Protein folding: simple models for a complex process

Twenty-eight years after its original publication, the diffusion-collision model has successfully been applied to describe the folding kinetics of two proteins with the same native structure but different sequences (Islam et al., this issue of Structure). The calculations show the relative importance of the primary and tertiary structure on the sequence of events and folding. For both proteins, the model suggests parallel folding pathways, a finding which has wide implications for the interpretations of experiments.     

Complex folding pathways in a simple beta-hairpin

The determination of the folding mechanisms of proteins is critical to understand the topological change that can propagate Alzheimer and Creutzfeld-Jakobs diseases, among others. The computational community has paid considerable attention to this problem; however, the associated time scale, typically on the order of milliseconds or more, represents a formidable challenge. Ab initio protein folding from long molecular dynamics simulations or ensemble dynamics is not feasible with ordinary computing facilities and new techniques must be introduced. Here we present a detailed study of the folding of a 16-residue beta-hairpin, described by a generic energy model and using the activation-relaxation technique. From a total of 90 trajectories at 300 K, three folding pathways emerge. All involve a simultaneous optimization of the complete hydrophobic and hydrogen bonding interactions. The first two pathways follow closely those observed by previous theoretical studies (folding starting at the turn or by interactions between the termini). The third pathway, never observed by previous all-atom folding, unfolding, and equilibrium simulations, can be described as a reptation move of one strand of the beta-sheet with respect to the other. This reptation move indicates that non-native interactions can play a dominant role in the folding of secondary structures. Furthermore, such a mechanism mediated by non-native hydrogen bonds is not available for study by unfolding and Gō model simulations. The exact folding path followed by a given beta-hairpin is likely to be influenced by its sequence and the solvent conditions. Taken together, these results point to a more complex folding picture than expected for a simple beta-hairpin.     

Malleability of the folding mechanism of the outer membrane protein PagP: parallel pathways and the effect of membrane elasticity

Understanding the interactions between membrane proteins and the lipid bilayer is key to increasing our ability to predict and tailor the folding mechanism, structure and stability of membrane proteins. Here, we have investigated the effects of changing the membrane composition and the relative concentrations of protein and lipid on the folding mechanism of the bacterial outer membrane protein PagP. The folding pathway, monitored by tryptophan fluorescence, was found to be characterized by a burst phase, representing PagP adsorption to the liposome surface, followed by a time course that reflects the folding and insertion of the protein into the membrane. In 1,2-dilauroyl-sn-glycero-3-phosphocholine (diC(12:0)PC) liposomes, the post-adsorption time course fits well to a single exponential at high lipid-to-protein ratios (LPRs), but at low LPRs, a second exponential phase with a slower folding rate constant is observed. Interrupted refolding assays demonstrated that the two exponential phases reflect the presence of parallel folding pathways. Partitioning between these pathways was found to be modulated by the elastic properties of the membrane. Folding into mixed 1,2-dilauroyl-sn-glycero-3-phosphoethanolamine:diC(12:0)PC liposomes resulted in a decrease in PagP adsorption to the liposomes and a switch to the slower folding pathway. By contrast, inclusion of 1,2-dilauroyl-sn-glycero-3-phosphoserine into diC(12:0)PC liposomes resulted in a decrease in the folding rate of the fast pathway. The results highlight the effect of lipid composition in tailoring the folding mechanism of a membrane protein, revealing that membrane proteins have access to multiple, competing folding routes to a unique native structure.     

