ABSTRACT
Machupo virus (MACV), the causative agent of Bolivian hemorrhagic fever (BHF), is a New World arenavirus that was first isolated in Bolivia from a human spleen in 1963. Due to the lack of a specific vaccine or therapy, this virus is considered a major risk to public health and is classified as a category A priority pathogen by the U.S. National Institutes of Health. In this study, we used DNA vaccination against the MACV glycoprotein precursor complex (GPC) and murine hybridoma technology to generate 25 mouse monoclonal antibodies (MAbs) against the GPC of MACV. Out of 25 MAbs, five were found to have potent neutralization activity in vitro against a recombinant vesicular stomatitis virus expressing MACV GPC (VSV-MACV) as well as against authentic MACV. Furthermore, the five neutralizing MAbs exhibited strong antibody-dependent cellular cytotoxicity (ADCC) activity in a reporter assay. When tested in vivo using VSV-MACV in a Stat2−/− mouse model, three MAbs significantly lowered viral loads in the spleen. Our work provides valuable insights into epitopes targeted by neutralizing antibodies that could be potent targets for vaccines and therapeutics and shed light on the importance of effector functions in immunity against MACV.
IMPORTANCE MACV infections are a significant public health concern and lead to high case fatality rates. No specific treatment or vaccine for MACV infections exist. However, cases of Junin virus infection, a related virus, can be treated with convalescent-phase serum. This indicates that a MAb-based therapy for MACV could be effective. Here, we describe several MAbs that neutralize MACV and could be used for this purpose.
INTRODUCTION
Machupo virus (MACV) is the etiological agent of Bolivian hemorrhagic fever, a devastating viral disease with case fatality rates ranging between 25 and 35% (1). MACV is classified as a New World (NW) arenavirus (genus Mammarenavirus), and the reservoir host of MACV is Calomys callosus, the large vesper mouse (2, 3). MACV was first isolated in 1963 from the spleen of a human patient who died from the consequences of viral pathogenesis (4). MACV was later described in 1964 as a hemorrhagic fever virus (5). While the virus does not actively circulate in humans, sporadic outbreaks have occurred in 2000 (18 cases), 2007 (20 cases), and 2008 (200 cases) in Bolivia (6). Human contact with aerosolized excretions or consumption of food contaminated with excretions of the rodent host is the main transmission mode for zoonotic MACV infections (7). Currently, there is no approved vaccine or therapy available for disease caused by MACV. It has been shown in nonhuman primate studies that vaccination against the related Junin virus (JUNV) confers protection against a lethal MACV challenge (8). However, the candidate vaccine strain for JUNV has not been approved for use in Bolivia or the United States, and its effectiveness in humans against MACV infections is unclear. Hence, there is an urgent need to develop therapeutics and vaccines for use during outbreaks of MACV.
Members of the genus Mammarenavirus (family Arenaviridae) are known to infect mammals, and some species can cause lethal hemorrhagic fever in humans (9). Different mammarenaviruses are found worldwide and are typically subdivided into Old World (OW) and NW viruses based on geographic distribution, genetic divergence, and epidemiology (10, 11). NW mammarenaviruses are grouped into four clades: A, B, C, and A/rec (recombinant, clade D) (7). There are five NW mammarenaviruses that cause hemorrhagic fevers in humans, including MACV, JUNV, Chapare virus (CAPV), Guanarito virus (GTOV), and Sabia virus (SABV). All five of these viruses cluster together in clade B.
The RNA genome of Mammarenavirus is negative sense and bi-segmented. The large (L) segment codes for both the matrix protein (Z) and the large RNA-dependent RNA polymerase (LP). The small (S) segment codes for both the nucleoprotein (NP) and the glycoprotein precursor complex (GPC) (12). GPC is the only viral protein found on the surface of mammarenaviruses and undergoes proteolytic cleavage to form a trimer with each subunit consisting of a stable signal peptide (SSP), GP1, and GP2 (13). The GPC of arenaviruses is important for attachment and entry into host cells (14). For NW arenaviruses, transferrin receptor 1 (TfR1) serves as the cellular entry receptor in natural hosts and humans due to its interaction with GP1 (15, 16). Given the crucial role the GPC plays in the viral life cycle of arenaviruses, it represents a promising target for vaccine and therapeutic development.
In this study, we generated 25 murine monoclonal antibodies (MAbs) from mice vaccinated with DNA encoding the full-length GPC of MACV. The majority of the generated MAbs exhibited specific binding to MACV GPC while a few cross-reacted against other NW arenavirus GPCs. Five antibodies could in vitro neutralize recombinant vesicular stomatitis virus which expresses the MACV GPC instead of its own G protein (VSV-MACV), and the same five antibodies also elicited antibody-dependent cellular cytotoxicity (ADCC) in a reporter assay. Furthermore, these antibodies were neutralizing against authentic MACV in vitro. When tested for efficacy against VSV-MACV in a Stat2−/− mouse model, three antibodies significantly reduced viral load in the spleen when administered prophylactically.
RESULTS
A majority of the generated MAbs are highly specific to MACV GPC with little cross-reactivity to other arenavirus GPCs.To obtain anti-MACV GPC antibodies, mice were immunized twice with a plasmid encoding the MACV GPC and then boosted with a recombinant MACV GPC protein generated via the baculovirus expression system. A hybridoma fusion was then performed 3 days postboost. Twenty-five IgG-secreting hybridoma clones were obtained from the hybridoma fusion, and all 25 MAbs were further characterized (Table 1). Out of 25 MAbs, 4 belonged to subclass IgG1, 18 MAbs belonged to the IgG2a subclass, and the remaining 3 MAbs belonged to subclass IgG2b.
