Screening and Identification of Lujo Virus Entry Inhibitors From an Food and Drug Administration-Approved Drugs Library

Lujo virus (LUJV) belongs to the Old World (OW) genus Mammarenavirus (family Arenaviridae). It is categorized as a biosafety level (BSL) 4 agent. Currently, there are no U.S. Food and Drug Administration (FDA)-approved drugs or vaccines specifically for LUJV or other pathogenic OW mammarenaviruses. Here, a high-throughput screening of an FDA-approved drug library was conducted using pseudotype viruses bearing LUJV envelope glycoprotein (GPC) to identify inhibitors of LUJV entry. Three hit compounds, trametinib, manidipine, and lercanidipine, were identified as LUJV entry inhibitors in the micromolar range. Mechanistic studies revealed that trametinib inhibited LUJV GPC-mediated membrane fusion by targeting C410 [located in the transmembrane (TM) domain], while manidipine and lercanidipine inhibited LUJV entry by acting as calcium channel blockers. Meanwhile, all three hits extended their antiviral spectra to the entry of other pathogenic mammarenaviruses. Furthermore, all three could inhibit the authentic prototype mammarenavirus, lymphocytic choriomeningitis virus (LCMV), and could prevent infection at the micromolar level. This study shows that trametinib, manidipine, and lercanidipine are candidates for LUJV therapy and highlights the critical role of calcium in LUJV infection. The presented findings reinforce the notion that the key residue(s) located in the TM domain of GPC provide an entry-targeted platform for designing mammarenavirus inhibitors.


INTRODUCTION
Lujo virus (LUJV) was identified in Zambia and South Africa in 2008. It is the pathogen of Lujo hemorrhagic fever (LUHF; Briese et al., 2009). LUJV is an enveloped, negative-sense, bi-segmented RNA virus belonging to the genus Mammarenavirus (family Arenaviridae; Oldstone, 2002;Briese et al., 2009). The mammarenavirus RNA genome encodes viral polymerase, nucleoproteins, matrix protein (Z), and glycoprotein complex (GPC). GPC is synthesized as a polypeptide precursor that is sequentially cleaved by signal peptidase and the cellular protease subtilisin kexin isozyme-1/site-1 protease to generate the three subunits of the mature complex: the retained stable signal peptide (SSP), the receptor-binding subunit GP1, and the membrane fusion subunit GP2 (Lenz et al., 2001;Eichler et al., 2003;Igonet et al., 2011;Wang et al., 2016). These three non-covalently bound subunits form a (SSP/GP1/GP2)3 trimeric complex that is present at the surface of the mature virion. It plays an essential role in virus entry. The relatively conserved mammarenavirus SSP and GP2 form an interface that not only contributes to the stabilization of the prefusion conformation of GPC, but also provides an "Achilles' heel" that can be targeted by the entry inhibitors.
To date, no vaccines or specific antiviral agents against LUJV have been developed. To address this issue, here we screened a U.S. Food and Drug Administration (FDA)-approved drug library of 1,775 compounds. Drug repurposing is a strategy that is used to accelerate the discovery and development of new and emerging pathogens. Meanwhile, understanding the antiviral mechanisms of potential drugs that could combat LUJV could provide novel insights into its pathogenesis. The safety, pharmacokinetics, and mechanisms of approved drugs have been intensively investigated. Here, therefore, we screened drugs that targeted the entry step of LUJV infection, as this could block viral replication and spreading at an early stage. After three rounds of screening, trametinib [mitogen-activated protein kinase (MAPK) inhibitor used for the treatment of melanoma], manidipine, and lercanidipine (calcium channel antagonists used for the treatment of hypertension), were found to be highly effective against LJUV entry, offering potential new therapies to treat LUHF.

