Introduction

Biomolecular condensates are related to many cellular processes, including the compartmentalization of biochemical reactions, the sensing and response to stress, mechanical regulation, and signal transduction1,2. Increasing evidence suggests that phase separation is a fundamental principle for the formation of biomolecular condensates3,4,5,6. The assembly components of condensates are typically complex and usually involve proteins and nucleic acids, which lead to phase separation via their multivalent interactions7,8,9,10. As obligate intracellular parasites, viruses rely on cellular functions and processes to enable efficient replication. During infection, condensate-like structures commonly referred to as replication compartments (RCs) or inclusion bodies (IBs) are formed by viral proteins. Importantly, recent studies have implicated phase separation as a driving force for the formation of these biomolecular condensates11,12. The condensates play key roles in viral life cycles, including viral entry13, genome synthesis and replication14, assembly of new particles15, and escape from antiviral defense pathways16.

Condensates formed by phase separation have attracted much attention in the pharmaceutical industry and academia as novel targets in drug discovery17,18,19. Although some small molecules have been shown to regulate the aggregation of viral biomolecular condensates in recent years, the discovery of new structures is still relatively rare20,21,22,23. Moreover, such modulators have not been reported in the agricultural field. Replication-related proteins encoded by several plant viruses, such as barley yellow striate mosaic virus (BYSMV), tomato yellow mottle-associated virus (TYMaV), tomato bushy stunt virus (TBSV), and carnation Italian ringspot virus (CIRV), have all been found to form condensates, which have been associated with phase separation in the past two years24,25,26. Recently, phytobacterial type III effectors have been shown to exhibit a propensity for phase separation27. Thus, targeting biomolecular condensates may be promising for the successful development of pesticides for treating plant diseases that are difficult to control. Tomato spotted wilt virus (TSWV) belongs to the family Tospoviridae and is one of the most devastating plant viruses worldwide, causing an annual economic loss of over one billion US dollars28. The use of antiviral agents can effectively slow down the occurrence and spread of viral diseases, making it one of the effective control strategies. But currently, commercialized drugs cannot actually serve the purpose of managing TSWV-induced disease29. Therefore, developing an effective anti-TSWV agents is highly important. The few proteins encoded by the TSWV genome play important roles in virus infection. Among them, the nucleocapsid protein (N) is an important component of the core template ribonucleoprotein (RNP) for viral replication30 and has multiple biological functions in the viral life cycle. The genomic RNA of viruses can be well protected in RNPs against RISC-mediated cleavage31. Similar to animal-infecting Bunyaviruses32, TSWV relies on a cap-snatching mechanism to initiate transcription of its genome33. The N proteins partially colocalize with processing bodies (PBs) and stress granules (SGs), which are the sources of 5′ capped RNA34. Notably, the N protein is capable of forming highly motile cytoplasmic IBs that traffic on actin filaments35. Therefore, the N protein can serve as an important molecular target for the creation of anti-TSWV agents.

Chromones are an important class of natural active ingredients produced during secondary metabolism in plants. Their structure is easy to modify and is considered a good lead skeleton for drug discovery36. Chromone compounds derived from natural products have potential broad-spectrum antiviral effects against human viruses, including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), herpes simplex virus type 1 (HSV1), influenza A virus (IAV), and hepatitis B virus (HBV)37,38. We previously synthesized several types of chromone compounds that possess good inhibitory activity against plant viruses and found that they have good binding ability to coat proteins (Fig. 1a)29,39. Dithiocarbamates (DTCs) serve as a broad-spectrum pharmacophore widely present in marketed drugs across the medical and agricultural fields40. By integrating DTC with various bioactive scaffolds, the compounds’ interaction with targets can be enhanced, thereby significantly boosting their biological activity41,42,43. Cationic amphiphilic drugs (CADs) are a class of commercialized drugs with a wide range of medical activities44. They are characterized by a common structure, namely, hydrophobic aromatic side chains containing an amine functional group, which is crucial for their pharmacological activity by increasing the pH in acidic intracellular compartments45. The phase separation modulators obtained through high-throughput screening methods appear to possess the properties of CADs (Fig. 1b)46,47,48,49,50. Notably, an increasing number of studies emphasize the crucial role of pH in the formation of condensates51,52,53,54. Therefore, we propose that designing a compound with CAD properties may be more successful in discovering a biomolecular condensate modulator.

Fig. 1: Design strategy for biomolecular condensate-inhibiting phytovirucides.
figure 1

a The structures of chromone and its antiviral analogs. b The structures of reported aggregation regulators containing tertiary amines. c Rational design of phase-separated (PS) condensates-inhibiting phytovirucides in the study.

In this work, we design a series of chromone compounds containing hydrophilic amine groups with piperazine dithiocarbamate as the bond bridge and obtain the target compounds in high yields (Fig. 1c). We find that the introduction of tertiary amine groups significantly enhances the anti-TSWV activities of these compounds. The highly active compound Z9 can target the N protein to inhibit the replication level of the viral ribonucleoprotein (RNP), thereby affecting virus proliferation in vivo. Application of Z9 can lead to a decrease in N protein condensation in plant cells. Furthermore, Z9 acts on condensates with phase separation properties, not on large solid-like granules. Interestingly, the reduction in the number of N condensates affects the stability of the N protein in plants. Moreover, we demonstrate that key sites or the addition of Z9 significantly affect the ability of the N protein to bind with RNA to form condensates. Overall, this study points to an important direction for pesticide discovery and crop virus control.

Results

Anti-TSWV activity

The inhibitory activities of the target compounds Z1-Z25 against TSWV were assessed via the half-leaf method, with Nicotiana glutinosa serving as the model plant. Preliminary screening at a concentration of 500 µg/mL revealed that the curative activity range of compounds Z1-Z25 against TSWV ranged from 43.5% to 64.4%, protective activities varied between 43.1% and 65.4%, and inactivation activities from 41.4% to 77.3%. Among them, compounds Z9 (77.3%), Z13 (72.8%), Z14 (70.6%), and Z17 (74.2%) had excellent inactivation activity against TSWV, outperforming the control ningnanmycin (69.3%), and were significantly superior to ribavirin (43.7%). We found that the inhibitory activities of compounds (Z1-Z7) without tertiary amine groups on TSWV were average. The EC50 values of the compounds were further tested, and the results revealed that compound Z9 exhibited the best inactivation activity against TSWV, with inhibitory activity of 65.3 µg/mL, which was significantly better than that of the best anti-plant viral agent currently available on the market, ningnanmycin (149.4 µg/mL), and another commonly used antiviral agent, ribavirin (801.5 µg/mL) (Table 1). The above results indicate that the introduction of a tertiary amine substantially enhances the anti-TSWV activities of these compounds.

Table 1 Anti-TSWV activity assay in vivo

Furthermore, the inhibitory effect of compound Z9 on TSWV was validated in tomatoes in addition to model plants. The tomato plants sprayed with 3% Tween 80 presented severe symptoms, including curled leaves, dark brown spots, delayed development, and yellowing. In contrast, plants treated with compound Z9 showed a significant alleviation in symptom severity, with improved growth and larger, greener leaves compared to untreated controls (Fig. 2a). These findings could indicate that compound Z9 significantly reduces the proliferation of TSWV in plants. The broad-spectrum antiviral activity of compound Z9 was further evaluated against a panel of plant viruses, and the results revealed that the inhibition rates of compound Z9 at 500 µg/mL against CMV, PVY, PMMoV, and TMV were 76.6%, 73.4%, 70.8%, and 79.0%, respectively. These rates are comparable to those of ningnanmycin (74.2%, t-test, P  =  0.5168; 77.6%, t-test, P  =  0.0422; 72.0%, t-test, P  =  0.6586; and 83.1%, t test, P  =  0.2275) and significantly better than those of ribavirin (52.4%, t-test, P  =  0.0013; 55.3%, t-test, P  =  0.0059; 56.5%, t test, P  =  0.0049; and 54.0%, t test, P  =  0.0036) (Supplementary Fig. 1ah). This result indicates that Z9 is a compound with broad-spectrum antiviral activity against plant viruses.

