Loop-closure events during protein folding: rationalizing the shape of Phi-value distributions

In the past years, the folding kinetics of many small single-domain proteins has been characterized by mutational Phi-value analysis. In this article, a simple, essentially parameter-free model is introduced which derives folding routes from native structures by minimizing the entropic loop-closure cost during folding. The model predicts characteristic folding sequences of structural elements such as helices and beta-strand pairings. Based on few simple rules, the kinetic impact of these structural elements is estimated from the routes and compared to average experimental Phi-values for the helices and strands of 15 small, well-characterized proteins. The comparison leads on average to a correlation coefficient of 0.62 for all proteins with polarized Phi-value distributions, and 0.74 if distributions with negative average Phi-values are excluded. The diffuse Phi-value distributions of the remaining proteins are reproduced correctly. The model shows that Phi-value distributions, averaged over secondary structural elements, can often be traced back to entropic loop-closure events, but also indicates energetic preferences in the case of a few proteins governed by parallel folding processes.     

Transition states and the meaning of Phi-values in protein folding kinetics

What is the mechanism of two-state protein folding? The rate-limiting step is typically explored through a Phi-value, which is the mutation-induced change in the transition state free energy divided by the change in the equilibrium free energy of folding. Phi-values ranging from 0 to 1 have been interpreted as meaning the transition state is denatured-like (0), native-like (1) or in-between. But there is no classical interpretation for the experimental Phi-values that are negative or >1. Using a rigorous method to identity transition states via an exact lattice model, we find that nonclassical Phi-values can arise from parallel microscopic flow processes, such as those in funnel-shaped energy landscapes. Phi < 0 results when a mutation destabilizes a slow flow channel, causing a backflow into a faster flow channel. Phi > 1 implies the reverse: a backflow from a fast channel into a slow one. Using a 'landscape mapping' method, we find that Phi correlates with the acceleration/deceleration of folding induced by mutations, rather than with the degree of nativeness of the transition state.     

Folding circular permutants of IL-1β: route selection driven by functional frustration

Interleukin-1β (IL-1β) is the cytokine crucial to inflammatory and immune response. Two dominant routes are populated in the folding to native structure. These distinct routes are a result of the competition between early packing of the functional loops versus closure of the β-barrel to achieve efficient folding and have been observed both experimentally and computationally. Kinetic experiments on the WT protein established that the dominant route is characterized by early packing of geometrically frustrated functional loops. However, deletion of one of the functional loops, the β-bulge, switches the dominant route to an alternative, yet, as accessible, route, where the termini necessary for barrel closure form first. Here, we explore the effect of circular permutation of the WT sequence on the observed folding landscape with a combination of kinetic and thermodynamic experiments. Our experiments show that while the rate of formation of permutant protein is always slower than that observed for the WT sequence, the region of initial nucleation for all permutants is similar to that observed for the WT protein and occurs within a similar timescale. That is, even permutants with significant sequence rearrangement in which the functional-nucleus is placed at opposing ends of the polypeptide chain, fold by the dominant WT "functional loop-packing route", despite the entropic cost of having to fold the N- and C- termini early. Taken together, our results indicate that the early packing of the functional loops dominates the folding landscape in active proteins, and, despite the entropic penalty of coalescing the termini early, these proteins will populate an entropically unfavorable route in order to conserve function. More generally, circular permutation can elucidate the influence of local energetic stabilization of functional regions within a protein, where topological complexity creates a mismatch between energetics and topology in active proteins.     

The structure of the major transition state for folding of an FF domain from experiment and simulation

We have analysed the transition state of folding of the four-helix FF domain from HYPA/FBP11 by high-resolution experiment and simulation as part of a continuing effort to understand the principles of folding and the refinement of predictive methods. The major transition state for folding was subjected to a Phi-value analysis utilising 50 mutants. The transition state contained a nucleus for folding centred around the end of helix 1 (H1) and the beginning of helix 2 (H2). Secondary structure in this region was fully formed (PhiF=0.9-1) and tertiary interactions were well developed. Interactions in the distal part of the native structure were weak (PhiF=0-0.2). The hydrophobic core and other parts of the protein displayed intermediate Phi-values, suggesting that interactions coalesce as the end of H1 and beginning of H2 are in the process of being formed. The distribution of Phi-values resembled that of barnase, which folds via an intermediate, rather than that of CI2 which folds by a concerted nucleation-condensation mechanism. The overall picture of the transition state structure identified in molecular dynamics simulations is in qualitative agreement, with the turn connecting H1 and H2 being formed, a loosened core, and H4 partially unfolded and detached from the core. There are some differences in the details and interpretation of specific Phi-values.     

