> top > docs > CORD-19:a8448e0c9df61442dd233ee8ee838ea3c7e59a63

CORD-19:a8448e0c9df61442dd233ee8ee838ea3c7e59a63 JSONTXT

Antibody Treatment of Ebola and Sudan Virus Infection via a Uniquely Exposed Epitope within the Glycoprotein Receptor-Binding Site Antibody Treatment of Ebola and Sudan Virus Infection via a Uniquely Exposed Epitope within the Glycoprotein Receptor-Binding Site Abstract Graphical Abstract Highlights d Monoclonal antibody FVM04 blocks interaction of ebolavirus with its host receptor d FVM04 cross-neutralizes and protects against EBOV and SUDV in rodents d A new ZMapp version that includes FVM04 protects guinea pigs from EBOV and SUDV d Specific GP mutations expose epitopes for FVM04 and other cross-neutralizing mAbs In Brief Howell et al. examine a mAb, FVM04 , that binds the ebolavirus receptor-binding site and find that FVM04 protects against EBOV and SUDV. When combined with two ZMapp TM components, the antibody cocktail retains EBOV protection similar to that of ZMapp TM and extends protection against SUDV. Specific glycoprotein mutations that enhance the exposure of cross-neutralizing epitopes are described. Previous efforts to identify cross-neutralizing antibodies to the receptor-binding site (RBS) of ebolavirus glycoproteins have been unsuccessful, largely because the RBS is occluded on the viral surface. We report a monoclonal antibody (FVM04) that targets a uniquely exposed epitope within the RBS; cross-neutralizes Ebola (EBOV), Sudan (SUDV), and, to a lesser extent, Bundibugyo viruses; and shows protection against EBOV and SUDV in mice and guinea pigs. The antibody cocktail ZMapp TM is remarkably effective against EBOV (Zaire) but does not cross-neutralize other ebolaviruses. By replacing one of the ZMapp TM components with FVM04, we retained the anti-EBOV efficacy while extending the breadth of protection to SUDV, thereby generating a cross-protective antibody cocktail. In addition, we report several mutations at the base of the ebolavirus glycoprotein that enhance the binding of FVM04 and other cross-reactive antibodies. These findings have important implications for pan-ebolavirus vaccine development and defining broadly protective antibody cocktails. Filoviruses are the causative agents of severe hemorrhagic fever in humans and nonhuman primates (NHPs) (Kuhn et al., 2014) . Members of the family Filoviridae include two marburgviruses: Marburg virus (MARV) and Ravn virus (RAVV), and five ebolaviruses: Ebola virus (EBOV), Sudan virus (SUDV), Bundibugyo virus (BDBV), Reston virus (RESTV), and Taï Forest virus (TAFV) (Kuhn et al., 2014) . The EBOV (Zaire) has caused the largest number of outbreaks, including the 2014 EBOV disease (EVD) epidemic that led to over 28,637 cases and 11,315 deaths. Due to the higher frequency of outbreaks caused by EBOV, most efforts toward vaccine and therapeutic development have focused on this agent. Several studies have shown remarkable efficacy of antibody therapeutics against EBOV Marzi et al., 2012; Olinger et al., 2012; Pettitt et al., 2013; Qiu et al., 2012a Qiu et al., , 2012b Qiu et al., , 2013a Qiu et al., , 2014 . However, until recently (Bounds et al., 2015; Flyak et al., 2016; Frei et al., 2016; Holtsberg et al., 2015; Keck et al., 2015) , the development of cross-protective monoclonal antibodies (mAbs) targeting multiple species of ebolavirus has been lagging behind. The filovirus surface glycoprotein, comprising disulfide-linked subunits GP1 and GP2, is the primary target for vaccines and immunotherapeutics (Marzi and Feldmann, 2014) . The crystal structures of the trimeric EBOV GP 1,2 spike (henceforth termed GP) in complex with KZ52 (Lee et al., 2008) , a neutralizing mAb derived from an EVD human survivor (Maruyama et al., 1999) , as well as SUDV GP in complex with the neutralizing mouse mAb 16F6 (Dias et al., 2011) , have revealed a key mechanism of neutralization. The three GP1 subunits form a chalice-like structure, with GP2 wrapping around GP1 and the N terminus of GP1 forming the base of the chalice (Lee et al., 2008) . Both KZ52 and 16F6 contact residues within GP1 and GP2 at the base and neutralize the virus by blocking the viral fusion with the endosomal membrane (Dias et al., 2011; Lee et al., 2008) . When administered prophylactically or 1 hr after infection, KZ52 protected guinea pigs from lethal EBOV challenge (Parren et al., 2002) . However, in a single study, KZ52 did not protect against EBOV in NHPs at the tested dosing and regimen (Oswald et al., 2007) . Several recent studies have revealed that effective post-exposure protection against EBOV in primates requires a cocktail of mAbs (Pettitt et al., 2013; Qiu et al., 2012a Qiu et al., , 2013a or a combination of mAbs and interferon alpha (IFNa) (Qiu et al., 2013b (Qiu et al., , 2013c . Further testing of various combinations in the guinea pig model of EBOV infection identified a highly effective cocktail of three EBOV-specific mAbs, known as ZMapp TM . ZMapp TM showed 100% efficacy in NHPs when treatment was initiated as late as 5 days post-infection (dpi) . Single-particle electron microscopy (EM) reconstructions of GP complexed with individual ZMappTM components (c2G4, c4G7, and c13C6) revealed two sites of vulnerability on the EBOV GP and elucidated the structural basis for their remarkable efficacy (Murin et al., 2014) . Of the three components of ZMapp TM , c2G4 and c4G7 target an epitope shared with KZ52 at the ''base'' of the chalice near the interface of GP1 and GP2, whereas c13C6 binds to a highly glycosylated domain on the top of a GP molecule known as the glycan cap (Davidson et al., 2015; Murin et al., 2014) . While the combination of the base and glycan cap binders thus far appeared to be most effective against EBOV, these antibodies are virus specific, and it is not clear if the same paradigm can be applied to broadly protective immunotherapeutics. Although the epitopes engaged by EBOV-specific KZ52 and SUDV-specific 16F6 overlap by ten residues (Dias et al., 2011; Lee et al., 2008) , these base binders do not crossreact with other ebolaviruses. Neutralizing antibodies targeting the receptor-binding site (RBS) have been described for several viruses, including influenza (Lee and Wilson, 2015) , HIV (Georgiev et al., 2013) , SARS (severe acute respiratory syndrome) coronaviruses (Coughlin and Prabhakar, 2012), and Chikungunya virus (van Duijl-Richter et al., 2015) . However, no neutralizing antibodies have been identified that would target the RBS within the ebolavirus glycoproteins. The filovirus RBS consists of a relatively exposed hydrophilic ''crest'' and a hydrophobic ''trough'' that is exposed on marburgvirus but occluded by the glycan cap in EBOV GP Wang et al., 2016) . RBS-binding mAbs were recently identified for marburgvirus and shown to bind to the RBS trough . Some of these