Protein-like Sequences Research Articles

Profiles of Natural and Designed Protein-Like Sequences Effectively Bridge Protein Sequence Gaps: Implications in Distant Homology Detection.

Sequence-based approaches are fundamental to guide experimental investigations in obtaining structural and/or functional insights into uncharacterized protein families. Powerful profile-based sequence search methods rely on a sequence space continuum to identify non-trivial relationships through homology detection. The computational design of protein-like sequences that serve as "artificial linkers" is useful in identifying relationships between distant members of a structural fold. Such sequences act as intermediates and guide homology searches between distantly related proteins. Here, we describe an approach that represents natural intermediate sequences and designed protein-like sequences as HMM (Hidden Markov Models) profiles, to improve the sensitivity of existing search methods. Searches made within the "Profile database" were shown to recognize the parent structural fold for 90% of the search queries at query coverage better than 60%. For 1040 protein families with no available structure, fold associations were made through searches in the database of natural and designed sequence profiles. Most of the associations were made with the Alpha-alpha superhelix, Transmembrane beta-barrels, TIM barrel, and Immunoglobulin-like beta-sandwich folds. For 11 domain families of unknown functions, we provide confident fold associations using the profiles of designed sequences and a consensus from other fold recognition methods. For two DUFs (Domain families of Unknown Functions), we performed detailed functional annotation through comparisons with characterized templates of families of known function.

A Specific Set of Heterogeneous Native Interactions Yields Efficient Knotting in Protein Folding.

Native interactions are crucial for folding, and non-native interactions appear to be critical for efficiently knotting proteins. Therefore, it is important to understand both their roles in the folding of knotted proteins. It has been proposed that non-native interactions drive the correct order of contact formation, which is essential to avoid backtracking and efficiently self-tie. In this study, we ask if non-native interactions are strictly necessary to tangle a protein or if the correct order of contact formation can be assured by a specific set of native, but otherwise heterogeneous (i.e., having distinct energies), interactions. In order to address this problem, we conducted extensive Monte Carlo simulations of lattice models of protein-like sequences designed to fold into a preselected knotted conformation embedding a trefoil knot. We were able to identify a specific set of heterogeneous native interactions that drives efficient knotting and is able to fold the protein when combined with the remaining native interactions modeled as homogeneous. This specific set of heterogeneous native interactions is strictly enough to efficiently self-tie. A distinctive feature of these native interactions is that they do not backtrack because their energies ensure the correct order of contact formation. Furthermore, they stabilize a knotted intermediate state, which is en route to the native structure. Our results thus show that-at least in the context of the adopted model-non-native interactions are not necessary to knot a protein. However, when they are taken into account in protein energetics, it is possible to find specific, nonlocal non-native interactions that operate as a scaffold that assists the knotting step.

How Deep Learning Tools Can Help Protein Engineers Find Good Sequences.

The deep learning revolution introduced a new and efficacious way to address computational challenges in a wide range of fields, relying on large data sets and powerful computational resources. In protein engineering, we consider the challenge of computationally predicting properties of a protein and designing sequences with these properties. Indeed, accurate and fast deep network oracles for different properties of proteins have been developed. These learn to predict a property from an amino acid sequence by training on large sets of proteins that have this property. In particular, deep networks can learn from the set of all known protein sequences to identify ones that are protein-like. A fundamental challenge when engineering sequences that are both protein-like and satisfy a desired property is that these are rare instances within the vast space of all possible ones. When searching for these very rare instances, one would like to use good sampling procedures. Sampling approaches that are decoupled from the prediction of the property or in which the predictor uses only post-sampling to identify good instances are less efficient. The alternative is to use sampling methods that are geared to generate sequences satisfying and/or optimizing the predictor's desired properties. Deep learning has a class of architectures, denoted as generative models, which offer the capability of sampling from the learned distribution of a predicted property. Here, we review the use of deep learning tools to find good sequences for protein engineering, including developing oracles/predictors of a property of the proteins and methods that sample from a distribution of protein-like sequences to optimize the desired property.

