 46 dpi (data not shown). In parallel, we observed that symptoms after the inoculation with the TuMV-Svr isolate were more severe, and for disease severity comparisons among the treatments, we chose the TuMV-Svr isolate for further experiments. TuMV-infected plants displayed strong chlorosis, internode elongation and leaf surface reduction, which was less evident in the dsCP-HCPro- or dsCP-HCPro: NPs-treated plants (Fig. 8A, Fig. S5). The results indicated that plants treated with either dsCP-HCPro: CDs or dsCP-HCPro: PMSNs maintained chlorophyll levels comparable to uninoculated controls at both 46 and 66 dpi, indicating effective mitigation of TuMV infection (Fig. 8B). In contrast, plants treated with naked dsRNA displayed significantly lower chlorophyll content at 66 dpi. Analysis of viral load across treatments at 7 and 37 dpi showed significant differences between the treated and untreated plants (Fig. 8C). At 37 dpi, both dsCP-HCPro: CDs and dsCP-HCPro: PMSNs treatments significantly reduced viral load compared to the untreated controls, supporting the sustained antiviral effect of nanoparticle-mediated dsRNA delivery over an extended period. At 7 dpi, TuMV-inoculated untreated plants showed an average Cq of 18.7, compared with 20.9, 24.81 and 24.79 of naked dsRNA, dsCP-HCPro: PMSNs and dsCP-HCPro: CDs, respectively. Accordingly, the virus load in the plants treated with dsRNA was reduced by 11-fold when compared to no treatments, and the virus loads in plants treated with dsCP-HCPro: PMSNs and dsCP-HCPro: CDs were reduced by 57.8- and 25-fold, respectively. When virus titers were quantified at 37 dpi, the average Cq was 17.7 for the virus-inoculated untreated plants, and 20.1, 21.34 and 22.4 for the naked dsRNA, dsCP-HCPro: PMSNs and dsCP-HCPro: CDs, respectively. As a result, the viral titer was reduced by 2.4-times in dsRNA-treated plants and 13.5- and 17.3-times, in the dsCP-HCPro: PMSNs- and dsCP-HCPro: CDs-treated plants, respectively.

Fig. 8
Fig. 8
Full size image

Effects of the dsRNA treatments on TuMV-Svr-infected plants. (A) Disease symptoms in TuMV-Svr-inoculated N. benthamiana plants treated with naked dsRNA and the nanocomposites at 25 dpi. The treatments were as follows: Mock (treated with DEPC-treated water); dsRNA (treated with naked dsCP-HCPro); dsRNA: CDs (treated with dsCP-HCPro-loaded CDs); dsRNA: PMSNs (treated with the dsCP-HCPro-loaded PMSNs). (B) Chlorophyll content in TuMV-inoculated plants in different treatments compared with non-inoculated plants. Determinations were made at 46 and 66 dpi. (C) TuMV accumulation among the treatments with naked and dsCP-HCPro-loaded NPs determined at 7 and 37 dpi. *P < 0.05, **P < 0.01 and ***P < 0.005 represent significance after one-way ANOVA test. The effect of the treatment with naked dsCP/HC-Pro at 37 dpi resulted non-significant (P = 0.062). Bars represent mean ± SE.

Discussion

In this study, we have investigated SIGS-derived resistance to the DNA virus BCTV and the RNA virus TuMV in N. benthamiana, using carbon dots (CDs) and polyethylenimine-functionalized mesoporous silica nanoparticles (PMSNs) to overcome the key delivery barriers of dsRNA.

