Antibody-Based Inhibition of Pathogenic New World Hemorrhagic Fever Mammarenaviruses by Steric Occlusion of the Human Transferrin Receptor 1 Apical Domain

Abstract

Pathogenic clade B New World mammarenaviruses (NWM) can cause Argentine, Venezuelan, Brazilian, and Bolivian hemorrhagic fevers. Sequence variability among NWM glycoproteins (GP) poses a challenge to the development of broadly neutralizing therapeutics against the entire clade of viruses. However, blockade of their shared binding site on the apical domain of human transferrin receptor 1 (hTfR1/CD71) presents an opportunity for the development of effective and broadly neutralizing therapeutics. Here, we demonstrate that the murine monoclonal antibody OKT9, which targets the apical domain of hTfR1, can sterically block cellular entry by viral particles presenting clade B NWM glycoproteins (GP1-GP2). OKT9 blockade is also effective against viral particles pseudotyped with glycoproteins of a recently identified pathogenic Sabia-like virus. With nanomolar affinity for hTfR1, the OKT9 antigen binding fragment (OKT9-Fab) sterically blocks clade B NWM-GP1s and reduces infectivity of an attenuated strain of Junin virus. Binding of OKT9 to the hTfR1 ectodomain in its soluble, dimeric state produces stable assemblies that are observable by negative-stain electron microscopy. A model of the OKT9-sTfR1 complex, informed by the known crystallographic structure of sTfR1 and a newly determined structure of the OKT9 antigen binding fragment (Fab), suggests that OKT9 and the Machupo virus GP1 share a binding site on the hTfR1 apical domain. The structural basis for this interaction presents a framework for the design and development of high-affinity, broadly acting agents targeting clade B NWMs. IMPORTANCE Pathogenic clade B NWMs cause grave infectious diseases, the South American hemorrhagic fevers. Their etiological agents are Junin (JUNV), Guanarito (GTOV), Sabiá (SABV), Machupo (MACV), Chapare (CHAV), and a new Sabiá-like (SABV-L) virus recently identified in Brazil. These are priority A pathogens due to their high infectivity and mortality, their potential for person-to-person transmission, and the limited availability of effective therapeutics and vaccines to curb their effects. While low homology between surface glycoproteins of NWMs foils efforts to develop broadly neutralizing therapies targeting NWMs, this work provides structural evidence that OKT9, a monoclonal antibody targeting a single NWM glycoprotein binding site on hTfR1, can efficiently prevent their entry into cells.

Keywords: Junin; Machupo; Sabiá-like; X-ray crystallography; antiviral agents; electron microscopy; mammarenavirus; monoclonal antibodies; transferrin receptor.

FIG 1

Characterization of OKT9 binding to hTfR1. (A) Formation of an hTfR1-OKT9 complex evaluated by SEC coelution. Samples include hTfR1 (purple), OKT9-Fab (blue), and sTfR1+OKT9-Fab (maroon). An overlay of all traces is shown for comparison. Gray bars indicate collected fractions. The red star denotes the expected peak fraction for OKT9-Fab alone. (B) SDS-PAGE of fractions collected from isolated and complex SEC runs. On the left is shown the SDS-PAGE of the SEC fractions of the hTfR1 run, in the center the fractions of the OKT9-Fab run, and on the right the fractions of the complex. Colored boxes indicate the presence of the protein of interest, coordinated with the color used in SEC traces. The red arrow denotes the expected band for OKT9-Fab. (C) ELISA binding of OKT9 to hTfR1. An indirect ELISA was performed decorating the plate with sTfR1 and then incubating with different concentrations of OKT9-Fab and OKT9. Anti-mouse IgG conjugated to HRP was used as a secondary antibody. The EC₅₀ calculated over the normalized OD₄₅₀ for OKT9 was 0.411 nM with a 95% confidence interval (CI) of 0.213 to 0.633, and for OKT9-Fab the EC₅₀ was 3.695 nM with a 95% CI of 2.463 to 6.028. (D) Kinetics of OKT9-Fab interaction with hTfR1 immobilized on a biosensor surface. The receptor was exposed to increasing concentrations of OKT9-Fab as labeled (31.3 nM, 62.5 nM, 125 nM, and 250 nM). One hundred eighty seconds of biolayer recordings show binding and dissociation. (E) Assessment of OKT9-Fab and MACV GP1-Fc for binding to sTfR1 by biolayer interferometry. MACV GP1-Fc immobilized onto anti-human Fc biosensor tips and incubated with sTfR1 (top left) or buffer alone (bottom left), sTfR1 in complex with OKT9-Fab (top center), OKT9-Fab alone (bottom center), sTR1 in complex with transferrin (top right), or transferrin alone (bottom right). The arrows indicate the time points at which the indicated proteins were added and the dissociation step. The data are representative of two replicates for each of the experimental conditions shown.

