Further evidence for the likely completeness of the library of solved single domain protein structures

On Protein Loops, Prior Molecular States and Common Ancestors of Life

The principle of continuity demands the existence of prior molecular states and common ancestors responsible for extant macromolecular structure. Here, we focus on the emergence and evolution of loop prototypes - the elemental architects of protein domain structure. Phylogenomic reconstruction spanning superkingdoms and viruses generated an evolutionary chronology of prototypes with six distinct evolutionary phases defining a most parsimonious evolutionary progression of cellular life. Each phase was marked by strategic prototype accumulation shaping the structures and functions of common ancestors. The last universal common ancestor (LUCA) of cells and viruses and the last universal cellular ancestor (LUCellA) defined stem lines that were structurally and functionally complex. The evolutionary saga highlighted transformative forces. LUCA lacked biosynthetic ribosomal machinery, while the pivotal LUCellA lacked essential DNA biosynthesis and modern transcription. Early proteins therefore relied on RNA for genetic information storage but appeared initially decoupled from it, hinting at transformative shifts of genetic processing. Urancestral loop types suggest advanced folding designs were present at an early evolutionary stage. An exploration of loop geometric properties revealed gradual replacement of prototypes with α-helix and β-strand bracing structures over time, paving the way for the dominance of other loop types. AlphFold2-generated atomic models of prototype accretion described patterns of fold emergence. Our findings favor a ‛processual' model of evolving stem lines aligned with Woese's vision of a communal world. This model prompts discussing the 'problem of ancestors' and the challenges that lie ahead for research in taxonomy, evolution and complexity.

Tracing the birth of structural domains from loops during protein evolution

The structures and functions of proteins are embedded into the loop scaffolds of structural domains. Their origin and evolution remain mysterious. Here, we use a novel graph-theoretical approach to describe how modular and non-modular loop prototypes combine to form folded structures in protein domain evolution. Phylogenomic data-driven chronologies reoriented a bipartite network of loops and domains (and its projections) into 'waterfalls' depicting an evolving 'elementary functionome' (EF). Two primordial waves of functional innovation involving founder 'p-loop' and 'winged-helix' domains were accompanied by an ongoing emergence and reuse of structural and functional novelty. Metabolic pathways expanded before translation functionalities. A dual hourglass recruitment pattern transferred scale-free properties from loop to domain components of the EF network in generative cycles of hierarchical modularity. Modeling the evolutionary emergence of the oldest P-loop and winged-helix domains with AlphFold2 uncovered rapid convergence towards folded structure, suggesting that a folding vocabulary exists in loops for protein fold repurposing and design.

A review of visualisations of protein fold networks and their relationship with sequence and function

Proteins form arguably the most significant link between genotype and phenotype. Understanding the relationship between protein sequence and structure, and applying this knowledge to predict function, is difficult. One way to investigate these relationships is by considering the space of protein folds and how one might move from fold to fold through similarity, or potential evolutionary relationships. The many individual characterisations of fold space presented in the literature can tell us a lot about how well the current Protein Data Bank represents protein fold space, how convergence and divergence may affect protein evolution, how proteins affect the whole of which they are part, and how proteins themselves function. A synthesis of these different approaches and viewpoints seems the most likely way to further our knowledge of protein structure evolution and thus, facilitate improved protein structure design and prediction.

AlphaFold 2: Why It Works and Its Implications for Understanding the Relationships of Protein Sequence, Structure, and Function

AlphaFold 2 (AF2) was the star of CASP14, the last biannual structure prediction experiment. Using novel deep learning, AF2 predicted the structures of many difficult protein targets at or near experimental resolution. Here, we present our perspective of why AF2 works and show that it is a very sophisticated fold recognition algorithm that exploits the completeness of the library of single domain PDB structures. It has also learned local side chain packing rearrangements that enable it to refine proteins to high resolution. The benefits and limitations of its ability to predict the structures of many more proteins at or close to atomic detail are discussed.

