Hydrophobic core formation and dehydration in protein folding studied by generalized-ensemble simulations

Abstract

Despite its small size, chicken villin headpiece subdomain HP36 folds into the native structure with a stable hydrophobic core within several microseconds. How such a small protein keeps up its conformational stability and fast folding in solution is an important issue for understanding molecular mechanisms of protein folding. In this study, we performed multicanonical replica-exchange simulations of HP36 in explicit water, starting from a fully extended conformation. We observed at least five events of HP36 folding into nativelike conformations. The smallest backbone root mean-square deviation from the crystal structure was 1.1 A. In the nativelike conformations, the stably formed hydrophobic core was fully dehydrated. Statistical analyses of the simulation trajectories show the following sequential events in folding of HP36: 1), Helix 3 is formed at the earliest stage; 2), the backbone and the side chains near the loop between Helices 2 and 3 take nativelike conformations; and 3), the side-chain packing at the hydrophobic core and the dehydration of the core side chains take place simultaneously at the later stage of folding. This sequence suggests that the initial folding nucleus is not necessarily the same as the hydrophobic core, consistent with a recent experimental phi-value analysis.

Figure 1

Time courses of backbone RMSD (a, d, and g), the number of nonlocal interresidue native contacts formed (N_n) (b, e, and h), and radius of gyration (R_g) (c, f, and i) of HP36. Here, backbone RMSD was calculated using N, C^α, and C atoms of the backbone, and R_g was calculated with respect to heavy atoms. Three of 16 replicas were chosen here: replica 5 of MUCAREM1 (a–c), replica 8 of MUCAREM1 (d–f), and replica 5 of MUCAREM2 (g–i). For native contacts, nonlocal means that the two contacting residues are in different secondary-structure elements (Helix 1: residues 4–11, Loop 1: residues 12–14, Helix 2: residues 15–18, Loop 2: residues 19–22, and Helix 3: residues 23–32) and not next to each other in the primary sequence. Two side chains are defined as in contact if the distance between any two heavy atoms in the side chains is ≤4.5 Å. In the native structure (PDB code: 1YRF), there are 18 nonlocal native side-chain contacts. The horizontal dashed lines in a, d, and g indicate backbone RMSD = 3.0 Å. Those in b, e, and h represent N_n = 10, and those in c, f, and i represent the R_g value of the native conformation (9.4 Å).

Figure 2

Low-RMSD conformations observed in the MUCAREM1 and MUCAREM2 production runs (orange). The x-ray structure (PDB code 1YRF) is also shown (blue and green). Here, the α-helices in the x-ray structure are green and the others are blue. Three phenylalanine side chains (Phe⁷, Phe¹¹, and Phe¹⁸) are shown in ball-and-stick representation. (a) Lowest backbone RMSD (with respect to N, C^α, and C atoms) conformation observed in the simulations presented here (replica 5 of MUCAREM2). The backbone RMSD value is 1.1 Å (for nonterminal 34 residues). (b) A low-RMSD conformation observed in MUCAREM1 (replica 8). The RMSD value is 1.0 Å for residues 9–32 and 3.3 Å for the nonterminal 34 residues. These figures were prepared using VMD (61).

Figure 3

(a and b) The probability that the number of water molecules lying within 3.5 Å of side-chain nonhydrogen atoms in Phe⁷, Phe¹¹, or Phe¹⁸ is <4. The solid curves represent the results at 300 K and the dashed curves those at 350 K. (c) The probability that the number of water molecules lying within 3.5 Å of side-chain nonhydrogen atoms in Phe¹⁸, Leu²¹, Gln²⁶, or Leu²⁹ is <9. The probability values were evaluated as functions of the backbone RMSD (in a) or N_n′ (in b and c), where N_n′ is defined as the integer part of N_n/2. These results were obtained by WHAM applied to the data from MUCAREM1 and MUCAREM2.

Figure 4

(a–c) Results of secondary-structure analysis of HP36. Canonical expectation values of α-helix contents for each residue evaluated by WHAM applied to the data from MUCAREM1 and MUCAREM2. Values at five temperature values are shown: 300 K (solid line), 350 K (dashed line), 400 K (short-dashed line), 500 K (dotted line), and 600 K (dash-dotted line). HP36 conformations are divided into three groups according to the number of nonlocal interside-chain native contacts formed (N_n): average helix contents for conformations with N_n ≤ 7 (a), N_n = 8 or 9 (b), and N_n ≥ 10 (c). (d) Canonical expectation values of backbone (N, C^α, and C) RMSDs for the 15-residue regions of HP36 as functions of N_n′ at 350 K evaluated by WHAM, where N_n′ is the integer part of N_n/2, for residues 4–18 (solid line); 9–23 (long-dashed line), 14–28 (short-dashed line), and 19–33 (dotted line).

Figure 5

Representative snapshot structures with different N_n values. Snapshot structures with potential energy values typical for the canonical NVT ensemble at T = 350 K or lower are chosen from MUCAREM1 and MUCAREM2. α-helices are purple, side chains of Phe⁷, Phe¹¹, and Phe¹⁸ are green, and side chains of Leu²¹, Gln²⁶, and Leu²⁹ are blue. Numbers written near the snapshot structures represent their N_n values. Side chains of the core phenylalanine residues (Phe⁷, Phe¹¹, and Phe¹⁸) and of Phe¹⁸, Leu²¹, Gln²⁶, and Leu²⁹, and oxygen atoms in water molecules that are near (≤3.5 Å from) those side chains are shown by space-filling representation in a–c, respectively. In c, water oxygen atoms that are near a side chain of Phe⁷, Phe¹¹, or Phe¹⁸ are red, and those near a side chain of Leu²¹, Gln²⁶, or Leu²⁹ are orange.