Discovery of three small-molecule inhibitors targeting Ebolavirus genome replication and transcription

Abstract

The 2013 Ebola virus (EBOV) outbreak was the largest in history. Despite recent advances in both vaccines and monoclonal antibody therapies, 12 years later, EBOV still poses a substantial threat. Previously, we published a ligand discovery pipeline combining in silico screening of compounds with a robust and rapid EBOV minigenome assay for early-stage inhibitor validation at Biological Safety Level 2. Here, we present the further use of this pipeline to identify three compounds that also inhibit EBOV minigenome transcription and replication. They are efficacious in the nM range, exhibited low cytotoxicity and were specific, with no effect on either a T7 RNA polymerase-driven firefly luciferase or a Bunyamwera virus minigenome. Furthermore, these small-molecule inhibitors exhibited the ability to block EBOV minigenome activity when applied after establishment of replication complexes, with implications for potential post-exposure EBOV treatment.

Data Availability

The datasets generated during and/or analysed as part of this study are available from the corresponding author on reasonable request.

Introduction

From the initial identification of Ebola virus (EBOV) in 1976, there have been 52 outbreaks leading to 34,945 cases, resulting in 15,407 deaths in 11 Sub-Saharan African countries [1]. The high pathogenicity, ease of transmission via bodily fluids [2] and the rapid infection progression have resulted in its classification as a Biosafety Level 4 (BSL4) pathogen, hampering development of effective therapies. As a result, until 2020, there were no licensed treatments for patients suffering from EBOV disease (EVD), but there are now two treatments approved by the US Food and Drug Administration to treat EVD in adults and children. Inmazeb^™ (REGN-EB3) was approved in October 2020 and is a combination of three mAbs [3]. Ebanga^™ (mAb114) is a single monoclonal antibody and was approved in December 2020 [4]. All of these mAbs bind to the surface glycoprotein, preventing virus entry. However, both Inmazeb^™ (REGN-EB3) and Ebanga^™ (mAb114) only reduce the mortality rate to 33.5 and 35.1%, from a control group mortality of 51.3 and 49.7%, respectively [5]. These mAbs also have little effect on viral clearance times compared to standard supportive care [6], demonstrating that there is still a need to develop effective small-molecule therapeutics.

Small molecule-based medicines are often stable at room temperature for long periods and compatible with oral administration [7], unlike the mAb-based treatments, and as such could be made quickly and widely available as a post-exposure prophylactic or as a treatment for symptomatic patients.

The Ebolavirus genus, together with Cuevavirus and Marburgvirus, is classified within the Filoviridae family (the order Mononegavirales) [8]. There are six species within the Ebolavirus genus: Bundibugyo (BDBV), Sudan (SUDV), Taï Forest (TAFV) and Zaire (previously known as ZEBOV), the type species now referred to as EBOV, are able to cause EVD in humans, while Reston (RESTV) and Bombali virus (BOMV) are not able to cause EVD [9,10].

EBOV is a filamentous enveloped virus with a non-segmented, negative-sense ssRNA genome of ~19 kb [8]. The genome encodes seven proteins: a nucleoprotein (NP), a glycoprotein, four viral proteins (VP24, VP30, VP35 and VP40) and the L protein (RNA-dependent RNA polymerase) [11]. The NP forms a complex with VP35, VP30 and L which is essential for genome replication and transcription [12,14]. This complex is the basis for the EBOV minigenome (MG) system [11], in which plasmids expressing these four proteins under the control of a T7 promoter are transfected into cells constitutively expressing T7 RNA polymerase, together with a plasmid with a T7 promoter driving production of an RNA containing the reverse complement of the reporter firefly luciferase (FF-luc) flanked by EBOV genome 5′ and 3′ terminal sequences. These are recognized by the replication complex, which transcribes the reporter and allows translation of FF-luc as an indirect measurement of EBOV-specific gene expression. Because no infectious virus can be produced, this system allows for the investigation of EBOV genome replication and transcription at BSL2.

The structure of the NP and the interactions with VP35 have been characterized for EBOV [15,16] and Marburg virus (MARV) [14]. A hydrophobic pocket on EBOV NP comprising three ‘hotspots’ for protein–protein interactions (Fig. S1A, available in the online Supplementary Material) binds either with an NP binding peptide of VP35 (NPBP, residues 20–48) which maintains NP as a monomer, or with a flexible arm of other NP molecules (helix-20) to oligomerize. The two binding states thus control the binding and release of RNA and oligomerization – essential to viral replication [17]. We have previously reported a small-molecule inhibitor (SMI) termed MCCB4 [18] which putatively bound to hotspots 1 and 2 within the hydrophobic pocket of NP and reduced EBOV-specific gene expression with an EC₅₀ value of 4.8 µM.

