English

• Since the first identification of helicase in Escherichia coli (E. coli) in 1976, a large number of different kinds of helicases have been isolated from animals, plants, and viruses, and have been intensively explored for three-dimensional structural characterization and physiological function [1-4]. Helicase is a micromotor protein widely present in living organisms that enables the destruction of double-stranded DNA/RNA by utilizing ATP-hydrolyzed energy to strip one strand from the corresponding complementary strand [4,5]. DNA helicases play an important role in replication, recombination, repair, transcription, and the maintenance of chromosomal integrity [6,7]. The majority of helicases encoded by single-stranded viruses are classified within the RNA helicase family. RNA helicases typically unfold local secondary RNA structures and act as RNA chaperones in the nucleus to facilitate gene transcription, ribosome assembly, RNA splicing, and translation [8-10]. Besides, several representative refractory viruses, such as coronavirus disease 2019 (COVID-19), dengue virus (DENV), hepatitis C virus (HCV), Japanese encephalitis virus (JEV), tobacco mosaic virus (TMV), their helicases also have a closely-connected relationship with viral virulence, symptom degree of the host, and induced resistance [11-16]. These representative biological features and functions drive helicases as an intriguing target in pathology fields and innovative drug or pesticide exploitation.

  Drug repurposing, alternatively termed new uses for old drugs or drug repositioning, is a highly effective and convenient methodology for discovering new availability for approved or investigational drugs that go beyond the boundaries of intended indications [17,18]. This strategy offers plenty of advantages over the burdensome traditional drug discovery. Firstly, pharmacokinetic, pharmacodynamic, and toxicity properties of drugs have been determined in previous studies. Second, drug development cycles can be shortened by using established authoritative models for evaluation and sufficient experimental support. Third, this strategy facilitates lower costs and cycles for drug development and reduces out-of-pocket costs for the public, resulting in significant price competitiveness in the market. Finally, repurposed medications provide novel prospective targets that deserve to be investigated [19-22]. Some successful repurposed drugs, such as allopurinol or gemcitabine, were originally used as antivirals, but were later evaluated for Gout or cancer indications [23]. For potentially functional proteins or enzymes, target-based drug repurposing strategies perform receptor-based molecular virtual screening from drug libraries to predict the complementarity of binding sites and possible action patterns between receptor-ligands. Moreover, this strategy can screen for almost any known drug structure and link targets to disease mechanisms, thus accelerating the likelihood of drug discovery [24-28]. Research progress on helicase as a drug target has been severely slower than other viral pivotal proteins or enzymes, and no helicase inhibitor has yet reached the stage of batch application [29,30]. Dishearteningly, reports of helicase in phytopathogenic viruses are even more scarce. In our previous studies, some alkaloids and ferulic acid structures have been identified as having potential anti-TMV profiles or TMV helicase inhibitory activity, but research on these molecules is still in the preliminary exploration stage [31,32].

  Exploitation of receptor-ligand interactions is a crucial procedure in pharmaceutical development. Computational simulations hold promise as a high-efficiency candidate to insipid and costly wet-lab experiments for accelerating the exploration of different alternatives [26-28]. For the sake of efficiently and quickly discovering underlying helicase inhibitors and their effects on virus replication, in this work, we take the TMV helicase target as an example, two strategies (SWISS modeling and AlphaFold2 modeling) were used to confirm the reliability of target TMV helicase modeling (Figs. S1–S4, Tables S1 and S2 in Supporting information). Subsequently, six potential helicase inhibitors with the highest affinity were screened from the small drug molecule library approved by Food and Drug Administration (FDA) (https://zinc15.docking.org/). According to the root mean square deviations (RMSD), root mean square fluctuations (RMSF), molecular mechanics generalized Born surface area (MM-PBSA) values, and density functional theory (DFT) calculations of these complexes, eltrombopag, lumacaftor, and raltegravir were successfully screened as promising helicase inhibitors. Furthermore, to further evaluate the potential targeting of the screened compounds and the predictability of the screening platform, these hits have been profoundly validated by binding force measurements, enzyme inhibition activity evaluations, and in vivo bioactivity determinations.

