CryoGuard: De Novo Design of a Biomimetic Chimeric Cryoprotectant via In Silico Biophysical Modeling

Cryopreservation of therapeutic cells relies on dimethyl sulfoxide (DMSO), a membrane-permeant cryoprotectant whose cytotoxicity and infusion-related adverse effects are well documented, whereas DMSOfree protocols expose cells to mechanical injury from ice growth and recrystallization. Ice-binding proteins (IBPs) inhibit ice recrystallization very effectively but do not cross cell membranes, and to our knowledge no single molecule has been described that couples intracellular delivery, ice-growth control and anti-apoptotic activity. Here we report the in silico design of CryoGuard, a single-chain chimeric protein that combines a non-viral cell-penetrating module, a pH-responsive endosomolytic module, the mature domain of the hyperactive ice-binding protein from Flavobacterium frigoris PS1, and a humanin-derived BAX-directed cytoprotective module, with rigid α-helical spacers that are intended to decouple the functional units. Using an entirely computational workflow (ExPASy ProtParam, AlphaFold 3, CABS-flex 2.0, HDOCK docking against a GenIce-generated hexagonal-ice lattice and against BAX, and a 20 ns explicit-solvent moleculardynamics simulation in OpenMM), we obtained a 376-residue design with a predicted instability index of 27.51, GRAVY of −0.039 and pI of 6.72; a confidently predicted ice-binding core (pTM 0.61, mean pLDDT>85); sub-ångström predicted core fluctuations; geometrically plausible ice- and BAX-interface poses; and a backbone that remained below 2.0 Å RMSD over 20 ns. These are predictions, not measurements: no experimental evidence of function is reported. We state falsifiable hypotheses, the limitations of each method, and a staged validation plan. Because a patent application is intended, the complete sequence and other enabling details are deliberately withheld; a cryptographic commitment to the sequence is provided instead.