Robustness of atomistic Go models in predicting native-like folding intermediates.

Go models are exceedingly popular tools in computer simulations of protein folding. These models are native-centric, i.e., they are directly constructed from the protein's native structure. Therefore, it is important to understand up to which extent the atomistic details of the native structure dictate the folding behavior exhibited by Go models. Here we address this challenge by performing exhaustive discrete molecular dynamics simulations of a Go potential combined with a full atomistic protein representation. In particular, we investigate the robustness of this particular type of Go models in predicting the existence of intermediate states in protein folding. We focus on the N47G mutational form of the Spc-SH3 folding domain (x-ray structure) and compare its folding pathway with that of alternative native structures produced in silico. Our methodological strategy comprises equilibrium folding simulations, structural clustering, and principal component analysis.

Structural similarity enhances interaction propensity of proteins.

We study statistical properties of interacting protein-like surfaces and predict two strong, related effects: (i) statistically enhanced self-attraction of proteins; (ii) statistically enhanced attraction of proteins with similar structures. The effects originate in the fact that the probability to find a pattern self-match between two identical, even randomly organized interacting protein surfaces is always higher compared with the probability for a pattern match between two different, promiscuous protein surfaces. This theoretical finding explains statistical prevalence of homodimers in protein-protein interaction networks reported earlier. Further, our findings are confirmed by the analysis of curated database of protein complexes that showed highly statistically significant overrepresentation of dimers formed by structurally similar proteins with highly divergent sequences ("superfamily heterodimers"). We suggest that promiscuous homodimeric interactions pose strong competitive interactions for heterodimers evolved from homodimers. Such evolutionary bottleneck is overcome using the negative design evolutionary pressure applied against promiscuous homodimer formation. This is achieved through the formation of highly specific contacts formed by charged residues as demonstrated both in model and real superfamily heterodimers.

Identifying critical residues in protein folding: Insights from phi-value and P(fold) analysis.

We apply a simulational proxy of the phi-value analysis and perform extensive mutagenesis experiments to identify the nucleating residues in the folding "reactions" of two small lattice Go polymers with different native geometries. Our findings show that for the more complex native fold (i.e., the one that is rich in nonlocal, long-range bonds), mutation of the residues that form the folding nucleus leads to a considerably larger increase in the folding time than the corresponding mutations in the geometry that is predominantly local. These results are compared to data obtained from an accurate analysis based on the reaction coordinate folding probability P(fold) and on structural clustering methods. Our study reveals a complex picture of the transition state ensemble. For both protein models, the transition state ensemble is rather heterogeneous and splits up into structurally different populations. For the more complex geometry the identified subpopulations are actually structurally disjoint. For the less complex native geometry we found a broad transition state with microscopic heterogeneity. These findings suggest that the existence of multiple transition state structures may be linked to the geometric complexity of the native fold. For both geometries, the identification of the folding nucleus via the P(fold) analysis agrees with the identification of the folding nucleus carried out with the phi-value analysis. For the most complex geometry, however, the applied methodologies give more consistent results than for the more local geometry. The study of the transition state structure reveals that the nucleus residues are not necessarily fully native in the transition state. Indeed, it is only for the more complex geometry that two of the five critical residues show a considerably high probability of having all its native bonds formed in the transition state. Therefore, one concludes that, in general, the phi-value correlates with the acceleration/deceleration of folding induced by mutation, rather than with the degree of nativeness of the transition state, and that the "traditional" interpretation of phi-values may provide a more realistic picture of the structure of the transition state only for more complex native geometries.

Statistically enhanced promiscuity of structurally correlated patterns.

We predict that patterns with correlated surface density of atoms have statistically higher promiscuity (ability to bind stronger to an arbitrary pattern) as compared with noncorrelated patterns with the same average surface density. We suggest that this constitutes a generic design principle for highly connected proteins (hubs) in protein interaction networks. We develop an analytical theory for this effect. We show that our key predictions are generic and independent, qualitatively, on the specific form of the interatomic interaction potential, provided it has a finite range.

Theoretical studies of protein-folding thermodynamics and kinetics.

