Screening of Botanical Drugs against Lassa Virus Entry

Currently, there is no approved therapy to treat Lassa fever (LASF). Our goal was to identify potential candidate molecules for LASF therapy. Herein, we screened a botanical drug library and identified two compounds, casticin and bergamottin, that inhibited LASV entry via different mechanisms.

subunits SSP, GP1, and GP2 form a (SSP/GP1/GP2) 3 trimeric complex that is present at the surface of the mature virion and is essential for virus entry. The unusual retained SSP interacts with the membrane-proximal external region as well as the transmembrane domain of GP2 and thus stabilizes the prefusion conformation of GPC and provides an interface with the potential to be targeted by entry inhibitors (8)(9)(10).
Lassa hemorrhagic fever is an annual epidemic in West Africa and peaks during the dry seasons. In 2020, Nigeria faced a large outbreak, with 4,761 suspected and 1,006 confirmed cases as of 17 May, according to the Nigeria Centre for Disease Control. To date, no vaccines or specific antiviral agents against LASV are available. Therapeutic strategies are limited to the administration of ribavirin in the early course of the illness. To address this issue, we screened a botanical drug library of 1,058 compounds. Natural-product screening could increase the speed of discovery and development for treatment, provide a strategy for drug optimization, and facilitate the elucidation of the mechanism underlying the infection (11). All the compounds were extracted from botanicals with a purity of $98%. By using a pseudotype of LASV in BSL2 facilities, we determined that the compounds bergamottin and casticin inhibit LASV entry.

RESULTS
Bergamottin and casticin block LASV entry. To construct viral systems to study viral entry in BSL2 laboratories and facilitate high-throughput screening, recombinant LASV (LASVrv) based on the vesicular stomatitis virus (VSV) backbone containing the gene for green fluorescent protein (GFP), as well as the LASV GPC gene, was constructed. LASVrv was replication competent at a titer of 1.6 Â 10 7 PFU/ml. A pseudotype of LASV (LASVpv) was also constructed with the VSV backbone, and the GPC gene was replaced with a luciferase gene. LASVpv was competent for a single round of viral entry and infection (9).
As shown in Fig. 1A, bergamottin and casticin were selected as successful hits after three rounds of screening. Both compounds exhibited robust inhibition of LASVrv, whereas they had mild effects on VSV, suggesting that they inhibit LASV infection at the entry step. The 50% inhibitory concentrations (IC 50 s) of bergamottin and casticin against LASVpv on Vero cells were 3.615 and 0.6954 mM, respectively, while the values obtained for A549 cells were 4.300 and 1.696 mM, respectively, indicating that both compounds exhibit similar inhibitory effects on different cell types ( Fig. 1C and E). As neither bergamottin nor casticin showed cytotoxicity at the highest tested concentration (100 mM), the selective index (SI; the ratio of the 50% cytotoxic concentration to the IC 50 ) values were calculated to be .27.6 and .143.8, respectively. This suggests that both drugs are promising possibilities to combat Lassa fever (LASF).
Casticin inhibits LASV GPC-mediated membrane fusion. As the unique SSP-GP2 interface plays critical roles in stabilizing the prefusion conformation of GPC and thus constitutes the "Achilles heel" targeted by most mammarenavirus entry inhibitors, the effects of bergamottin and casticin on LASV GPC-mediated membrane fusion were studied. As shown in Fig. 2A, casticin effectively blocked low-pH-triggered membrane fusion in a dose-dependent manner, whereas bergamottin showed little effect at all tested concentrations.
To further quantitatively evaluate inhibition, fusion efficacy was determined using the dual-luciferase assay (9,(12)(13)(14). As shown in Fig. 2B, casticin exhibited a dose-dependent inhibition of GPC-mediated membrane fusion. Furthermore, the inhibition effect shown in the membrane fusion assay was similar to that obtained from the infection assay (Fig. 1E). Notably, casticin inhibited LASV GPC-mediated membrane fusion by ;65% at 25 mM, while the inhibition by bergamottin was ,50% at the same concentration. Bergamottin had no inhibitory effect on LASV GPC-mediated fusion at concentrations as high as 400 mM (Fig. 3A), suggesting that bergamottin blocks LASV entry via a different mechanism than casticin.
