Cross-reactive antibodies targeting surface-exposed non-structural protein 1 (NS1) of dengue virus-infected cells recognize epitopes on the spaghetti loop of the β-ladder domain

Abstract

Non-structural protein 1 (NS1) is a glycoprotein component of dengue virus (DENV) that is essential for viral replication, infection and immune evasion. Immunization with NS1 has been shown to elicit antibody-mediated immune responses which protect mice against DENV infections. Here, we obtained peripheral blood mononuclear cells from human subjects with secondary dengue infections, which were used to construct a dengue immune phage library displaying single-chain variable fragments. Phage selective for DENV NS1 were obtained by biopanning. Twenty-one monoclonal antibodies (mAbs) against DENV NS1 were generated from the selected phage and characterized in detail. We found most anti-NS1 mAbs used IGHV1 heavy chain antibody genes. The mAbs were classified into strongly and weakly-reactive groups based on their binding to NS1 expressed in dengue virus 2 (DENV2)-infected cells. Antibody binding experiments with recombinant NS1 proteins revealed that the mAbs recognize conformational epitopes on the β-ladder domain (amino acid residues 178-273) of DENV NS1. Epitope mapping studies on alanine-substituted NS1 proteins identified distinct but overlapping epitopes. Protruding amino acids distributed around the spaghetti loop are required for the binding of the strongly-reactive mAbs, whereas the recognition residues of the weakly-reactive mAbs are likely to be located in inaccessible sites facing toward the cell membrane. This information could guide the design of an NS1 epitope-based vaccine that targets cross-reactive conserved epitopes on cell surface-associated DENV NS1.

Fig 1

B-cell germline genes and CDR3 lengths represented among scFv clones of the (A) heavy chain and (B) light chains. The germline antibody family representation is shown in the pie charts with the total of scFv clones (top). The gene frequency of germline IGHV and IGLV/IGKV genes represented among scFv clones (middle) and the amino acid lengths of CDR3H and CDR3L (bottom) are shown in the bar graphs. The grey and white bars represent scFv clones isolated from the library before and after selection for DENV-NS1 scFv-binders, respectively.

Fig 2. NS1-specific phage clones.

(A) The reactivity of the scFv phages biopanned with pooled DENV NS1 was measured by monoclonal phage ELISA using pooled DENV1-4 NS1. (B) & (C) Selected scFv phage clones from the first (pooled DENV1-4 NS1) and second biopanning (separated NS1 from each serotype) experiments were tested with each DENV1-4 NS1 by phage ELISA. The cut-off value shown in (A-C) is OD_{450/620 nm.} = 0.1. The values are from a single experiment. (D) ELISA signal intensity of the scFv phage’s binding to DENV NS1 obtained from two experiments is shown in a heat map. The spectrum of red to white indicates the range of good to no binding of scFv phages on DENV NS1.

Fig 3. Characterization of anti-NS1 mAbs specific to DENV NS1.

(A) Reformatted IgG1 antibodies were tested for NS1 serotype specificity by ELISA using each purified sNS1 derived from dengue infected Vero cells. The cut-off value is 0.1 and the values are from a single experiment. (B) Intracellular NS1 staining (blue) and surface-exposed NS1 (pink) on DENV2- infected imHC cells are shown as % positive cells (mean ± S.D.) from three independent experiments by flow cytometry. The solid line indicates the threshold for antibody binding activity (20% of positive cells) and the dot line represents the cut-off value of detection (5% of positive cells). (C) Either purified DENV2 sNS1 or BSA was incubated with uninfected imHC cells. Binding reactivity of the mAbs to binding-back NS1 was tested in three independent experiments by flow cytometry. The percentage of positive cells are represented as mean ±S.D. The dot line represents the cut-off value of detection (5% of positive cells).

Fig 4. Identification of recognition sites of anti-NS1 mAbs on chimeric rDENV2-ZIKV NS1 proteins.

(A) Ten soluble rDENV2-ZIKV NS1 chimeras, wt DENV2 and ZIKV NS1 were spotted onto nitrocellulose membrane and detected with different anti-NS1 mAbs. The mAbs were classified into groups A–D according to the detection patterns observed. Cell supernatant of non-transfected cells (mock) was used as the negative control. Expression of the NS1 proteins was confirmed by anti-6xHis.tag mAb and anti-ZIKV NS1 pAb. (B) The putative recognition sites of the indicated mAbs are shown with respect to the rDENV2 NS1 residues numbered from N to C-terminus with three structural domains indicated: β-roll (aa 1–29; purple), wing (aa 38–151; orange), and β-ladder (aa 181–352; red), respectively. The recognition sites of each antibody group are indicated by the colored regions (yellow, group A; green, group B).

Fig 5. 3D model illustration for the putative recognition sites on DENV2 NS1 structure.

The putative binding sites for the mAbs in groups A and B are highlighted in yellow and green, respectively. The NS1 structure is shown as outer and inner faces as well as the side view. NS1 domains including the β-roll (purple), wing (orange), and β-ladder (red) are indicated in the NS1 structure.

Fig 6. Epitope mapping of anti-NS1 mAbs by alanine substitution.

Nine NS1 mutants, with predicted surface residues replaced with alanine in the positions indicated, were used to identify the critical binding sites of the strongly and weakly-reactive mAbs to surface-exposed NS1 on DENV infected cells. (A) Core and peripheral amino acid residues for the antibody recognition identified from western blot experiments (S7 Fig) are shown as red and yellow boxes, respectively. A pool of mouse anti- DENV NS1 mAbs was used as a positive control. (B) Location of core residues (marked in red) on the DENV2 dimeric NS1 structure on the top and side views. (C) Protein sequence alignment of NS1 is shown for DENV1 (Hawaii), DENV2 (16681), DENV3 (H87), and DENV4 (H241). DENV2 conserved residues are highlighted in yellow. Colored bars above the alignment indicated the β-ladder domains (aa 181–352). The dashed line indicated the surface-exposed spaghetti loop (aa 219–272). Core amino acid residues of the strong and weak reactive mAbs are indicated by the red and blue boxes, respectively.