On the possible origin of protein homochirality, structure, and biochemical function

Living systems have chiral molecules, e.g., native proteins that almost entirely contain L-amino acids. How protein homochirality emerged from a background of equal numbers of L and D amino acids is among many questions about life's origin. The origin of homochirality and its implications are explored in computer simulations examining the stability and structural and functional properties of an artificial library of compact proteins containing 1:1 (termed demi-chiral), 3:1, and 1:3 ratios of D:L and purely L or D amino acids generated without functional selection. Demi-chiral proteins have shorter secondary structures and fewer internal hydrogen bonds and are less stable than homochiral proteins. Selection for hydrogen bonding yields a preponderance of L or D amino acids. Demi-chiral proteins have native global folds, including similarity to early ribosomal proteins, similar small molecule ligand binding pocket geometries, and many constellations of L-chiral amino acids with a 1.0-Å RMSD to native enzyme active sites. For a representative subset containing 550 active site geometries matching 457 (2) 4-digit (3-digit) enzyme classification (E.C.) numbers, native active site amino acids were generated at random for 472 of 550 cases. This increases to 548 of 550 cases when similar residues are allowed. The most frequently generated sequences correspond to ancient enzymatic functions, e.g., glycolysis, replication, and nucleotide biosynthesis. Surprisingly, even without selection, demi-chiral proteins possess the requisite marginal biochemical function and structure of modern proteins, but were thermodynamically less stable. If demi-chiral proteins were present, they could engage in early metabolism, which created the feedback loop for transcription and cell formation.

Multistep mutational transformation of a protein fold through structural intermediates

New protein folds may evolve from existing folds through metamorphic evolution involving a dramatic switch in structure. To mimic pathways by which amino acid sequence changes could induce a change in fold, we designed two folded hybrids of Xfaso 1 and Pfl 6, a pair of homologous Cro protein sequences with ~40% identity but different folds (all-α vs. α + β, respectively). Each hybrid, XPH1 or XPH2, is 85% identical in sequence to its parent, Xfaso 1 or Pfl 6, respectively; 55% identical to its noncognate parent; and ~70% identical to the other hybrid. XPH1 and XPH2 also feature a designed hybrid chameleon sequence corresponding to the C-terminal region, which switched from α-helical to β-sheet structure during Cro evolution. We report solution nuclear magnetic resonance (NMR) structures of XPH1 and XPH2 at 0.3 Å and 0.5 Å backbone root mean square deviation (RMSD), respectively. XPH1 retains a global fold generally similar to Xfaso 1, and XPH2 retains a fold similar to Pfl 6, as measured by TM-align scores (~0.7), DALI Z-scores (7-9), and backbone RMSD (2-3 Å RMSD for the most ordered regions). However, these scores also indicate significant deviations in structure. Most notably, XPH1 and XPH2 have different, and intermediate, secondary structure content relative to Xfaso 1 and Pfl 6. The multistep progression in sequence, from Xfaso 1 to XPH1 to XPH2 to Pfl 6, thus involves both abrupt and gradual changes in folding pattern. The plasticity of some protein folds may allow for "polymetamorphic" evolution through intermediate structures.

Using molecular simulation to explore the nanoscale dynamics of the plant kinome

Eukaryotic protein kinases (PKs) are a large family of proteins critical for cellular response to external signals, acting as molecular switches. PKs propagate biochemical signals by catalyzing phosphorylation of other proteins, including other PKs, which can undergo conformational changes upon phosphorylation and catalyze further phosphorylations. Although PKs have been studied thoroughly across the domains of life, the structures of these proteins are sparsely understood in numerous groups of organisms, including plants. In addition to efforts towards determining crystal structures of PKs, research on human PKs has incorporated molecular dynamics (MD) simulations to study the conformational dynamics underlying the switching of PK function. This approach of experimental structural biology coupled with computational biophysics has led to improved understanding of how PKs become catalytically active and why mutations cause pathological PK behavior, at spatial and temporal resolutions inaccessible to current experimental methods alone. In this review, we argue for the value of applying MD simulation to plant PKs. We review the basics of MD simulation methodology, the successes achieved through MD simulation in animal PKs, and current work on plant PKs using MD simulation. We conclude with a discussion of the future of MD simulations and plant PKs, arguing for the importance of molecular simulation in the future of plant PK research.