High-Throughput Screening Assay of an FDA-Approved Compound Library
A library of 1,775 FDA-approved drugs was purchased from Selleck Chemicals (Houston, TX, United States). The compounds were stored as 10 mM stock solutions in dimethyl sulfoxide (DMSO) at −80°C until use. High-throughput screening (HTS) was performed as shown in Figure 1A. Vero cells were treated in duplicate with the compounds (10 μM). Then, 1 h later, cells were infected with LUJVrv (MOI, 0.1) for 1 h. After 23 h, cells were fixed with 4% paraformaldehyde and stained with 4', 6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St. Louis, MO, United States). Nine fields per well were imaged on an Operetta high-content imaging system (PerkinElmer), and the percentages of infected and DAPIpositive cells were calculated using the associated Harmony 3.5 software.
After primary screening, 145 inhibitors were defined as prime candidates with inhibition >90% and no apparent cytotoxicity. Then, a counter-screening was conducted to rule out the inhibitors of VSVrv, and 21 compounds with <10% VSVrv inhibition of were selected. Finally, four hit compounds (0.23%) were selected as they exerted the dosedependent inhibition.
LUJVpv was used to determine 50% inhibitory concentration (IC 50 ) values. Vero cells were treated with compounds of the indicated concentrations. Then, 1 h later, cells were infected with LUJVpv, and the supernatant was removed 1 h postinfection. Luciferase activity was measured using the Rluc assay system (Promega, Madison, WI, United States), and the IC 50 was calculated using GraphPad Prism 8 (GraphPad Inc., La Jolla, CA, United States).

Membrane Fusion Assay
293 T cells co-transfected with pCAGGS-LUJV GPC, pcDNA 3.1-CD63 GY234AA , and pEGFP-N1 were treated with compounds or a vehicle (DMSO) for 1 h, followed by incubation for 20 min in acidified (pH 5.0) medium. The cells were then placed in neutral medium, and syncytium formation was visualized 2 h later via light microscopy. Images were processed using ImageJ software for quantification. The boundaries and coverage areas were traced and calculated using the analyzed particles.

Selection of the Adaptive Mutants
Drug-resistant viruses were generated by passaging LUJVrv or LASVrv on vero cells in the presence of 10 μM trametinib. Parallels were passaged in the presence of 2% DMSO as a control. Passaging was terminated when no further improvement in the resistance was detected. RNA from the resistant viruses was extracted using TRIzol (TaKaRa, Kusatsu, Shiga, Japan), followed by reverse transcribed using the PrimeScript™ RT reagent Kit (TaKaRa). The GPC segment was amplified and sequenced using primers, as previously described (Wang et al., 2018). Mutant sites were introduced to LUJVpv or LASVpv, as previously described (Wang et al., 2018), and trametinib sensitivity was determined by Rluc activity.

HTS for LUJV Entry Inhibitors
As studies of LUJV require BSL-4 containment, we utilized a replication-competent recombinant virus of LUJV (LUJVrv, comprising a VSV backbone with a genome containing green fluorescent protein and LUJV GPC) for HTS. LUJVrv possesses the entry characteristics of LUJV and can be handled under BSL-2 conditions (Garbutt et al., 2004;Wang et al., 2018). The HTS assay conditions, including the seeding cell density and LUJVrv infective dose, were optimized at 1.5 × 10 4 cells and 1.5 × 10 2 PFU per 96-well plate, respectively. Under these optimized conditions, the coefficient of variation (CV) and Z' factor were 4.21 and 0.95%, respectively, demonstrating that this assay represents a promising approach for the large-scale screening of inhibitors. A schematic of the HTS is shown in Figure 1A. Inhibitors were defined as prime candidates with >90% inhibition and no apparent cytotoxicity at a concentration of 10 μM. Of the 1,775 tested compounds, 145 (8.17%) were considered as prime candidates. Then, the prime candidates were subjected to counterscreening to rule out the inhibition of VSV genome replication. Twenty-one compounds (1.18%) were selected with inhibition of VSV < 10% at 25 μM. Screening was then performed to reconfirm the results using these compounds over a broader concentration range (0.024-75.0 μM). Four compounds (0.23%) were selected based on their concentration-dependent inhibitory effects. Among these four compounds, the top three hits (trametinib, manidipine, and lercanidipine) were selected for further investigation, and ospemifene was eliminated because of its cytotoxicity ( Figure 1B).
Trametinib is an inhibitor of MAPK, while manidipine and lercanidipine are DHP voltage-gated Ca 2+ channel antagonists. We evaluated the IC 50 of the hit compounds using LUJVpv ( Figure 1B). Furthermore, the inhibitory effects were confirmed using LUJVrv on both Vero and A549 human epithelial cell lines ( Figure 1C). We evaluated the 90% inhibitory concentration (IC 90 ) of hit compounds in Figure 1C. To validate the antiviral effects, trametinib, manidipine, and lercanidipine were purchased from other commercial sources and tested. The cytotoxic and antiviral effects were similar to the results shown in Figure 1.