antibodies also bind to EBOV GP but only after the glycan cap is proteolytically removed by thermolysin , a process that mimics the cathepsin-mediated cleavage in endosomes . In contrast, no antibodies have been reported to date that bind the prominent crest region of the RBS. Recently, we reported that pan-ebolavirus and pan-filovirus antibodies, including two broadly neutralizing mAbs isolated from mice and macaques that were immunized with a mixture of engineered glycoproteins for EBOV, SUDV, and MARV and boosted with virus-like particles (VLPs) for the three viruses (Holtsberg et al., 2015; Keck et al., 2015) . In contrast to the base binders, these antibodies all bind, to the apex of the GP trimer, either the inner chalice or the glycan cap. Here, we report that one of these mAbs, FVM04, binds to the tip of the RBS crest and blocks the interaction of EBOV GP with its endosomal receptor Niemann-Pick C1 (NPC-1). EM reconstructions of FVM04 complexed with EBOV GP show a unique asymmetric mode of engagement with a single antibody per trimer. FVM04 neutralizes EBOV, SUDV, and, to a lesser extent, BDBV and protects mice and guinea pigs from lethal challenge with EBOV and SUDV. Furthermore, replacement of one of the base binders in ZMapp TM with FVM04 retains the potency of the cocktail toward EBOV while expanding the protective breadth of the cocktail to include SUDV. FVM04 Binds to an Exposed Region of the Filovirus RBS We recently described several macaque-derived pan-ebolavirus mAbs (Keck et al., 2015) . One of these mAbs (FVM04) neutralized both EBOV and SUDV and showed significant efficacy in mice when administered at two doses, one on the day of challenge and one 3 dpi (Keck et al., 2015) . Our studies suggested that FVM04 targets a conformational epitope shared among all ebolaviruses with a low level of cross-reactivity to MARV (Keck et al., 2015) . In order to define the FVM04 epitope, we used an alanine scanning approach, where FVM04 binding was evaluated against a ''shotgun mutagenesis'' mutation library of EBOV GP with 641 of 644 target residues individually mutated. Human HEK293T cells were transfected with the entire library in a 384-well array format (one clone per well; Figure 1A ) and assessed for reactivity to FVM04 by high-throughput flow cytometry. The epitope mapping identified EBOV GP residues K115, D117, and G118 as critical for FVM04 binding (Figure 1 ). Alanine substitutions at these residues reduced FVM04 binding to 29%, 1%, and 2% of wild-type, respectively, suggesting that these residues constitute key contact sites for FVM04, with D117 and G118 having the greatest energetic contribution to FVM04 binding ( Figures 1B and 1C ). In contrast, binding of two other panebolavirus antibodies FVM02 and FVM09 (Keck et al., 2015) was not affected by these mutations ( Figure 1C ). The putative epitope of FVM04 is positioned in a previously described region with a crest-and-trough morphology (Bornholdt et al., 2016; Hashiguchi et al., 2015) within the RBS and constitutes the tip of the hydrophilic crest (red dotted outline in Figure 1D ), which interacts with a loop from the endosomal filovirus receptor NPC1 (Wang et al., 2016) . In contrast to this exposed tip of the crest, the trough is lined with hydrophobic residues and occluded by the b14-b15 loop within the glycan cap (black dotted outline in Figure 1D ). This occlusion explains why trough-binding mAbs do not bind and neutralize EBOV unless the glycan cap is removed by proteolysis . Interestingly, while FVM04 binds well to EBOV GP, this binding was moderately enhanced by alanine substitution of N238, T240, N257, T259, N268, and T270, mutations that delete three out of four glycosylation sites on the glycan cap ( Figure S1 ), suggesting that these glycans may modestly interfere with FVM04 binding. Nonetheless, these data show that FVM04 represents a prototypic pan-ebolavirus antibody that recognizes a uniquely exposed epitope within the RBS. The crest region is highly conserved among all ebolavirus species, and the three residues critical for FVM04 binding are 100% identical among ebolaviruses ( Figure 1E ). An overlay of the crystal structures of EBOV GP (Lee et al., 2008) with the structure of thermolysin-cleaved GP (GP CL ) (Bornholdt et al., 2016; Wang et al., 2016) and SUDV GP (Dias et al., 2011) showed high structural conservation within this region and prominent exposure of these residues on the surface of GP CL ( Figure 1F ) and SUDV GP (Figure 1G) . While only one out of the three residues in this putative epitope match between EBOV and MARV ( Figure 1E ), this region still shows a high degree of conformational similarity between EBOV and MARV, with the three critical EBOV residues overlaying well with MARV D99, S101, and G102 ( Figure 1H ). This may explain the low affinity of FVM04 for MARV GP as we previously reported (Keck et al., 2015) . EBOV-neutralizing antibodies KZ52 (Lee et al., 2008; Parren et al., 2002) , 2G4, and 4G7 (Murin et al., 2014; Qiu et al., 2012b) bind to overlapping epitopes at the base of the GP trimer consisting of residues from both GP2 and the base of GP1 (Figure 2A ). In contrast, FVM04 appears to bind the apex of the trimer, between the glycan cap and the trimer center (Figure 2A ), therefore representing a distinct class of neutralizing antibodies. In addition to epitope-mapping information (loss-of-function analysis), the alanine-scanning experiments revealed a striking difference between the GP-binding patterns of these ''base-binding'' and ''apex-binding'' classes of neutralizers. While the binding of the base binder KZ52 to all individual mutants remained consistently below 150% of wild-type GP ( Figure 2B ), several single-alanine mutations of GP had a dramatic enhancing effect (as high as 200%-300%) on FVM04 binding ( Figure 2C ). Out of 217 amino acids forming the base, alanine mutation of 23 residues increased FVM04 binding to GP by more than 2-fold ( Figure 2C ; Table S1 ). Most of these residues are hydrophobic, highly networked, and not surface exposed ( Figure S2 ; Table S1 ). In addition to the 23 base residues, mutation of R498 and R501, within the furin cleavage site separating GP1 and GP2 (Volchkov et al., 1998) , also led Human HEK293T cells expressing the mutation library were tested for reactivity to mAbs using an Intellicyt flow cytometer. A typical reactivity pattern (red wells) is shown for a representative assay plate. Eight positive (wild-type EBOV GP) and eight negative (mock-transfected) control wells were included on each plate. (B) The library was tested for reactivity with FVM04. Clones with <30% binding relative to that of wild-type (WT) EBOV GP yet >65% reactivity for a control mAb were initially identified to be critical for FVM04 binding. (C) Mutation of three individual residues reduced FVM04 binding (red bars) but had little effect on the binding of FVM02 and FVM09 (gray bars; Keck et al., 2015) . Bars represent the mean and range ([maxÀmin]/2) of at least two replicate data points. (D) The FVM04 binding residues are shown in red in the crystal structure of EBOV GP. The glycan cap is shown in cyan, and the attachment points for N-linked glycans in orange. The GP1 core is shown in purple, and parts of GP2 are seen in yellow. The RBS crest is shown in red outline and the occluded trough region is shown in black outline. (E) Sequence homology between filoviruses within the RBS crest region containing putative FVM04 epitope. Identical sequences among ebolavirus species and between ebolavirus and marburgvirus are shown in red. The FVM04 binding site is boxed. (F-H) The EBOV GP monomer is depicted as a cartoon overlay with GP CL (F), SUDV GP (G), and MARV RAVN GP (H). Putative critical EBOV contacts made by FVM04 are shown as sticks (red) overlaid with corresponding contact residues from the overlay structure (green). to an increase over 2-fold in FVM04 binding ( Figure 2C ; Table S1 ). An alanine mutagenesis scan of several other cross-reactive apex-binding antibodies (m8C4, 4B8, FVM09, FVM17, and FVM20) (Holtsberg et al., 2015; Keck et al., 2015) also revealed a similar enhancement of binding to the same alanine mutants that enhance FVM04 binding (Table S1 ). These data suggest that specific mutations in the GP base may have a global impact on the exposure of cross-reactive epitopes in the GP1 head domain. This finding may have major implications for the design of pan-ebolavirus vaccines. To further characterize FVM04 binding to EBOV GP, FVM04 Fab in complex with GPDMuc was analyzed by negative-stain EM. The binding location of FVM04 revealed an epitope consistent with the crest residues derived by mutagenesis studies. The class averages suggest that only one FVM04 Fab binds to each GP trimer (Figures 3A and 3B; Figure S3 ). The binding of one FVM04 Fab per trimer was also further confirmed by size exclusion chromatography-multiangle light scattering (SEC-MALS) analysis ( Figure S4 ). It is likely that the binding orientation and proximity to the 3-fold axis precludes additional FVM04 Fabs from binding. In contrast, the previously characterized glycancap-binding antibody c13C6, which binds in a nearby region to FVM04, has a greater occupancy (two to three Fabs per GP trimer) (Murin et al., 2014 ) ( Figure 3C ). Although several attempts were made to generate a 3D reconstruction of FVM04 binding to GP, the data failed to converge on an interpretable model, suggesting that the crest epitope is flexible. We used several assays to evaluate the cross-neutralizing activity of FVM04 against multiple ebolaviruses. These assays included replication-incompetent vesicular stomatitis viruses (VSVs) pseudotyped with filovirus GP and expressing Luciferase (VSV-GP-Luc), as well as replication-competent VSV pseudotypes expressing GFP (VSV-GP-GFP) and wild-type live virus. While the first assay identifies only direct-entry inhibitors, antibodies inhibiting either the entry or other stages of viral replication can be identified by the second assay or live virus. As shown in Figure 4A , FVM04 effectively neutralized the entry of both VSV-pseudotyped EBOV and SUDV, with half maximal effective concentration (EC 50 ) values of 3.4 and 4.3 mg/ml, respectively. We also examined whether the FVM04 neutralization is dependent on bivalent binding of full immunoglobulin (Ig) G or whether FVM04 Fab fragment would also mediate neutralization. EBOV entry was effectively inhibited by FVM04 Fab with an EC 50 similar to that of full IgG; however, the neutralizing potency of FVM04 Fab toward SUDV was reduced compared to that of full IgG ( Figure 4B ). FVM04 also effectively neutralized replication-competent VSV pseudotyped with EBOV or SUDV but not TAFV ( Figure 4C ). While FVM04 showed low neutralizing activity toward VSV-BDBV GP-GFP, this neutralization plateaued at 50%, leaving a non-neutralized subset of infectious virions remaining ( Figure 4C ). We further tested the neutralizing activity of FVM04 against authentic (wild-type) EBOV, SUDV, and BDBV, using a plaque reduction neutralization (PRNT) assay under BioSafety Level 4 (BSL-4) containment. The highest neutralizing activity was observed against SUDV, while 3-to 4-fold and 20-to 30-fold higher concentrations of FVM04 were required to effectively neutralize wild-type EBOV and BDBV, respectively ( Figure 4D ). As expected, FVM04 did not neutralize MARV ( Figure 4D ). Previous reports showed that mAbs targeting the trough region of the RBS do not or poorly neutralize VSV-EBOV GP but can neutralize the pseudotyped virus after cleavage with thermolysin (VSV-EBOV GP CL ), which mimics the cathepsin cleavage in endosomes (Bornholdt et al., 2016; Flyak et al., 2015; Hashiguchi et al., 2015) . To this end, we tested whether the crest binder FVM04 could also neutralize the endosomal form of the virus and whether proteolytic removal of the glycan cap and mucinlike domain (MLD) impacts its neutralizing potency. As shown in Figure 4E , FVM04 neutralized both VSV-EBOV GP CL and VSV-SUDV GP CL nearly 100-fold more effectively than the VSV expressing full-length GP (compare with Figure 4C ), suggesting that cathepsin cleavage may further expose the FVM04-binding site. Conversely, no increase in the partial neutralization of BDBV by FVM04 was observed upon thermolysin cleavage ( Figure 4E ). Given the location of the FVM04 epitope and the antibody's dramatically increased neutralizing efficacy upon thermolysin cleavage, we hypothesized that FVM04 blocks binding of cleaved EBOV GP (GP CL ) to its endosomal receptor NPC1. Binding of FLAG-tagged soluble NPC1 domain C to biotinylated VSV-EBOV GP CL immobilized on streptavidin-coated plates was tested in the presence and absence of FVM04. As shown in Figure 4F, FVM04 exhibited concentration-dependent inhibition of NPC-1 binding to GP CL , suggesting that the inhibition of GP binding to its endosomal receptor is one likely mechanism of action for FVM04. We used biolayer interferometry (BLI) to measure the affinity of FVM04 binding to EBOV, SUDV, or BDBV GP ectodomains (GPDTM). FVM04 was bound to protein G sensors and exposed to a range of GPDTM concentrations ( Figure 5 ). The association and dissociation of FVM04 to EBOV and SUDV GPDTM fit a 1:1 binding model ( Figures 5A and 5B ). The K D for EBOV GPDTM was calculated to be 2.2 ± 0.1 nM, with a k a (association rate) of (2.5 ± 0.1) 3 10 4 /M sec and a k d (dissociation rate) of (5.4 ± 0.1) 3 10 À5 /s. The affinity of FVM04 for SUDV GPDTM was slightly higher, with a K D of 9.1 ± 0.2 nM, a k a of (6.0 ± 0.1) 3 10 4 /M sec, and a k d of (5.4 ± 0.1) 3 10 À4 /s. The affinity of FVM04 to EBOV and SUDV GPDTM was also analyzed in an additional