Subclasses of anti-MACV GPC MAbs
In order to assess the binding profiles of these MAbs, enzyme-linked immunosorbent assays (ELISAs) were performed using dilutions of each MAb starting at 30 μg/ml against several recombinant arenavirus GPCs such as MACV GPC, Tacaribe virus (TCRV) GPC, JUNV GPC, GTOV GPC, Tamiami (TAMV) GPC, and Lassa virus (LASV) GPC. Figure 1 shows the minimal binding concentration (MBC) of each MAb against these recombinant GPCs. MAb KL-AV-2A1, a pan-arenavirus MAb, was used as a positive control (14). All MAbs bound to MACV GPC even at low concentrations, except for MAb KL-MM-1B7, which had a high minimal binding concentration (suggesting low affinity). When tested against other closely related NW arenavirus GPCs such as JUNV GPC and TCRV GPC, only a few MAbs had low MBC values. A low MBC value is indicative of binding, even at low concentrations of the MAb against a respective GPC. Only three MAbs bound to GTOV GPC in ELISAs. TAMV, another NW arenavirus in clade D, was bound by only two MAbs. When tested against NW arenaviruses, the 25 MAbs overall appear highly specific to MACV GPC. To characterize the binding of the MAbs even further, the GPC of LASV, an OW virus, was also used in the ELISA panel. Surprisingly, there were some antibodies that bound a range of NW and OW (LASV) GPCs, such as KL-MM-2C2, KL-MM-2D10, KL-MM-2H5, and KL-MM-4C12. Some MAbs, such as KL-MM-1B7 and KL-MM-3A3, bound only MACV GPC and LASV GPC. Although very different in terms of amino acid sequence, LASV GPC and MACV GPC may share unique epitopes that are exclusive and not present on other NW GPCs.
Most antibodies are highly specific and bind only to MACV GPC. Binding of antibodies to several NW GPCs and LASV GPC is shown. A standard ELISA was performed using several dilutions of each respective antibody starting at 30 μg/ml against various NW arenavirus GPCs such as MACV GPC, TCRV GPC, JUNV GPC, GTOV GPC, TAMV GPC, and LASV GPC. An anti-influenza H4 virus antibody was used as a negative control while a pan-arenavirus GPC antibody, KL-AV-2A1, was used as a positive control. The values in the graph correspond to minimal binding concentration, the lowest concentration at which the antibody still demonstrates binding signal. Phylogenetic differences between the GPCs is indicated by a phylogenetic tree (based on amino acid sequence). The scale bar represents a 5% amino acid change.
Most isolated antibodies bound native MACV glycoprotein as expressed on the surface of an infected cell.To ensure that the antibodies can also bind MACV glycoprotein when expressed on the cell surface, Vero.E6 cells were infected with replication-competent recombinant VSV-MACV and fixed. These infected cells were then immunostained with each antibody at a concentration of 30 μg/ml (Fig. 2A). As the images in Fig. 2A demonstrate, most antibodies bound to MACV glycoprotein as expressed on infected cells in this immunofluorescence (IF) assay. A few MAbs, though, showed little to no binding. These nonbinders include KL-MM-1B7 (which also had a high MBC in the ELISA), KL-MM-2B4, KL-MM-2D7, KL-MM-2D10, and KL-MM-2G6. It is well documented in the literature that some MAbs bind recombinant GPCs but not infected/transfected cells and vice versa (17). This could occur due to exposure of epitopes in the recombinant form of the GPC which may be masked in the native, mature conformation. In addition, we fixed the infected cells in the IF assay, which could also have had a negative influence on binding. IF assays with nonfixed cells were unsuccessful. To further investigate if the MAbs were binding to linear or microconformational epitopes, Western blotting was performed using recombinant MACV GPC under denaturing conditions (Fig. 2B). An irrelevant protein, influenza A virus H2 hemagglutinin (HA), was used as control to ensure that the binding observed is specific to MACV GPC and not the hexahistidine tag itself. Most MAbs showed no binding on the Western blot; it is likely that these MAbs target a conformational epitope that has been destroyed under denaturing conditions. Only three MAbs had weak reactivity on the Western blot: KL-MM-1D4, KL-MM-2H5, and KL-MM-4A2.
Antibodies bind to cells infected with VSV-MACV. (A) Binding of antibodies to MACV glycoprotein in an immunofluorescence assay. Vero.E6 cells were infected with VSV-MACV at an MOI of 1.0 overnight and then fixed and immunostained using each antibody at a concentration of 30 μg/ml. An irrelevant antibody, anti-influenza H4 virus (KL-H4-4A11), was used as a negative control while the positive control used was KL-AV-2A1. (B) Western blot analysis. A standard Western blot was performed using recombinant MACV GPC and an irrelevant protein, influenza virus H2 HA. Each blot was stained with 30 μg/ml of each respective antibody, and an anti-mouse IgG conjugated to alkaline phosphatase was used as a secondary antibody at a dilution of 1:3,000. A separate Western blot was run using anti-histidine antibody to ensure proper transfer of protein to the nitrocellulose membrane. The ladder used was a ColorPlus Prestained Protein Ladder, Broad Range (New England BioLabs).
