Antibody responses are generated to immunodominant ELK/KLE-type motifs on the nonstructural-1 glycoprotein during live dengue virus infections in mice and humans: implications for diagnosis, pathogenesis, and vaccine design

Abstract

Antibodies generated to the purified dengue type 2 virus (D-2V) nonstructural-1 (NS1) protein in mice and rabbits were compared with those generated to this protein in congeneic (H-2 class II) mouse strains and humans after D-2V infections. Unlike the profiles observed with the rabbits, similar antibody reaction profiles were generated by mice and humans with severe D-2V disease (dengue hemorrhagic fever [DHF]/dengue shock syndrome [DSS]). Many of these epitopes contained the core acidic-hydrophobic-basic (tri-amino-acid; ELK-type) motifs present in the positive or negative orientations. Antibody responses generated to these ELK/KLE-type motifs and the epitope LX1 on this protein were influenced by class II molecules in mice during D-2V infections; but these antibodies cross-reacted with human fibrinogen and platelets, as implicated in DHF/DSS pathogenesis. The core LX1 epitope (113YSWKTWG119), identified by the dengue virus complex-specific monoclonal antibody (MAb) 3D1.4, was prepared so that it contained natural I-Ad-binding and ELK-type motifs. This AFLX1 peptide, which appropriately displayed the ELK-type and LX1 epitopes in solid-phase immunoassays, generated a similar, but lower, immunodominant anti-ELK-motif antibody reaction in I-Ad-positive mice, as generated in mice and humans during D-2V infections. These antibody responses were much stronger in the high-responding mouse strains and each of the DHF/DSS patients tested and may therefore account for the association of DHF/DSS resistance or susceptibility with particular class II molecules and autoantibodies, antibody-stimulating cytokines (e.g., interleukin-6), and complement product C3a being implicated in DHF/DSS pathogenesis. These results are likely to be important for the design of a safe vaccine against this viral disease and showed the AFLX1 peptide and MAb 3D1.4 to be valuable diagnostic reagents.

FIG. 1.

Reactions of rabbit, mouse, and human antibodies generated to the D-2V NS1 glycoprotein against 174 overlapping synthetic peptides sequentially spanning the D-2V NS1 protein sequence. Pools of antisera generated to purified D-2V NS1 protein in outbred strains of mice and rabbits or during live D-2V infections in human DSS patients were diluted and reacted with 174 overlapping synthetic (9-amino-acid) peptides sequentially spanning the entire D-2V (strain PR159S1) NS1 protein sequence. The results are expressed as ELISA absorbance (492 nm); the peptides are marked at intervals of 20; and the locations of the linear (sequential) epitopes LD2 (D; peptide 13), 24A (A; peptide 31), LX1 (X; peptides 56 and 57), 24B (B; peptide 125), and 24C (C; peptides 150 and 151) are marked.

FIG. 2.

Antibody responses of congeneic mouse strains to the D-2V NS1 glycoprotein after repeated infections with live D-2V. Three mice of each congeneic strain (strains B10.G, B10.RIII, B10.M, B10.S, C57BL/BJ, B10.BR, B10.A, and B10.D2N) were immunized twice with live D-2V or virus-free medium [B10.S(C) (control)], and the reciprocal log₁₀ t₅₀ values against the nonreduced form of the purified D-2V NS1 protein were determined by using sera collected from each mouse 2 weeks after the first (gray bars) and second (black bars) immunizations. Strains that were high and low responders were identified when the log₁₀ t₅₀ ELISA titers of each mouse/group were >1.0 and 2.0 or <1.0 and 2.0 after the first and second infections, respectively.

FIG. 3.

Cross-reactions of PAbs generated to live D-2V in congeneic mice with human platelets and fibrinogen. The reciprocal log₁₀ t₅₀ values for pools of sera from congeneic mouse strains B10.RIII, B10.S, C57BL/BJ, B10.BR, B10.A, and B10.D2N, infected twice with live D-2V or with virus-free medium [B10.S(C) (control)], against the nonreduced forms of the purified D-2V NS1 protein, human platelets, human fibrinogen, human serum albumin [ALBUMIN (H)], or chicken egg albumin [ALBUMIN (O)] were determined.

FIG. 4.

Design of AFLX1 peptide. The AFLX1 peptide contained the 110- to 129-amino-acid sequence of the D-2V (strain PR159S1 and TR1751) NS1 protein, with the LX1 epitope (shown in boldface), the ¹²¹AKMLST¹²⁶ sequence predicted to be an H-2 I-A^d binding motif (cumulative score, >400) (27), a natural glutamic acid (E) (negative [-ve] charge) residue at the amino terminus and natural histidine (H) (positive [+ve] charge), and an unnatural cysteine (C) residue at the carboxy terminus. The molecular shape (C, random coil; H, alpha helix; E, extended strand; X, no consensus), predicted by using a consensus of eight computer algorithms, is shown. The amino acid substitutions which occur within the LX1 epitope from the NS1 proteins of D-1V, D-3V, and D-4V are underlined, and the cumulative I-A^d binding scores in their corresponding 6-amino-acid sequences are shown.

FIG. 5.

Immunoblot of the nonreduced and reduced AFLX1 peptide on nitrocellulose and nylon membranes. Two concentrations (10 and 2.5 μg) of the AFLX1 peptide were nonreduced (lanes −) or reduced (lanes +) with 2-mercaptoethanol (2ME), subjected to 15% (wt/vol) SDS-polyacrylamide gel electrophoresis, electroblotted onto nitrocellulose or nylon membranes, and detected with MAb 3D1.4. The locations of the monomer (2.5-kDa) and dimer (5.0-kDa) forms of the peptide are shown.