Studies with other negative-strand viruses have also shown that NP is a valid target for SMIs, for example, the influenza inhibitor Nucleozin, which triggers aggregation of NP [19,20], and inhibitors which promote influenza NP oligomerization [19]. Another reason why NP is an attractive antiviral target is the high level of conservation of the VP35 binding pocket between EBOV and the related Marburgviruses [14].

The MG system has been used previously to identify SMIs of EBOV replication [21,24]; however, these studies have involved high-throughput screens of pre-existing libraries of known bioactive compounds. Here, we used the previously defined pipeline [18] to identify three more SMIs predicted to bind to the NP hydrophobic pocket. These compounds exhibited a range of EC₅₀ values from 4.4 to 0.198 µM and were characterized at a variety of time points, in multiple cell lines and through binding site mutations.

Methods

Cell lines and plasmids

BSR-T7 cells are derived from BHK-21 cells and stably express T7 RNA polymerase [25]. Huh7-Lunet-T7 cells are a derivative of Huh7 cells and also express T7 RNA polymerase [26]. Cells were grown in DMEM (Sigma-Aldrich), 10% FBS (Sigma-Aldrich), 1% non-essential amino acids (Lonza) and penicillin-streptomycin (100 units ml⁻¹) (Sigma) with either G418 (600 μg ml⁻¹) (Life Technologies) (BSR-T7) or Zeocin (Invitrogen) (5 μg ml⁻¹) (Huh7-Lunet-T7) added at every second passage. Cells were maintained at 37 °C with 5% CO₂.

The EBOV MG (Makona strain, GenBank KJ660347.2) [27] expressing FF-luc used in this study was kindly donated by Julian Hiscox (University of Liverpool) following the MG model developed by Mühlberger et al. [11] (see Fig. S1B). The control plasmid pT7FFLuc contains FF-luc under the direct control of a T7 promoter. The plasmids for the Bunyamwera virus (BUNV) MG express Renilla luciferase (R-luc) and were described previously [28,29].

Compounds

Commercially available small molecule compounds CB6 and MCCB5 were purchased from ChemBridge (https://www.hit2lead.com/) and Asinex ChemDiv (http://www.chemdiv.com/), respectively, and compound purity was assessed by LC-MS (Thermo Scientific Ultimate 3000 HPLC system equipped with a Bruker amaZon Speed mass spectrometer). Samples were run in positive mode over a 2 min gradient at a column temperature of 40 °C with mobile phases of 0.1% formic acid in water and 0.1% formic acid in MeCN, and a flow rate of 1.334 ml min⁻¹. Both compounds were >95% pure. Compound MJM364 was synthesized in-house, and details of the chemical synthesis are available on request.

Structure-based virtual screening and molecular design

Structures of the EBOV NP (PDB 4YPI and 4Z9P) and MARV NP (PDB 5XSQ) were accessed from the Protein Data Bank and imported into the Maestro interface (version 10.3, Schrodinger LLC, New York, NY, 2018 Schrodinger). All NP structures were subjected to the Protein Preparation Wizard (default settings), and the outputs were used directly for the virtual screening process as described previously [18]. Excel files of the docking triage process are available on request. The de novo molecular design software SPROUT [30] was used to build (atom by atom) putative inhibitors of the hotspots within the EBOV NP hydrophobic pocket. The obtained molecular designs were visualized using the PyMOL Molecular Graphics System (Version 1.8 Schrödinger, LLC).

EBOV and BUNV MG and pT7FFLuc transfections

BSR-T7 cells (5×10⁴ cells per well) were seeded in a 48-well plate and allowed to adhere overnight. Per well, cells were transfected with EBOV MG (0.25 µg), NP (0.125 µg), VP30 (0.62 µg), VP35 (0.62 µg) and L (0.62 µg); pT7FFLuc (0.15 µg); or BUNV(S) REN (0.2 µg), L-sup (0.05 µg) and S-sup (0.05 µg), as described [18]. Transfections were performed with Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Compounds for testing were made to the required concentrations in serum-free DMEM directly before use and added to cells immediately before transfection or at defined points during the experiment.

