Isolation of potent neutralizing antibodies from a survivor of the 2014 Ebola virus outbreak

Science  18 Feb 2016:

DOI: 10.1126/science.aad5788


Antibodies targeting the Ebola virus surface glycoprotein (EBOV GP) are implicated in protection against lethal disease, but the characteristics of the human antibody response to EBOV GP remain poorly understood. We isolated and characterized 349 GP-specific monoclonal antibodies (mAbs) from the peripheral B cells of a convalescent donor who survived the 2014 EBOV Zaire outbreak. Remarkably, 77% of the mAbs neutralize live EBOV, and several mAbs exhibit unprecedented potency. Structures of selected mAbs in complex with GP reveal a site of vulnerability located in the GP stalk region proximal to the viral membrane. Neutralizing antibodies (NAbs) targeting this site show potent therapeutic efficacy against lethal EBOV challenge in mice. The results provide a framework for the design of new EBOV vaccine candidates and immunotherapies.

In recent years, EBOV outbreaks have increased in frequency, duration, and geographical spread, underscoring the need for pre- and post-exposure treatments (1). The membrane-anchored EBOV GP trimer is the sole known target for protective antibodies and is currently the primary target for antiviral vaccines and therapies (2, 3). A small number of protective anti-GP mAbs have been isolated from immunized mice, and recent structures of these antibodies in complex with GP have illuminated key sites of vulnerability on the EBOV glycoprotein (37). However, only a small number of GP-specific mAbs have been isolated from human EBOV survivors (810), and therefore the characteristics of the human antibody response to EBOV GP remain largely undefined.

In this study, we aimed to comprehensively profile the human B cell response to EBOV GP by cloning an extensive panel of anti-GP mAbs from the peripheral B cells of a convalescent donor (subject 45) who survived the 2014 EBOV Zaire outbreak. Three months after primary infection, the donor plasma showed strong IgG binding reactivity to EBOV GP and potent neutralizing activity, suggesting that this subject had mounted a robust anti-GP NAb response by this time point (fig. S1, A and B). To assess the magnitude of the B cell response to EBOV GP, B cells were stained with a fluorescently labeled EBOV GP ectodomain (GP∆TM) (4) and analyzed by flow cytometry. Approximately 3% of IgG+ B cells were specific for GP∆TM (fig. S2), which is comparable to the percentage of circulating antigen-specific peripheral B cells observed during chronic HIV infection and after primary dengue infection (11, 12). Cognate antibody heavy- and light-chain pairs were rescued from 420 individual GP∆TM-reactive B cells by single-cell PCR and subsequently cloned and expressed as full-length IgGs in an engineered strain of Saccharomyces cerevisiae (13). Of the 420 cloned mAbs, 349 bound to EBOV GP in preliminary binding screens (table S1). Analysis of the heavy- and light-chain variable regions (VH and Vκ, respectively) revealed that the anti-GP repertoire was highly diverse, containing 294 independent clonal lineages (fig. S3A and table S2). This result contrasts with previously described anti-HIV and anti-influenza repertoires, which show a significantly higher degree of clonal restriction (11, 14). Comparison to non-GP reactive antibodies (15) revealed that the EBOV GP-specific repertoire was skewed toward immunoglobulin light-chain kappa (Igκ) versus immunoglobulin light-chain lambda (Igλ) and longer heavy chain complementarity-determining region 3 (CDRH3) lengths (fig. S3, B and C, and table S2). Similar biases have also been observed in HIV-1 infected patient repertoires (11, 12). VH and Vκ germline gene usage in the GP-specific repertoire was similar to non-GP–specific repertoires (15, 16) (fig. S3, D and E, and table S2). As expected for antibodies derived from IgG+ B cells, almost all of the GP-specific clones were somatically mutated, with an average of 5.1 and 2.7 nucleotide substitutions in VH and VL, respectively (fig. S3F and table S2).