To improve dsRNA delivery to plants, two nanoparticle formulations, CDs and PMSNs, were used. The CDs used in this study were synthesized following Delgado-Martín et al.28 and displayed properties such as small size (< 5 nm), high photoluminescence, and capacity to bind to dsRNA. The successful synthesis and functionalization of PMSNs were confirmed through a combination of TEM, DLS, FT-IR, nitrogen adsorption-desorption, ensuring that the nanoparticles were suitable for dsRNA binding and delivery. The hydrodynamic size of the MSNs and their complexes (RNA-loaded PMSNs) analyzed using the DLS technique showed significantly higher average size values compared to those obtained by TEM. this discrepancy can be attributed to several factors. First, the DLS technique measures the hydrodynamic diameter, which represents the size of a hypothetical hard sphere that diffuses at the same rate as the particle being measured. This includes not only the particle core but also the surrounding hydration layers and potential aggregation in solution. Functionalized particles and RNA-loaded nanoparticles tend to exhibit larger hydrodynamic sizes because the functional groups and RNA molecules are hydrophilic and capable of absorbing water, thereby increasing their apparent size. Moreover, functionalization can enhance the tendency of nanoparticles to aggregate in aqueous media, as also indicated by our DLS results31. The positive surface charge of PMSNs, crucial for stable electrostatic binding of dsRNA, was verified by ζ potential (surface charge) measurements. Also, the negative value of the ζ potential of the dsRNA: PMSNs complex confirmed successful dsRNA molecules’ coverage on the positive PMSNs, which is in line with previously reported findings32. While FT-IR confirmed functional group addition at each step, the characteristic nucleic acid signatures in the dsRNA: PMSNs nanocomposites were masked, probably because of the high nanoparticle to dsRNA ratio. The XPS results of the PMSNs aligned with previous studies on PEI-coated nanoparticles, which have demonstrated similar spectral features confirming successful PEI functionalization. For example, Choi et al.33 reported a comparable XPS spectrum for PEI-modified silica nanoparticles, where nitrogen peaks were similarly observed due to the amine-rich polymer coating. Porosity analysis revealed that the decrease in surface area of PMSNs as compared to BMSNs indicates that functional groups blocked the pores on BMSNs. The fact that BMSNs had higher micropore volume than PMSNs suggests that functionalization preferentially fills micropores (with sizes of < 2 nm), thus decreasing their contribution to total surface area. By entering the pore structures of MSNs and accumulating there, PEI groups conferred a more positive surface charge to nanoparticles34. Herein, instead of loading dsRNA into the pores, we exploited the porosity capacity of PMSNs to enhance their positive surface charge, subsequently leading to a stronger electrostatic interaction between dsRNA and PMSNs. Also, increased mesopore accessibility along with the increase in external surface area should allow PMSNs to be beneficial for carrying larger molecules like dsRNAs. In such cases, surface chemistry plays a more important role than porosity. The adsorption behavior best fit a Langmuir-Freundlich model, indicating a complex interaction conducive to multilayer adsorption35.

We observed that increasing the relative amount of PMSNs compared to dsRNA increased the loading efficiency up to 69% in a ratio of 1:40. When decreasing the PMSNs ratios from 1:10 to 1:5 and 1:1, a noticeable decrease in the upward trend (the slope of the graph) of the loading capacity was observed; and inversely, the loading efficiency decreased from 42% to 22.8% and 4.5%, respectively, as also evidenced in retardation agarose gels. A similar pattern was reported when different PMSNs were loaded with dsRNA13 or pDNA36. Recently, it was shown that the capacity of loading siRNAs was conversely proportional to MSN diameter15.

The NP-assisted delivery on plants aims to enhance the delivery and stability of dsRNAs in plant tissues. RT-qPCR analysis at three days post-application revealed that there was a significant increase in dsRNA accumulation compared to naked treatments in cucumber plants treated with a 1:10 ratio of dsRNA: PMSNs composites. Similarly, in N. benthamiana, a 3.5-fold increase in dsRNA accumulation was observed with a 1:10 ratio compared to naked treatments. These results are consistent with previous studies showing that nanoparticle-mediated delivery systems can enhance the uptake and stability of exogenous nucleic acids in plants28,37. The stability of delivered dsRNAs in the CD- and PMSNs-nanocomposites forms over time was also assessed by RT-qPCR at seven days post-application. We observed that dsRNA movement within the plant is restricted except when the dsRNA is delivered in the form of dsRNA: CDs nanocomposites, in line with a prior report showing the systemic movement of dsRNA: CDs26.

Since viral replication cycles often extend beyond initial treatment periods, the increase in delivery and stability mediated by NPs is desirable7,13,38. Most of the studies reported to date on the use of exogenous dsRNA against plant viruses have been performed on RNA viruses, given that, up to our knowledge, effective control for DNA viruses with this method is scarce, as has been reported for the begomoviruses tomato leaf curl virus (ToLCV), tomato yellow leaf curl virus (TYLCV) and ToLCNDV13,27,39,40. In general, the infection method for the virus was agroinoculation into the abaxial side of the leaf, followed by the simultaneous application of dsRNA to the adaxial surface of the same leaf. However, it has been recently shown that dsRNA delivered by amine-functionalized MSNPs could decrease viral titers in ToLCNDV-infected N. benthamiana plants when supplied to roots or leaves13. In contrast, when another DNA virus, as is the case for the begomovirus tomato severe rugose virus (ToSRV), was inoculated by its insect vector, the whitefly, that directly transfers the virus to the vascular tissue of the plant, the dsRNA treatment was unsuccessful41. In this case, the failure of the SIGS method in generating resistance may be due to the possibility that the virus enters the vascular tissue before being inhibited by the interfering RNA induced by dsRNAs in plant leaves41. In a recent work, it has been shown that dsRNA derived from the AC2 gene, when loaded onto MSNPs, produced amelioration of symptoms and decreased virus titers after infection with ToLCNDV13. A BCTV gene segment chosen in our study for SIGS was previously used by Montazeri et al. 43 to transiently express the viral sense, anti-sense, and hairpin in N. benthamiana and sugar beet to confer the resistance against BCTV and BCTIV. They observed that the hairpin form of RNA could significantly delay the disease symptom appearance compared to non-transgenic N. benthamiana, as well as the transgenic N. benthamiana transformed with sense and anti-sense constructs. In our preliminary assays, when using naked dsRNAs targeting agroinoculated BCTV, we observed that symptom onset was delayed by three days compared to untreated controls. However, by 14-dpi, all plants exhibited symptoms with comparable severity, indicating that while naked dsRNA treatment could delay disease progression slightly through SIGS, it was insufficient for long-term protection against BCTV infection. Noteworthy, the use of nanoparticle loaded with dsRNAs showed more interesting results. In particular, plants treated with dsRNA: PMSNs formulations exhibited a statistically significant reduction in disease severity compared to untreated controls and those treated with naked dsRNA. Although at 7 dpi, no significant differences in viral load were observed between treatments, by 14-dpi, plants treated with dsRNA: PMSNs exhibited a two-fold reduction in viral load compared to untreated controls. Even though the SIGS method did not provide the plants with a desired long resistance against this DNA virus, it is conceivable that multiple applications of the dsRNA: NPs could boost their effectiveness15.

