Comparison of Zaire and Bundibugyo Ebolavirus Polymerase Complexes and Susceptibility to Antivirals through a Newly Developed Bundibugyo Minigenome System

ABSTRACT

Members of the genus Ebolavirus cause lethal disease in humans, with Zaire ebolavirus (EBOV) being the most pathogenic (up to 90% morality) and Bundibugyo ebolavirus (BDBV) the least pathogenic (∼37% mortality). Historically, there has been a lack of research on BDBV, and there is no means to study BDBV outside of a high-containment laboratory. Here, we describe a minigenome replication system to study BDBV transcription and compare the efficacy of small-molecule inhibitors between EBOV and BDBV. Using this system, we examined the ability of the polymerase complex proteins from EBOV and BDBV to interact and form a functional unit as well as the impact of the genomic untranslated ends, known to contain important signals for transcription (3′-untranslated region) and replication (5′-untranslated region). Various levels of compatibility were observed between proteins of the polymerase complex from each ebolavirus, resulting in differences in genome transcription efficiency. Most pronounced was the effect of the nucleoprotein and the 3′-untranslated region. These data suggest that there are intrinsic specificities in the polymerase complex and untranslated signaling regions that could offer insight regarding observed pathogenic differences. Further adding to the differences in the polymerase complexes, posttransfection/infection treatment with the compound remdesivir (GS-5734) showed a greater inhibitory effect against BDBV than EBOV. The delayed growth kinetics of BDBV and the greater susceptibility to polymerase inhibitors indicate that disruption of the polymerase complex is a viable target for therapeutics.

IMPORTANCE Ebolavirus disease is a viral infection and is fatal in 25 to 90% of cases, depending on the viral species and the amount of supportive care available. Two species have caused outbreaks in the Democratic Republic of the Congo, Zaire ebolavirus (EBOV) and Bundibugyo ebolavirus (BDBV). Pathogenesis and clinical outcome differ between these two species, but there is still limited information regarding the viral mechanism for these differences. Previous studies suggested that BDBV replicates slower than EBOV, but it is unknown if this is due to differences in the polymerase complex and its role in transcription and replication. This study details the construction of a minigenome replication system that can be used in a biosafety level 2 laboratory. This system will be important for studying the polymerase complex of BDBV and comparing it with other filoviruses and can be used as a tool for screening inhibitors of viral growth.

INTRODUCTION

The family Filoviridae consists of the genera Ebolavirus, Marburgvirus, Cuevavirus, Striavirus, and Thamnovirus (1). There are six different species of ebolaviruses: Zaire ebolavirus (EBOV), Sudan ebolavirus (SUDV), Bundibugyo ebolavirus (BDBV), Tai Forest ebolavirus (TAFV), Reston ebolavirus (RESTV), and the newly discovered Bombali ebolavirus (BOMV) (2). With the exception of RESTV and BOMV, ebolaviruses are known to cause severe disease in humans, and EBOV, SUDV, and BDBV all have caused outbreaks of hemorrhagic fever across Africa. The largest ebolavirus outbreak recorded, caused by EBOV, occurred in West Africa from 2013 to 2016 and resulted in over 28,000 cases and a case fatality rate (CFR) of 63% for confirmed cases (3). BDBV, the most recently emerged ebolavirus, is responsible for two outbreaks, one in Bundibugyo, Uganda, in 2007 and one in the DRC in 2012. The BDBV outbreaks had an average CFR of 37% compared to 55% for SUDV and 76% for EBOV outbreaks in central Africa (4,5). Importantly, EBOV, SUDV, and BDBV are all endemic to this region at the border of the DRC and Uganda, indicating a need to fully understand the virology and pathology of these three species in order to gain insight for the development of effective medical countermeasures.

Filoviruses are nonsegmented, negative-stranded RNA viruses that replicate using a combination of packaged viral proteins and host cell proteins. The genome of BDBV, as with the other ebolaviruses, contains seven structural proteins, four of which make up the polymerase complex. These proteins are the nucleoprotein (NP), viral protein 35 (VP35), viral protein 30 (VP30), and the catalytic subunit of the polymerase complex encoded by the large gene (L). NP encapsidates the viral RNA, VP35 acts as a polymerase cofactor and a bridge between NP and L, and VP30 is a transcriptional coactivator (reviewed in references 6 and 7). Interactions between these proteins, specifically between NP and VP35 (8, 9) and between VP35 and L (10), are vital to the proper function of the polymerase. VP30 is only necessary for transcription and is not required for genomic replication (11, 12). The efficiency of the polymerase complex is an important component to determining viral replication rates as transcription and genome replication creates not only new virion components but also proteins involved in evading the immune response.

The genomic untranslated regions (UTR), the 3′-leader and 5′-trailer, are also implicated in the regulation of transcription and replication. There is evidence suggesting that the terminal nucleotides in the 3′-UTR are specific to the genera for initiation of transcription, as the EBOV and Lloviu virus (LLOV) polymerase complexes cannot transcribe chimeric minigenomes with the terminal nucleotides from Marburg virus (MARV) (13, 14). In addition, EBOV cannot transcribe minigenomes containing the 3′-UTR from MARV, although MARV can transcribe in the presence of the EBOV 3′-UTR (12). The ability of the polymerase complex and host proteins to interact with the 5′-UTR is necessary for antigenome to genome replication (7, 15, 16).