  Repurposing of previously exploited drugs to address frequently stubborn diseases caused by causative agents is progressively emerging as a beneficial inspiration, which has the advantages of lower development costs and timelines and more available experimental data on the physicochemical properties of "old" drugs [18,21]. To find out candidate inhibitors targeting TMV helicase with excellent antiviral profiles, the approved drugs (in major jurisdictions, including the FDA approved) were used as a small molecule ligand library. The vast space and structural diversity of the chemical dataset are conducive to the discovery of novel drug structures. To examine the chemical skeleton diversity of the ligand molecule library, we evaluated the Tanimoto similarity of these molecules here using two molecular fingerprints (Chem.RDKFingerprint and MACCS). Figs. 1A and B show that the fingerprint similarity coefficient of most compounds in the ligand library is below 0.4, indicating obvious structural differences in the dataset. Meanwhile, Fig. 1C shows that the AlogP values of compounds with different molecular weights are mainly between −4 and 8, which means their chemical spatial distribution is relatively wide. Subsequently, through t-SNE dimensionality reduction, K-means clustering, and Calinski-Harabasz scoring, the ligand molecule library was reasonably divided into 14 clusters (from cluster 0–13), and the central molecule of each cluster is displayed in the corresponding color region in Figs. 1D–F. Autodock Vina (1.2.1) software was used to conduct virtual screening and molecular docking between dynamically optimized helicases and molecules of different clusters. Ningnanmycin, a broad-spectrum class of antiviral agents considered to have a potential binding effect on TMV helicase, was used as a control [33]. For the docking screen, a total of 6405 results (containing different conformational ligands) were generated. Compared with that of ningnanmycin (docking score was −5.8 kcal/mol), six compounds with the highest affinity were selected (Table S3 in Supporting information), namely ZINC11679756 (eltrombopag, −10.4 kcal/mol, originally used to treat immune thrombocytopenia), ZINC64033452 (lumacaftor, −10.1 kcal/mol, the second candidate compound for oral study in cystic fibrosis treatment), ZINC13831130 (raltegravir, −9.7 kcal/mol, the first-in-class integrase inhibitor among human immunodeficiency virus (HIV) antiviral drugs), ZINC19632618 (imatinib, −9.6 kcal/mol, was used clinically for the treatment of chronic myeloid leukemia and malignant gastrointestinal mesenchymal tumors), ZINC6716957 (nilotinib, −9.6 kcal/mol, an orally available Bcr-Abl tyrosine kinase inhibitor with antitumor activity), and ZINC8215434 (fad, −9.5 kcal/mol, as a redox cofactor is involved in enzymatic reactions) as preliminary preferred hits (Figs. S5 and S6 in Supporting information). Surprisingly, all of these preferred molecules are from cluster 9 except fad (cluster 1).

  Figure 1

  

  Figure 1.  The frequency distribution of Molecular fingerprint similarity. (A) Tanimoto coefficient and (B) MACCS fingerprint heat map analysis. (C) Distribution of chemical spatial diversity. (D) Hierarchical clustering of molecules in the ligand library was performed based on the Ward method. (E) Division and demonstration of compound clusters. Each color represents a kind of molecular cluster, and the center of the scatter plot corresponds to the central molecule of that cluster. (F) The proportional quantity of each cluster.

  DownLoad: Full-Size Img PowerPoint

  For further evaluating the potential helicase binding ability and the stability of these hits in the pocket, molecular dynamics simulations (50 ns) of these optimal binding conformations resulting from docking were performed. The helicase-imatinib group for these complexes presented the biggest protein RMSD, whereas the helicase-fad group has the largest ligand RMSD (Fig. S7 in Supporting information). In the RMSF comparison, amino acid residues between 110 and 140 of all complexes produced the most pronounced fluctuations (Fig. S8 in Supporting information). From the visualized VMD, we found that this sequence happened to be the loop region of helicase, where helicase-eltrombopag and helicase-lumacaftor exhibited lower peak intensities. Furthermore, the binding free energy calculation (MM-PBSA approach) was employed to identify the binding ability of each receptor-ligand. The results were shown in this order: helicase-lumacaftor (ΔG_bind = −29.0 kcal/mol) > helicase-raltegravir (ΔG_bind = −22.1 kcal/mol) > helicase-imatinib (ΔG_bind = −17.2 kcal/mol) > helicase-eltrombopag (ΔG_bind = −15.4 kcal/mol) > helicase-nilotinib (ΔG_bind = −12.6 kcal/mol) > helicase-fad (ΔG_bind = −8.4 kcal/mol). Combined with these results, eltrombopag, lumacaftor, imatinib, and raltegravir were again preferentially selected for follow-up studies.