Fig. 2: Action mode of compound Z9 inhibiting TSWV.
figure 2

a Compound Z9 could inhibit the proliferation of TSWV in tomatoes from the appearance of symptoms. b Schematic diagrams of the constructs SR(+)eGFP and L(+)opt. L(+)opt, a full-length infectious clone of TSWV L genomic RNA fragment with optimized RNA-dependent RNA polymerase (RdRpopt); SR(+)eGFP, an infectious clone of TSWV S genomic RNA with nucleocapsid protein (N), in which NSs of TSWV S genome RNA fragment is replaced by eGFP. Plus sign (+) and 3′ to 5′ designation indicate the antigenomic RNA of TSWV; 2×35S, a double 35S promoter; HH, hammerhead ribozyme; RZ, hepatitis delta virus (HDV) ribozyme; NOS, nopaline synthase terminator. c The amounts of fluorescence in the infiltrated leaves were observed at the 5th day. Bars, 500 µm. The experiment was repeated three times showing similar results. d Western blot was used to detect the accumulation level of protein N in the infiltrated leaves. Staining of RuBisCO with Ponceau S was used as a sample loading control. The experiment was repeated three times showing similar results. e Pattern diagram of continuous detection of binding signal values between small molecules and N protein using BLI. f The binding signal values between compounds with different activities and N protein. Compounds Z1Z6 are compounds with average activities, while compounds Z8, Z9, Z12, Z13, Z14, and Z16 are compounds with good activity. g The binding ability of compound Z9 with wild-type N protein (NWT) was measured by BLI. Source data are provided as a Source Data file.

Compound Z9 inhibits normal replication of RNPs by targeting the N protein

The ribonucleoprotein complex (RNP) of TSWV is responsible for executing the transcription and replication of the viral genome, playing a crucial role in the viral lifecycle28,30,31. The competent minigenome replication system (L(+)opt+ SR(+)eGFP + VSRs) (Fig. 2b) can be used to study the replication levels of reconstituted RNP55. To investigate the mode of action of compound Z9, a replication system was used to detect the effect of compound Z9 on the replication of RNP in Nicotiana benthamiana. We observed that the fluorescence levels in the infiltrated leaves, monitored 5 days post-application of compound Z9, were significantly lower than those in the untreated leaves, as determined by fluorescence microscopy (Fig. 2c). Western blot analysis further revealed that the accumulation levels of N protein in infiltrating leaves were consistent and considerably decreased relative to those in the untreated leaves (Fig. 2d). These findings indicate that compound Z9 effectively inhibits the replication of the RNP of TSWV in plants. The N protein, on the other hand, serves as the main assembly protein of the viral RNP28,30,31 and was explored as a potential target for these compounds. To assess this, His-tagged N protein was fixed to the sensor (Ni-NTA), and compounds with different inhibitory activities were used for rapid detection of binding signal values (Fig. 2e). The results of the bio-layer interferometry (BLI) assay revealed that compounds with poor activity presented weak binding signals, essentially below 0.10 nm. Conversely, compounds with potent activity presented higher binding signal values with the N protein (Fig. 2f), although a consistently strong positive correlation between activity and signal values was not apparent for every compound. For example, compounds Z1 and Z8 presented lower binding signal values than compounds Z2 and Z12 did. The introduction of amine groups greatly improved the binding of the compounds to the N protein. The binding signal between compound Z9 and the N protein was the highest, with an affinity constant of 7.6 µM (Fig. 2g). These findings suggest that the N protein may serve as a target for this class of compounds. In addition, this method is beneficial for rapid screening of small molecules that target N proteins in vitro. To confirm that the N protein is a direct target of compound Z9, a probe was synthesized by introducing an alkyne group into the side chain of compound Z9 (Supplementary Fig. 2a, b). The results of the activity assay indicated that the probe possessed good inhibitory activity against TSWV (Supplementary Table 1). The competitive binding experiment result showed that the probe and compound Z9 had the same binding region for the N protein (Supplementary Fig. 2c). We found that the protein labeled by the probe in the total protein of healthy leaves and diseased leaves seem to be similar. However, pull-down assays clearly demonstrated that the probe significantly captured the N protein from diseased leaf extracts (Supplementary Fig. 2d). These results provide direct evidence that the N protein is a target of compound Z9.

Compound Z9 significantly reduces N protein condensates

The TSWV N protein can form motile cytoplasmic IBs to promote the proliferation of viruses33,34,35. We aimed to observe the impact of compound Z9 on the behavior of N protein condensates in vivo. N-YFP was transiently expressed in N. benthamiana leaves, and after 18 h, the diluted phase of N-YFP was clearly visible. Compound Z9 or DMSO was injected into the left side of the same infiltrated leaves, and the formation of IBs was monitored at 36 and 72 h post-injection via z-stack scanning (Fig. 3a). Compared with DMSO, compound Z9, at a concentration of 100 µg/mL, significantly reduced the number of small IBs or their diffuse distribution in the leaf cell infiltration area at 36 h, and the number of IBs in the leaves treated with compound Z9 was reduced by more than half at 72 h (t-test, P  =  0.0001) (Fig. 3b, c). The accumulation levels of the N-YFP protein were further detected. The results indicated a significant reduction in the accumulation level of the N-YFP protein following the addition of compound Z9 (Fig. 3d). In N. benthamiana cells, highly motile cytoplasmic IBs formed by N-YFP can be clearly observed at 36 h (Supplementary Movie 1). To explore whether compound Z9 has an elimination effect on small particles, N-YFP was transiently expressed in N. benthamiana leaves, and after 36 h, drugs were injected into the infiltration area, and the IBs 5 h post-injection were observed (Supplementary Fig. 3a). Compound Z9, at a concentration of 100 µg/mL, significantly reduced the number of IBs in the leaf cell infiltration area (t test, P  =  0.0001), whereas compound Z3, with less effective anti-TSWV activity, had a relatively weaker ability to disperse the condensates (t-test, P  =  0.0075) (Supplementary Fig. 3bd). The above results revealed that compound Z9, with an amine group, could effectively inhibit the formation of aggregates involving the N protein.

Fig. 3: Compound Z9 significantly disrupt the formation of IBs.
figure 3

a Schematic method of exploring the effect of compound Z9 on the formation of IBs in vivo. b Representative fluorescence images showing compound Z9 could significantly reduce the formation of IBs in N. benthamiana leaves. DMSO was used as a control. The experiments were repeated at least three times with similar results. Bars, 50 μm. c The total number of IBs in confocal images at 72 h. Data are shown as individual data points and mean ± SD (n = 4). Statistical P-values were calculated using the two-tailed independent samples t test (P = 0.0001), ***,P < 0.001. d Western blot was used to detect the accumulation level of protein N-YFP in the infiltrated leaves. Staining of RuBisCO with Ponceau S was used as a sample loading control. N. benthamiana plants were infiltrated with Agrobacterium containing empty vector (EV) as the mock control. The experiment was repeated three times with similar results. e Representative fluorescence images showing compound Z9 could significantly degrade small IBs in N. benthamiana protoplasts. Bars, 10 μm. The experiment was repeated three times with similar results. Source data are provided as a Source Data file.

Although several types of anti-TSWV compounds targeting N proteins have been previously reported, their impact on IBs has not been explored29,56,57. Herein, we further investigated whether other highly active compounds similar to CADs can disrupt the aggregation of the N protein. Piperazine derivatives containing ethyl morpholine indeed inhibited IB formation (Supplementary Fig. 3e, f). Further extraction and transformation of N. benthamiana protoplasts revealed that N-eGFP formed condensates within the cells, and within 30 min, larger N-eGFP particles remained condensed in intact cells after treatment with compound Z9 at a concentration of 10 µg/mL, whereas smaller N-eGFP particles disappeared in intact cells (Fig. 3e). Therefore, the above experimental results reveal the potential mechanism by which drugs targeting the N protein may act by degrading small condensates. Although compound Z9 could act on plant proteins, we found that Z9 had not any adverse effects on the growth and development of plants (Supplementary Fig. 4a). To investigate whether compound Z9 have the same effects of on the other viral proteins, the binding capabilities of compound Z9 with PVY CP and PMMoV CP were first tested. The results showed that the binding constants of compound Z9 with these two proteins were 67 μM and 18 μM, respectively (Supplementary Fig. 4bd), indicating a good binding capability between compound Z9 and PMMoV CP. Thus, PMMoV CP in leaves was treated using the same method. Through fluorescence observation, we found that PMMoV CP did not significantly decrease at 5 h and 18 h after compound Z9 treatment (Supplementary Fig. 4e, f), while compound Z9 could significantly reduce the number of TSWV N protein condensates in cells. This seems to suggest that Z9 is more easily recruited to condensates rather than dilute phases and Z9 has a unique mechanism of action against TSWV.