To map the antigenic specificities of the anti-GP mAbs, we produced 321 IgGs in larger quantities and performed biolayer interferometry (BLI) binding experiments with several GP variants. We first tested binding to EBOV GP∆TM and a mucin-like domain deletion construct (GP∆muc) (6). Unexpectedly, only two mAbs failed to bind to GP∆muc, indicating that less than 1% of the GP-specific antibody response in this donor is directed against epitopes within or dependent on the mucin-like domain (Fig. 1A and table S3). Interestingly, ~30% of the mAbs showed increased binding responses and faster association rates to GP∆muc compared to GP∆TM (fig. S4), suggesting that these mAbs likely recognize epitopes that are partially occluded by the mucin-like domain. We next tested the mAbs for binding to a secreted GP isoform, sGP, which is expressed as a disulfide-linked GP1 dimer containing the majority of the non-mucin GP1 core and glycan cap sequence (fig. S5) (17, 18). This analysis revealed that 39% of GP∆muc-reactive mAbs failed to bind to sGP, 2% bound with similar apparent affinity to both GP∆muc and sGP, and 59% reacted with both proteins but bound with higher apparent affinity to sGP (Fig. 1, B and C, and table S3). The latter result is consistent with previous studies showing that sGP is secreted in large quantities during natural infection and may behave as an antigenic decoy by redirecting the immune response toward epitopes that are either inaccessible on surface GP or shared between the two proteins (17, 19).

Fig. 1 Antigen-binding properties of anti-GP mAbs.

(A) Apparent binding affinities of GP-specific IgGs to Zaire GP∆TM and Zaire GP∆muc constructs as determined by BLI measurements. Newly discovered anti-GP mAbs are shown as red circles. KZ52 IgG (yellow diamond), 13C6 IgG (green triangle), 1H3 IgG (orange square), and 2G4 IgG (purple hexagon) are included for comparison. (B) Apparent binding affinities of GP-specific IgGs to Zaire sGP and Zaire GP∆muc as determined by BLI measurements. (C) Pie chart summarizing antibody binding profiles. Cross-reactive mAbs refer to those that bind to both GP and sGP. N.B., non-binder; W.B., weak binder. IgG KDs were calculated for mAbs with BLI responses >0.1 nm. MAbs with BLI responses <0.05 nm were designated as N.B.; MAbs with BLI responses between 0.05 and 0.1 nm were designated as W.B. All data are representative of two or more independent experiments.

To further define the epitopes targeted by the anti-GP mAbs, we performed competitive binding experiments (20). We first tested the 321 mAbs for competition with two well-characterized murine mAbs, 1H3 and 13C6, that recognize overlapping epitopes in the glycan cap (4). The vast majority of sGP cross-reactive binders competed with one or both mAbs, suggesting that they also bind within the glycan cap (Fig. 2A and fig. S6A). We next tested the GP-specific mAbs for competition with KZ52, a human antibody that binds at the interface of GP1 and GP2 (6, 8). Approximately half of the GP-specific binders competed with KZ52 (Fig. 2A and fig. S6B), suggesting that this antigenic site is a common target for antibodies elicited by natural EBOV infection, at least for the donor studied. Since KZ52 has been shown to exhibit specificity for Zaire GP (6), we next tested selected KZ52 competitors for cross-reactivity with Sudan (SUDV) GP and Bundibugyo (BDBV) GP. Similar to KZ52, most of these mAbs did not show broad species cross-reactivity (Fig. 2B and fig. S7A). However, in contrast to KZ52 and other well-characterized GP base binders (4), the vast majority of KZ52 competitor mAbs failed to react with a minimal thermolysin processed GP core in which both the mucin domain and glycan cap regions have been proteolytically removed (GPCL) (4) (Fig. 2C). Thus, this class of antibodies appear to target unique epitopes that either directly overlap with the KZ52 epitope or are sterically inhibited by the KZ52 Fab. To estimate the number of different antigenic sites recognized by the remaining GP-specific mAbs, we performed competitive binding experiments with four high affinity mAbs from the panel (ADI-15974, ADI-15933, ADI-15810, and ADI-15983) that did not show significant competition with KZ52 or with each other (table S4). Eighty percent of the non-KZ52 competitive GP-specific mAbs bound to epitopes overlapping that of ADI-15974 or ADI-15810 (Fig. 2A and fig. S6B). This group of mAbs also showed significantly broader cross-species GP-binding reactivity than the KZ52 competitors (Fig. 2D and fig. S7B). Overall, these data show that the anti-GP repertoire in this patient is primarily composed of clones that target four non-overlapping antigenic sites on EBOV GP (Fig. 2E).