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.     

Using stochastic roadmap simulation to predict experimental quantities in protein folding kinetics: folding rates and phi-values.

This paper presents a new method for studying protein folding kinetics. It uses the recently introduced Stochastic Roadmap Simulation (SRS) method to estimate the transition state ensemble (TSE) and predict the rates and the Phi-values for protein folding. The new method was tested on 16 proteins, whose rates and Phi-values have been determined experimentally. Comparison with experimental data shows that our method estimates the TSE much more accurately than an existing method based on dynamic programming. This improvement leads to better folding-rate predictions. We also compute the mean first passage time of the unfolded states and show that the computed values correlate with experimentally determined folding rates. The results on Phi-value predictions are mixed, possibly due to the simple energy model used in the tests. This is the first time that results obtained from SRS have been compared against a substantial amount of experimental data. The results further validate the SRS method and indicate its potential as a general tool for studying protein folding kinetics.     

Localized nature of the transition-state structure in goat alpha-lactalbumin folding

To investigate whether the structure partially formed in the molten globule folding intermediate of goat alpha-lactalbumin is further organized in the transition state of folding, we constructed a number of mutant proteins and performed Phi-value analysis on them. For this purpose, we measured the equilibrium unfolding transitions and kinetic refolding and unfolding reactions of the mutants using equilibrium and stopped-flow kinetic circular dichroism techniques. The results show that the mutants with mutations located in the A-helix (V8A, L12A), the B-helix (V27A), the beta-domain (L52A, W60A), the C-helix (K93A, L96A), the C-D loop (Y103F), the D-helix (L105A, L110A), and the C-terminal 3(10)-helix (W118F), have low Phi-values, less than 0.2. On the other hand, D87N, which is located on the Ca(2+)-binding site, has a high Phi-value, 0.91, indicating that tight packing of the side-chain around Asp87 occurs in the transition state. One beta-domain mutant (I55V) and three C-helix mutants (I89V, V90A, and I95V) demonstrated intermediate Phi-values, between 0.4 and 0.7. These results indicate that the folding nucleus in the transition state of goat alpha-LA is not extensively distributed over the alpha-domain of the protein, but very localized in a region that contains the Ca(2+)-binding site and the interface between the C-helix and the beta-domain. This is apparently in contrast with the fact that the molten globule state of alpha-lactalbumin has a partially formed structure inside the alpha-domain. It is concluded that the specific docking of the alpha and beta-domains at a domain interface is necessary for this protein to organize its native structure from the molten globule intermediate.     

Different secondary structure elements as scaffolds for protein folding transition states of two homologous four-helix bundles

Comparison of the folding processes for homologue proteins can provide valuable information about details in the interactions leading to the formation of the folding transition state. Here the folding kinetics of 18 variants of yACBP and 3 variants of bACBP have been studied by Phi-value analysis. In combination with Phi-values from previous work, detailed insight into the transition states for folding of both yACBP and bACBP has been obtained. Of the 16 sequence positions that have been studied in both yACBP and bACBP, 5 (V12, I/L27, Y73, V77, and L80) have high Phi-values and appear to be important for the transition state formation in both homologues. Y31, A34, and A69 have high Phi-values only in yACBP, while F5, A9, and I74 have high Phi-values only in bACBP. Thus, additional interactions between helices A2 and A4 appear to be important for the transition state of yACBP, whereas additional interactions between helices A1 and A4 appear to be important for the transition state of bACBP. To examine whether these differences could be assigned to different packing of the residues in the native state, a solution structure of yACBP was determined by NMR. Small changes in the packing of the hydrophobic side-chains, which strengthen the interactions between helices A2 and A4, are observed in yACBP relative to bACBP. It is suggested that different structure elements serve as scaffolds for the folding of the 2 ACBP homologues.     