Bergamottin blocks LASV endocytic trafficking. To explore the mechanism underlying bergamottin inhibition, we further investigated the effects of both compounds on virus binding and endocytic trafficking. First, binding efficacy was evaluated in the absence and presence of the compounds in A549 cells, and the IIH6 antibody (a blocking antibody targeting the glycan epitope on a-dystroglycan [DG]) was used as the positive control (15,16). As shown in Fig. 3B, IIH6 significantly blocked LASVrv transduction to the cell, while neither compound showed inhibitory effects, suggesting that neither bergamottin nor casticin interferes with the binding of LASV GP1 to the cell surface receptor.
Next, we tested whether both compounds could bind to LASV GPC and exert a virucidal effect. To this end, LASVpv (2 Â 10 5 PFU) was mixed with either compound at 25 mM for 1 h; the mixture was then diluted 200-fold to the noninhibitory concentration and added to the cells for 1 h (17). Luciferase activity was determined 23 h later. As shown in Fig. 3C, the luciferase activities observed in the compound-treated groups (E) Drug-drug interactions of bergamottin with NH 4 Cl. Vero cells were infected with LASVrv (MOI, 0.01) and treated with various concentrations of bergamottin in combination with NH 4 Cl. Differential surface plots at the 95% confidence level were calculated and generated using MacSynergy II for the drug-drug interactions. Data are means and SD from 3 to 8 independent experiments. ***, P , 0.001, compared to DMSO.
were similar to those in the dimethyl sulfoxide (DMSO) control group, suggesting that neither compound had a virucidal effect on LASVpv.
As bergamottin did not exert a virucidal effect or an inhibitory effect on the binding or fusion step, we hypothesized that it might inhibit endocytic trafficking. To address this, we quantitatively compared the inhibition of the endocytic efficiency by bergamottin with the acidification antagonist NH 4 Cl (18,19). As shown in Fig. 3D, 1 h of incubation with LASVrv (multiplicity of infection [MOI] of 0.01) at 4°C led to approximately 2,000 cells being infected in a 96-well plate, while treatment with NH 4 Cl significantly decreased the number of infected cells to ,200, especially within the first 30 min, in which LASV is supposed to undergo pH-dependent fusion (15,20). Similar to the inhibitory effect of NH 4 Cl, bergamottin suppressed the early stage of LASVrv infection. These results suggest that bergamottin inhibits LASV entry by acting on the endocytic trafficking of the virus. Intriguingly, the combination of bergamottin and NH 4 Cl had significantly synergistic antiviral effects, suggesting that they might act on distinct targets in the endocytic trafficking step (Fig. 3E). Notably, we tried to select adaptive mutants by serially passaging LASVrv in the presence of bergamottin, and no mutant was obtained after 12 rounds of passage, which correlated with the results that bergamottin acted on the cellular endocytic pathway as opposed to targeting the virion.
We also conducted a time-of-addition assay with 25 mM casticin. At that concentration, casticin exerted an inhibitory effect similar to that of bergamottin and NH 4 Cl. Of note, casticin at lower concentrations (#12.5 mM) showed little inhibition of VSVpv (Fig. 1E), which utilizes the endocytic pathway for entry (21). We suggest that high concentrations of casticin could exert nonspecific inhibition on the endocytic pathway. As  casticin showed robust dose-dependent inhibition of GPC-mediated membrane fusion, it can be concluded that casticin inhibits LASV entry primarily by blocking fusion.
The LASV GPC F446L mutation confers resistance to casticin. Since most arenavirus fusion inhibitors serve as direct antiviral agents by targeting the SSP-GP2 interface, we further investigated the viral targeting of casticin by selection for resistance mutations. As shown in Fig. 4A, after five rounds of passaging in the presence of 6.25 mM casticin (corresponding to an approximate IC 90 ), LASVrv exhibited a steady increase of resistance to casticin which was absent in the DMSO control group. We sequenced the casticin-resistant LASVrvs from passage 6 (P6) to P10 without plaque purification. As shown in Fig. 4B, a resistant mutant with a change of a phenylalanine (TTC) to leucine (CTC) emerged at P7 and completed the substitution by P10. F446 is located at the transmembrane domain of GP2, which is highly conserved in mammarenaviruses (Fig.  4D). We introduced the F446L mutation into the GPC and generated a pseudotype of the mutant, LASV F446L pv. LASV F446L pv exhibited resistance to casticin. The inhibition efficiency against wild-type LASVpv infection reached ;98% and ;80% with casticin at 25 mM and 1.5625 mM, respectively, while only ;15% inhibition of LASV F446L pv was observed with 25 mM casticin (Fig. 4C). Intriguingly, the F446 (corresponding to F438 in JUNV GPC) mutation has been reported to confer resistance to other structurally distinct compounds (22)(23)(24)(25), suggesting that phenylalanine at this position might be accessible to different inhibitors and that the F446 mutation might contribute to the stabilization of GPC induced by the inhibitors.
