Thermodynamic effects of proline introduction on protein stability

The amino acid Pro is more rigid than other naturally occurring amino acids and, in proteins, lacks an amide hydrogen. To understand the structural and thermodynamic effects of Pro substitutions, it was introduced at 13 different positions in four different proteins, leucine-isoleucine-valine binding protein, maltose binding protein, ribose binding protein, and thioredoxin. Three of the maltose binding protein mutants were characterized by X-ray crystallography to confirm that no structural changes had occurred upon mutation. In the remaining cases, fluorescence and CD spectroscopy were used to show the absence of structural change. Stabilities of wild type and mutant proteins were characterized by chemical denaturation at neutral pH and by differential scanning calorimetry as a function of pH. The mutants did not show enhanced stability with respect to chemical denaturation at room temperature. However, 6 of the 13 single mutants showed a small but significant increase in the free energy of thermal unfolding in the range of 0.3-2.4 kcal/mol, 2 mutants showed no change, and 5 were destabilized. In five of the six cases, the stabilization was because of reduced entropy of unfolding. However, the magnitude of the reduction in entropy of unfolding was typically several fold larger than the theoretical estimate of -4 cal K(-1) mol(-1) derived from the relative areas in the Ramachandran map accessible to Pro and Ala residues, respectively. Two double mutants were constructed. In both cases, the effects of the single mutations on the free energy of thermal unfolding were nonadditive.     

Large-scale prediction of protein geometry and stability changes for arbitrary single point mutations

We have developed a method to both predict the geometry and the relative stability of point mutants that may be used for arbitrary mutations. The geometry optimization procedure was first tested on a new benchmark of 2141 ordered pairs of X-ray crystal structures of proteins that differ by a single point mutation, the largest data set to date. An empirical energy function, which includes terms representing the energy contributions of the folded and denatured proteins and uses the predicted mutant side chain conformation, was fit to a training set consisting of half of a diverse set of 1816 experimental stability values for single point mutations in 81 different proteins. The data included a substantial number of small to large residue mutations not considered by previous prediction studies. After removing 22 (approximately 2%) outliers, the stability calculation gave a standard deviation of 1.08 kcal/mol with a correlation coefficient of 0.82. The prediction method was then tested on the remaining half of the experimental data, giving a standard deviation of 1.10 kcal/mol and covariance of 0.66 for 97% of the test set. A regression fit of the energy function to a subset of 137 mutants, for which both native and mutant structures were available, gave a prediction error comparable to that for the complete training set with predicted side chain conformations. We found that about half of the variation is due to conformation-independent residue contributions. Finally, a fit to the experimental stability data using these residue parameters exclusively suggests guidelines for improving protein stability in the absence of detailed structure information.     

Structure-based prediction of the effects of a missense variant on protein stability

Predicting the effects of amino acid substitutions on protein stability provides invaluable information for protein design, the assignment of biological function, and for understanding disease-associated variations. To understand the effects of substitutions, computational models are preferred to time-consuming and expensive experimental methods. Several methods have been proposed for this task including machine learning-based approaches. However, models trained using limited data have performance problems and many model parameters tend to be over-fitted. To decrease the number of model parameters and to improve the generalization potential, we calculated the amino acid contact energy change for point variations using a structure-based coarse-grained model. Based on the structural properties including contact energy (CE) and further physicochemical properties of the amino acids as input features, we developed two support vector machine classifiers. M47 predicted the stability of variant proteins with an accuracy of 87 % and a Matthews correlation coefficient of 0.68 for a large dataset of 1925 variants, whereas M8 performed better when a relatively small dataset of 388 variants was used for 20-fold cross-validation. The performance of the M47 classifier on all six tested contingency table evaluation parameters is better than that of existing machine learning-based models or energy function-based protein stability classifiers.     

Selection of mutations for increased protein stability

There are many ways to select mutations that increase the stability of proteins, including rational design, functional screening of randomly generated mutant libraries, and comparison of naturally occurring homologous proteins. The protein engineer's toolbox is expanding and the number of successful examples of engineered protein stability is increasing. Still, the selection of thermostable mutations is not a standard process. Selection is complicated by lack of knowledge of the process that leads to thermal inactivation and by the fact that proteins employ a large variety of structural tricks to achieve stability.     