Complete change of the protein folding transition state upon circular permutation

Reversing the loop lengths of the small protein S6 by circular permutation has a dramatic effect on the transition state structure: it changes from globally diffuse to locally condensed. The phenomenon arises from a biased dispersion of the contact energies. Stability data derived from point mutations throughout the S6 structure show that interactions between residues that are far apart in sequence are stronger than those that are close. This entropy compensation drives all parts of the protein to fold simultaneously and produces the diffuse transition-state structure typical for two-state proteins. In the circular permutant, where strong contacts and short sequence separations are engineered to concur, the transition state becomes atypically condensed and polarized. Taken together with earlier findings that S6 may also fold by a 'collapsed' trajectory with an intermediate, the results suggest that this protein may fold by a multiplicity of mechanisms. The observations indicate that the diffuse transition state of S6 is not required for folding but could be an evolutionary development to optimize cooperativity.     

Transition states for folding of circular-permuted proteins

To explore the role of entropy and chain connectivity in protein folding, a particularly interesting scheme, namely, the circular permutation, has been used. Recently, experimental observations showed that there are large differences in the folding mechanisms between the wild-type proteins and their circular permutants. These differences are strongly related to the change in the intrachain connectivity. Some results obtained by molecular dynamics simulations also showed a good agreement with the experimental findings. Here, we use a topology-based free-energy functional method to study the role of the chain connectivity in folding by comparing features of transition states of the wild-type proteins with those of their circular permutants. We concentrate our study on 3 small globular proteins, namely, the alpha-spectrin SH3 domain (SH3), the chymotrypsin inhibitor 2 (CI2), and the ribosomal protein S6, and obtain exciting results that are consistent with the available experimental and simulation results. A heterogeneity of the interaction energies between contacts for protein CI2 and for protein S6 is also introduced, which characterizes the strong interactions between contacts with long loops, as speculated from experiments for protein S6. The comparison between the folding nucleus of the wild-type proteins and those of their circular permutants indicates that chain connectivity affects remarkably the shapes of the energy profiles and thus the folding mechanism. Further comparisons between our theoretical calculated phi(th) values and the experimental observed phi(exp) values for the 3 proteins and their permutants show that our results are in good agreement with experimental ones and that correlations between them are high. These indicate that the free-energy functional method really provides a way to analyze the folding behavior of the circular-permuted proteins and therefore the folding mechanism of the wild-type proteins.     

The transition state for folding of a peripheral subunit-binding domain contains robust and ionic-strength dependent characteristics

The denaturant dependencies of the folding and unfolding kinetics were used to characterize the structure of the transition state for folding of E3BD, a peripheral subunit-binding domain. For the majority of E3BD mutants, the Phi-values calculated at 298 K from the analysis of chevron plots were in good agreement with those previously determined at 325 K using Arrhenius analysis. This agreement further demonstrates the general robustness of Phi-value analyses, since different experiments, methods of denaturation and thermodynamic assumptions were used to determine each set of Phi(F) values. The structure of the transition state for folding was grossly conserved at 298 K and 325 K, with residues in Helix I playing a lesser role in folding than those located in the 3(10) helix, disordered loop and Helix II. However, the energetic contributions of a cluster of basic residues close to the N-terminus and Helix I, which are an integral part of the ligand-binding site, were susceptible to ionic strength effects because of electrostatic strain in native and transition states of E3BD at low ionic strength. We found no evidence of the downhill folding previously proposed for E3BD, even though the conditions employed in this study significantly increased the energetic bias towards the native state.     

