Pangenome-guided immunoinformatics design and in silico characterization of a multi-epitope vaccine candidate against Acinetobacter baumannii with nanoparticle assembly potential.

Abstract

Acinetobacter baumannii is a critical multidrug-resistant pathogen causing severe healthcare infections with high mortality, yet no licensed vaccine exists. This study aims to identify universally conserved surface antigens through pangenome analysis, predict immunogenic epitopes using integrated machine learning, and computationally design and in silico characterize a self-assembling nanoparticle vaccine with dual adjuvants. A computational framework integrating pangenome analysis of 712 complete genomes, epitope prediction, and structural vaccinology was employed to design a multi-epitope nanoparticle vaccine candidate for experimental evaluation. Pangenome analysis identified 3894 core genes with 42 outer membrane proteins, prioritizing OmpA, BamA, and OmpW as antigen targets. Protein language models predicted conformational B-cell epitopes, while NetMHCpan-4.2 predicted T-cell epitopes across 125 HLA alleles. The final construct (AB-VAX-01, 289 amino acids) incorporates 15 epitopes fused with dual adjuvants (RS09 TLR4 and cGAMP STING agonists) and a foldon domain for nanoparticle assembly. Microsecond molecular dynamics simulations with replicates demonstrated stability with TLR4 and STING. Conservation analysis across all 712 genomes showed 96.2-100% epitope identity. Immune simulations predicted Th1-biased responses with 94.2% global population coverage. In silico cloning confirmed favorable codon adaptation parameters. This in silico characterized vaccine construct represents a promising candidate requiring experimental validation against A. baumannii infections.