Fig. 2 Epitope mapping.

(A) Percentage of sGP-reactive and sGP non-reactive mAbs directed against each antigenic site on EBOV GP. Epitope binning was performed using a previously described yeast-based competition assay (20). (B) Percentage of selected KZ52 competitors that cross-react with SUDV GP and BDBV GP. Binding cross-reactivity was assessed by ELISA. (C) ELISA binding of selected KZ52 competitors to a minimal GP core that contains deletions in the mucin-like domain and glycan cap (GPCL). ELISA binding is expressed as the OD405 reading at a concentration of 0.2 μg/ml. (D) Percentage of selected KZ52 non-competitors that cross-react with SUDV GP and BDBV GP. Binding cross-reactivity was assessed by ELISA. (E) Summary of the antigenic sites targeted by the anti-GP mAbs. All data are representative of two or more independent experiments.

We next measured the neutralizing activity of the B cell-derived mAbs using a live virus plaque reduction neutralization (PRNT) assay. Due to the large number of mAbs and high-throughput nature of our study, initial neutralization screening was performed using a single concentration of purified IgG (table S5 and fig. S9). Remarkably, 77 and 63% of the mAbs reduced viral infectivity by 50 and 80% (PRNT50 and PRNT80), respectively, at concentrations ≤50 μg/ml (Fig. 3A and table S5). Control experiments with yeast-produced and CHO-produced IgGs demonstrated that functional activity is likely not affected by the host production system (fig. S8). Analysis of neutralizing activity by competition group revealed that the majority of competition groups contained a proportion of NAbs, with the KZ52 and 13C6/1H3 competition groups containing the highest proportion of NAbs (Fig. 3A and table S5). The latter result was unexpected, as 13C6 and 1H3 only weakly neutralize in the absence of complement (7, 21). We next performed neutralization titration experiments in order to evaluate neutralization potency. These results showed that several NAbs, particularly those in the ADI-15974 and KZ52 competition groups, exhibited extraordinary potency. Notably, half of the NAbs tested from the ADI-15974 competition group, and two of the NAbs tested from the KZ52 competition group, neutralized with PRNT50 values ≤0.05 μg/ml (Fig. 3B and table S5). In contrast, the majority of 13C6 and/or 1H3 competitor mAbs neutralized with relatively modest potency, with PRNT50 values averaging 3.3 μg/ml. We conclude that the GP-specific antibody repertoire in the donor studied contains a high proportion of NAbs, the most potent of which bind to epitopes overlapping that of KZ52 or ADI-15974.

Fig. 3 Neutralizing activity of anti-GP mAbs.

(A) Percentage of mAbs in each competition group that reached PRNT50 or PRNT80 at concentrations ≤50 μg/ml. The total number of mAbs tested from each competition group are shown at the top of the corresponding bar. (B) PRNT50 and PRNT80 values of selected mAbs from each competition group. KZ52 IgG is included for comparison (green inverted triangle). Red bars indicate median PRNT50 and PRNT80 values. Neutralization assays were performed using a live virus plaque reduction assay. PRNT50 and PRNT80 values represent the concentration of IgG required to reduce viral infectivity by 50 and 80%, respectively. All data are representative of two independent experiments.