Phi-analysis of the folding of the B domain of protein A using multiple optical probes

We examined the co-operativity of ultra-fast folding of a protein and whether the Phi-value analysis of its transition state depended on the location of the optical probe. We incorporated in turn a tryptophan residue into each of the three helices of the B domain of Protein A. Each Trp mutant of the three-helix bundle protein was used as a pseudo-wild-type parent for Phi-analysis in which the intrinsic Trp fluorescence probed the formation of each helix during the transition state. Apart from local effects in the immediate vicinity of the probe, the three separate sets of Phi-values were in excellent agreement, demonstrating the overall co-operativity of folding and the robustness of the Phi-analysis. The transition state of folding of Protein A contains the second helix being well formed with many stabilizing tertiary hydrophobic interactions. In contrast, the first and the third helices are more poorly structured in the transition state. The mechanism of folding thus involves the concurrent formation of secondary and tertiary interactions, and is towards the nucleation-condensation extreme in the nucleation-condensation-framework continuum of mechanism, with helix 2 being the nucleus. We provide an error analysis of Phi-values derived purely from the kinetics of two-state chevron plots.     

Do proteins with similar folds have similar transition state structures? A diffuse transition state of the 169 residue apoflavodoxin

Apoflavodoxin from Anabaena PCC 7119 is a 169 residue globular protein of known structure and energetics. Here, we present a comprehensive Phi-value analysis to characterize the structure of its transition state. A total of 34 non-disruptive mutations are made throughout the structure and a range of Phi-values from zero to one are observed. In addition, a small set of eight aliphatic small-to-large mutations have been introduced in the hydrophobic core of the protein and they have been analyzed to investigate the feasibility of stabilizing the unfolding transition state by creating new non-native interactions. We find that the transition state of apoflavodoxin (so far the largest protein subjected to Phi-analysis) is diffuse and that it can be stabilized by unspecific hydrophobic interactions that can speed up the folding reaction. The data gathered on the apoflavodoxin transition state are compared with results from experimental studies in other proteins to revisit the relationship between the native state topology and transition state structure.     

The effect of increasing the stability of non-native interactions on the folding landscape of the bacterial immunity protein Im9

How stabilising non-native interactions influence protein folding energy landscapes is currently not well understood: such interactions could speed folding by reducing the conformational search to the native state, or could slow folding by increasing ruggedness. Here, we examine the influence of non-native interactions in the folding process of the bacterial immunity protein Im9, by exploiting our ability to manipulate the stability of the intermediate and rate-limiting transition state (TS) in the folding of this protein by minor alteration of its sequence or changes in solvent conditions. By analysing the properties of these species using Phi-value analysis, and exploration of the structural properties of the TS ensemble using molecular dynamics simulations, we demonstrate the importance of non-native interactions in immunity protein folding and demonstrate that the rate-limiting step involves partial reorganisation of these interactions as the TS ensemble is traversed. Moreover, we show that increasing the contribution to stability made by non-native interactions results in an increase in Phi-values of the TS ensemble without altering its structural properties or solvent-accessible surface area. The data suggest that the immunity proteins fold on multiple, but closely related, micropathways, resulting in a heterogeneous TS ensemble that responds subtly to mutation or changes in the solvent conditions. Thus, altering the relative strength of native and non-native interactions influences the search to the native state by restricting the pathways through the folding energy landscape.     

Cooperativity in two-state protein folding kinetics

We present a solvable model that predicts the folding kinetics of two-state proteins from their native structures. The model is based on conditional chain entropies. It assumes that folding processes are dominated by small-loop closure events that can be inferred from native structures. For CI2, the src SH3 domain, TNfn3, and protein L, the model reproduces two-state kinetics, and it predicts well the average Phi-values for secondary structures. The barrier to folding is the formation of predominantly local structures such as helices and hairpins, which are needed to bring nonlocal pairs of amino acids into contact.     

Simulation and experiment at high temperatures: ultrafast folding of a thermophilic protein by nucleation-condensation

We used Phi-value analysis to characterise the transition state for folding of a thermophilic protein at the relatively high temperature of 325 K. PhiF values for the folding of the three-helix bundle, peripheral subunit binding domain from Bacillus stearothermophilus (E3BD) were determined by temperature-jump experiments in the absence of chemical denaturants. E3BD folded in microseconds through a highly diffuse transition state. Excellent agreement was observed between experiment and the results from eight (independent) molecular dynamics simulations of unfolding at 373 K. We used a combination of heteronuclear NMR experiments and molecular dynamics simulations to characterise the denatured ensemble, and found that it contained very little persistent, residual structure. However, those regions that adopt helical structure in the native state were found by simulation to be poised for helix formation in the denatured state. These regions also had significant structure in the transition state for folding. The overall folding pathway appears to be nucleation-condensation.     