According to the number of successful reports of induced resistance to RNA viruses by exogenous application of dsRNA in the literature, this approach seems more efficient for RNA viruses than DNA viruses. The 3′ end of the genome of potyviruses such as TuMV is a highly conserved region that includes the NIa, NIb and CP genes43. Several reports have shown that with external application of exogenous dsRNA from CP and NIb genes, dsRNACP had better efficiency in reducing potyvirus infection44,45. The HC-Pro protein encoded in the TuMV genome is an example of a multifunctional protein that carries at least three different functions: plant-to-plant transmission of the virus, polyprotein maturation, and suppression of the RNA silencing process46. There are several reports of successful application of dsRNAHC-Pro with different lengths from 500 to 1100 bp to reduce the infection caused by the potyviruses44,47,48. Consequently, we used combined CP + HC-Pro TuMV-derived dsRNAs in our study. Recently, a reduction of symptoms in tomato, pepper and tobacco by 60% and a decrease of virus titers mediated by nanoparticle (including CDs) loaded with dsRNAs have been reported in potato virus Y-infected plants38. Plants treated with naked dsRNA displayed significantly lower chlorophyll content compared to those receiving dsRNA-loaded NPs at 66 dpi. Accordingly, analysis of viral load across treatments at 7 and 37 dpi showed significant differences among dsRNA-loaded NPs, naked dsRNA, as well as untreated plants. This corresponds to an approximate reduction in viral load by 57.8- and 25-fold for dsRNA: PMSNs and dsRNA: CDs treatments compared to untreated controls, respectively. At 37 dpi, both nanoparticle-based treatments continued to show significant reductions in viral load compared to untreated controls. However, the difference between the naked dsRNA treatment and the untreated controls was negligible. It can be concluded that in the long term, the efficient dsRNA delivery mediated by nanoparticles plays a key role in resistance to this potyvirus.

In this work we have not investigated specific effects of the nanoparticles alone. However, we have never seen any toxicity or phenotypic effects on plants of any of the species so far studied, including N. benthamiana, N. tabacum, squash, cucumber, or turnip following NP application at the same dosages as those employed in this investigation. Workers have shown that biomass-derived carbon dots enhance photosynthetic efficiency, nutrient uptake and biomass accumulation in several plant species without growth inhibition at effective doses, confirming their high biocompatibility49,50. It has also been shown that citric acid/β-alanine carbon dots used as DNA-delivery carriers into plant tissues do not reduce regeneration efficiency relative to conventional spermidine-based protocols, supporting their safe use in plant tissue culture systems51. Thus, we acknowledge that CDs and PMSNs could, in principle, influence cuticle properties or plant stress signaling independently of dsRNA, and future work including nanoparticle-only treatments and detailed physiological analyses will be important to fully disentangle nanoparticle-derived effects from dsRNA-mediated RNAi responses.

The results confirm that nanocarriers like CDs and PMSNs represent a plausible strategy for enhancing SIGS effects against both DNA and RNA viruses. However, several challenges need to be solved before this technology can be widely adopted in greenhouse and field scales. The environmental fate and degradation kinetics of dsRNA-loaded NPs need further investigation to ensure their ecological safety52. Microbial-based production systems for generating cost-effective sources of dsRNAs may help produce large amounts of dsRNAs for use in commercial settings, in particular for agricultural applications53. Finally, comprehensive regulatory frameworks for assessing the environmental risks associated with nanoparticle-mediated RNAi products are necessary to ensure the safe use of dsRNA and nanoparticles in agriculture.

