High-throughput screening of small compounds targeting EBOV VP30Expression and Purification of VP30

The sequences of VP30110–272 were codon-optimized and harbored an N-terminal MBP tag, and a 6 × His tag and TEV digestive sequence were inserted between MBP and VP30. The recombinant plasmid pET28a-MBP-VP30110–272 was transformed into E. coli to express the VP30110‒272 protein. After the cells were crushed and centrifuged, the supernatant was successively purified by nickel column affinity, anion exchange, nickel column affinity and size exclusion chromatography (SEC) to high purity. The analysis of SEC and SDS‒PAGE confirmed the purity of expressed VP30 with a molecular weight of 18.3 kDa (Fig. 1a).

Fig. 1: The establishment of FP method.

a Purification profile from size exclusion chromatography on a Superdex 200 column and SDS-PAGE with EBOV VP30110-272 at a molecular weight of 18.3 kDa. b The schematic diagram to illustrate the principle of fluorescence polarization (FP) assay and competitive fluorescence polarization-based assay for screening inhibitors that compete with NP-derived peptide for binding with VP30. Unbound fluorescent labeled NP peptide (NP-FITC) has a rapidly rotating of small molecule fluorophore and gives a low FP signal. The binding of NP-FITC with a large VP30 protein slows down the rotation of the fluorophore, leading to an increased FP signal. The competitive binding to other small molecules to VP30 will free NP-FITC from binding and result in a decreased FP signal. c Fluorescence polarization curves given by VP30 in the presence of NP-FITC at different concentrations (0.031–120 μM). d Fluorescence polarization curves at 300 nM of NP-FITC mixed with various concentrations of the VP30 protein. e Fluorescence polarization changes in competition by adding unlabeled NP peptide in the presence of 300 nM NP-FITC and 7.68 μM of VP30. f Compound libraries of 8004 compounds at 100 μM were screened against VP30-NP binding by fluorescence polarization (FP) assay. Z scores indicate the prominence of potential inhibitors discovered. The percentage of effect is the inhibition rate of the interface activity of VP30/NP by the compounds.

Competitive binding assay between the NP-derived peptide and small molecules for binding to VP30

To screen for small molecules binding to the same interaction pocket between NP and VP30, we established a high-throughput competitive assay using fluorescence polarization (FP). The binding competition between the fluorescently labeled NP-derived peptide and a small molecule can be monitored by changes in polarization and measured by the generation of a fluctuating FP signal [31] (Fig. 1b). EBOV VP30 binding peptide (NP) was conjugated to fluorescein 5-isothiocyanate (FITC), which could bind to VP30 and lead to a slower tumbling speed, resulting in an increase in fluorescence. When the tested compounds competitively bind to the VP30 protein, the fluorescent small molecule NP-FITC is in a free state, and the rotation speed becomes faster, resulting in a decrease in the fluorescence signal. To establish this assay, we first determined the optimal concentration for fluorescein 5-isothiocyanate (FITC)-labeled NPs (NP-FITC). Using various concentrations of NP-FITC as the substrate and VP30 as the target protein, 300 nM NP-FITC was selected as the optimal concentration in this assay (Fig. 1c). The trend observed for the curve is better under this condition, and the concentration of NP-FITC is moderate. Then, we determined the optimal concentration of VP30 after using various concentrations of VP30 protein to react with 300 nM NP-FITC, and the IC50 was 3.07 μM (Fig. 1d). The optimal concentration of VP30 was determined to be 7.68 μM by concentration titration, which is one gradient value above the IC50 value as the optimum concentration. Afterward, we employed unlabeled NP peptide to confirm the reliability of this competitive binding assay. A constant concentration of VP30 was incubated with increasing concentrations of unlabeled NP peptide as a competitor. After incubation, NP-FITC was added to measure the changes in FP signals, which successfully reproduced the binding between NP-derived peptide and VP30 (Fig. 1e). The IC50 value of the NP polypeptide in the competitive FP assay was 0.33 μM. The above established FP method was performed to conduct high-throughput screening for compound libraries including 8004 compounds containing clinically approved drugs and natural products. Compounds were screened at a final concentration of 100 μM in each well. The results of high-throughput screening based on the FP method are displayed in a volcano plot (Fig. 1f). In this screening experiment, compounds with Z scores less than −4 and inhibitory effect values greater than 80% were defined as potential antiviral inhibitors [18].

