Structural Determinants of Misfolding in Multidomain Proteins

Recent single molecule experiments, using either atomic force microscopy (AFM) or Förster resonance energy transfer (FRET) have shown that multidomain proteins containing tandem repeats may form stable misfolded structures. Topology-based simulation models have been used successfully to generate models for these structures with domain-swapped features, fully consistent with the available data. However, it is also known that some multidomain protein folds exhibit no evidence for misfolding, even when adjacent domains have identical sequences. Here we pose the question: what factors influence the propensity of a given fold to undergo domain-swapped misfolding? Using a coarse-grained simulation model, we can reproduce the known propensities of multidomain proteins to form domain-swapped misfolds, where data is available. Contrary to what might be naively expected based on the previously described misfolding mechanism, we find that the extent of misfolding is not determined by the relative folding rates or barrier heights for forming the domains present in the initial intermediates leading to folded or misfolded structures. Instead, it appears that the propensity is more closely related to the relative stability of the domains present in folded and misfolded intermediates. We show that these findings can be rationalized if the folded and misfolded domains are part of the same folding funnel, with commitment to one structure or the other occurring only at a relatively late stage of folding. Nonetheless, the results are still fully consistent with the kinetic models previously proposed to explain misfolding, with a specific interpretation of the observed rate coefficients. Finally, we investigate the relation between interdomain linker length and misfolding, and propose a simple alchemical model to predict the propensity for domain-swapped misfolding of multidomain proteins.Author Summary: Multidomain proteins with tandem repeats are abundant in eukaryotic proteins. Recent studies have shown that such domains may have a propensity for forming domain-swapped misfolded species which are stable for long periods, and therefore a potential hazard in the cell. However, for some types of tandem domains, no detectable misfolding was observed. In this work, we use coarse-grained structure-based folding models to address two central questions regarding misfolding of multidomain proteins. First, what are the possible structural topologies of the misfolds for a given domain, and what determines their relative abundance? Second, what is the effect of the topology of the domains on their propensity for misfolding? We show how the propensity of a given domain to misfold can be correlated with the stability of domains present in the intermediates on the folding and misfolding pathways, consistent with the energy landscape view of protein folding. Based on these observations, we propose a simplified model that can be used to predict misfolding propensity for other multidomain proteins.