De novo protein design. I. In search of stability and specificity

A generic framework for hierarchical de novo protein design

De novo protein design enables the exploration of novel sequences and structures absent from the natural protein universe. De novo design also stands as a stringent test for our understanding of the underlying physical principles of protein folding and may lead to the development of proteins with unmatched functional characteristics. The first fundamental challenge of de novo design is to devise "designable" structural templates leading to sequences that will adopt the predicted fold. Here, we built on the TopoBuilder (TB) de novo design method, to automatically assemble structural templates with native-like features starting from string descriptors that capture the overall topology of proteins. Our framework eliminates the dependency of hand-crafted and fold-specific rules through an iterative, data-driven approach that extracts geometrical parameters from structural tertiary motifs. We evaluated the TopoBuilder framework by designing sequences for a set of five protein folds and experimental characterization revealed that several sequences were folded and stable in solution. The TopoBuilder de novo design framework will be broadly useful to guide the generation of artificial proteins with customized geometries, enabling the exploration of the protein universe.

Key aspects of the past 30 years of protein design

Proteins are the workhorse of life. They are the building infrastructure of living systems; they are the most efficient molecular machines known, and their enzymatic activity is still unmatched in versatility by any artificial system. Perhaps proteins' most remarkable feature is their modularity. The large amount of information required to specify each protein's function is analogically encoded with an alphabet of just ∼20 letters. The protein folding problem is how to encode all such information in a sequence of 20 letters. In this review, we go through the last 30 years of research to summarize the state of the art and highlight some applications related to fundamental problems of protein evolution.

A sequence space search engine for computational protein design to modulate molecular functionality

De-novo protein design explores the untapped sequence space that is otherwise less discovered during the evolutionary process. This necessitates an efficient sequence space search engine for effective convergence in computational protein design. We propose a greedy simulated annealing-based Monte-Carlo parallel search algorithm for better sequence-structure compatibility probing in protein design. The guidance provided by the evolutionary profile, the greedy approach, and the cooling schedule adopted in the Monte Carlo simulation ensures sufficient exploration and exploitation of the search space leading to faster convergence. On evaluating the proposed algorithm, we find that a dataset of 76 target scaffolds report an average root-mean-square-deviation (RMSD) of 1.07 Å and an average TM-Score of 0.93 with the modeled designed protein sequences. High sequence recapitulation of 48.7% (59.4%) observed in the design sequences for all (hydrophobic) solvent-inaccessible residues again establish the goodness of the proposed algorithm. A high (93.4%) intra-group recapitulation of hydrophobic residues in the solvent-inaccessible region indicates that the proposed protein design algorithm preserves the core residues in the protein and provides alternative residue combinations in the solvent-accessible regions of the target protein. Furthermore, a COFACTOR-based protein functional analysis shows that the design sequences exhibit altered molecular functionality and introduce new molecular functions compared to the target scaffolds.Communicated by Ramaswamy H. Sarma.

An Evolutionary Profile Guided Greedy Parallel Replica-Exchange Monte Carlo Search Algorithm for Rapid Convergence in Protein Design

Protein design, also known as the inverse protein folding problem, is the identification of a protein sequence that folds into a target protein structure. Protein design is proved as an NP-hard problem. While researchers are working on designing heuristics with an emphasis on new scoring functions, we propose a replica-exchange Monte Carlo (REMC) search algorithm that ensures faster convergence using a greedy strategy. Using biological insights, we construct an evolutionary profile to encode the amino acid variability in different positions of the target protein from its structural homologs. The evolutionary profile guides the REMC search, and the greedy approach confirms appreciable exploration and exploitation of the sequence-structure fitness surface. We allow termination of a simulation trajectory once stagnant situation is detected. A series of sequence and structure level validations establish the goodness of our design. On a benchmark dataset, our algorithm reports an average root-mean-square deviation of 1.21Å between the target and the design proteins when modeled with an existing protein folding software. Besides, our algorithm assures 6.16 times overall speedup. In Molecular Dynamics simulations, we observe that four out of selected five design proteins report better to comparable stability to the corresponding target proteins.

Hybrid MC/MD for protein design

Computational protein design relies on simulations of a protein structure, where selected amino acids can mutate randomly, and mutations are selected to enhance a target property, such as stability. Often, the protein backbone is held fixed and its degrees of freedom are modeled implicitly to reduce the complexity of the conformational space. We present a hybrid method where short molecular dynamics (MD) segments are used to explore conformations and alternate with Monte Carlo (MC) moves that apply mutations to side chains. The backbone is fully flexible during MD. As a test, we computed side chain acid/base constants or pK_a's in five proteins. This problem can be considered a special case of protein design, with protonation/deprotonation playing the role of mutations. The solvent was modeled as a dielectric continuum. Due to cost, in each protein we allowed just one side chain position to change its protonation state and the other position to change its type or mutate. The pK_a's were computed with a standard method that scans a range of pH values and with a new method that uses adaptive landscape flattening (ALF) to sample all protonation states in a single simulation. The hybrid method gave notably better accuracy than standard, fixed-backbone MC. ALF decreased the computational cost a factor of 13.

Protein design under competing conditions for the availability of amino acids

Isolating the properties of proteins that allow them to convert sequence into the structure is a long-lasting biophysical problem. In particular, studies focused extensively on the effect of a reduced alphabet size on the folding properties. However, the natural alphabet is a compromise between versatility and optimisation of the available resources. Here, for the first time, we include the impact of the relative availability of the amino acids to extract from the 20 letters the core necessary for protein stability. We present a computational protein design scheme that involves the competition for resources between a protein and a potential interaction partner that, additionally, gives us the chance to investigate the effect of the reduced alphabet on protein-protein interactions. We devise a scheme that automatically identifies the optimal reduced set of letters for the design of the protein, and we observe that even alphabets reduced down to 4 letters allow for single protein folding. However, it is only with 6 letters that we achieve optimal folding, thus recovering experimental observations. Additionally, we notice that the binding between the protein and a potential interaction partner could not be avoided with the investigated reduced alphabets. Therefore, we suggest that aggregation could have been a driving force in the evolution of the large protein alphabet.