Confirmation of Embelin and Kobe2602 inhibitors identified from high-throughput screeningDose-dependent competitive assay

The above screening led to a total of 70 compounds with Z scores < −4 and inhibitory effect values > 80% (Fig. 1f). Among them, we first excluded self-fluorescence interference according to the properties of the compounds and then removed those small molecules with poor medicinal properties according to the pharmacological effects. Finally, we chose 15 potential inhibitors for low-throughput dose-dependent experiments to determine their IC50 values. The Z values and inhibition rates of these compounds are moderate but do not deviate too much from the median to exclude the fluorescence interference of the drug themselves. Among them, Embelin and Kobe2602 were prominent due to their relatively strong inhibitory activity in the FP assay. Embelin is an active benzoquinone compound that is extracted from an herb of the Myrsinaceae family [32] (Fig. 2a). Kobe2602 generally serves as an effective inhibitor of the Ras signaling pathway and is a novel antitumoral agent [33, 34] (Fig. 2b). The IC50 values of Embelin and Kobe2602 from the competitive assay were 33.19 and 26.95 μM, respectively (Fig. 2c, d).

Fig. 2: Specific binding of small molecule inhibitors to VP30 protein.

a Chemical structure of Embelin. b Chemical structure of Kobe2602. c, d The dose‒response curve of Embelin and Kobe2602 in fluorescence polarization (FP) assay. The IC50 values of Embelin and Kobe2602 are 33.19 and 26.95 μM, respectively. e, f Sensorgrams of Embelin and Kobe2602 in the SPR assay at various compound concentrations. The affinity constant (KD) of Embelin and Kobe2602 are 4.62 and 0.88 μM, respectively. g, h Tm changes of VP30 in the absence or presence of small molecule inhibitors in the thermal shift assay (TSA) at various pH. Under various pH conditions, Embelin and Kobe2602 both can specifically bind to VP30 and change its melting temperature Tm.

Binding assay

The binding between VP30 and small molecules can be confirmed by surface plasmon resonance (SPR). The binding affinity (KD) of Embelin and Kobe2602 were found to be 4.62 and 0.88 μM, respectively (Fig. 2e, f).

Thermal shift assay

TSA experiments were performed to characterize the melting temperature (Tm) of a protein after adding a potential inhibitor to further confirm the binding interaction [35]. As shown in Fig. 2g, Embelin increased the stability of VP30 at pH 6.5, 7.0, 7.5 and 8.0. Kobe2602, by comparison, can increase the stability of VP30 at all pH ranges tested (from 5.5 to 8.5) (Fig. 2h). Notably, under different pH conditions, the number of ions in the solution alters the surface charge of the protein, which affects the binding affinity between the protein and the small molecule compound and produces different changes in Tm values. Moreover, some compounds exert different effects on protein stability and decrease or increase the melting temperature (Tm) under different pH conditions. Zhao et al. reported similar results in Nature, 2016 [36].

Interaction mechanism of Embelin and Kobe2602 with EBOV VP30

To further clarify the interaction mechanism between VP30 and the small molecule discovered (Embelin and Kobe2602), we performed docking and molecular dynamics (MD). Previously, we reported that there are two key residues in the interaction interface of VP30 and NP [12]. Mutating one of these two key residues (Glu197, Trp230) to alanine will result in complete loss of binding between VP30 and NP. It is interesting to know if the binding locations of Embelin and Kobe2602 are near Glu197 and Trp230. We employed AutoDock Vina to generate initial conformations and subsequent long 200 ns MD simulations to examine the stable binding conformations. As shown in Fig. 3, the average distance between the center of mass (COM) of Embelin and the COM of the sidechain of Trp230 is approximately 6 Å, suggesting that both Embelin and Kobe2602 bind in or near the NP-VP30 binding pocket. The 2D diagrams describing the distance between small molecules and the key amino acid residues of the VP30 protein are now presented in the supplementary materials (Fig. S1).

