Cell-penetrating peptides predictors: A comparative analysis of methods and datasets.

Cell-Penetrating Peptides (CPP) are emerging as an alternative to small-molecule drugs to expand the range of biomolecules that can be targeted for therapeutic purposes. Due to the importance of identifying and designing new CPP, a great variety of predictors have been developed to achieve these goals. To establish a ranking for these predictors, a couple of recent studies compared their performances on specific datasets, yet their conclusions cannot determine if the ranking obtained is due to the model, the set of descriptors or the datasets used to test the predictors. We present a systematic study of the influence of the peptide sequence's similarity of the datasets on the predictors' performance. The analysis reveals that the datasets used for training have a stronger influence on the predictors performance than the model or descriptors employed. We show that datasets with low sequence similarity between the positive and negative examples can be easily separated, and the tested classifiers showed good performance on them. On the other hand, a dataset with high sequence similarity between CPP and non-CPP will be a hard dataset, and it should be the one to be used for assessing the performance of new predictors.

Antimicrobial peptides with cell-penetrating activity as prophylactic and treatment drugs.

Health is fundamental for the development of individuals and evolution of species. In that sense, for human societies is relevant to understand how the human body has developed molecular strategies to maintain health. In the present review, we summarize diverse evidence that support the role of peptides in this endeavor. Of particular interest to the present review are antimicrobial peptides (AMP) and cell-penetrating peptides (CPP). Different experimental evidence indicates that AMP/CPP are able to regulate autophagy, which in turn regulates the immune system response. AMP also assists in the establishment of the microbiota, which in turn is critical for different behavioral and health aspects of humans. Thus, AMP and CPP are multifunctional peptides that regulate two aspects of our bodies that are fundamental to our health: autophagy and microbiota. While it is now clear the multifunctional nature of these peptides, we are still in the early stages of the development of computational strategies aimed to assist experimentalists in identifying selective multifunctional AMP/CPP to control nonhealthy conditions. For instance, both AMP and CPP are computationally characterized as amphipatic and cationic, yet none of these features are relevant to differentiate these peptides from non-AMP or non-CPP. The present review aims to highlight current knowledge that may facilitate the development of AMP's design tools for preventing or treating illness.

Saturation Mutagenesis of the Transmembrane Region of HokC in Escherichia coli Reveals Its High Tolerance to Mutations.

Cells adapt to different stress conditions, such as the antibiotics presence. This adaptation sometimes is achieved by changing relevant protein positions, of which the mutability is limited by structural constrains. Understanding the basis of these constrains represent an important challenge for both basic science and potential biotechnological applications. To study these constraints, we performed a systematic saturation mutagenesis of the transmembrane region of HokC, a toxin used by Escherichia coli to control its own population, and observed that 92% of single-point mutations are tolerated and that all the non-tolerated mutations have compensatory mutations that reverse their effect. We provide experimental evidence that HokC accumulates multiple compensatory mutations that are found as correlated mutations in the HokC family multiple sequence alignment. In agreement with these observations, transmembrane proteins show higher probability to present correlated mutations and are less densely packed locally than globular proteins; previous mutagenesis results on transmembrane proteins further support our observations on the high tolerability to mutations of transmembrane regions of proteins. Thus, our experimental results reveal the HokC transmembrane region high tolerance to loss-of-function mutations that is associated with low sequence conservation and high rate of correlated mutations in the HokC family sequences alignment, which are features shared with other transmembrane proteins.

Selective Moonlighting Cell-Penetrating Peptides.

Cell penetrating peptides (CPPs) are molecules capable of passing through biological membranes. This capacity has been used to deliver impermeable molecules into cells, such as drugs and DNA probes, among others. However, the internalization of these peptides lacks specificity: CPPs internalize indistinctly on different cell types. Two major approaches have been described to address this problem: (i) targeting, in which a receptor-recognizing sequence is added to a CPP, and (ii) activation, where a non-active form of the CPP is activated once it interacts with cell target components. These strategies result in multifunctional peptides (i.e., penetrate and target recognition) that increase the CPP's length, the cost of synthesis and the likelihood to be degraded or become antigenic. In this work we describe the use of machine-learning methods to design short selective CPP; the reduction in size is accomplished by embedding two or more activities within a single CPP domain, hence we referred to these as moonlighting CPPs. We provide experimental evidence that these designed moonlighting peptides penetrate selectively in targeted cells and discuss areas of opportunity to improve in the design of these peptides.

