An Engineered N-Glycosylated Dengue Envelope Protein Domain III Facilitates Epitope-Directed Selection of Potently Neutralizing and Minimally Enhancing Antibodies

Abstract

The envelope protein of dengue virus (DENV) is a primary target of the humoral immune response. The domain III of the DENV envelope protein (EDIII) is known to be the target of multiple potently neutralizing antibodies. One such antibody is 3H5, a mouse antibody that binds strongly to EDIII and potently neutralizes DENV serotype 2 (DENV-2) with unusually minimal antibody-dependent enhancement (ADE). To selectively display the binding epitope of 3H5, we strategically modified DENV-2 EDIII by shielding other known epitopes with engineered N-glycosylation sites. The modifications resulted in a glycosylated EDIII antigen termed "EDIII mutant N". This antigen was successfully used to sift through a dengue-immune scFv-phage library to select for scFv antibodies that bind to or closely surround the 3H5 epitope. The selected scFv antibodies were expressed as full-length human antibodies and showed potent neutralization activity to DENV-2 with low or negligible ADE resembling 3H5. These findings not only demonstrate the capability of the N-glycosylated EDIII mutant N as a tool to drive an epitope-directed antibody selection campaign but also highlight its potential as a dengue immunogen. This glycosylated antigen shows promise in focusing the antibody response toward a potently neutralizing epitope while reducing the risk of antibody-dependent enhancement.

Keywords: dengue; domain III; envelope; epitope; glycan; shield.

Figure 1

Structure and epitopes of DENV EDIII. (A) Structure of DENV-2 EDIII with three epitopes, AG-strand, AB-loop, and lateral-ridge epitopes circled in dotted line. The structure was extracted from PDB 1OAN (amino acid residues 298–397). (B) Epitopes of anti-EDIII antibodies used in this study. The epitope residues are highlighted in colored corresponding to each antibody.

Figure 2

Structure-guided selection of the N-glycosylation sites for selective epitope shielding. (A) Overlayed DENV-2 EDIII structures with an antigen–antibody complex of each template antibody. The selected residues are shown as sticks with labels. (B) Table summary of the selected N-glycosylation sites and shielded epitopes. The native residues of DENV-2 EDIII (amino acids 298–394) subjected to an NxS/T sequon mutation are represented as bold letters. The potential N-glycosylation residues are shown in red.

Figure 3

Analysis of purified monoglycosylated EDIII antigens. (A) Apparent size of EDIII antigens on SDS-PAGE under denaturing conditions (reduced and heat) stained with Coomassie blue. (B) Binding of the antibody panel of 3H5, 2C8, 513, and 2H12 to EDIII WT compared with monoglycosylated EDIII antigens (three technical replicates).

Figure 4

Analysis of purified monoglycosylated EDIII antigen mutant 298N_300T. (A) Apparent size of EDIII antigens on SDS-PAGE under denaturing conditions (reduced and heat) stained with Coomassie blue. (B) Binding of the antibody panel of 3H5, 2C8, 513, and 2H12 to EDIII WT compared with EDIII antigen mutant 298N_300T (three technical replicates).

Figure 5

Multiple glycosylation sites on EDIII antigens. (A) Summary table of sequential incorporation of N-glycans onto the EDIII antigen. (B) Model of EDIII Mut N illustrating glycosylation at engineered residues 298N, 309N, and 317N. Epitope residues of 3H5 are displayed in cyan. The model was created from Glycoprotein Builder. (C, D) Binding of the template antibodies 2C8, 2H12, 513, and 3H5 to EDIII WT compared with EDIII antigens with one, two, and three glycosylation sites (two technical replicates, error bars representing SD of replicates in the same plate. ND = not detected). The K_D values of each antibody are displayed in panel (D). (E) Apparent size of monoglycosylated 298N_300T, diglycosylated Mut J, and triglycosylated Mut N EDIII antigens on SDS-PAGE under denaturing conditions (reduced and heat) stained with Coomassie blue. The arrows indicate the two overlapping bands of the EDIII Mut N. (F) Binding of EDIII 298N_300T, Mut N, and Mut J to 3H5 and 513 on a western blot assay. The arrows indicate the EDIII N mutant antigen detected by each antibody.

Figure 6

scFv-phage selection campaigns. (A) Selection scheme of the Mut N-selection and WT-selection. Hit clones from each selection and their binding to EDIII antigens (WT or Mut N) and a negative control (Neg ctrl is an unrelated Fc-fusion protein). Hit clones with identical scFv sequence from each selection are highlighted in the same color. (B) Immunoglobulin germline gene analysis of each distinct hit scFv. (C) Venn diagram summarizing distinct scFv hits (reactive to both EDIII WT and Mut N) from the two selection campaigns. The number of times the clone was identified as a hit are indicated in the parentheses following the clone names.

Figure 7

Epitope mapping by mutagenesis on 3H5 epitope. (A) Epitope mapping by point mutagenesis on 3H5 epitope residues. Binding residues of each antibody are defined point mutation that leads to either severe or moderate binding reduction. Mutations with % relative binding less than 25% are defined as severe binding reduction while mutations with the % relative binding from 75 to 25% are defined as moderate binding reduction. (B) Binding of scFv-derived antibodies to D2E80 in the presence of template antibodies (Competitive ELISA). An anti-E antibody 4G2 that binds to domain II of the DENV envelope protein was used as a negative control. 3H5* detecting antibody was a chimeric 3H5 for all competition assays except for an experiment with 513 where a mouse 3H5 antibody was used. (C) Equilibrium dissociation constant (K_D) of scFv-derived antibodies measured with indirect ELISA using soluble E80 protein of four DENV serotypes (the representing values are the average of two technical replicates). A control anti-EDIII antibody 513, which is cross-reactive, showed K_D values in line with literature reported values.

Figure 8

Neutralization and antibody-dependent enhancement (ADE) of the scFv-derived anti-EDIII mAbs. (A) Neutralization of scFv-derived IgG1 in comparison with chimeric 3H5(ch3H5) antibody. (B) ADE of the scFv-derived antibodies in comparison with 4G2, 2C8, 3H5, and ch3H5 antibodies.