Trametinib Inhibits LUJV GPC-Mediated Membrane Fusion
The unique retained SSP, together with GP2, provides an "Achilles's heel" that can be targeted by many arenavirus entry inhibitors. We carried out a membrane fusion assay to test whether the three hit compounds act by targeting the SSP-GP2 interface.
Notably, in addition to transfecting 293 T cells with LUJV GPC, plasmids expressing pEGFP and CD63 GY234AA (a mutant leading to CD63 localization on the cell surface) were also transfected. This indicates the formation of syncytium and the triggering of LUJV GPC-mediated membrane fusion (Raaben et al., 2017;Cohen-Dvashi et al., 2018). As shown in Figure 2A, low pH induced obvious membrane fusion in cells transfected with both GPC and mutated CD63, whereas no syncytium formed in cells treated with neutral pH, or in cells lacking CD63 GY234AA .
Mild inhibition was observed under treatment with manidipine even at the highest tested concentration (75 μM). Notably, lercanidipine exerted little inhibition on membrane fusion, even at higher concentrations (15-375 μM; Figure 2D). Although the three hit compounds had similar IC 50 values (all <10 μM), they might inhibit LUJV entry through different mechanisms. Trametinib blocks LUJV entry by inhibiting GPC-mediated membrane fusion, but the other two calcium channel antagonists might use alternative mechanism(s).
A luciferase-based fusion assay has previously been used to quantify membrane fusogenicity (Wang et al., 2018;Zhang et al., 2019;Cao et al., 2021;Liu et al., 2021). However, this luciferase-based fusion assay failed to evaluate LUJV GPC-mediated membrane fusion, as there no difference in luciferase activity was observed between the acidic pH treatment group and the control group. One possible reason for this might be that one additional plasmid, (CD63 GY234AA ) needed to be co-transfected with other plasmids (Wang et al., 2018;Zhang et al., 2019;Cao et al., 2021;Liu et al., 2021), which led to a low transfection efficacy. To this end, here we used ImageJ to quantify the average size of each syncytium. In line with the qualification analysis, trametinib exerted dose-dependent inhibition on the LUJV GPC-mediated membrane fusion (Figures 2B,E). Neither manidipine nor lercanidipine inhibited this effect (Figures 2C,D,F,G).

Mutation C410A Confers Resistance to Trametinib
As most arenavirus entry inhibitors target the SSP-GP2 interface to prevent membrane fusion, we further investigated the viral targets of the hit compounds. We selected the adaptive mutant virus by serially passaging the replication-competed with LUJVrv, in the presence of the hit compounds. The starting concentration was 10 μM for trametinib and manidipine and 20 μM for lercanidipine, corresponding to the IC 90 values obtained in the LUJVrv inhibition assay ( Figure 1C). Parallel passaging of LUJVrv was also conducted in DMSO as a control. In the trametinib-treated group, resistance was detected after three rounds of passaging, which increased in passage P4 and then remained stable through passages P5-P6 ( Figure 3A). To identify the viral genetic determinant(s) responsible for the resistance, we sequenced the trametinib-resistant virus and identified the cysteine (C)-to-glycine (G) mutation at amino acid position 410, which is located at the C terminus of the transmembrane domain (TM) on GP2 (Figure 3B). To confirm that the C410G mutation conferred trametinib resistance, we introduced a mutation into the GPC gene and generated a pseudotype mutant virus. Compared with the wild-type (WT) LUJVpv, the C410G mutant virus exhibited robust resistance to trametinib, but remained sensitive to manidipine and lercanidipine ( Figure 3C). Resistance was confirmed in the membrane fusion assay by the obvious formation of syncytium in the GPC C410G group, when treated with trametinib. Meanwhile, syncytium formation was robustly inhibited in WT ( Figure 3D). The cysteine residues located within the border region between the transmembrane domain and the cytoplasmic tail of glycoprotein were usually found to be palmitoylated (Veit, 2012). Furthermore, LUJV GPC possessed the motif C 410 IC, which was reported to be palmitoylated in Ebola virus (EBOV) and Marburg virus (MARV) glycoprotein (Ito et al., 2001). Thus, we hypothesize that both C410 and C412 might serve as potential palmitoylated sites in LUJV GPC and that palmitic acid might be the target of trametinib. To this end, C412 was substituted with G, and its drug sensitivity/resistance was evaluated. As shown in Figure 4A, both the single C412G mutant and the combined C410G/C412G mutant still underwent membrane fusion with acid triggering, similar to that of C410G ( Figure 3D). When treated with trametinib, fusogenicity was inhibited in the C412G group, but not in the C410G/C412G group, suggesting that only C410 served as the viral target of trametinib. Furthermore, LUJV C412G pv was still as sensitive to trametinib as that of WT, whereas LUJV C410G/C412G pv was resistant to trametinib, similar to LUJV C410G pv ( Figure 4B). This confirms that only C410G contributed to the resistance to trametinib.
Notably, we attempted to identify the resistance of LUJVrv to manidipine and lercanidipine. However, after 12 rounds of passaging with increasing concentrations of either compound (10-50 μM for manidipine, 20-50 μM for lercanidipine), the drug-resistant phenotype was absent in both groups, and no adaptive mutation was obtained for either compound. Combined with the finding that manidipine and lercanidipine had little effect on GPC-mediated membrane fusion, we thus speculate that manidipine and lercanidipine inhibited virus entry by targeting cellular compounds, rather than viral proteins.