In vitro and in vivo data examining BDBV infection indicates that the viral kinetics and pathology are distinct from EBOV and SUDV. BDBV replicates slower and to lower peak titers compared to EBOV (17). In addition, nonhuman primates (NHPs) are less susceptible to BDBV infection compared to EBOV with a lower lethality rate observed during BDBV infection (18–21). A slower disease course is observed in human cases, with a delayed time to death in patients infected with BDBV compared to EBOV (22, 23). The prolonged disease course is mimicked in animal models, including NHPs (20, 21) and ferrets (24, 25). The characteristic cytokine storm induced during infection with EBOV has not been seen to the same extent during infection with BDBV and suggests that the course of infection and disease progression may be distinct from that of EBOV (17, 22). The reason for the differences in disease course and growth kinetics between EBOV and BDBV are unclear and could be due to underlying variations in the efficiency of the polymerase complex.

Research involving replication-competent filoviruses is limited to high-containment biosafety level 4 (BSL4) laboratories due to their classification as risk group 4/National Institute of Allergy and Infectious Diseases (NIAID) Category A pathogens. These classifications are based on the high mortality rate, the ease of spread, and the need for specialized public health preparedness measures. This limitation minimizes not only the number of individuals who can study the viruses but also the amount of work that can take place, as the BSL-4 setting is not conducive to high-throughput experimentation. In order to accelerate research on these pathogens, surrogate systems must be developed that allow for less restrictive use.

One system that has become popular for studying these highly lethal viruses is the minigenome replication system (26). This system relies on a minigenome made up of a reporter gene flanked by the 3′ and 5′ genomic ends from the virus of interest that provide the signaling for gene transcription and genomic replication. The replication system consists of a plasmid encoding the minigenome along with the support plasmids encoding the proteins necessary for transcription and replication; in the case of ebolaviruses, these are NP, VP35, VP30, and L. Since the minigenome replication system does not contain a full-length genome and cannot produce infectious particles, it is safe for use in BSL-2 laboratories. Currently, minigenome replication systems are available for the ebolaviruses EBOV and RESTV as well as the related MARV and LLOV (EBOV [12], RESTV [27], MARV [28], and LLOV [14]). The work described here details the construction and optimization of a minigenome replication system for BDBV and investigates the ability of BDBV polymerase complex proteins to functionally interact with EBOV proteins. The minigenome was further used to screen potential polymerase inhibitors and compare the efficacy of the inhibitor remdesivir against viral infection. Exploration into similarities and differences of the polymerase complexes will provide information regarding potential therapeutic targets and potentially explain differences in viral growth kinetics that are a possible mechanism for the observed contrasts in disease course.

RESULTS

BDBV infection results in reduced titers and intracellular genomic RNA compared to EBOV infection.

A single report has demonstrated that BDBV grows slower and to lower titers than EBOV in vitro in a mixed peripheral blood mononuclear cell (PBMC) population (17). We wanted to confirm this finding in two additional cell types representing the primary and secondary target cells. THP-1 cells, a protomonocytic cell line, were differentiated to macrophages to represent the primary target cells, and the HepG2 liver-derived cell line was used to represent secondary target cells. Delayed growth kinetics and lower titers for BDBV were observed in both cell types when analyzed by plaque assay (Fig. 1A and B). In both cell types, BDBV grew at least a log lower than EBOV and reached peak titers 24 h after EBOV. In the THP-1-derived macrophages, EBOV reached a peak titer of 6.4 × 10⁵ PFU/ml by 72 h postinfection (hpi), while BDBV reached a peak titer of 3.3 × 10⁴ PFU/ml by 96 hpi. In the HepG2 cells, EBOV reached a titer of 1.2 × 10⁷ PFU/ml at 48 hpi with a slight increase to 1.5 × 10⁷ PFU/ml by 72 hpi, while BDBV reached a peak titer of 1.6 × 10⁶ PFU/ml by 72 hpi. In addition, intracellular genomic RNA was measured and, although the input titers were higher for EBOV in the THP-1 macrophages, the amount of genomic RNA was similar for both viruses (Fig. 1C). Over the course of infection, lower quantities of genomic RNA were detected in the THP-1 macrophages (Fig. 1C). In the HepG2 cell line, more BDBV than EBOV genomic RNA was detected at 1 hpi, but a similar number of copies was detected at later time points, showing an increased rate of production for EBOV over BDBV (Fig. 1D).

FIG 1.

Growth kinetics in primary and secondary target cell types. Viral titer of EBOV (red) and BDBV (blue) over time at an MOI of 1.0 in THP-1 cells differentiated to macrophages (A) and HepG2 cell line (B). Peak titer indicated to the right of each curve. Copies of genomic RNA in cell lysates shown as genome equivalents (GE) per ng of RNA in THP-1 differentiated macrophages (C) and the HepG2 cell line (D). All experiments were performed in triplicate and are graphed as means ± standard deviations.

Generation of a BDBV minigenome replication system.