N protein condensates in vivo possess phase separation properties

We further delved into the mechanism behind the reduction in the number of condensates observed upon the addition of compound Z9. Within cells, proteins and nucleic acids, which are biological macromolecules, can undergo a process known as phase separation to form membraneless organelles3,4,5,6. With this premise, the phase separation potential of the N protein was assessed using the Predictor of Natural Disordered Regions (PONDR) VL-XT algorithm. This sequence-based prediction is based on the IDR of the N protein, which is located primarily within its central core domain (Supplementary Fig. 5a). Encouraged by these bioinformatic insights, we pursued experimental verification to discern whether N protein condensates could indeed engage in phase separation activities. To this end, a construct encoding N-YFP was transiently expressed in N. benthamiana leaves. We observed that after 36 h, the N protein formed a relatively large number of small particles within the leaves, and by 72 h, the particles had significantly increased in size, but their number had significantly decreased (Fig. 4a). Time-lapse confocal imaging revealed these large inclusions at 36 h occasionally divided slowly, spawning smaller ones that dispersed (Fig. 4b and Supplementary Movie 2). Indeed, the behavior of small, dynamic inclusions that frequently merged and coalesced into larger formations (Fig. 4c and Supplementary Movie 3). The characteristics of these inclusions, such as their sphericity, motility, and fusion/fission, as observed in N-YFP-infected cells, reveal that the IBs may exhibit properties akin to those of liquid droplets. At 36 h, the small IBs exhibited high-speed mobility within the cells, making it difficult to perform fluorescence recovery after photobleaching (FRAP) experiments. However, at 72 h, the particle mobility was significantly reduced. The results of the FRAP assay indicated that, after 72 h, the fluorescence intensity of small individual puncta was recovered, whereas the intensity of large individual puncta was difficult to recover (Fig. 4d–g).

Fig. 4: N protein condensates possess gel-like phase separation properties.
figure 4

a The granules size of N-YFP formed in the N. benthamiana leaf cells at 36 hpi and 72 hpi. b Fluorescence time lapse confocal images showing the fission of IBs. The white circles indicated the IBs undergoing fission at 36 hpi. Bars, 10 μm. c Fluorescence time-lapse confocal images showing the fusion of IBs. The white circles indicated the IBs undergoing fusion at 36 hpi. Bars, 10 μm. d Fluorescence recovery after photobleaching (FRAP) of small IBs formed in N. benthamiana cells. The images were taken every 2.5 s for 240 s to document fluorescence recovery, each time point was normalized to before photobleaching. Bars, 20 μm. e Quantification of the small N–YFP condensates in FRAP assays. (n  =  4 puncta). f Fluorescence recovery after photobleaching (FRAP) of large IBs formed in N. benthamiana cells. The images were taken every 2.5 s for 220 s to document fluorescence recovery, each time point was normalized to before photobleaching. Bars, 20 μm. g Quantification of the large N–YFP condensates in FRAP assays. (n  =  1 puncta). h The effect of 1,6-hexanediol on N-YFP granules at 36 hpi and 72 hpi, and the effect of compound Z9 on N-YFP granules at 72 hpi. The experiments were repeated at least three times with similar results. Bars, 20 μm. Source data are provided as a Source Data file.

Typically, phase-separated proteins achieve a photobleaching recovery rate of over 80%, but we observed consistently low levels of recovery ( 10% to  30%), which seems to indicate that the N protein condensates tend towards gel-like properties. To further confirm this, we conducted FRAP experiments with N-eGFP puncta from protoplasts, and the results revealed that complete bleaching of small individual puncta occurred at similarly low levels ( 10% to  30%), and large puncta were difficult to recover (Supplementary Fig. 5b-e). This phenomenon is consistent with linker histone H1 in the chromatin or Early Heading Date 6 (EHD6), which can bind to nucleic acids to form RNP58,59. To further explore whether these granules possess phase-separated properties, we treated the samples with 1,6-hexanediol, and the results revealed that treating the leaves at 36 h led to a significant reduction in particle number, whereas treating the leaves at 72 h made it difficult to observe a significant decrease in the number of large particles. Concurrently, we also treated the leaves with compound Z9 and found that large particles were hard to eliminate (Fig. 4h). The phenomenon is consistent with the treatment of large N-eGFP particles in protoplasts (Fig. 3e). Together, these data demonstrate that the N protein undergoes phase separation from the surrounding cytoplasm, forming small phase-separated condensates with gel-like properties or transitions into larger solid-like granules, whereas compound Z9 is more effective in the early stages and has difficulty disrupting the formation of large granules.

The absence of potential binding sites reduces the proliferation of viruses in vivo

To explore how compound Z9 affects the reduction of N protein condensates, we investigated the key binding sites of compound Z9 on the N protein. We used different molecular docking software to predict the binding sites of compound Z9 with the N protein and found that compound Z9 could potentially interact with amino acids Arg94 (R94) and Tyr184 (Y184) (Fig. 5a and Supplementary Fig. 6a, b). TSWV N protein possesses a compact central domain, flanked by extended N- and C-terminal lobes on either side, forming a positively charged groove that is essential for RNA binding28,60. In the crystal structure of the N protein trimer complexed with nucleic acid (PDB ID: 5IP3), amino acids R94 and Y184 of each chain interact with the nucleic acid61. Further analysis revealed that the position at which Z9 interacts with RNA in the N protein has a high degree of overlap (Supplementary Fig. 6c). To validate the effects of the potential binding sites of compound Z9 on RNP proliferation, reverse PCR amplification and recombination were employed to construct the SR(+)eGFPR94A, SR(+)eGFPG147A, SR(+)eGFPY184A, and SR(+)eGFPR94A&Y184A mini-replicons (Fig. 5b and Supplementary Fig. 7ac). N. benthamiana plants were then infected using a competent minigenome replication system (RNPeGFP), and fluorescence was observed on the 5th day. The results showed that the absence of the amino acid R94 marginally reduced the fluorescence level. Conversely, the absence of G147 did not alter the intensity of fluorescence in leaf cells, whereas the absence of Y184 or R94&Y184 diminished fluorescence levels (Fig. 5c). The average fluorescence intensity was analyzed according to three independent biological replicates. The results revealed that RNPeGFPWT and RNPeGFPG147A had the strongest average fluorescence intensity in leaf cells (t test, P = 0.3054), RNPeGFPR94A slightly decreased (t-test, P = 0.0064), and RNPeGFPY184A was the weakest among all single-point mutations (t test, P = 0.0001). Compared with the control, RNPeGFPR94A&Y184A weakened more significantly (t test, P < 0.0001) (Fig. 5d). Additionally, the accumulation levels of N protein were further detected. Compared with RNPeGFPWT, the lack of Y184 or R94&Y184 significantly reduced the accumulation level of N protein, whereas the absence of R94 slightly reduced the accumulation of N protein (Fig. 5e).

Fig. 5: The potential binding sites of compound Z9 and N protein significantly affect the replication of RNP and the proliferation of TSWV.
figure 5

a Analysis of binding sites of compound Z9 and N protein by molecular docking. b Schematic representation of constructs expressing TSWV full-length antigenomic L(+)opt with optimized RdRp, full-length genomic M(−)opt with optimized glycoproteins (GP) and the viral movement protein (NSm), full-length antigenomic SR(+)eGFPMut, and the sequence encoding amino acids (R94A, G147A, Y184A, or R94A&Y184A) on the N protein gene were mutated to that of alanine. Minus sign (−) and 5′ to 3′ designation indicate the genomic RNA of TSWV. c Constructs (L(+)opt + SR(+)eGFPWT) or its mutant (L(+)opt + SR(+)eGFPMut) with three RNA silencing virus inhibitors VSRs (P19, Hc-Pro, and γb) co-expression in plants together with at least three plants in each experiment, the expression of eGFP protein in leaves was observed by fluorescence microscope on the 5th day. Bars, 500 μm. d The average fluorescence intensities in infiltrated leaves on the 5th day) was measured by ImageJ. Error bars indicate ± SD of the mean three independent experiments (n  =  3). Statistical P-values were calculated using the two-tailed independent samples t-tests (**, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, P > 0.05 not significant). R94A (P = 0.0064), G147A (P = 0.3054), Y184A (P = 0.0001), R94A&Y184A (P < 0.0001). e The accumulation level of N protein in the infiltrated leaves was detected by western blot. Staining of RuBisCO with Ponceau S was used as a sample loading control. The experiment was repeated three times with similar results. f Phenotypes of TSWVWT and TSWVMut (TSWVR94A, TSWVG147A, TSWVY184A, or TSWVR94A&Y184A) -infected plants on the 21st day. Constructs (L(+)opt + SR(+)eGFPWT + M(-)opt) or its mutant (L(+)opt + SR(+)eGFPMut + M(-)opt) with three RNA silencing virus inhibitors VSRs (P19, Hc-Pro, and γb) co-expression in plants together with at least ten plants in each experiment. g The infection rates of TSWVWT and TSWVMut -infected plants on days 14, 17, and 21. Error bars indicate  ±  SD of the mean three independent experiments (n  =  3). h The percentages of TSWVWT and TSWVMut -infected plants with different disease symptom grades. Error bars indicate ± SD of the mean three independent experiments (n  =  3). i The accumulation level of N protein in the systemic leaves was detected by western blot. Staining of RuBisCO with Ponceau S was used as a sample loading control. The experiment was repeated three times with similar results. Source data are provided as a Source Data file.