Fig. 3: Binding of Embelin and Kobe2602 in the VP30-NP binding pocket.

a, b Embelin/Kobe2602 RMSD represents the heavy-atom RMSD of Embelin/Kobe2602 with the docking conformation as the reference. Embelin/Kobe2602 and Glu197/Trp230 represent the distance between the center of mass (COM) of Embelin/Kobe2602 and the COM of Glu197/Trp230 side chain. c, d The stable conformation of Embelin and Kobe2602 binding with VP30 found by MD simulations, along with the NP-derived peptides. VP30 is shown in gray, Embelin and Kobe2602 in magenta, the VP30 residues within 5 Å of NP peptide in cyan.

Inhibiting the transcription and replication of Ebola virus in the minigenome system

Due to the high lethality of EBOV, the minigenome system was used as a surrogate of the authentic virus for use in BSL-2 laboratories. The MG system can replicate for months and stably express critical key proteins of the virus life cycle. We observed that Embelin and Kobe2602 have a similar capability to inhibit virus reproduction, with EC50 values of 16.85 μM and 22.93 μM, respectively (Fig. 4a, b). Simultantously, the cytotoxicity of these drugs was negligible in Huh7-4P cells, with a 50% cytotoxic concentration (CC50 value) of 494.1 μM for Embelin and 382.5 μM for Kobe2602. Thus, these two small molecule drugs exhibit anti-Ebola virus activities, making them attractive candidates for the development of antiviral inhibitors.

Fig. 4: Validation on the antiviral activities of small molecule inhibitors in the Ebola minigenome system.

a, b The inhibitory curve and cytotoxic measurement in the minigenome assay. The inhibition concentration on virus transcription and replication of Embelin and Kobe2602 (EC50) are 16.85 μM and 22.25 μM, respectively. Cell viabilities (green) were determined by CCK-8 assay. The toxicities of the two compounds are all greater than 200 μM.

Embelin and Kobe2602 analogsEmbelin analog

Based on the relationship between drug core structural frameworks and pharmacological activities, we tested a structural analog of Embelin, 8-gingerol (Fig. 5a). It has been reported that Embelin is an inhibitor of XIAP (inhibitor of apoptosis) and simultaneously modulates multiple signaling pathways, such as NF-κB, PI3K/AKT, P53 and JAK/STAT3, to induce apoptosis [32, 37, 38]. As a structural analog of Embelin, 8-gingerol can inhibit the generation of reactive oxygen species (ROS) and shows good antioxidant and anti-inflammatory effects [39]. More interestingly, in terms of the mechanism of pharmacological activity, 8-gingerol is consistent with multiple signaling pathways regulated by Embelin, which increases the credibility of its study as an analog [40]. Competitive binding assays based on fluorescence polarization indicated that 8-gingerol has an IC50 value of 61.95 μM (Fig. 5b). 8-Gingerol has a much weaker IC50 than Embelin (33.19 μM). Thus, it is not surprising that 8-gingerol showed weaker antiviral activity in the MG assay with an EC50 of 32.00 μM (Fig. 5c). The cytotoxicity of this small molecule displayed negligible toxicity to Huh7-4P cells with a CC50 of 274.50 μM for 8-gingerol. Unlike Embelin, 8-gingerol negatively affects the stability of VP30 for all pH ranges tested (Fig. 5d) with a poorer binding affinity at 17.90 μM, compared to 4.62 μM for Embelin (Fig. 5e). This reduction in binding affinity may be associated with the reduction of the carbonyl groups on the benzene ring in 8-gingerol.

Fig. 5: Binding assay of Embelin analog: 8-gingerol.

a Chemical structure of 8-gingerol, compared to Embelin. b The inhibitory curve and cytotoxic effect of 8-gingerol in minigenome assay. The inhibition activity of 8-gingerol (red) are 32.00 μM. Cell viabilities (green) were determined by CCK-8 assay. The toxicity of 8-gingerol is greater than 200.00 μM based on CC50. c The dose-response curve of 8-gingerol in the competitive fluorescence polarization (FP) assay. The IC50 of 8-gingerol is 61.96 μM. d Changes in melting temperatures (Tm) of VP30 in the absence or presence of small molecule inhibitors in the thermal shift assay (TSA) at various pH. Unlike Embelin, 8-gingerol destabilizes, rather than stabilizes VP30. e Sensorgrams of 8-gingerol in the SPR assay at different compound concentrations. The affinity constant (KD) of 8-gingerol is 17.90 μM.