Five MAbs neutralize VSV-MACV as well as pathogenic rMACV.To assess if these antibodies had any functional role and biological relevance, all MAbs were tested in a plaque reduction neutralization assay (PRNA) using VSV-MACV. In this assay, each MAb was incubated with virus at various concentrations for 1 h, and this virus-antibody mixture was then used to infect Vero.E6 cells. Plaques were counted, and the percent inhibition of VSV-MACV was calculated from each MAb. These data are shown in Fig. 3A. Out of 25 MAbs, 5 antibodies, KL-MM-1B6, KL-MM-1F9, KL-MM-2C8, KL-MM-4C12, and KL-MM-4G7, had potent neutralization activity. KL-MM-2C8, KL-MM-4C12, and KL-MM-4G7 completely neutralized the virus, with no plaques visible at 100 μg/ml. All five of these MAbs belong to the IgG2a subclass and were able to bind cells infected with VSV-MACV in the IF assay shown in Fig. 2A. A few other antibodies also neutralized VSV-MACV but at lower levels (KL-MM-1B11, KL-MM-2C2, and KL-MM-3F6). Of note, the weak neutralization observed from KL-MM-3F6 correlates with low binding to the surface of infected cells.
Five antibodies have potent neutralizing activity and ADCC reporter activity against VSV-MACV and can also neutralize pathogenic MACV at low concentrations. (A) PRNA against VSV-MACV. PRNAs were performed with each antibody starting at a concentration of 100 μg/ml with subsequent 1:5 dilutions. Vero.E6 cells were infected with the antibody-virus mixture, and similar concentrations of antibody were also present in the overlay for the duration of the assay. An anti-influenza virus H4 antibody was utilized as a negative control. Percent inhibition with each antibody is plotted. (B) PRNAs against authentic MACV. PRNAs were run with several dilutions of the five MAbs that neutralized VSV-MACV. An antibody-virus mixture was used to infect Vero.E6 cells, and cells were overlaid for 8 days. No antibody was added to the overlay. An irrelevant antibody which does not bind to MACV GPC was used as a negative control for this assay. (C) PRNAs against VSV-LASV. PRNAs were also performed with VSV-LASV to test the few antibodies that had cross-reactivity toward LASV GP, as shown in Fig. 1. Cells were overlaid with antibody present in the overlay for 48 h, and plaques were counted to calculate percent inhibition. (D) ADCC reporter activity using VSV-MACV. An in vitro ADCC reporter assay was performed using a commercial kit (Promega). Vero.E6 cells were infected overnight with VSV-MACV at an MOI of 1.0, and the next day, antibody dilutions and effector cells were added. Six hours later, luciferase substrate was added, and luminescence was measured to assess fold induction of ADCC activity due to FcR engagement.
To ensure that antibodies could also neutralize authentic MACV, another PRNA was performed in a biosafety level 4 (BSL-4) facility with the five potent neutralizing antibodies (Fig. 3B). The five MAbs also neutralized authentic MACV with very potent neutralizing activity. MAbs KL-MM-1B6 and KL-MM-1F9 neutralized MACV slightly better than VSV-MACV (50% inhibitory concentrations [IC50s] of 0.47 μg/ml versus 5.9 μg/ml and 0.49 μg/ml versus and 2.1 μg/ml, respectively, for MACV and VSV-MACV). However, the opposite was true for KL-MM-2C8 (IC50s of 1.3 μg/ml versus 0.54 μg/ml, respectively, for MACV and VSV-MACV). KL-MM-4C12 and KL-MM-4G7 had nearly identical levels of activity against both viruses (IC50s of 0.5 μg/ml versus 0.45 μg/ml and 0.7 μg/ml versus 0.4 μg/ml, respectively). The IC50 values of each antibody against MACV and authentic VSV-MACV are listed in Table 2.
IC50 values of MAbs
As described above, some antibodies showed cross-reactivity toward LASV GPC, and these same antibodies had neutralizing activity against VSV-MACV, albeit at lower levels (Fig. 1). These few MAbs were tested in a PRNA using VSV-LASV (Fig. 3C). The percent inhibition seen by KL-MM-1B7, KL-MM-2C2, and KL-MM-4G7 against VSV-LASV was approximately 40% at 100 μg/ml. This could be due to weaker binding of these MAbs against LASV GPC than against MACV, given the lower MBC value in the ELISA against the former. This is especially true for KL-MM-4G7 since very little binding activity was observed.
Several neutralizing antibodies have Fc-mediated effector functions.Fc-mediated effector functions are known to play an important role in immunity against many viruses, such as influenza viruses and Ebola viruses (18, 19). Furthermore, nonneutralizing antibodies with effector functions are capable of inhibiting viral spread and can protect against viral challenge in vivo (17, 19, 20). Hence, an in vitro ADCC reporter assay was set up to assess if any of the generated MAbs are able to engage mouse Fc gamma receptor IV (FcγRIV), and the results are depicted in Fig. 3D. ADCC activity has never previously been reported for any NW arenavirus antibody. Interestingly, the same five MAbs that were able to neutralize VSV-MACV and pathogenic recombinant MACV (rMACV) exhibited high ADCC reporter activity against VSV-MACV. Antibodies that inhibited the virus at lower levels, such as KL-MM-1B7 and KL-MM-1B11, in the PRNA also had ADCC activity but in a considerably lower range.