To verify the impact of missing binding sites on TSWV pathogenicity in plants, the movement-competent minigenome replication system (M(-)opt) was added to complement the cell-to-cell and systemic movements of reconstituted ribonucleoprotein complexes (RNPs). By the 14th day, noticeable pathogenic symptoms, such as leaf wrinkling and spotting, were evident in the systemic leaves of TSWVWT-infected plants. The severity of symptoms in TSWV-infected plants was divided into three grades (Supplementary Fig. 8a, b). By the 21st day, nearly all of the plants had significant wilting and mortality (grade III), with a disease susceptibility rate of 100%. In the case of TSWVG147A, more than 50% of the plants displayed severe disease symptoms (grade III) in systemic leaves, with an overall infection rate surpassing 70%. These findings suggest that the absence of G147 somewhat delayed the systemic infection of plants by the virus. For TSWVR94A or TSWVY184A, symptom onset in system leaves was slower and considerably milder than that of TSWVWT, with a susceptibility rate of approximately 50% (grade I and grade II). In contrast, plants infected with TSWVR94A&Y184A presented fewer spots on systematic leaves and a lower incidence rate of approximately 20% (grade I and grade II). Therefore, through a comprehensive comparative analysis of symptoms and infection rates, we found that the sole absence of R94 or Y184 effectively hindered the virus’s systemic spread, and their combined absence drastically reduced it (Fig. 5f–h). The order of N protein accumulation level in the infected system leaves was TSWVWT > TSWVG147A > TSWVR94A > TSWVY184A > TSWVR94A&Y184A (Fig. 5i). Further treatment of TSWVWT and TSWVR94A&Y184A with compound Z9 revealed that the pathogenic symptoms of TSWVWT were significantly improved, whereas those of TSWVR94A&Y184A remained unchanged (Supplementary Fig. 8c). Together, these findings underscore that compound Z9 may target nucleic acid binding sites and exert a crucial impact on viral proliferation in vivo by affecting RNP replication.

Condensates are beneficial for the stability of the N protein

To further explore the interaction between compound Z9 and the N protein, mutant proteins (NR94A, NY184A, and NR94A&Y184A) were obtained via prokaryotic expression and purification (Supplementary Fig. 9a, b). BLI was employed to examine the binding dynamics between compound Z9 and the N protein. Compared with that of NWT (7.6 µM), the affinity ability of compound Z9 for the double mutant NR94A&Y184A was significantly diminished, with a KD value exceeding 400 µM (Fig. 6a). The microscale thermophoresis (MST) results indicated that the affinity constants of NR94A&Y184A with compound Z9 decreased 180-fold (Fig. 6b). Similarly, the affinity constants for the single mutant proteins NR94A and NY184A with compound Z9 were reduced by approximately 103 and 135 times, respectively (Supplementary Fig. 9c, 9d). To further verify that the absence of the binding sites affects the interaction between the N protein and Z9, a labeling experiment was performed by incubating equal amounts of the probe with equimolar concentrations of NWT and NR94A&Y184A. We found the probe that NWT captured significantly more than that of NR94A&Y184A (Fig. 6c). These results indicate that mutations at residues (R94 and Y184) within the nucleic acid binding region of the N protein significantly reduced the binding affinity of the compound Z9, suggesting that Z9 targets the nucleic acid binding region of the N protein.

Fig. 6: The key binding sites significantly reduce the aggregation of N protein in vivo.
figure 6

a The binding ability of compound Z9 with NR94A& Y184A was measured by BLI. b The binding ability of compound Z9 with NR94A& Y184A was measured by MST. The fitted curves and KD values calculated by Mo. Affinity Analysis (v.2.2.4). Each binding assay was repeated three times independently (n = 3), and the bars represent SD. c Labeling experiment indicated that the probe bound by NWT significantly more than that of NR94A& Y184A. d Confocal micrographs showing the localization patterns of green fluorescence derived from NWT-YFP and NR94A& Y184A-YFP. Bars, 20 μm. The experiments were repeated at least three times with similar results. e The accumulation level of N-YFP protein in the infiltrated leaves was detected by western blot. Staining of RuBisCO with Ponceau S was used as a sample loading control. f Protein stability assay. Total protein was crudely extracted using PBS, allowed to stand at 0, 30, 100, and 160 min, and then the accumulation level of N protein was detected by western blot. Actin antibody was used as the internal control. g The purified protein (sfGFP-NWT or sfGFP-NR94A& Y184A) was identified by SDS-PAGE and showed green fluorescence. h R94 and Y184 mutations disrupt RNA mediates N condensation. The condensates of sfGFP-NWT and RNA were larger than that of sfGFP-NR94A& Y184A and RNA under the enlarged field of view. 50 ng/µL extracted RNA was incubated with 10 μM sfGFP-NWT or sfGFP-NR94A& Y184A protein at 25 °C for 10 min. Bars, 20 μm. The experiments were repeated three times with similar results. Source data are provided as a Source Data file.

To investigate the influence of potential binding sites on the formation of IBs, we engineered a construct in which the NR94A&Y184A protein was transformed into a yellow fluorescent protein (NR94A&Y184A-YFP) (Supplementary Fig. 10ac). When transiently expressed in N. benthamiana leaves, NWT-YFP formed numerous discrete cytoplasmic inclusions; in stark contrast, NR94A&Y184A-YFP was predominantly dispersed throughout the cytoplasm at 36 h, indicating a diffuse distribution pattern. At 72 h post-treatment, we observed that NWT-YFP condensed into larger particles, whereas the distribution of NR94A&Y184A-YFP was significantly diminished (Fig. 6d). Western blot analysis revealed that the accumulation level of NR94A&Y184A-YFP markedly decreased after 72 h, suggesting that NR94A&Y184A-YFP might be degraded within the plant (Fig. 6e). These observations suggest that mutations at binding sites markedly decrease IB formation and are essentially consistent with the phenomenon of adding compound Z9.

To verify whether the stability of the N protein was reduced after the R94 and Y184 mutations, we conducted a protein stability assay. The results indicated that NWT, NR94A, and NY184A maintained relatively good stability over different time periods, whereas NR94A&Y184A tended to degrade (Fig. 6f), implying that NR94A&Y184A is more susceptible to degradation in the body. During this process, NR94A&Y184A-YFP never formed many particles within the cells. Considering that NR94A&Y184A-YFP is subject to degradation in plants, we purified recombinant His-sfGFP-NWT from Escherichia coli and examined its aggregation under conditions of relatively high concentration and low salinity (Fig. 6g). Confocal microscopy analysis revealed no significant aggregation of the His-sfGFP-NWT protein. RNP granules assemble through phase separation of their molecular components, and RNA is thought to be a critical element for the sizing and composition of the condensates62,63,64,65,66. N protein is a nucleic acid-binding protein, and the condensates it forms are similar to some RNPs, possessing gel-like properties58,59. Considering these preliminary studies, our focus shifted to investigating whether nucleic acids could mediate the formation of condensates. To explore this possibility, total RNA was extracted from TSWV-infected N. benthamiana leaves. His-sfGFP-NWT displayed a phase transition in the presence of RNA, rapidly forming larger, aggregate-like structures at high concentrations of RNA. RNA extracted from healthy N. benthamiana could also induce the aggregation of N protein, indicating that the binding of N protein to RNA is not specific (Supplementary Fig. 11a), which is consistent with previous conclusion67. However, these condensates exhibited a solid-like physical state rather than a fluidic state (Supplementary Fig. 11b). This rapid solidification parallels phenomena observed in liquid-to-solid phase separation-induced oskar ribonucleoprotein granules or N-terminal Toll/interleukin-1 receptor (TIR) domain proteins68,69. Their formation depends not only on the aggregating protein but also on environmental factors, such as pH, salt concentration, crowding, and the presence of membranes and condensates70. To verify whether mutations at the binding sites R94 and Y184 in the N protein significantly affect RNA-induced N protein condensation, we generated His-sfGFP-N mutants by substituting the residues with alanine residues R94A/Y184A (His-sfGFP-NR94A&Y184A). Confocal microscopy data revealed that both the number and size of N protein condensates were greatly reduced by binding site mutation (Fig. 6h). These results suggest that the stability of the N protein in plants is related to RNA-mediated N protein condensation.