Hierarchical protein folding pathways: a computational study of protein fragments

We have previously presented a building block folding model. The model postulates that protein folding is a hierarchical top-down process. The basic unit from which a fold is constructed, referred to as a hydrophobic folding unit, is the outcome of combinatorial assembly of a set of "building blocks." Results obtained by the computational cutting procedure yield fragments that are in agreement with those obtained experimentally by limited proteolysis. Here we show that as expected, proteins from the same family give very similar building blocks. However, different proteins can also give building blocks that are similar in structure. In such cases the building blocks differ in sequence, stability, contacts with other building blocks, and in their 3D locations in the protein structure. This result, which we have repeatedly observed in many cases, leads us to conclude that while a building block is influenced by its environment, nevertheless, it can be viewed as a stand-alone unit. For small-sized building blocks existing in multiple conformations, interactions with sister building blocks in the protein will increase the population time of the native conformer. With this conclusion in hand, it is possible to develop an algorithm that predicts the building block assignment of a protein sequence whose structure is unknown. Toward this goal, we have created sequentially nonredundant databases of building block sequences. A protein sequence can be aligned against these, in order to be matched to a set of potential building blocks.     

Predicting protein folding pathways

A structured folding pathway, which is a time ordered sequence of folding events, plays an important role in the protein folding process and hence, in the conformational search. Pathway prediction, thus gives more insight into the folding process and is a valuable guiding tool to search the conformation space. In this paper, we propose a novel 'unfolding' approach to predict the folding pathway. We apply graph-based methods on a weighted secondary structure graph of a protein to predict the sequence of unfolding events. When viewed in reverse this yields the folding pathway. We demonstrate the success of our approach on several proteins whose pathway is partially known.     

Equilibrium folding and unfolding pathways for a model protein

The folding–unfolding process of reduced bovine pancreatic trypsin inhibitor was investigated with an idealized model employing approximate free energies. The protein is regarded to consist of only C^α and C^β atoms. The backbone dihedral angles are the only conformational variables and are permitted to take discrete values at every 10°. Intraresidue energies consist of two terms: an empirical part taken from the observed frequency distributions of (?,ψ) and an additional favorable energy assigned to the native conformation of each residue. Interresidue interactions are simplified by assuming that there is an attractive energy operative only between residue pairs in close contact in the native structure. A total of 230,000 molecular conformations, with no atomic overlaps, ranging from the native state to the denatured state, are randomly generated by changing the sampling bias. Each conformation is classified according to its conformational energy, F; a conformational entropy, S(F) is estimated for each value of F from the number of samples. The dependence of S(F) on energy reveals that the folding–unfolding transition for this idealized model is an “all-or-none” type; this is attributable to the specific long-range interactions. Interresidue contact probabilities, averaged over samples representing various stages of folding, serve to characterize folding intermediates. Most probable equilibrium pathways for the folding–unfolding transition are constructed by connecting conformationally similar intermediates. The specific details obtained for bovine pancreatic trypsin inhibitor are as follows: (1) Folding begins with the appearance of nativelike medium-range contacts at a β-turn and at the α-helix. (2) These grow to include the native pair of interacting β-strands. This state includes intact regular secondary conformations, as well as the interstrand sheet contacts, and corresponds to an activated state with the highest free energy on the pathway. (3) Additional native long-range contacts are completely formed either toward the amino terminus or toward the carboxyl terminus. (4) In a final step, the missing contacts appear. Although these folding pathways for this model are not consistent with experimental reports, it does indicate multiple folding pathways. The method is general and can be applied to any set of calculated conformational energies and furthermore permits investigation of gross folding features.     

Optimum folding pathways for growing protein chains

The folding of a protein is studied as it grows residue by residue from the N-terminus and enters an environment that stabilizes the folded state. This mode of folding of a growing chain is different from refolding where the full chain folds from a disordered initial configuration to the native state. We propose a sequential dynamic optimization method that computes the evolution of optimum folding pathways as amino acid residues are added to the peptide chain one by one. The dynamic optimization formulation is deterministic and uses Newton's equations of motion and a Go-type potential that establishes the native contacts and excluded volume effects. The method predicts the optimal energy-minimizing path among all the alternative feasible pathways. As two examples, the folding of the chicken villin headpiece, a 36-residue protein, and chymotrypsin inhibitor 2 (CI2), a 64-residue protein, are studied. Results on the villin headpiece show significant differences from the refolding of the same chain studied previously. Results on CI2 mostly agree with the results of refolding experiments and computational work.     