THE-DB: a threading model database for comparative protein structure analysis of the E. coli K12 and human proteomes

New methodology must be developed to improve the ability to characterize the growing number of amino acid sequences, which vastly exceeds the number of experimentally determined protein structures. Homologous proteins can be used as structural templates for modeling proteins that do not have experimentally determined structures. However, in many cases, there are no homologous proteins (typically <30% sequence identity) with determined structures from which a query sequence can be reliably modeled. The aim of protein threading is to use features, such as secondary structure, solvent accessibility and torsional angles, in addition to sequence patterns to identify structural templates from the protein databank to assist for full-length atomic-level structural modeling. However, there are still numerous protein sequences for which correct templates cannot be recognized. This raises the question as to what attributes allow query sequences to be matched to the correct but distantly homologous templates. To aid the investigation into this question and to provide genome-score protein structure for the biological community, a database called THE-DB (threading hard and easy protein database) has been developed in which it becomes possible to analyze over 15 000 query sequences from the Escherichia coli (E. coli) K12 and human proteomes, as well as to find their three-dimensional templates derived from the state-of-the-art threading algorithms which is not feasible with existing protein template databases. The E. coli K12 and human data can be downloaded in bulk from the THE-DB page.

Enhanced unbiased sampling of protein dynamics using evolutionary coupling information

One of the major challenges in atomistic simulations of proteins is efficient sampling of pathways associated with rare conformational transitions. Recent developments in statistical methods for computation of direct evolutionary couplings between amino acids within and across polypeptide chains have allowed for inference of native residue contacts, informing accurate prediction of protein folds and multimeric structures. In this study, we assess the use of distances between evolutionarily coupled residues as natural choices for reaction coordinates which can be incorporated into Markov state model-based adaptive sampling schemes and potentially used to predict not only functional conformations but also pathways of conformational change, protein folding, and protein-protein association. We demonstrate the utility of evolutionary couplings in sampling and predicting activation pathways of the β 2-adrenergic receptor (β 2-AR), folding of the FiP35 WW domain, and dimerization of the E. coli molybdopterin synthase subunits. We find that the time required for β 2-AR activation and folding of the WW domain are greatly diminished using evolutionary couplings-guided adaptive sampling. Additionally, we were able to identify putative molybdopterin synthase association pathways and near-crystal structure complexes from protein-protein association simulations.

Why Is There a Glass Ceiling for Threading Based Protein Structure Prediction Methods?

Despite their different implementations, comparison of the best threading approaches to the prediction of evolutionary distant protein structures reveals that they tend to succeed or fail on the same protein targets. This is true despite the fact that the structural template library has good templates for all cases. Thus, a key question is why are certain protein structures threadable while others are not. Comparison with threading results on a set of artificial sequences selected for stability further argues that the failure of threading is due to the nature of the protein structures themselves. Using a new contact map based alignment algorithm, we demonstrate that certain folds are highly degenerate in that they can have very similar coarse grained fractions of native contacts aligned and yet differ significantly from the native structure. For threadable proteins, this is not the case. Thus, contemporary threading approaches appear to have reached a plateau, and new approaches to structure prediction are required.

Evaluating the accuracy of protein design using native secondary sub-structures

BACKGROUND

According to structure-dependent function of proteins, two main challenging problems called Protein Structure Prediction (PSP) and Inverse Protein Folding (IPF) are investigated. In spite of IPF essential applications, it has not been investigated as much as PSP problem. In fact, the ultimate goal of IPF problem or protein design is to create proteins with enhanced properties or even novel functions. One of the major computational challenges in protein design is its large sequence space, namely searching through all plausible sequences is impossible. Inasmuch as, protein secondary structure represents an appropriate primary scaffold of the protein conformation, undoubtedly studying the Protein Secondary Structure Inverse Folding (PSSIF) problem is a quantum leap forward in protein design, as it can reduce the search space. In this paper, a novel genetic algorithm which uses native secondary sub-structures is proposed to solve PSSIF problem. In essence, evolutionary information can lead the algorithm to design appropriate amino acid sequences respective to the target secondary structures. Furthermore, they can be folded to tertiary structures almost similar to their reference 3D structures.

RESULTS

The proposed algorithm called GAPSSIF benefits from evolutionary information obtained by solved proteins in the PDB. Therefore, we construct a repository of protein secondary sub-structures to accelerate convergence of the algorithm. The secondary structure of designed sequences by GAPSSIF is comparable with those obtained by Evolver and EvoDesign. Although we do not explicitly consider tertiary structure features through the algorithm, the structural similarity of native and designed sequences declares acceptable values.

CONCLUSIONS

Using the evolutionary information of native structures can significantly improve the quality of designed sequences. In fact, the combination of this information and effective features such as solvent accessibility and torsion angles leads IPF problem to an efficient solution. GAPSSIF can be downloaded at http://bioinformatics.aut.ac.ir/GAPSSIF/ .