In order to examine whether there are differences in the ability of the polymerase complex to carry out RNA synthesis, a potential mechanism for decreased viral titers, we set out to establish a BDBV minigenome replication system. PCR amplicons were produced from RNA from BDBV, strain 200706291, isolated from a fatal human case in the 2007 outbreak in Uganda (19, 21, 29). The genes encoding the polymerase complex proteins NP, VP35, VP30, and L, which are essential for transcription and/or replication and, therefore, essential for minigenome activity, were cloned in the pTM1 vector, which contains a T7 RNA polymerase promoter upstream of the gene of interest. To create the minigenomes, the untranslated regions of BDBV and EBOV were cloned into the p2,0 vector flanking a luciferase reporter gene (Fig. 2A). Addition of the T7 RNA polymerase promoter and a hepatitis delta ribozyme, a self-cleaving site that ensures an exact 3′ end, were added in by PCR amplification to the corresponding 3′-leader and 5′-trailer amplicons. All plasmids were confirmed by Sanger sequencing, and no mutations were found. The minigenome construct is identified as 3B5B, indicating that the sequences from BDBV (B) are used for the 3′-leader (3) and 5′-trailer (5).

FIG 2.

Generation and optimization of BDBV minigenome replication system and chimeric minigenomes. (A) Map of EBOV, BDBV, and chimeric minigenomes. Restriction sites for RsrII and NdeI are located outside the leader region, and NotI and XmaI are outside the trailer region. The T7 promoter is located upstream of the 5′ end to initiate transcription, and the hepatitis delta ribozyme self-cleaving site is located downstream of the 3′ end to ensure an exact 3′ end. Naming refers to the 3′-leader (3) or 5′-trailer (5) from BDBV (B) or EBOV (E). (B) BDBV support plasmids were titrated from 0 ng to 2 μg. Each plasmid was titrated with the other four plasmids remaining constant. Plasmids that were kept constant were used at the following concentrations: NP at 500 ng, VP35 at 1 μg, VP30 at 100 ng, L at 500 ng, minigenome 3B5B at 2 μg. Each condition was tested in duplicate and graphed as the mean percent maximal luminescence for each curve. (C) A narrowed range of concentrations was tested to determine the optimal input of each support plasmid and the minigenome plasmid. The amount of each plasmid is shown in nanograms, and the minigenome activity from each combination is shown as percent maximal luminescence. Each condition was tested in triplicate and graphed as means ± standard deviations.

Titration of BDBV minigenome and support plasmids.

The four proteins of the polymerase complex, NP, VP35, VP30, and L, are essential for transcription and replication of the minigenome (12, 27). Each support plasmid was titrated from 0 ng to 4 μg, while the concentration of the other three support plasmids and the minigenome-coding plasmid remained constant. Initial plasmid concentrations were chosen based on the published concentrations for the EBOV minigenome system (30). Titrations of the BDBV plasmids resulted in a range of activity (Fig. 2B), indicating an optimal concentration of each plasmid and showing that too little or too much of an individual plasmid renders the system inactive. In the case of the minigenome plasmid, increasing concentrations of this plasmid resulted in increased luciferase activity, as would be expected. All plasmids were essential for the transcription of luciferase, as removal of any single plasmid resulted in no luminescence signal. Further testing was completed to determine the optimal concentration of each plasmid in the full system (Fig. 2C). Based on these titration experiments, the optimized BDBV minigenome required 500 ng NP, 1,000 ng VP35, 50 ng VP30, 250 ng L, and 2,000 ng minigenome 3B5B.

Exchange of polymerase complex proteins.

To determine if the proteins of the EBOV and BDBV polymerase complexes could interact and function, all combinations of polymerase complex proteins were exchanged to test for the ability to transcribe minigenome 3B5B or 3E5E (Fig. 3). VP30 was interchangeable between the systems regardless of the other proteins, and there was no significant difference between the complete system and the combination in which VP30 was exchanged (Fig. 3A and B, columns 2 versus 5). The EBOV system was amenable to substitution with BDBV NP, whereby addition of this protein significantly increased minigenome expression 2-fold (Fig. 3A, column 2 versus 16). This increase in minigenome expression was also seen when both BDBV NP and VP30 were exchanged into the EBOV system (Fig. 3A, column 14). In contrast, the presence of EBOV NP resulted in a significant decrease in minigenome expression to 27% of the maximum when used in the BDBV system (Fig. 3B, column 2 versus 16). The polymerase unit consisting of EBOV VP35 and L transcribed and replicated the minigenome 3E5E most efficiently (Fig. 3A, columns 2, 5, 14, and 16). When either or both components were replaced by the BDBV protein, a significant reduction in minigenome activity was observed. The 3B5B minigenome, though, was most efficiently transcribed and replicated when the system contained NP, VP35, and L from BDBV (Fig. 3B, columns 2 and 5).

FIG 3.

Exchange of polymerase complex proteins. Polymerase complex proteins were exchanged in the optimized EBOV (A) or BDBV (B) system. BDBV (B) and EBOV (Z) plasmids were exchanged at equal concentrations. Each condition was tested in triplicate and graphed as means ± standard deviations. Results are expressed as percent relative luminescence units (RLU) compared to column 2 which represents 100%. ns, nonsignificant; #, P < 0.0001 compared to column 2 by one-way ANOVA with Dunnett’s post hoc analysis.

