Chemical physics of protein folding

In 2008, to the consternation of some, one of the editors of this special issue on the “Chemical Physics of Protein Folding,” was quoted as saying, “What was called the protein folding problem 20 years ago is solved” (1). One purpose of this special issue is to drive home this point. The other, more important purpose is to illustrate how workers on the protein folding problem, by moving beyond their early obsession with seeming paradoxes (2), are developing a quantitative understanding of how the simpler biological structures assemble both in vitro and in vivo. The emerging quantitative understanding reveals simultaneously the richness of folding phenomena and the elegant simplicity of the underlying principles of spontaneous biomolecular assembly. The appreciation of these contrasting aspects of the folding problem has come about through the cooperation of theorists and experimentalists, a theme common to all the contributions to this special issue. Although the basic ideas about the folding energy landscape have turned out to be quite simple, entering even into some undergraduate textbooks (3), exploring their consequences in real systems has required painstaking intellectual analysis, as well as detailed computer simulations and experiments that still stretch the bounds of what is feasible. The backgrounds of the contributors to this issue reflect the breadth of the folding field and range from computer science and theoretical physics to molecular biology and organic chemistry. A great deal of the progress in the field can thus be traced to a fairly successful effort to develop a common language and conceptual framework for describing folding.

The conceptual framework is provided by energy landscape theory, which describes the diversity of structural possibilities in statistical mechanical terms. The main paradoxes of folding are resolved by the consistency principle (4) or, more generally, by the principle of minimal frustration (5), which quantifies the dominance of interactions stabilizing the specific native structure over other interactions that would favor nonnative, topologically distinct traps. In other words, the energy landscape of evolved proteins appears to be funneled (6, 7).

The overall funneled nature of the folding landscape provides a first guess of how folding begins and continues: Proteins fold by assembling primarily native substructures, whereas they only transiently sample misfolds. This insight explains the success of protein engineering in providing detailed structural information on the transition state ensemble, the so-called “φ-value analysis” (8, 9). In some cases, changing solution conditions changes the position of the transition state along the reaction coordinate, such that φ-value studies can even tell us in what order native parts assemble along the dominant transition path.

The ensemble nature of the transition states was the first clue that mechanistic complexity still remains on funneled landscapes. Multiple choices of a precise folding mechanism are possible, and several experimental studies herein demonstrate the malleability of specific folding mechanisms for proteins of either related structure or related sequence; on a funneled landscape, many roads lead to Rome (10, 11). After a molecule embarks on a route, even on a funneled energy landscape, assembly may not be completely straightforward. Occasionally, a greedy attempt to make native contacts early on can lead to topological traps (12); in that case, some early native interactions must be undone to allow complete folding and one must backtrack (13). Also, evolution toward a funneled landscape cannot repeal the universal character of the physics of specific molecular interactions; thus, no real folding landscape is ever perfectly funneled. If the nonnative interactions are fairly weak, they just provide a source of “friction,” a topic quantified here in several papers (14, 15). If the nonnative interactions become stronger, as theory predicts, frustrated interactions can allow specific intermediates to form with substantial nonnative contacts along with some native structure. This phenomenon also receives attention here (16).

Many of the most intriguing questions of mechanism can only be addressed feebly at the level of ensemble-averaged experiments; thus, folding science has called forth some extremely challenging experiments in which molecules are studied individually one at a time (17–19). On the computational side, although effective theories of folding can often use simplified models that make interesting and surprising predictions (20, 21), many questions of detail can now be answered satisfactorily with fully atomistic simulations that challenge computational power to the limit (22–24).

As the title of this special issue suggests, most of the contributions focus on exploring the general principles of folding. These principles emerge as clearly in test tube studies as they do in the cell. However, the cell provides its own challenges to folding science. The effects of heterogeneous intracellular environments on folding landscapes (25) are beginning to be explored. Also, despite the inborn tendency of proteins to do the right thing (arising from the minimal frustration principle), alternate folding, misfolding, and aggregation for a few proteins do lead to pathology, a topic explored in this issue (26) with the superoxide dismutase system that is involved in amyotrophic lateral sclerosis.

The topics discussed in this issue are only a small part of the work in the folding field. Nevertheless, they make clear that protein folding is a vibrant, living, interdisciplinary part of the natural sciences. We hope this snapshot of the field will encourage others to bring new approaches to contribute to our understanding of biomolecular assembly and encourage the use of the ideas and strategies that have already proved successful in the study of assembling the simplest biomolecules to look at the full complexity of living systems at higher levels of complexity.