Luciferase and MTT assays

Cells were harvested at 24 h post-transfection (hpt) (unless otherwise specified) by washing in PBS then lysing in Passive Lysis Buffer (Promega). Samples were read using the FF-luc or R-luc reporter assay systems (Promega) and a FLUOStar Optima microplate reader. The dose–response curves used to calculate EC₅₀ and CC₅₀ values are presented as percentages of in-plate DMSO-only transfection controls. These data were generated from three independent experiments performed in triplicate. MTT assays were performed alongside luciferase assays. After washing the cells in PBS, MTT reagent (1 mg ml⁻¹ in H₂O) (Alfa Aesar) was added for 1 h. DMSO was used to lyse cells prior to being read on a Tecan Infinite F50 microplate reader. All plates contained a transfection-only control and a cell-only control to show 100% luciferase signal and background signal, respectively.

Site-directed mutagenesis

Mutagenesis was performed on the NP plasmid using a Q5 Site-Directed Mutagenesis Kit (NEB) to generate four NP mutants (V250A, K274A, L280A and F284A). Primers were designed using NEBaseChanger^™ (sequences available on request) and contained silent restriction sites for identification.

Results

In silico identification of SMIs of NP-VP35 interaction

Our previous virtual screening study identified a number of chemically diverse putative inhibitors of EBOV NP, including the previously described compound MCCB4 [18]. Here, we present data obtained for three further compounds that showed promising biological activity in the initial screen against the EBOV MG at 100 µM. Of those, compounds MCCB5 and CB6 were prioritized based on a low host cell toxicity and therefore favourable selectivity index (Table 1). Compound MJM364 was also taken forward, identified via de novo molecular design and then chemically synthesized at the University of Leeds. The structures and predicted docking poses of these three compounds are shown in Fig. 1.

Table 1. Summary of EC₅₀ and CC₅₀ data.

Compound	Target	Cell	EC50	CC50	SI*
MCCB5	WT	BSR-T7	4.4†‡	>100	22.7
MCCB5	WT	Huh7-Lunet-T7	10.4†	80	7.6
MCCb5	F280A	BSR-T7	13.8	nd	nd
MCCB5	L284A	BSR-T7	15.0	nd	nd
MCCB5	V250A	BSR-T7	15.4	nd	nd
CB6	WT	BSR-T7	0.19†‡	>100	526
CB6	WT	Huh7-Lunet-T7	0.41†	>100	222
CB6	F280A	BSR-T7	2.5	nd	nd
CB6	L284A	BSR-T7	2.6	nd	nd
CB6	V250A	BSR-T7	31	nd	nd
MJM364	WT	BSR-T7	0.84†‡	>100	119
MJM364	WT	Huh7-Lunet-T7	1.2†	18.8	15.6
MJM364	F280A	BSR-T7	8.6	nd	nd
MJM364	L284A	BSR-T7	8.6	nd	nd
MJM364	V250A	BSR-T7	6.9	nd	nd

*Selective index (CC₅₀/EC₅₀).

†Unpaired two-tailed t-test of significance between EBOV and T7FFLuc control. ****P<0.0001.

‡Unpaired two-tailed t-test of significance between EBOV and BUNV control. ****P<0.0001.

nd, not determined.

Fig. 1. Identification of SMIs. Chemical structures of MCCB5, CB6 and MJM364 and predicted docking poses. Three hydrophobic hotspots were identified as important for the NP:VP35 interaction. Compounds were screened in silico using the eHiTs exhaustive docking engine, AutoDock cluster analysis and pose prediction software or de novo designed (see also Fig. S1A, available in the online Supplementary Material).

MCCB5, CB6 and MJM364 inhibit EBOV genome replication and transcription

As previously reported [18], the EBOV MG system was used to assess whether MCCB5, CB6 and MJM364 were able to inhibit EBOV genome replication and transcription. The BUNV MG system and a plasmid in which FF-luc expression was directly driven by a T7 promoter were used as controls.