Cavity-creating mutations in Pseudomonas aeruginosa azurin: effects on protein dynamics and stability

Changes in flexibility and structural stability of Pseudomonas aeruginosa azurin in response to cavity-creating mutations were probed by the phosphorescence emission of Trp-48, which was deeply buried in the compact hydrophobic core of the macromolecule, and by measurements of guanidinum hydrochloride unfolding, respectively. Replacement of the bulky side chains Phe-110, Phe-29, and Tyr-108 with the smaller Ala introduced cavities at different distances from the hydrophobic core. The phosphorescence lifetime (τ₀) of Trp-48, buried inside the protein core, and the acrylamide quenching rate constant (k_q) were used to monitor local and global flexibility changes induced by the introduction of the cavity. The results of this work demonstrate the following: 1), the effect on core flexibility of the insertion of cavities is not correlated readily to the distance of the cavity from the core; 2), the protein global flexibility results are related to the cavity distance from the packed core of the macromolecule; and 3), the increase in protein flexibility does not correspond necessarily to a comparable destabilizing effect of some mutations.     

Artificial intelligence challenges for predicting the impact of mutations on protein stability

Stability is a key ingredient of protein fitness, and its modification through targeted mutations has applications in various fields, such as protein engineering, drug design, and deleterious variant interpretation. Many studies have been devoted over the past decades to build new, more effective methods for predicting the impact of mutations on protein stability based on the latest developments in artificial intelligence. We discuss their features, algorithms, computational efficiency, and accuracy estimated on an independent test set. We focus on a critical analysis of their limitations, the recurrent biases toward the training set, their generalizability, and interpretability. We found that the accuracy of the predictors has stagnated at around 1 kcal/mol for over 15 years. We conclude by discussing the challenges that need to be addressed to reach improved performance.     

Integrated prediction of the effect of mutations on multiple protein characteristics

Site-directed mutagenesis is routinely used in modern biology to elucidate the functional or biophysical roles of protein residues, and plays an important role in the field of rational protein design. Over the past decade, a number of computational tools have been developed that can predict the effect of point mutations on a protein's biophysical characteristics. However, these programs usually provide predictions for only a single characteristic. Furthermore, online versions of these tools are often impractical to use for examination of large and diverse sets of mutants. We have created a new web application, (http://enzyme.ucd.ie/PEAT_SA), that can simultaneously predict the effect of mutations on stability, ligand affinity and pK(a) values. PEAT-SA also provides an expanded feature-set with respect to other online tools which includes the ability to obtain predictions for multiple mutants in one submission. As a result, researchers who use site-directed mutagenesis can access state-of-the-art protein design methods with a fraction of the effort previously required. The results of benchmarking PEAT-SA on standard test-sets demonstrate that its accuracy for all three prediction types compares well to currently available tools. We illustrate PEAT-SA's potential by using it to investigate the influence of mutations on the activity of Subtilisin BPN'. This example demonstrates how the ability to obtain a wide range of information from one source, that can be combined to obtain deeper insight into the influence of mutations, makes PEAT-SA a valuable service to both experimental and computational biologists.     

Effect of proline to alanine mutation on the thermal stability of the all-beta-sheet protein tendamistat

The temperature dependent denaturation of wild-type tendamistat and of the proline-free triple mutant P7A/P9A/P50A was investigated using Fourier-transform infrared (FTIR) spectroscopy. Whereas the temperature-induced unfolding is reversible in the wild type, aggregation was observed for the proline-free tendamistat when studied under the same conditions. The midpoint unfolding temperature T(m) was found as 82.3+/-0.5 degrees C in (2)H2O. The thermal stability of the proline-free mutant is reduced by 15 degrees C as compared to the wild type. Changes in the strength of hydrogen bonding of tyrosine O-H groups upon unfolding and aggregation are reflected in small shifts of the C-C stretching mode of the aromatic ring near 1515 cm(-1). Evaluation of data from different infrared (IR) bands sensitive to changes in secondary structure as well as to changes in tertiary structure strongly supports a two-state model for the unfolding process of wild-type tendamistat.     

Context-dependent effects of proline residues on the stability and folding pathway of ubiquitin.