CryoProtect: A Web Server for Classifying Antifreeze Proteins from Nonantifreeze Proteins

Antifreeze protein (AFP) is an ice-binding protein that protects organisms from freezing in extremely cold environments. AFPs are found across a diverse range of species and, therefore, significantly differ in their structures. As there are no consensus sequences available for determining the ice-binding domain of AFPs, thus the prediction and characterization of AFPs from their sequence is a challenging task. This study addresses this issue by predicting AFPs directly from sequence on a large set of 478 AFPs and 9,139 non-AFPs using machine learning (e.g., random forest) as a function of interpretable features (e.g., amino acid composition, dipeptide composition, and physicochemical properties). Furthermore, AFPs were characterized using propensity scores and important physicochemical properties via statistical and principal component analysis. The predictive model afforded high performance with an accuracy of 88.28% and results revealed that AFPs are likely to be composed of hydrophobic amino acids as well as amino acids with hydroxyl and sulfhydryl side chains. The predictive model is provided as a free publicly available web server called CryoProtect for classifying query protein sequence as being either AFP or non-AFP. The data set and source code are for reproducing the results which are provided on GitHub.

Designing peptide sequences in flexible chain conformations to bind RNA: a search algorithm combining Monte Carlo, self-consistent mean field and concerted rotation techniques

A search algorithm combining Monte Carlo, self-consistent mean field, and concerted rotation techniques was developed to discover peptide sequences that are reasonable HIV drug candidates due to their exceptional binding to human tRNAUUU(Lys3), the primer of HIV replication. The search algorithm allows for iteration between sequence mutations and conformation changes during sequence evolution. Searches conducted for different classes of peptides identified several potential peptide candidates. Analysis of the energy revealed that the asparagine and cysteine at residues 11 and 12 play important roles in "recognizing" tRNA(Lys3) via van der Waals interactions, contributing to binding specificity. Arginines preferentially attract the phosphate linkage via charge-charge interaction, contributing to binding affinity. Evaluation of the RNA/peptide complex's structure revealed that adding conformation changes to the search algorithm yields peptides with better binding affinity and specificity to tRNA(Lys3) than a previous mutation-only algorithm.

Protein side chain conformation predictions with an MMGBSA energy function

The prediction of protein side chain conformations from backbone coordinates is an important task in structural biology, with applications in structure prediction and protein design. It is a difficult problem due to its combinatorial nature. We study the performance of an "MMGBSA" energy function, implemented in our protein design program Proteus, which combines molecular mechanics terms, a Generalized Born and Surface Area (GBSA) solvent model, with approximations that make the model pairwise additive. Proteus is not a competitor to specialized side chain prediction programs due to its cost, but it allows protein design applications, where side chain prediction is an important step and MMGBSA an effective energy model. We predict the side chain conformations for 18 proteins. The side chains are first predicted individually, with the rest of the protein in its crystallographic conformation. Next, all side chains are predicted together. The contributions of individual energy terms are evaluated and various parameterizations are compared. We find that the GB and SA terms, with an appropriate choice of the dielectric constant and surface energy coefficients, are beneficial for single side chain predictions. For the prediction of all side chains, however, errors due to the pairwise additive approximation overcome the improvement brought by these terms. We also show the crucial contribution of side chain minimization to alleviate the rigid rotamer approximation. Even without GB and SA terms, we obtain accuracies comparable to SCWRL4, a specialized side chain prediction program. In particular, we obtain a better RMSD than SCWRL4 for core residues (at a higher cost), despite our simpler rotamer library. Proteins 2016; 84:803-819. © 2016 Wiley Periodicals, Inc.

Introducing folding stability into the score function for computational design of RNA-binding peptides boosts the probability of success

A computational strategy that integrates our peptide search algorithm with atomistic molecular dynamics simulation was used to design rational peptide drugs that recognize and bind to the anticodon stem and loop domain (ASL(Lys3)) of human tRNAUUULys3 for the purpose of interrupting HIV replication. The score function of the search algorithm was improved by adding a peptide stability term weighted by an adjustable factor λ to the peptide binding free energy. The five best peptide sequences associated with five different values of λ were determined using the search algorithm and then input in atomistic simulations to examine the stability of the peptides' folded conformations and their ability to bind to ASL(Lys3). Simulation results demonstrated that setting an intermediate value of λ achieves a good balance between optimizing the peptide's binding ability and stabilizing its folded conformation during the sequence evolution process, and hence leads to optimal binding to the target ASL(Lys3). Thus, addition of a peptide stability term significantly improves the success rate for our peptide design search.

Protein sequence design and its applications

Design of proteins has far-reaching potentials in diverse areas that span repurposing of the protein scaffold for reactions and substrates that they were not naturally meant for, to catching a glimpse of the ephemeral proteins that nature might have sampled during evolution. These non-natural proteins, either in synthesized or virtual form have opened the scope for the design of entities that not only rival their natural counterparts but also offer a chance to visualize the protein space continuum that might help to relate proteins and understand their associations. Here, we review the recent advances in protein engineering and design, in multiple areas, with a view to drawing attention to their future potential.

De novo design of stereochemically-bent sixteen-residue β-hairpin as a hydrolase mimic

Stepwise design of sixteen-residue β-hairpin as a hydrolase mimic involving fold design by stereochemical mutation followed by inverse-design of sequence.

