De Novo Design of Specific Heterotrimeric Collagen‐Like Peptides via Genetic Algorithm

Abstract Collagen, the most abundant protein by mass in mammals, features a triple‐helical structure composed of three intertwined peptide strands. Nonequivalent strands can assemble, forming heterotrimeric structures, which are more prevalent among natural collagens. However, the design and application of heterotrimeric collagen mimetic peptides (CMPs) are constrained by the potential formation of competing compositions and registers, which impede the formation of target assemblies. Herein, a computational protocol is described, GRACE, utilizing a genetic algorithm to reproducibly generate peptide sequences having a strong tendency to self‐assemble into heterotrimeric triple helices with high specificity. The approach leverages SCEPTTr1.2, a refined scoring function tailored to estimate CMP stability and structure, as the basis for fitness evaluation. Four sets of peptide sequences generated by the algorithm, including two with bio‐relevant integrin‐binding motifs, are experimentally synthesized and characterized. All computationally designed peptides are experimentally shown to self‐assemble into heterotrimeric triple helices with expected registers having a minimum specificity of 13.5 °C. The study represents a robust method to overcome barriers in heterotrimeric CMP design providing a versatile framework for the fundamental study of collagen as well as engineering collagen‐like materials.