Recognition of related minigenomes.

To better compare the ability of the EBOV and BDBV polymerase as a whole to synthesize RNA, we examined whether each complex could recognize and transcribe the minigenome from a different species. To do this, BSR-T7/5 cells were transfected with the complete polymerase complex of either BDBV or EBOV and the heterologous minigenome. Samples were collected 48 h posttransfection and luciferase activity was measured. The EBOV polymerase complex was able to transcribe the minigenome 3B5B, but there was a statistically significant decrease in minigenome expression compared to the 3E5E minigenome (51% maximum expression, P = 0.0002) (Fig. 4A). The BDBV polymerase complex was able to transcribe the minigenome 3E5E, but, similar to EBOV, there was a statistically significant decrease in minigenome expression compared to the 3B5B minigenome (40% maximum expression, P < 0.0001) (Fig. 4B).

FIG 4.

Recognition of related and chimeric minigenomes. (A and B) Minigenomes 3E5E (red) and 3B5B (blue) were used with the optimized polymerase complex of EBOV (A) or BDBV (B). Comparisons were made using a t test. (C and D) Chimeric minigenome 3E5B (green) and 3B5E (purple) were used with the optimized polymerase complex of EBOV (C) or BDBV (D). Comparisons were made using a one-way ANOVA with Dunnett’s post hoc analysis. Each condition was tested in triplicate and graphed as means ± standard deviations. Results are expressed as percent relative luminescence units (RLU) compared to the complete system, which represents 100%. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Recognition of chimeric minigenomes.

Based on the results examining transcription of the heterologous minigenome, it was of interest to test if the decrease in minigenome reporter activity was due to differences in the 3′-leader and 5′-trailer regions. Chimeric minigenomes were designed to have the 3′-leader of EBOV and the 5′-trailer of BDBV (3E5B) or the 3′-leader of BDBV and 5′-trailer of EBOV (3B5E) (Fig. 2A). Plasmids encoding each minigenome were transfected with the optimized concentration of support plasmids for each system and luciferase activity was measured. The EBOV polymerase complex was able to transcribe both the 3E5B and 3B5E chimeric minigenomes. The chimera containing the BDBV trailer 3E5B was replicated at 54% maximal activity (P = 0.0002) while the chimera 3B5E, containing the BDBV leader, had increased transcription to 125% maximal activity (P = 0.0110) (Fig. 4C). The BDBV polymerase complex was also able to transcribe both chimeric minigenomes, although expression was significantly lower than that of the homologous minigenome (3B5E, 77% maximal activity, P = 0.0053; 3E5B, 57% maximal activity, P < 0.0001) (Fig. 4D).

Use of minigenome system to screen polymerase inhibitors.

Minigenome systems offer a rapid and accessible means to study the efficacy of replication/transcription-targeted therapeutics in vitro. This method has been used to screen novel and repurposed compounds against EBOV (35–38). Based on the results comparing the polymerase complexes of BDBV and EBOV, which suggested slower and less efficient growth kinetics for BDBV, it was of interest to examine whether compounds tested against EBOV had a similar or better efficacy profile against BDBV. The compound remdesivir (GS-5734) was tested in the context of the BDBV and EBOV minigenome systems and viral infection (39–41). Remdesivir treatment showed a reduction in minigenome activity in this study (Fig. 5). When remdesivir was added to the medium 1 h after transfection, there was a clear dose response that was equivalent between BDBV and EBOV (Fig. 5A). Interestingly, although a dose response still existed, when treatment was delayed to 12 or 24 h after transfection, BDBV minigenome activity was reduced more than that of EBOV minigenome activity (Fig. 5B and C, respectively).

FIG 5.

Remdesivir inhibition of minigenome activity. EBOV (red) and BDBV (blue) minigenome activity was measured in the presence of the nucleoside analog remdesivir. Remdesivir was added to wells 1 h posttransfection (A), 12 h posttransfection (B), or 24 h posttransfection (C). Samples were lysed 48 h posttransfection and luciferase activity measured. All concentrations compared to DMSO vehicle control which was set to 100% luciferase activity. Means ± standard errors of the means from 2 replicates are graphed. A t test was used to compare the percentage of total luciferase activity between systems for the indicated concentration of remdesivir. **, P < 0.01; ***, P < 0.001.

Remdesivir effect on BDBV and EBOV in vitro.

The minigenome system indicated a more robust benefit for use of remdesivir in inhibiting BDBV polymerase activity, even when treatment was delayed, compared to EBOV polymerase activity. To understand if the results from the minigenome studies were relevant to virus infection in target cell types of human origin, remdesivir treatment was further characterized in the context of viral infection in HepG2 cells. Remdesivir titration studies where treatment was initiated at 1 h postinfection determined the 90% effective concentration (EC₉₀) to be 109.6 nM for BDBV and 284.1 nM for EBOV when the readout was infectious virus 5 days postinfection (Fig. 6A). To test the delayed treatment efficacy of remdesivir, two concentrations were tested with treatment initiated either 24 or 48 hpi. These concentrations were the EC₉₀ and 12.5 μM. The latter was chosen based on results from the minigenome experiments. Since the active metabolite of remdesivir has a half-life of 20 to 25 h, we decided to add fresh drug every 24 h. Interestingly, while the EC₉₀ was effective as a single dose given 1 hpi (Fig. 6A), there was no difference from the vehicle control when treatment was delayed 24 or 48 hpi (Fig. 6B for 24 hpi delayed treatment; data not shown for 48 hpi delayed treatment). A concentration of 12.5 μM, on the other hand, showed a significant decrease in viral titers when treatment was delayed either 24 or 48 hpi (Fig. 6C and D, respectively). This difference was more pronounced for BDBV than EBOV. A single delayed treatment at 12.5 μM was also tested, but this showed no difference from the vehicle control, indicating that continuous replenishment of remdesivir is necessary to reduce viral titers (data not shown).