Three neutralizing MAbs significantly reduce viral titers in vivo in a Stat2−/− VSV-MACV mouse model.To further study the functionality of these MAbs, in vivo experiments were performed in Stat2−/− mice which are immunodeficient and, thus, allow recombinant VSV viruses to replicate (19, 21). Animals were infected with 1,000 PFU via the intraperitoneal route, and organs were harvested from three mice at day 3 and day 6. A standard plaque assay was performed using each organ homogenate on a confluent monolayer of Vero.E6 cells, and day 3 (Fig. 4A) and day 6 (Fig. 4B) viral titers were plotted. As illustrated, viral titers in the spleen and serum were much higher than those in other organs such as liver, lung, and kidney. By day 6, virus had been entirely cleared from several organs, and the viral titer in the spleen dropped from 106 PFU/ml to approximately 104 PFU/ml. Many organs contained little to no infectious virus at day 6, and no morbidity was observed. The tropism of VSV-MACV we observed in Stat2−/− mice was similar to the tropism observed in Stat1−/− mice with authentic MACV, but the titers in Stat1−/− mice were significantly higher on day 5 and day 7 (22).
VSV-MACV is prevalent in the serum and spleen, and prophylactic administration of neutralizing MAbs leads to reduction in viral titer in vivo. (A and B) Assessment of viral load in several organs and the spleen. Stat2−/− mice (n = 3) were infected with 1,000 PFU of VSV-MACV via the intraperitoneal route, and organs were harvested on day 3 and day 6, as indicated, to assess viral titer via standard plaque assay with organ homogenate. An anti-influenza virus H4 antibody was used as a negative control. (C and D) Prophylactic administration of neutralizing antibodies. Each neutralizing MAb was administered at 10 mg/kg via the intraperitoneal route, and Stat2−/− mice (n = 4) were then infected with 1,000 PFU 2 h later. Viral titer in the spleen was used as a measure to see if antibody administration can reduce viral titers in vivo. The negative control was a running control as experiments were performed in batches due to limited availability of mice. Spleen was harvested on day 3 and day 6, as indicated. Statistical significance was determined via one-way ANOVA test (***, P ≤ 0.001; ns, not significant).
Having shown that VSV-MACV can efficiently replicate in Stat2−/− mice, we wanted to use this model to assess if the isolated MAbs have a protective effect. Hence, an experiment was set up with prophylactic antibody administration at 10 mg/kg. We then used virus titers in the spleen at day 3 and day 6 as readout to evaluate if antibody administration can inhibit virus replication in vivo. Only the five neutralizing antibodies were tested due to limited availability of Stat2−/− mice. Two hours prior to infection, each antibody was administered via the intraperitoneal route. Mice were then infected with 1,000 PFU each via the intraperitoneal route. Day 3 (Fig. 4C) and day 6 (Fig. 4D) virus titers in the spleen are shown for each respective antibody. While KL-MM-1F9 and KL-MM-2C8 led to little reduction in viral titer in the spleen, KL-MM-1B6 had a significant effect, and KL-MM-1B6-treated mice had much lower titers by day 6. KL-MM-4C12 and KL-MM-4G7 completely inhibited the virus, and no virus was found in the spleen at day 3. At day 6, viral titers in the spleen dropped in all of the groups, and even the negative-control animals had lower viral titers than at day 3. Little to no virus was found in the KL-MM-1B6 group and KL-MM-4C12 group at day 6 while the other groups had lower titers than those of the negative-control animals.
Neutralizing antibodies bind to epitopes overlapping the receptor binding site including the MACV-specific loop 10.To pinpoint the epitope that is targeted by each neutralizing MAb, escape mutant viruses were generated by continuously passaging each antibody with VSV-MACV at increasing concentrations. Each antibody was passaged with the virus starting at 1× IC50, and the antibody concentration was doubled until the total concentration of antibody reached 128× IC50. Three single virus clones were selected, and escape was confirmed using PRNAs (Fig. 5A to E). Sequencing was then performed to see where mutations had occurred. MAbs KL-MM-1B6, KL-MM-2C8, and KL-MM-4G7 selected the same virus escape mutants which had a leucine in position 226 instead of a phenylalanine (F226L) (Fig. 5F to I). The same site seems to have been targeted also by KL-MM-1F9; but in this case the three clones differed, and two of the clones had additional mutations (F226L/K169E, F226I, and F226L/K170N) (Fig. 5F to I). Similarly, escape mutants isolated from a KL-MM-4C12 culture showed K170N, F226I, and K170N/N174S mutations (Fig. 5F to I). F226, K169, and K170 make contact with the MACV receptor TfR1, while N174 is adjacent to the receptor binding site (RBS) (23). Of note, F226 is part of loop 10, a unique feature of MACV that is not found in other NW arenaviruses (24).