Methods

Plants and virus sources

Seeds of cucumber (Cucumis sativus cv. Bellpuig) were purchased from Semillas Fitó (Barcelona, Spain) and N. benthamiana seeds were already available in the lab. Two plant viruses, TuMV and BCTV, were evaluated in this study. Regarding TuMV, two strains UK1 and DSMZ were used as mild and severe strains, respectively. The isolate UK1 was used in preliminary tests; and the DSMZ isolate, because of its higher severity, was used in subsequent experiments. For clarity and brevity, the DSMZ isolate will be referred to as “TuMV-Svr” throughout the text. TuMV inocula were obtained from different sources: the infectious clone p35tunos54 was used as the source of the TuMV-UK155 (Acc. No. AB194802) to inoculate N. benthamiana plants that were used as source of inoculum in the experiments. The TuMV-Svr was originally a field isolate from lettuce kindly provided by S. Winter and W. Menzel (DSMZ, Germany; Acc. No. MZ405633). This isolate was provided freeze-dried and was mechanically inoculated in N. benthamiana using the DSMZ buffer (0.05 M sodium/potassium phosphate pH 7.0, 1 mM EDTA, 5 mM DIECA, 5 mM thioglycolic acid) to obtain the inoculum source. For the SIGS-TuMV experiment described in this work, 5 g of infected and symptomatic N. benthamiana leaves were smashed in 5 ml of DSMZ buffer using a cold pestle and mortar. Then, 5 ml buffer was added and mixed well. Two expanded healthy N. benthamiana leaves were first dusted with carborundum and then inoculated uniformly by rubbing the pestle on each leaf. Plants were kept in the growth room at 22 °C and 16 h/8 h light/dark cycles as preliminary observations showed us that TuMV symptoms were better observed at that temperature rather than 25 °C.

The infectious clone BCTV-Svr (including the full BCTV-Svr genome, Acc. No. X97203) originally collected from sugar beet plants in Iran was kindly provided by Dr. Behjatnia, University of Shiraz, Shiraz, Iran56. For the agroinoculations, Agrobacterium tumefaciens LBA4404 harboring the binary plasmids was cultured overnight in 25 ml of LB medium containing kanamycin (50 µg/ml) and rifampicin (10 µg/ml). When the OD of the culture reached 1, the bacteria were precipitated and the bacterial sediment was resuspended in 25 ml of sterile distilled water and Acetosyringone (Sigma) was added to it at a final concentration of 200 µM. The bacteria were stirred for one hour at room temperature and 180 rpm. Two expanded leaves from each plant were agroinoculated with 1 ml of the liquid in the abaxial side using an insulin syringe until the liquid entered the entire leaf blade. Plants were kept in the growth room at 25 °C and 16 h/8 h light/dark cycles.

Design of constructs and dsRNA production

Constructs for the synthesis of dsRNAs were derived from the BCTV-Svr and the conserved sequences of TuMV-UK1 and TuMV-Svr. The dsRNAs were produced in Escherichia coli HT115(DE3) cells that carried L4440-derived plasmids harboring T7 promoters at both sides of the virus gene segments. For obtaining the constructs, RNA extractions of plants infected either with the TuMV-UK1 and TuMV-Svr using Trizol (Invitrogen) following the manufacturer’s recommendations. The quality of the RNAs was estimated by gel electrophoresis and the quantity with the Nanodrop-2000 spectrophotometer (Thermo Fisher). Next, the cDNA synthesis was carried out using the MultiScribe Reverse Transcriptase (Applied Biosystems) in the presence of 1 mM dNTPs and RNase inhibitor for 1 h at 42 °C. The cDNAs were used as templates for amplification by PCR of the viral gene segments to be cloned in vector L4440gtwy (G. Caldwell: Addgene plasmid # 11,344). For both TuMV-UK1 and TuMV-Svr, two constructs each were designed for targeting the respective viruses isolates. For the TuMV-UK1 one was derived from a segment of 510 bp of the HC-Pro gene within positions 2,066 − 2,575, rendering the dsHCPro. The other construct was derived from a 612 bp section of the CP gene in positions 8,854-9,465, for generating the dsCP. The amplicons were obtained from the cDNA of an infected plant using the sets of primers disclosed in Table S2. These primers included a region overlapping vector L4440gtwy for InFusion cloning (Takara). For that, plasmid L4440gtwy was digested with restriction enzymes HindIII and BglII and purified. Next, the assembly reaction was carried out using the digested vector and the respective CP and HCPro amplicons. The products of the reactions were used to transform E. coli Top10 competent cells. Plasmids were extracted and checked by PCR, restriction analysis and sequencing and used to transform E. coli HT115(DE3). Similar plasmids in equivalent genomic regions were obtained for the TuMV-Svr. For the targeting of BCTV-Svr, a chimeric construct of 342 bp was obtained that included a 120 bp fragment covering parts of Rep (C1) and TrAP (C2) genes from BCTV-Svr genome fused to a 222 bp fragment covering parts of CP (V1) and MP (V2) genes from BCTIV genome42. The construct was cloned into L4440 after RE digestion and the resulting plasmid was used to transform E. coli HT115(DE3). For the dsRNA delivery experiments in cucumber, an L4440gtwy-derived plasmid that included a 550-bp segment of the N. benthamiana magnesium chelatase subunit h gene (Acc. No. XM_004149349) was obtained by Gateway (Thermo Fisher) cloning, following manufacturer’s protocols and was used to produce the dsRNA(NbCHE) named as dsNbCHE. Since this selected segment was not present in cucumber genome, the RT-qPCR-amplified amplicons in dsNbCHE-sprayed cucumbers belonged to the sprayed dsNbCHE, and no other off-targets in the cucumber genome. All bacterially produced dsRNAs in this study were obtained from the IPTG-induced E. coli HT115(DE3) cells transformed with the corresponding plasmids. The cells were grown in LB media supplemented with carbenicillin (100 µg/mL) and IPTG (1 mM final concentration) for 7 h at 37 °C. The dsRNAs were extracted using Trizol, as previously described, and quantified by densitometry analysis following agarose gel electrophoresis.

