Identification of candidate serotype-specific B-cell epitopes from dengue virus non-structural proteins using an integrated in silico approach.

Dengue is a vector-borne disease transmitted through mosquitoes that pose serious health issues globally. Dengue infection is caused by dengue virus (DENV) that circulates as four serotypes in endemic regions and accurate serotype differentiation is hindered by widespread serological cross-reactivity. DENV is composed of three structural and seven non-structural proteins and in the current study, all seven non-structural (NS) proteins were analyzed using an integrated in silico approach to identify candidate serotype-specific B-cell epitopes carrying potential diagnostic relevance. Multiple sequence alignment, pairwise identity assessment, and phylogenetic relationship of all four DENV serotype amino acid sequences were carried out. Linear B-cell epitopes were predicted using the BepiPred tool with parameters optimized to balance sensitivity and specificity and the ABCPred tool for further sequence-based epitope prediction. This was later followed by structure-based surface epitope prediction using ElliPro. The predicted epitopes were subsequently subjected to a two-step conservancy analysis to evaluate intra-serotype conservation and pan-serotype divergence. The multiple sequence alignment showed that high sequence conservation (~ 98-99%) was observed overall across NS proteins, yet specific aminoacid substitutions were identified that enabled serotype-level discrimination. Initial B-cell epitope prediction resulted in 355 epitopes and 1067 epitopes from BepiPred and ABCPred respectively. Antigenicity, allergenicity and toxicity filters were applied to retain potential antigenic epitopes. Further, initial stringent filtering criteria (≤ 75% intra-serotype conservation and ≤ 30% pan-serotype minimum identity) were applied to increase serotype-specific strictness of the candidate epitopes. Altogether all non-structural proteins yielded 121 epitopes from the BepiPred, ABCPred and ElliPro tools with stricter 50-75% intra-serotype conservation, and ≤ 25% pan-serotype minimum identity. Also, few of the predicted epitopes were in concordance with previous in silico or experimentally validated antigenic regions reported, supporting their biological relevance. Overlap with surface epitopes and stringent filtering suggest that the DENV NS proteins are made of distinct antigenic regions that makes them as potential candidates for serotype-specific differentiation. This workflow for epitope prediction and filtering provides a computational basis for utilizing epitopes for future experimental validation and diagnostic assay development.