NrichD database: sequence databases enriched with computationally designed protein-like sequences aid in remote homology detection

NrichD (http://proline.biochem.iisc.ernet.in/NRICHD/) is a database of computationally designed protein-like sequences, augmented into natural sequence databases that can perform hops in protein sequence space to assist in the detection of remote relationships. Establishing protein relationships in the absence of structural evidence or natural ‘intermediately related sequences’ is a challenging task. Recently, we have demonstrated that the computational design of artificial intermediary sequences/linkers is an effective approach to fill naturally occurring voids in protein sequence space. Through a large-scale assessment we have demonstrated that such sequences can be plugged into commonly employed search databases to improve the performance of routinely used sequence search methods in detecting remote relationships. Since it is anticipated that such data sets will be employed to establish protein relationships, two databases that have already captured these relationships at the structural and functional domain level, namely, the SCOP database and the Pfam database, have been ‘enriched’ with these artificial intermediary sequences. NrichD database currently contains 3 611 010 artificial sequences that have been generated between 27 882 pairs of families from 374 SCOP folds. The data sets are freely available for download. Additional features include the design of artificial sequences between any two protein families of interest to the user.

Pairwise decomposition of an MMGBSA energy function for computational protein design

Computational protein design (CPD) aims at predicting new proteins or modifying existing ones. The computational challenge is huge as it requires exploring an enormous sequence and conformation space. The difficulty can be reduced by considering a fixed backbone and a discrete set of sidechain conformations. Another common strategy consists in precalculating a pairwise energy matrix, from which the energy of any sequence/conformation can be quickly obtained. In this work, we examine the pairwise decomposition of protein MMGBSA energy functions from a general theoretical perspective, and an implementation proposed earlier for CPD. It includes a Generalized Born term, whose many-body character is overcome using an effective dielectric environment, and a Surface Area term, for which we present an improved pairwise decomposition. A detailed evaluation of the error introduced by the decomposition on the different energy components is performed. We show that the error remains reasonable, compared to other uncertainties.

An evolution-based approach to De Novo protein design and case study on Mycobacterium tuberculosis

Computational protein design is a reverse procedure of protein folding and structure prediction, where constructing structures from evolutionarily related proteins has been demonstrated to be the most reliable method for protein 3-dimensional structure prediction. Following this spirit, we developed a novel method to design new protein sequences based on evolutionarily related protein families. For a given target structure, a set of proteins having similar fold are identified from the PDB library by structural alignments. A structural profile is then constructed from the protein templates and used to guide the conformational search of amino acid sequence space, where physicochemical packing is accommodated by single-sequence based solvation, torsion angle, and secondary structure predictions. The method was tested on a computational folding experiment based on a large set of 87 protein structures covering different fold classes, which showed that the evolution-based design significantly enhances the foldability and biological functionality of the designed sequences compared to the traditional physics-based force field methods. Without using homologous proteins, the designed sequences can be folded with an average root-mean-square-deviation of 2.1 Å to the target. As a case study, the method is extended to redesign all 243 structurally resolved proteins in the pathogenic bacteria Mycobacterium tuberculosis, which is the second leading cause of death from infectious disease. On a smaller scale, five sequences were randomly selected from the design pool and subjected to experimental validation. The results showed that all the designed proteins are soluble with distinct secondary structure and three have well ordered tertiary structure, as demonstrated by circular dichroism and NMR spectroscopy. Together, these results demonstrate a new avenue in computational protein design that uses knowledge of evolutionary conservation from protein structural families to engineer new protein molecules of improved fold stability and biological functionality.

The goal of computational protein design is to create new protein sequences of desirable structure and biological function. Most protein design methods are developed to search for sequences with the lowest free-energy based on physics-based force fields following Anfinsen's thermodynamic hypothesis. A major obstacle of such approaches is the inaccuracy of the force-field design, which cannot accurately describe atomic interactions or correctly recognize protein folds. We propose a novel method which uses evolutionary information, in the form of sequence profiles from structure families, to guide the sequence design. Since sequence profiles are generally more accurate than physics-based potentials in protein fold recognition, a unique advantage lies on that it targets the design procedure to a family of protein sequence profiles to enhance the robustness of designed sequences. The method was tested on 87 proteins and the designed sequences can be folded by I-TASSER to models with an average RMSD 2.1 Å. As a case study of large-scale application, the method is extended to redesign all structurally resolved proteins in the human pathogenic bacteria, Mycobacterium tuberculosis. Five sequences varying in fold and sizes were characterized by circular dichroism and NMR spectroscopy experiments and three were shown to have ordered tertiary structure.

The design of a peptide sequence to inhibit HIV replication: a search algorithm combining Monte Carlo and self-consistent mean field techniques

We developed a search algorithm combining Monte Carlo (MC) and self-consistent mean field techniques to evolve a peptide sequence that has good binding capability to the anticodon stem and loop (ASL) of human lysine tRNA species, tRNA(Lys3), with the ultimate purpose of breaking the replication cycle of human immunodeficiency virus-1. The starting point is the 15-amino-acid sequence, RVTHHAFLGAHRTVG, found experimentally by Agris and co-workers to bind selectively to hypermodified tRNA(Lys3). The peptide backbone conformation is determined via atomistic simulation of the peptide-ASL(Lys3) complex and then held fixed throughout the search. The proportion of amino acids of various types (hydrophobic, polar, charged, etc.) is varied to mimic different peptide hydration properties. Three different sets of hydration properties were examined in the search algorithm to see how this affects evolution to the best-binding peptide sequences. Certain amino acids are commonly found at fixed sites for all three hydration states, some necessary for binding affinity and some necessary for binding specificity. Analysis of the binding structure and the various contributions to the binding energy shows that: 1) two hydrophilic residues (asparagine at site 11 and the cysteine at site 12) "recognize" the ASL(Lys3) due to the VDW energy, and thereby contribute to its binding specificity and 2) the positively charged arginines at sites 4 and 13 preferentially attract the negatively charged sugar rings and the phosphate linkages, and thereby contribute to the binding affinity.

