In contrast to the prevailing belief that, without evolutionary selection, protein folding would take longer than the age of the universe, we showed that the tendency for hydrophobic parts of the protein to preferentially avoid water is sufficient to reduce the relevant conformational landscape such that protein can randomly search for their native fold within seconds. The Lin lab is actively pursuing the structure of the much larger conformation space of protein aggregation and assembly.
Protein self-assembly into large complexes is a ubiquitous and crucial aspect of sub-cellular function. However, protein aggregates can also misfold or become infection agents. Understanding the molecular structure and aggregation pathway is the key to specifically combating neurodegenerative diseases and viral infection. Yet the size of the conformation space for protein self-assembly can be tens or hundreds of orders of magnitude larger than that for protein folding. We are pursuing a new computational approach to solve the search problem by taking advantage of the observation that identical copies of proteins are often identically folded when assembled, allowing for orders of magnitude speedup while avoiding problems of non-ergodicity and thermodynamic bias.
The energy scale of thermal fluctuations at ambient conditions are comparable to the constituent interaction energies that stabilize macromolecular structures. Therefore, thermal noise is a fundamental bottleneck to achieving functional efficiency and structural stability. We’ve recently shown that intramolecular correlations in timing allows proteins to exceed this thermodynamic communication limit, explaining how long-range intra-protein communication can occur without measureable conformational change. We devised a new function, called the conditional activity, which measures the correlations in timing between events, which is applicable for systems that are non-Markovian (i.e. history-dependent). In addition to proteins, we are applying this methodology to find collective dynamical modes in gene regulatory and neural networks that are not captured by existing structural correlation functions such as the mutual information.
At the micron scale, the relevant self-assembly process is not the intramolecular structure of the individual proteins, but rather the intermolecular architecture of the entire network. In particular, the collective interactions of self-assembly can result in phase transitions, enabling small changes in the analog input to effect all-or-none binary transitions (ultrasensitivity). The most ubiquitous protein assembly process involves the actin cytoskeleton, which controls many aspects of mechanical motion, transport, and cell shape. We are developing analytical models as well as generalized Ising simulations of the dynamical assembly process which incorporates the effects of the auxiliary proteins affecting actin filament stability and branching. The goal is to predict and control the phase transitions in actin networks as a function of actin and auxiliary protein concentration.