Residue Cluster Classes: A Unified Protein Representation for Efficient Structural and Functional Classification.

Proteins are characterized by their structures and functions, and these two fundamental aspects of proteins are assumed to be related. To model such a relationship, a single representation to model both protein structure and function would be convenient, yet so far, the most effective models for protein structure or function classification do not rely on the same protein representation. Here we provide a computationally efficient implementation for large datasets to calculate residue cluster classes (RCCs) from protein three-dimensional structures and show that such representations enable a random forest algorithm to effectively learn the structural and functional classifications of proteins, according to the CATH and Gene Ontology criteria, respectively. RCCs are derived from residue contact maps built from different distance criteria, and we show that 7 or 8 Å with or without amino acid side-chain atoms rendered the best classification models. The potential use of a unified representation of proteins is discussed and possible future areas for improvement and exploration are presented.

Antimicrobial Peptide against Mycobacterium Tuberculosis That Activates Autophagy Is an Effective Treatment for Tuberculosis.

Mycobacterium tuberculosis (MTB) is the principal cause of human tuberculosis (TB), which is a serious health problem worldwide. The development of innovative therapeutic modalities to treat TB is mainly due to the emergence of multi drug resistant (MDR) TB. Autophagy is a cell-host defense process. Previous studies have reported that autophagy-activating agents eliminate intracellular MDR MTB. Thus, combining a direct antibiotic activity against circulating bacteria with autophagy activation to eliminate bacteria residing inside cells could treat MDR TB. We show that the synthetic peptide, IP-1 (KFLNRFWHWLQLKPGQPMY), induced autophagy in HEK293T cells and macrophages at a low dose (10 µM), while increasing the dose (50 µM) induced cell death; IP-1 induced the secretion of TNFα in macrophages and killed Mtb at a dose where macrophages are not killed by IP-1. Moreover, IP-1 showed significant therapeutic activity in a mice model of progressive pulmonary TB. In terms of the mechanism of action, IP-1 sequesters ATP in vitro and inside living cells. Thus, IP-1 is the first antimicrobial peptide that eliminates MDR MTB infection by combining four activities: reducing ATP levels, bactericidal activity, autophagy activation, and TNFα secretion.

Relevant Features of Polypharmacologic Human-Target Antimicrobials Discovered by Machine-Learning Techniques.

Polypharmacologic human-targeted antimicrobials (polyHAM) are potentially useful in the treatment of complex human diseases where the microbiome is important (e.g., diabetes, hypertension). We previously reported a machine-learning approach to identify polyHAM from FDA-approved human targeted drugs using a heterologous approach (training with peptides and non-peptide compounds). Here we discover that polyHAM are more likely to be found among antimicrobials displaying a broad-spectrum antibiotic activity and that topological, but not chemical features, are most informative to classify this activity. A heterologous machine-learning approach was trained with broad-spectrum antimicrobials and tested with human metabolites; these metabolites were labeled as antimicrobials or non-antimicrobials based on a naïve text-mining approach. Human metabolites are not commonly recognized as antimicrobials yet circulate in the human body where microbes are found and our heterologous model was able to classify those with antimicrobial activity. These results provide the basis to develop applications aimed to design human diets that purposely alter metabolic compounds proportions as a way to control human microbiome.

Protein-Protein Interactions Efficiently Modeled by Residue Cluster Classes.