BSR-T7 cells were transfected and treated with MCCB5, CB6 or MJM364 over a 10⁵-fold range of concentrations from 1 nM to 100 µM. MCCB5, CB6 and MJM364 all exhibited a dose-dependent inhibition of the EBOV MG with EC₅₀ values of 4.4, 0.194 and 0.843 µM, respectively (Fig. 2). EC₅₀ values for both the BUNV MG and the T7FFLuc controls were significantly higher (Table 1; ****P<0.0001), confirming specificity for inhibition of EBOV replication. Cytotoxicity was assessed in cells transfected with the various combinations of plasmids to account for any additional toxicity due to expression of exogenous viral proteins. None of the three compounds exhibited any significant toxicity (CC₅₀ values>100 µM; Fig. 2), with the exception of MCCB5 in cells transfected with the BUNV MG, where the CC₅₀ value was 89 µM.

Fig. 2. Activity of MCCB5, CB6 and MJM364. Compounds were tested for activity against the T7FFLuc control, the BUNV or the EBOV MG. The compounds were added to BSR-T7 cells at the indicated concentrations immediately before transfection. Cells were harvested at 24 hpt and analysed for FF-luc (EBOV MG and T7FFLuc control) or R-luc (BUNV MG), or cell viability by MTT assay. Values are represented as percentages of a transfection-only control. Graphs show dose–response curves for % RLU (left) and cytotoxicity (right). (a) MCCB5, (b) CB6 and (c) MJM364. EBOV, red circles. T7FFLuc control, blue squares. BUNV control, black diamonds. Averages plotted from three independent experiments performed in triplicate. Error bars show sd.

The three compounds were then investigated further over a time course. BSR-T7 cells were transfected in the presence or absence of the EC₅₀ concentration of the compounds and harvested at the time points indicated. The three compounds inhibited the EBOV MG from 12 hpt and displayed a consistent 50% inhibition from 24 to 48 hpt (Fig. 3a). Between 0 and 6 hpt, the luciferase signal was indistinguishable from background levels apart from MCCB5, which demonstrated a modest inhibition at 6 hpt (Fig. 3a).

Fig. 3. Time-course analysis of MCCB5, CB6 and MJM364 activity. (a) Compounds were analysed at their relative EC₅₀ values. Samples were harvested at the indicated times hpt, and values are represented as percentages of a transfection-only control. (b) A time-of-addition study was undertaken at the relative EC₇₅ concentrations of the compounds. The top panel is a schematic demonstrating when the compounds were present during transfection. The graph shows the data as a percentage of a transfection-only DMSO control, harvested at the same time. All averages were plotted from three independent experiments performed in triplicate and compared using an unpaired two-tailed t-test. Error bars show sd. ****P<0.0001, ***P 0.001 and **P 0.01.

We also conducted a time-of-addition study to test whether the compounds exhibited inhibitory activity against pre-existing NP-VP35 replication complexes or could prevent the formation of such complexes. To test this, MCCB5, CB6 or MJM364 at the EC₇₅ concentrations (19.3 µM, 948 nM and 14.9 µM, respectively) were added to BSR-T7 cells at time points pre- (−24 to 0 hpt), during (0–48 hpt and 0–24 hpt) and post-transfection (24–48 hpt) (Fig. 3b, top panel), prior to harvest of all cells at 48 hpt. Notably, when the compounds were added from −24 to 0 hpt and then removed, there was a statistically significant inhibition of luciferase production (Fig. 3b, −24 to 0 hpt), but this did not match the reduction expected based on the EC₇₅ dose (Fig. 3b, dashed line, bottom panel). As anticipated, the addition of MCCB5, CB6 or MJM364 at transfection was able to disrupt EBOV replication complex formation and so genome replication. This was shown by a reduction in luciferase by between 68% and 71% (Fig. 3b, 0–48 hpt). The addition of MCCB5 post-transfection (Fig. 3b, 24–48 hpt) also reduced the luciferase signal by 64%, suggesting that MCCB5 was able to effectively compete for binding to NP in established replication complexes. However, CB6 and MJM364 were less effective. As with our previously published compound MCCB4, the addition of MCCB5, CB6 or MJM364 at transfection and subsequent removal (Fig. 3b, 0–24 hpt) resulted in an inhibition of luciferase production that was significantly different from the DMSO control but higher than that displayed by the sustained addition (0–48 hpt) and the expected signal based on the EC₇₅ dose. Taken together, these data are consistent with the ability of MCCB5 to inhibit both pre-existing and newly formed replication complexes, whereas CB6 and MJM364 were less effective at inhibiting pre-existing complexes.