Biophysics of protein evolution and evolutionary protein biophysics

The study of molecular evolution at the level of protein-coding genes often entails comparing large datasets of sequences to infer their evolutionary relationships. Despite the importance of a protein's structure and conformational dynamics to its function and thus its fitness, common phylogenetic methods embody minimal biophysical knowledge of proteins. To underscore the biophysical constraints on natural selection, we survey effects of protein mutations, highlighting the physical basis for marginal stability of natural globular proteins and how requirement for kinetic stability and avoidance of misfolding and misinteractions might have affected protein evolution. The biophysical underpinnings of these effects have been addressed by models with an explicit coarse-grained spatial representation of the polypeptide chain. Sequence-structure mappings based on such models are powerful conceptual tools that rationalize mutational robustness, evolvability, epistasis, promiscuous function performed by 'hidden' conformational states, resolution of adaptive conflicts and conformational switches in the evolution from one protein fold to another. Recently, protein biophysics has been applied to derive more accurate evolutionary accounts of sequence data. Methods have also been developed to exploit sequence-based evolutionary information to predict biophysical behaviours of proteins. The success of these approaches demonstrates a deep synergy between the fields of protein biophysics and protein evolution.

On the role of physics and evolution in dictating protein structure and function

How many of the structural and functional properties of proteins are inherent? Computer simulations provide a powerful tool to address this question. A series of studies on QS, quasi-spherical, compact polypeptides which lack any secondary structure; ART, artificial, proteins comprised of compact homopolypeptides with protein-like secondary structure; and PDB, native, single domain proteins shows that essentially all native global folds, pockets and protein-protein interfaces are in the ART library. This suggests that many protein properties are inherent and that evolution is involved in fine-tuning. The completeness of the space of ligand binding pockets and protein-protein interfaces suggests that promiscuous interactions are intrinsic to proteins and that the capacity to perform the biochemistry of life at low level does not require evolution. If so, this has profound consequences for the origin of life.

A comparative assessment and analysis of 20 representative sequence alignment methods for protein structure prediction

Protein sequence alignment is essential for template-based protein structure prediction and function annotation. We collect 20 sequence alignment algorithms, 10 published and 10 newly developed, which cover all representative sequence- and profile-based alignment approaches. These algorithms are benchmarked on 538 non-redundant proteins for protein fold-recognition on a uniform template library. Results demonstrate dominant advantage of profile-profile based methods, which generate models with average TM-score 26.5% higher than sequence-profile methods and 49.8% higher than sequence-sequence alignment methods. There is no obvious difference in results between methods with profiles generated from PSI-BLAST PSSM matrix and hidden Markov models. Accuracy of profile-profile alignments can be further improved by 9.6% or 21.4% when predicted or native structure features are incorporated. Nevertheless, TM-scores from profile-profile methods including experimental structural features are still 37.1% lower than that from TM-align, demonstrating that the fold-recognition problem cannot be solved solely by improving accuracy of structure feature predictions.

Evolution and structural organization of the C proteins of paramyxovirinae

The phosphoprotein (P) gene of most Paramyxovirinae encodes several proteins in overlapping frames: P and V, which share a common N-terminus (PNT), and C, which overlaps PNT. Overlapping genes are of particular interest because they encode proteins originated de novo, some of which have unknown structural folds, challenging the notion that nature utilizes only a limited, well-mapped area of fold space. The C proteins cluster in three groups, comprising measles, Nipah, and Sendai virus. We predicted that all C proteins have a similar organization: a variable, disordered N-terminus and a conserved, α-helical C-terminus. We confirmed this predicted organization by biophysically characterizing recombinant C proteins from Tupaia paramyxovirus (measles group) and human parainfluenza virus 1 (Sendai group). We also found that the C of the measles and Nipah groups have statistically significant sequence similarity, indicating a common origin. Although the C of the Sendai group lack sequence similarity with them, we speculate that they also have a common origin, given their similar genomic location and structural organization. Since C is dispensable for viral replication, unlike PNT, we hypothesize that C may have originated de novo by overprinting PNT in the ancestor of Paramyxovirinae. Intriguingly, in measles virus and Nipah virus, PNT encodes STAT1-binding sites that overlap different regions of the C-terminus of C, indicating they have probably originated independently. This arrangement, in which the same genetic region encodes simultaneously a crucial functional motif (a STAT1-binding site) and a highly constrained region (the C-terminus of C), seems paradoxical, since it should severely reduce the ability of the virus to adapt. The fact that it originated twice suggests that it must be balanced by an evolutionary advantage, perhaps from reducing the size of the genetic region vulnerable to mutations.