Substitution of trans-proline at three positions in ubiquitin (residues 19, 37 and 38) produces significant context-dependent effects on protein stability (both stabilizing and destabilizing) that reflect changes to a combination of parameters including backbone flexibility, hydrophobic interactions, solvent accessibility to polar groups and intrinsic backbone conformational preferences. Kinetic analysis of the wild-type yeast protein reveals a predominant fast-folding phase which conforms to an apparent two-state folding model. Temperature-dependent studies of the refolding rate reveal thermodynamic details of the nature of the transition state for folding consistent with hydrophobic collapse providing the overall driving force. Br?nsted analysis of the refolding and unfolding rates of a family of mutants with a variety of side chain substitutions for P37 and P38 reveals that the two prolines, which are located in a surface loop adjacent to the C terminus of the main alpha-helix (residues 24-33), are not significantly structured in the transition state for folding and appear to be consolidated into the native structure only late in the folding process. We draw a similar conclusion regarding position 19 in the loop connecting the N-terminal beta-hairpin to the main alpha-helix. The proline residues of ubiquitin are passive spectators in the folding process, but influence protein stability in a variety of ways.     

Amyloid fibril formation in the context of full-length protein: effects of proline mutations on the amyloid fibril formation of beta2-microglobulin

Beta2-microglobulin (beta2-m), a typical immunoglobulin domain made of seven beta-strands, is a major component of amyloid fibrils formed in dialysis-related amyloidosis. To understand the mechanism of amyloid fibril formation in the context of full-length protein, we prepared various mutants in which proline (Pro) was introduced to each of the seven beta-strands of beta2-m. The mutations affected the amyloidogenic potential of beta2-m to various degrees. In particular, the L23P, H51P, and V82P mutations significantly retarded fibril extension at pH 2.5. Among these, only L23P is included in the known "minimal" peptide sequence, which can form amyloid fibrils when isolated as a short peptide. This indicates that the residues in regions other than the minimal sequence, such as H51P and V82P, determine the amyloidogenic potential in the full-length protein. To further clarify the mutational effects, we measured their stability against guanidine hydrochloride of the native state at pH 8.0 and the amyloid fibrils at pH 2.5. The amyloidogenicity of mutants showed a significant correlation with the stability of the amyloid fibrils, and little correlation was observed with that of the native state. It has been proposed that the stability of the native state and the unfolding rate to the amyloidogenic precursor as well as the conformational preference of the denatured state determine the amyloidogenicity of the proteins. The present results reveal that, in addition, stability of the amyloid fibrils is a key factor determining the amyloidogenic potential of the proteins.     

MICO: A meta-tool for prediction of the effects of non-synonymous mutations

The Next Generation Sequencing (NGS) is a state-of-the-art technology that produces high throughput data with high resolution mutation information in the genome. Numerous methods with different efficiencies have been developed to predict mutational effects in the genome. The challenge is to present the results in a balanced manner for better biological insights and interpretation. Hence, we describe a meta-tool named Mutation Information Collector (MICO) for automatically querying and collecting related information from multiple biology/bioinformatics enabled web servers with prediction capabilities. The predicted mutational results for the proteins of interest are returned and presented as an easy-to-read summary table in this service. MICO also allows for navigating the result from each website for further analysis.

Availability

http: //mico.ggc.org /MICO     

PremPS: Predicting the impact of missense mutations on protein stability

Computational methods that predict protein stability changes induced by missense mutations have made a lot of progress over the past decades. Most of the available methods however have very limited accuracy in predicting stabilizing mutations because existing experimental sets are dominated by mutations reducing protein stability. Moreover, few approaches could consistently perform well across different test cases. To address these issues, we developed a new computational method PremPS to more accurately evaluate the effects of missense mutations on protein stability. The PremPS method is composed of only ten evolutionary- and structure-based features and parameterized on a balanced dataset with an equal number of stabilizing and destabilizing mutations. A comprehensive comparison of the predictive performance of PremPS with other available methods on nine benchmark datasets confirms that our approach consistently outperforms other methods and shows considerable improvement in estimating the impacts of stabilizing mutations. A protein could have multiple structures available, and if another structure of the same protein is used, the predicted change in stability for structure-based methods might be different. Thus, we further estimated the impact of using different structures on prediction accuracy, and demonstrate that our method performs well across different types of structures except for low-resolution structures and models built based on templates with low sequence identity. PremPS can be used for finding functionally important variants, revealing the molecular mechanisms of functional influences and protein design. PremPS is freely available at https://lilab.jysw.suda.edu.cn/research/PremPS/, which allows to do large-scale mutational scanning and takes about four minutes to perform calculations for a single mutation per protein with ~ 300 residues and requires ~ 0.4 seconds for each additional mutation.     