Predicting protein-protein interactions (PPI) represents an important challenge in structural bioinformatics. Current computational methods display different degrees of accuracy when predicting these interactions. Different factors were proposed to help improve these predictions, including choosing the proper descriptors of proteins to represent these interactions, among others. In the current work, we provide a representative protein structure that is amenable to PPI classification using machine learning approaches, referred to as residue cluster classes. Through sampling and optimization, we identified the best algorithm-parameter pair to classify PPI from more than 360 different training sets. We tested these classifiers against PPI datasets that were not included in the training set but shared sequence similarity with proteins in the training set to reproduce the situation of most proteins sharing sequence similarity with others. We identified a model with almost no PPI error (96-99% of correctly classified instances) and showed that residue cluster classes of protein pairs displayed a distinct pattern between positive and negative protein interactions. Our results indicated that residue cluster classes are structural features relevant to model PPI and provide a novel tool to mathematically model the protein structure/function relationship.

An automatic representation of peptides for effective antimicrobial activity classification.

Antimicrobial peptides (AMPs) are a promising alternative to small-molecules-based antibiotics. These peptides are part of most living organisms' innate defense system. In order to computationally identify new AMPs within the peptides these organisms produce, an automatic AMP/non-AMP classifier is required. In order to have an efficient classifier, a set of robust features that can capture what differentiates an AMP from another that is not, has to be selected. However, the number of candidate descriptors is large (in the order of thousands) to allow for an exhaustive search of all possible combinations. Therefore, efficient and effective feature selection techniques are required. In this work, we propose an efficient wrapper technique to solve the feature selection problem for AMPs identification. The method is based on a Genetic Algorithm that uses a variable-length chromosome for representing the selected features and uses an objective function that considers the Mathew Correlation Coefficient and the number of selected features. Computational experiments show that the proposed method can produce competitive results regarding sensitivity, specificity, and MCC. Furthermore, the best classification results are achieved by using only 39 out of 272 molecular descriptors.

Heterologous Machine Learning for the Identification of Antimicrobial Activity in Human-Targeted Drugs.

The emergence of microbes resistant to common antibiotics represent a current treat to human health. It has been recently recognized that non-antibiotic labeled drugs may promote antibiotic-resistance mechanisms in the human microbiome by presenting a secondary antibiotic activity; hence, the development of computer-assisted procedures to identify antibiotic activity in human-targeted compounds may assist in preventing the emergence of resistant microbes. In this regard, it is worth noting that while most antibiotics used to treat human infectious diseases are non-peptidic compounds, most known antimicrobials nowadays are peptides, therefore all computer-based models aimed to predict antimicrobials either use small datasets of non-peptidic compounds rendering predictions with poor reliability or they predict antimicrobial peptides that are not currently used in humans. Here we report a machine-learning-based approach trained to identify gut antimicrobial compounds; a unique aspect of our model is the use of heterologous training sets, in which peptide and non-peptide antimicrobial compounds were used to increase the size of the training data set. Our results show that combining peptide and non-peptide antimicrobial compounds rendered the best classification of gut antimicrobial compounds. Furthermore, this classification model was tested on the latest human-approved drugs expecting to identify antibiotics with broad-spectrum activity and our results show that the model rendered predictions consistent with current knowledge about broad-spectrum antibiotics. Therefore, heterologous machine learning rendered an efficient computational approach to classify antimicrobial compounds.

Systematic Identification of Machine-Learning Models Aimed to Classify Critical Residues for Protein Function from Protein Structure.

Protein structure and protein function should be related, yet the nature of this relationship remains unsolved. Mapping the critical residues for protein function with protein structure features represents an opportunity to explore this relationship, yet two important limitations have precluded a proper analysis of the structure-function relationship of proteins: (i) the lack of a formal definition of what critical residues are and (ii) the lack of a systematic evaluation of methods and protein structure features. To address this problem, here we introduce an index to quantify the protein-function criticality of a residue based on experimental data and a strategy aimed to optimize both, descriptors of protein structure (physicochemical and centrality descriptors) and machine learning algorithms, to minimize the error in the classification of critical residues. We observed that both physicochemical and centrality descriptors of residues effectively relate protein structure and protein function, and that physicochemical descriptors better describe critical residues. We also show that critical residues are better classified when residue criticality is considered as a binary attribute (i.e., residues are considered critical or not critical). Using this binary annotation for critical residues 8 models rendered accurate and non-overlapping classification of critical residues, confirming the multi-factorial character of the structure-function relationship of proteins.

