Pathways to folding, nucleation events and native geometry

We perform extensive Monte Carlo simulations of a lattice model and the Go potential to investigate the existence of folding pathways at the level of contact cluster formation for two native structures with markedly different geometries. Our analysis of folding pathways revealed a common underlying folding mechanism, based on nucleation phenomena, for both protein models. However, folding to the more complex geometry (i.e. that with more non-local contacts) is driven by a folding nucleus whose geometric traits more closely resemble those of the native fold. For this geometry folding is clearly a more cooperative process.

💡 Research Summary

In this work the authors employ extensive Monte Carlo simulations of a lattice protein model equipped with a Go‑type potential to interrogate how folding pathways emerge at the level of contact‑cluster formation. Two distinct native topologies are examined: a relatively simple structure in which most native contacts are local along the chain, and a more intricate geometry characterized by a high proportion of long‑range (non‑local) contacts. For each topology thousands of independent folding trajectories are generated, and the order and timing of contact formation are recorded. Contacts that appear together or in close succession are grouped into “clusters,” allowing the authors to reconstruct a stepwise picture of the folding process and to identify recurring pathways.

The central finding is that both systems share a common underlying mechanism: a nucleation event in which a small subset of native contacts—designated the folding nucleus—forms early and stabilizes a partially folded core. Subsequent contacts then collapse onto this core in a rapid, cooperative manner. However, the composition and geometry of the nucleus differ markedly between the two topologies. In the simple topology the nucleus consists mainly of short‑range, sequential contacts and involves roughly 10–15 % of the total native contacts. In contrast, the complex topology’s nucleus is larger (≈25–30 % of contacts) and is dominated by long‑range interactions that closely resemble the overall native fold. Because the nucleus of the complex protein already mirrors the final geometry, its formation sharply lowers the free‑energy barrier, leading to a highly cooperative folding transition: once the nucleus appears, the remaining contacts are added almost simultaneously.

Kinetic analysis further reveals that while the average folding times for the two proteins are comparable, the distribution of folding times is much narrower for the complex topology, indicating a more deterministic pathway. The simple topology displays a broader distribution, reflecting multiple possible routes after nucleus formation. Transition‑state analysis shows that the complex protein reaches a transition state with 60–70 % of native contacts already formed, whereas the simple protein’s transition state contains only 30–40 % of contacts. This underscores the greater “pre‑organization” of the complex system.

The authors interpret these results in the context of real proteins. They argue that proteins rich in non‑local contacts are likely to fold via a nucleus that is geometrically similar to the final structure, producing a sharp, cooperative folding event. Proteins with predominantly local contacts may follow a more gradual, stepwise pathway, offering greater conformational diversity and potentially higher robustness to mutations or environmental fluctuations.

Methodologically, the study demonstrates that a coarse‑grained lattice model combined with a Go potential can capture essential aspects of nucleation‑driven folding while remaining computationally tractable. The contact‑cluster framework provides a clear, quantitative way to compare folding pathways across different native geometries. The authors acknowledge that the model omits many atomic‑level details, but they suggest that extending the approach to off‑lattice or all‑atom representations, and correlating the identified nuclei with experimental φ‑value analyses, would be valuable next steps.

In summary, the paper establishes that nucleation is a universal feature of folding in these simplified models, but the structural character of the nucleus—and consequently the cooperativity of the folding transition—is strongly dictated by the native geometry, especially the prevalence of long‑range contacts. This insight bridges the gap between abstract folding models and the diverse folding behaviors observed in real proteins.