Nanoparticles’ synthesis and characterization

Synthesis of CDs was done using a hydrothermal method, according to Delgado-Martín et al.28. Briefly, 2 g of branched polyethyleneimines (bPEI, 10000 Da, Alpha Aesar, Germany) dissolved thoroughly in a solution of 2 g glucose in 10 ml Milli-Q water. Then, the solution was sealed in a stainless-steel autoclave lined with Teflon (100 ml capacity) and heated at 210 °C for 8 h. When the autoclave was cooled down by remaining in room temperature, the solution was taken out and purified using a 0.22 μm filter (Millipore, Merck, Darmstadt, Germany). The obtained filtrates were dialyzed using 1000 Da molecular weight cut-off dialysis bags (Spectra/Por, Fisher, MA, USA) against 40 ml of Milli-Q water by shaking for 24 h at 85 rpm and room temperature. Next, the dialysis bag was moved to a glass beaker containing 2000 ml double-deionized water and stirred overnight. Then, the water in the beaker was renewed and the previous step was repeated for 6 h. Finally, the nanoparticles inside the dialysis bag were taken, lyophilized, weighted, and used for subsequent characterizations and applications. The CDs were resuspended in Milli-Q water for the obtention of the dsRNA composites.

MSNs were synthesized using our previously reported protocol based on the template removing approach with some modifications57. For that, 100 mg CTAB was dissolved in 48.65 ml deionized water and then stirred at 14,000 rpm. As soon as the solution was clear, 350 µl NaOH (2 M) was added to the solution in a dropwise manner and then stirred at temperature set at 138 °C until the solution temperature reached 80 °C. Then, 1 ml of TEOS (98%) was added dropwise, the lid was closed thoroughly and constant stirring was maintained for 2 h. To remove the CTAB, the NPs were dispersed in an acidic alcoholic solution of HCl (37%) : ethanol (99%) (1:10 ratio) and incubated at 65 °C with constant stirring at 1000 rpm for 16 h. After 16 h, the NPs were collected by centrifugation and the previous step was repeated for 3 h. The synthesized Bare-MSNs (BMSNs) were washed three times with absolute ethanol and subsequently three times with deionized water. In the next step, the amine functionalization was carried out by adding 200 µl deionized water and 100 µl glacial acetic acid to 50 mg BMSNs dispersed in 4.6 ml absolute ethanol and stirred at 1000 rpm. Afterwards, 100 µl ethylenediamine N-(3-(trimethoxysilyl)propyl) (EDS) was added to the stirring solution and the condition was kept for at least one hour. The resulting Amine-functionalized MSNs (AMSNs) were washed three times with absolute ethanol and deionized water. To add carboxyl groups to the AMSNs, 200 mg of succinic anhydride (SA) dissolved in 20 ml of dimethylformamide (DMF) and solution was stirred for 20 min under a stream of nitrogen gas to evacuate the oxygen. Then, 100 mg of AMSNs that were previously well-dispersed in DMF, were added to the SA solution in a dropwise manner. The lid was sealed completely and the mixture was stirred overnight. The carboxyl-functionalized MSNs (CMSNs) were washed three times with absolute ethanol and deionized water. In the last step, positively charged MSNs prepared by coating CMSNs with polyethyleneimine (PEI) (PMSNs). First, CMSNs were precipitated in microtubes and redispersed in 1.5 ml of phosphate-buffer saline (PBS, 10 mM, pH 7.4) containing N-Hydroxysuccinimide (NHS) (0.2 mg/ml) and EDC (1 mg/ml) and were manually shaken for about 5 min. The nanoparticles were immediately centrifuged for 3 min at 13,000 rpm. Then, each pellet was dispersed in 1.5 ml of PBS containing PEI (2 mg/ml) and the microtubes were inverted at 15 rounds per minute for 1 h at room temperature. Finally, the obtained PEI-functionalized MSNs (PMSNs) were washed three times with PBS (10 mM) and deionized water.