Artificial protein sequences enable recognition of vicinal and distant protein functional relationships.

High divergence in protein sequences makes the detection of distant protein relationships through homology-based approaches challenging. Grouping protein sequences into families, through similarities in either sequence or 3-D structure, facilitates in the improved recognition of protein relationships. In addition, strategically designed protein-like sequences have been shown to bridge distant structural domain families by serving as artificial linkers. In this study, we have augmented a search database of known protein domain families with such designed sequences, with the intention of providing functional clues to domain families of unknown structure. When assessed using representative query sequences from each family, we obtain a success rate of 94% in protein domain families of known structure. Further, we demonstrate that the augmented search space enabled fold recognition for 582 families with no structural information available a priori. Additionally, we were able to provide reliable functional relationships for 610 orphan families. We discuss the application of our method in predicting functional roles through select examples for DUF4922, DUF5131, and DUF5085. Our approach also detects new associations between families that were previously not known to be related, as demonstrated through new sub-groups of the RNA polymerase domain among three distinct RNA viruses. Taken together, designed sequences-augmented search databases direct the detection of meaningful relationships between distant protein families. In turn, they enable fold recognition and offer reliable pointers to potential functional sites that may be probed further through direct mutagenesis studies.

Boltzmann Machine Learning and Regularization Methods for Inferring Evolutionary Fields and Couplings From a Multiple Sequence Alignment

The inverse Potts problem to infer a Boltzmann distribution for homologous protein sequences from their single-site and pairwise amino acid frequencies recently attracts a great deal of attention in the studies of protein structure and evolution. We study regularization and learning methods and how to tune regularization parameters to correctly infer interactions in Boltzmann machine learning. Using L2 regularization for fields, group L1 for couplings is shown to be very effective for sparse couplings in comparison with L2 and L1. Two regularization parameters are tuned to yield equal values for both the sample and ensemble averages of evolutionary energy. Both averages smoothly change and converge, but their learning profiles are very different between learning methods. The Adam method is modified to make stepsize proportional to the gradient for sparse couplings and to use a soft-thresholding function for group L1. It is shown by first inferring interactions from protein sequences and then from Monte Carlo samples that the fields and couplings can be well recovered, but that recovering the pairwise correlations in the resolution of a total energy is harder for the natural proteins than for the protein-like sequences. Selective temperature for folding/structural constrains in protein evolution is also estimated.

Evidence for a novel negative-stranded RNA mycovirus isolated from the plant pathogenic fungus Fusarium graminearum

Here we describe a novel (−)ssRNA mycovirus, Fusarium graminearum negative-stranded RNA virus 1 (FgNSRV-1), isolated from Fusarium graminearum strain HN1. The genome of FgNSRV-1 is 9072 nucleotides in length, with five discontinuous but linear ORFs (ORF I-V). Phylogenetic analysis based on entire L polymerase sequences indicated that FgNSRV-1 is related to the (−)ssRNA mycovirus Sclerotinia sclerotiorum negative-stranded RNA virus 1 (SsNSRV-1), and other mycoviruses. Our data suggest that FgNSRV-1 can be classified into the family Mymonaviridae, order Mononegavirales. Putative enveloped virion-like structures with filamentous morphology similar to SsNSRV-1 were observed in virion preparation samples. The L proteins of FgNSRV-1, and other fungal mononegaviruses, were found to be related to L protein-like sequences in some fungal genome, supporting the hypothesis that there is coevolution occurring between mycoviruses and fungi. Besides, clearing the virus from the infected host fungus resulted in no discernable phenotypic change.

Innovative scattering analysis shows that hydrophobic disordered proteins are expanded in water.

A substantial fraction of the proteome is intrinsically disordered, and even well-folded proteins adopt non-native geometries during synthesis, folding, transport, and turnover. Characterization of intrinsically disordered proteins (IDPs) is challenging, in part because of a lack of accurate physical models and the difficulty of interpreting experimental results. We have developed a general method to extract the dimensions and solvent quality (self-interactions) of IDPs from a single small-angle x-ray scattering measurement. We applied this procedure to a variety of IDPs and found that even IDPs with low net charge and high hydrophobicity remain highly expanded in water, contrary to the general expectation that protein-like sequences collapse in water. Our results suggest that the unfolded state of most foldable sequences is expanded; we conjecture that this property was selected by evolution to minimize misfolding and aggregation.

