Understanding the determinants of stability and folding of small globular proteins from their energetics

Abstract

The results of minimal model calculations indicate that the stability and the kinetic accessibility of the native state of small globular proteins are controlled by few "hot" sites. By means of molecular dynamics simulations around the native conformation, which describe the protein and the surrounding solvent at the all-atom level, an accurate and compact energetic map of the native state of the protein is generated. This map is further simplified by means of an eigenvalue decomposition. The components of the eigenvector associated with the lowest eigenvalue indicate which hot sites are likely to be responsible for the stability and for the rapid folding of the protein. The comparison of the results of the model with the findings of mutagenesis experiments performed for four small proteins show that the eigenvalue decomposition method is able to identify between 60% and 80% of these (hot) sites.Keywords: protein folding; protein stability; molecular dynamics; local elementary structures The study of how the structure and stability of a protein is connected with its amino acid sequence has been, during recent years, a major focus of research. In particular, the problem of protein folding and its relation to stability has been studied through a variety of experimental and theoretical techniques. These studies have highlighted the central role the free energy landscape plays in determining the properties displayed by proteins. The discussion has then been turned to the problem of determining this landscape for specific proteins (Shea and Brooks 2001).Experimental techniques have yielded detailed information on macroscopic features of protein dynamical behavior, such as folding times and stability, and on some specific issues at the level of amino acids, such as the sensitivity to mutations (cf. Fersht 1999). However, there is still no experimental procedure capable of providing insight at the amino acid level into either the folding process in its completeness or the stabilization determinants of proteins. To gain insight into these questions, one must turn to theoretical and computational methods.Because of the high computational costs involved, realistic models that provide an exhaustive description of the free energy landscape have been used in the study of only short peptides. Ferrara and Caflish (2000), for instance, could reconstruct the whole free energy landscape of a small designed ␤-sheet peptide with an all-atom representation of the solute and an implicit model of the solvent. Daura et al. (1998) were able to demonstrate the reversible folding of a small (seven residues) helix forming ␤-peptide in methanol solvent using long molecular dynamics (MD) simulations.On the other hand, minimal models (Chan and Dill 1991;Sali et al. 1994;Mirny and Shakhnovich 2001;Micheletti 2003) have provided important results about the general features of the free energy landscape of proteins. They give an approximate description of both the interaction energy among amino acids (usually through a contact potential en-

Selecting sequences that fold into a defined 3D structure: A new approach for protein design based on molecular dynamics and energetics

“…The energy decomposition method is based on the calculation of an interaction matrix M ij on a representative protein structure derived from an MD trajectory [15][16][17][18][19]. The matrix contains the interaction energies between residue pairs, comprising all the nonbonded interresidue atomic energy components (namely, van der Waals and electrostatic couplings between all atoms of two residues).…”

“…This approach is based on a recently introduced Energy Decomposition Method, aimed at identifying the key residues (interaction hot spots) for the stabilization and folding of the protein to a defined 3D structure [15][16][17][18][19]. Previously, we showed the ability of this method to capture the essential changes occurring in the energetics of a protein upon single amino acid mutation [15][16][17][18][19].…”

“…Previously, we showed the ability of this method to capture the essential changes occurring in the energetics of a protein upon single amino acid mutation [15][16][17][18][19]. These changes are mainly related to stability variations in an ensemble of single-point mutants of a certain protein [19].…”

“…In this method, for a protein of N residues, the matrix of average non-bonded interactions between pairs of residues is built from an MD trajectory. The energy map is then simplified through eigenvalue decomposition (Principal Component Analysis) [15][16][17][18][19].…”

“…In particular, pair interactions between two peak-related residues provide a significant stabilizing energy to the protein. As a natural consequence, the SE reports therefore also on the residue-residue couplings defining the network of interactions that contribute to the folding core of the protein, whose participating residues are indicated by the SE regions above a defined threshold [20].The SE of the YAP65 WW domain ( Figure 2) indicates as energetically relevant residues a subset comprising the segments E8-S13 and Q17-D25 (numbering as in Socolich et al [11]). Relevant residues correspond to positions in the principal eigenvector characterized by a high value of the The results showed generally high values, in the range between 0.7 and 1 for the Pearson's correlation (see Table 1), and an average value of 0.80, indicating that the specific pattern of interactions defined by the Sequence Eigenvector (SE) profile may actually be considered important in the discovery of the energetic determinants of the folding features of the protein.…”