Template-based structure modeling of protein-protein interactions

The structure of protein-protein complexes can be constructed by using the known structure of other protein complexes as a template. The complex structure templates are generally detected either by homology-based sequence alignments or, given the structure of monomer components, by structure-based comparisons. Critical improvements have been made in recent years by utilizing interface recognition and by recombining monomer and complex template libraries. Encouraging progress has also been witnessed in genome-wide applications of template-based modeling, with modeling accuracy comparable to high-throughput experimental data. Nevertheless, bottlenecks exist due to the incompleteness of the protein-protein complex structure library and the lack of methods for distant homologous template identification and full-length complex structure refinement.

Detecting protein candidate fragments using a structural alphabet profile comparison approach

Predicting accurate fragments from sequence has recently become a critical step for protein structure modeling, as protein fragment assembly techniques are presently among the most efficient approaches for de novo prediction. A key step in these approaches is, given the sequence of a protein to model, the identification of relevant fragments - candidate fragments - from a collection of the available 3D structures. These fragments can then be assembled to produce a model of the complete structure of the protein of interest. The search for candidate fragments is classically achieved by considering local sequence similarity using profile comparison, or threading approaches. In the present study, we introduce a new profile comparison approach that, instead of using amino acid profiles, is based on the use of predicted structural alphabet profiles, where structural alphabet profiles contain information related to the 3D local shapes associated with the sequences. We show that structural alphabet profile-profile comparison can be used efficiently to retrieve accurate structural fragments, and we introduce a fully new protocol for the detection of candidate fragments. It identifies fragments specific of each position of the sequence and of size varying between 6 and 27 amino-acids. We find it outperforms present state of the art approaches in terms (i) of the accuracy of the fragments identified, (ii) the rate of true positives identified, while having a high coverage score. We illustrate the relevance of the approach on complete target sets of the two previous Critical Assessment of Techniques for Protein Structure Prediction (CASP) rounds 9 and 10. A web server for the approach is freely available at http://bioserv.rpbs.univ-paris-diderot.fr/SAFrag.

Protein folding and de novo protein design for biotechnological applications

In the postgenomic era, the medical/biological fields are advancing faster than ever. However, before the power of full-genome sequencing can be fully realized, the connection between amino acid sequence and protein structure, known as the protein folding problem, needs to be elucidated. The protein folding problem remains elusive, with significant difficulties still arising when modeling amino acid sequences lacking an identifiable template. Understanding protein folding will allow for unforeseen advances in protein design; often referred to as the inverse protein folding problem. Despite challenges in protein folding, de novo protein design has recently demonstrated significant success via computational techniques. We review advances and challenges in protein structure prediction and de novo protein design, and highlight their interplay in successful biotechnological applications.

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.

eVolver: an optimization engine for evolving protein sequences to stabilize the respective structures

BACKGROUND

Many structural bioinformatics approaches employ sequence profile-based threading techniques. To improve fold recognition rates, homology searching may include artificially evolved amino acid sequences, which were demonstrated to enhance the sensitivity of protein threading in targeting midnight zone templates.

FINDINGS

We describe implementation details of eVolver, an optimization algorithm that evolves protein sequences to stabilize the respective structures by a variety of potentials, which are compatible with those commonly used in protein threading. In a case study focusing on LARG PDZ domain, we show that artificially evolved sequences have quite high capabilities to recognize the correct protein structures using standard sequence profile-based fold recognition.

CONCLUSIONS

Computationally design protein sequences can be incorporated in existing sequence profile-based threading approaches to increase their sensitivity. They also provide a desired linkage between protein structure and function in in silico experiments that relate to e.g. the completeness of protein structure space, the origin of folds and protein universe. eVolver is freely available as a user-friendly webserver and a well-documented stand-alone software distribution at http://www.brylinski.org/evolver.

