The evolutionary landscape and serotypic dynamics of avian infectious bronchitis virus from spike protein

Avian infectious bronchitis virus (IBV), a gammacoronavirus with substantial agricultural impact, offers a tractable model for dissecting coronavirus evolution. Here, we integrated 20 years of epidemiological surveillance with whole-genome sequence analysis of 624 IBV strains, including 136 newly isolated field samples, to investigate the evolutionary and structural dynamics of N-linked glycosylation at the spike protein. We identified three dominant glycosylation haplotypes defined by residues 51 and 77 of spike protein, which correlate with receptor-binding interfaces, clinical phenotypes, and spatiotemporal transmission patterns. Molecular modeling and docking analyses provided insights into potential mechanistic links between glycan positioning and Neu5Acα2-3Galβ1-3GlcNAc receptor engagement. Complementing these findings, we developed a proof-of-concept machine learning model that shows potential for predicting clinical serotypes directly from the spike protein sequence, achieving high accuracy on a preliminary independent validation set. These findings support the use of glycosylation motifs as structural-genomic markers and highlight the potential of sequence-based serotype prediction. Our work establishes a scalable genomic-structural framework that leverages glycosylation motifs and sequence features as evolutionary markers, providing a powerful approach for forecasting coronavirus adaptation and informing vaccine design and outbreak preparedness.