  The highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) energy gap is valuable for assessing the interaction between ligand and receptor. Usually, a higher total energy or lower HOMO-LUMO gap means that the ligand may have potential biological activity for the target [34-36]. The molecular total energy (MTE) and LUMO-HOMO energy gap of eltrombopag were −1485.859 and 0.12 hartree, and its electron transfer was from HOMO of pyrazolone, benzene ring and hydroxyphenylhydrazide to HUMO of pyrazolone and hydroxyphenylhydrazide, exhibiting the highest total energy (Fig. S9 and Table S4 in Supporting information). Intriguingly, lumacaftor displayed the adverse DFT calculation values, affording the lowest total energy (−1610.717 hartree) and highest LUMO-HOMO energy gap (0.17 hartree), whose electron was transferred from the HOMO of the pyridine ring, benzene ring, and amide group to the LUMO of the pyridine ring and benzoic acid fragment. Besides, raltegravir (−1579.257 and 0.16 hartree) and imatinib (−1582.290 and 0.14 hartree) exhibited similar MTE and LUMO-HOMO energy gap. raltegravir has also been developed as an inhibitor of the virus target protein, with applications similar to the purpose of this screening. With these results in mind, lumacaftor, eltrombopag, and raltegravir were further evaluated to excavate hits and compare DFT calculations with bioactivity.

  The number and density of hydrogen bonds generated between the ligand-receptor have a meaningful impact on the stability of the binding pocket as well as the RMSD [36]. As seen in Fig. S10 (Supporting information), the TMV helicase and ligand were in a state of motion in the simulation, and the number of H-bonds formed varies slightly at different times. In helicase-raltegravir, the number of H-bonds in the complex was mostly scattered in the 0–1 range, with the lowest overall hydrogen bond density. In helicase-lumacaftor, complex H-bonds were mainly concentrated in the 0–2 range, and their density distribution in the 0–1 range was the densest among the three groups. In contrast, the number and density of hydrogen bonds in helicase-eltrombopag were mainly concentrated between ranges 2–4, indicating that eltrombopag and helicase produced the most intensive non-covalent interactions, significantly more than the other complexes. Among these complexes, the average radius of gyration (Rg) values for helicase-eltrombopag, helicase-raltegravir, and helicase-lumacaftor were 2.07 ± 0.02, 2.07 ± 0.03, 2.11 ± 0.06 nm, signifying an approximate protein flexibility and foldability, which was consistent with the measurement results of their solvent accessible surface area (SASA) values (149.7 ± 2.4, 150.5 ± 2.8, and 151.9 ± 2.9 nm², respectively). Additionally, the bound state of complexes and PCA were analyzed using Discovery Studio and the Bio3D package in R (Figs. S11–S13 in Supporting information).