As EBOV can infect and replicate in both fibroblasts and hepatocytes [31], we also tested these compounds in the human hepatocellular carcinoma cell line Huh7-Lunet-T7, which constitutively expresses T7 RNA polymerase [26]. The transfection efficacy of the EBOV MG and the T7FFLuc control in the different cell lines was similar (data not shown). Reassuringly, in Huh7-Lunet-T7 cells, all three compounds were effective (Fig. 4), although the EC₅₀ values were marginally higher than those observed in BSR-T7 cells (see Table 1 for comparison). For MCCB5 and MJM364, this coincided with a higher cytotoxicity, resulting in a lower SI. The reasons for the differences are unclear but may be due to differences in uptake or turnover of the compounds in the two cell types.

Fig. 4. MCCB5, CB6 and MJM364 activity in Huh7-Lunet-T7 cells. Compounds were tested for activity against the EBOV MG or T7FFLuc control. The compounds were added to Huh7-Lunet-T7 cells at the indicated concentrations immediately before transfection. Cells were harvested at 24 hpt and analysed for FF-luc, or cell viability by MTT assay. Values are represented as percentages of a transfection-only control. Graphs show dose–response curves for % RLU (left) and cytotoxicity (right). (a) MCCB5, (b) CB6 and (c) MJM364. EBOV, red circles. T7FFLuc control, blue squares. Averages were plotted from three independent experiments performed in triplicate. Error bars show sd.

Mutation of the NP hydrophobic pocket provides evidence that this is the protein target

The compounds were selected based on in silico predicted binding to the EBOV NP, but they had the potential to bind to alternative cavities on the NP protein and/or other molecular targets. In order to validate the predicted binding of MCCB5, CB6 or MJM364 to NP, the effect of specific mutants of NP on the inhibitory activity of the compounds was investigated. The residues V250, F280 and L284 within NP hotspot 2 (Fig. 5a), which were predicted to interact with the compounds, were therefore substituted with alanines (NP V250A, F280A and L284A). As expected, these mutants exhibited a tenfold reduction in replication and FF-luc expression compared to WT (Fig. S2) and therefore could be tested for the effects of the compounds. A fourth mutant in hotspot 3 (K274A), predicted to contact MJM364, was replication-defective and thus could not be tested further.

Fig. 5. Mutagenic analysis of NP confirms the interaction of MCCB5, CB6 and MJM365. (a) The EBOV NP hotspot residues: Val250, Lys274, Phe280 and Leu284 were mutated to alanine (V250A, K274A, F280A and L284A, respectively). The compounds were added to cells at the indicated concentrations immediately before transfection with either WT EBOV NP, mutant V250A, F280A or L284A. Cells were harvested at 24 hpt and analysed for FF-luc. (b–d) Cells were also treated with mycophenolic acid (MPA) at the previously defined [33] EC₉₀ concentration, and this is shown as a single point for WT and the three mutants. Dose–response curves for MCCB5 (b), CB6 (c) and MJM364 (d). Left-hand graphs show % RLU represented as % of a transfection-only control. Right-hand graphs show a comparison of EC₅₀ values generated from the WT or mutant NP transfections. EC₅₀ values shown are from three independent experiments performed in triplicate compared using an unpaired two-tailed t-test. Error bars show sd. ***P<0.001 and ****P<0.0001.

Cells were transfected with the EBOV MG plasmids including either WT or the 3 mutant NPs and treated with the 3 compounds over a 10,000-fold range of concentrations. As a control, cells were treated with MPA, an inosine monophosphate dehydrogenase inhibitor which inhibits replication of several RNA viruses [32], including EBOV [33]. MPA at the EC₉₀ concentration inhibited all three mutant and WT NPs equally.

The EC₅₀ values for all three mutants were significantly higher than those shown by the WT NP (Fig. 5b–d, Table 1), demonstrating that these mutations resulted in partial resistance to the compounds. This difference was more pronounced for CB6 and MJM364 compared to MCCB5, suggesting that CB6 and MJM364 were more dependent on these residues for binding into the hydrophobic pocket. Taken together, these data are consistent with the predicted binding of all three compounds into the hydrophobic pocket within NP.

Discussion

The availability of high-resolution structures of the EBOV NP protein used in conjugation with computer-aided drug design approaches has led to the identification of three molecules (MCCB5, CB6 and MJM364) that showed potent inhibition of EBOV MG replication, but no effect on an MG system derived from another negative-strand virus, BUNV, or against T7-driven transcription.