Morphogenesis of a protein: folding pathways and the energy landscape

Current knowledge on the reaction whereby a protein acquires its native three-dimensional structure was obtained by and large through characterization of the folding mechanism of simple systems. Given the multiplicity of amino acid sequences and unique folds, it is not so easy, however, to draw general rules by comparing folding pathways of different proteins. In fact, quantitative comparison may be jeopardized not only because of the vast repertoire of sequences but also in view of a multiplicity of structures of the native and denatured states. We have tackled the problem of the relationships between the sequence information and the folding pathway of a protein, using a combination of kinetics, protein engineering and computational methods, applied to relatively simple systems. Our strategy has been to investigate the folding mechanism determinants using two complementary approaches, i.e. (i) the study of members of the same family characterized by a common fold, but substantial differences in amino acid sequence, or (ii) heteromorphic pairs characterized by largely identical sequences but with different folds. We discuss some recent data on protein-folding mechanisms by presenting experiments on different members of the PDZ domain family and their circularly permuted variants. Characterization of the energetics and structures of intermediates and TSs (transition states), obtained by Φ-value analysis and restrained MD (molecular dynamics) simulations, provides a glimpse of the malleability of the dynamic states and of the role of the topology of the native states and of the denatured states in dictating folding and misfolding pathways.     

Prediction of protein folding pathways.

Recent 1H nuclear magnetic resonance (n.m.r.) hydrogen exchange experiments on five different proteins have delineated the secondary structures formed in trapped, partially folded intermediates. The early forming structural elements are identifiable through a technique described in this work to predict folding pathways. The method assumes that the sequential selection of structural fragments such as alpha-helices and beta-strands involved in the folding process is founded upon the maximal burial of solvent accessible surface from both the formation of internal structure and substructure association. The substructural elements were defined objectively by major changes in main-chain direction. The predicted folding pathways are in complete correspondence with the n.m.r. results in that the formed structural fragments found in the folding intermediates are those predicted earliest in the pathways. The technique was also applied to proteins of known tertiary structure and with fold similar to one of the five proteins examined by 1H n.m.r. The pathways for these structures also showed general consistency with the n.m.r. observations, suggesting conservation of a secondary structural framework or molten globule about which folding nucleates and proceeds.     

An analysis of protein folding pathways

We have developed a model of the protein folding process based on three primary assumptions: that burying of hydrophobic area is the dominant contribution to the relative free energy of a conformation, that a record of the folding process is largely preserved in the final structure, and that the denatured state is a random coil. Detailed folding pathways are identified for 19 protein structures. The picture of the folding process that emerges from this analysis is one of nucleation by regions of 8-16 residues. Nucleation sites then lead to larger structures by two mechanisms: propagation and diffusion/collision. A Monte Carlo simulation is used to follow the folding pathway when propagation is the dominant mechanism. Because detailed pathways are derived for each protein, the models are susceptible to experimental verification.     

Prediction of protein folding pathways: Bovine pancreatic trypsin inhibitor

A previous theory for the prediction of protein folding pathways [1] is applied to bovine pancreatic trypsin inhibitor and the resulting sequence of folding events shown to corroborate well with available experimental data.     

Constructing templates for protein structure prediction by simulation of protein folding pathways

How a one-dimensional protein sequence folds into a specific 3D structure remains a difficult challenge in structural biology. Many computational methods have been developed in an attempt to predict the tertiary structure of the protein; most of these employ approaches that are based on the accumulated knowledge of solved protein structures. Here we introduce a novel and fully automated approach for predicting the 3D structure of a protein that is based on the well accepted notion that protein folding is a hierarchical process. Our algorithm follows the hierarchical model by employing two stages: the first aims to find a match between the sequences of short independently-folding structural entities and parts of the target sequence and assigns the respective structures. The second assembles these local structural parts into a complete 3D structure, allowing for long-range interactions between them. We present the results of applying our method to a subset of the targets from CASP6 and CASP7. Our results indicate that for targets with a significant sequence similarity to known structures we are often able to provide predictions that are better than those achieved by two leading servers, and that the most significant improvements in comparison with these methods occur in regions of a gapped structural alignment between the native structure and the closest available structural template. We conclude that in addition to performing well for targets with known homologous structures, our method shows great promise for addressing the more general category of comparative modeling targets, which is our next goal.     

