Two-State Folding, Folding through Intermediates, and Metastability in a Minimalistic Hydrophobic-Polar Model for Proteins

Within the frame of an effective, coarse-grained hydrophobic-polar protein model, we employ multicanonical Monte Carlo simulations to investigate free-energy landscapes and folding channels of exemplified heteropolymer sequences, which are permutations of each other. Despite the simplicity of the model, the knowledge of the free-energy landscape in dependence of a suitable system order parameter enables us to reveal complex folding characteristics known from real bioproteins and synthetic peptides, such as two-state folding, folding through weakly stable intermediates, and glassy metastability.

💡 Research Summary

In this work the authors explore how far a minimalist hydrophobic‑polar (HP) lattice model can capture the rich folding behavior observed in real proteins. They generate a set of twenty heteropolymer sequences that are permutations of one another – each sequence contains the same number of hydrophobic (H) and polar (P) residues but differs in the order of these residues along the chain. By keeping composition constant while varying only the sequence pattern, the study isolates the effect of residue ordering on the thermodynamic landscape.

The computational engine is a multicanonical Monte Carlo (MUCA) algorithm. Unlike conventional canonical simulations that sample at a fixed temperature and often become trapped in deep minima, MUCA assigns a weight to each energy level so that the resulting histogram of visited energies is flat. Consequently, rare high‑energy conformations and low‑energy native‑like states are sampled with comparable frequency. From the MUCA trajectory the authors reconstruct the density of states, and by reweighting they obtain the free‑energy surface F(Q,R; T)=−kBT ln P(Q,R) as a function of two order parameters: Q, the structural overlap with a chosen native reference, and R, the fraction of hydrophobic‑hydrophobic contacts (a measure of core formation).

Analysis of the free‑energy landscapes reveals three distinct folding scenarios.

1. Two‑state folding – For several sequences the surface exhibits a broad, high‑entropy basin at high temperature (the unfolded ensemble) and a single deep minimum at low temperature (the native state). The barrier separating them is narrow and high, leading to an abrupt, cooperative transition reminiscent of the classic two‑state protein fold.

2. Folding through weak intermediates – Some permutations display two well‑defined minima (unfolded and native) connected by a shallow saddle or a secondary minimum. This intermediate corresponds to a partially formed hydrophobic core that retains considerable conformational flexibility. In the temperature window where the intermediate is thermodynamically competitive, the population shifts transiently, reproducing the experimentally observed “folding‑through‑intermediate” pathways seen in many small peptides.

3. Glassy metastability – A third group of sequences shows a rugged landscape populated by many shallow minima and low barriers. The system wanders among numerous quasi‑stable conformations, failing to reach a unique native basin within realistic simulation times. This behavior mimics the kinetic trapping and slow relaxation characteristic of glassy protein states or misfolded aggregates. The underlying cause is the specific arrangement of H residues that frustrates the formation of a compact hydrophobic core, thereby generating competing folding routes.

The study demonstrates that even a coarse‑grained HP model, when combined with an exhaustive sampling technique such as MUCA, can reproduce the full spectrum of folding phenomenology—from highly cooperative two‑state transitions to multi‑pathway, metastable behavior. The authors argue that the sequence order, not merely composition, is a decisive factor shaping the free‑energy topography. Moreover, the ability to map the landscape in the (Q,R) plane provides a clear visual tool for identifying folding routes, transition states, and kinetic traps.

From a methodological perspective, the work highlights the superiority of multicanonical sampling over traditional temperature‑scan approaches for detecting rare events like intermediate formation or glassy trapping. It also suggests that systematic variation of model parameters (hydrophobic interaction strength, chain length, lattice geometry) could be used to tune the landscape and thereby emulate specific experimental systems.

In conclusion, the paper establishes that minimalistic HP models, despite their simplicity, are capable of capturing essential aspects of protein folding thermodynamics and kinetics. By focusing on the free‑energy landscape rather than on a single temperature trajectory, the authors provide a unifying framework that bridges the gap between abstract lattice models and the complex behavior of real biopolymers, offering a valuable reference point for future studies that aim to integrate coarse‑grained theory with experimental observations.