Using Confocal Microscopy and Computational Modeling to Investigate the Cell-Penetrating Properties of Antimicrobial Peptides.

Antimicrobial peptides (AMPs) may display the ability to penetrate cells, which may be relevant for their antibiotic activity. To investigate the relevance of the penetrating activity for the antibiotic activity of AMPs, here we describe a method based on the combined use of confocal microscopy and computational modeling coupled with cell death kinetics.

Effective Design of Multifunctional Peptides by Combining Compatible Functions.

Multifunctionality is a common trait of many natural proteins and peptides, yet the rules to generate such multifunctionality remain unclear. We propose that the rules defining some protein/peptide functions are compatible. To explore this hypothesis, we trained a computational method to predict cell-penetrating peptides at the sequence level and learned that antimicrobial peptides and DNA-binding proteins are compatible with the rules of our predictor. Based on this finding, we expected that designing peptides for CPP activity may render AMP and DNA-binding activities. To test this prediction, we designed peptides that embedded two independent functional domains (nuclear localization and yeast pheromone activity), linked by optimizing their composition to fit the rules characterizing cell-penetrating peptides. These peptides presented effective cell penetration, DNA-binding, pheromone and antimicrobial activities, thus confirming the effectiveness of our computational approach to design multifunctional peptides with potential therapeutic uses. Our computational implementation is available at http://bis.ifc.unam.mx/en/software/dcf.

Autophagy as a target for therapeutic uses of multifunctional peptides.

The emergence of complex diseases is promoting a change from one-target to multitarget drugs and peptides are ideal molecules to fulfill this polypharmacologic role. Here we review current status in the design of polypharmacological peptides aimed to treat complex diseases, focusing on tuberculosis. In this sense, combining multiple activities in single molecules is a two-sided sword, as both positive and negative side effects might arise. These polypharmacologic compounds may be directed to regulate autophagy, a catabolic process that enables cells to eliminate intracellular microbes (xenophagy), such as Mycobacterium tuberculosis (MBT). Here we review some strategies to control MBT infection and propose that a peptide combining both antimicrobial and pro-autophagic activities would have a greater potential to limit MBT infection. This endeavor may complement the knowledge gained in understanding the mechanism of action of antibiotics and may lead to the design of better polypharmacological peptides to treat complex diseases such as tuberculosis.

Machine Learnable Fold Space Representation based on Residue Cluster Classes.

MOTIVATION: Protein fold space is a conceptual framework where all possible protein folds exist and ideas about protein structure, function and evolution may be analyzed. Classification of protein folds in this space is commonly achieved by using similarity indexes and/or machine learning approaches, each with different limitations. RESULTS: We propose a method for constructing a compact vector space model of protein fold space by representing each protein structure by its residues local contacts. We developed an efficient method to statistically test for the separability of points in a space and showed that our protein fold space representation is learnable by any machine-learning algorithm. AVAILABILITY: An API is freely available at https://code.google.com/p/pyrcc/.

Quality control test for sequence-phenotype assignments.

Relating a gene mutation to a phenotype is a common task in different disciplines such as protein biochemistry. In this endeavour, it is common to find false relationships arising from mutations introduced by cells that may be depurated using a phenotypic assay; yet, such phenotypic assays may introduce additional false relationships arising from experimental errors. Here we introduce the use of high-throughput DNA sequencers and statistical analysis aimed to identify incorrect DNA sequence-phenotype assignments and observed that 10-20% of these false assignments are expected in large screenings aimed to identify critical residues for protein function. We further show that this level of incorrect DNA sequence-phenotype assignments may significantly alter our understanding about the structure-function relationship of proteins. We have made available an implementation of our method at http://bis.ifc.unam.mx/en/software/chispas.