Antibodies target epitopes on GP1 to inhibit entry into host cell. PRNAs were performed with escape viruses. (A to E) Escape mutant viruses were generated for each neutralizing MAb, and these viruses were used to confirm that the virus had escaped and that the respective MAb could not neutralize the virus any longer. PRNAs were performed using each MAb against VSV-MACV as well as each respective escape virus (EV) as indicated in the figure. (F to I) Structural visualization of the antibody epitopes. Using a previously defined structure of MACV GP1 (PDB accession number 5W1M) and LASV GPC (PDB 5VK2), critical epitopes on GP1, which are targeted by neutralizing MAbs, are shown highlighted via PyMOL. MACV GP1 (gray) was aligned and overlaid onto the LASV GPC trimer (gold) (G). The RBS is shown in green, loop 10 of MACV GP1 is shown in orange (H), and the epitopes are marked with unique colors (I). For ease of visualization, one monomer is shown as a solid surface while the other two monomers are depicted as ribbon structures.
To investigate if the most common escape mutations ablated binding, ELISAs were performed using mutant MACV GP K170N, MACV GP F226L, and wild-type MACV GPC as the substrate and each antibody as a primary stain (Fig. 6). All MAbs demonstrated reduced binding to both escape variants while the positive-control MAb KL-AV-2A1 bound all three GPs equally well. Of note, while reduced binding was observed, it was not completely ablated except for KL-MM-2C8 binding to the F226L mutant. These data suggest that while the MAbs were not able to neutralize the escape viruses further, they were still able to bind to various degrees.
ELISAs using recombinant mutant MACV GPCs. (A to F) To assess if binding was altered after escape virus mutations were introduced in the recombinant MACV GPC, ELISAs were performed using MACV GPC, MACV GPC K170N, and MACV GPC F226L with the antibodies indicated at the top of each panel. MAb KL-AV-2A1 was used as the positive control.
DISCUSSION
At present, there are no approved vaccines or therapeutics available for MACV. According to anecdotal evidence, the vaccine strain of JUNV, Candid#1, might provide some protection against MACV (8). Furthermore, a MAb isolated from a Candid#1 vaccinee showed strong cross-neutralization from JUNV to MACV (25), while other isolated human and murine (26) anti-JUNV GPC monoclonal and polyclonal antibodies (27) did not. However, efficient induction of cross-reactive antibodies against these two viruses is hindered by their low amino acid identity and similarity. Specifically, the RBSs in these viruses are notably dissimilar, and the presence of an additional loop (loop 10) in the MACV GPC compared to the structure of the JUNV GPC drastically changes RBS architecture (23, 28). Given the high level of pathogenicity of MACV and its ability to sporadically spill over from rodents to humans, vaccines and therapeutics against MACV are urgently needed. In this study, DNA vaccination with full-length MACV GPC in mice led to isolation of 25 unique antibodies. Five of these antibodies had high neutralizing activity against recombinant VSV-MACV and authentic MACV. Furthermore, these neutralizing MAbs had effector functions and exhibited strong ADCC reporter activity in vitro. Hence, vaccine regimens consisting of DNA expressing MACV GPC followed by a recombinant GPC boost could be an effective way to combat MACV as this strategy led to induction of GPC-specific neutralizing MAbs with Fc-mediated effector functions (29).
For JUNV, plasma/IgG transfer has been shown to be effective if initiated before day 8 post-symptom onset (30), and MAb therapeutics for this virus are currently under development (31). The five neutralizing MAbs described here have potent neutralizing activity against not only the recombinant VSV-MACV but also the authentic MACV and could be developed for use as therapeutics to more quickly and safely resolve human infections with MACV.
Over the years, several neutralizing MAbs against New World arenaviruses have been reported (25–28, 31–33). Typically, due to low amino acid conservation between GPCs these MAbs are relatively species specific (25–28, 31–33). The presence of an additional loop (loop 10) in MACV which is absent from other NW arenavirus RBSs adds to this problem (23, 24). Only one human MAb from a Candid#1 vaccinee has recently been reported that effectively cross-neutralizes MACV. This MAb, CR1-07, mostly avoids making contacts with loop 10 and therefore can bind to both viruses (25). Based on escape mutagenesis, all five neutralizing MAbs in this study inhibit the virus by blocking interactions between the RBS and its receptor, TfR1, similar to other previously described anti-JUNV GP antibodies (34). It seems that, for all five MAbs, interactions with loop 10 are important, which explains their species specificity. The five identified neutralizers had potent activity in in vitro neutralization assays. It is interesting that the IC50 values we obtained for authentic MACV were lower for KL-MM-1F9 and KL-MM-1B6 than the IC50 values for VSV-MACV. Usually, IC50 values for antibodies are higher for authentic viruses than for recombinant or pseudotyped viruses, reflecting differences in binding and surface compatibility (25, 35). We also report that our antibodies are able to elicit ADCC reporter activity in vitro. This is the first time ADCC activity of antibodies has been reported against NW arenaviruses. Surprisingly, only antibodies which neutralized the virus had high ADCC reporter activity. This is a distinct finding compared to activity of MAbs elicited against other virus proteins, including ebolavirus GP and influenza virus HA. Against these viruses, nonneutralizing MAbs are often able to protect through various effector functions (18, 20, 36–39). Fc-mediated effector functions could play an important role in protection from MACV infection, and this avenue needs to be further explored.