Effects of proline mutations on the folding of staphylococcal nuclease

Effects of proline isomerizations on the equilibrium unfolding and kinetic refolding of staphylococcal nuclease were studied by circular dichroism in the peptide region (225 nm) and fluorescence spectra of a tryptophan residue. For this purpose, four single mutants (P11A, P31A, P42A, and P56A) and four multiple mutants (P11A/P47T/P117G, P11A/P31A/P47T/P117G, P11A/P31A/P42A/P47T/P117G, and P11A/P31A/P42A/P47T/P56A/P117G) were constructed. These mutants, together with the single and double mutants for Pro47 and Pro117 constructed in our previous study, cover all six proline sites of the nuclease. The P11A, P31A, and P42A mutations did not change the stability of the protein remarkably, while the P56A mutation increased protein stability to a small extent by 0.5 kcal/mol. The refolding kinetics of the protein were, however, affected remarkably by three of the mutations, namely, P11A, P31A, and P56A. Most notably, the amplitude of the slow phase of the triphasic refolding kinetics of the nuclease observed by stopped-flow circular dichroism decreased by increasing the number of the proline mutations; the slow phase disappeared completely in the proline-free mutant (P11A/P31A/P42A/P47T/P56A/P117G). The kinetic refolding reactions of the wild-type protein assessed in the presence of Escherichia coli cyclophilin A showed that the slow phase was accelerated by cyclophilin, indicating that the slow phase was rate-limited by cis-trans isomerization of the proline residues. Although the fast and middle phases of the refolding kinetics were not affected by cyclophilin, the amplitude of the middle phase decreased when the number of the proline mutations increased; the percent amplitudes for the wild-type protein and the proline-free mutants were 43 and 13%, respectively. In addition to these three phases detected with stopped-flow circular dichroism, a very fast phase of refolding was observed with stopped-flow fluorescence, which had a shorter dead time (3.6 ms) than the stopped-flow circular dichroism. The following conclusions were drawn. (1) The effects of the P11A, P31A, and P56A mutations on the refolding kinetics indicate that the isomerizations of the three proline residues are rate-limiting, suggesting that the structures around these residues (Pro11, Pro31, and Pro56) may be organized at an early stage of refolding. (2) The fast phase corresponds to the refolding of the native proline isomer, and the middle phase whose amplitude has decreased when the number of proline mutations was increased may correspond to the slow refolding of non-native proline isomers. The occurrence of the fast- and slow-refolding reactions together with the slow phase rate-limited by the proline isomerization suggests that there are parallel folding pathways for the native and non-native proline isomers. (3) The middle phase did not completely disappear in the proline-free mutant. This suggests that the slow-folding isomer is produced not only by the proline isomerizations but also by another conformational event that is not related to the prolines. (4) The very fast phase detected with the fluorescent measurements suggests that there is an intermediate at a very early stage of kinetic refolding.     

Effects of pathogenic proline mutations on myosin assembly

Laing distal myopathy (MPD1) is a genetically dominant myopathy characterized by early and selective weakness of the distal muscles. Mutations in the MYH7 gene encoding for the β-myosin heavy chain are the underlying genetic cause of MPD1. However, their pathogenic mechanisms are currently unknown. Here, we measure the biological effects of the R1500P and L1706P MPD1 mutations in different cellular systems. We show that, while the two mutations inhibit myosin self-assembly in non-muscle cells, they do not prevent incorporation of the mutant myosin into sarcomeres. Nevertheless, we find that the L1706P mutation affects proper antiparallel myosin association by accumulating in the bare zone of the sarcomere. Furthermore, bimolecular fluorescence complementation assay shows that the α-helix containing the R1500P mutation folds into homodimeric (mutant/mutant) and heterodimeric [mutant/wild type (WT)] myosin molecules that are competent for sarcomere incorporation. Both mutations also form aggregates consisting of cytoplasmic vacuoles surrounding paracrystalline arrays and amorphous rod-like inclusions that sequester WT myosin. Myosin aggregates were also detected in transgenic nematodes expressing the R1500P mutation. By showing that the two MPD1 mutations can have dominant effects on distinct components of the contractile apparatus, our data provide the first insights into the pathogenesis of the disease.     