Compound Z9 inhibits nucleic acid-mediated N protein condensation

To further investigate the impact of these specific amino acids on the ability of the N protein to bind RNA, a digoxigenin-labeled RNA probe was prepared, and an electrophoretic mobility shift assay (EMSA) was conducted. We found that in the presence of NWT, the rate of RNA migration began to change slightly at a protein dosage of 0.1 µg, with a significant decrease observed at dosages exceeding 0.1875 µg. Conversely, for the NR94A&Y184A mutant, a slight slowing in RNA migration was noted at a protein dosage of 1.5 µg, with a substantial decrease only evident at dosages greater than 3.0 µg (Supplementary Fig. 11c). These results indicate that the absence of these binding sites significantly compromises the RNA binding capability of the N protein. We next immobilized biotin-labeled RNA (100 nM) onto streptavidin (SA) biosensors, and the binding affinities of NWT and NR94A&Y184A with RNA were measured via BLI assay. The results suggested that the affinity constants (KDs) for NWT and NR94A&Y184A binding to RNA were 7.3 nM and 36 nM, respectively (Fig. 7a, b), indicating that the mutation reduced the RNA-binding affinity of the N protein by approximately 5-fold. We observed that NWT exhibited faster association kinetics and slower dissociation kinetics compared to NR94A&Y184A at low concentrations. Furthermore, when the biosensor was continuously loaded with NR94A&Y184A and NWT, a significant increase in the binding curve was detected (Fig. 7c). Upon reversing the loading sequence, the binding curve decreased significantly (Supplementary Fig. 11d). These findings demonstrate that the absence of these binding sites significantly impairs the RNA-binding capacity of the N protein.

Fig. 7: Compound Z9 inhibits nucleic acid mediate N protein condensation.
figure 7

a The binding ability of RNA with NWT was measured by BLI. b The binding ability of RNA with NR94A& Y184A was measured by BLI. c Binding curve of continuous injection of NR94A& Y184A and NWT protein. d Competitive binding experiment confirmed that Z9 blocking the binding of RNA and N protein. The experiments were repeated three times with similar results. 5 µM Z9 was first incubated with 12.5 nM NWT protein at 25 °C for 30 min, and then detected using SA sensors with fixed RNA. e The experiment where compound Z9 blocked the binding of protein N to RNA, leading to the degradation of protein N by proteases. In this experiment, protein N, RNA, and Z9 were incubated together at 25 °C for 30 min, then added to the total protein extracted from healthy leaf tissue in PBS buffer, and further incubated at 25 °C for 0, 30, 100, and 160 min respectively. Detection was carried out by immunoblotting. The use of a ultraNuclease to remove nucleic acids serves as a control. f The accumulation level of N protein was detected by western blot. g Quantitative analysis of the accumulation levels of N protein in each treatment group using ImageJ. Each treatment was independently repeated in triplicate experiments. Error bars indicate ± SD of the mean three independent experiments (n  =  3). Statistical P-values were calculated using the one-tailed paired samples t-tests (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, P > 0.05 not significant). The same group of samples was compared for differences at three time points (30, 100, and 160 min) relative to 0 min. NWT + RNA (30 min, P = 0.2445; 100 min, P = 0.0997; 160 min, P = 0.4256), NWT + Z9 + RNA (30 min, P = 0.2867; 100 min, P = 0.0199; 160 min, P = 0.0039), NWT + RNA + UltraNuclease (30 min, P = 0.0449; 100 min, P = 0.0214; 160 min, P = 0.0004), NWT + Z9 + RNA + UltraNuclease (30 min, P = 0.0427; 100 min, P = 0.0086; 160 min, P = 0.0027). h Proposed model for Z9 as phase-separated (PS) biomolecular condensate inhibitor. Source data are provided as a Source Data file.

To evaluate the binding efficiency of compound Z9 with the N-RNA complex, we first determined the affinity constant (KD = 150 µM) between compound Z9 and RNA (Supplementary Fig. 11e). The interaction between compound Z9 and the N protein (KD = 7.6 µM) was significantly stronger than that of Z9-RNA, indicating a preference of compound Z9 for binding to the N protein. In the competitive binding assay, we discovered that the binding signal value of the N protein and RNA (0.26 nm) was markedly greater than that of the mixture of compound Z9 and N protein (0.17 nm) (Fig. 7d). The constants of the binding rate and dissociation rate revealed that prebinding of compound Z9 to the N protein significantly blocked RNA binding. To further explore whether compound Z9 could destabilize the N protein-RNA complex, a mixture of NWT and RNA or a mixture of NWT, RNA, and Z9 was added to total protein extracts, with or without UltraNuclease, and incubated for different periods of time. The stability of the N protein was subsequently assessed by western blotting (Fig. 7e). We found that the addition of compound Z9 indeed caused degradation of the N protein (NWT + Z9 + RNA, one-tailed paired t-test, 30 min, P = 0.2867; 100 min, P = 0.0199; 160 min, P = 0.0039), and the stability of the N protein significantly decreased after the removal of nucleic acids (NWT + RNA + UltraNuclease, one-tailed paired t test, 30 min, P = 0.0449; 100 min, P = 0.0214; 160 min, P = 0.0004) (Fig. 7f, g). These results suggest that Z9 can block the binding of the N protein to RNA, leading to the degradation of the N protein. Overall, it was confirmed that condensates are caused by the interaction between the N protein and RNA, and these condensates can protect the protein from being extensively degraded within the cell. Compound Z9 is an inhibitor targeting N, which can act on the dilute phase of the N protein or non-solidified condensates, disrupting the formation of these condensates by interfering with the normal protein-nucleic acid interactions, thereby further accelerating the degradation of the N protein (Fig. 7h).

Discussion

Biomolecular condensate modulators have received much attention in the pharmaceutical field17,18,19,20,21,22,23; however, their application in agriculture remains highly limited. Consequently, based on the advantages of the privileged scaffold (benzopyrone and dithiocarbamate) in drug discovery36,37,38,39,40,41,42,43, a series of benzopyrone compounds were rationally designed and synthesized by analyzing the structures of biomolecular condensate modulators in the pharmaceutical field (Fig. 1). We discovered that the introduction of amine groups into the side chain significantly improved the inhibitory activity of the compounds against TSWV (Table 1). Further investigations into the mechanism of action revealed that the highly active compound Z9 could inhibit the replication of viral RNPs (a complex of N protein, RNA, and RdRp) in vivo. Subsequent BLI screening indicated that the introduction of an amine group significantly improved the binding of the N protein to the compounds. The binding signal values were positively correlated with their activity to some extent, with compound Z9 displaying the highest signal value (Fig. 2f). This suggests that the N protein is a potential target protein for Z9. The N protein is a direct target of Z9 was further confirmed by pull-down of the probe (Supplementary Fig. 2). Currently, in the drug discovery for treating plant viral diseases, the concentrations used in in vivo activity tests are relatively high, which may be related to the difficulty in control71. Z9 shows better activity compared to other types of compounds against TSWV29,56,72,73. We found that Z9 exhibits a certain degree of broad-spectrum antiviral activity against plant viruses due to the diverse biological activities of the privileged scaffolds, which indicates that Z9 is a valuable lead compound. Through continuous optimization, it can provide an important research foundation for the subsequent development of candidate compounds that remain effective at low concentrations and are cost-effective.