Interplay of physics and evolution in the likely origin of protein biochemical function

The intrinsic ability of protein structures to exhibit the geometric and sequence properties required for ligand binding without evolutionary selection is shown by the coincidence of the properties of pockets in native, single domain proteins with those in computationally generated, compact homopolypeptide, artificial (ART) structures. The library of native pockets is covered by a remarkably small number of representative pockets (∼400), with virtually every native pocket having a statistically significant match in the ART library, suggesting that the library is complete. When sequences are selected for ART structures based on fold stability, pocket sequence conservation is coincident to native. The fact that structurally and sequentially similar pockets occur across fold classes combined with the small number of representative pockets in native proteins implies that promiscuous interactions are inherent to proteins. Based on comparison of PDB (real, single domain protein structures found in the Protein Data Bank) and ART structures and pockets, the widespread assumption that the co-occurrence of global structure, pocket similarity, and amino acid conservation demands an evolutionary relationship between proteins is shown to significantly underestimate the random background probability. Indeed, many features of biochemical function arise from the physical properties of proteins that evolution likely fine-tunes to achieve specificity. Finally, our study suggests that a repertoire of thermodynamically (marginally) stable proteins could engage in many of the biochemical reactions needed for living systems without selection for function, a conclusion with significant implications for the origin of life.

EvoDesign: De novo protein design based on structural and evolutionary profiles

Protein design aims to identify new protein sequences of desirable structure and biological function. Most current de novo protein design methods rely on physics-based force fields to search for low free-energy states following Anfinsen’s thermodynamic hypothesis. A major obstacle of such approaches is the inaccuracy of the force field design, which cannot accurately describe the atomic interactions or distinguish correct folds. We developed a new web server, EvoDesign, to design optimal protein sequences of given scaffolds along with multiple sequence and structure-based features to assess the foldability and goodness of the designs. EvoDesign uses an evolution-profile–based Monte Carlo search with the profiles constructed from homologous structure families in the Protein Data Bank. A set of local structure features, including secondary structure, torsion angle and solvation, are predicted by single-sequence neural-network training and used to smooth the sequence motif and accommodate the physicochemical packing. The EvoDesign algorithm has been extensively tested in large-scale protein design experiments, which demonstrate enhanced foldability and structural stability of designed sequences compared with the physics-based designing methods. The EvoDesign server is freely available at http://zhanglab.ccmb.med.umich.edu/EvoDesign.

The utility of artificially evolved sequences in protein threading and fold recognition

Template-based protein structure prediction plays an important role in Functional Genomics by providing structural models of gene products, which can be utilized by structure-based approaches to function inference. From a systems level perspective, the high structural coverage of gene products in a given organism is critical. Despite continuous efforts towards the development of more sensitive threading approaches, confident structural models cannot be constructed for a considerable fraction of proteins due to difficulties in recognizing low-sequence identity templates with a similar fold to the target. Here we introduce a new modeling stratagem, which employs a library of synthetic sequences to improve template ranking in fold recognition by sequence profile-based methods. We developed a new method for the optimization of generic protein-like amino acid sequences to stabilize the respective structures using a combined empirical scoring function, which is compatible with these commonly used in protein threading and fold recognition. We show that the artificially evolved sequences, whose average sequence identity to the wild-type sequences is as low as 13.8%, have significant capabilities to recognize the correct structures. Importantly, the quality of the corresponding threading alignments is comparable to these constructed using conventional wild-type approaches (the average TM-score is 0.48 and 0.54, respectively). Fold recognition that uses data fusion to combine ranks calculated for both wild-type and synthetic template libraries systematically improves the detection of structural analogs. Depending on the threading algorithm used, it yields on average 4-16% higher recognition rates than using the wild-type template library alone. Synthetic sequences artificially evolved for the template structures provide an orthogonal source of signal that could be exploited to detect these templates unrecognized by standard modeling techniques. It opens up new directions in the development of more sensitive threading methods with the enhanced capabilities of targeting difficult, midnight zone templates.

Toward a "structural BLAST": using structural relationships to infer function

We outline a set of strategies to infer protein function from structure. The overall approach depends on extensive use of homology modeling, the exploitation of a wide range of global and local geometric relationships between protein structures and the use of machine learning techniques. The combination of modeling with broad searches of protein structure space defines a "structural BLAST" approach to infer function with high genomic coverage. Applications are described to the prediction of protein-protein and protein-ligand interactions. In the context of protein-protein interactions, our structure-based prediction algorithm, PrePPI, has comparable accuracy to high-throughput experiments. An essential feature of PrePPI involves the use of Bayesian methods to combine structure-derived information with non-structural evidence (e.g. co-expression) to assign a likelihood for each predicted interaction. This, combined with a structural BLAST approach significantly expands the range of applications of protein structure in the annotation of protein function, including systems level biological applications where it has previously played little role.

