Kinetics of Folding of Proteins and RNA

The assembly of biological molecules, most notably globular proteins1 and RNA,2,3 into unique threedimensional structures with well-defined topology is a complex and fascinating phenomenon in molecular biology. There are two aspects to the problem of folding of proteins and RNA. The first is the prediction of the three-dimensional structure of the folded state from the one-dimensional primary sequence of amino acids (for proteins) and nucleotides (for RNA). The second question is concerned with the kinetics of approach to the essentially unique folded state (also referred to as the native state) starting from an initial ensemble of disordered structures. In this Account we describe recent advances in our understanding of the kinetics of in vitro folding of globular proteins in terms of the underlying energy landscape. We further show that similar considerations can be usefully applied to describe the general features of the folding of RNA molecules. The pioneering experiments of Anfinsen4,5 and subsequent studies established that in many cases protein folding is a self-assembly process; i.e., the information needed for obtaining the three-dimensional structure is encoded in the primary sequence. These experiments did not provide the mechanisms of folding to the native conformation. The intellectual impetus to understand the kinetic mechanisms of protein folding came from Levinthal6 who wondered how a protein molecule searches the astronomically large number of conformations to reach the native state on a biologically relevant time scale. It has been proposed recently, through statistical mechanical studies of several classes of minimal models, that the key to resolving the Levinthal paradox lies in elucidating the ways in which proteins explore the energy landscape.7-11 In the minimal models only those aspects of a polypeptide chain which are thought to be crucial for describing the folding process are retained. These include chain connectivity, approximate representation of hydrophobic interactions, and self-avoidance between the various residues. Theoretical studies using the minimal models have shown that the free energy surface of typical proteins is rugged; i.e., there are many minima besides the one corresponding to the native state, which are separated by free energy barriers of varying heights. An examination of the dynamics in such a complex energy hypersurface leads to general kinetic scenarios for protein folding which are just beginning to be confirmed experimentally. More recently, interest in the problem of RNA folding has been renewed by the discovery of catalytic RNA.12,13 Pioneering work on the structure of transfer RNA first established that RNAs could form complex structures.14-16 The expectation that catalytic RNAs should also fold into well-defined, compact structures has been borne out by a battery of biochemical, spectroscopic, and crystallographic experiments designed to probe RNA structure.3,17,18 From the energy landscape perspective it is natural to suggest that the considerations that lead to the theoretical developments of protein folding should also apply to RNA folding. In general terms, the requirements for RNA folding are analogous to those of protein folding. As is true for polypeptides, the number of conformations in the fully denatured state (the Levinthal limit) is large. For RNA sequences, the kinetic problem consists of forming the correct secondary structure, that is, Watson-Crick base pairs between complementary sequences, and achieving the correct three-dimensional organization of the structural elements. In this context the challenge is to understand how the interplay of interactions among polynucleotides establishes an energy surface such that the native state is activated in a biologically meaningful time.