[Paper Review] Folding Mechanism of Small Proteins

Seung‐Yeon Kim, Julian Lee+1 more|arXiv (Cornell University)|Jun 20, 2003

Protein Structure and Dynamics22 citations

TL;DR

This study presents a novel Monte Carlo folding simulation method using a single, optimized atomistic potential to fold four small proteins (betanova, 1fsd, HP-36, and protein A) into their native-like structures. The approach reveals that folding is governed by both thermodynamic stability and kinetic factors, with early folding trajectories converging to specific collapsed states that dictate the folding pathway—horizontal or diagonal—relative to the native state, a dynamic not captured by free energy surfaces alone.

ABSTRACT

Extensive Monte Carlo folding simulations for four proteins of various structural classes are carried out, using a single atomistic potential. In all cases, collapse occurs at a very early stage, and proteins fold into their native-like conformations at appropriate temperatures. The results demonstrate that the folding mechanism is controlled not only by thermodynamic factors but also by kinetic factors: The way a protein folds into its native structure, is also determined by the convergence point of early folding trajectories, which cannot be obtained by the free energy surface.

Motivation & Objective

• To develop a unified simulation method capable of folding multiple small proteins into their native states using a single atomistic potential.
• To investigate the interplay between kinetic and thermodynamic factors in protein folding beyond traditional free energy surface analysis.
• To understand how early folding trajectories influence the final folding pathway and convergence to the native structure.
• To examine the role of structural class (e.g., β-sheet vs. α-helix bundle) in determining folding mechanisms and cooperativity.
• To explore glassy behavior and non-ergodic dynamics in low-temperature folding simulations.

Proposed method

• Employed united-residue (UNRES) force field with continuous, atomistic-like interactions including electrostatic, van der Waals, and multibody terms.
• Used conformational space annealing to identify low-energy minima and optimize force field parameters simultaneously for four proteins.
• Applied Metropolis Monte Carlo dynamics with small-angle perturbations to simulate folding from non-native conformations over up to 10⁹ MCS.
• Tracked folding using RMSD from native structure, radius of gyration (Rg), and native contact fractions (Q and ρ) as key observables.
• Defined ρ as the weighted average of native contact probabilities from simulations of the native state, capturing fluctuation effects.
• Analyzed trajectory convergence in (ρ, Rg) space to distinguish kinetic folding routes (horizontal vs. diagonal).

Experimental results

Research questions

• RQ1 How does the early convergence point of folding trajectories influence the overall folding pathway in (ρ, Rg) space?
• RQ2 To what extent do kinetic factors, rather than just thermodynamic stability, determine the folding mechanism?
• RQ3 Can a single, optimized atomistic potential successfully fold multiple proteins of different structural classes into their native-like conformations?
• RQ4 What role does secondary structure content (β-sheet vs. α-helix) play in shaping folding kinetics and cooperativity?
• RQ5 How does temperature affect the emergence of glassy, non-ergodic behavior in folding simulations?

Key findings

• All four proteins folded into native-like conformations with RMSD values as low as 0.78 Å (betanova) and 1.07 Å (1fsd), indicating high structural accuracy.
• Rapid collapse occurred within ~10⁴ MCS, followed by slower folding, with distinct folding pathways observed in (ρ, Rg) space.
• For betanova and 1fsd, early trajectories converged to (ρ, Rg) ≈ (0.35, 8.5 Å) and (0.3, 9 Å), respectively, followed by horizontal movement toward the native state.
• For HP-36 and protein A, trajectories converged to (ρ, Rg) ≈ (0.2, 11 Å) and (0.25, 12 Å), respectively, with more diagonal folding paths.
• Glassy, non-ergodic behavior was observed at low temperatures (T ≤ 50 for betanova, T ≤ 60 for HP-36), disappearing at T ≥ 70–90.
• A first-order-like collapse transition was observed near T = 120 for protein A, shifting from (Q, Rg) ≈ (0.15, 18 Å) to (0.15, 12 Å).