Prediction of protein structure by simulating coarse-grained folding pathways: a preliminary report

A set of software tools designed to study protein structure and kinetics has been developed. The core of these tools is a program called Folding Machine (FM) which is able to generate low resolution folding pathways using modest computational resources. The FM is based on a coarse-grained kinetic ab initio Monte-Carlo sampler that can optionally use information extracted from secondary structure prediction servers or from fragment libraries of local structure. The model underpinning this algorithm contains two novel elements: (a) the conformational space is discretized using the Ramachandran basins defined in the local phi-psi energy maps; and (b) the solvent is treated implicitly by rescaling the pairwise terms of the non-bonded energy function according to the local solvent environments. The purpose of this hybrid ab initio/knowledge-based approach is threefold: to cover the long time scales of folding, to generate useful 3-dimensional models of protein structures, and to gain insight on the protein folding kinetics. Even though the algorithm is not yet fully developed, it has been used in a recent blind test of protein structure prediction (CASP5). The FM generated models within 6 A backbone rmsd for fragments of about 60-70 residues of alpha-helical proteins. For a CASP5 target that turned out to be natively unfolded, the trajectory obtained for this sequence uniquely failed to converge. Also, a new measure to evaluate structure predictions is presented and used along the standard CASP assessment methods. Finally, recent improvements in the prediction of beta-sheet structures are briefly described.     

A unified mechanism for protein folding: predetermined pathways with optional errors

There is a fundamental conflict between two different views of how proteins fold. Kinetic experiments and theoretical calculations are often interpreted in terms of different population fractions folding through different intermediates in independent unrelated pathways (IUP model). However, detailed structural information indicates that all of the protein population folds through a sequence of intermediates predetermined by the foldon substructure of the target protein and a sequential stabilization principle. These contrary views can be resolved by a predetermined pathway--optional error (PPOE) hypothesis. The hypothesis is that any pathway intermediate can incorporate a chance misfolding error that blocks folding and must be reversed for productive folding to continue. Different fractions of the protein population will then block at different steps, populate different intermediates, and fold at different rates, giving the appearance of multiple unrelated pathways. A test of the hypothesis matches the two models against extensive kinetic folding results for hen lysozyme which have been widely cited in support of independent parallel pathways. The PPOE model succeeds with fewer fitting constants. The fitted PPOE reaction scheme leads to known folding behavior, whereas the IUP properties are contradicted by experiment. The appearance of a conflict with multipath theoretical models seems to be due to their different focus, namely on multitrack microscopic behavior versus cooperative macroscopic behavior. The integration of three well-documented principles in the PPOE model (cooperative foldons, sequential stabilization, optional errors) provides a unifying explanation for how proteins fold and why they fold in that way.     

Effective protein folding in simple random search

Here I show the following facts using a simple random search model without including any sophisticated energy term. As the size of elements exponentially affects the efficiency of folding, it can be remarkably enhanced by dividing the elements into small blocks. As the folding of the blocks is completely independent, the total folding time can be reduced to the folding time of the single hardest block. This result gives the simplest and most straightforward answer to the Levinthal paradox.     