Computational protein design: the Proteus software and selected applications

We describe an automated procedure for protein design, implemented in a flexible software package, called Proteus. System setup and calculation of an energy matrix are done with the XPLOR modeling program and its sophisticated command language, supporting several force fields and solvent models. A second program provides algorithms to search sequence space. It allows a decomposition of the system into groups, which can be combined in different ways in the energy function, for both positive and negative design. The whole procedure can be controlled by editing 2-4 scripts. Two applications consider the tyrosyl-tRNA synthetase enzyme and its successful redesign to bind both O-methyl-tyrosine and D-tyrosine. For the latter, we present Monte Carlo simulations where the D-tyrosine concentration is gradually increased, displacing L-tyrosine from the binding pocket and yielding the binding free energy difference, in good agreement with experiment. Complete redesign of the Crk SH3 domain is presented. The top 10000 sequences are all assigned to the correct fold by the SUPERFAMILY library of Hidden Markov Models. Finally, we report the acid/base behavior of the SNase protein. Sidechain protonation is treated as a form of mutation; it is then straightforward to perform constant-pH Monte Carlo simulations, which yield good agreement with experiment. Overall, the software can be used for a wide range of application, producing not only native-like sequences but also thermodynamic properties with errors that appear comparable to other current software packages.

Protein WISDOM: a workbench for in silico de novo design of biomolecules

The aim of de novo protein design is to find the amino acid sequences that will fold into a desired 3-dimensional structure with improvements in specific properties, such as binding affinity, agonist or antagonist behavior, or stability, relative to the native sequence. Protein design lies at the center of current advances drug design and discovery. Not only does protein design provide predictions for potentially useful drug targets, but it also enhances our understanding of the protein folding process and protein-protein interactions. Experimental methods such as directed evolution have shown success in protein design. However, such methods are restricted by the limited sequence space that can be searched tractably. In contrast, computational design strategies allow for the screening of a much larger set of sequences covering a wide variety of properties and functionality. We have developed a range of computational de novo protein design methods capable of tackling several important areas of protein design. These include the design of monomeric proteins for increased stability and complexes for increased binding affinity. To disseminate these methods for broader use we present Protein WISDOM (http://www.proteinwisdom.org), a tool that provides automated methods for a variety of protein design problems. Structural templates are submitted to initialize the design process. The first stage of design is an optimization sequence selection stage that aims at improving stability through minimization of potential energy in the sequence space. Selected sequences are then run through a fold specificity stage and a binding affinity stage. A rank-ordered list of the sequences for each step of the process, along with relevant designed structures, provides the user with a comprehensive quantitative assessment of the design. Here we provide the details of each design method, as well as several notable experimental successes attained through the use of the methods.

Computer-aided antibody design

Recent clinical trials using antibodies with low toxicity and high efficiency have raised expectations for the development of next-generation protein therapeutics. However, the process of obtaining therapeutic antibodies remains time consuming and empirical. This review summarizes recent progresses in the field of computer-aided antibody development mainly focusing on antibody modeling, which is divided essentially into two parts: (i) modeling the antigen-binding site, also called the complementarity determining regions (CDRs), and (ii) predicting the relative orientations of the variable heavy (V(H)) and light (V(L)) chains. Among the six CDR loops, the greatest challenge is predicting the conformation of CDR-H3, which is the most important in antigen recognition. Further computational methods could be used in drug development based on crystal structures or homology models, including antibody-antigen dockings and energy calculations with approximate potential functions. These methods should guide experimental studies to improve the affinities and physicochemical properties of antibodies. Finally, several successful examples of in silico structure-based antibody designs are reviewed. We also briefly review structure-based antigen or immunogen design, with application to rational vaccine development.

Dynamics in Sequence Space for RNA Secondary Structure Design

We have implemented a method for the design of RNA sequences that should fold to arbitrary secondary structures. A popular energy model allows one to take the derivative with respect to composition, which can then be interpreted as a force and used for Newtonian dynamics in sequence space. Combined with a negative design term, one can rapidly sample sequences which are compatible with a desired secondary structure via simulated annealing. Results for 360 structures were compared with those from another nucleic acid design program using measures such as the probability of the target structure and an ensemble-weighted distance to the target structure.

Computational design of a PDZ domain peptide inhibitor that rescues CFTR activity

The cystic fibrosis transmembrane conductance regulator (CFTR) is an epithelial chloride channel mutated in patients with cystic fibrosis (CF). The most prevalent CFTR mutation, ΔF508, blocks folding in the endoplasmic reticulum. Recent work has shown that some ΔF508-CFTR channel activity can be recovered by pharmaceutical modulators (“potentiators” and “correctors”), but ΔF508-CFTR can still be rapidly degraded via a lysosomal pathway involving the CFTR-associated ligand (CAL), which binds CFTR via a PDZ interaction domain. We present a study that goes from theory, to new structure-based computational design algorithms, to computational predictions, to biochemical testing and ultimately to epithelial-cell validation of novel, effective CAL PDZ inhibitors (called “stabilizers”) that rescue ΔF508-CFTR activity. To design the “stabilizers”, we extended our structural ensemble-based computational protein redesign algorithm to encompass protein-protein and protein-peptide interactions. The computational predictions achieved high accuracy: all of the top-predicted peptide inhibitors bound well to CAL. Furthermore, when compared to state-of-the-art CAL inhibitors, our design methodology achieved higher affinity and increased binding efficiency. The designed inhibitor with the highest affinity for CAL (kCAL01) binds six-fold more tightly than the previous best hexamer (iCAL35), and 170-fold more tightly than the CFTR C-terminus. We show that kCAL01 has physiological activity and can rescue chloride efflux in CF patient-derived airway epithelial cells. Since stabilizers address a different cellular CF defect from potentiators and correctors, our inhibitors provide an additional therapeutic pathway that can be used in conjunction with current methods.

Cystic fibrosis (CF) is an inherited disease that causes the body to produce thick mucus that clogs the lungs and obstructs the breakdown and absorption of food. The cystic fibrosis transmembrane conductance regulator (CFTR) is mutated in CF patients, and the most common mutation causes three defects in CFTR: misfolding, decreased function, and rapid degradation. Drugs are currently being studied to correct the first two CFTR defects, but the problem of rapid degradation remains. Recently, key protein-protein interactions have been discovered that implicate the protein CAL in CFTR degradation. Here we have developed new computational protein design algorithms and used them to successfully predict peptide inhibitors of the CAL-CFTR interface. Our algorithm uses a structural ensemble-based evaluation of protein sequences and conformations to calculate accurate predictions of protein-peptide binding affinities. The algorithm is general and can be applied to a wide variety of protein-protein interface designs. All of our designed inhibitors bound CAL with high affinity. We tested our top binding peptide and observed that the inhibitor could successfully rescue CFTR function in CF patient-derived epithelial cells. Our designed inhibitors provide a novel therapeutic path which could be used in combination with existing CF therapeutics for additive benefit.