Casticin extends the antiviral spectrum to other mammarenaviruses. As F446 is highly conserved in mammarenaviruses (Fig. 4D), we asked whether casticin could inhibit entry of other arenaviruses. To address this, pseudotypes of pathogenic OW viruses (including the prototype lymphocytic choriomeningitis virus [LCMV], LUJV, and the most closely related Mopeia virus [MOPV]) and NW viruses (including JUNV, MACV, GTOV, SBAV, and CHAPV) were generated with the VSV backbone. As shown in Fig. 5, casticin effectively inhibited both NW and OW pseudotype virus infections in a dosedependent manner. Moreover, we constructed the corresponding phenylalanine-toleucine mutation in each virus pseudotype and tested the sensitivities of these mutant viruses to casticin. Notably, all the mutant NW pseudotype viruses, similar to LASV F446L pv, exhibited resistance to casticin, while the pathogenic OW viruses (LCMV and LUJV) bearing the leucine mutation were still sensitive to casticin. We hypothesize that casticin might inhibit LASV and NW mammarenavirus entry by a similar mechanism. To this end, we tested whether casticin inhibited membrane fusion caused by NW GPC. As shown in Fig. 5B and C, casticin at 80 mM inhibited NW GPC-mediated membrane fusion, while the corresponding F446L mutation conferred various degrees of resistance to casticin inhibition by GPC-mediated membrane fusion.
Both casticin and bergamottin inhibit authentic LCMV infection. To test the antiviral effects of both compounds on an authentic mammarenavirus, the BSL-2-compatible arenavirus, LCMV was utilized. As shown in Fig. 6A, bergamottin inhibited LCMV infection in a dose-dependent manner with an IC 50 of 2.624 mM. Similarly, casticin inhibited LCMV infection with an IC 50 of 7.388 mM (Fig. 6B), which was in line with casticin inhibition of LCMVpv infection (Fig. 5A). We further tested the inhibitory effects of both compounds against LCMV GPC-mediated membrane fusion. Intriguingly, both bergamottin and casticin showed inhibition against membrane fusion ( Fig. 6C and D), suggesting that bergamottin might inhibit LCMV through a different mechanism than LASV (Fig. 3A). We attempted to obtain adaptive mutants of LCMV by serially passaging with increasing concentrations of bergamottin, but no mutants were observed in the entire genome.
Both casticin and bergamottin inhibit the activity of LASV mini-genome. To test whether casticin and bergamottin could inhibit the replication step of LASV, a LASV minigenome (MG) cell-based assay was conducted (26). As shown in Fig. 7, bergamottin exerted dose-dependent inhibition of LASV MG, whereas casticin showed a lessened inhibitory effect at concentrations above 50 mM, suggesting that bergamottin, and potentially casticin, might target RNA synthesis directed by the virus ribonucleoprotein complex.

DISCUSSION
In this study, we screened a botanical drug library and identified two compounds, bergamottin and casticin, that blocked entry of pseudotype and recombinant LASV infection. Casticin is a constituent of Vitex agnus-castus seeds; natural products containing casticin have been extensively used to treat premenstrual syndrome (PMS) with few adverse effects (27,28). Intravenous injection of 50 mg/kg in rats has been shown to lead to a maximum plasma concentration (C max ) of ;34 mM (29), which is far higher than the IC 50 reported here. Notably, casticin has also been shown to inhibit CYP3A4 (27,28), which is the most abundant CYP450 in the liver and intestines. Herein, we demonstrated that casticin inhibits LASVpv entry by acting as a direct antiviral agent (DAA). Casticin blocked the LASV GPC-mediated membrane fusion, and the viral target was the conserved phenylalanine located in the transmembrane domain of GP2. Furthermore, we found that the phenylalanine-to-leucine mutation at the corresponding site caused the NW pathogenic mammarenaviruses to confer resistance to casticin. Of note, when sequencing the resistant viruses, we observed only the mutant emerging in GPC. However, when we constructed the F446L mutant LASVpv de novo by introducing the mutant site into GPC and generated LASV F446L pv, resistance to casticin was conferred, indicating that the adaptive mutation in GPC made LASV resistant to casticin. Additionally, mutations at this position have been noted to confer resistance to distinct entry inhibitors (22)(23)(24). Whether the phenylalanine at this site was directly accessible to the compounds or worked allosterically to prevent viral entry or whether the mutation increased the stability of the glycoprotein requires further investigation.
