Surprising Abundance of Misfolding during Refolding of Multidomain Proteins

Abstract

The covalently linked domains constituting a multidomain protein typically share a similar fold and sequence, are generally stable in isolation, and are largely capable of independent folding. However, the elegant interplay of forces leading an amino acid sequence to its native state is more complex in proteins with multiple domains because of inter-domain interactions, which complicate the folding energy landscape.In previous experiments, employing immunoglobulin-like (Ig-like) domains of the human multidomain protein titin, we showed that tandem repeats of domains with high sequence identity can form a stable misfolded state upon refolding in physiological solution conditions. Conversely, tandem constructs of natural neighbouring domains, which display a low sequence identity, did not misfold. This supported the hypothesis that sequence identity between neighbouring domains in multidomain proteins is reduced as a result of evolutionary pressure to avoid misfolding.With a combination of single-molecule Forster resonance energy transfer, microfluidic mixing, stopped-flow kinetics and molecular dynamics simulations, we now demonstrate that Ig-like domains can transiently populate a surprisingly broad range of misfolded conformations on the sub-second timescale. Using tandem repeats of Ig-like domains, we can resolve both strand-swapped misfolds dominated by native-like interactions and, remarkably, a non-native-like, largely disordered type of misfolded state which so far was never observed experimentally, characterized by promiscuous interactions. Even more surprisingly, both types of misfolding are detected also for the naturally occurring tandem repeat, showing how finely the propensities of folding and misfolding have been balanced by co-evolution of adjacent domains to avoid stable misfolded states formation. On longer timescales, however, all or most of the protein molecules are able to reach the native state, demonstrating that the overall free energy surface is still sufficiently optimized for the protein to efficiently reach its correctly folded state.