Yeast mating and image-based quantification of spatial pattern formation.

Communication between cells is a ubiquitous feature of cell populations and is frequently realized by secretion and detection of signaling molecules. Direct visualization of the resulting complex gradients between secreting and receiving cells is often impossible due to the small size of diffusing molecules and because such visualization requires experimental perturbations such as attachment of fluorescent markers, which can change diffusion properties. We designed a method to estimate such extracellular concentration profiles in vivo by using spatiotemporal mathematical models derived from microscopic analysis. This method is applied to populations of thousands of haploid yeast cells during mating in order to quantify the extracellular distributions of the pheromone α-factor and the activity of the aspartyl protease Bar1. We demonstrate that Bar1 limits the range of the extracellular pheromone signal and is critical in establishing α-factor concentration gradients, which is crucial for effective mating. Moreover, haploid populations of wild type yeast cells, but not BAR1 deletion strains, create a pheromone pattern in which cells differentially grow and mate, with low pheromone regions where cells continue to bud and regions with higher pheromone levels and gradients where cells conjugate to form diploids. However, this effect seems to be exclusive to high-density cultures. Our results show a new role of Bar1 protease regulating the pheromone distribution within larger populations and not only locally inside an ascus or among few cells. As a consequence, wild type populations have not only higher mating efficiency, but also higher growth rates than mixed MATa bar1Δ/MATα cultures. We provide an explanation of how a rapidly diffusing molecule can be exploited by cells to provide spatial information that divides the population into different transcriptional programs and phenotypes.

Cell penetrating peptides and cationic antibacterial peptides: two sides of the same coin.

Cell penetrating peptides (CPP) and cationic antibacterial peptides (CAP) have similar physicochemical properties and yet it is not understood how such similar peptides display different activities. To address this question, we used Iztli peptide 1 (IP-1) because it has both CPP and CAP activities. Combining experimental and computational modeling of the internalization of IP-1, we show it is not internalized by receptor-mediated endocytosis, yet it permeates into many different cell types, including fungi and human cells. We also show that IP-1 makes pores in the presence of high electrical potential at the membrane, such as those found in bacteria and mitochondria. These results provide the basis to understand the functional redundancy of CPPs and CAPs.

Moonlighting peptides with emerging function.

Hunter-killer peptides combine two activities in a single polypeptide that work in an independent fashion like many other multi-functional, multi-domain proteins. We hypothesize that emergent functions may result from the combination of two or more activities in a single protein domain and that could be a mechanism selected in nature to form moonlighting proteins. We designed moonlighting peptides using the two mechanisms proposed to be involved in the evolution of such molecules (i.e., to mutate non-functional residues and the use of natively unfolded peptides). We observed that our moonlighting peptides exhibited two activities that together rendered a new function that induces cell death in yeast. Thus, we propose that moonlighting in proteins promotes emergent properties providing a further level of complexity in living organisms so far unappreciated.

An FPGA implementation to detect selective cationic antibacterial peptides.

Exhaustive prediction of physicochemical properties of peptide sequences is used in different areas of biological research. One example is the identification of selective cationic antibacterial peptides (SCAPs), which may be used in the treatment of different diseases. Due to the discrete nature of peptide sequences, the physicochemical properties calculation is considered a high-performance computing problem. A competitive solution for this class of problems is to embed algorithms into dedicated hardware. In the present work we present the adaptation, design and implementation of an algorithm for SCAPs prediction into a Field Programmable Gate Array (FPGA) platform. Four physicochemical properties codes useful in the identification of peptide sequences with potential selective antibacterial activity were implemented into an FPGA board. The speed-up gained in a single-copy implementation was up to 108 times compared with a single Intel processor cycle for cycle. The inherent scalability of our design allows for replication of this code into multiple FPGA cards and consequently improvements in speed are possible. Our results show the first embedded SCAPs prediction solution described and constitutes the grounds to efficiently perform the exhaustive analysis of the sequence-physicochemical properties relationship of peptides.