To structurally define the epitopes recognized by the NAbs, we used single-particle electron microscopy (EM) to examine five potent NAbs, representing each of the four major competition groups, in complex with fully glycosylated EBOV GP∆TM. These NAbs included ADI-15731 (a 13C6 competitor), ADI-15734 and ADI-15762 (KZ52 competitors), ADI-15758 (an ADI-15974 competitor), and ADI-15859 (an ADI-15810 competitor). We were able to obtain negative stain 2D class averages for all five complexes of Fabs bound to EBOV GPΔTM (fig. S10) and three-dimensional reconstructions for four of the Fab:EBOV GPΔTM complexes at 18-24 Å resolution (Fig. 4 and fig. S11). In agreement with competitive binding data, the EM reconstruction of ADI-15731 showed that this NAb binds within the glycan cap, with a footprint approximately between the epitopes recognized by13C6 and 1H3 and with a similar angle of approach (Fig. 4 and fig. S12A). We next examined the two KZ52 competing NAbs, ADI-15734 and ADI-15762. As anticipated, ADI-15734 bound to EBOV GPΔTM at the GP1/GP2 interface, slightly adjacent to the KZ52 epitope and at a similar angle of approach (Fig. 4 and fig. S12C). In contrast, ADI-15762 actually binds within the glycan cap, but with a shallow binding angle that likely sterically occludes the KZ52 epitope (Fig. 4 and fig. S12B). Lastly, we determined the structure of EBOV GP∆TM in complex with ADI-15758 (an ADI-15974 competitor), one of the most potent NAbs described in this panel. The EM reconstruction shows that ADI-15758 binds to a region proximal to the viral membrane, distal to all previously described epitopes, and below the body of the trimeric EBOV GP structure (Fig. 4 and fig. S12D). While this region has not yet been structurally characterized at high resolution in the pre-fusion GP context, it corresponds to the α-helical heptad repeat 2 (HR2; residues 613 to 637) defined in the post-fusion conformation (22). Docking of the EBOV GP crystal structure into the reconstruction suggests that the ADI-15758 epitope is within the C-terminal 24 residues of GP2 contained in the EBOV GP∆TM construct (6). Three Fab molecules could be visualized in the 2D class averages (fig. S10), suggesting that the HR2 region may exist as a three-helix bundle in the pre-fusion GP structure (6). Additionally, while we were not able to generate a three-dimensional reconstruction of ADI-15859 (an ADI-15810 competitor) bound to EBOV GP∆TM, the negative stain 2D class averages indicate that this mAb also binds within the GP stalk region. Collectively, these data suggest that the GP stalk region containing the HR2 helices is an accessible antigenic region targeted by NAbs, a proportion of which exhibit remarkable neutralization potency.

Fig. 4 Negative stain electron microscopy of Fab:EBOV GP∆TM complexes.

(A) A structure-based (PDB 3CSY and 3S88) (5) surface representation of the ebolavirus GP trimer. The mucin domain (gray), glycan cap domain of GP1 (aqua green), GP1 core (blue), GP2 (light blue), fusion loop region of GP2 (pink), and the stalk/HR2 region (orange) have been mapped onto the structure. The residues comprising the trimeric body and the stalk region of the ebolavirus GP are displayed on the right. The mucin domains are modeled only as spheres as they are largely unstructured and poorly defined (27). Residues 613-637 corresponding to the stalk/HR2 region were modeled in silico using threefold symmetry and peptide structure prediction for the HR2 region (28). (B) Corresponding three-dimensional reconstructions of four Fab:EBOV GPΔTM complexes are shown in transparent surface representation (gray) with the model from (A) fitted in the density. Additionally, structural models for each Fab variable region were generated using the ROSIE server (29, 30) and then fitted into the density maps as surface representations. Each structure is shown as side (left) and top (right) views with the exception of ADI-15758, which is shown from the bottom up, respective to the viral membrane.

Finally, we sought to determine whether certain anti-GP NAb specificities showed greater in vivo efficacy than others. For this experiment, several NAbs were selected from each competition group and evaluated for post-exposure therapeutic efficacy against lethal EBOV challenge in a mouse infection model (23). Groups of mice were challenged with a target dose of 100 plaque-forming units (p.f.u.) of mouse-adapted EBOV (MA-EBOV), followed by a single 100 μg dose of mAb at two days post-infection (dpi). The previously described neutralizing mAb 2G4, a component of the ZMapp cocktail, was also included for comparison (3). Of significance, all of the ADI-15974 competitor NAbs (GP stalk binders) provided significant post-exposure protection, with survival rates ranging from 60 to 100% and average weight loss ranging between 8 and 10% (Fig. 5A, fig. S13, and table S6). Five out of the six NAbs in the KZ52 competition group were also highly effective in protection, with survival rates ranging between 60 and 100% (Fig. 5B and table S6). With the exception of ADI-15818 (a KZ52 competitor), all of the NAbs in these two competition groups showed greater therapeutic efficacy than 2G4, which only provided 40% protection under these conditions. In contrast to the ADI-15974 and KZ52 competitor NAbs, the NAbs targeting the glycan cap (13C6/1H3 competitors) and undefined epitopes generally showed little to no therapeutic efficacy (Fig. 5, C and D, fig. S13, and table S6). Only ADI-16037 (a 13C6 competitor NAb) provided potent protection, yielding 80% survival and 7% average weight loss. The remaining NAbs in these groups yielded ≤50% survival, which in most cases was not a statistically significant increase in protection over the negative control (Fig. 5, C and D, and table S6). This result is consistent with previous studies showing that mAbs targeting the glycan cap generally do not afford significant protection when administered at 1-2 dpi (3, 24). In summary, the vast majority of NAbs targeting the GP stalk region (ADI-15974 competitors) and the GP1/GP2 interface (KZ52 competitors) provided significant post-exposure protection against lethal EBOV challenge, whereas NAbs targeting the glycan cap (13C6/1H3 competitors) and undefined regions generally showed little to no therapeutic efficacy under these conditions.