FIG 6.

Remdesivir inhibition of viral infection. (A) Titration of a single dose of remdesivir added at 1 h postinfection. Supernatant was collected 5 days postinfection. (B) Wells were treated every 24 h with the EC₉₀ beginning at 24 hpi. EBOV EC₉₀, 284.1 nM; BDBV EC₉₀, 109.6 nM. (C and D) Wells were treated every 24 h with 12.5 μM remdesivir starting 24 hpi (C) or 48 hpi (D).

DISCUSSION

In this study, we describe the first minigenome replication system for the newly emerged BDBV that can be used in a BSL-2 setting and used the system to screen polymerase inhibitors. We selectively compared the BDBV polymerase complex machinery with the more widely studied EBOV. BDBV is of great concern as the geographic range of the virus overlaps with both EBOV and SUDV. As BDBV appears to be less lethal than either EBOV or SUDV in humans and NHPs with a delayed time to death in humans, NHPs, and ferrets, it was of great interest to develop a system in which to compare these viruses directly (18–25). As observed in previous publications in the development of filovirus minigenome systems, an optimal total concentration of each plasmid as well as relative concentrations were necessary for maximum minigenome activity. BDBV NP appeared to be optimal at 500 ng with a dramatic loss of activity observed with an increase or decrease in plasmid concentration. The concentration of VP35 seemed to have the greatest influence in determining optimal minigenome activity. Even as other plasmids were optimized, increasing the amount of BDBV VP35 from 500 ng to 1 μg consistently resulted in increased activity. The amount of BDBV VP30 needed was limited, with a concentration of 50 ng being optimal. The amount of BDBV L plasmid was increased compared to the EBOV system, but this is perhaps due to the increased amount of VP35 and the need for a proper ratio of these two proteins (27). One limitation of these experiments is the inability to detect the amount of protein produced by these plasmids. Due to limitations in antibody cross-reactivity between EBOV and BDBV and a lack of BDBV-specific antibodies, we were unable to accurately detect and quantify BDBV protein production.

At optimized conditions, the BDBV polymerase complex was most efficient when the minigenome contained both the BDBV leader and trailer. In contrast, only changes in the trailer region affected the efficiency of the EBOV polymerase complex. Several possibilities exist for these variations using the heterologous and chimeric minigenomes. First, it is possible that there are differences in the efficiency of RNA binding by NP from BDBV compared to EBOV. As seen when the NP plasmid was exchanged between systems, BDBV NP enhanced the EBOV system, while EBOV NP decreased the efficiency of the BDBV system. Sequence analysis comparing the residues engaged in the RNA binding indicates no difference between these two species, but more distant residues could influence the binding efficiency (32). The interaction of NP with the heterologous polymerase complex could also influence the binding efficiency of NP with the RNA, leading to the observed differences. Future studies should examine the divergent sequences in NP for possible effects on binding efficiency.

A second possibility is that the polymerase complex was unable to efficiently transcribe the reporter gene due to differences in the leading 3′-UTR. The first 3 to 4 nucleotides in the 3′-UTR have been shown to be specific to filovirus genera, with neither EBOV nor LLOV able to transcribe and replicate a minigenome containing the 3′-UTR of MARV (14). The first 90 nucleotides of EBOV and BDBV are nearly identical, with 100% similarity in the first 17 nucleotides as well as in the NP gene start signal. Within the 3′-UTR, including the noncoding region of the NP gene, are two replication promoter elements (33). The first element consists of the first 55 nucleotides of the 3′-UTR and folds on itself to form a stable hairpin structure (33). The second promoter element is a sequence of UN₅ hexamers with a minimum requirement of three hexamers (33). The EBOV UTR has eight while BDBV has six. Interestingly, the EBOV polymerase complex could transcribe and replicate the minigenome with the BDBV leader more efficiently than with its own leader. An additional hairpin loop incorporating nucleotides 56 to 78 is necessary for VP30-dependent transcription (11). There are 2 nucleotide variations (A68U and U69C) in BDBV, but the hairpin formation is likely to remain intact, as these changes occur in the hairpin loop, not in the stem (Fig. 7A).

FIG 7.

Sequence variations between EBOV and BDBV. (A) 3′-Leader including initiation motif (blue), NP gene start signal (purple), NP gene start signal hairpin, and UN₅ motifs (boxed). Differing nucleotides are shown in red. Differing BDBV nucleotides are shown outside the hairpins in red. (B) Alignment of EBOV and BDBV trailer region with differences shown in red.