Folding rates and low-entropy-loss routes of two-state proteins

We develop a simple model for computing the rates and routes of folding of two-state proteins from the contact maps of their native structures. The model is based on the graph-theoretical concept of effective contact order (ECO). The model predicts that proteins fold by "zipping up" in a sequence of small-loop-closure events, depending on the native chain fold. Using a simple equation, with a few physical rate parameters, we obtain a good correlation with the folding rates of 24 two-state folding proteins. The model rationalizes data from Phi-value analysis that have been interpreted in terms of delocalized or polarized transition states. This model indicates how much of protein folding may take place in parallel, not along a single reaction coordinate or with a single transition state.     

Analysis of the pH-dependent folding and stability of histidine point mutants allows characterization of the denatured state and transition state for protein folding

pH-Dependent studies of the folding kinetics and stability of a set of His to Gln point mutants were used to characterize the denatured state and transition state ensembles for the C-terminal domain of the ribosomal protein L9 (CTL9). CTL9 contains three histidine residues, two of which, H106 and H134, are buried in the native state, while the third, H144, is more exposed. Comparison of the pH-dependent stability calculated using the Tanford-Wyman linkage relationship to the measured values demonstrates that the apparent pK(a) values of the three histidine residues are not significantly perturbed in the denatured state ensemble. Kinetic measurements show that mutation of H134 has a larger effect on the folding process than does mutation of H106 and H144. The Phi-value for H134 is significantly larger than the Phi-values for the other histidine residues, which are near zero at both pH 5.45 and pH 8.0. The Phi-value for H134 is higher, 0.55, at pH 8.0 than at pH 5.45, 0.39. At pH 5.45, H134 is protonated in the unfolded state but deprotonated in the native state, while at pH 8.0 it is deprotonated in both. There is an excellent linear relationship between stability (logK) and folding rates (logk(f)) over the range of pH 5-9 for all mutants. From these plots, the ratio of DeltaQ( not equal)/DeltaQ can be calculated for each mutant. DeltaQ( not equal) is the difference in the number of protons bound to the transition state and to the unfolded state, while DeltaQ represents the difference between folded and denatured state. The linear plots indicate that the relative position of the transition state ensemble as judged by DeltaQ( not equal)/DeltaQ is independent of pH. The linkage analysis is consistent with the Phi-value analysis, showing that H134 is the most critical contributor to the development of pH-dependent interactions, including desolvation effects in the transition state ensemble.     

Mechanical unfolding of a titin Ig domain: structure of transition state revealed by combining atomic force microscopy,protein engineering and molecular dynamics simulations

Titin I27 shows a high resistance to unfolding when subject to external force. To investigate the molecular basis of this mechanical stability, protein engineering Phi-value analysis has been combined with atomic force microscopy to investigate the structure of the barrier to forced unfolding. The results indicate that the transition state for forced unfolding is significantly structured, since highly destabilising mutations in the core do not affect the force required to unfold the protein. As has been shown before, mechanical strength lies in the region of the A' and G-strands but, contrary to previous suggestions, the results indicate clearly that side-chain interactions play a significant role in maintaining mechanical stability. Since Phi-values calculated from molecular dynamics simulations are the same as those determined experimentally, we can, with confidence, use the molecular dynamics simulations to analyse the structure of the transition state in detail, and are able to show loss of interactions between the A' and G-strands with associated A-B and E-F loops in the transition state. The key event is not a simple case of loss of hydrogen bonding interactions between the A' and G-strands alone. Comparison with Phi-values from traditional folding studies shows differences between the force and "no-force" transition states but, nevertheless, the region important for kinetic stability is the same in both cases. This explains the correspondence between hierarchy of kinetic stability (measured in stopped-flow denaturant studies) and mechanical strength in these titin domains.