N protein is capable of forming highly motile cytoplasmic IBs that traffic on actin filaments, and partially colocalize with processing bodies (PBs) and stress granules (SGs), which is important for the TSWV33,34,35. When the N-YFP protein was transiently expressed in plants and treated with compound Z9, a significant reduction in the number of N protein condensates was observed (Fig. 3a–d and Supplementary Fig. 3). Furthermore, N-eGFP was expressed in protoplasts, and changes in condensates in specific cells after treatment with compound Z9 over time were observed. Large granules were difficult to eliminate, while the number of small granules significantly decreased (Fig. 3e). Considering that Z9 possess good inhibitory activity against the other plant viruses, we further explore whether Z9 could eliminate the other viral proteins. The binding constant of Z9 with PMMoV CP was 18 μM, indicating a good binding capability. However, we found that the CP in the dilute phase did not seem to show a significant decrease, while Z9 could significantly reduce the number of TSWV N protein condensates in cells (Supplementary Fig. 4). This suggests that Z9 has a unique mechanism of action against TSWV. Recent study indicated that drugs are more easily recruited to condensates with phase separation to enhance their efficacy, rather than dilute phases74. This may provide a reasonable explanation for Z9 could more effectively target N protein condensates. Phase separation is recognized as a fundamental principle driving the formation of condensates3,4,5,6. Based on these results, we explored the formation process of these condensates and found that at 36 h, the granules were numerous and small, and by 72 h, the granules were increasingly larger, with further observation revealing a process of fusion (Fig. 4a–c). FRAP experiments revealed that the small granules mostly possessed gel-like phase separation properties, whereas larger granules are mostly in the solid state. Treatment with 1,6-hexanediol also affected the small granules, whereas the large granules seemed unaffected. Moreover, treatment with compound Z9 had little effect on large particles (Fig. 4h). These results elucidate the role of compound Z9 designed with CADs in the early stages of condensate formation, which includes the dilution phase and the process of small granules fusing to form larger granules. Currently, in the existing research on phase separation modulators, the pathogenic condensates targeted possess liquid-like properties, and drugs mostly serve to harden them or blocking interactions between macromolecules18,19,20,21. Similar to the condensates formed by TSWV N in vivo, some condensates are found to develop characteristics of gel-like properties or undergo a phase transition to a solid state58,59,68,69, and there is a significant lack of drug discovery for similar pathogenic condensates. A recent study has indicated that gel-droplets have a higher affinity for drugs than liquid-droplets, facilitating drug distribution75; in fact, the physicochemical properties of drugs and condensates play a key role in the distribution of drugs to the condensates where their targets reside76. We discovered that the introduction of basic amine groups is critical, enabling their classification as pH-responsive drugs (CADs), which significantly alters the physicochemical properties of Z9. Our observations show that most biomolecular condensate inhibitors identified through high-content screening methods in the medical field possess alkalic groups20,42,43,44,45,46. Research has increasingly emphasized that physiological pH plays a significant role in regulating the formation of certain biomolecular condensates51,52,53,54, and of course, such studies involve the physiological environment inside cells, which adds to the complexity. As the mechanisms of condensate formation are increasingly discovered77, how to consider the physiological pH of condensate formation during the rational design of drug structures is highly important for the further development of biomolecular condensate modulators.

To further elucidate the mechanism of action of compound Z9, we found that the amino acids targeted by compound Z9 likely tend to be in the regions associated with nucleic acid binding. Further experimental results indicated that the absence of these potential binding sites (R94 and Y184) significantly inhibited the replication of RNP and alleviated the pathological symptoms of TSWV, which was consistent with the mode of action of Z9 (Fig. 5b–i). In vivo experiments demonstrated that at 36 h, the absence of R94 and Y184 led to a diffuse distribution of the N protein within the cell, significantly reducing the number of condensates. Interestingly, after 72 h, NR94A&Y184A-YFP seemed difficult to observe, and immunoblot experiments confirmed that the accumulation level of the mutated protein was significantly reduced (Fig. 6d–f). These findings suggest that NR94A&Y184A-YFP is more prominently degraded in plants. This degradation mechanism could be a way for plants to recognize and degrade pathogen effector proteins through their protein degradation mechanisms, preventing their function within plant cells78. Therefore, we speculate that the formed condensates favor the stability of the N protein. We further found that sfGFP-NWT alone does not spontaneously form condensates, but the addition of nucleic acids rapidly induces the aggregation of the N protein, and under the same conditions, the aggregation ability of sfGFP-NR94A&Y184A is significantly reduced (Fig. 6f). Thus, these condensates are not solely the N protein, but likely RNP particles formed after the N protein binds to nucleic acids. The addition of compound Z9 inhibited the binding of the N protein to RNA (Fig. 7d). We further designed experiments to verify that reducing nucleic acids or adding compound Z9 increased the susceptibility of the N protein to degradation (Fig. 7e–g). Thus, through comprehensive research, condensates were confirmed to form through the interaction of the N protein with RNA, and this condensate plays a protective role for the protein. Compound Z9, an inhibitor that targets condensates, disrupts the formation of these condensates by directly blocking protein‒nucleic acid interactions, further leading to the degradation of the N protein. Under normal circumstances, more attention is paid to the protective role of RNPs on nucleic acids, as well as transcription and translation65,66. A recent study found that phase-separated condensates can reduce protein degradation by compartmentalizing components of the proteasome machinery79. Leveraging the cell’s innate protein degradation systems to selectively degrade disease-associated pathogenic proteins represents an important strategy in drug development80,81. The utilization of protein degradation systems to disrupt pathogenic condensates has begun to attract attention in the pharmaceutical field82,83,84.

Previous studies have shown that the N proteins of other viruses belong to the family Tospoviridae have an identical mechanism of RNA encapsidation with TSWV N, with R94 being highly conserved in Tospoviruses and Y184 present in groundnut yellow spot virus (GYSV) and tomato chlorotic spot virus (TCSV)60,61. This suggests that Z9 may be effective in preventing and controlling viruses in the family Tospoviridae. Based on the effect of Z9 in disrupting the normal binding of N-RNA, how to design more effective drugs will arouse our interest. Given the increasing evidence that many plant viruses and phytopathogenic bacteria rely on the condensation of replication proteins for proliferation24,25,26,27,85, the discovery of biomolecular condensate inhibitors could play a significant role in controlling plant diseases that were hard to manage.

Methods

Materials and virus sources

The seeds of N. glutinosa and N. benthamiana were preserved by our laboratory. All of the plants were cultured in an artificial climate chamber (25 °C during the day, 50% relative humidity/23 °C at night, 40% relative humidity). The full-length infectious clone vectors (L(+)opt, M(-)opt, and SR(+)eGFP), the expression vector of silencing suppressors (VSRs), and vector pCambia2300-N-YFP were provided by Professor Xiaorong Tao (Nanjing Agricultural University). The TSWV lettuce isolate was maintained in N. benthamiana, and the infected leaves were maintained at − 80 °C.

Plasmid construction

To generate the mutant plasmids SR(+)eGFPMut (SR(+)eGFPR94A, SR(+)eGFPY184A, SR(+)eGFPG147A, and SR(+)eGFPR94A&Y184A), the insertion fragments and vectors were obtained with SR(+)eGFPWT as a template via reverse polymerase chain reaction. For the p2300-NR94A&Y184A-YFP construct, the cDNA of the NR94A&Y184A gene was amplified from SR(+)eGFPR94A&Y184A and then inserted into the binary vector pCambia2300-YFP. The plasmid pAN580-N-eGFP was constructed by Biorun Biosciences Co., Ltd. (Wuhan, China). All of the constructs were verified via DNA sequencing. The primers used are shown in Supplementary Table 2.

Synthesis

All commercial reagents and solvents were used without further purification. Thin layer chromatography (TLC) was performed on precoated silica gel GF254 plates. Visualization of TLC was achieved using UV light (wavelength of 254 nm). Column chromatography was performed on silica gel (300–400 mesh) using the proper eluent. NMR data were recorded on a Bruker DPX 400 MHz (Bruker, Germany). Chemical shifts were reported in parts per million (ppm) referenced to the appropriate solvent peaks (δ 7.26 ppm for CDCl3 in the proton spectra, 77.0 ppm for CDCl3 in the fully decoupled 13C spectra). The following abbreviations were used to describe the peak splitting patterns: s = singlet, d = doublet, t = triplet, and m = multiplet. The coupling constants J are reported in Hertz (Hz). High-resolution mass spectra were obtained with a Thermo Scientific Q Exactive instrument (Thermo Scientific, USA). The specific synthesis steps can be found in Supplementary Method 1 and Supplementary Figs. 1214, and the structural identification of the compounds can be found in Supplementary Note 1 and Supplementary Figs. 1592.