Kobe2602 analog

We also examined a structural analog of Kobe2602, Kobe0065 (Fig. 6a). Kobe2602 and Kobe0065 are both inhibitors of Ras proteins that were discovered by structure-based molecular drug design in silico [33]. Competitive binding assays based on fluorescence polarization indicated that Kobe0065 has an IC50 value of 29.57 μM (Fig. 6b), a value similar to that of Kobe2602 (26.95 μM). However, Kobe0065 exhibited a weaker binding affinity to VP30 (KD = 65.60 μM) than that of Kobe2602 (0.88 μM) (Fig. 6c). Kobe0065 caused a similar stabilization effect on VP30 in TSA analysis across all pH ranges (Fig. 6d). Despite a similar IC50 for the competitive binding assay, similar stabilization effect, and weaker binding affinity to VP30, Kobe0065 exhibits a much stronger MG activity with an EC50 of 1.33 μM compared to 22.25 μM for Kobe2602 (Fig. 6e). Kobe0065 is also relatively nontoxic, with a CC50 of 278.20 μM.

Fig. 6: Binding assay of Kobe0065, the analog of Kobe2602.

a Chemical structure of Kobe0065, compared to Kobe2602. b The inhibitory curve and cytotoxic effect of 8-gingerol in minigenome assay. The inhibition activity of Kobe0065 (red) are 1.33 μM. Cell viabilities (green) were determined by CCK-8 assay. The toxicity of Kobe0065 is greater than 200.00 μM based on CC50. c The dose‒response curve of 8-gingerol in fluorescence polarization (FP) assay. The IC50 of Kobe0065 is 29.57 μM. d Changes in melting temperatures (Tm) of VP30 in the absence or presence of small molecule inhibitors in the thermal shift assay (TSA) at various pH. Similar to Kobe2602, Kobe0065 stabilizes VP30. e Sensorgrams of Kobe0065 in the SPR assay at different compound concentrations. The affinity constant (KD) of Kobe0065 is 65.60 μM.

Compounds that disrupt the VP30-NP interaction in vitro

We tested the abilities of selected small molecule inhibitors to block the binding of VP30 and NP using purified MBP-His-VP30110–272 and NP600 proteins by pull-down assays with MBP beads [41]. As shown in Fig. S2, selected compounds can effectively abolish or attenuate the NP-VP30 interaction. In small-molecule competition assays, different small molecule inhibitors competitively bind to the VP30 protein; as a result, NP proteins are in a free state and washed out. Thus, there was no corresponding NP protein band in the final beads. We further reconstructed the full-length viral plasmids of HA-VP30 and FLAG-NP. Coimmunoprecipitation experiments with anti-HA beads were performed on lysates of HEK293T cells cotransfected with plasmids of HA-VP30 and FLAG-NP [42], Western blotted with mouse anti-FLAG antibody and anti-HA antibody, followed by horseradish peroxidase-conjugated goat anti-mouse antibody. Different levels of FLAG-NP noted in the control group relative to experimental groups may reflect inhibition levels (Fig. S3). Selected compounds can abolish or attenuate the NP-VP30 interaction with full-length viral proteins according to a coimmunoprecipitation assay.

Synergy of inhibitor combinations in inhibiting Ebola virus transcription and replication

Drug combination therapeutic strategies aim to reduce toxicity, enhance efficacy and improve selectivity. These strategies play an important role in antiviral therapies [43]. In the drug combination experiment, we combined two structural types of compounds to evaluate the synergistic effect of their antiviral effects. The synergy scores of two-drug combinations were calculated using the SynergyFinder tool [44, 45]. Considering the antiviral EC50 of the four drugs alone in the minigenome system, we established a drug cross design to evaluate the synergies and sensitivity of antiviral activity of the two drug combinations. Huh7-4P cells were treated with increasing doses of combination drugs in a 7 × 7 concentration checkerboard format [46, 47]. To fully understand the synergy of the drug combination, avoid the calculation deviation caused by a certain calculation mode, HSA, Bliss, Loewe and ZIP synergy metrics were compared for combinations and single agents calculated with the SynergyFinder tool [48].