Importance of sequence specificity for predicting protein folding pathways: perturbed Gaussian chain model

Recent experimental and theoretical studies suggest that rates and pathways of protein folding are largely decided by topology of the native structures, at least for small proteins. However, some exceptions are known; for example, protein L and protein G have the same topology, but exhibit different characteristics of the TSE. Thus, folding pathways of some proteins are critically affected by detailed information on amino acid sequences. To investigate the sequence specificity, we calculate folding pathways of 20 small proteins using the perturbed Gaussian chain model developed by Portman et al. (Phys Rev Lett 1998;81:5237-5240; J Chem Phys 2001;114:5069-5081). Characteristics of the TSE predicted by the model are in good agreement with experimental phi-value data for many proteins at coarse-grained level. Especially, estimation of folding TSE for protein G and protein L based on both topology and additional sequence information are consistent with experimental phi-value data. With only topology information, however, the model predicts the TSE of protein G incorrectly. Moreover, the model that uses topology and sequence information describes free energy profiles of two-state and three-state folders consistently with experiment, whereas the topology only model predicts free energy profiles of some proteins incorrectly. This indicates that sequence specificity also has critical roles in determining the folding pathways for some proteins.     

Conformational propagation with prion-like characteristics in a simple model of protein folding

Protein refolding/misfolding to an alternative form plays an aetiologic role in many diseases in humans, including Alzheimer's disease, the systemic amyloidoses, and the prion diseases. Here we have discovered that such refolding can occur readily for a simple lattice model of proteins in a propagatable manner without designing for any particular alternative native state. The model uses a simple contact energy function for interactions between residues and does not consider the peculiarities of polypeptide geometry. In this model, under conditions where the normal (N) native state is marginally stable or unstable, two chains refold from the N native state to an alternative multimeric energetic minimum comprising a single refolded conformation that can then propagate itself to other protein chains. The only requirement for efficient propagation is that a two-faced mode of packing must be in the ground state as a dimer (a higher-energy state for this packing leads to less efficient propagation). For random sequences, these ground-state dimeric configurations tend to have more beta-sheet-like extended structure than almost any other sort of dimeric ground-state assembly. This implies that propagating states (such as for prions) are beta-sheet rich because the only likely propagating forms are beta-sheet rich. We examine the details of our simulations to see to what extent the observed properties of prion propagation can be predicted by a simple protein folding model. The formation of the alternative state in the present model shows several distinct features of amyloidogenesis and of prion propagation. For example, an analog of the phenomenon of conformationally distinct strains in prions is observed. We find a parallel between 'glassy' behavior in liquids and the formation of a propagatable state in proteins. This is the first report of simulation of conformational propagation using any heteropolymer model. The results imply that some (but not most) small protein sequences must maintain a sequence signal that resists refolding to propagatable alternative native states and that the ability to form such states is not limited to polypeptides (or reliant on regular hydrogen bonding per se) but can occur for other protein-like heteropolymers.     

Pathway of oxidative folding of alpha-lactalbumin: a model for illustrating the diversity of disulfide folding pathways

The pathway of oxidative folding of alpha-lactalbumin (alpha LA) (four disulfide bonds) has been characterized by structural and kinetic analysis of the acid-trapped folding intermediates. In the absence of calcium, oxidative folding of alpha LA proceeds through highly heterogeneous species of one-, two-, three-, and four-disulfide (scrambled) intermediates to reach the native structure. In the presence of calcium, the folding intermediates of alpha LA comprise two predominant isomers (alpha LA-IIA and alpha LA-IIIA) adopting exclusively native disulfide bonds, including the two disulfide bonds (Cys(61)-Cys(77) and Cys(73)-Cys(91)) located within the beta-sheet calcium binding domain. alpha LA-IIA is a two-disulfide species consisting of Cys(61)-Cys(77) and Cys(73)-Cys(91) disulfide bonds. alpha LA-IIIA contains Cys(61)-Cys(77), Cys(73)-Cys(91), and Cys(28)-Cys(111) disulfide bonds. The underlying mechanism of the contrasting folding pathways of calcium-bound and calcium-depleted alpha LA is congruent with the cause of diversity of disulfide folding pathways observed among many well-characterized three-disulfide proteins, including bovine pancreatic trypsin inhibitor and hirudin. Our study also reveals novel aspects of the folding mechanism of alpha LA that have not been described previously.