experiment using anti-human Fc-coated sensors, and the K D values were found to be similar. Interestingly, the binding of FVM04 to BDBV GPDTM was best described with a 2:1 binding model, with two dissociation constants of K D 1 = 570 ± 50 nM and K D 2 = 31 ± 1 nM ( Figure 5C ). The initial association with FVM04 is within the same range of EBOV GPDTM and SUDV GPDTM, with a k a 1 of (3.8 ± 0.1) 3 10 4 /M sec, but the second association step is 4-fold slower, with a k a 2 of (1.1 ± 0.1) 3 10 4 /M sec. The dissociation rate of BDBV GPDTM from FVM04 is the fastest among all three proteins, with a k d 1 of (2.2 ± 0.1) 3 10 À2 /s, and a k d 2 of (3.3 ± 0.1) 3 10 À4 /s. Collectively, this kinetics analysis offers a biophysical explanation for the differential neutralization profiles between the three viruses. FVM04 exhibits potent neutralization of both EBOV and SUDV, consistent with a high affinity for both glycoproteins in vitro (nanomolar range, K D ), but only a modest neutralization activity against BDBV consistent with poorer binding profiles toward this GP. The difference in binding between EBOV/SUDV and BDBV GP is most evident in the biphasic dissociation phase for BDBV, which is dominated by a dissociation rate (k d ) that is two orders of magnitude faster than for EBOV or SUDV GP. We previously reported the efficacy of FVM04 in a mouse model of EBOV when administered at two doses starting immediately after infection (Keck et al., 2015) . Here, we expanded these studies to post-exposure treatment in both the EBOV model and a recently developed mouse model for SUDV . First, we evaluated the delayed administration of a single dose of FVM04. Groups of ten mice were infected with 100 plaque-forming units (PFU) of mouse-adapted EBOV (MA-EBOV) (Bray et al., 1999) and treated with a single intraperitoneal (i.p.) injection of 10 mg/kg (200 mg per mouse) of FVM04 either 1, 2, or 3 dpi. A single injection of FVM04 at 1 dpi led to full protection from lethal challenge (p < 0.0001; determined by Manel-Cox method), while delayed treatment on day 2 or day 3 (peak of viremia) resulted in 80% (p = 0.0012) and 30% (p = 0.108) protection, respectively ( Figure 6A ). Consistent with survival data, mice treated at 1 dpi showed no weight loss or sign of disease ( Figure 6A ). Mice treated on 2 or 3 dpi lost a maximum of 8% and 10% body weight, respectively, compared to 18% weight loss in the control group, and milder clinical signs of disease were observed ( Figure 6A ). Then, we evaluated the dose response by treating the mice 2 dpi with 10, 5, or 2.5 mg/kg (200, 100, or 50 mg per mouse) of FVM04 or PBS as control. In this study, both 10 and 5 mg/kg FVM04 provided full protection (p < 0.0001), while 70% of the mice receiving 2.5 mg/kg survived the challenge (p = 0.0004) (Figure 6B ). PBS-treated mice lost about 13% of their body weight before succumbing to infection, while mice treated with FVM04 showed less weight loss and less severe disease, as determined by health scores ( Figure 6B ). These data clearly indicate the post-exposure efficacy of FVM04 at relatively low doses. Efficacy of FVM04 was further tested in mice in which the genes for IFNa/IFNb (interferon beta) receptor are knocked out (IFNabR À/À ) . Groups of seven 4-weekold IFNabR À/À mice were infected with 1,000 PFU of SUDV followed by i.p. injection of 10 mg/kg of anti-SUDV GP mAb 16F6 at 1 and 3 dpi, or FVM04 at 1 dpi. A control group of six mice received no treatment after the infection. The SUDV-specific 16F6 fully protected mice with minimal weight loss or signs of disease ( Figure 6C ). Five out of seven mice treated with FVM04 were protected from lethal challenge, while the effect on average weight loss and health scores was not apparent ( Figure 6C ). Efficacy of FVM04 was also examined in guinea pigs using guineapig-adapted EBOV and SUDV (GPA-EBOV and GPA-SUDV, respectively) (Volchkov et al., 2000; Wong et al., 2015) . Four groups of six guinea pigs were challenged with either 1,000 3 LD 50 (the median lethal dose) of GPA-SUDV or GPA-EBOV followed by a single i.p. injection of 5 mg FVM04 ($15 mg/kg) or vehicle control at 1 dpi. Animals were monitored for 16 days for weight change and 28 days for survival. A single injection of FVM04 protected all guinea pigs from GPA-SUDV challenge, while the controls succumbed to infection within 10-13 days (p = 0.0004) ( Figure 7A ). While the controls lost up to 40% body weight after the GPA-SUDV challenge, no weight loss or sign of disease was observed among FVM04-treated animals (Figure 7A ). All DPBS-treated guinea pigs infected with GPA-EBOV succumbed to infection within 6-7 days, while two out of six FVM04-treated animals survived the challenge, and the remaining died between 9 and 11 dpi (p = 0.0012) ( Figure 7B ). The median survival was 7 days for the control group and 11 days for the FVM04-treated group. Control animals lost over 25% of their weight before dying, while the FVM04-treated animals initially gained weight followed by a moderate weight loss ($5%) 10-15 days post-challenge ( Figure 7B ). The partial protection against EBOV is consistent with previous reports indicating that an antibody cocktail is required for effective post-exposure protection against EBOV in guinea pigs and NHPs . ZMapp TM , consisting of the two base binders c2G4 and c4G7 and the glycan cap binder c13C6, was selected for testing in NHPs based on significant, but partial, protection in guinea pigs (four out of six) when administered once at 3 dpi . Based on the aforementioned study, we hypothesized that replacing one of the ZMapp TM components with FVM04 would lead to an effective EBOV cocktail that is also protective against SUDV. We selected the mAb c4G7 to be replaced, since it binds to an epitope closely overlapping the c2G4 epitope Davidson et al., 2015; Murin et al., 2014 ). First, we tested whether FVM04 alone (5 mg) or a cocktail of c2G4/c13C6/cFVM04 (1.6 mg each) would protect against GPA-SUDV when administered at 3 dpi. While all GPA-SUDV-infected control animals died within 10-14 days, all FVM04-treated animals and five out of six animals treated with the cocktail survived the challenge ( Figure 7C ). The protection was highly significant, with p = 0.0008 for both treatment groups compared with the controls. Animals treated with FVM04 exhibited no weight loss, while control animals lost an average of 25% body weight ( Figure 7C ). While the animals treated with the cocktail also showed no weight loss on average ( Figure 7C ), the only fatal case in this group lost 17% of its body weight before dying on day 14 post-infection. The cocktail consisting of FVM04, c13C6, and c2G4 was also tested in the GPA-EBOV model. Four out of six animals treated with a single dose of 5 mg cocktail ($1.6 mg of each component) at 3 dpi survived the challenge, while all control