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.

Filling-in Void and Sparse Regions in Protein Sequence Space by Protein-Like Artificial Sequences Enables Remarkable Enhancement in Remote Homology Detection Capability

Protein functional annotation relies on the identification of accurate relationships, sequence divergence being a key factor. This is especially evident when distant protein relationships are demonstrated only with three-dimensional structures. To address this challenge, we describe a computational approach to purposefully bridge gaps between related protein families through directed design of protein-like “linker” sequences. For this, we represented SCOP domain families, integrated with sequence homologues, as multiple profiles and performed HMM-HMM alignments between related domain families. Where convincing alignments were achieved, we applied a roulette wheel-based method to design 3,611,010 protein-like sequences corresponding to 374 SCOP folds. To analyze their ability to link proteins in homology searches, we used 3024 queries to search two databases, one containing only natural sequences and another one additionally containing designed sequences. Our results showed that augmented database searches showed up to 30% improvement in fold coverage for over 74% of the folds, with 52 folds achieving all theoretically possible connections. Although sequences could not be designed between some families, the availability of designed sequences between other families within the fold established the sequence continuum to demonstrate 373 difficult relationships. Ultimately, as a practical and realistic extension, we demonstrate that such protein-like sequences can be “plugged-into” routine and generic sequence database searches to empower not only remote homology detection but also fold recognition. Our richly statistically supported findings show that complementary searches in both databases will increase the effectiveness of sequence-based searches in recognizing all homologues sharing a common fold.

A Coarse-Grained Protein Model in a Water-like Solvent

Simulations employing an explicit atom description of proteins in solvent can be computationally expensive. On the other hand, coarse-grained protein models in implicit solvent miss essential features of the hydrophobic effect, especially its temperature dependence, and have limited ability to capture the kinetics of protein folding. We propose a free space two-letter protein (“H-P”) model in a simple, but qualitatively accurate description for water, the Jagla model, which coarse-grains water into an isotropically interacting sphere. Using Monte Carlo simulations, we design protein-like sequences that can undergo a collapse, exposing the “Jagla-philic” monomers to the solvent, while maintaining a “hydrophobic” core. This protein-like model manifests heat and cold denaturation in a manner that is reminiscent of proteins. While this protein-like model lacks the details that would introduce secondary structure formation, we believe that these ideas represent a first step in developing a useful, but computationally expedient, means of modeling proteins.

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.

Interplay between native topology and non-native interactions in the folding of tethered proteins

We assessed the interplay of native topology and non-native interactions on surface-tethered protein folding via extensive Monte Carlo simulations of a simple lattice model. In particular, we investigated the thermodynamics and kinetics of protein-like sequences enclosing different amounts of non-native interactions to protein energetics, and which were designed to fold to distinct native topologies. Our results show that the high-contact order (CO) structure renders a folding transition that is robust to (external) steric constraints and non-native interactions. On the other hand, the folding process of the simple low-CO topology can be easily hampered by the presence of a nearby chemically inert plane. In this case, if non-native interactions are highly conspicuous during folding they can actually drive chain collapse into a very native-like trapped state, which impedes the formation of the native structure. The analysis of folding kinetics reveals that the empirical correlation between folding rate and CO may not apply to surface-tethered protein folding. Indeed, results reported here show that depending on the native environment of the tethered chain terminus the folding rate of a low-CO topology can become so drastically small that the high-CO topology actually folds faster under the same conditions. We predict that complex topologies are more likely to conserve their bulk folding mechanism upon surface tethering.

