Z. Liu, Y. Zhou, C. Liu, C. T. Brown, L. Wang
Vibrio parahaemolyticus (Vp) is the major foodborne pathogen transmitted via shellfish products, which has posed significant threats to modern public health and resulted in significant economic damage to the seafood industry. Numerous studies have documented diverse aspects of Vp pathogenicity, among which thermostable direct haemolysin (TDH) and TDH related haemolysin (TRH) are considered as the major virulence biomarkers of Vp. Despite their established roles as key biomarkers, a systematic understanding of the divergence of TDH and TRH across sequence, structure, and function remains limited. In this study, a multi scale analysis of TDH and TRH was performed using publicly available (2131 and 99 records from NCBI and Uniprot database, respectively) amino acid sequence data combined with representation learning and structure prediction. Global alignment of curated TDH and TRH sequences revealed extensive, distributed mutations and clear separation between TDH and TRH at the amino acid level (percentage identity of between TDH and TRH ranging from 56.1 to 67.4%). In contrast, protein language model derived embeddings showed high global functional similarity while preserving distinct clustering patterns, indicating conserved core functionality alongside nuanced divergence echoed with the structural inference by AlphaFold. Importantly, modeling of mutation trajectories demonstrated that the transition from TDH to TRH is driven by accumulated, genome-wide residue changes rather than a small set of key mutations. Together, these results suggested that TDH and TRH represent functionally conserved yet evolutionarily diverged toxins driven by accumulated sequence variation throughout the full-length amino acid sequence. These accumulated point mutations lead to major structural difference: TRH forms an alpha-helical tail that TDH lacks, which suggests that TDH and TRH disrupt host membranes by different mechanisms despite their conserved core function, such as pore-forming and ion flux induction capability. These methods provided unprecedented detailed insights into the functional and structural properties of Vp haemolysin, offering critical information on how multi-dimensional variations in sequence might influence their role in Vp pathogenicity. Insights from this study reinforce the rationale for using Vp strains harboring tdh and trh genes in experimental design for environmental fitness investigation.