Computational protein design with explicit consideration of surface hydrophobic patches

De novo protein design requires the identification of amino-acid sequences that favor the target-folded conformation and are soluble in water. One strategy for promoting solubility is to disallow hydrophobic residues on the protein surface during design. However, naturally occurring proteins often have hydrophobic amino acids on their surface that contribute to protein stability via the partial burial of hydrophobic surface area or play a key role in the formation of protein-protein interactions. A less restrictive approach for surface design that is used by the modeling program Rosetta is to parameterize the energy function so that the number of hydrophobic amino acids designed on the protein surface is similar to what is observed in naturally occurring monomeric proteins. Previous studies with Rosetta have shown that this limits surface hydrophobics to the naturally occurring frequency (∼28%), but that it does not prevent the formation of hydrophobic patches that are considerably larger than those observed in naturally occurring proteins. Here, we describe a new score term that explicitly detects and penalizes the formation of hydrophobic patches during computational protein design. With the new term, we are able to design protein surfaces that include hydrophobic amino acids at naturally occurring frequencies, but do not have large hydrophobic patches. By adjusting the strength of the new score term, the emphasis of surface redesigns can be switched between maintaining solubility and maximizing folding free energy.

PROTEIN SIDE-CHAIN PACKING PROBLEM: A MAXIMUM EDGE-WEIGHT CLIQUE ALGORITHMIC APPROACH

"Protein Side-chain Packing" has an ever-increasing application in the field of bio-informatics, dating from the early methods of homology modeling to protein design and to the protein docking. However, this problem is computationally known to be NP-hard.

In this regard, we have developed a novel approach to solve this problem using the notion of a maximum edge-weight clique. Our approach is based on efficient reduction of protein side-chain packing problem to a graph and then solving the reduced graph to find the maximum clique by applying an efficient clique finding algorithm developed by our co-authors. Since our approach is based on deterministic algorithms in contrast to the various existing algorithms based on heuristic approaches, our algorithm guarantees of finding an optimal solution.

We have tested this approach to predict the side-chain conformations of a set of proteins and have compared the results with other existing methods. We have found that our results are favorably comparable or better than the results produced by the existing methods. As our test set contains a protein of 494 residues, we have obtained considerable improvement in terms of size of the proteins and in terms of the efficiency and the accuracy of prediction.

Computational protein design with a generalized Born solvent model: application to Asparaginyl-tRNA synthetase

Computational Protein Design (CPD) is a promising method for high throughput protein and ligand mutagenesis. Recently, we developed a CPD method that used a polar-hydrogen energy function for protein interactions and a Coulomb/Accessible Surface Area (CASA) model for solvent effects. We applied this method to engineer aspartyl-adenylate (AspAMP) specificity into Asparaginyl-tRNA synthetase (AsnRS), whose substrate is asparaginyl-adenylate (AsnAMP). Here, we implement a more accurate function, with an all-atom energy for protein interactions and a residue-pairwise generalized Born model for solvent effects. As a first test, we compute aminoacid affinities for several point mutants of Aspartyl-tRNA synthetase (AspRS) and Tyrosyl-tRNA synthetase and stability changes for three helical peptides and compare with experiment. As a second test, we readdress the problem of AsnRS aminoacid engineering. We compare three design criteria, which optimize the folding free-energy, the absolute AspAMP affinity, and the relative (AspAMP-AsnAMP) affinity. The sequences and conformations are improved with respect to our previous, polar-hydrogen/CASA study: For several designed complexes, the AspAMP carboxylate forms three interactions with a conserved arginine and a designed lysine, as in the active site of the AspRS:AspAMP complex. The conformations and interactions are well maintained in molecular dynamics simulations and the sequences have an inverted specificity, favoring AspAMP over AsnAMP. The method is not fully successful, since experimental measurements with the seven most promising sequences show that they do not catalyze at a detectable level the adenylation of Asp (or Asn) with ATP. This may be due to weak AspAMP binding and/or disruption of transition-state stabilization.

Discovery of entry inhibitors for HIV-1 via a new de novo protein design framework

A new (to our knowledge) de novo design framework with a ranking metric based on approximate binding affinity calculations is introduced and applied to the discovery of what we believe are novel HIV-1 entry inhibitors. The framework consists of two stages: a sequence selection stage and a validation stage. The sequence selection stage produces a rank-ordered list of amino-acid sequences by solving an integer programming sequence selection model. The validation stage consists of fold specificity and approximate binding affinity calculations. The designed peptidic inhibitors are 12-amino-acids-long and target the hydrophobic core of gp41. A number of the best-predicted sequences were synthesized and their inhibition of HIV-1 was tested in cell culture. All peptides examined showed inhibitory activity when compared with no drug present, and the novel peptide sequences outperformed the native template sequence used for the design. The best sequence showed micromolar inhibition, which is a 3-15-fold improvement over the native sequence, depending on the donor. In addition, the best sequence equally inhibited wild-type and Enfuvirtide-resistant virus strains.

Computational protein design: validation and possible relevance as a tool for homology searching and fold recognition

BACKGROUND

Protein fold recognition usually relies on a statistical model of each fold; each model is constructed from an ensemble of natural sequences belonging to that fold. A complementary strategy may be to employ sequence ensembles produced by computational protein design. Designed sequences can be more diverse than natural sequences, possibly avoiding some limitations of experimental databases.