We also identified bergamottin via high-throughput sequencing (HTS) because of its strong inhibition against LASV entry. Bergamottin is a natural furanocoumarin found principally in grapefruit, which has been shown to inhibit cytochrome P450 (CYP450) enzymes and thus affects the metabolism of many drugs (30,31). Herein, we showed that bergamottin inhibits LASVpv entry by blocking viral endocytic trafficking, which seems to be independent of the effect on CYP450. Of note, we were unable to identify the specific target of bergamottin in anti-LASV inhibition. Although bergamottin showed a similar time course of inhibition to NH 4 Cl, it might act via a different mechanism from blocking the endosomal acidification, because the combination of bergamottin and NH 4 Cl had significantly synergistic antiviral effects. Both NW and OW arenaviruses utilize the endocytic pathway for entry. Pathogenic NW arenaviruses use TfR1 as the primary receptor and consistently engage in clathrin-mediated endocytosis, in which the gradually lower pH leads to endosome maturation (32). In contrast to NW, entry by the OW arenaviruses bypasses the early endocytic pathway and is instead initiated by binding to the primary receptor a-DG (33) and transported with the multivesicular bodies to the late endosome (20,34). Notably, the OW species LUJV utilizes NRP2 as the primary receptor that is needed to switch the receptor to CD63 to complete productive entry (35,36). Similarly, LASV undergoes receptor switching and engages LAMP1 in the late endosome for efficient fusion (15,37). As both the NW and OW mammarenaviruses use receptor-mediated endocytosis to enter cells, bergamottin is thought to inhibit infections by blocking endocytic trafficking. Moreover, combination of bergamottin with other inhibitors might offer benefits beyond a single monotherapy.
In total, this study identified bergamottin and casticin as LASV entry inhibitors, highlighting two compounds derived from natural products that inhibit by different mechanisms. This study also highlighted the role of F446 in the SSP-GP2 interface in drug resistance in mammarenaviruses and identified the broad-spectrum antiviral effects of casticin.  (38,39). 293T cells transfected with pCAGGSGPC were infected at an MOI of 0.1 for 2 h with pseudotype VSV (described below) in which the G gene was replaced with a luciferase gene. The culture supernatants were harvested 24 h later, centrifuged to remove cell debris, and stored at 280°C. The recombinant VSV expressing LASV GPC was generated as described previously (9,39,40). The plasmid used for constructing the recombinant virus was pVSVDG-eGFP (where eGFP is enhanced green fluorescent protein) (plasmid 31842; Addgene, Watertown, MA, USA). GPC was cloned into the DG site, and the construct was designated pVSVDG-eGFP-GPC. BHK-21 cells grown in 6-well plates were infected with a recombinant vaccinia virus (vTF7-3) encoding T7 RNA polymerase at an MOI of 5. After 45 min, cells were transfected with 11 mg of mixed plasmids with a 5:3: 5:8:1 ratio of pVSVDG-eGFP-GPC (pVSVDG-Rluc for generating pseudotype VSV), pBS-N, pBS-P, pBS-G, and pBS-L, respectively. After 48 h, the supernatants were filtered to remove the vaccinia virus and inoculated into BHK-21 cells that had been transfected with pCAGGS-VSV G 24 h previously. The pseudotype and recombinant viruses bearing LASV GPC were designated LASVpv and LASVrv, respectively. The titer of pseudotype virus was measured by infecting BHK-21 cells previously transfected with pCAGGS-VSV G and determined by plaque assay 24 h postinfection. The titer of the recombinant virus was determined by plaque assay. The titers of LASVpv and LASVrv were 3 Â 10 7 /ml and 1.6 Â 10 7 /ml, respectively.