Effects of Hit Compounds Against Other Mammarenaviruses
We further tested whether the hit compounds exhibited their inhibitory effects against the entry of other pseudotypes of other pathogenic mammarenaviruses, such as LASV, LCMV, MOPV, JUNV, MACV, GTOV, SBAV, and CHAPV. Intriguingly, although C410 was found to be unique among the mammarenaviruses (Figure 3B), trametinib inhibited all of the tested pseudotypes of mammarenaviruses, with IC 50 values ranging from 0.756 to 10.36 μM (Figures 5A-H). We used the BSL-2 level-compatible LCMV Cl-13 strain to test the antiviral effects of trametinib against an authentic mammarenavirus. As shown in Figure 5I, trametinib inhibited the authentic LCMV infection, with an IC 50 value of 3.919 μM, similar to that of LCMVpv (4.822 μM, Figure 5H).
Trametinib has been reported to repress the expression of transferrin 1 (Horniblow et al., 2017), which might result in the inhibition of NW mammarenavirus entry (Figures 5A-E).
For the other OW mammarenaviruses that do not possess the corresponding C410, we investigated the viral target by serially passaging LASVrv in the presence of 10 μM trametinib, corresponding to the IC 85 value obtained in the LASVpv inhibition assay ( Figure 5F). As shown in Figure 6A, resistance emerged after two rounds of passaging. Sequencing of trametinibresistant LASVrv revealed that the viral target was F446L, which is located in the TM of GP2 ( Figure 6B). F446 is A B FIGURE 4 | Only C410G, but not C412G, conferred resistance to trametinib. (A) Both C410G and C410G/C412G mutants conferred resistance to the inhibition of trametinib, against LUJV GPC-mediated membrane fusion. Images are representative fields from three independent experiments. (B) Both C410G and C410G/C412G mutants conferred the resistance to trametinib, while both WT and C412G retained their sensitivity. Data are means ± SD from 3 to 4 independent experiments. The corresponding F446L mutant LCMVpv was still sensitive to trametinib. Data are presented as means ± SD from 3 to 4 independent experiments ( *** p < 0.001; ** p < 0.01; and * p < 0.05).
highly conserved in mammals and has been reported to confer resistance to other structurally distinct membrane fusion inhibitors . The sensitivity/resistance was further tested in the presence of trametinib by introducing the F446L mutant into LASVpv, MOPVpv, and LCMVpv. As shown in Figures 6C-E, LASV F446L pv, MOPVpv, and LCMVpv (containing the corresponding F446L mutant) conferred resistance to trametinib, confirming that F446 served as the viral target of trametinib in LASV, MOPV, and LCMV.
We also investigated the broad-spectrum antiviral effects of manidipine and lercanidipine. Both showed robust inhibition against NW mammarenavirus entry, with IC 50 values ranging from ~0.1 to 1 μM. At the tested concentration, the doseresponse curves of both manidipine and lercanidipine for NW viruses showed the typical "s" model, with 2-log spans (Figures 7A-E,J-N). These results are in line with published reports, in which voltage-gated calcium channel (VGCC) antagonists were found to be critical for the cellular binding Frontiers in Microbiology | www.frontiersin.org and entry of the NW arenaviruses JUNV and Tacaribe virus (Lavanya et al., 2013;Sarute and Ross, 2020). Notably, although manidipine and lercanidipine exerted ambiguous inhibition against OW pseudoviruses (Figures 7F-H,O-Q), both showed promising inhibition against authentic LCMV infections, with IC 50 values of 2.344 μM for manidipine ( Figure 7I) and 1.093 μM for lercanidipine (Figure 7R).