Recently, protein-folding models have advanced to the point where folding simulations of protein-like chains of reasonable length (up to 125 amino acids) are feasible, and the major physical features of folding proteins, such as cooperativity in thermodynamics and nucleation mechanisms in kinetics, can be reproduced. This has allowed deep insight into the physical mechanism of folding, including the solution of the so-called 'Levinthal paradox'.

Different circular permutations produced different folding nuclei in proteins: a computational study.

There have been many studies about the effect of circular permutation on the transition state/folding nucleus of proteins, with sometimes conflicting conclusions from different proteins and permutations. To clarify this important issue, we have studied two circular permutations of a lattice protein model with side-chains. Both permuted sequences have essentially the same native state as the original (wild-type) sequence. Circular permutant 1 cuts at the folding nucleus of the wild-type sequence. As a result, the permutant has a drastically different nucleus and folds more slowly than wild-type. In contrast, circular permutant 2 involves an incision at a site unstructured in the wild-type transition state, and the wild-type nucleus is largely retained in the permutant. In addition, permutant 2 displays both two-state and multi-state folding, with a native-like intermediate state occasionally populated. Neither the wild-type nor permutant 1 has a similar intermediate, and both fold in an apparently two-state manner. Surprisingly, permutant 2 folds at a rate identical with that of the wild-type. The intermediate in permutant 2 is stabilised by native and non-native interactions, and cannot be classified simply as on or off-pathway. So we advise caution in attributing experimental data to on or off-pathway intermediates. Finally, our work illuminates the results on alpha-spectrin SH3, chymotrypsin inhibitor 2 and beta-lactoglobulin, and supports a key assumption in the experimental efforts to locate potential nucleation sites of real proteins via circular permutations.

Understanding hierarchical protein evolution from first principles.

We propose a model that explains the hierarchical organization of proteins in fold families. The model, which is based on the evolutionary selection of proteins by their native state stability, reproduces patterns of amino acids conserved across protein families. Due to its dynamic nature, the model sheds light on the evolutionary time-scales. By studying the relaxation of the correlation function between consecutive mutations at a given position in proteins, we observe separation of the evolutionary time-scales: at short time intervals families of proteins with similar sequences and structures are formed, while at long time intervals the families of structurally similar proteins that have low sequence similarity are formed. We discuss the evolutionary implications of our model. We provide a "profile" solution to our model and find agreement between predicted patterns of conserved amino acids and those actually observed in nature.

Characterization of the folding kinetics of a three-helix bundle protein via a minimalist Langevin model.

We use a simple off-lattice Langevin model of protein folding to characterize the folding and unfolding of a fast-folding, 46 residue three-helix bundle. Under conditions at which the C-terminal helix is 30 % stable, we observe a clear three-state folding mechanism. In the on-pathway intermediate state, the middle and C-terminal helices are folded and in contact with each other, while the N-terminal region remains disordered. Nevertheless, under these conditions this intermediate is thermodynamically unstable relative to its unfolded state. The first and highest folding barrier corresponds to the organization of the hinge between the middle and C-terminal helices. A subsequent major barrier corresponds to the organization of the hinge between the middle and N-terminal helices. Hyperstabilizing the hinge regions leads to twice the folding rate that is obtained from hyperstabilizing the helices, even though much fewer contacts are involved in hinge hyperstabilization than in helix hyperstabilization. Unfolding follows single-exponential kinetics, even at temperatures only slightly above the folding transition temperature.

How to derive a protein folding potential? A new approach to an old problem.

In this paper we introduce a novel method of deriving a pairwise potential for protein folding. The potential is obtained by an optimization procedure that simultaneously maximizes thermodynamic stability for all proteins in the database. When applied to the representative dataset of proteins and with the energy function taken in pairwise contact approximation, our potential scored somewhat better than existing ones. However, the discrimination of the native structure from decoys is still not strong enough to make the potential useful for ab initio folding. Our results suggest that the problem lies with pairwise amino acid contact approximation and/or simplified presentation of proteins rather than with the derivation of potential. We argue that more detail of protein structure and energetics should be taken into account to achieve energy gaps. The suggested method is general enough to allow us to systematically derive parameters for more sophisticated energy functions. The internal control of validity for the potential derived by our method is convergence to a unique solution upon addition of new proteins to the database. The method is tested on simple model systems where sequences are designed, using the preset "true" potential, to have low energy in a dataset of structures. Our procedure is able to recover the potential with correlation r approximately 91% with the true one and we were able to fold all model structures using the recovered potential. Other statistical knowledge-based approaches were tested using this model and the results indicate that they also can recover the true potential with high degree of accuracy.

