Biological Sciences: Biophysics

In this study, we explore nucleation and the transition state ensemble of the ribosomal protein S6 using a Monte Carlo Go model in conjunction with restraints from experiment. The results are analyzed in the context of extensive experimental and evolutionary data. The roles of individual residues in the folding nucleus are identified and the order of events in the S6 folding mechanism is explored in detail. Interpretation of our results agrees with, and extends the utility of, experiments that shift φ-values by modulating denaturant concentration and presents strong evidence for the realism of the mechanistic details in our Monte Carlo Go model and the structural interpretation of experimental φvalues. We also observe plasticity in the contacts of the hydrophobic core that support the specific nucleus. For S6, which binds to RNA and protein after folding, this plasticity may result from the conformational flexibility required to achieve biological function. These results present a theoretical and conceptual picture that is relevant in understanding the mechanism of nucleation in protein folding. Understanding the transition state (TS) is among the major technical and intellectual milestones towards understanding the protein folding reaction (1). Several recent studies (2-6) have attempted to construct transition-state ensemble (TSE) structures by utilizing φ-values as structural restraints in unfolding simulations. Through extensive studies of the experimentally and computationally benchmarked protein G (7), we have shown that experimental φ-values (φ) may be employed in simulation to construct a putative TSE, but that measurement of a conformation’s transmission coefficient (“probability to fold”, pfold) is the only means by which a structure may be classified as a member of the TSE. However, one must also be cautious in choosing which φ to use since the point mutations on which they are based may alter protein stability or structural features of the TSE, making normalization to the wild-type data ambiguous (8). Given that our method for studying the structure of the TSE has been validated in the complicated case of protein G folding (7), we are now able to carry out such analysis for other proteins on a comparative basis to aid in the distinction between experimental inconsistencies, noise, or artifacts and to determine the common denominators of the critical nucleus in protein folding. The split β-α-β ribosomal protein S6 from Thermus thermophilus consists of 97 residues in a four-stranded β-sheet packed against two α-helices with a hydrophobic core (9). Functionally, S6 binds to both RNA and its protein partner S18 in a cooperative manner during the intermediate stage of 30S ribosomal subunit formation (10). S6 is an ideal candidate for computational study, due to the large body of high-quality experimental data available, including extensive kinetic and φ-value data at varying denaturant concentrations (11), circular-permutant studies that reflect rearrangements in the TSE (12), and studies of salt-induced off-pathway intermediates (13). Detailed structural information also exists for the function of S6 (10). We use simulation to structurally interpret the combined set of S6 φ-values. We begin by generating an ensemble of structures consistent with structural restraints based on φ. After characterizing each ensemble conformation by measuring its pfold, we construct a detailed model of the TSE and the events occurring before and after nucleation. This formalized treatment of φ allows microscopic analysis and reconstruction of the folding nucleation process. Our results support the idea that the experimentally observed denaturant-induced shifts in φ-values shift the TSE along the free-energy profile and hence probe events earlier or later along the folding pathway (as interpreted through the Hammond postulate). We also analyze residue conservation patterns of S6 to determine the evolutionary history of nucleus residues in the split β-α-β family. Through a combination of experimental data, evolutionary information, and allatom simulation we are able to extend interpretation of experimental data and create unified and ordered, atomic level description of nucleation, the transition state and folding in the S6 protein. Theory and Methods Model system. Our protocol was previously implemented for reconstructing the TSE of CI2 (6) and protein G (7) and has been used to simulate the complete folding of protein G (14) and crambin (15) from random coil to < 1Å distance RMS (dRMS) from the native state. The model includes a hard sphere representation of all non-hydrogen atoms in the backbone and side chains, a full representation of backbone and side chain rotational degrees of freedom, a square well Go potential (16) with native contacts having a -1 attraction and non-native contacts having a +1 repulsion, and a MC move set (with localized backbone and side chain moves) that maintains chain connectivity, planar peptide bonds, and excluded volume at each step. As a measure of structural similarity to the native state, dRMS is computed as D− D0 ( 2 ) where D and D0 are pairwise Cα distances in a selected and native conformation. This model has the advantage of allowing for a statistically significant number of trajectories to be collected while including atomic-resolution details, such as side chain packing. Go potentials adequately represent the thermodynamics and kinetics of proteins with minimal energetic frustration and allow the complete folding process to be studied (17). Their use is also motivated by the theoretical and experimental finding that transition state is robust with respect to selection of specific sequences that fold to a given structure and potentials sets used to design and fold sequences (18-20). Go models have been used to propose folding mechanisms (21), predict folding rates, and interpret φ-values (22-24). They have also successfully predicted φ-values for several proteins (25-27). Presently, there is no general potential capable of folding α/β proteins so Go potentials present the best option for studying the folding of small proteins (28). Constructing putative TS conformations. Structures were constructed via constrained unfolding simulations (7) from the native PDB structure (1RIS). A common interpretation of φ is the fraction of native contacts made by a particular residue in the TSE. We define a simulation φ-value (φ), which may be calculated for any residue, as the fraction of native contacts made by residue i in conformation k.