MCCB5, CB6 and MJM364 were shown to be efficacious in reducing the luciferase signal detected in EBOV MG-transfected BSR-T7 cells and Huh7-Lunet-T7 cells. These effects were specific against EBOV replication as shown by significantly different EC₅₀ values demonstrated by both the T7FFLuc control transfection and the BUNV MG transfection (Table 1, ****P<0.0001). All three compounds were well tolerated by cells, although MCCB5 and MJM364 appear to have a greater cytotoxicity in Huh7-Lunet-T7 cells. The MCCB5 and MJM364 reduction in luciferase signal presented by the T7FFLuc and BUNV MG control transfections tracks closely to cell viability and therefore most likely reflects a reduction in the viability of transfected cells.

We envision that if these compounds were taken forward for treatment of EBOV infection, it is likely that they would be used in combination with other treatment modalities such as the approved mAb therapies [Inmazeb^™ (REGN-EB3) [3] and Ebanga^™ (mAb114) [4]]. The mAbs are used for EBOV post-exposure prophylaxis, or the treatment of symptomatic patients, and we believe that the compounds described here could also function to reduce virus replication and allow the adaptive and innate immune systems time to overcome infection [9,34]. Although the three compounds were able to inhibit EBOV replication when applied at the time of transfection, MCCB5, like MCCB4 [18], was also able to effectively inhibit replication when added at 24 hpt. This is consistent with its ability to compete for NP:VP35 binding in pre-existing replication complexes, and making it a plausible candidate for use as a post-exposure drug. A caveat to this is that in the MG system, there will be continuous expression of VP30, VP35, L and NP, so even at 24 hpt, there will still be active synthesis of replication complexes. We cannot therefore formally exclude the possibility that MCCB5 is only interfering with the interaction between VP35 and NP during replication complex formation. CB6 and MJM364 also demonstrated a significant reduction compared to the DMSO control when added at 24 hpt, but the effect was lower than the expected EC₇₅ effect. Both of these compounds exhibit low cytotoxicity, so there is the possibility of increasing the dose of CB6 or MJM364 if used post-exposure.

To confirm that the compounds interact as predicted with EBOV NP at the targeted hydrophobic binding pocket, three alanine substitution mutants were generated (NP V250A, L280A and F284A) in this region. These changes were predicted to reduce binding efficiency.

There was a significant EC₅₀ shift between WT and mutants in response to all three compounds, which was consistent with the hypothesis that they bind to the NP as predicted by in silico screening. MCCB5 and MJM364 showed no difference between the V250A, L280A and F284A mutants, while CB6 demonstrated over a log increase in EC₅₀ difference between NP V250A and L280A or F284A. This suggested that CB6 may interact directly with V250. MJM364 was predicted to interact with hotspot 3, but the 2-log increase in EC₅₀ shown by the hotspot 2 mutants (V250A, L280A and F284A) strongly implied an interaction with hotspot 2. Given the fragment-like size of this molecule and the depth of hotspot 2, molecular docking of MJM364 within the hydrophobic groove of NP also favours binding into hotspot 2. Of note, MJM364 is a bespoke de novo designed compound which is of smaller molecular weight compared to MCCB5 and CB6. As such, there is the possibility to grow this fragment into lead-like space and pick up additional binding interactions with the NP hydrophobic pocket.

Importantly, the NP protein is essential for genome replication and is highly conserved throughout the filoviruses. In particular, the EBOV and MARV NP proteins exhibit a 40% overall similarity, rising to 95% for the residues located within the NP hotspots 1, 2 and 3. The SMIs described here are also predicted, in silico, to bind to the Marburg NP (Fig. S3), suggesting that this pipeline could lead to a potential pan-filovirus therapeutic agent. In this regard, the WHO has stated a desire for a multivalent filovirus vaccine (WHO, 2016), demonstrating an appetite for multivalent therapeutics specifically for EBOV Zaire, EBOV Sudan and MARV. Further development of the compounds described here could therefore involve the use of a MARV MG system [11] to establish pan-filovirus efficacy, followed by screening against infectious EBOV and MARV at BSL4.

Abbreviations

BSL4

biosafety level 4

BUNV

Bunyamwera virus

EBOV

Ebola virus

EVD

EBOV disease

FF-luc

firefly luciferase

hpt

hour post-transfection

MARV

Marburg virus

MG

minigenome

MPA

mycophenolic acid

NP

nucleoprotein

R-luc

Renilla luciferase

SMI

small-molecule inhibitor