A perspective on the evolution of artificial silk fibre research

For over a century, artificial silk spinning has pursued the exceptional mechanical performance of natural fibres, largely through a molecular-centric strategy. Limitations in artificial systems have been attributed to insufficient molecular weight, incomplete sequence architecture, or loss of native terminal domains, factors also thought to disrupt liquid-liquid phase separation (LLPS) and hierarchical assembly. LLPS is proposed to concentrate and pre-organise silk proteins, facilitating alignment during spinning and formation of hierarchical structures that underpin mechanical performance. Here, we reevaluate this paradigm through comparisons of regenerated silk fibroin (RSF), recombinant silk proteins, and regenerated undegummed silk (RUS). Advances across these systems have substantially narrowed the molecular gap with native silk, and under optimised conditions, all can produce fibres with comparable mechanical properties. While LLPS and hierarchical organisation can be induced in RSF and recombinant systems, these features do not consistently improve mechanics, suggesting native-like assembly alone is insufficient. RUS, which retains multicomponent interactions, most closely reflects the native system, yet still exhibits distinct rheology. Collectively, this indicates that artificial dopes lack compositional complexity and ability to respond to dynamic physiochemical gradients of the silk gland. Future progress in artificial spinning will require reconstructing the multicomponent, non-equilibrium environment governing silk assembly in nature.