Manidipine and Lercanidipine Inhibit LUJV Entry by Acting as Calcium Inhibitors
Both manidipine and lercanidipine are DHP VGCC antagonists. VGCC has been reported to be indispensable for the entry of NW mammarenaviruses (Lavanya et al., 2013;Sarute and Ross, 2020). To this end, we tested the role of siRNAs in depleting the expression of VGCC genes in U-2 OS cells to determine their effect on LUJV entry. U-2 OS cells have been reported to express VGCC genes (CACNA1S and CACNA2D2; Lavanya et al., 2013;Sarute and Ross, 2020). First, siRNA knockdown efficiency was validated using a WB assay. As shown in Figures 8A,B, the siRNA knockdown of CACNA1S, encoding the VGCC pore-forming α1 subunit targeted by most DHPs. This decreased the protein levels of both CACNA1S and CACNA2D2, whereas the knockdown of CACNA2DA decreased CACNA2D2 protein levels. Furthermore, the siRNA knockdown of CACNA1S effectively blocked infection by MACVpv, but not by VSVpv. This finding is in line with previous reports that VGCCs are essential for NW mammarenavirus entry (Lavanya et al., 2013;Sarute and Ross, 2020). Similarly, LUJV infection was inhibited following the knockdown by siRNA of either CACNA1S or CACNA2D2, indicating that VGCC is critical for LUJV entry (Figures 8C,D).