In the course of this study, we took advantage of Stat2−/− mice, which allowed for the efficient replication of VSV. In Stat2−/− mice generated on the B6 background, VSV-MACV was not lethal (even at high doses), but virus was found in serum and in several tissues such as spleen, kidney, liver, and brain. Tissue tropism in these mice is consistent with findings from nonhuman primates such as adult marmosets (Saguinus geoffroyi) (40). In these animals, MACV infection led to virus being present in several tissues, such as brain, spleen, kidney, heart, and liver. In another study conducted in African green monkeys (Cercopithecus aethiops), virus was found in liver, spleen, and kidney tissues (41). Of note, the virus was initially isolated from a human spleen (4). However, viral titers were lower on day 6 in Stat2−/− mice, indicating that the virus was effectively cleared. Authentic MACV has been shown to be lethal in other mouse models such as Stat1−/− and interferon αβ/γ receptor knockout (IFN R αβ/γ−/−) mice (7, 22). It would be interesting to check whether authentic MACV could cause lethal disease in Stat2−/− mice. The five isolated neutralizing MAbs led to reduction in viral titers in the spleen in the Stat2−/− mouse model when mice were challenged with VSV-MACV. However, more work in another animal model is needed to better study the effect of antibody administration in prophylactic therapeutic settings. Specifically, experiments in immunocompetent animal models, e.g., guinea pigs or nonhuman primates (40–47), might shed light on the therapeutic efficacy of the isolated MAbs. These models might also be a better reflection of human disease since it is not clear if MACV is using TfR1 for entry in mice (48, 49).
In summary, we isolated 25 anti-MACV GPC MAbs with various degrees of cross-reactivity. Five of the MAbs had strong neutralizing activity but were species specific. Escape mutation analysis suggests that these MAbs neutralize the virus by binding to the receptor binding site and blocking interactions between GPC and its receptor TfR1. The MAbs were protective in a Stat2−/− mouse VSV-MACV challenge model, and their potential as prophylactic and therapeutic treatment options for MACV should be further explored in the future.
MATERIALS AND METHODS
Cells and viruses.Vero.E6 cells (ATCC CRL-1586) were grown and maintained in culture using Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies). DMEM was supplemented with 10% fetal bovine serum (FBS; HyClone), antibiotics (100 units/ml penicillin–100 μg/ml streptomycin [Pen-Strep; Gibco]), and buffer solution [1 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES); Gibco]. Sf9 cells were grown and maintained in TNM-FH medium (Gemini Bioproducts) with antibiotics (Pen-Strep) and 10% FBS, while BTI-TN-5B1-4 cells were grown in serum-free SFX insect medium (HyClone) supplemented with antibiotics (Pen-Strep). The process has been described previously (50). VSV-MACV, a recombinant VSV expressing the MACV glycoprotein on the viral surface, was generously provided by Sean Whelan at Harvard Medical School (51).
Replication-competent recombinant VSV (VSV-LASV) was provided by Heinrich Feldmann at Rocky Mountain Laboratories (RML), National Institute of Allergy and Infectious Diseases (NIAID), and has been described previously (52–54). VSV-MACV was grown and propagated in Vero.E6 cells. Vero.E6 cells were infected with VSV-MACV diluted in 1× minimal essential medium (MEM; Gibco) supplemented with antibiotics, HEPES, glutamine (200 mM l-glutamine; Gibco), and sodium bicarbonate (sodium bicarbonate 7.5% solution; Gibco) for 1 h. The virus was removed, and 1× MEM was added to the cells. Cytopathic effect was assessed after 24 h, and supernatant from these cells was collected after centrifugation at 4,000 × g. Virus was stored at –80°C, and the titer was measured via a classic plaque assay as described below.
Plaque assay.Various concentrations of the viruses were prepared via 10-fold dilutions and added to confluent monolayers of Vero.E6 cells for 1 h. Virus was removed from cells, and cells were overlaid with MEM containing 2% Oxoid agar (Oxoid purified agar; Thermo Scientific). Cells were incubated at 37°C for 24 h and were then fixed with 3.7% paraformaldehyde (PFA) for 1 h. Afterwards, the agar overlay was removed, and crystal violet solution (Sigma-Aldrich) was added to the cells for 20 min and later removed with water. Finally, plaques were counted, and the viral titer was measured.
Recombinant arenavirus GPCs.All recombinant proteins used in the experiments have been described previously. The ectodomain of JUNV XJ13 glycoprotein (GenBank accession number ACO52428.1) was cloned into the same insect cell expression vector with a trimerization domain and expressed via the baculovirus expression system (55).
Generation of mouse monoclonal antibodies.Six- to 8-week-old, female BALB/c mice (Jackson Laboratories) were vaccinated with pCAGGS, a mammalian expression vector, containing the complete open reading frame of the MACV Carvallo strain glycoprotein (GenBank accession number KM198592.1). DNA plasmid administration was performed by injection via the intramuscular route followed by application of an electrical stimulus immediately afterwards (TriGrid delivery system; Ichor Medical Systems). Mice were vaccinated twice sequentially with 40 μg of DNA, and 3 days prior to harvesting of the spleen, mice were boosted via the intraperitoneal route with 100 μg of recombinant MACV glycoprotein supplemented with 10 μg of poly(I·C). To perform the fusion, one mouse was euthanized, and the spleen was removed. The hybridoma fusion was performed using established protocols (56, 57). Splenocytes were washed three times and then mixed with SP2/0 myeloma cells at a ratio of 5:1. Cell fusion was performed using polyethylene glycol (PEG; Sigma-Aldrich), and the fused cells were grown on semisolid Clonacell-HY medium D (StemCell Technologies) for 10 days. Hybridoma clones were picked and cultured in 96-well cell culture plates for 5 days in medium E (StemCell Technologies). Clones were screened for reactivity against MACV glycoprotein using an enzyme-linked immunosorbent assay (ELISA), which is described below. Clones which had reactivity to MACV glycoprotein were then typed via a Pierce rapid antibody isotyping kit (Life Technologies). IgG-secreting clones were kept in culture in medium E and then adapted to serum-free hybridoma medium (Hybridoma-SFM; Gibco). Once the culture volume reached 300 ml, the supernatant was collected, and each antibody was purified using protein G Sepharose 4 Fast Flow (GE Healthcare). This protocol has been described previously in detail (58). All animal procedures were performed in accordance with protocols approved by the Icahn School of Medicine at Mount Sinai Animal Use and Care Committee.