  Considering that there is a strong interaction between ligands and helicase, and this binding force may drive the conformation of helicase to change, we randomly simulated the secondary structure of helicase with or without ligand hits. As shown in Figs. S14A and B (Supporting information), the α-helix, β-folding, β-rotation, random curl and bending of helicase in the complex were obviously compared with pure helicase, indicating the binding of eltrombopag may drive helicase secondary structure changes. To identify whether these hits have effects on the conformations of the TMV helicase, a circular dichroism (CD) spectroscopy was used to ongoing analyze. Figs. S14C–F (Supporting information) showed that helicase produced the typical positive and negative Cotton at 195, 200, 205, and 209 nm, respectively, which considered to be related to α-helix, β-sheet, and random curl in the secondary structure of protein [37]. After treated with ningnanmycin, helicase characteristic absorption peak position was shifted to the right by 1–2 units, such as β-fold (195–197 nm) or α-helix (209–210 nm), and its absorption intensity was obviously reduced, especially the negative Cotton at 205 nm. The CD spectra of helicase showed significant changes after treatment by lumacaftor, such as the disappearance of negative absorption peaks at 205 and 209 nm, and the creation of a strong negative absorption peak at 216 nm. In addition, its positive absorption peak was replaced by the opposite negative absorption peak with greater intensity at 195 nm, while the intensity of the random curl absorption peak decreased (200 nm). These phenomena indicate that lumacaftor can induce helicase conformational changes, including α-helix, β-folding, β-rotation, and random curling. Comparatively, helicase treated with raltegravir showed similar changes, but the trend was not as pronounced as in the lumacaftor group. On the other hand, the displacement and intensity of characteristic absorption peaks of helicase-eltrombopag were almost unchanged, except for the random curl peak at 200 nm.

  Microscale thermophoresis (MST) is widely used to evaluate intermolecular interactions based on the displacement changes of molecules in a microscopic temperature gradient field [38,39]. To assess the affinity between the screened molecules and target enzymes, lumacaftor, raltegravir, eltrombopag, and ningnanmycin (positive control) were selected for MST assays with TMV helicase, respectively. Among these compounds, lumacaftor showed the most brilliant binding intensity, surpassing that of ningnanmycin (Figs. 2A–D). However, eltrombopag demonstrated an adverse binding capability compared to silicon simulation. Their corresponding K_d values were ranked as follows: lumacaftor (K_d = 0.22 µmol/L) > raltegravir (K_d = 6.53 µmol/L) > ningnanmycin (K_d = 9.35 µmol/L) > eltrombopag (K_d = 19.17 µmol/L), which roughly supports the outcomes predicted by simulation analysis. To further illuminate the binding affinity of the screened hits with TMV helicase, fluorescence titration determination was conducted and analyzed (by the Stern-Volmer method) in this bioassay [40]. Fig. S15 (Supporting information) showed the strongest fluorescence peak at 384 nm, which is the characteristic fluorescence peak of TMV helicase in the absence of ligand. As the concentration of the compounds in the solution increased, TMV helicase fluorescence intensity showed a regular decrease in each group. The corresponding binding constants K_a and binding sites were obtained by linear regression analysis (Table S5 in Supporting information) and ranked as: raltegravir (K_a = 10^7.804) > lumacaftor (K_a = 10^7.102) > ningnanmycin (K_a = 10^6.401) > eltrombopag (K_a = 10^5.195), which resembled the MST result, proving that hits raltegravir and lumacaftor possessed a superior binding profile with helicase compared to that of ningnanmycin. Furthermore, eltrombopag showed a lower binding ratio (about 1:1) to helicase, suggesting that eltrombopag may more readily enter the active center and bind key residue sites of helicase.

  Figure 2

  

  Figure 2.  (A–D) MST measurement of hit molecules with TMV helicase. (E, H) Representative illustration of the interaction mechanism between lumacaftor and helicase. Different concentrations (50–250 µmol/L) of ningnanmycin (F), raltegravir (G), eltrombopag (I), and lumacaftor (J) anti-TMV helicase potency in vitro. Error bars indicate the mean ± standard deviation (SD) (n = 3).

