Nucleation of ordered solid phases of proteins triggers numerous phenomena in laboratory, industry, and in healthy and sick organisms. Recent simulations and experiments with protein crystals suggest that the formation of an ordered crystalline nucleus is preceded by a disordered high-density cluster, akin to a droplet of high-density liquid that has been observed with some proteins; this mechanism allowed a qualitative explanation of recorded complex nucleation kinetics curves. Here, we present a simple phenomenological theory that takes into account intermediate high-density metastable states in the nucleation process. Nucleation rate data at varying temperature and protein concentration are reproduced with high fidelity using literature values of the thermodynamic and kinetic parameters of the system. Our calculations show that the growth rate of the near-critical and supercritical ordered clusters within the dense intermediate is a major factor for the overall nucleation rate. This highlights the role of viscosity within the dense intermediate for the formation of the ordered nucleus. The model provides an understanding of the action of additives that delay or accelerate nucleation and presents a framework within which the nucleation of other ordered protein solid phases, e.g., the sickle cell hemoglobin polymers, can be analyzed.

We probe the transport properties in protein solutions stable with respect to any, solid or liquid, phase separation as a step in the understanding of transport in the cytosol of live cells. We determine the mean-squared displacement of probe particles in the time range ${10}^{-3}–10\mathrm{s}$ in solutions of a model protein. The tested solutions exhibit significant elasticity at high frequencies, while at low frequencies, they are purely viscous. We attribute this viscoelasticity to a dense network of weakly-bound chains of protein molecules with characteristic lifetime of 10–100 ms. The found intrinsic viscoelasticity of protein solutions should be considered in biochemical kinetics models.