Anti-TSWV assay

Healthy N. glutinosa was used as the model plant, and the inhibitory activities of the compounds against TSWV were tested via the half-leaf method. For curative activity, carborundum powder was precoated on the leaves of the plant, and the extracted viral mixture was inoculated onto the leaves via a soft brush. After 30 min, the leaves were rinsed thoroughly with water, and the agent was evenly applied to the right side of each leaf. For protective activity, the prepared agent was first evenly applied to the right side of the leaves. After 12 h, the extracted viral mixture was inoculated onto the leaves. For inactivation activity, the prepared agent with an equal amount of extracted virus mixture was diluted on ice, and the mixture was inoculated onto the right leaves after 30 min. At that time, the extracted virus was diluted with the same amount of buffer solution and inoculated onto the left leaves. The plants were subsequently placed in an artificial climate chamber (25 °C during the day, 50% relative humidity/23 °C at night, 40% relative humidity) for cultivation. The number of lesions was counted after approximately 2 days. For the EC50 test of inactivation activity, the concentrations of the compounds were set to 31.25, 62.5, 125, 250, and 500 µg/mL. The method is consistent with the description provided above. The extracted TSWV was inoculated on the lower leaves of the tomatoes. The plants were cultured in an artificial climate chamber (25 °C during the day, 50% relative humidity/23 °C at night, 40% relative humidity). After 24 h, 100 µg/mL agent (2.0 mg of compound Z9, 50 µL of DMSO, 20 mL of 2% (v/v) Tween 80 water) was sprayed onto the tomato leaves, with 20 mL of 3% Tween 80 water containing DMSO (50 µL) as a control. Pathogenic symptoms were observed after ~ 20 days.

Agrobacterium infiltration

Agrobacterium tumefaciens strain GV3101 was transfected with the recombinant plasmids, and the monoclonal clone was selected and cultured at 28 °C for approximately 24–36 h. A. tumefaciens cells were resuspended in agroinfiltration buffer (10 mM MES, pH = 5.6, 10 mM MgCl2, and 100 μM acetosyringone) and incubated for 2 to 3 h in the dark at room temperature after the optical density at 600 nm (OD600) was adjusted. The cells were used for agroinfiltration of N. benthamiana leaves. For the competent minigenome replication system (L(+)opt + SR(+)eGFP + VSRs) or reconstituted infectious TSWV (L(+)opt + SR(+)eGFP + M(-)opt + VSRs), equal volumes of Agrobacterium cultures harboring the full-length infectious clone vector (final concentration OD600 = 0.20) were mixed with the bacterial mixture (final concentration OD600 = 0.05) containing the VSRs (P19-HcPro-γb)55. For the expression vector p2300-N-YFP, the cells were resuspended in an agroinfiltration buffer adjusted to an optical density (OD600) of 0.50. Leaves of N. benthamiana plants at the 5–6-leaf stage were infiltrated with Agrobacterium cultures using syringes. The infiltrated plants were grown in an artificial climate chamber. The operation of the mutants is consistent with the above description.

Pull-down assay

TSWV-systemically infected leaves of N. benthamiana were collected, and extraction buffer (25 mM Tris-HCl, pH = 7.5, 150 mM NaCl, 1 mM EDTA, 2% w/v polyvinylpolypyrrolidone, 10 mM dithiothreitol, 0.5% v/v Triton X-100, 10% v/v glycerol, and 1 × protease inhibitor cocktail) was added to obtain the total crude protein extract. 10 μL of the probe solution (100 mM) was added into 1 mL of the total extracted protein solution and incubated in the dark for 2 h. Sequentially 25 μL of 10 mM N3-biotin, 10 μL of 100 mM BTTAA, 10 μL of 100 mM CuSO4, and 5 μL of 500 mM vitamin C were added into the reaction system. After mixing, the system was reacted at 37 °C for 3 h. 0.5 mL the treated magnetic beads were added into the mixture, and incubated for 12 h. After enrichment, the centrifuge tubes were placed on a magnetic stand, and the supernatant was used as the input group. The magnetic beads were washed with 0.5 mM PBS (containing 2 mM UA) and mass spectrometry water. Proteins bound to streptavidin beads were fractionated by SDS-PAGE and analyzed by immunoblotting using an N protein antibody. For the competitive binding experiment between compound Z9 and the probe, 1.0 μL of 50 μM the probe and 20 μL of N protein (0.5 mg/mL) were incubated with different concentrations of compound Z9 (50, 500, 1000, and 1500 μM) for 2 h. Sequentially 0.5 μL of 10 mM N3-biotin, 0.2 μL of 100 mM BTTAA, 0.2 μL of 100 mM CuSO4, and 0.1 μL of 500 mM vitamin C were added into the reaction system. After mixing, the system was reacted at 37 °C for 3 h. The mixture was fractionated by SDS-PAGE and analyzed by immunoblotting using HRP-Streptavidin (1:6000, Sangon Biotech, China, Cat. No. D111054).

Isolation of protoplasts and transformation with N-eGFP

N. benthamiana protoplasts were prepared and transformed via the plant protoplast preparation and transformation kit (Real-Times, China). Briefly, 0.20 g of N. benthamiana leaves were cut into strips 0.5–1 mm in width, submerged in 10 mL of enzymatic hydrolysate [5 mL of solution I, 0.15 g of cellulose R-10, 0.04 g of pectinase R-10, 5 μL of β-mercaptoethanol, 0.2 mL of 50 mg/mL BSA] and treated at 20–25 °C for 3 h. The mixtures were then added to 10 mL of solution II and centrifuged at 100 × g at room temperature for 2 min, followed by resuspension in 1 mL of solution III. Five hundred microlitres of protoplast cells and 500 µL of transformation mixture were mixed and transfected with 75 μg of the pAN580-N-eGFP plasmid at room temperature for 15 min. After 16 h, the transformed protoplasts were resuspended in solution II and cultured in solution V at 25 °C for the expression of N-eGFP. The obtained protoplasts were detected via a confocal microscope with a × 63 oil immersion objective (Zeiss, Oberkochen, Germany).

Drug treatments

When N-YFP was significantly expressed in the leaves of N. benthamiana after 18 h, the compound Z9 (2.0 mg) was dissolved in 100 µL of DMSO, followed by the addition of 20 mL of ddH2O or 0.1% tween 80 aqueous solution to prepare a 100 µg/mL agent, which was rapidly injected into the infiltration area below the left side of the leaf. Concurrently, 0.5% DMSO (v/v) was injected into the infiltration area above the left side of the leaf. After 18 and 54 h, confocal microscopy with Z-Stack was used to capture micrographs of the left side drug-treated and right side of untreated areas within the same leaf. The drug treatment at 36 h is consistent with the description above. The treatment of PMMoV CP-YFP with the compound Z9 was consistent with the above description. To determine the impact of drug treatment on N-eGFP in protoplasts, compound Z9 was diluted to 200 µg/mL. Then, 10 µL of the drug mixture was quickly and evenly mixed with 200 µL of protoplasts, and a confocal microscope was immediately used to locate intact cells expressing N-eGFP. The effects of Z9 on N-eGFP condensates were observed within 30 min.

Fluorescence microscopy

Agro-infiltrated N. benthamiana leaves were examined for fluorescence expression via an OLYMPUS model BX53F fluorescence microscope (OLYMPUS, Tokyo, Japan) with a green barrier filter. The samples were fixed in water on a microslider under a coverslip to detect eGFP fluorescence. The fluorescence images were collected using CellSens Standard 1.18 (OLYMPUS, Tokyo, Japan).

Confocal microscopy and fluorescence recovery after photobleaching

Confocal laser scanning micrographs of Agro-infiltrated leaves were captured with a Zeiss LSM900 confocal laser scanning microscope (Zeiss, Oberkochen, Germany). The fluorescent protein was excited at 488 nm, and time-lapse images were taken over a 1- or 2-min period. For fluorescence recovery after photobleaching (FRAP) experiments, IB droplets formed in infected leaf cells were bleached with a laser at 488 nm (50% or 100% intensity) under a confocal microscope with × 20 water immersion, and the field of view was magnified × 2.5. After bleaching, at least 40 frames of recovery images were collected in time-lapse mode. The Images were collected and the fluorescence intensity of the bleaching region was measured via ZEN 3.1 (Zeiss, Oberkochen, Germany). The data were normalized, and recovery curves were generated with GraphPad Prism 8 (GraphPad, San Diego, USA).