  DownLoad: Full-Size Img PowerPoint

  In the actual situation, the hit molecule with a strong binding force to the target protein may have two opposite results: either causing appropriate changes in the steric conformation of the target protein, which in turn affects protein physiological activity, or making the target protein more tightly bound to the hit and difficult to separate, but not affecting functional protein bioactivity [41]. To identify whether these hits have TMV helicase inhibitory activity, relevant helicase ATPase activity inhibitory assays were performed. The bioassay results are shown in Figs. 2E–H. At 200 µmol/L, lumacaftor showed the most outstanding helicase ATPase inhibition activity, significantly exceeding the control ningnanmycin by 1.4-fold and the control Ribavirin by 2.7-fold, with the corresponding activities listed in this order: lumacaftor (61.2%) > raltegravir (53.7%) > ningnanmycin (42.7%) > Ribavirin (22.9%) > eltrombopag (19.1%). Unexpectedly, eltrombopag showed outstanding performance in virtual screening, dynamic simulations, and enzyme binding assays. However, its helicase ATPase inhibition activity was rather unsatisfactory, especially at 200 µmol/L (19.1%). To accurately assess these hits anti-helicase ATPase profiles, the inhibitory activities of lumacaftor, Ribavirin, eltrombopag, ningnanmycin, and raltegravir were tested at different concentrations (50, 100, 150, 200, and 250 µmol/L), respectively. Furthermore, the half maximal inhibitory concentration (IC₅₀) values of lumacaftor (162.5 µmol/L) and raltegravir (185.5 µmol/L) were obtained by calculating their bioactivities (Table S6 in Supporting information). Combined with the results of MST and fluorescence analysis, this reaffirms lumacaftor and raltegravir have a strong binding character to TMV helicase, blunting its ability to hydrolyze ATP.

  Besides, green fluorescent protein (GFP)-labeled TMV was used to visualize the therapeutic efficacy of screened drugs for TMV. For GFP-labeled TMV group, GFP-labeled TMV containing different drugs group, and DMSO control, those were cultured in a greenhouse for 5–10 days respectively. It could be found that TMV-GFP was distributed almost uniformly over a large area of the leaf in the absence of the drug group after 5 days (Fig. 3A). In the drug dealt group, the fluorescence distribution of the leaves was mainly concentrated in the veins, but did not show a diffuse distribution. Contrastively, the fluorescence intensity and area in the lumacaftor group was smaller than those in the ningnanmycin group (positive group), which indicated that lumacaftor can effectively inhibit the replication and diffusion of TMV in tobacco in some time. However, after 10 days, the fluorescence distribution of the ningnanmycin group, the lumacaftor group and the raltegravir group spread from vein to the entire leaf, especially the eltrombopag group. To observe the relative expression levels of helicase genes (Table S7 in Supporting information) in different groups of drugs, leaves treated with TMV for 5 days were validated by quantitative polymerase chain reaction (qPCR) with different concentrations of small molecules. The results showed that the levels of helicase gene expression in the ningnanmycin, lumacaftor and raltegravir groups decreased significantly compared with the control group, to 60%, 58% and 74% at 250 µg/mL, and 47%, 46% and 62% at 500 µg/mL, respectively (Fig. 3B). Among them, the relative expression capacity of the helicase gene in tobacco was more significantly down-regulated in the lumacaftor group, indicating a dose-dependent inhibition relationship.

  Figure 3

  

  Figure 3.  (A) Visualization and distribution of the TMV-GFP virus after 5–10 days of different drugs treatment. (B) The related transcript levels of viral helicase gene treated with ningnanmycin, ribavirin, raltegravir, eltrombopag, and lumacaftor at 250 or 500 µg/mL, respectively. Error bars indicate the mean ± SD (n = 3). ns, no significance. P < 0.05, ***P < 0.001, ****P < 0.0001.

  DownLoad: Full-Size Img PowerPoint

  New drugs designed and developed for target proteins or repurposed old drugs may have potential bioactivity against target proteins in vitro, but their activity in vivo is unknown. To validate whether these hits can block TMV replication and propagation on the host by inhibiting the physiological function of TMV helicase, we determined the anti-TMV activities of lumacaftor, eltrombopag, and raltegravir at different concentrations (100, 200, 300, 400 and 500 µg/mL) in Nicotiana glutinosa L. As shown in Fig. S16 (Supporting information), the left and right sides of the leaves showed similar numbers of viral spots in the absence of drug. Encouragingly, after pre-treatment with either lumacaftor or raltegravir, the TMV viral spots were significantly smaller than those on the left side, achieving similar therapeutic effects to those achieved with commercially available ningnanmycin. The corresponding curative effect ranking order was lumacaftor (61.7%) > ningnanmycin (55.6%) > raltegravir (48.1%) > ribavirin (39.9%) > eltrombopag (27.2%) at 500 µg/mL (Fig. S18A in Supporting information). Among them, lumacaftor exhibited the most obvious concentration-dependent anti-TMV therapeutic activity (16.6%–58.7%) in the concentration range of 100–500 µg/mL, which significantly overpassed those of ningnanmycin (24.2%–55.6%) and Ribavirin (9.8%–39.9%), affording the concentration for 50% of maximal effect (EC₅₀) value of 386.3 µg/mL (Fig. S18D and Table S8 in Supporting information). Taken together, these results demonstrate that lumacaftor possesses optimal anti-TMV curative performance along with outstanding helicase inhibitory capability.