A third possibility is that replication of the minigenome between the genome and antigenome states was limited due to differences in the trailing 5′-untranslated regions. This was examined using chimeric minigenomes containing the 3′-UTR from one species and the 5′-UTR from the other species. The BDBV support plasmids transcribed both chimeric minigenomes, although to a significantly lower level than the 3B5B minigenome. Similarly, when the EBOV support plasmids were used, there was a significant decrease in the amount of minigenome activity with the chimera containing the EBOV leader and the BDBV trailer (3E5B). In contrast, when the chimera 3B5E, containing the BDBV leader and the EBOV trailer, was included with the EBOV support plasmids, there was a significant increase in the amount of luciferase activity. As the 5′-trailer region is important for replication, the 51% nucleotide divergence between the BDBV and EBOV trailer regions has the potential to impede efficient replication (Fig. 7B). A limitation to the minigenome assay is that the read-out of luminescence looks specifically at transcription. While replication of the minigenome may occur, adding to available copies of the luciferase gene for transcription, we were unable to measure replication independent of transcription. Analysis by sequencing or strand-specific reverse transcription-PCR would aid in answering this question. The possibility that the BDBV 5′-trailer region limits genome replication should be further examined in the context of a full-length recombinant virus where trailer regions can be exchanged.

Since work in high-containment laboratories is limited both in access and feasibility, it is necessary to develop model systems that can allow for experimentation at lower biosafety levels and at a larger scale. Artificial transcription and replication systems are of great use because they allow for testing therapeutics targeted at these processes. We used the newly developed BDBV minigenome system to test the in vitro efficacy of remdesivir, which was previously reported in the literature to inhibit EBOV minigenome or viral replication. Remdesivir treatment had an inhibitory effect on both minigenome activity and viral infection, and this inhibition was maintained even with a delay of treatment, indicating this minigenome system acts as a suitable tool for initial screening of therapeutic compounds. The results from the minigenome experiments indicate that remdesivir offers a greater inhibitory effect on BDBV than that of EBOV. This finding was maintained in the context of viral infection: concentrations above 10 μM had an inhibitory effect on both EBOV and BDBV regardless of when treatment was initiated. However, with delayed treatment, the effect was greater against BDBV than EBOV, and by day 5 postinfection there was a near complete clearance of BDBV, while EBOV titers plateaued over time. A possibility for the differential effect of remdesivir on these two ebolavirus species is the slower growth kinetics observed in our study and by Gupta et al. (17). Since BDBV appears to grow slower than EBOV and there is a lower viral titer for BDBV than EBOV at all time points measured, it is expected that less of the active compound of remdesivir is necessary to inhibit viral growth. This was demonstrated in the titration of remdesivir, with a nearly 3-fold lower EC₉₀ for BDBV than EBOV. Interestingly, the utility of delayed remdesivir treatment was seen only with continuous replenishment of the compound, indicating a need to maintain a surplus of the active metabolite as the virus continues to replicate. The need for continuous treatment provides supporting data for the 10-day clinical treatment regimen of remdesivir for treating an ebolavirus infection.

One possible mechanism for the lack of utility of a single delayed treatment is that remdesivir may preferentially act on viral gene transcription. The efficacy of a single treatment 1 hpi points to this mechanism, as transcription would be the dominant viral intracellular process occurring at this time. By 24 hpi there is a substantial increase in genome copies, indicating genome replication is taking place. This increase in genome copies occurs at a later time point for BDBV than EBOV and points to the possibility that BDBV is undergoing transcription for a longer period of time. While it is likely that remdesivir can be incorporated during replication and result in early termination of new genomes, the overall effects on viral growth would be greater if transcription was inhibited. This is due to the necessity of the four protein components of the polymerase complex. Inhibition of transcription would lead to an inhibition or delay in the production of these proteins and, therefore, a delay in further transcription and replication. Future analyses should look at the incorporation of remdesivir in viral mRNA, genome copies, and antigenome copies.

In conclusion, in this study we have described a novel minigenome system to study the polymerase complex of BDBV. We directly compared the BDBV polymerase complex with the polymerase complex of EBOV and found that the efficiency of the BDBV polymerase complex appears to be lower than that of EBOV, as shown by the luciferase activity when the heterologous minigenomes were used. Differences seen using the minigenome system should be further teased apart in the context of the full-length virus in known target cell types. We utilized the minigenome system as a tool to screen small molecules for therapeutic use and compared the minigenome results to inhibitory effects against infection. Remdesivir proved efficacious in the minigenome system and showed continued in vitro efficacy against viral infection. Remdesivir showed greater inhibitory effects at lower concentrations and with delayed treatment for BDBV than EBOV. These results indicate that remdesivir has a greater therapeutic benefit for treatment of infections with BDBV than infectious with EBOV and that the minigenome system for BDBV can be a good surrogate to assess treatments targeting the polymerase complex machinery.

MATERIALS AND METHODS

Cells.

THP-1 cells were maintained in complete RPMI 1640 (cRPMI; Gibco, Carlsbad, CA) with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin (P/S; Gibco) at 37°C and 5% CO₂. For differentiation, cells were plated in cRPMI with 200 nM phorbol 12-myristate 13-acetate (PMA) on a 24-well plate at a density of 2 × 10⁵ cells per well. Three days afer plating, cells were observed for adherence and the medium was replaced with fresh cRPMI. Two days later, cells were transferred into the BSL-4 laboratory for inoculation.