Transmission electron microscopy (TEM) (Thermofisher, FEI Talos F200X) was used for studying the size and morphology of the MSNs dispersed in deionized water (40 µg/ml) and a population was used to obtain the average size with the ImageJ software58. The hydrodynamic diameter and Zeta (ζ) potential of MSNs (40 µg/ml) were measured using the Zetasizer (Zetasizer Nano ZS, Malvern, UK). Surface functionalization of BMSNs, AMSNs, CMSNs, and PMSNs was evaluated on 5 mg of lyophilized MSNs using Fourier-transform infrared (FT-IR) analysis of the spectra collected with the Tensor 27 spectrophotometer (Bruker, Bremen, Germany) using a Gate Single Reflection Diamond ATR System accessory. A standard spectral resolution of 4 cm−1 in the spectral range 4000–400 cm− 1 and 64 accumulations were used.

The surface chemical composition of the CMSNs and PMSNs were analyzed using X-ray photoelectron spectroscopy (XPS). Measurements were performed on a PHI VersaProbe II X-ray photoelectron spectrometer (Physical Electronics, Chanhassen, MN, USA) equipped with a monochromatic Al Kα X-ray source (photon energy of 1486.6 eV) at the SCAI (Universidad de Málaga, Spain). The spectra were recorded in the binding energy range of 0–1400 eV under ultra-high vacuum conditions. Survey scans were acquired to identify the elemental composition, while high-resolution scans were performed for detailed analysis of specific elements. Data processing and peak fitting were conducted using standard software provided with the spectrometer.

The surface area and porosity of the BMSNs and PMSNs were determined using nitrogen adsorption/desorption isotherms measured at liquid nitrogen temperature (77 K). These experiments were conducted with an ASAP 2240 system (Micromeritics, Norcross, GA, USA) available at the SCAI. Prior to the measurements, the samples were degassed under vacuum at 60 °C for 12 h to eliminate adsorbed moisture and contaminants, ensuring that only the intrinsic surface properties of the material were analyzed. The adsorption/desorption isotherms were recorded over a relative pressure (p/p0) range from 0 to 1. The specific surface areas were calculated using the Brunauer-Emmett-Teller (BET) method, which involves applying the BET equation to the linear region of the adsorption isotherm at low relative pressures (p/p0 < 0.3). The slope and intercept of the BET plot were used to determine the monolayer adsorption volumes (Vm), which were subsequently converted into specific surface areas (SSA) using standard constants for nitrogen gas. The pore size distribution was analyzed using the Barrett-Joyner-Halenda (BJH) method, applied to the desorption branch of the isotherm. This method uses the Kelvin equation to relate relative pressure to pore size, assuming cylindrical pores. The BJH analysis provided detailed information on mesoporous volumes and pore size distribution in the range of 1–100 nm.

Characterization of the dsRNA: PMSNs composites

PMSNs were well dispersed using a sonicator bath with medium intensity for 3 min prior to adding the dsRNAs. To determine the optimum dsRNA: PMSNs ratio, 2.5 µg of dsRNAs derived from the N. benthamiana magnesium chelatase (dsNbCHE) were electrostatically loaded on PMSNs with different ratios (dsRNA: PMSNs: 1:1, 1:5, 1:10, 1:20, 1:30, and 1:40) in 20 µl final volume for 1 h. In addition to the loading treatments, control groups with dsRNA alone and PMSNs alone (2.5 and 25 µg) were included for comparison. The assay was repeated three times. One replicate was electrophoresed on a 2% agarose gel for visualization. The other samples were centrifuged for 10 min at 6000 rpm and the amount of RNA in supernatant was measured with the Nanodrop. Loading capacity (LC) and loading efficiency percent (LE%) of each loading ratio were calculated using Eqs. 1 and 2, respectively.

$$\:Loading\:capacity\:\left(\raisebox{1ex}{$ng$}\!\left/\:\!\raisebox{-1ex}{$\mu\:g$}\right.\right)=\frac{{C}_{i}-{C}_{f}}{m}$$
(1)

Where, Ci is the initial concentration of dsRNA (ng), Cf is the final concentration of dsRNA in supernatant (ng) and m is the PMSNs mass (µg).

$$\:Loading\:efficiency\:(\%)=\frac{{C}_{i}-{C}_{f}}{{C}_{i}}\times\:100$$
(2)

The equilibrium concentrations of dsRNA on the PMSNs (qe) were calculated using Eq. 3 to plot the dsRNA adsorption isotherm on PMSNs.

$$\:{q}_{e}\:\left(\raisebox{1ex}{$ng\:of\:dsRNA$}\!\left/\:\!\raisebox{-1ex}{$\mu\:g\:of\:PMSNs$}\right.\right)=\frac{\left({C}_{i}-{C}_{e}\right)\times\:V}{m}$$
(3)

Where V is the loading reaction volume (µl).