DISCUSSION
Here, we screened an FDA-approved drug library and identified three hit compounds, trametinib, manidipine, and lercanidipine, which prohibited the entry step of pseudotype and recombinant LUJV infections. Among these three hits, trametinib exerted the most effective inhibition, with an IC 50 of 0.3566 μM and an SI of >560. Trametinib, which is an MAPK inhibitor, is used for to treat patients with unresectable or metastatic melanoma carrying the BRAF V600E mutation. Trametinib has also been shown to play a role as an anti-coronavirus After 6 h, the duplicate cell lysates were subjected to qPCR (C) and Rluc assays (D), respectively. Data are presented as presented as means ± SD from more than five independent experiments (****p < 0.0001; ***p < 0.001; **p < 0.01; and *p < 0.05).
agent that can inhibit the Middle East respiratory syndrome coronavirus infection by modulating the MAPK/ERK signaling pathway (Kindrachuk et al., 2015). It has been shown to inhibit severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) by downregulating the expression of ACE2 (Zhou et al., 2020). Intriguingly, investigating the antiviral mechanism of trametinib revealed that it could robustly block LUJV GPC-mediated membrane fusion and its viral target was C410, which was embedded in the TM domain of GP2. In mammarenavirus GPC, the retained SSP interacts with the membrane-proximal external region, as well as the TM domain of GP2. Thus, it stabilizes the prefusion conformation of GPC and provides an "Achilles' heel" that can be targeted by distinct entry inhibitors (Larson et al., 2008;Lee et al., 2008;Shankar et al., 2016;Wang et al., 2018;Zhang et al., 2019;Tang et al., 2020;Liu et al., 2021). To our knowledge, this is the first study to report entry inhibitors targeting the LUJV SSP-GP2 interface and to identify their viral target(s). Notably, trametinib extended its antiviral spectrum to the pseudotypes of other pathogenic mammarenavirus infections. We demonstrated that trametinib inhibited LASVpv infection by targeting F446 (embedded in the SSP-GP2 interface). Trametinib inhibited both the pseudotype and authentic LCMV infection with similar IC 50 values. For NW mammarenaviruses, we hypothesize that inhibition might occur due to a reduction in the surface expression of the NW mammarenavirus receptor TfR1 (Horniblow et al., 2017).
Besides trametinib, the other two hits, manidipine and lercanidipine, were VGCC inhibitors. VGCC inhibitors have been widely reported to inhibit different types of viruses with various antiviral mechanisms (Chen et al., 2019). Notably, VGCC has been reported to play essential roles in NW mammarenavirus infection (Lavanya et al., 2013;Sarute and Ross, 2020). This is in line with our results, which showed that both manidipine and lercanidipine robustly inhibited NW mammarenavirus entry, with IC 50 values ranging from 0.1-1 μM. Lacidipine, which is a VGCC inhibitor, has been demonstrated to inhibit LASVpv and GTOVpv infections, as well as GPC-mediated membrane fusion, by targeting LASV T40 (in the ectodomain of SSP), GTOV V36 (in the ectodomain of SSP), and V436 (in TM domain of GP2; Wang et al., 2018). To investigate whether manidipine and lercanidipine inhibit LUJV entry by acting as calcium inhibitors or fusion inhibitors, we studied the effects of both drugs on LUJV GPC-mediated membrane fusion and neither showed promising inhibition of fusion. We tried to select the adaptive mutant by serially passaging with increasing concentrations of both drugs (10-50 μM), but no mutant was observed in GPC. Notably, after reviewing all 29 calcium inhibitors included in the current FDA drug library, we found that besides manidipine and lercanidipine, 12 additional calcium channel blockers (10 μM) inhibited LUJVrv infection to >90% inhibition, suggesting that calcium channels are potential antiviral targets. By using siRNA to knockdown VGCC genes, we found VGCC to be essential for LUJV entry. Based on these results, the calcium channel might serve as a potential antiviral target in LUJV treatment. Manidipine and lercanidipine inhibited LUJVpv entry by acting as calcium inhibitors.
The fourth hit compound, ospemifene, is a selective estrogen receptor modulator (SERM). It robustly inhibited LUJV entry in a dose-dependent manner. Notably, during the primary screen all nine SERMs [ospemifene, bazedoxifene HCl, bazedoxifene acetate, clomifene citrate, chlorotrianisene, raloxifene (Evista), tamoxifen, tamoxifen citrate, and toremifene citrate] showed effective inhibition (75-99% inhibition) on LUJVrv infection. However, none of these nine drugs were researched further because of either their cytotoxicity or their inhibition of VSVpv. Notably, a recently published study showed that four SERMs [bazedoxifene HCl, raloxifene (Evista), tamoxifen citrate, and toremifene citrate] could inhibit LUJV infection (Welch et al., 2021). Similar to our results, these SERMs showed cytotoxicity, with relatively low CC 50 values (5-10 μM; Welch et al., 2021). Although their associated cytotoxicity limits the use of SERMs in antiviral treatment, their promising inhibitory effects suggest that their action mechanisms, that is, the activation or repression of the estrogen target genes (Riggs and Hartmann, 2003), might be involved in LUJV infection. Furthermore, the core structures of SERMs might serve as a backbone for structure-activity relationship optimization for drug development.
To date, only one LUJV outbreak has recorded, in 2008, which resulted in a fatality rate of 80% (4/5 cases; Briese et al., 2009). There is an urgent need to develop therapeutic options to treat LUJV and other highly pathogenic mammarenavirus infections. Here, we screened an FDA-approved drug library to develop novel therapeutic options for LUJV treatment. Revealing the mechanisms of action of these antiviral inhibitors can facilitate the understanding of LUJV pathogenesis.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
WW, JC, and SD conceived and designed the experiments. JC, SD, YL, MZ, JG, XJ, YZ, YH, and MT performed the experiments. WW, JC, SD, and YL analyzed the data. WW and JC wrote the paper. WW and GX supervised the study. All authors contributed to the article and approved the submitted version.