HepG2 cells were maintained in Eagle’s minimum essential medium (EMEM) with 10% FBS and 1% P/S at 37°C and 5% CO₂. For infection experiments, cells were plated at a density of 5 × 10⁵ cells per well of a 24-well plate.

Vero E6 cells were used for titration of viruses. Cells were maintained in EMEM with 10% FBS and 1% P/S at 37°C and 5% CO₂.

The baby hamster kidney cell line BSR-T7/5 was used for all transfection experiments. BSR-T7/5 cells stably express the T7 polymerase (34) under positive selection with the antibiotic Geneticin, which was added to the medium every other passage. Cells were cultured in Dulbecco’s modified Eagle medium (DMEM; Gibco) with 10% FBS and 1% P/S. Cells were maintained at 37°C and 5% CO₂.

Virus.

A laboratory seed stock of Zaire ebolavirus strain Mayinga was grown from the serum of a fatal human case in 1976 in the Democratic Republic of the Congo (Zaire ebolavirus/H.sapiens-tc/COD/1976/Yambuku-Mayinga, accession number NC_002549) and passaged twice in authenticated Vero E6 cells.

A laboratory seed stock of Bundibugyo ebolavirus was grown from the serum of a fatal human case in 2007 in Uganda (Bundibugyo virus/H.Sapiens-tc-UGA/2007/Bundibugyo-200706291, accession number KU182911) and passaged twice in authenticated Vero E6 cells. Supernatant was inactivated in TRIzol (Ambion, Carlsbad, CA) before removal from BSL-4 facilities.

Infection.

All experiments using infectious virus were performed in BSL-4 facilities at the Galveston National Lab in Galveston, Texas. Virus inoculum was made up in serum-free EMEM at a multiplicity of infection (MOI) of 1.0. Medium was removed from wells, and inoculum was added and adsorbed for 1 h with gentle rocking every 15 m. After adsorption, wells were washed five times with phosphate-buffered saline (PBS) and fresh medium was added. At each time point supernatant was collected, clarified, and frozen before determining titers.

Titer determination.

Vero E6 cells were plated in 6-well plates to form a continuous monolayer, and 200 μl inoculum was added and adsorbed for 1 h with gentle rocking every 15 m. Cells were overlaid with 0.8% agarose (Lonza) and 5% MEM. Five percent neutral red (Sigma-Aldrich, St. Louis, MO) stain in PBS was added after 5 days for EBOV and 7 days for BDBV. After 24 h, stain was removed and plaques were counted.

Plasmids.

The Zaire ebolavirus (EBOV) minigenome system, including the minigenome plasmid and four support plasmids, were a gift from Elke Muhlberger (Addgene plasmids 68121, 69119, 69120, 69121, and 69358) (10). This system is based on EBOV strain Mayinga. The transfection control plasmid pGL4.74[hRluc/TK] (Promega, Madison, WI), encoding Renilla luciferase, was used for all luciferase assays.

Cloning.

(i) Support plasmids. Construction of the BDBV support plasmids pTM1_NP_BDBV, pTM1_VP30_BDBV, and pTM1_VP35_BDBV was completed using the backbone of the pTM1_VP30_ZEBOV plasmid after restriction enzyme digestion with EcoRI-HF and PacI (New England Biolabs Inc. [NEB], Ipswich, MA). BDBV RNA was harvested by TRIzol extraction using the Direct-zol RNA miniplus kit (Zymo Research, Irvine, CA). Reverse transcription was performed using the SuperScript IV first-strand synthesis system (Invitrogen, Carlsbad, CA) using a gene-specific forward primer. The gene of interest was then amplified with the Platinum SuperFi PCR master mix (Invitrogen) using gene-specific primers with the addition of the desired 5′ and 3′ restriction sites. Following PCR amplification, the product was column purified using the PureLink PCR purification kit (Invitrogen) and underwent restriction enzyme digestion and a second column purification. Ligation of vector and insert was performed using the Fast-Link DNA ligation kit (Lucigen, Middlesex, UK). Ligated plasmids were transformed into chemically competent C600 E. coli cells (NEB) and selected based on ampicillin resistance. Colonies were screened by PCR and restriction digestion, and positive clones were confirmed by Sanger sequencing.

The support plasmid pTM1_L_BDBV was constructed using pTM1_VP30_ZEBOV as a backbone after restriction enzyme digestion with AgeI and XhoI (NEB). The L gene was reverse transcribed and amplified in two segments. Reverse transcription was performed using the SuperScript IV first-strand synthesis kit (Invitrogen) using a gene-specific forward primer for each segment, and amplification was completed using the Platinum SuperFi PCR master mix (Invitrogen). Primers were designed encoding the end of the gene with an overlapping region complementary to the pTM1 plasmid backbone and reconstructing the restriction sites per the principles of Gibson cloning (31). All PCR products were purified using the PureLink PCR purification kit (Invitrogen). The final plasmid was constructed using the NEBuilder HiFi DNA assembly cloning kit (NEB), and the ligation mixture was transformed into chemically competent NEB5α E. coli cells (NEB) and selected based on ampicillin resistance. Colonies were screened by PCR and restriction digestion, and positive clones were confirmed by Sanger sequencing.