FIG 2

Inhibition of internalization of NWHF pseudovirus by OKT9. (A) Fluorescence microscopy images of HEK-293T cells. The images were taken from the inhibition of JUNV, MACV, SABV-L, and LASV pseudovirus internalization assays mentioned for panel B: pseudovirus + buffer control, pseudovirus + 100 nM OKT9, pseudovirus + 100 nM OKT9-Fab, and pseudovirus + 100 nM anti-EGFR as a nonrelevant antibody control. The reference bar indicates 50 μm. (B) Fluorescence microscopy of HEK-293T cells treated 48 h with buffer control, 100 nM OKT9, 100 nM OKT9-Fab, or 100 nM anti-EGFR, labeled with the DiO membrane dye (green), and incubated for 30 s with transferrin-TMR (red). The reference bar indicates 34 μm. (C) Flow cytometry dot plot analysis of JUNV, MACV, SABV-L, and LASV pseudovirus internalization in HEK-293T cells showing eGFP expression in the absence and presence of 100 nM OKT9, OKT9-Fab, and anti-EGFR. In the case of MACV, JUNV, and SABV-L, a reduction of the percent eGFP-positive events occurs in the presence of OKT9 and OKT9-Fab compared to buffer control and the nonrelevant antibody. In the case of LASV, similar levels of eGFP events are observed independently of the treatment. The percentages of eGFP-positive events are indicated inside the gates. (D) Inhibition of JUNV, MACV, and SABV-L pseudovirus internalization. Relative entry rate of JUNV, MACV, the recently reported SABV-L, and control LASV pseudovirus in HEK-293T cells was quantified in the presence of 100 nM OKT9-Fab, 100 nM OKT9, and 100 nM anti-EGFR. The data were 100% normalized with the cells without treatment, and the significant differences are indicated by comparing Fab OKT9, OKT9, and the negative-control anti-EGFR versus no treatment (*, P < 0.005, Student's t test for unpaired data of triplicate determinations). (E) Relative entry rate of pseudoviruses decorated with the GP1/GP2 complex of SABV, JUNV, MACV, GTOV, CHAV, SABV-L, and LASV to HEK-293T cells in the presence of OKT9 (0.01 to 100 nM). Pseudoviruses were loaded with an eGFP expression vector to express once internalized. After 48 h, the cells were fixed and the percentage of positive internalization events quantified by flow cytometry. The data are expressed as the means ± standard deviations (SD) from the sample. The data were normalized to 100% with the cells without treatment. (F) OKT9 and OKT9-Fab inhibit the infectivity of JUNV IV4454 virus strain. A549 cells were preincubated for 1 h with 200 nM OKT9, OKT9-Fab, anti-EGFR, or medium alone (control) and then infected with JUNV at an MOI of 0.01. After 1 h of incubation, viral inocula were replaced with the respective antibody-supplemented medium or medium alone, and 24 h postinfection, total JUNV production in A549 cell supernatants was measured using a PFU assay in Vero cells. The graph shows means ± SD from a representative experiment (from four independent experiments). The statistical analysis performed was ANOVA followed by Duncan’s test (**, P < 0.05). (G) Set of 200 nM OKT9-, OKT9-Fab-, anti-EGFR-, or medium alone (control)-treated A549 cells monolayers was harvested with TRIzol for RNA extraction 18 h postinfection with JUNV IV4454. Viral RNA (z gene) was quantified using RT-PCR, using actin as a housekeeping gene. The graph shows means ± SD from two independent experiments. The statistical analysis performed was ANOVA followed by Dunnett’s test (*, P < 0.0001).

FIG 3

Crystallographic structure of OKT9-Fab and visualization of the hTfR1-OKT9 complex. (A) Amino acid sequence of OKT9 heavy-chain (top) and light-chain (bottom) variable region. In blue are the heavy-chain CDRs H1, H2, and H3 and in light blue the light-chain CDRs L1, L2, and L3. (B) X-ray crystallographic structure of OKT9-Fab with heavy chain colored in slate and light chain in cyan. Loops corresponding to variable light chain CDRs are labeled L1, L2, and L3, and CDRs for the variable heavy chain are labeled H1, H2, and H3, where unmodeled regions are shown as dashed lines. An unmodeled tyrosine residue in CDR_H3 is shown in white. (C) Micrograph of negatively stained sample containing hTfR1-OKT9 complex. Red triangles indicate particles selected for 2D classification (scale bar, 50 nm). (D) Representative particles and 2D class averages of negatively stained hTfR1-OKT9 complexes on ultrathin carbon (scale bar, 200 Å). (E) Projections of 3D hTfR1-OKT9 complex density along two orthogonal directions. 3D density was obtained by ab initio reconstruction from negative-stain images of the complex.

FIG 4

Docking model of hTfR1-OKT9 complex and relationship to natural ligands and human pathogen molecular ligands. (A) Isosurface rendering of hTfR1-OKT9 3D density map with an estimated resolution of 11 Å. Inset shows a well-resolved region of the 3D density, corresponding to a monomeric transferrin receptor bound by an OKT9-Fab. Front (left) and side (right) views are shown. (B) Front views of all 10 ClusPro-generated models using OKT9-Fab and 3KAS hTfR1 apical domain monomer (white). Best-fit models used for analysis are highlighted in red; other models are shown in hues of blue. (C) Overlay of ClusPro generated model using OKT9-Fab and crystal structure PDB entry 3KAS with hTfR1-OKT9 3D density. (D) Surface representation of the 3KAS apical domain (dark gray) colored to illustrate the binding interface with MACV GP1 protein (salmon), OKT9-Fab model (blue), or both (black). (E, top) Transferrin receptor (gray) bound to natural ligands (transferrin, PDB entry 1SUV; hereditary hemochromatosis factor [HFE] [43], PDB entry 1DE4; and ferritin, PDB entry 6GSR). (Bottom) Transferrin receptor bound to OKT9-Fabs and human pathogens (MACV-GP1, PDB entry 3KAS; P. vivax, PDB entry 6D04).