animals succumbed to infection within 7-9 days (p = 0.0061) ( Figure 7D ). The control animals lost an average of 20% of their weight, while the average weight within the cocktail-treated group showed a steady increase over 16 dpi ( Figure 7D ). Of the two cocktailtreated animals that died, one lost about 9% body weight by the day of death (7 dpi), and the second animal actually gained 12% body weight before dying on day 8. As a comparison, Figure 7E shows compiled survival and weight loss data from three studies that we have performed with ZMapp TM (5 mg per animal; n = 20). A survival rate of 67% in guinea pigs for the cocktail of FVM04/c13C6/c2G4 is well within the range of protection afforded by ZMapp TM , as shown here and reported previously . The devastating 2014 EVD epidemic in West Africa is a sobering reminder of the global threat of filovirus infections. This outbreak reached unprecedented dimensions, despite the fact that the majority of efforts to develop vaccines and therapeutics over the past decade was focused on the same ebolavirus that caused this epidemic. Given the uncertainties about the natural reservoir and the zoonotic dynamics of filoviruses, it is impossible to predict the species or location of future outbreaks. Thus, broadly protective vaccines and therapeutics are urgently needed to cope with this emerging threat. Recent reports indicate that mAbs against GP are effective as post-exposure treatments for ebolavirus hemorrhagic fever (Pettitt et al., 2013; Qiu et al., 2014) . ZMapp TM , a cocktail of three mAbs against EBOV GP, is among the most advanced therapeutic candidates and has exhibited remarkable efficacy in symptomatic EBOV-infected NHPs . However, these immunotherapeutic candidates are species specific and mostly target EBOV (Zaire) only. According to the current dogma in the field, a combination of antibodies is required for effective control of EBOV infection, and antibodies in current cocktails include those that target the base of the trimeric GP, as well as those that bind the apex of the trimer in the glycan cap (Murin et al., 2014) . Recently, engineered bispecific antibodies targeting the GP base of both EBOV and SUDV were shown to provide post-exposure protection in mice (Frei et al., 2016) . Despite the structural overlap of 16F6 (SUDV-specific) and KZ52 (EBOV-specific) to date, no cross-binding antibody that targets the GP base epitope on a canonical IgG framework has been described. Thus, antibody cocktails with expanded neutralizing and protective breadth are likely to engage novel epitopes. We have recently identified a set of pan-ebolavirus and pan-filovirus mAbs that target phylogenetically conserved sites within the glycan cap and core GP1, including two mAbs (m8C4 and FVM04) that effectively crossneutralize EBOV and SUDV, the two most divergent filoviruses (Holtsberg et al., 2015; Keck et al., 2015) . Here, we report full characterization of the pan-ebolavirus mAb FVM04 and demonstrate that it targets a uniquely exposed epitope within the RBS and conveys post-exposure protection against both EBOV and SUDV. The receptor-binding unit of filovirus GP has an ocean-wave morphology with a recessed trough and a rising crest (Bornholdt (Bornholdt et al., 2016; Hashiguchi et al., 2015) . In contrast, the wave crest (strands b7 and b9 and their connecting loops in ebolavirus GP) is hydrophilic and contains basic residues that undergo electrostatic interactions with the filovirus host receptor NPC-1 (Bornholdt et al., 2016; Wang et al., 2016) . The recessed trough is occluded by specific interaction of residues from the glycan cap, with the hydrophobic lining of the trough blocking its accessibility on the surface of EBOV GP . Only after cathepsin-mediated cleavage within endosomes removes the glycan cap are the residues within the ebolavirus GP trough unmasked for specific interaction with NPC1 (Bornholdt et al., 2016; Krishnan et al., 2012; . These findings provide a structural explanation for the observation that MARV-neutralizing antibodies such as MR78 and MR72 bind fulllength MARV GP, in which the trough is apparently exposed, but bind and neutralize EBOV only after thermolysin cleavage and removal of the glycan cap . In contrast, the tip of the crest is accessible on the surface of GP trimer (Bornholdt et al., 2016; Hashiguchi et al., 2015) . Our alanine-scanning mutagenesis analysis reveals K115, D117, and G118 (positioned at the tip of the crest) as critical residues for FVM04 binding. Bornholdt et al. recently demonstrated that mutations of K114 and K115 to alanine significantly reduce NPC-1 binding by GP CL , while mutating these residues to glutamic acid completely abrogates GP interactions with NPC-1, suggesting that these basic residues are involved in receptor binding (Bornholdt et al., 2016) . Moreover, the recently solved X-ray crystal structure of a GP CL -NPC-1 domain C protein complex reveals that a loop in NPC1 domain C directly contacts the basic crest (Wang et al., 2016) . Here, we show that FVM04 binding specifically blocks the interaction of GP CL with NPC-1. Lately, Corti et al. (2016) also reported an EBOV-specific antibody that binds to the EBOV GP glycan cap and the RBS region and blocks GP interaction with NPC-1 . Thus, FVM04 represents a distinct cross-neutralizing antibody that binds an exposed epitope within the ebolavirus RBS and neutralizes the virus, at least in part, through blockade of receptor interaction. The primary amino acid sequence of the crest region is highly conserved among all ebolaviruses and modestly homologous between ebolaviruses and marburgviruses. The tip of the crest, which includes FVM04 binding sites, also shows high structural homology between EBOV and SUDV and, to a lesser extent, (C) Three groups of IFNabR À/À mice were infected with 1,000 PFU of SUDV; one group (n = 7) received two i.p. injections of 16F6 at 1 and 3 dpi, a second group received FVM04 once at 1 dpi, and a third group was left untreated. Mice were monitored for 21 days for survival, weight change, and signs of disease. between EBOV and MARV GP. While FVM04 effectively neutralized both EBOV and SUDV, it only exhibited partial neutralization of BDBV. Our previous results indicated that FVM04 binds to BDBV GP with an ELISA EC 50 comparable to those of EBOV and SUDV (Keck et al., 2015) . However, our kinetic analysis in this report showed that FVM04 has a very rapid off rate for BDBV GP, providing a possible explanation for low neutralizing activity toward BDBV. The amino acid sequence of the EBOV and BDBV GP crest regions only differ in positions 112 (glutamic acid in EBOV versus aspartic acid in BDBV) and 116 (proline in EBOV versus alanine in BDBV). In our alanine-scanning analysis, the E112A and P116A mutations in EBOV GP did not alter the binding to FVM04 (106% and 