(ii) Minigenome plasmids. The vector backbone for the minigenome plasmids was derived from p2,0_3E5E_luciferase. The restriction enzymes RsrII, NdeI, NotI, and XhoI (NEB) were used to digest out the unwanted EBOV fragments (Fig. 1, 3E5E). The BDBV leader and trailer regions were synthesized as described for L using genome-specific forward primers for reverse transcription, addition of an overlapping region to the vector, and reconstruction of restriction sites. The hepatitis delta virus (HDV) ribozyme self-cleaving site was added to the leader region, and the T7 promoter was added to the trailer region during PCR amplification. Construction of the plasmids was completed using NEBuilder HiFi DNA assembly cloning kit (NEB), and the ligation mixture was transformed into chemically competent NEB5α E. coli cells (NEB) and selected based on ampicillin resistance. Colonies were screened by PCR and restriction digestion, and positive clones were confirmed by Sanger sequencing. The minigenome plasmids consist of the 3′-leader (3) and 5′-trailer (5) regions of either EBOV (E) or BDBV (B) and are annotated as 3E5E, 3E5B, 3B5E, and 3B5B (Fig. 1).

Primers used for all cloned plasmids are listed in Table 1.

TABLE 1.

Primers used for generation of amplicons needed for cloning support and minigenome plasmids^a

Primer name	Sequence
NP forward	TCA GAT GAA TTC GCT ACA TTC TCT ATC CAA GAC C
NP reverse	TGT CCA TTA ATT AAC CAT CAC CTG TGA TGC TGG
VP35 forward	TTA GCA GAA TTC TAT GAC CTC TAA CAG AGC AAG
VP35 reverse	TGT ACC TTA ATT AAC CAA CCT TAG ATT TTG AGT CCG AG
VP30 forward	TTA GGT GAA TTC ATC TTG GGG ATT TCT CTG AAC
VP30 reverse	AGT CAC TTA ATT AAC CAT CTT ATC TGC GTT GAA TAG GG
L forward 1	TAG TGG ATC CGC GAA ATG GCA ACT CAA CAT ACA C
L reverse 1	CCA GGT ATA TTC ACG TAA AAT TTG CGC CAA ATC AAC TGT ACA AGA AAC
L forward 2	GTT TCT TGT ACA GTT GAT TTG GCG CAA ATT TTA CGT GAA TAT ACC TGG
L reverse 2	CCG GAT CGT CGA CTT AAT CTC TAA GGG GAT CTT AAG CG
BDBV leader forward	GAT GCC CAG GTC GGA CCG CGA GGA GGT GGA GAT GCC ATG C CG ACC C CG GAC ACA CAA AAA G
BDBV leader reverse	TTT TGG CGT CTT CCA TAT G CA TTT TGA GGT CTT G
BDBV trailer forward	GAT CGC CGT GTA AGC GGC CGC GAT CCC CTT AGA GG
BDBV trailer reverse	CAG GGG GAT ATC GAT CCC GGG TAA TAC GAC TCA CTA TAG TGG ACA CAC AAA AAA G

^aUnderlined sequences correspond to the BDBV genome, italicized sequences correspond to minigenome features T7 promoter and hepatitis delta ribozyme, and boldface sequences correspond to vector overlapping region needed for Gibson cloning. (32–34).

Transfection.

BSR-T7/5 cells were plated in 6-well tissue culture-treated plates at a density of 4 × 10⁵ cells/well per recommendations for the EBOV minigenome system (30). When cells reached 70% confluence, they were transfected using Lipofectamine 3000 (Invitrogen) per the manufacturer’s protocol. Cell lysates were harvested 48 h after transfection for analysis of either viral protein expression or luciferase expression.

Minigenome expression by luciferase assay.

Minigenome expression was quantified using the Dual-Luciferase assay system (Promega). Forty-eight hours after transfection, medium was removed and cell monolayers were washed once with phosphate-buffered saline (PBS). Cell lysis was completed using passive lysis buffer per the manufacturer’s protocol, and lysates were collected for downstream analysis. All conditions were tested in triplicate and standardized as relative luminescence units (RLU) calculated as a ratio of Firefly luciferase to Renilla luciferase. The complete minigenome system for each species was considered 100% activity.

Compound.

GS-5734 (remdesivir) was purchased from MedChemExpress (Monmouth Junction, NJ). For all experiments, the compound was dissolved in dimethyl sulfoxide (DMSO) per the manufacturer’s recommendations and stored until use. All dilutions for treatment experiments were in complete media before addition to cells.

Statistical analysis.

All statistical analyses were performed using Prism (v. 8; GraphPad, San Diego, CA) with an alpha level set at 0.05. A t test, corrected for multiple comparisons, was used for the comparison of minigenome activity between systems. For the comparison of minigenome activity within each system, an analysis of variance (ANOVA) was performed followed by Dunnett’s method adjusting alpha risk for multiple comparisons. Comparisons were made to the homologous minigenome (3E5E for EBOV support plasmids, 3B5B for BDBV support plasmids). Details of statistical tests can be found in the figure legends.