Correlation of the equilibrium data with Langmuir (Eq. 4), Freundlich (Eq. 5), and Langmuir-Freundlich (Eq. 6) models was studied by fitting the empirical adsorption isotherm on the mentioned models. These isotherms relate the quantity of adsorbed material per adsorbent mass (qe) with the equilibrium concentration Ce in the fluid phase.

$$\:{q}_{e}=\frac{{q}_{max}{K}_{L}{C}_{e}}{1\:+\:{K}_{L}{C}_{e}}$$
(4)

Where, qmax is the maximum dsRNA adsorption capacity of PMSNs (ng/µg), KL is the Langmuir’s constant (µl/ng) and Ce is the equilibrium concentration of dsRNA (ng).

$$\:{q}_{e}={\text{K}}_{\text{F}}{\text{C}}_{\text{e}}^{\raisebox{1ex}{$1$}\!\left/\:\!\raisebox{-1ex}{$\text{n}$}\right.}$$
(5)

KF is the Freundlich’s constant (ng/mg), Ce is the equilibrium concentration of dsRNA (ng) and n is the heterogeneity index.

$$\:{\text{q}}_{\text{e}}=\frac{{{\text{q}}_{\text{m}\text{a}\text{x}}{(\text{K}}_{\text{L}\text{F}}{\text{C}}_{\text{e}})}^{n}}{{1+{(\text{K}}_{\text{L}\text{F}}{\text{C}}_{\text{e}})}^{n}}$$
(6)

q max in the L-F isotherm model is the maximum dsRNA adsorption capacity of PMSNs (ng/µg), KLF is the affinity constant for adsorption (L/mg), Ce is the equilibrium concentration of dsRNA (ng) and n is the heterogeneity index. The values of the constants in the different models were estimated using the solver add-in in Microsoft Excel59.

Release of dsRNA from PMSNs

Sodium dodecyl sulfate, SDS (0.1%, 0.2%, and 0.3%) and ethylenediaminetetraacetic acid, EDTA (5 and 10 mM) treatments on dsRNA: PMSNs were used to study the dsNbCHE release from the PMSNs. For that, we used in these assays a 1:10 loading ratio in 20 µl final volume that were incubated for 30-min incubation at room temperature. Then, solutions of SDS or EDTA were added to the nanocomposite solutions and mixed by pipetting. The microtubes were next incubated at room temperature for another 30 min and the samples were inspected on 2% agarose gels.

DsRNA delivery to plant leaves

DsRNA delivery was studied through three experiments either on cucumber or N. benthamiana leaves at 3 and 5–6 leaf stages, respectively. In all experiments of this section, each leaf was sprayed with 60 µg of (ds)RNA extract that included 3.5 µg (~ 5%) of the corresponding specific dsRNA, according to our estimations from the densitometry analyses by an airbrush at 2.5 bar pressure. The final volume of sprayed liquid containing CDs and PMSNs nanocomposites were 400 µl and 2000 µl, respectively. Noteworthy, PMSNs are much bigger in size as compared to CDs; hence, to avoid the precipitation of dsRNA: PMSNs nanocomposites (especially in higher loading ratios) it was necessary to further dilute with Milli-Q water. There were no viral inoculations involved in these experiments. Three days post-treatment, the leaves were washed thoroughly with double-distilled water using a wash bottle and then a clean airbrush to wash away any remaining surface dsRNA. After the leaves got dried, one hollow metal cylinder was used to take a sample from each leaf area.

In the first delivery experiment, to optimize the dsRNA loading ratios on PMSNs (dsRNA: PMSNs), we tested various ratios 1:1, 1:10, 1:20 for dsNbCHE: PMSNs and 1:0.5 for dsNbCHE: CDs. The latter nanocomposite was served as a benchmark to evaluate the dsRNA delivery performance of PMSNs. Moreover, naked dsNbCHE and Milli-Q water were sprayed on plants as a negative control of NPs and a negative control of dsRNA, respectively. In this test, cucumber plant was used due to its larger leaf surface and ease of spraying. Each leaf was considered as divided into upper (near the stem) and lower (near the tip of the leaf) halves. While spraying the lower half, the upper half was covered with an aluminum foil fixed on it. Based on their distance from the sprayed point, the lower half (sprayed) and upper half (unsprayed and covered) were referred as proximal and distal areas, respectively. Three days post-treatment (dpt), samples were collected from both halves of each treated leaf. There were three replications per treatment. The content of dsRNA detected in the upper half indicated the extent to which they moved from the sprayed lower half toward the upper half. In the second delivery experiment, nanocomposites containing dsBCTV – dsRNA derived from BCTV – were sprayed on N. benthamiana leaves without virus inoculation to investigate the dsBCTV’s entrance and its relative amount in sprayed leaves. This step was a prerequisite to conducting subsequent experiments investigating the impact of topical dsRNA application on viral disease development on N. benthamiana. The treatments were: dsBCTV: CDs (1:0.5), dsBCTV: PMSNs (1:10 and 1:30), naked dsBCTV, and Milli-Q water. There were three replications per treatment. The dsRNA: PMSNs loading ratio of 1:10 was chosen from the previous experiment and the 1:30 ratio was also tested to study the presence of excessive amount of PMSNs. The third experiment was performed to study the stability of dsRNA in sprayed leaves over time. For that, one leaf of each cucumber plant was sprayed with dsNbCHE and the nanocomposites as above. Tissue samples were taken from the sprayed area of the leaves at 7 dpt. In this case, the treatments were as follows: dsNbCHE: CDs (1:0.5), dsNbCHE: PMSNs (1:10), naked dsNbCHE, and dsCP: PMSNs (1:10) as negative control of NPs and non-specific dsRNA-loaded PMSNs, respectively. There were five replications per treatment.