115% of wild-type respectively). However, it is possible that the residues 112 and 116 influence the positioning of the key residues 115, 117, and 118 or are part of the epitope directly but at a lower energetic level that cannot be captured by the cell-surface staining assay. FVM04 did not neutralize VSV pseudotyped with TAFV GP, which also differs from EBOV GP in the same crest residues as BDBV with alanine and valine in positions 112 and 116, respectively. Nonetheless, FVM04 is an effective neutralizer of the most virulent ebolaviruses EBOV and SUDV. An interesting feature of FVM04 is its asymmetric binding to GP in contrast to antibodies such as ZMapp TM components (Murin et al., 2014) and KZ52 (Lee et al., 2008) . Class averages from the analysis of negative-stain EM of FVM04 Fab bound to GPDMuc revealed that it binds with a stoichiometry of one Fab per trimer. The position of FVM04 in EM studies is consistent with the mutational analysis. Together, these results suggest that this mAb binds with a more angled pose than c13C6, which binds at the top of GP as well but perpendicular to the plane of the viral membrane (Murin et al., 2014) . Although we have not quantitatively established with solution studies that only a single FVM04 Fab is capable of binding, even at saturating concentrations, it seems reasonable from the EM-derived images that the angle of approach for a single FVM04 Fab occludes the other potential binding sites on the other two subunits. Further studies are required to elucidate the implications of this binding morphology on neutralization and in vivo efficacy, but a tantalizing possibility is that fewer FVM04 IgG molecules bind the GP trimer than other IgGs on average. Lower requirement for occupancy could, in theory, provide a therapeutic advantage because less antibody would be required for neutralization and render the IgG less susceptible to decoy antigens such as sGP or GP spike on defective particles. For example, studies performed with anti-EBOV equine Ig clearly showed rapid reduction in EBOV-specific equine IgG titer during the peak of viremia in NHPs, while total equine IgG titer remained constant (Jahrling et al., 1996) , suggesting high antibody consumption during infection. Single-site occupancy could also potentially leave more of the GP surface exposed in the mAb-GP complex, providing additional opportunities for engagement of other epitopes by other antibodies in a cocktail. While the FVM04 epitope is accessible, the binding is probably not entirely free of steric hindrance. Our previous report indicated ZMapp TM (n = 20) at 3 dpi. Challenge was performed with 1,000 3 LD 50 of GPA-SUDV or GPA-EBOV as indicated. Survival was monitored for 21 dpi, and weights were monitored for 16 days dpi. that FVM04 binding to GP was slightly increased after removal of MLD and the glycan cap (Keck et al., 2015) , and data in Figure 4E show that FVM04 neutralizes the virus carrying GP CL more effectively than full-length GP. Interestingly, our mutagenesis studies revealed that, in contrast to KZ52, binding of FVM04 was substantially enhanced when certain residues within the base of the GP (GP2 and N terminus of GP1) were mutated to alanine. Most of these mutations also increased binding of other crossreactive antibodies that we had previously reported (Holtsberg et al., 2015; Keck et al., 2015) , such as 4B8, m8C4, FVM09, and FVM20. These residues are mostly hydrophobic and buried in the structure, suggesting that single-point mutations within the base may loosen up the apex, exposing cross-reactive epitopes. This finding may have important implications for the development of pan-ebolavirus vaccines if such mutations can be incorporated into vaccines to improve their potency or breadth of protection. Efficacy studies in mice and guinea pigs showed the high therapeutic value of FVM04. Full protection from EBOV infection was observed 2 dpi with as little as 100 mg per mouse, and a lower dose of 50 mg per mouse showed partial but significant protection. A small-scale study in IFNabR À/À mice showed 71% protection from lethal SUDV challenge when FVM04 was administered 1 dpi. These data show that FVM04 compares well with the post-exposure efficacy (2 dpi) of EBOV-specific mAbs 1H3, 2G4, and 4G7 as reported (Qiu et al., 2012b) , while expanding the breadth of protection to include SUDV. The guinea pig model of EBOV infection is considered a substantially more stringent model for screening of therapeutics (Cross et al., 2015; Parren et al., 2002; Qiu et al., 2012b) , Particularly, the testing of therapeutics at 3 dpi in this model is particularly stringent, with good predictive value for efficacy in NHPs . Recently, Wong et al. also reported the development of a guinea pig model of SUDV infection (Wong et al., 2015) . The GPA-SUDV causes a uniformly lethal infection, with the major hallmarks of SUDV infection, including lymphadenopathy, increased liver enzyme activities, and coagulation abnormalities (Wong et al., 2015) . We used both of these models in the present study to evaluate the efficacy of FVM04. A single injection of FVM04 (5 mg) at 1 or 3 dpi fully protected guinea pigs against GPA-SUDV infection with no signs of disease. This suggests that FVM04 alone may be sufficient to control SUDV infection. To date, no SUDV antibodies have been tested in NHP studies; thus, it is not known how protection in rodent models relates to efficacy against SUDV in NHPs. In contrast, in the EBOV guinea pig model, a protection level of more than $60%, 3 dpi, at a total dose of 5 mg is a good indicator of efficacy in NHPs. In fact, the ZMapp TM cocktail that fully protected NHPs at 5 dpi was selected based on partial efficacy at 3 dpi in guinea pigs . In the GPA-EBOV-infected guinea pigs, FVM04 protected two out of six animals when administered at 1 dpi. While the protection level of 33% was statistically significant, it was lower than the reported protection level of 60% at 1 dpi afforded by c2G4 or c4G7 (Qiu et al., 2012b) . These data suggested that, while FVM04 alone may be sufficient for SUDV, more efficient protection against EBOV would require a cocktail. Given that c2G4 and c4G7 target overlapping epitopes (Murin et al., 2014) , we reasoned that 4G7 may be a good candidate for replacement by FVM04. This led to a cocktail of FVM04/c2G4/c13C6 that protected five out of six guinea pigs against SUDV and four out of six animals against EBOV. This level of protection against EBOV is comparable with historical efficacy data of ZMapp TM ( Figure 7E ) . These data demonstrate that a cocktail of FVM04, c13C6, and c2G4, is as effective as ZMapp TM against EBOV while expanding the breadth of protection to include SUDV. The efficacy of these cocktails must be further tested in future NHP studies; however, these data position FVM04 as a strong