Haipeng Wang, Chunsuo Tian, Hongda Yang, Jia Xu, Dingbin Liu, Longhai Guo, Xueguang Shao, Wensheng Cai
Cryopreservation is challenged by both ice recrystallization damage and microbial contamination. Although antifreeze peptides (AFPTs) offer excellent sequence programmability and biocompatibility, integrating ice-growth inhibition and antimicrobial membrane disruption within a single short peptide scaffold remains a major molecular design challenge. Here, comparative structural analysis of collected α-helical AFPTs and α-helical antimicrobial peptides (AMPs) revealed a shared amphipathic α-helical organization characterized by directional radial segregation of hydrophilic and hydrophobic surfaces. We term this anisotropic architecture the Janus α-helix and propose it as a common interfacial framework for coupling antifreeze and antimicrobial functions. Guided by this framework, we reengineered a deep learning-identified AMP into P-Janus, a short α-helical peptide with enhanced Janus amphiphilicity. Molecular dynamics simulations and experiments demonstrated that P-Janus exhibits stable adsorption at the ice-water interface, suppresses ice recrystallization, and promotes persistent pore formation in bacterial membranes, thereby enhancing antimicrobial activity while maintaining low hemolytic activity. We further established a virtual screening workflow integrating Janus α-helix structural constraints with antifreeze activity prediction to mine natural AMP space. Combined computational and experimental validation identified five additional AMP-derived peptides with pronounced antifreeze activity. Comparison of these candidates revealed a graded structure-function relationship, in which stronger Janus-like radial face segregation generally promotes more effective interfacial recognition and antifreeze performance. Collectively, these findings establish the Janus α-helix as a transferable interfacial design principle for engineering short multifunctional peptides with integrated antifreeze and antimicrobial activities.