  In recent decades, the FDA has approved a majority of high-performance and action-specific direct acting antivirals, prompting pharmaceutical chemists to focus on targeted viral inhibitors. The exploration of receptor-ligand interactions using computer calculations and biological module analysis is a vital phase in drug discovery. In this work, six top-ranked hits were obtained through homologous modeling and virtual screening based on the FDA-approved commercially available molecule library. Through further assessment by dynamic simulation, free energy calculation, Gaussian computation, binding force determination, enzyme inhibitory evaluation and in vivo bioassay, two optimal hits lumacaftor (K_d = 0.22 µmol/L, IC₅₀ = 162.5 µmol/L) and raltegravir (K_d = 6.53 µmol/L, IC₅₀ = 185.5 µmol/L), were recognized to potentially inhibit TMV replication and assembly by interfering with TMV helicase hydrolyzes ATP. Especially lumacaftor, affording a considerable curative effect (EC₅₀ = 386.3 µg/mL) and is equivalent to that of control ningnanmycin (EC₅₀ = 415.1 µg/mL). From the trajectory extraction analysis of 500 ns, the non-covalent binding force of the lumacaftor-helicase was found to change from hydrogen bond and electrostatic interaction to hydrophobic interaction with time (Figs. S17, S18E–J and Table S8 in Supporting information). In addition, literature has reported that lumacaftor can be repurposed as an inhibitor against SARS-CoV-2 replication by targeting its highly conserved superfamily 1 helicase [16]. Current integrated outcomes highlight this multi-computational screening strategy combined with bioactivity identification is reasonable and fruitful, and has the potential to be engaged in rapid mining of other target inhibitors.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  CRediT authorship contribution statement

  Zhenxing Li: Writing – original draft, Methodology, Conceptualization. Yue Ding: Software, Investigation. Xinxin Tuo: Software, Investigation. Jinhong Hu: Formal analysis. Taihong Zhang: Supervision, Conceptualization. Xiang Zhou: Writing – review & editing, Validation, Project administration, Funding acquisition. Liwei Liu: Validation. Song Yang: Validation, Resources, Project administration, Funding acquisition.

  Acknowledgments

  This research was financially supported by National Natural Science Foundation of China (Nos. 32372610, U23A20201, 32160661, 32202359), National Key Research and Development Program of China (No. 2022YFD1700300), and the Central Government Guides Local Science and Technology Development Fund Projects (Nos. [Qiankehezhongyindi (2023) 001] and [Qiankehezhongyindi [2024]007]).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110737.

  
  
• 1. [1]

     G.X. Xu, Curr. Med. Chem. 14 (2007) 883–915. doi: 10.1111/j.1365-2478.2007.00634.x

  2. [2]

     D.W. Li, R. Zhou, W. Lilyestrom, et al., Nature 423 (2003) 512–518. doi: 10.1038/nature01691

  3. [3]

     J.M. Ontoria, S.D. Marco, I. Conte, et al., J. Med. Chem. 47 (2004) 6443–6446. doi: 10.1021/jm049435d

  4. [4]

     M. Högbom, R. Collins, S.V.D. Berg, et al., J. Mol. Biol. 372 (2007) 150–159. doi: 10.1016/j.jmb.2007.06.050

  5. [5]

     Y.S. Keum, Y.J. Jeong, Biochem. Pharmacol. 84 (2012) 1351–1358. doi: 10.1016/j.bcp.2012.08.012

  6. [6]

     A. Datta, R.M. Brosh, Front. Mol. Biosci. 5 (2018) 59. doi: 10.3389/fmolb.2018.00059

  7. [7]