candidate to be a component of a pan-ebolavirus therapeutic cocktail. Antibody Production FVM04 was produced in HEK293 cells transfected with the plasmid encoding IgH and IgL using polyethylenimine (Polysciences) and purified by protein A chromatography, as described previously (Keck et al., 2015) . For ZMapp TM production, N. benthamiana plants genetically modified to produce homogenous mammalian N-glycans of the GnGn glycoform were grown for 4 weeks in an enclosed growth room (20-23 C) and used for vacuum infiltration as described previously . Seven days later, the mAb was extracted from the leaf tissue and purified via protein A chromatography, and endotoxin was removed using an Acrodisc Unit with Mustang Q Membrane (Pall Life Sciences). For Fab production, FVM04 IgG was incubated in 2% papain for 2 hr at 37 C, and the reaction was quenched with 50 mM iodoacetamide. The Fc portion was removed via protein A affinity, and the Fab was further purified via S75 SEC. Alanine-scanning mutagenesis of an expression construct for full-length EBOV GP (strain Mayinga-76) (Davidson et al., 2015) changed residues 33-676 to alanine (and alanine residues to serine) to create a library of clones, each representing an individual point mutant, covering 641 of 644 residues. Clones were individually arrayed into 384-well plates, transfected into HEK293T cells, and allowed to express for 22 hr. Cells were incubated with primary antibody and then with an Alexa-Fluor-488-conjugated secondary antibody (Jackson ImmunoResearch Laboratories). After washing, cellular fluorescence was detected using the Intellicyt high-throughput flow cytometer (Intellicyt). Background fluorescence was determined in vector-transfected control cells. mAb reactivity against each mutant GP clone was calculated relative to wild-type GP reactivity by subtracting the signal from mock-transfected controls and normalizing to the signal from wild-type GP-transfected controls. Mutated residues within clones were identified as critical to a test mAb epitope if they did not support reactivity of the mAb but did support the reactivity of other control EBOV mAbs. This counter-screen facilitates the exclusion of GP mutants that are locally misfolded or have an expression defect. The detailed algorithms used to interpret shotgun mutagenesis data are described elsewhere (patent application 61/938,894; Davidson and Doranz, 2014) . All cell lines were obtained from ATCC. Vero and HEK293T cells were maintained in DMEM (Life Technologies) and supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals) and 1% penicillin-streptomycin (Life Technologies). BHK-21 and Vero cells were grown in Eagle's minimal essential medium (EMEM) (Corning) supplemented with 10% FBS and 1% penicillin-streptomycin. All cell lines were maintained in a humidified 37 C, 5% CO 2 incubator. Negative-Stain EM Studies EBOV GPDMuc and FVM04 were each individually purified. EBOV GPDMuc was combined with a 10-fold molar excess of FVM04 Fab and incubated overnight at 4 C. The complex was further purified by SEC with an S200i column (GE Healthcare) equilibrated in Tris-buffered saline (TBS). The complex was deposited onto a carbon-coated 400 copper mesh grid and subsequently stained with 1% uranyl formate. Grids were loaded into a Tecnai T12 Spirit at 120keV and imaged using this Tietz TemCam-F416 CMOS camera at 52,0003 magnification at a nominal defocus of $1.5 mM. Micrographs were collected automatically using Leginon (Suloway et al., 2005) and processed within Appion (Lander et al., 2009) . DogPicker ) was used to automatically pick particles in the raw micrographs and placed into a 2D stack. Initial reference-free 2D classification was undertaken in XMIPP (Sorzano et al., 2004) , and particles that did not correspond to EBOV GPDMuc bound to FVM04 (e.g., noise or amorphous particles) were removed at this point, resulting in a final stack of 13,139 particles. The remaining 13,139 particles were then processed in two subsequent rounds of reference-free 2D classification in Relion 1.4 (Scheres, 2012) , resulting in a final particle count of 8,208. Approximately half of the final 2D class averages had identifiable complexes of EBOV GPDMuc bound to FVM04. Images were created in UCSF Chimera and Adobe Photoshop. To determine the absolute molecular weight of EBOV GPDmuc, we first performed SEC-MALS using conjugate analysis to account for glycan content. To determine the number of FVM04 Fab molecules bound to EBOV GPDmuc trimer, SEC-MALS was repeated with EBOV GPDmuc that had been complexed with excess FVM04 Fab for 6 hr at room temperature. A Superdex 200 SEC column was coupled in line with the following calibrated detectors: (1) a MiniDawn Treos MALS detector (Wyatt Corporation) and (2) an Optilab T-reX refractive index (RI) detector (Wyatt Corporation). The Astra VI software (Wyatt Corporation) was used to combine these measurements to determine the absolute molar mass of the eluted proteins. Two different neutralization assays based on pseudotyped recombinant VSV-expressing filovirus GP were used: a replication-defective recombinant Indiana VSV (rVSV)-GP-Luc expressing firefly luciferase and a replicationcompetent rVSV expressing GFP (rVSV-GFP). rVSV-GP Luciferase Pseudotype Assay Pseudotyped viruses were generated based on a modification of a previously published method (Whitt, 2010) . HEK293T cells (80% confluency) were transfected with plasmids encoding EBOV GP, MARV GP, or SUDV GP using Fugene HD (Promega) according to the manufacturer's protocol. The next day, these cells were infected with rVSV-DGP pseudotype (Kerafast) at an MOI of 3, and the virus was washed off after 1 hr with DPBS. The next day, the supernatant was collected and clarified by centrifugation. To titer the pseudotyped virus, BHK-21 cells were transfected in six-well plates with pCAGGS VSV-G (Kerafast), and after 48 hr, serial dilution of VSV-EBOV-GP-Luc or VSV-SUDV-GP-Luc pseudotype was added to each well for 1 hr before the addition of 0.9% agar in DMEM. The next day, wells were fixed with 5% glutaraldehyde for 30 min before removing the agar and staining with crystal violet to count the plaques. The details of the luciferase assay for the determination of VSV-GP-Luc infectivity were previously described (Keck et al., 2015) . Data were fit to a 4PL curve using GraphPad Prism 6. Percent neutralization was calculated based on wells containing virus only. rVSV GP-GFP Assay rVSVs expressing EGFP, as well as EBOV or MARV GP in place of VSV-G, were described previously Ng et al., 2014; Wong et al., 2010) . rVSVs bearing TAFV, SUDV GP, or BDBV GPDMuc were generated by the same method. VSV particles containing cleaved GP (GP CL ) were generated by incubating rVSV-GPs with thermolysin (200 mg/ml) for 1 hr at 37 C, followed by treatment with phosphoramidon (1 mM), and reaction