METHODOLOGY/PRINCIPAL FINDINGS

WE EXPLORE THIS STRATEGY FOR FOUR SCOP FAMILIES: Small Kunitz-type inhibitors (SKIs), Interleukin-8 chemokines, PDZ domains, and large Caspase catalytic subunits, represented by 43 structures. An automated procedure is used to redesign the 43 proteins. We use the experimental backbones as fixed templates in the folded state and a molecular mechanics model to compute the interaction energies between sidechain and backbone groups. Calculations are done with the Proteins@Home volunteer computing platform. A heuristic algorithm is used to scan the sequence and conformational space, yielding 200,000-300,000 sequences per backbone template. The results confirm and generalize our earlier study of SH2 and SH3 domains. The designed sequences ressemble moderately-distant, natural homologues of the initial templates; e.g., the SUPERFAMILY, profile Hidden-Markov Model library recognizes 85% of the low-energy sequences as native-like. Conversely, Position Specific Scoring Matrices derived from the sequences can be used to detect natural homologues within the SwissProt database: 60% of known PDZ domains are detected and around 90% of known SKIs and chemokines. Energy components and inter-residue correlations are analyzed and ways to improve the method are discussed.

CONCLUSIONS/SIGNIFICANCE

For some families, designed sequences can be a useful complement to experimental ones for homologue searching. However, improved tools are needed to extract more information from the designed profiles before the method can be of general use.

Standardized evaluation of protein stability

We compare mean force potential values of a large series of PDB models of proteins and peptides and find that, either as monomers or polymers, proteins longer than 200-250 residues have equivalent MFP values that are averaged to -65+/-3 kcal/aa. This value is named the standard or stability value. The standard value is reached irrespective of sequences and 3D folds. Peptides are too short to follow the rule and frequently exist as populations of conformers; one exception is peptides in amyloid fibrils. Fibrils surpass the standard value in accordance with their uppermost stability. In parallel, we calculate median MFP values of amino acids in stably folded PDB models of proteins: median values vary from -25 for Gly to -115 kcal/aa for Trp. These median values are used to score primary sequences of proteins: all sequences converge to a mean value of -63.5+/-2.5 kcal/aa, i.e., only 1.5 kcal less than the folded model standard. Sequences from unfolded proteins have lower values. This supports the conclusion that sequences carry in an important message and more specifically that diversity of amino acids in sequences is mandatory for stability. We also use the median amino acid MFP to score residue stability in 3D folds. This demonstrates that 3D folds are compromises between fragments of high and fragments of low scores and that functional residues are often but not always in the extreme score values. The approach opens to possibilities of evaluating any 3D model and of detecting functional residues and should help in conducting mutation assays.

Computational protein design as a tool for fold recognition

Computationally designed protein sequences have been proposed as a basis to perform fold recognition and homology searching. To investigate this possibility, an automated procedure is used to completely redesign 24 SH3 proteins and 22 SH2 proteins. We use the experimental backbone coordinates as fixed templates in the folded state and a molecular mechanics model to compute the pairwise interaction energies between all sidechain types and conformations. Energy calculations are done with the Proteins@Home volunteer computing platform. A heuristic algorithm is then used to scan the sequence and conformational space for optimal solutions. We produced 200,000-450,000 sequences for each backbone template. The designed sequences ressemble moderately-distant, natural homologues of the initial templates, according to their identity scores and their similarity with respect to the Pfam sets of SH2 and SH3 domains. Standard homology detection tools document their native-like character: the Conserved Domain Database recognizes 61% (52%) of our low-energy sequences as SH3 (SH2) domains; the SUPERFAMILY, Hidden-Markov Model library recognizes 81% (84%). Conversely, position specific scoring matrices (PSSMs) derived from our designed sequences can be used to detect natural homologues in sequence databases. Within SwissProt, a set of natural SH3 PSSMs detects 772 SH3 domains, for example; our designed PSSMs detect 67% of these, plus one additional sequence and two false positives. If six amino acids involved in substrate binding (a selective pressure not accounted for in our design) are reset to their experimental types, then 77% of the experimental SH3 domains are detected. Results for the SH2 domains are similar. Several directions to improve the method further are discussed.

A search for energy minimized sequences of proteins

In this paper, we present numerical evidence that supports the notion of minimization in the sequence space of proteins for a target conformation. We use the conformations of the real proteins in the Protein Data Bank (PDB) and present computationally efficient methods to identify the sequences with minimum energy. We use edge-weighted connectivity graph for ranking the residue sites with reduced amino acid alphabet and then use continuous optimization to obtain the energy-minimizing sequences. Our methods enable the computation of a lower bound as well as a tight upper bound for the energy of a given conformation. We validate our results by using three different inter-residue energy matrices for five proteins from protein data bank (PDB), and by comparing our energy-minimizing sequences with 80 million diverse sequences that are generated based on different considerations in each case. When we submitted some of our chosen energy-minimizing sequences to Basic Local Alignment Search Tool (BLAST), we obtained some sequences from non-redundant protein sequence database that are similar to ours with an E-value of the order of 10^-7. In summary, we conclude that proteins show a trend towards minimizing energy in the sequence space but do not seem to adopt the global energy-minimizing sequence. The reason for this could be either that the existing energy matrices are not able to accurately represent the inter-residue interactions in the context of the protein environment or that Nature does not push the optimization in the sequence space, once it is able to perform the function.

Challenges in the computational design of proteins

Protein design has many applications not only in biotechnology but also in basic science. It uses our current knowledge in structural biology to predict, by computer simulations, an amino acid sequence that would produce a protein with targeted properties. As in other examples of synthetic biology, this approach allows the testing of many hypotheses in biology. The recent development of automated computational methods to design proteins has enabled proteins to be designed that are very different from any known ones. Moreover, some of those methods mostly rely on a physical description of atomic interactions, which allows the designed sequences not to be biased towards known proteins. In this paper, we will describe the use of energy functions in computational protein design, the use of atomic models to evaluate the free energy in the unfolded and folded states, the exploration and optimization of amino acid sequences, the problem of negative design and the design of biomolecular function. We will also consider its use together with the experimental techniques such as directed evolution. We will end by discussing the challenges ahead in computational protein design and some of their future applications.