FINDSITE(comb): a threading/structure-based, proteomic-scale virtual ligand screening approach

Virtual ligand screening is an integral part of the modern drug discovery process. Traditional ligand-based, virtual screening approaches are fast but require a set of structurally diverse ligands known to bind to the target. Traditional structure-based approaches require high-resolution target protein structures and are computationally demanding. In contrast, the recently developed threading/structure-based FINDSITE-based approaches have the advantage that they are as fast as traditional ligand-based approaches and yet overcome the limitations of traditional ligand- or structure-based approaches. These new methods can use predicted low-resolution structures and infer the likelihood of a ligand binding to a target by utilizing ligand information excised from the target's remote or close homologous proteins and/or libraries of ligand binding databases. Here, we develop an improved version of FINDSITE, FINDSITE(filt), that filters out false positive ligands in threading identified templates by a better binding site detection procedure that includes information about the binding site amino acid similarity. We then combine FINDSITE(filt) with FINDSITE(X) that uses publicly available binding databases ChEMBL and DrugBank for virtual ligand screening. The combined approach, FINDSITE(comb), is compared to two traditional docking methods, AUTODOCK Vina and DOCK 6, on the DUD benchmark set. It is shown to be significantly better in terms of enrichment factor, dependence on target structure quality, and speed. FINDSITE(comb) is then tested for virtual ligand screening on a large set of 3576 generic targets from the DrugBank database as well as a set of 168 Human GPCRs. Excluding close homologues, FINDSITE(comb) gives an average enrichment factor of 52.1 for generic targets and 22.3 for GPCRs within the top 1% of the screened compound library. Around 65% of the targets have better than random enrichment factors. The performance is insensitive to target structure quality, as long as it has a TM-score ≥ 0.4 to native. Thus, FINDSITE(comb) makes the screening of millions of compounds across entire proteomes feasible. The FINDSITE(comb) web service is freely available for academic users at http://cssb.biology.gatech.edu/skolnick/webservice/FINDSITE-COMB/index.html.

eThread: a highly optimized machine learning-based approach to meta-threading and the modeling of protein tertiary structures

Template-based modeling that employs various meta-threading techniques is currently the most accurate, and consequently the most commonly used, approach for protein structure prediction. Despite the evident progress in this field, accurate structure models cannot be constructed for a significant fraction of gene products, thus the development of new algorithms is required. Here, we describe the development, optimization and large-scale benchmarking of eThread, a highly accurate meta-threading procedure for the identification of structural templates and the construction of corresponding target-to-template alignments. eThread integrates ten state-of-the-art threading/fold recognition algorithms in a local environment and extensively uses various machine learning techniques to carry out fully automated template-based protein structure modeling. Tertiary structure prediction employs two protocols based on widely used modeling algorithms: Modeller and TASSER-Lite. As a part of eThread, we also developed eContact, which is a Bayesian classifier for the prediction of inter-residue contacts and eRank, which effectively ranks generated multiple protein models and provides reliable confidence estimates as structure quality assessment. Excluding closely related templates from the modeling process, eThread generates models, which are correct at the fold level, for >80% of the targets; 40–50% of the constructed models are of a very high quality, which would be considered accurate at the family level. Furthermore, in large-scale benchmarking, we compare the performance of eThread to several alternative methods commonly used in protein structure prediction. Finally, we estimate the upper bound for this type of approach and discuss the directions towards further improvements.

When a domain is not a domain, and why it is important to properly filter proteins in databases: conflicting definitions and fold classification systems for structural domains make filtering of such databases imperative

Membership in a protein domain database does not a domain make; a feature we realized when generating a consensus view of protein fold space with our consensus domain dictionary (CDD). This dictionary was used to select representative structures for characterization of the protein dynameome: the Dynameomics initiative. Through this endeavor we rejected a surprising 40% of the 1,695 folds in the CDD as being non-autonomous folding units. Although some of this was due to the challenges of grouping similar fold topologies, the dissonance between the cataloguing and structural qualification of protein domains remains surprising. Another potential factor is previously overlooked intrinsic disorder; predictions suggest that 40% of proteins have either local or global disorder. One thing is clear, filtering a structural database and ensuring a consistent definition for protein domains is crucial, and caution is prescribed when generalizations of globular domains are drawn from unfiltered protein domain datasets.