Protein extracts and immunoblot assays

Healthy, Agro-infiltrated leaf patches or TSWV-systemically infected leaves of N. benthamiana were collected, and total protein was extracted in 1 mL of extraction buffer (25 mM Tris-HCl, pH = 7.5, 150 mM NaCl, 1 mM EDTA, 2% w/v polyvinylpolypyrrolidone, 10 mM dithiothreitol, 0.5% v/v Triton X-100, 10% v/v glycerol, and 1 × protease inhibitor cocktail)55. The samples were transferred to PVDF membranes after separation by SDS‒PAGE. The membranes were blocked with 5% w/v skim milk solution and probed with a rabbit polyclonal anti-N antibody (1:5000, Donated by Professor Xiaorong Tao from Nanjing Agricultural University) or a rabbit polyclonal anti-PMMoV CP antibody (1:5000, Donated by Professor Fei Yan from Ningbo University). RuBisCO stained with Ponceau S was used as a sample loading control. Finally, the signal was visualized with a ChemiDoc MP Imaging System (Bio-Rad, Hercules, USA). Quantitative analysis was performed using ImageJ v1.52.

Molecular docking

The binding sites of compound Z9 with the protein in the TSWV N crystal structure (PDB: 5IP1) were predicted using Schrodinger maestro 13.5, LeDock, and AutoDock Vina 1.1.2 for the monomer. PyMOL was used as the plotting software.

Protein purification and microscale thermophoresis (MST) assay

The prokaryotic expression vectors (pET-32a-NWT, pET-32a-NR94A, pET-32a-NY184A, pET-32a-NR94A&Y184A, pET-21a-sfGFP-NWT, and pET-21a-sfGFP-NR94A&Y184A) were successfully introduced into Escherichia coli BL21 (DE3). The expression strains were induced by isopropyl-β-D-thiogalactopyranoside at 28 °C overnight, after which the precipitates were lysed (power 35%, 40 min, 4 s ultrasound, and 6 s interval) with buffer (0.869 g NaCl, 3.75 mL PBS, 36 μL mercaptoethanol, 5 mL glycerol, 41.25 mL H2O, pH = 7.4). The lysates were clarified by centrifugation at 16,000 × g for 30 min. The proteins were purified via affinity chromatography using a Trap high-performance column (GE Healthcare, USA). The binding affinities of compound Z9 to the proteins were determined with a Monolith NT.115 instrument (NanoTemper, Munich, Germany). The Mo. Affinity Analysis 2.2.4 software (NanoTemper, Munich, Germany) was used to fit the MST curve and obtain the affinity constants (KD). The specific operation was performed according to the reported literature29.

In vitro RNA-induced N protein aggregation

Total RNA from healthy or TSWV-infected N. benthamiana leaves was extracted with TRIzol and quantified at a concentration of 1000 ng/μL. The purified 6×His-tagged sfGFP-NWT or sfGFP-NR94A&Y184A was concentrated to 10 μM in desalting buffer (20 mM Tris-HCl, pH 7.5, and 150 mM NaCl). Twenty microlitres of protein and 2 μL or 1 μL of RNA were incubated for 10 min, and the mixture was observed with a confocal microscope.

Electrophoretic mobility shift assay (EMSA)

The digoxigenin-labeled RNA probe was prepared following the previously described protocol31. Briefly, a DNA template with the T7 promoter sequence was generated via polymerase chain reaction and purified using a gel recovery kit. The RNA was then transcribed in vitro using T7 RNA polymerase (Promega, Madison, USA, Cat. No. P2075). One microlitre of RNA probe at a concentration of approximately 1200 ng/µL was placed in a PCR tube and diluted to 40 ng/µL with DEPC water. One microlitre of RNA probe (40 ng/µL) was added to the PCR tubes and mixed with 9 µL of different qualities of the purified NWT or NR94A&Y184A (0, 0.025, 0.05, 0.1, 0.1875, 0.375, 0.75, 1.5, or 3.0 µg). The mixtures were incubated in an ice water bath for 10 min. The samples were transferred onto a nylon membrane. The membrane was exposed to UV light, probed with an anti-digoxin alkaline phosphatase antibody (1:5000, Roche, Basel, Switzerland, Cat. No. 11585614910) for 1–2 h, and then incubated with BCIP-NBT for staining at 37 °C. The signal was directly visualized and photographed.

Biolayer interferometry (BLI) assays

The purified His-tagged N protein was loaded onto Ni-NTA biosensors, and the binding signal values of different compounds (50 µg/mL) were measured via a biolayer interferometry method with Octet RED 96 (ForteBio, San Francisco, USA). All of the compounds were diluted in PBST. The affinity constants (KD) between compound Z9 and the recombinant wild-type and mutant proteins were measured using Octet BLI Discovery 12.2 (ForteBio, San Francisco, USA). The binding constants of compound Z9 with proteins were determined via global fit analyses of experimental binding curves using the Octet Analysis Studio 12.2 (ForteBio, San Francisco, USA). To analyze the effects of binding sites on N-RNA interactions, 100 nM of biotinylated RNA (bio-GAGAGCAAUUGUGUCAGAA) was loaded onto streptavidin (SA) biosensors and balanced with PBS. The biosensors were then exposed to diluted recombinant NWT or NR94A&Y184A, followed by washing (dissociation) with PBS. The affinity constant (KD) values were calculated via Octet Red software. For the binding inhibition experiments, the affinity constant (KD) value of compound Z9 with RNA was first measured via BLI, and compound Z9 (5 µM) was then combined with the N protein (12.5 nM) and incubated for 30 min at room temperature. The biosensors were exposed to recombinant N protein (12.5 nM), incubated with a mixture of compound Z9 and N protein, and then with compound Z9 (5 µM), followed by washing (dissociation) with PBST. Association rate constants (Kon) and dissociation rate constants (Koff) were measured using Octet BLI Discovery 12.2 (ForteBio, San Francisco, USA).

Protein stability

Systematically infected leaves (1.0 g) from plants infected with TSWVWT, TSWVR94A, TSWVY184A, or TSWVR94A& Y184A were collected. A total of 1.0 mL of 5 mM PBS buffer (pH=8.0) was added, and the mixture was ground finely. Then, 100 μL of each mixture was added, and the mixture was distributed into 4 separate 1.5 mL centrifuge tubes. The samples were incubated at 25 °C for 0, 30, 100, or 160 min and then centrifuged at 13,198 × g for 10 min. Eighty microlitres of the supernatant from each sample was mixed with 20 μL of 5 × loading buffer and then denatured at 100 °C for 10 min. The accumulation level of N protein and actin (1:2000-5000, ABclonal, Wuhan, China, Cat. No. AC009) was detected via an immunoblotting assay. To test the ability of compound Z9 to disrupt N-RNA binding and destabilize the N protein, 1.0 g of healthy leaf tissue was collected, and total protein was extracted with 1.0 mL of PBS. UltraNuclease (25 U/mL) was added to 0.5 mL of the extracted solution. Compound Z9 (5 µM) was incubated with purified NWT (12.5 nM) and the above 100 nM biotin-labeled RNA at room temperature for 30 min and then added to 0.5 mL of the extracted mixture containing UltraNuclease. The same mixture (5 µM Z9, 12.5 nM NWT, and 100 nM RNA) was added to 0.5 mL of extracted solution without Omni-Nuclease as a control. After they were incubated at 25 °C for 0, 30, 100, and 160 min, they were denatured at 100 °C for 10 min. The accumulation level of N protein was detected by immunoblotting. The treatment without the addition of compound Z9 followed the same procedure as described above.

Data analysis

The data in the article was processed using Microsoft Office Excel 2019 or GraphPad Prism 8.0 software (GraphPad Software, San Diego, CA, USA). Statistically significant differences between groups were assessed using GraphPad Prism 8.0 software (GraphPad Software, San Diego, CA, USA). P-values were calculated using the two-tailed t-test or one-tailed paired t test from triplicated independent experiments. Data are presented as mean ± standard deviation. Four levels of significance (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001) were used for all tests.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.