     A.A. Riccio, J. Bouvette, M.J. Longley, et al., Methods 205 (2022) 263–270. doi: 10.1016/j.ymeth.2022.06.012

  8. [8]

     R. Weber, M. McCullagh, J. Phys. Chem. B 125 (2021) 8787–8796. doi: 10.1021/acs.jpcb.1c04528

  9. [9]

     D. Bonafoux, S. Nanthakumar, U.K. Bandarage, et al., J. Med. Chem. 59 (2016) 7138–7151. doi: 10.1021/acs.jmedchem.6b00486

  10. [10]

      S. Habtemariam, S.F. Nabavi, M. Banach, et al., Arch. Med. Res. 51 (2020) 733–735. doi: 10.1016/j.arcmed.2020.05.024

  11. [11]

      J. Chen, B. Malone, E. Llewellyn, et al., Cell 182 (2020) 1560–1573. doi: 10.1016/j.cell.2020.07.033

  12. [12]

      W. Xiao, K. Zvi, J.N. Culver, Virology 402 (2010) 292–302. doi: 10.1016/j.virol.2010.03.019

  13. [13]

      N.L. Sweeney, A.M. Hanson, S. Mukherjee, et al., ACS Infect. Dis. 1 (2015) 140–148. doi: 10.1021/id5000458

  14. [14]

      Z. Li, J.M. Xu, Y.K. Lang, et al., Acta Pharm. Sin. B 12 (2022) 1662–1670. doi: 10.1016/j.apsb.2021.10.017

  15. [15]

      Y.L. Shi, X.B. Zhang, K.J. Mu, et al., Acta Pharm. Sin. B 10 (2022) 1239–1248.

  16. [16]

      Y. Masoudi-Sobhanzadeh, A. Salemi, M.M. Pourseif, et al., Brief. Bioinform. 22 (2021) bbab113. doi: 10.1093/bib/bbab113

  17. [17]

      L. Shen, H.M. Liu, M. Jin, D.R. Houston, Chin. Chem. Lett. 35 (2024) 109572. doi: 10.1016/j.cclet.2024.109572

  18. [18]

      Y.X. Xu, C. Zhang, K. Jiang, et al., Acta Pharm. Sin. B 13 (2023) 1671–1685. doi: 10.1016/j.apsb.2022.08.023

  19. [19]

      L.L. Fu, W.K. Jin, J.H. Zhang, et al., Acta Pharm. Sin. B 12 (2022) 532–557. doi: 10.1016/j.apsb.2021.09.006

  20. [20]

      V. Parvathaneni, N.S. Kulkarni, A. Muth, V. Gupta, Drug Discov. Today 24 (2019) 2076–2085. doi: 10.1016/j.drudis.2019.06.014

  21. [21]

      S. Pushpakom, F. Iorio, P.A. Eyers, et al., Nat. Rev. Drug Discov. 18 (2019) 41–58. doi: 10.1038/nrd.2018.168

  22. [22]

      Z. Zhang, L. Zhou, N. Xie, et al., Signal Transduct. Tar. 5 (2020) 113.

  23. [23]

      R.M. Plenge, E.M. Scolnick, D. Altshuler, Nat. Rev. Drug Discov. 12 (2013) 581–594. doi: 10.1038/nrd4051

  24. [24]

      C.B. Mishra, P. Pandey, R.D. Sharma, et al., Brief. Bioinform. 22 (2021) 1346–1360. doi: 10.1093/bib/bbaa378

  25. [25]

      J.Q. Yang, X.X. Hu, Y.Y. Zhang, et al., Chin. Chem. Lett. 36 (2025) 110128. doi: 10.1016/j.cclet.2024.110128

  26. [26]

      X.Y. Shao, S.J. Xua, X.F. Wan, et al., Chin. Chem. Lett. 34 (2023) 108349. doi: 10.1016/j.cclet.2023.108349

  27. [27]

      C.R. Wu, Y. Liu, Y.Y. Yang, et al., Acta Pharm. Sin. B 10 (2020) 766–788. doi: 10.1016/j.apsb.2020.02.008