mixtures were used immediately. Infectivity of rVSVs was measured by counting EGFP-positive cells at 12-14 hr post-infection using a CellInsight CX5 automated microscope and onboard software (Thermo Scientific). For neutralization experiments, serial dilutions of mAbs were incubated with the rVSV-GP for 1 hr at room temperature. Monolayers of Vero cells seeded in 96-well plates were inoculated with the mAb-virus mixture in triplicate and then incubated at 37 C overnight. Infection was scored 12 hr to 16 hr post-infection by the enumeration of eGFP + cells and normalized to no-antibody control taken to represent 100%. PRNT Assay Antibody at indicated concentrations was incubated with 100 PFU of EBOV (Kikwit-95) or SUDV (Boniface) at 37 C, 5% CO 2 , and 80% humidity for 1 hr. The mixture was then added to Vero cells in six-well plates and incubated for 1 hr. Cells were then overlaid with a mixture of one part 1% agarose (Seakem) and one part 2X Eagle's basal medium (EBME), 30 mM HEPES buffer, and 5% DFBS and incubated at 37 C, 5% CO 2 , and 80% humidity. A second overlay containing 5% neutral red was added 6-7 days later, and plates were incubated overnight. Plaques were counted the following day, and percent neutralization was determined by comparing to control wells. Thermolysin-cleaved VSV-GP (rVSV-GP CL ) was incubated with a functionalspacer-lipid reagent conjugated to biotin (FSL-biotin; Sigma) to allow incorporation into viral membrane, and the resulting biotinylated viral particles were then captured onto streptavidin-coated plates, as described previously (Ng et al., 2014) . Plates were then washed and blocked with 3% BSA-PBS buffer. Serial dilutions of antibodies were added to virus-coated plates, and after washing, soluble FLAG-tagged NPC1 domain C protein (2 nM) was added to each well. Plates were washed, and the extent of NPC1 binding to rVSV-GP CL was detected using an anti-FLAG-HRP (horseradish peroxidase) antibody conjugate. All incubations were performed for 1 hr at 37 C. Binding was expressed as percentage of the maximal binding signal obtained with no-antibody control. Kinetics experiments were performed on the ForteBio Octet Red96 platform. Data were collected at 25 C with orbital shaking at 1,000 rpm in 200 ml. Protein G or Anti-human Fc sensors (ForteBio) were equilibrated in kinetics buffer (13 PBS, 0.1% BSA, and 0.02% Tween-20) for 10 min prior to loading with 10 mg/ml FVM04 antibody for 2 min. A stable baseline was established in kinetics buffer for 1 min before the FVM04-coated sensors were added to a range of EBOV, SUDV, or BDBV GPDTM concentrations. The association step of GPDTM with FVM04 proceeded for 2 min before allowing the GPDTM to dissociate into kinetics buffer for 10 min. A reference sensor without FVM04 was used to account for nonspecific binding of GP to the sensor. The data were fit globally to a 1:1 or 2:1 Langmuir binding model using ForteBio data analysis software 9.0. Mouse Challenge Studies with EBOV Female BALB/c mice (6-8 weeks old) were purchased from Charles River Laboratories. Mice were exposed to an i.p. dose of 100 PFU of (Mayinga strain) MA-EBOV suspended in PBS. Antibodies were delivered at the indicated i.p. doses and time points after exposure. Control mice were treated with PBS or left untreated. Mice were observed daily for clinical signs of disease, including but not limited to, reduced grooming and hypoactivity, and group weights were recorded through day 14. Observations were increased to a minimum of twice daily when mice exhibited signs of disease. Moribund mice were humanely euthanized based on Institutional Animal Care and Use Committee (IACUC)-approved criteria. Mice were observed a minimum of 21 days after exposure. Mouse Challenge Study with SUDV IFNabR À/À mice (4 weeks old) (B6.129S2-Ifnar1tm1Agt/Mmjax) on the C57BL/6 background were purchased from Jackson Laboratories. Upon arrival, mice were housed in microisolater cages and provided chow and water ad libitum. Mice were challenged intraperitoneally with 1,000 PFU of SUDV and treated at indicated times with antibodies by i.p. injection. Mice were observed daily for lethality or clinical signs of disease. Moribund mice were humanely euthanized based on IACUC-approved criteria. Guinea Pig Challenge Studies Female Hartley guinea pigs (4-6 weeks old, 250-300 g) were purchased from Charles River Laboratories and randomly assigned into different groups. All guinea pigs were challenged with an i.p. dose of 1,000 3 LD 50 GPA-EBOV (Volchkov et al., 2000) or 1,000 3 LD 50 GPA-SUDV (Wong et al., 2015) in 1 ml DMEM. The mAbs or mAb cocktail were given in an i.p. dose once at 1 or 3 dpi with 5 mg of each individual mAb or mAb cocktail (at a 1:1:1 ratio of each mAb) per animal. The control group was given 1 ml PBS. All animals were monitored for signs of disease, survival, and weight change for 15-16 days, and survival was monitored for 12 additional days. Animal research using mice was conducted under a protocol approved by the U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID) IACUC in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals. The USAMRIID facility is fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International and adheres to the principles stated in the Guide for the Care and Use of Laboratory Animals. Challenge studies were conducted under maximum containment in an animal BSL-4 facility. The guinea pig experiments were performed at the National Microbiology Laboratory in Winnipeg, MB, Canada. All animal experiments have been approved by the Animal Care Committee at the Canadian Science Center for Human and Animal Health in accordance with the guidelines outlined by the Canadian Council on Animal Care. Relative bindings of FVM04 and KZ52 to the mutants affecting the four glycan cap glycosylation sites are shown as percent of binding of the same antibody to wild type (wt) GP expressed on 293T cells and determined by flow cytometry. Related to Figure 2 . Figure S2 : The location of residues whose mutation increased mAb binding (Table S1) plotted on models of the EBOV GP trimer and monomer structures. Residues R498 and R501 are not present on the EBOV crystal structure, but their general location is indicated here by showing residue 502. GP1 is shown in shades of yellow, GP2 is shown in shades of red, except for GP1 Cys53 and GP2 Cys601 and Cys 609, which are shown as orange. The model was created by overlaying the available crystal structures for EBOV (PDB ID 3CSY) and SUDV (PDB ID 3S88). Related to Figure 2 . While unbound FVM04 Fab shows a molecular weight (MW) of 47.1KDa, the MW of GPmuc trimer shifted from 192.7 kd to 242.0 KDa upon binding to FVM04, indicating the binding of a single Fab per trimer. Related to Figure 3 .

projects that include this document

Unselected / annnotation Selected / annnotation