The dose‒response map of Embelin and Kobe2602 is shown in Fig. 7a. The abscissa and ordinate represent the concentration gradients of the two drugs, respectively, and each grid contains two drugs at different doses. The darker the color, the higher the synergy score. In the experimental results, we observed the dose and corresponding antiviral responses of Embelin and Kobe2602 alone or in combination. In the presence of Kobe2602, Embelin can reach or even exceed its highest response at lower concentrations. Similarly, due to Embelin, Kobe2602 could produce a higher response at lower doses. Fig 7b further shows that the toxicity of the combination on Huh7-4P cells continues to show low toxicity. Loewe in 3D showed an obvious synergistic effect in Fig. 7c. Four synergy metrics of HSA, Bliss, Loewe and ZIP are also reported in Fig. S4a–d and provided a consistent trend. Notably, in different calculation models, Embelin and Kobe2602 show the highest synergistic effects in the condition of 1:1 or 1:2. These results provided a reference for our follow-up studies on the antiviral synergistic effects of combining the two drugs.

Fig. 7: Synergic binding assay.

a The dose–response table of Embelin and Kobe2602 mixture in a 7 × 7 dose matrix with a series of one blank and six half-dilution concentrations. b Cytotoxicity assay of the drug combination of Embelin and Kobe2602. Cell viability matric shows that the two drug combinations have low cytotoxicity. c The synergistic score of Embelin and Kobe2602 mixture in Loewe 3D matrix. d The dose response table of Embelin and Kobe0065 mixture in a 7 × 7 dose matrix with a series of one blank and six half-dilution concentration. e Cytotoxicity assay of the drug combination of Embelin and Kobe0065. Cell viability matric shows that the two drug combinations have low cytotoxicity. f The synergistic score of Embelin and Kobe0065 mixture in Loewe 3D matrix.

We also examined the synergistic effect when Kobe0065 was used in combination with Embelin. In the dose‒response table, Kobe0065 at 1.56 μM and Embelin exhibited excellent antiviral responses at concentrations greater than 0.78 μM (Fig. 7d). For example, at a concentration of Kobe0065 of 3.12 μM, the inhibition rate increased by 10 times with the addition of Embelin. It should be mentioned that the combination of these drugs caused no apparent cytotoxicity to Huh7-4P cells; even at the maximal concentration ratio conditions, these two drug combinations exhibited low cytotoxicity (Fig. 7e). Loewe in 3D for Embelin and Kobe0065 showed a prominent synergistic effect in Fig. 7f. This highest score is better than the combination of Embelin and Kobe2602. Notably, the combination of Embelin and Kobe0065 provided a better synergy score at low concentrations, while the synergy at high concentration ratios was not obvious (Fig. S4e–h). These results highlight the advantages of the synergistic effect when the two drugs are used in combination, that is, the maximum therapeutic effect is achieved with the smallest dose.

We optimized the drug concentration ratios based on the synergy scores obtained above and calculated the drug combination index (CI) under various concentration ratios using CompuSyn software [49]. When the inhibition rate was 50%, the combination index of Embelin and Kobe2602 and that of Embelin and Kobe0065 at various concentration ratios were less than 1, indicating that they produced synergistic effects with each other. Synergies were divided into different intensities according to CI values, in which CI values between 0.1–0.3 represented strong synergies, 0.3–0.7 represented moderate synergies, and 0.8–0.9 indicated weak synergies [50]. According to the calculated CI values and the synergy scores obtained in SynergyFinderPlus, we chose the optimal concentration ratios of each drug combination. After that, the inhibitory effects and antiviral effects of several drug combinations with the selected optimal concentration ratios were verified on the fluorescence polarization (FP) system and minigenome system, respectively (Fig. 8). In the FP assay, the IC50 of Embelin alone was 33.19 μM. The IC50 value was reduced to 21.89 and 2.68 μM when Embelin was combined with Kobe2602 at a ratio of 1:2 and with Kobe0065 at a ratio of 1:4, respectively (Fig. 8a, b). Similarly, the IC50 values of 8-gingerol decreased from 61.96 μM to 22.54 μM and 3.15 μM when combined with Kobe2602 (1:2) and Kobe0065 (1:2), respectively (Fig. 8c, d). The IC50 of Kobe0065 decreases from more than 20 μM to 2–3 μM when used in combination with Embelin at a ratio of 4:1. A large decrease in the IC50 is summarized in Fig. 8i. The reduction in IC50 when combining Kobe0065 with Embelin or 8-gingerol is larger than when Kobe2602 is combined with the two inhibitors.