Ligand effects on protein thermodynamic stability

A simple, partition-function formalism is used to describe the coupling between ligand binding and protein equilibrium unfolding. This general theoretical framework is shown to provide an adequate basis for the analysis of experimental ligand effects on the unfolding of complex protein systems. Nevertheless, the most important consequences of ligand binding for protein thermodynamic stability, as exposed by the partition-function approach, are found to be those demonstrated by Julian Sturtevant about 20 years ago.     

Understanding mutations and protein stability through tripeptides

A novel methodology to predict the local conformational changes in a protein as a consequence of missense mutations is proposed. A pentapeptide at the locus of mutation plays the dominant role and it is analyzed in terms of tripeptides. A measure for spatial and temporal fluctuations in a pentapeptide is devised and validated. The method does not involve any prior knowledge of structural templates from sequence homology studies. Structural deformations can be predicted with about 70-80% reliability in any protein. Disease causing mutations and benign mutations have been addressed. In particular, p53, retinoblastoma protein and lipoprotein lipase are studied in detail.     

Impact of proline residues on parvalbumin stability

Despite its higher net charge and reduced opportunities for favorable tertiary interactions, Ca(2+)-free rat beta-parvalbumin is more stable than rat alpha-parvalbumin. Under conditions wherein alpha denatures at 45.8 degrees C, beta denatures at 53.6 degrees. The homologous chicken beta isoform known as CPV3 also exhibits heightened stability-prompting an inquiry into the stabilizing influence of Pro-21 and Pro-26. Individual P21A and P26A mutations lower the T(m) of rat beta by 3.2 degrees, decreasing conformational stability by 0.74 kcal/mol. Simultaneous replacement of Pro-21 and Pro-26 essentially abolishes the excess stability (DeltaT(m) = -7.6 degrees; DeltaDeltaG(conf) = -1.77 kcal/mol). Significantly, the P21A/P26A variant displays Ca(2+) affinity virtually indistinguishable from wild-type beta, implying that structural alterations in the AB domain do not necessarily influence the divalent ion affinity of the CD-EF domain. The consequences of introducing proline at positions 21 and 26 in rat alpha were also examined. Whereas the H26P mutation raises the T(m) by 5.6 degrees (DeltaDeltaG(conf) = 1.25 kcal/mol), A21P lowers the T(m) by 8.5 degrees (DeltaDeltaG(conf) = -1.9 kcal/mol). Replacement of Ala-21 by proline in an alpha AB/beta CD-EF chimera increases the T(m) by 5.8 degrees (DeltaDeltaG(conf) = 0.95 kcal/mol), implying that the destabilization of alpha by Pro-21 results from steric conflict with a residue in the CD-EF domain. Consistent with that hypothesis, the K80S mutation markedly stabilizes alpha A21P, yielding a protein with a T(m) 2.0 degrees higher than wild-type alpha. The observed differences in stability resulting from proline addition/removal are largely consistent with alterations in main-chain and side-chain conformational entropy.     

Importance of surrounding residues for protein stability of partially buried mutations

For understanding the factors influencing protein stability, we have analyzed the relationship between changes in protein stability caused by partially buried mutations and changes in 48 physico-chemical, energetic and conformational properties of amino acid residues. Multiple regression equations were derived to predict the stability of protein mutants and the efficiency of the method has been verified with both back-check and jack-knife tests. We observed a good agreement between experimental and computed stabilities. Further, we have analyzed the effect of sequence window length from 1 to 12 residues on each side of the mutated residue to include the sequence information for predicting protein stability and we found that the preferred window length for obtaining the highest correlation is different for each secondary structure; the preferred window length for helical, strand and coil mutations are, respectively, 0, 9 and 4 residues on both sides of the mutant residues. However, all the secondary structures have significant correlation for a window length of one residue on each side of the mutant position, implying the role of short-range interactions. Extraction of surrounding residue information for various distances (3 to 20A) around the mutant position showed the highest correlation at 8A, 6A and 7A, respectively, for mutations in helical, strand and coil segments. Overall, the information about the surrounding residues within the sphere of 7 to 8A, may explain better the stability in all subsets of partially buried mutations implying that this distance is sufficient to accommodate the residues influenced by major intramolecular interactions for the stability of protein structures.