ELISA.Flat-bottom 96-well plates (Immulon 4 HBX; Thermo Scientific) were coated with recombinant protein prepared in 1× coating solution (10× coating concentrate; Seracare) at a concentration of 2 μg/ml overnight (50 μl/well). The next day, the coating solution was discarded, and 100 μl/well of 3% nonfat milk prepared in phosphate-buffered saline (PBS) containing 0.1% Tween 20 (TPBS) was added onto the plates. Next, the blocking solution was discarded, and 100 μl/well of various antibody dilutions starting at 30 μg/ml prepared in TPBS containing 1% nonfat milk was added to the plates. The plates were then washed rigorously with 250 μl/well TPBS three times. Next, 100 μl/well of horseradish peroxidase (HRP)-labeled anti-mouse IgG (GE Healthcare) was used as a secondary antibody, and this was also prepared in TPBS containing 1% nonfat milk. The plates were then washed three times with TPBS to get rid of any unbound secondary antibody. Plates were developed using 100 μl/well of SigmaFast o-phenylenediamine dihydrochloride (OPD; Sigma-Aldrich) for 10 min. The reaction was stopped using 50 μl/well of 3 M hydrochloric acid (HCl), and the optical density at 490 nm was measured via a Synergy 4 (BioTek) plate reader. To screen hybridomas for reactivity, 50 μl of each hybridoma supernatant was added as primary antibody, but the remaining part of the protocol was the same as described previously.
Phylogenetic analysis.The phylogenetic tree of the different GPCs shown in Fig. 1 was built using the neighbor-joining method. Amino acid sequences of the GPCs of MACV (GenBank accession number KM198592.1), JUNV XJ13 (GenBank accession number ACO52428.1), TCRV (GenBank accession number M20304), GTOV (GenBank accession number AY129247.1), TAMV (GenBank accession number AF485263), and LASV (GenBank accession number GU481072.1) were aligned using Clustal Omega. The tree was visualized and labeled in Figtree, version 1.4.1.3.
Immunofluorescence assays.Approximately 50,000 cells/well of Vero.E6 cells were plated in a 96-well cell culture plate and infected at a multiplicity of infection (MOI) of 1 overnight with VSV-MACV. The next morning, cells were fixed with 3.7% PFA for 1 h. Next, 100 μl/well of blocking solution, as described above in the ELISA section, was added to each well. The cells were then stained with 100 μl of 30 μg/ml of each respective antibody for 1 h. The cells were then washed with TPBS three times. Next, 100 μl/well of a 1:1,000 dilution of goat anti-mouse IgG heavy plus light chains (H+L)-Alexa Fluor 488 (Abcam) was added as secondary antibody for 1 h. The cells were again washed three times with TPBS. Finally, 100 μl of PBS was added to each well, and immunofluorescence was observed under a microscope (Olympus IX-70). This protocol is adapted from previously described methods (59, 60).
Western blotting.The Western blotting protocol was adapted from previously described methods (60, 61). Recombinant MACV GPC was used for this analysis. Ten nanograms of recombinant MACV glycoprotein alongside a negative control (influenza A virus H2 protein) was loaded onto a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel. Samples were prepared by mixing the respective protein with 2× Laemmli buffer containing 2% beta-mercaptoethanol (βME) at a 1:1 ratio. Samples were then heated to 100°C for 15 to 20 min and loaded onto an SDS-PAGE gel. The gel was then transferred to a nitrocellulose membrane, and each blot was first blocked for 1 h and then stained with 30 μg/ml of each respective antibody for 1 h. Membranes were then washed with TPBS twice for 5 min, and then anti-mouse IgG (whole molecule)–alkaline phosphatase (AP) antibody produced in goat (Sigma-Aldrich) was added onto the blot at a 1:3,000 dilution. Membranes were again washed three times with TPBS for 5 min and then developed using an AP conjugate substrate kit (Bio-Rad). Since the recombinant proteins have a hexahistidine tag, an anti-His antibody (TaKaRa Bio Company) was used as a positive control to check for the presence of protein on the membranes.
PRNA.Plaque reduction neutralization assays (PRNAs) were run in duplicates using confluent monolayers of Vero.E6 cells. Antibodies were serially diluted 1:5 starting at 100 μg/ml, and 50 μl of virus was added at a concentration of 1,500 PFU/ml. This antibody-virus mixture was incubated and shaken for 1 h at room temperature. Next, the mixture was added onto Vero.E6 cells and incubated at 37°C for 1 h. After 1 h, cells were overlaid with 2% Oxoid agar (described above) containing various dilutions of each antibody. The monolayer was fixed and stained with crystal violet, and the plaques were counted.