Evidence for negative-strand RNA virus infection in fungi

Fungal viruses comprise two groups: a major group of five families with double-stranded RNA genomes and a minor group with positive-sense single-stranded (ss)RNA genomes. Although many fungal viruses have been identified, no negative-stranded (−)ssRNA mycoviruses have been reported. Here we present two lines of evidence suggesting the presence of (−)ssRNA viruses in filamentous fungi based on an exhaustive search using extant (−)ssRNA viruses as queries. This revealed (−)ssRNA virus L protein-like sequences in the genome of a phytopathogenic obligate ascomycete, Erysiphe pisi. A similar search for (−)ssRNA viruses in fungal transcriptome shotgun assembly libraries demonstrated that two independent libraries from Sclerotinia homoeocarpa, another phytopathogenic ascomycete, contained several sequences considered to correspond to the entire mononegavirus L gene and likely originating from an infecting (−)ssRNA virus. These results provide strong evidence for both ancient and extant (−)ssRNA virus infections in fungi.

On the Structural Space of Protein-Protein Interfaces

All major cellular processes in living cells are dependent on protein-protein interactions. At the heart of protein-protein interactions are protein-protein interfaces where the direct physical interactions occur. One may view the collection of all possible protein interface structures as a structural space. Understanding the nature of this space not only advances our fundamental knowledge about proteins but has profound implications for protein-protein interaction prediction and design. Can one explain the observed space of structures just by the principles of physics or does evolution need to be invoked as well? Here, we focus on interface structures of dimers, i.e., a complex formed by two protein monomers. By developing and applying an efficient structural alignment method, iAlign, we study the structural similarity of >1000 representative protein-protein interfaces. We present results of comparing experimental (native) interface structures formed by proteins whose monomers adopt different structures and show that most native interfaces have a close structural neighbor with similar backbone C geometry and interfacial contact pattern. To understand the possible origin of this interface similarity, we build artificial complexes from a library of randomly generated, compact homopolymeric structures and compare the structure of their interfaces to native interfaces. We show that most of native complexes can find an artificial counterpart, and vice versa. Moreover, from protein-like sequences further designed to be thermodynamically compatible with the artificial structures, native-like artificial protein complexes with strong favorable interactions emerge, albeit at a small probability.

BlastR—fast and accurate database searches for non-coding RNAs

We present and validate BlastR, a method for efficiently and accurately searching non-coding RNAs. Our approach relies on the comparison of di-nucleotides using BlosumR, a new log-odd substitution matrix. In order to use BlosumR for comparison, we recoded RNA sequences into protein-like sequences. We then showed that BlosumR can be used along with the BlastP algorithm in order to search non-coding RNA sequences. Using Rfam as a gold standard, we benchmarked this approach and show BlastR to be more sensitive than BlastN. We also show that BlastR is both faster and more sensitive than BlastP used with a single nucleotide log-odd substitution matrix. BlastR, when used in combination with WU-BlastP, is about 5% more accurate than WU-BlastN and about 50 times slower. The approach shown here is equally effective when combined with the NCBI-Blast package. The software is an open source freeware available from www.tcoffee.org/blastr.html.

Non‐native interactions play an effective role in protein folding dynamics

Systematic Monte Carlo simulations of simple lattice models show that the final stage of protein folding is an ordered process where native contacts get locked (i.e., the residues come into contact and remain in contact for the duration of the folding process) in a well-defined order. The detailed study of the folding dynamics of protein-like sequences designed as to exhibit different contact energy distributions, as well as different degrees of sequence optimization (i.e., participation of non-native interactions in the folding process), reveals significant differences in the corresponding locking scenarios--the collection of native contacts and their average locking times, which are largely ascribable to the dynamics of non-native contacts. Furthermore, strong evidence for a positive role played by non-native contacts at an early folding stage was also found. Interestingly, for topologically simple target structures, a positive interplay between native and non-native contacts is observed also toward the end of the folding process, suggesting that non-native contacts may indeed affect the overall folding process. For target models exhibiting clear two-state kinetics, the relation between the nucleation mechanism of folding and the locking scenario is investigated. Our results suggest that the stabilization of the folding transition state can be achieved through the establishment of a very small network of native contacts that are the first to lock during the folding process.