  28. [28]

      J.Z. Zhu, C. Chen, J. Dong, et al., Chin. Chem. Lett. 34 (2023) 107514. doi: 10.1016/j.cclet.2022.05.028

  29. [29]

      K. Li, K.J. Frankowski, C.A. Belon, et al., J. Med. Chem. 55 (2012) 3319-3130. doi: 10.1021/jm300021v

  30. [30]

      M. Bassetto, P. Leyssen, J. Neyts, et al., Bioorg. Med. Chem. Lett. 27 (2017) 936–940. doi: 10.1016/j.bmcl.2016.12.087

  31. [31]

      Z.X. Li, H.S. Fang, W.B. Shao, et al., Phosph. Sulf. 192 (2017) 1279–1285. doi: 10.1080/10426507.2017.1321649

  32. [32]

      Z.X. Li, B.X. Yang, H.W. Liu, et al., Int. J. Mol. Sci. 23 (2022) 13991. doi: 10.3390/ijms232213991

  33. [33]

      D.M. Wang, M.X. Huang, D. Gao, et al., Pestic. Biochem. Physiol. 166 (2020) 104449. doi: 10.1016/j.pestbp.2019.07.017

  34. [34]

      X.M. Zhang, H. Xu, H.F. Su, et al., J. Agric. Food Chem. 70 (2022)1776–1787. doi: 10.1021/acs.jafc.1c06595

  35. [35]

      K. Gao, Y.K. Qin, L.S. Wang, et al., J. Agric. Food Chem. 70 (2022) 9520–9535. doi: 10.1021/acs.jafc.2c02507

  36. [36]

      X.H. Liu, W. Yu, L.J. Min, et al., J. Agric. Food Chem. 68 (2020) 7324–7332. doi: 10.1021/acs.jafc.0c01042

  37. [37]

      P. Pu, X. Zheng, L. Jiao, et al., Food Chem. 339 (2021) 128106. doi: 10.1016/j.foodchem.2020.128106

  38. [38]

      F. Samsudin, P. Raghuvamsi, G. Petruk, et al., J. Mol. Cell Biol. 14 (2023) mjac058. doi: 10.1093/jmcb/mjac058

  39. [39]

      C.J. Wienken, P. Baaske, U. Rothbauer, et al., Nat. Commun. 1 (2010) 100. doi: 10.1038/ncomms1093

  40. [40]

      R.Q. Zhai, G. Chen, G.Y. Liu, et al., J. Adv. Res. 37 (2022) 61–74. doi: 10.1016/j.jare.2021.08.008

  41. [41]

      Y. Yi, J.H. Li, X.Y. Lai, et al., J. Adv. Res. 36 (2022) 201–210. doi: 10.1016/j.jare.2021.11.012

• 

  Figure 1  The frequency distribution of Molecular fingerprint similarity. (A) Tanimoto coefficient and (B) MACCS fingerprint heat map analysis. (C) Distribution of chemical spatial diversity. (D) Hierarchical clustering of molecules in the ligand library was performed based on the Ward method. (E) Division and demonstration of compound clusters. Each color represents a kind of molecular cluster, and the center of the scatter plot corresponds to the central molecule of that cluster. (F) The proportional quantity of each cluster.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (A–D) MST measurement of hit molecules with TMV helicase. (E, H) Representative illustration of the interaction mechanism between lumacaftor and helicase. Different concentrations (50–250 µmol/L) of ningnanmycin (F), raltegravir (G), eltrombopag (I), and lumacaftor (J) anti-TMV helicase potency in vitro. Error bars indicate the mean ± standard deviation (SD) (n = 3).

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (A) Visualization and distribution of the TMV-GFP virus after 5–10 days of different drugs treatment. (B) The related transcript levels of viral helicase gene treated with ningnanmycin, ribavirin, raltegravir, eltrombopag, and lumacaftor at 250 or 500 µg/mL, respectively. Error bars indicate the mean ± SD (n = 3). ns, no significance. P < 0.05, ***P < 0.001, ****P < 0.0001.

  下载: 全尺寸图片 幻灯片
•