PRNAs with pathogenic recombinant MACV were performed at the University of Texas Medical Branch (UTMB) in a BSL-4 facility. Antibodies (200 μg/ml) were 5-fold diluted in DMEM–2% FBS (final dilution, 0.064 μg/ml), mixed with an equal volume of diluent containing 80 PFU of rMACV (Carvallo), and incubated for 1 h at 37°C in a CO2 incubator. The mixtures were then used to infect Vero.E6 monolayers grown on 12-well plates. After the incubation for 1 h at 37°C, the inoculum was replaced with tragacanth overlay (0.6% tragacanth in MEM containing 2% FBS and 1% Pen-Strep) and incubated for 8 days. The plates were fixed and stained with 1% crystal violet in 10% formalin to count distinct plaques. IC50 values were calculated from each PRNA graph using four-parameter logarithmic nonparametric regression, using a range from 0% to 100%.
ADCC reporter assay.Antibody-dependent cell-mediated cytotoxicity (ADCC) reporter activity of each antibody was assessed using an ADCC reporter bioassay kit (Promega). The assay was performed on white-bottom, 96-well cell culture plates (Corning Costar) with 50,000 Vero.E6 cells per well. Cells were infected with VSV-MACV diluted in MEM at an MOI of 1 overnight. The next morning, the medium containing virus was removed from the cells. Twofold antibody dilutions were prepared starting at 100 μg/ml, and these dilutions were added onto the cells. Additionally, 75,000 effector cells/well were added onto the cells, and cells were incubated at 37°C for 6 h. At the end of the incubation period, luciferase substrate supplied in the kit was added, and luminescence was measured using a Synergy Hybrid Reader (BioTek).
Generation of escape mutant viruses.Confluent monolayers of Vero.E6 cells were prepared, and 1× the half-maximal inhibitory concentration (IC50) of each MAb along with VSV-MACV at an MOI of 1 was added onto the cells for 24 h. The supernatant from this initial passage was collected, and 200 μl of this passage was then passaged onto new Vero.E6 cells with 2× IC50 of each MAb. This was continuously done, and antibody concentration was doubled until 128× IC50 was reached. The final passage was collected, and a classic plaque assay was performed with this final passage. Three distinct plaques were picked from each antibody escape virus (EV) and grown in culture. RNA was extracted from each plaque and submitted for next-generation sequencing. cDNA was generated with random hexamers and SuperScript III (ThermoFisher) according to the manufacturer’s instructions. The complete coding region of the MACV GPC was amplified using the primer pair 5′-CAGAGATCGATCTGTTTCCTTGAC-3′ (forward) and 5′-GGTTCAAACATGAAGAATCTGTGTGC-3′ (reverse) that produced an ∼1,600-bp PCR amplicon. Each sample was barcoded twice and further amplified using a sequence-independent single primer amplification approach (62). Samples were then pooled and sequenced using an Illumina MiSeq instrument (300-bp paired-end reads). Contigs were then assembled and analyzed using CLC Genomics Workbench, version 11.
In vivo studies in Stat2−/− mice.All animal procedures in this study were performed in accordance with the Institutional Animal Care and Use Committee (IACUC) guidelines. Six- to 8-week-old female and male Stat2−/− mice were mixed and randomized. Animals (n = 3) were infected with 1,000 PFU of VSV-MACV each via the intraperitoneal route, and organs were harvested at day 3 and day 6 to study viral tropism. Plaque assays were performed, as described above, using each organ homogenate. Each organ was homogenized using a BeadBlaster 24 (Benchmark) homogenizer. Experimental design was adapted from earlier published work (19).
To study the effect of each antibody, a similar experiment was performed as described above (n = 4 for each MAb at each time point). Due to limited availability of mice, a running negative control was used. However, 10 mg/kg of each antibody was administered via the intraperitoneal route 2 h prior to infection. At day 3 and day 6, organs were harvested and processed for use in a plaque assay to assess viral load. The protocol was published in greater detail previously (19).
Structural modeling.Using PyMOL, version 2.3, the MACV GP1 trimer (PDB accession number 5W1M [25]) was aligned and overlaid onto LASV GPC (PDB accession number 5VK2 [63]). In Fig. 5, the LASV GPC is shown in gold while the MACV GP1 is shown in gray. The alignment of MACV GP1 with the LASV GPC trimer yielded a root mean square deviation (RMSD) value of 5.41. Using PyMOL, version 2.3, the residues which were mutated in the escape viruses are indicated and labeled in the respective colors. The receptor binding site (RBS) is depicted in green while loop 10 of MACV GP1 is shown in orange.
Statistical analyses.Data were analyzed using GraphPad Prism, version 7. For ELISA results, the minimal binding concentration (MBC) was calculated as the last concentration at which antibody binding was above the average of background values multiplied by 3 times the standard deviation. Statistical significance was calculated using a one-way analysis of variance (ANOVA) with multiple comparisons in GraphPad Prism, version 7.
ACKNOWLEDGMENTS
We thank Sean Whelan for providing the VSV-MACV, which was an invaluable reagent for this study. We thank Erica Ollmann Saphire for advice and encouragement.
This work was partially supported by institutional seed funding to F.K. and Federal funds to G.S.T. from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services, under award number U19AI110819.
FOOTNOTES
- Received 9 October 2019.
- Accepted 1 December 2019.
- Accepted manuscript posted online 4 December 2019.
- Copyright © 2020 American Society for Microbiology.