What can disulfide bonds tell us about protein energetics, function and folding: simulations and bioninformatics analysis.

We study the impact of disulfide bonds on protein stability and folding. Using lattice model simulations, we show that formation of a disulfide bond stabilizes a protein to an extent that depends on the distance along the chain between linked cysteine residues. However, the impact of disulfide bonds on folding kinetics varies broadly, from acceleration when disulfides are introduced in or close to the folding nucleus, to slowing when disulfides are introduced outside the nucleus. Having established the effect of disulfide bonds on stability, we study the correlation between the number of disulfide bonds and the composition of certain amino acid classes with the goal to use it as a statistical probe into factors that contribute to stability of proteins. We find that the number of disulfides is negatively correlated with aliphatic hydrophobic but not aromatic content. It is surprising that we observe a strong correlation of disulfide content with polar (Q,S,T,N) amino acid content and a strong negative correlation with charged (E,D,K,R) content. These findings provide insights into factors that determine protein stability and principles of protein design as well as possible relations of disulfide bonds and protein function.

Protein structure prediction by threading. Why it works and why it does not.

We developed a novel Monte Carlo threading algorithm which allows gaps and insertions both in the template structure and threaded sequence. The algorithm is able to find the optimal sequence-structure alignment and sample suboptimal alignments. Using our algorithm we performed sequence-structure alignments for a number of examples for three protein folds (ubiquitin, immunoglobulin and globin) using both "ideal" set of potentials (optimized to provide the best Z-score for a given protein) and more realistic knowledge-based potentials. Two physically different scenarios emerged. If a template structure is similar to the native one (within 2 A RMS), then (i) the optimal threading alignment is correct and robust with respect to deviations of the potential from the "ideal" one; (ii) suboptimal alignments are very similar to the optimal one; (iii) as Monte Carlo temperature decreases a sharp cooperative transition to the optimal alignment is observed. In contrast, if the template structure is only moderately close to the native structure (RMS greater than 3.5 A), then (i) the optimal alignment changes dramatically when an "ideal" potential is substituted by the real one; (ii) the structures of suboptimal alignments are very different from the optimal one, reducing the reliability of the alignment; (iii) the transition to the apparently optimal alignment is non-cooperative. In the intermediate cases when the RMS between the template and the native conformations is in the range between 2 A and 3.5 A, the success of threading alignment may depend on the quality of potentials used. These results are rationalized in terms of a threading free energy landscape. Possible ways to overcome the fundamental limitations of threading are discussed briefly.

Universally conserved positions in protein folds: reading evolutionary signals about stability, folding kinetics and function.

Here, we provide an analysis of molecular evolution of five of the most populated protein folds: immunoglobulin fold, oligonucleotide-binding fold, Rossman fold, alpha/beta plait, and TIM barrels. In order to distinguish between "historic", functional and structural reasons for amino acid conservations, we consider proteins that acquire the same fold and have no evident sequence homology. For each fold we identify positions that are conserved within each individual family and coincide when non-homologous proteins are structurally superimposed. As a baseline for statistical assessment we use the conservatism expected based on the solvent accessibility. The analysis is based on a new concept of "conservatism-of-conservatism". This approach allows us to identify the structural features that are stabilized in all proteins having a given fold, despite the fact that actual interactions that provide such stabilization may vary from protein to protein. Comparison with experimental data on thermodynamics, folding kinetics and function of the proteins reveals that such universally conserved clusters correspond to either: (i) super-sites (common location of active site in proteins having common tertiary structures but not function) or (ii) folding nuclei whose stability is an important determinant of folding rate, or both (in the case of Rossman fold). The analysis also helps to clarify the relation between folding and function that is apparent for some folds.