Computational protein design with side-chain conformational entropy

Recent advances in modeling protein structures at the atomic level have made it possible to tackle "de novo" computational protein design. Most procedures are based on combinatorial optimization using a scoring function that estimates the folding free energy of a protein sequence on a given main-chain structure. However, the computation of the conformational entropy in the folded state is generally an intractable problem, and its contribution to the free energy is not properly evaluated. In this article, we propose a new automated protein design methodology that incorporates such conformational entropy based on statistical mechanics principles. We define the free energy of a protein sequence by the corresponding partition function over rotamer states. The free energy is written in variational form in a pairwise approximation and minimized using the Belief Propagation algorithm. In this way, a free energy is associated to each amino acid sequence: we use this insight to rescore the results obtained with a standard minimization method, with the energy as the cost function. Then, we set up a design method that directly uses the free energy as a cost function in combination with a stochastic search in the sequence space. We validate the methods on the design of three superficial sites of a small SH3 domain, and then apply them to the complete redesign of 27 proteins. Our results indicate that accounting for entropic contribution in the score function affects the outcome in a highly nontrivial way, and might improve current computational design techniques based on protein stability.

Evaluating and optimizing computational protein design force fields using fixed composition-based negative design

An accurate force field is essential to computational protein design and protein fold prediction studies. Proper force field tuning is problematic, however, due in part to the incomplete modeling of the unfolded state. Here, we evaluate and optimize a protein design force field by constraining the amino acid composition of the designed sequences to that of a well behaved model protein. According to the random energy model, unfolded state energies are dependent only on amino acid composition and not the specific arrangement of amino acids. Therefore, energy discrepancies between computational predictions and experimental results, for sequences of identical composition, can be directly attributed to flaws in the force field's ability to properly account for folded state sequence energies. This aspect of fixed composition design allows for force field optimization by focusing solely on the interactions in the folded state. Several rounds of fixed composition optimization of the 56-residue beta1 domain of protein G yielded force field parameters with significantly greater predictive power: Optimized sequences exhibited higher wild-type sequence identity in critical regions of the structure, and the wild-type sequence showed an improved Z-score. Experimental studies revealed a designed 24-fold mutant to be stably folded with a melting temperature similar to that of the wild-type protein. Sequence designs using engrailed homeodomain as a scaffold produced similar results, suggesting the tuned force field parameters were not specific to protein G.

Mutations affecting the oligomerization interface of G-protein-coupled receptors revealed by a novel de novo protein design framework

Specific functional and pharmacological properties have recently been ascribed to G-protein-coupled receptor (GPCR) dimers/oligomers. Because the association of two identical or two distinct GPCR monomers seems to be required to elicit receptor function, it is necessary to understand the exact nature of this interaction. We present here a novel method for de novo protein design and its application to the prediction of mutations that can stabilize or destabilize a GPCR dimer while maintaining the monomer's native fold. To test the efficacy of this new method, the dimer of the single-spanned transmembrane domain of glycophorin A was used as a model system. Experimental data from mutagenesis of the helix-helix interface are compared with computational predictions at that interface, and the model's results are found to be consistent with the experimental findings. A flexible template was developed for the rhodopsin homodimer at atomic resolution and used to predict sets of three and five mutations. The results are found to be consistent across eight case studies, with favored mutations at each position. Mutation sets predicted to be the most disruptive at the dimerization interface are found to be less specific to the flexible template than sets predicted to be less disruptive.

Toward full-sequence de novo protein design with flexible templates for human beta-defensin-2

In this article, we introduce and apply our de novo protein design framework, which observes true backbone flexibility, to the redesign of human beta-defensin-2, a 41-residue cationic antimicrobial peptide of the innate immune system. The flexible design templates are generated using molecular dynamics simulations with both Generalized Born implicit solvation and explicit water molecules. These backbone templates were employed in addition to the x-ray crystal structure for designing human beta-defensin-2. The computational efficiency of our framework was demonstrated with the full-sequence design of the peptide with flexible backbone templates, corresponding to the mutation of all positions except the native cysteines.

Rational design of new binding specificity by simultaneous mutagenesis of calmodulin and a target peptide

Calcium-saturated calmodulin (CaM) binds and influences the activity of a varied collection of target proteins in most cells. This promiscuity underlies the role of CaM as a shared participant in calcium-dependent signal transduction pathways but imposes a handicap on popular CaM-based calcium biosensors, which display an undesired tendency to cross-react with cellular proteins. Designed CaM/target pairs that retain high affinity for one another but lack affinity for wild-type CaM and its natural interaction partners would therefore be useful as sensor components and possibly also as elements of "synthetic" cellular-signaling networks. Here, we have adopted a rational approach to creating suitably modified CaM/target complexes by using computational design methods to guide parallel site-directed mutagenesis of both binding partners. A hierarchical design procedure was applied to suggest a small number of complementary mutations on CaM and on a peptide ligand derived from skeletal-muscle light-chain kinase (M13). Experimental analysis showed that the procedure was successful in identifying CaM and M13 mutants with novel specificity for one another. Importantly, the designed complexes retained an affinity comparable to the wild-type CaM/M13 complex. These results represent a step toward the creation of CaM and M13 derivatives with specificity fully orthogonal to the wild-type proteins and show that qualitatively accurate predictions may be obtained from computational methods applied simultaneously to two proteins involved in multiple-linked binding equilibria.

A maximum likelihood framework for protein design

Background

The aim of protein design is to predict amino-acid sequences compatible with a given target structure. Traditionally envisioned as a purely thermodynamic question, this problem can also be understood in a wider context, where additional constraints are captured by learning the sequence patterns displayed by natural proteins of known conformation. In this latter perspective, however, we still need a theoretical formalization of the question, leading to general and efficient learning methods, and allowing for the selection of fast and accurate objective functions quantifying sequence/structure compatibility.