HTS assay of the botanical drug library. A library of 1,058 botanical drug compounds was purchased from Weikeqi Biotech (Sichuan, China). The compounds were stored as 10 mM stock solutions until use. HTS was carried out as shown in Fig. 1A. Vero cells were treated in duplicate with the compounds (50 mM); 1 h later, cells were infected with LASVrv (MOI = 0.1) for 1 h. After 23 h, cells were fixed with 4% paraformaldehyde and stained with 49,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St. Louis, MO, USA). Nine fields per well were imaged on an Operetta high-content imaging system (PerkinElmer, Waltham, MA, USA), and the percentages of infected and DAPI-positive cells were calculated using Harmony 3.5 (PerkinElmer). Cell viability was evaluated using a 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. To determine the IC 50 , LASVpv at an MOI of 0.1 was used utilizing the timeline described above. Luciferase activity was measured using the Renilla luciferase (Rluc) assay system (Promega, Madison, WI, USA), and IC 50 was calculated using GraphPad Prism 6.
Membrane fusion assay. 293T cells cotransfected with pCAGGS-GPC and pEGFP-N1 were treated with either compounds or vehicle (DMSO) for 1 h, followed by incubation for 15 min with acidified (pH 5.0) medium. The cells were then placed in neutral medium, and syncytium formation was visualized 1 h later.
For quantification of the luciferase-based fusion assay, 293T cells in 24-well plates were transfected with both pCAGGS-GPC (0.25 mg) and plasmids expressing T7 RNA polymerase (pCAGT7, 0.25 mg) and cocultured at a ratio of 3:1 with targeted cells transfected with pT7EMCVLuc (2.5 mg per well in 6-well plates) and 0.1 mg pRL-CMV (plasmids used in this assay were kindly provided by Yoshiharu Matsuura, Osaka University, Osaka, Japan). After 12 h, compound treatment and pH induction were conducted as described above. Cell fusion activity was quantitatively determined after 24 h by measuring firefly luciferase activity expressed by pT7EMCVLuc and was standardized with Rluc activity expressed by pRL-CMV using the Dual-Glo luciferase assay (Promega) (9,(12)(13)(14).
IIH6 inhibition assay. A549 cells were pretreated with either 25 mM bergamottin or 25 mM casticin at 37°C for 1 h. As a control, A549 cells were pretreated either with IIH6 (sc-53987; Santa Cruz Biotechnology, Dallas, TX, USA) or a control IgM at 4°C for 2 h (15,16). The cells were then transferred to ice, and LASVrv (MOI = 1) was added for 1 h. After three washes with cold phosphate-buffered saline (PBS), the cells were incubated at 37°C for 24 h. GFP-positive cells were counted using an Operetta highcontent imaging system.
Virucidal assay. To study virucidal effects, approximately 2 Â 10 5 PFU of either LASVpv or VSVpv was incubated with compound (25 mM) or vehicle at 37°C for 1 h. The mixture was diluted 200-fold to a noninhibitory concentration to infect Vero cells in a 96-cell plate. Luciferase activity was determined 24 h later, as described above.
Time-of-addition assay. The experimental timeline is shown in Fig. 3D. At 21 h, A549 cells were infected with LASVrv (MOI = 0.01) at 4°C for 1 h and washed with PBS. The temperature was then increased to 37°C to synchronize the infections. The compounds were incubated with the cells for different time courses, as shown in Fig. 3D.
Selection of adaptive mutants. Drug-resistant viruses were generated by passaging LASVrv on Vero cells in the presence of 6.25 mM casticin. LASVrv was also passaged in the presence of 0.5% DMSO in parallel as a control. RNA from the resistant viruses was extracted using TRIzol (TaKaRa Bio, Inc., Kusatsu, Japan) and reverse transcribed using the PrimeScript RT reagent kit (TaKaRa). The GPC segment was amplified and sequenced as previously described (9). Mutant sites were introduced to LASVpv as previously described, and casticin sensitivities were determined by Rluc activity.
MG assay. The MG assay was conducted as previously reported (26). The LASV minigenome contains the antigenome 59 and 39 untranslated regions (UTRs) of the LASV S segment (Josiah strain; GenBank number HQ688673.1), in which the NP and GPC coding sequences have been replaced with those for ZsGreen (ZsG) and gLuc, respectively. Support plasmids encode LASV L and NP proteins. Vero cells in the 96-well plate were transfected with 90 ng of mixed plasmids with a 3:1:5 ratio of pLASV-S, pCAGGS-LASV-L, and pCAGGS-LASV-NP. After 1 h, the inhibitors were added at the indicated concentrations. Luciferase activity was evaluated 48 h later.