Fig. 8: Confirmation of the synergistic effects from drug combinations.

a–d Competitive FP assays of the drug combinations of Embelin and Kobe2602 at the ratio of 1:2, Embelin and Kobe0065 at the ration of 1:4, 8- gingerol and Kobe2602 at the ratio of 1:2 and 8-gingerol and Kobe0065 at the ratio of 1:2 on the VP30/NP binding interface. IC50 values are 21.89, 2.68, 22.54, and 3.15 μM, respectively. e–h Different drug combinations inhibitory activities of the transcription and replication of Ebola virus in minigenome assay. EC50 of Embelin and Kobe2602 (1:2) in mixture are 2316 and 463.7 nM respectively. EC50 of Embelin and Kobe0065 in mixture (1:4) are 98.80 and 395.10 nM respectively. EC50 of 8-gingerol and Kobe2602 (1:2) in mixture are 759.80 and 1520.00 nM respectively. EC50 of 8-gingerol and Kobe0065 (1:2) in mixture are 350.60 and 701.20 nM respectively. i Comparison IC50 given by each compound administered alone and in combination with another compound as labeled in FP assay. Embelin:Kobe2602 (1:2), Embelin:Kobe0065 (1:4), 8-gingerol:Kobe2602 (1:2), Embelin:Kobe0065 (1:2). j In MG assay, comparison of the EC50 when each compound administered alone or with another compound. Embelin:Kobe2602 (1:2), Embelin:Kobe0065 (1:4), 8-gingerol:Kobe2602 (1:2), Embelin:Kobe0065 (1:2).

We further examined the synergistic effect of using two compounds by employing the MG assay. When combining Embelin with Kobe2602 (1:2) or Kobe0065 (1:4), the EC50 values decreased from 16.85 μM to 2316 nM and 98.80 nM, respectively (Fig. 8e, f). After mixing 8-gingerol with Kobe2602 or Kobe0065 in a 1:2 proportion, the EC50 was reduced from 32 μM to 759.8 and 350.6 nM, respectively (Fig. 6g, h). The combined use of two compounds in MG analysis is summarized in Fig. 8j, indicating a large reduction in EC50 (as high as 90-fold) occurred. Simultaneously, the antiviral effects of Kobe2602 and Kobe0065 in combination with Embelin or 8-gingerol were better than those obtained with these compounds as single agents, and their EC50 values were all significantly reduced. For example, when combining Kobe2602 with Embelin (2:1) or 8-gingerol (2:1), the EC50 values of Kobe2602 decreased from 22.25 μM to 463.70 nM and 1520 nM, respectively (Fig. 8e, g). The drug combination of Kobe0065 with Embein and 8-gingerol decreased its EC50 from 1.33 μM to 395.1 nM and 701.2 nM, respectively (Fig. 8f, h). In the anti-Ebola virus experiment, the advantages of combined drugs were more prominent. The combined use of each drug combination in MG analysis resulted in several-fold or even dozens-fold improvement in antiviral efficacy (Fig. 8j).

To further confirm that binding to VP30 was synergistic rather than competitive between the two structural types found in this work (Embelin and Kobe2602), we performed the SPR binding assay as before but with a small-molecule chip as the stationary phase along with VP30 and another small molecule as the first and second mobile phases to bind in sequence. The observed changes in binding curves are shown in Fig. S5a. Embelin or 8-gingerol was first bound to VP30. The addition of Kobe2602 or Kobe0065 leads to a further increase in binding (Figure S5a). We also tested using VP30 as the stationary phase of the chip and another compound, GW0742, as a control. GW0742 is also a small molecule inhibitor targeting VP30 that we screened in the FP assay. GW0742 also exhibits good antiviral activity in the MG assay; however, GW0742 does not exhibit obvious synergistic effects in combination with other similar drugs. The results showed that VP30 could still efficiently bind to Kobe2602 and Kobe0065 after binding to Embelin or 8-gingerol (Fig. S5b). However, a reduction in binding was observed when GW0742 was added (Fig. S5b). This clearly confirmed the additive effect of Embelin or 8-gingerol with Kobe2602 or Kobe0065.