Fig. 5 Therapeutic efficacy of NAbs against MA-EBOV.

Kaplan–Meier survival curves for ADI-15974 competitor NAbs (A) KZ52 competitor NAbs (B) 13C6 competitor NAbs (C) and NAbs targeting undefined epitopes (D). Mice were infected with 100 p.f.u. of MA-EBOV and treated intraperitoneally with a single dose of the indicated mAbs at two dpi (dotted black line). Negative control mice were treated with PBS. MAb 2G4 is included for comparison. Data are representative of one experiment with 10 mice per group.

In conclusion, we show that the human B cell response to EBOV GP is composed of a broad diversity of clones that primarily target three non-overlapping antigenic sites on the GP spike: the glycan cap, the GP1/GP2 interface, and the stalk region inclusive of the HR2 helices (fig. S14). A substantial fraction of the mAbs cloned from GP-specific B cells show neutralizing activity, demonstrating that at least in this donor, NAb responses can develop relatively early after EBOV infection. The most potently neutralizing and therapeutically effective mAbs in our panel target the GP1/GP2 interface and the GP stalk region, suggesting that these epitopes may be promising targets for rational vaccine design. In addition, the observation that EBOV escape variants can emerge after treatment with the MB-003 antibody cocktail highlights the need for protective mAbs that target new antigenic sites, such as those described here targeting the GP stalk (25, 26).


Materials and Methods

Figs. S1 to S14,

Tables S1 to S7

References (3139)


  1. ACKNOWLEDGMENTSWe thank T. Boland and M. Vasquez for assistance with antibody sequence analysis, C. Williams and S. M. Eagol for assistance with figure preparation, and R. Pejchal for providing helpful comments on the manuscript. We also thank M. Haynes for assistance with flow cytometry. All the IgGs were sequenced by Adimab's Molecular Core and produced by the High Throughput Expression group. Biolayer interferometry binding experiments were performed by Adimab's protein analytics group. The ZMapp cocktail mAb 2G4 was generously provided by Mapp Biopharmaceutical. The data presented in this manuscript are tabulated in the main paper and in the supplementary materials. GenBank accession numbers for the antibody variable region gene sequences reported in this study can be found in table S7. The cryo-EM maps have been deposited to EMDB (accession numbers EMD-6586, EMD-6587, EMD-6588, and EMD-6589). E.O.S., Z.A.B., M.L.F., K.B.J.P., A.B.W., H.L.T., and C.D.M. acknowledge support from NIH/National Institute of Allergy and Infectious Diseases Center for Excellence in Translational Research Grant U19AI109762 “Consortium for Immunotherapeutics Against Viral Hemorrhagic Fevers.” E.O.S. was also supported by R01AI067927. C.D.M. was supported by a predoctoral fellowship from the National Science Foundation. This study was supported in part by U.S. NIH grants U19 AI109762 and R01 AI067927 awarded to E.O.S. Research was funded in part by the Defense Advanced Research Projects Agency (DARPA-BAA-13-03). D.R.B. and D.S. acknowledge support from the Center for HIV/AIDS Vaccine Immunology and Immunogen Discovery Grant UM1AI100663. This is manuscript no. 29237 from The Scripps Research Institute. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the U.S. Army.
View Abstract


Navigate This Article