Results

We propose a formulation of the protein design problem in terms of model-based statistical inference. Our framework uses the maximum likelihood principle to optimize the unknown parameters of a statistical potential, which we call an inverse potential to contrast with classical potentials used for structure prediction. We propose an implementation based on Markov chain Monte Carlo, in which the likelihood is maximized by gradient descent and is numerically estimated by thermodynamic integration. The fit of the models is evaluated by cross-validation. We apply this to a simple pairwise contact potential, supplemented with a solvent-accessibility term, and show that the resulting models have a better predictive power than currently available pairwise potentials. Furthermore, the model comparison method presented here allows one to measure the relative contribution of each component of the potential, and to choose the optimal number of accessibility classes, which turns out to be much higher than classically considered.

Conclusion

Altogether, this reformulation makes it possible to test a wide diversity of models, using different forms of potentials, or accounting for other factors than just the constraint of thermodynamic stability. Ultimately, such model-based statistical analyses may help to understand the forces shaping protein sequences, and driving their evolution.

Ultra-fast evaluation of protein energies directly from sequence

The structure, function, stability, and many other properties of a protein in a fixed environment are fully specified by its sequence, but in a manner that is difficult to discern. We present a general approach for rapidly mapping sequences directly to their energies on a pre-specified rigid backbone, an important sub-problem in computational protein design and in some methods for protein structure prediction. The cluster expansion (CE) method that we employ can, in principle, be extended to model any computable or measurable protein property directly as a function of sequence. Here we show how CE can be applied to the problem of computational protein design, and use it to derive excellent approximations of physical potentials. The approach provides several attractive advantages. First, following a one-time derivation of a CE expansion, the amount of time necessary to evaluate the energy of a sequence adopting a specified backbone conformation is reduced by a factor of 10⁷ compared to standard full-atom methods for the same task. Second, the agreement between two full-atom methods that we tested and their CE sequence-based expressions is very high (root mean square deviation 1.1–4.7 kcal/mol, R² = 0.7–1.0). Third, the functional form of the CE energy expression is such that individual terms of the expansion have clear physical interpretations. We derived expressions for the energies of three classic protein design targets—a coiled coil, a zinc finger, and a WW domain—as functions of sequence, and examined the most significant terms. Single-residue and residue-pair interactions are sufficient to accurately capture the energetics of the dimeric coiled coil, whereas higher-order contributions are important for the two more globular folds. For the task of designing novel zinc-finger sequences, a CE-derived energy function provides significantly better solutions than a standard design protocol, in comparable computation time. Given these advantages, CE is likely to find many uses in computational structural modeling.

Many applications in computational structural biology involve evaluating the energy of a protein adopting a specific structure. A variety of functions are used for this purpose. Statistical potentials are fast to evaluate but do not have a clear biophysical basis, whereas physics-based functions consist of well-defined terms that can be costly to compute. This paper describes how the theory of cluster expansion, originally developed to describe the energies of alloys, can be applied to generate a physical potential for proteins that is extremely fast to evaluate. Cluster expansion is a way of representing a property of a system as a discrete function of its degrees of freedom. In this paper, it is used for the problem of protein design, where the energy is determined by the identities and conformations of amino acids at different sites on a fixed protein backbone. Application of cluster expansion to three small protein folds—the α-helical coiled coil, the zinc finger, and the WW domain—shows that protein sequence can be mapped directly to energy using a surprisingly simple function that maintains high accuracy. Promising results on these small systems suggest that the theory may have utility for macromolecular modeling more generally.

Protein design simulations suggest that side-chain conformational entropy is not a strong determinant of amino acid environmental preferences

Loss of side-chain conformational entropy is an important force opposing protein folding and the relative preferences of the amino acids for being buried or solvent exposed may be partially determined by which amino acids lose more side-chain entropy when placed in the core of a protein. To investigate these preferences, we have incorporated explicit modeling of side-chain entropy into the protein design algorithm, RosettaDesign. In the standard version of the program, the energy of a particular sequence for a fixed backbone depends only on the lowest energy side-chain conformations that can be identified for that sequence. In the new model, the free energy of a single amino acid sequence is calculated by evaluating the average energy and entropy of an ensemble of structures generated by Monte Carlo sampling of amino acid side-chain conformations. To evaluate the impact of including explicit side-chain entropy, sequences were designed for 110 native protein backbones with and without the entropy model. In general, the differences between the two sets of sequences are modest, with the largest changes being observed for the longer amino acids: methionine and arginine. Overall, the identity between the designed sequences and the native sequences does not increase with the addition of entropy, unlike what is observed when other key terms are added to the model (hydrogen bonding, Lennard-Jones energies, and solvation energies). These results suggest that side-chain conformational entropy has a relatively small role in determining the preferred amino acid at each residue position in a protein.

Common attributes of native-state structures of proteins, disordered proteins, and amyloid

We show that a framework derived from the common character of globular proteins can be used to understand the design of protein sequences, the behavior of intrinsically unstructured proteins, and the formation of amyloid fibrils in a unified manner. Our studies provide compelling support for the idea that protein native-state structures, the structures adopted by intrinsically unstructured proteins on binding as well as those of amyloid aggregates, all reside in a physical state of matter in which the free energy landscape is sculpted not by the specific sequence of amino acids, but rather by considerations of geometry and symmetry. We elucidate the key role played by sequence design in selecting the structure of choice from the predetermined menu of putative native-state structures.

A search algorithm for fixed-composition protein design

We present a computational protein design algorithm for finding low-energy sequences of fixed amino acid composition. The search algorithms used in protein design typically do not restrict amino acid composition. However, the random energy model of Shakhnovich suggests that the use of fixed-composition sequences may circumvent defects in the modeling of the denatured state. Our algorithm, FC_FASTER, links fixed-composition versions of Monte Carlo and the FASTER algorithm. As proof of principle, FC_FASTER was tested on an experimentally validated, full-sequence design of the beta1 domain of protein G. For the wild-type composition, FC_FASTER found a lower energy sequence than the experimentally validated sequence. Also, for a different composition, FC_FASTER found the hypothetical lowest-energy sequence in 14 out of 32 trials.