Protein structure prediction: Concepts and applications

Abstract

Protein Structure Prediction: Concepts and Applications Anna Tramontano, Wiley-VHC Verlag GmbH & Co. KGaA, Weinheim, 2006, 228 pp., ISBN 3-527-31167-X, $70.00 paperback. Duane Sears [e-mail]*, * Department of Molecular, Cellular and Developmental Biology, University of California, Santa Barbara, CA 93106. This book is one of a series from this publisher covering contemporary topics in bioinformatics, protein structure and modeling, and protein design. The book's stated aim is to provide students and experimental researchers with a guide to various methods used for structure predictions. The author focuses on straightforward examples to describe the assumptions, limitations, and expected accuracies of different applications while delving somewhat lightly into the underlying mathematics so as to not overwhelm the reader with the theoretical complexities of some methods. This makes for easier reading, but it also limits the reader's exposure to the “nuts and bolts” of most applications. The progression of topics is logical and suitably juxtaposed with discussions of key historical developments that have shaped this field. The book is filled with simple, attractive figures to help illustrate key concepts, but unfortunately, many figures (e.g. Figs. 4.4, 4.12, 5.2, 7.6, and 7.15) are oversimplified or are not explained well enough for the reader to easily discern their intended messages. For example, why was the specific combination of sequence insertions chosen for the Needleman and Wunsch alignment shown in Fig. 4.4b when a different combination of insertions would obviously have produced a better initial alignment? Some figures (e.g. Figs. 4.15 and 7.6) illustrate very complicated concepts and needed even lengthier explanations for clarification. Chapters 4–7 are the most informative as they present the reader with a reasonably detailed and cohesive overview of several commonly used structure prediction methods based on comparative modeling, fold-recognition, fragment alignment, and/or secondary structure. Considering that the science of structure prediction is still very much in its infancy, Chapter 2 is particularly welcome because it addresses problems associated with the objectivity, assessment, reliability, and accuracy of structure predictions in general. Here, one learns about the so-called CASP (Critical Assessment of Structure Prediction) experiments as conducted by a worldwide consortium of scientists who are aiming to standardize critical assessment techniques for protein structure predictions. Considering the whole range of issues that surround the problem of predicting a protein's structure, the scope of this book is somewhat limited because it offers little critical discussion of several important conceptual hurdles that still confront modelers in their efforts to predict three-dimensional (3-D)1 conformations accurately. The classical boundaries of this problem are clearly laid out for the reader in terms of Anfinsen's experiments, on the one hand, which demonstrated that amino acid sequence is the major factor determining a protein's 3-D conformation, and Leventhal's paradox, on the other hand, which strongly suggests that proteins most likely arrive at their 3-D conformations by kinetically determined pathways possibly leading to metastable states rather than global free energy minimum states, as previously assumed. These half-century-old ideas are still provocative because accurate protein structure predictions ultimately depend on how effectively these ideas are implemented theoretically. Although the author takes an even narrower view by flatly stating on p. 24 that “there is no evidence that anything but the amino acid sequence determines the native protein structure in vivo,” this viewpoint is still a matter of conjecture for many reasons. In contrast to Anfinsen's protein folding model, nascent polypeptides begin folding during synthesis on the ribosome, well before the polypeptide has reached its completed length. Cellular proteins called chaperones appear to prevent, minimize, and/or reverse nascent polypeptide misfolding, thus shepherding the folding processes in some vaguely understood way. If chaperones act as the catalysts for initiating or even guiding the folding pathways of nascent polypeptide chains toward their final (metastable?) 3-D conformations, they would be inextricably linked to the folding process, at least for many proteins. The “inconvenient truth” is that the role of chaperones in protein folding is not understood well enough to model their behavior, which may explain why they are barely mentioned in the book. In the same vein, the thermodynamic aspects of protein folding are also difficult to model theoretically. Since the insightful work of Walter Kauzmann and others more than 50 years ago, it is known that thermodynamic “forces” drive protein folding in that different protein groups partition into different regions of the folded structure, much in the way that an immiscible mixture of oil and water spontaneously partitions into aqueous and nonaqueous phases. This explains why soluble proteins tend to have “oily” centers with higher concentrations of “buried” nonpolar groups, as pointed out in the book. Beyond this, however, the book provides only a standard introduction to the underlying thermodynamics (albeit not until Chapter 8, which is devoted to membrane proteins) without offering much insight into the difficulties associated with theoretically predicting the outcome of such partitioning events. Unlike the specific physical forces between groups that maintain a protein's 3-D structure, once formed, the partitioning of polar and nonpolar groups in a folded protein structure cannot be predicted by simply knowing the final geometrical arrangements and interactions between specific groups or charges in the native protein. However, progress is being made with predictive thermodynamic models like the “molten globule” model of protein folding [1], which is based on observations suggesting that the 3-D conformations of some proteins arise by spontaneous coalescence of pre-formed “chunks” of secondary structure, but this topic is surprisingly absent from the book. Other potentially important observations that bear on the protein folding problem are only briefly mentioned, including the observations that normal cells contain large percentages of intrinsically disordered proteins [2] and the observations that many native proteins undergo slow, spontaneous “unfolding” events under physiological conditions, often leading to the formation of larger polymers or aggregates such as amyloid fibers [3]. Such observations suggest that the same amino acid sequence can adopt different conformations under physiological conditions that differ in fundamental ways. Again, the “inconvenient truth” about such phenomena is that they are not understood well enough to be factored into the algorithms used for predicting 3-D structures. Judging from the many examples of “successful” protein predictions presented in this book (e.g. Figs. 3.11, 4.24, 5.5, and 6.2), one is forced to conclude that only marginal progress has been made toward accurate protein structure predictions, notwithstanding the author's attempt to paint a rosier picture with the “Applications and Examples” presented in Chapter 9. Most of these examples illustrate just how far the field has yet to go considering that biologically active structures depend on interactions and distances scaling to less than an Angstrom, well below the resolution of these examples. Given that many human diseases result from protein instability and unfolding [4], the problem of understanding what makes folded proteins unstable seems equally important to the daunting task of predicting a protein's native conformation ab initio (the topic of Chapter 3), but this problem was not addressed, even though it would seem to be far more tract-able, theoretically. In any case, those who may be contemplating using this book as a teaching resource will appreciate how well it summarizes the current methodology even though supplemental resources will likely be required for more detailed explanations of these methods in terms of their real-life applications and mathematical foundations.