RNA extractions from leaf tissues (100 mg) were performed using Trizol followed by cDNA synthesis as described above. The subsequent RT-qPCRs were carried out, in two technical replicates, using diluted cDNA (1:1), Kapa SYBR Fast qPCR Master Mix (2X) Universal (Kapa Biosystems), and the corresponding forward and reverse primers for the dsNbCHE, the C. sativus RNA 18 S gene (as endogenous reference in cucumber), the dsBCTV and the N. benthamiana elongation factor 1α (NbEF-1α), each with the final concentration adjusted to 500 nM (Table S2). The qPCR program was as follows: 3 min at 95 °C, 39 cycles of 10 s at 95 °C, 10 s at 55 °C and 30 s at 72 °C, followed by a denaturing step of 10 s at 95 °C and the melting program consisting of 5-sec ramps from 65 to 95 °C seconds with 0.5 °C/sec increments. The geometric mean of their expression ratios was used as the normalization factor in all samples for measuring the quantification cycle (Cq). The relative expressions (fold) of the (ds)RNA amounts were compared based on the calculations done with the 2−∆∆Cq method60. One-way ANOVA of the relative expression differences was performed followed by mean separation using Tukey’s HSD post hoc analysis when the data were distributed normally or Kruskal-Wallis when not followed by Dwass-Steel-Critchlow-Fligner two-to-one comparisons, with the software Jamovi v. 2.3.21 (https://www.jamovi.org).

Evaluation of TuMV and BCTV-inoculated plants after the dsRNA treatments

In two separate inoculation experiments, four weeks old N. benthamiana plants were sprayed with either naked dsRNA or dsRNA-nanocomposites on all the plant’s shoot system, following BCTV-Svr or TuMV-Svr inoculation.

The BCTV-Svr experiment was as follows: Treatments were naked dsBCTV, dsBCTV: CDs (1:0.5), dsBCTV: PMSNs (1:10), naked dsMP as dsRNA negative control – referred to as Mock, and dsBCTV alone as virus negative control – referred to as Uninoculated. dsMP was used as nonspecific dsRNA derived from the MP gene of CGMMV61. All plants except the ones for virus negative control treatment were agroinoculated with BCTV-Svr on the two lower leaves at 3 dpt. There were 10 replications. BCTV-induced symptoms on N. benthamiana were scored from 0 to 4 after simplifying the rating scale introduced by Montazeri et al.43 (Table S3). Plants were monitored regularly at 9, 14, 16, 18, and 23 days post-inoculation (dpi) and the average score symptoms were obtained for each treatment which were subsequently used to calculate the area under disease progress curve (AUDPC).

Regarding TuMV-Svr experiment, treatments were naked dsCP-HCPro, dsCP-HCPro: CDs (1:0.5), dsCP-HCPro: PMSNs (1:10), DEPC-treated water as dsRNA negative control – referred to as Mock, and dsCP-HCPro alone as virus negative control – referred to as Uninoculated. All treatments except the virus negative control treatment were mechanically inoculated with TuMV-Svr at 3 dpt as described above. There were 12 replications. Symptoms of TuMV-Svr inoculated plants were scored based on the chlorophyll content. Since TuMV-caused chlorosis was a quantifiable parameter, disease severity was assessed through chlorophyll quantification using the SPAD-502Plus meter (Konica-Minolta). For that, on each plant three measures per leaf were taken at the upper, medium and lower leaves, and the average of 12 biological replicates (plants) per treatment were considered in the statistical analyses. Then, one-way ANOVA was performed followed by mean separation using Tukey’s HSD post hoc analysis to investigate the relationship between the disease progress for the respective treatments.

Virus quantifications

Total RNAs extraction and cDNA synthesis were performed using Trizol (Invitrogen) and Multiple Reverse Transcriptase (Applied biosystems), respectively, from the TuMV-inoculated plants, including the untreated and dsRNA-treated plants at 7 and 37 dpi, as described earlier. The relative RNA amounts were compared similar to the description above. Samples for BCTV quantitation were taken at 9- and 16-days dpi from the first and second leaves above the last inoculated leaf, respectively, and subsequently the total DNA was extracted62. The qPCR reactions contained DNA (200 ng), Kapa Master Mix and, in independent reactions, the BCTV-Svr or NbEF-1α (as reference gene) forward and reverse specific primers (final concentration 500 nM) and carried out in technical replicates and with a final volume of 20 µl. The qPCR was set up as follows: 3 min at 95 °C, 39 cycles of 10 s at 95 °C, 10 s at 60 °C and 30 s at 72 °C, and the melting program setting as described above. The relative expression of viral DNAs using the Cq values was estimated as above.