Excluded volume in protein side-chain packing.

The excluded volume occupied by protein side-chains and the requirement of high packing density in the protein interior should severely limit the number of side-chain conformations compatible with a given native backbone. To examine the relationship between side-chain geometry and side-chain packing, we use an all-atom Monte Carlo simulation to sample the large space of side-chain conformations. We study three models of excluded volume and use umbrella sampling to effectively explore the entire space. We find that while excluded volume constraints reduce the size of conformational space by many orders of magnitude, the number of allowed conformations is still large. An average repacked conformation has 20 % of its chi angles in a non-native state, a marked reduction from the expected 67 % in the absence of excluded volume. Interestingly, well-packed conformations with up to 50 % non-native chi angles exist. The repacked conformations have native packing density as measured by a standard Voronoi procedure. Entropy is distributed non-uniformly over positions, and we partially explain the observed distribution using rotamer probabilities derived from the Protein Data Bank database. In several cases, native rotamers that occur infrequently in the database are seen with high probability in our simulation, indicating that sequence-specific excluded volume interactions can stabilize rotamers that are rare for a given backbone. In spite of our finding that 65 % of the native rotamers and 85 % of chi(1) angles can be predicted correctly on the basis of excluded volume only, 95 % of positions can accommodate more than one rotamer in simulation. We estimate that, in order to quench the side-chain entropy observed in the presence of excluded volume interactions, other interactions (hydrophobic, polar, electrostatic) must provide an additional stabilization of at least 0.6 kT per residue in order to single out the native state.

The folding thermodynamics and kinetics of crambin using an all-atom Monte Carlo simulation.

We present a novel Monte Carlo simulation of protein folding, in which all heavy atoms are represented as interacting hard spheres. This model includes all degrees of freedom relevant to folding, all side-chain and backbone torsions, and uses a Go potential. In this study, we focus on the 46 residue alpha/beta protein crambin and two of its structural components, the helix and helix hairpin. For a wide range of temperatures, we recorded multiple folding events of these three structures from random coils to native conformations that differ by less than 1 A C(alpha) dRMS from their crystal structure coordinates. The thermodynamics and kinetic mechanism of the helix-coil transition obtained from our simulation shows excellent agreement with currently available experimental and molecular dynamics data. Based on insights obtained from folding its smaller structural components, a possible folding mechanism for crambin is proposed. We observed that the folding occurs via a cooperative, first order-like process, and that many folding pathways to the native state exist. One particular sequence of events constitutes a "fast-folding" pathway where kinetic traps are avoided. At very low temperatures, a kinetic trap arising from the incorrect packing of side-chains was observed. These results demonstrate that folding to the native state can be observed in a reasonable amount of time on desktop computers even when an all-atom representation is used, provided the energetics sufficiently stabilize the native state.

Impact of local and non-local interactions on thermodynamics and kinetics of protein folding.

To address the question of how the geometry of a protein's native conformation affects its folding and stability, we studied three model 36-mers on a cubic lattice. The native structure of one of these model 36-mers consisted mostly of local contacts, while that of a second consisted mostly of non-local contacts. The third native structure had a typical compact native conformation, and served as our reference. For each protein, the amino acid sequence was designed to have a pronounced energy minimum at its native conformation. We observed dramatic differences in folding, dependent on the presence or absence of non-local contacts. For the proteins with a typical large number of non-local contacts, the folding transition was all-or-none, whereas for the one with mostly local contacts, it was not. Although the maximum rate of folding was similar for all three proteins, we found that under conditions at which each native conformation was stable, the structure with mostly non-local contacts folded two orders of magnitude faster than the one with mostly local contacts. The statistical analysis of protein structure agrees fully with the implications of the theory. We discuss the importance of cooperativity in protein folding for its stability.

Differential stabilization of two hydrophobic cores in the transition state of the villin 14T folding reaction.