Characterizing Energy Landscapes of Proteins and Identifying Shape-Determining Factors

Protein folding is a long standing problem in biology, whose mechanism is still not completely understood. Funnel-shape energy landscape has been proposed as a plausible folding mechanism. However, the factors that determine the funnel-shape energy landscapes is largely unknown. In this study, we use hydrophobic-hydrophilic (HP) model to investigate the factors that affect the funnel-shape of protein energy landscapes. We designed a clustering method based on graph theory to analyze the conformations sampled using a recently developed Monte Carlo method, FRESS. We found that the way conformations move from one to another defined by a particular sampling method (move set) has a significant effect on the shape of protein energy landscapes. To our surprise, both protein-like sequences and random sequences with around 50% hydrophobic residues have a stable state represented by a single dominant cluster, consisting of a large number of similar conformations. The energy landscapes resemble a funnel, where there are many paths to minimum energy conformations in the dominant cluster from conformations of higher energies. We also found that sequences with hydrophobic residues above or below the optimal range around 50% do not have a single stable state. In stead, there are many much smaller clusters, representing multiple local energy minima. Our finding is consistent with the compositions of hydrophobic and polar residues in globular proteins (fold to unique structures) and intrinsically disordered proteins (IDPs). Our study suggests that in computational simulations, move sets affect significantly the shape of protein energy landscapes; hydrophobic interaction is likely a major force leading to the funnel-shape energy landscape of proteins; and the composition of hydrophobic and polar residues is an important sequence feature for the formation of funnel-shape of protein energy landscapes.

Classifying proteinlike sequences in arbitrary lattice protein models using LatPack

Knowledge of a protein's three-dimensional native structure is vital in determining its chemical properties and functionality. However, experimental methods to determine structure are very costly and time-consuming. Computational approaches such as folding simulations and structure prediction algorithms are quicker and cheaper but lack consistent accuracy. This currently restricts extensive computational studies to abstract protein models. It is thus essential that simplifications induced by the models do not negate scientific value. Key to this is the use of thoroughly defined proteinlike sequences. In such cases abstract models can allow for the investigation of important biological questions. Here, we present a procedure to generate and classify proteinlike sequence data sets. Our LatPack tools and the approach in general are applicable to arbitrary lattice protein models. Identification is based on thermodynamic kinetic features and incorporates the sequential assembly of proteins by addressing cotranslational folding. We demonstrate the approach in the widely used unrestricted 3D-cubic HP-model. The resulting sequence set is the first large data set for this model exhibiting the proteinlike properties required. Our data tools are freely available and can be used to investigate protein-related problems.

Geometry of the energy landscape and folding transition in a simple model of a protein

A geometric analysis of the global properties of the energy landscape of a minimalistic model of a polypeptide is presented, which is based on the relation between dynamical trajectories and geodesics of a suitable manifold, whose metric is completely determined by the potential energy. We consider different sequences, some with a definite proteinlike behavior, a unique native state and a folding transition, and others undergoing a hydrophobic collapse with no tendency to a unique native state. The global geometry of the energy landscape appears to contain relevant information on the behavior of the various sequences: in particular, the fluctuations of the curvature of the energy landscape, measured by means of numerical simulations, clearly mark the folding transition and allow the proteinlike sequences to be distinguished from the others.

Protein Folding Dynamics via Quantification of Kinematic Energy Landscape

We study folding dynamics of proteinlike sequences on a square lattice using a physical move set that exhausts all possible conformational changes. By analytically solving the master equation, we follow the time-dependent probabilities of occupancy of all 802 075 conformations of 16 mers over 7 orders of time span. We find that (i) folding rates of these proteinlike sequences of the same length can differ by 4 orders of magnitude, (ii) folding rates of sequences of the same conformation can differ by a factor of 190, and (iii) parameters of the native structures, designability, and thermodynamic properties are weak predictors of the folding rates; rather, a basin analysis of the kinematic energy landscape defined by the moves can provide an excellent account of the observed folding rates.