We report the distribution of hydrophobic core contacts during the folding reaction transition state for villin 14T, a small 126-residue protein domain. The solution structure of villin 14T contains a central beta-sheet with two flanking hydrophobic cores; transition states for this protein topology have not been previously studied. Villin 14T has no disulfide bonds or cis-proline residues in its native state; it folds reversibly, and in an apparently two-state manner under some conditions. To map the hydrophobic core contacts in the transition state, 27 point mutations were generated at positions spread throughout the two hydrophobic cores. After each point mutation, comparison of the change in folding kinetics with the equilibrium destabilization indicates whether the site of mutation is stabilized in the transition state. The results show that the folding nucleus, or the sub-region with the strongest transition state contacts, is located in one of the two hydrophobic cores (the predominantly aliphatic core). The other hydrophobic core, which is mostly aromatic, makes much weaker contacts in the transition state. This work is the first transition state mapping for a protein with multiple major hydrophobic cores in a single folding unit; the hydrophobic cores cannot be separated into individual folding subdomains. The stabilization of only one hydrophobic core in the transition state illustrates that hydrophobic core formation is not intrinsically capable of nucleating folding, but must also involve the right specific interactions or topological factors in order to be kinetically important.

Identifying the protein folding nucleus using molecular dynamics.

Molecular dynamics simulations of folding in an off-lattice protein model reveal a nucleation scenario, in which a few well-defined contacts are formed with high probability in the transition state ensemble of conformations. Their appearance determines folding cooperativity and drives the model protein into its folded conformation. Amino acid residues participating in those contacts may serve as "accelerator pedals" used by molecular evolution to control protein folding rate.

Pseudodihedrals: simplified protein backbone representation with knowledge-based energy.

Pairwise contact energies do not explicitly take protein secondary structure into account, and so provide an incomplete description of conformational energy. In order to construct a Hamiltonian that specifically relates to protein backbone conformations, a simplified backbone angle is used. The pseudodihedral angle (the torsion angle between planes defined by 4 consecutive alpha-carbon atoms) provides a simplified backbone representation and continues to manifest information about secondary-structure elements: the pseudo-Ramachandran plot contains helical and sheetlike regions. The distribution of pseudodihedral angles is highly sensitive to the identity of the central pair of amino acids. Therefore, a sequence-dependent, knowledge-based potential energy was found according to a quasichemical approximation. These functions form complementary additions to the contact potentials currently in use. This pseudodihedral potential greatly enhances the ability to design sequences that are specific to a given conformation and also improves the ability to discriminate a native conformation from many other conformations.

Analysis of knowledge-based protein-ligand potentials using a self-consistent method.

We propose a self-consistent approach to analyze knowledge-based atom-atom potentials used to calculate protein-ligand binding energies. Ligands complexed to actual protein structures were first built using the SMoG growth procedure (DeWitte & Shakhnovich, 1996) with a chosen input potential. These model protein-ligand complexes were used to construct databases from which knowledge-based protein-ligand potentials were derived. We then tested several different modifications to such potentials and evaluated their performance on their ability to reconstruct the input potential using the statistical information available from a database composed of model complexes. Our data indicate that the most significant improvement resulted from properly accounting for the following key issues when estimating the reference state: (1) the presence of significant nonenergetic effects that influence the contact frequencies and (2) the presence of correlations in contact patterns due to chemical structure. The most successful procedure was applied to derive an atom-atom potential for real protein-ligand complexes. Despite the simplicity of the model (pairwise contact potential with a single interaction distance), the derived binding free energies showed a statistically significant correlation (approximately 0.65) with experimental binding scores for a diverse set of complexes.

Domains in folding of model proteins.

By means of Monte Carlo simulation, we investigated the equilibrium between folded and unfolded states of lattice model proteins. The amino acid sequences were designed to have pronounced energy minimum target conformations of different length and shape. For short fully compact (36-mer) proteins, the all-or-none transition from the unfolded state to the native state was observed. This was not always the case for longer proteins. Among 12 designed sequences with the native structure of a fully compact 48-mer, a simple all-or-none transition was observed in only three cases. For the other nine sequences, three states of behavior-the native, denatured, and intermediate states-were found. The contiguous part of the native structure (domain) was conserved in the intermediate state, whereas the remaining part was completely unfolded and structureless. These parts melted separately from each other.