Design of active analogues of a 15-residue peptide using D-optimal design, QSAR and a combinatorial search algorithm

Design, Screening, and Testing of Non-Rational Peptide Libraries with Antimicrobial Activity: In Silico and Experimental Approaches

One of the challenges of modern biotechnology is to find new routes to mitigate the resistance to conventional antibiotics. Antimicrobial peptides (AMPs) are an alternative type of biomolecules, naturally present in a wide variety of organisms, with the capacity to overcome the current microorganism resistance threat. Here, we reviewed our recent efforts to develop a new library of non-rationally produced AMPs that relies on bacterial genome inherent diversity and compared it with rationally designed libraries. Our approach is based on a four-stage workflow process that incorporates the interplay of recent developments in four major emerging technologies: artificial intelligence, molecular dynamics, surface-display in microorganisms, and microfluidics. Implementing this framework is challenging because to obtain reliable results, the in silico algorithms to search for candidate AMPs need to overcome issues of the state-of-the-art approaches that limit the possibilities for multi-space data distribution analyses in extremely large databases. We expect to tackle this challenge by using a recently developed classification algorithm based on deep learning models that rely on convolutional layers and gated recurrent units. This will be complemented by carefully tailored molecular dynamics simulations to elucidate specific interactions with lipid bilayers. Candidate AMPs will be recombinantly-expressed on the surface of microorganisms for further screening via different droplet-based microfluidic-based strategies to identify AMPs with the desired lytic abilities. We believe that the proposed approach opens opportunities for searching and screening bioactive peptides for other applications.

How Melittin Inserts into Cell Membrane: Conformational Changes, Inter-Peptide Cooperation, and Disturbance on the Membrane

Antimicrobial peptides (AMPs), as a key component of the immune defense systems of organisms, are a promising solution to the serious threat of drug-resistant bacteria to public health. As one of the most representative and extensively studied AMPs, melittin has exceptional broad-spectrum activities against microorganisms, including both Gram-positive and Gram-negative bacteria. Unfortunately, the action mechanism of melittin with bacterial membranes, especially the underlying physics of peptide-induced membrane poration behaviors, is still poorly understood, which hampers efforts to develop melittin-based drugs or agents for clinical applications. In this mini-review, we focus on recent advances with respect to the membrane insertion behavior of melittin mostly from a computational aspect. Membrane insertion is a prerequisite and key step for forming transmembrane pores and bacterial killing by melittin, whose occurrence is based on overcoming a high free-energy barrier during the transition of melittin molecules from a membrane surface-binding state to a transmembrane-inserting state. Here, intriguing simulation results on such transition are highlighted from both kinetic and thermodynamic aspects. The conformational changes and inter-peptide cooperation of melittin molecules, as well as melittin-induced disturbances to membrane structure, such as deformation and lipid extraction, are regarded as key factors influencing the insertion of peptides into membranes. The associated intermediate states in peptide conformations, lipid arrangements, membrane structure, and mechanical properties during this process are specifically discussed. Finally, potential strategies for enhancing the poration ability and improving the antimicrobial performance of AMPs are included as well.

Machine learning assisted design of highly active peptides for drug discovery

The discovery of peptides possessing high biological activity is very challenging due to the enormous diversity for which only a minority have the desired properties. To lower cost and reduce the time to obtain promising peptides, machine learning approaches can greatly assist in the process and even partly replace expensive laboratory experiments by learning a predictor with existing data or with a smaller amount of data generation. Unfortunately, once the model is learned, selecting peptides having the greatest predicted bioactivity often requires a prohibitive amount of computational time. For this combinatorial problem, heuristics and stochastic optimization methods are not guaranteed to find adequate solutions. We focused on recent advances in kernel methods and machine learning to learn a predictive model with proven success. For this type of model, we propose an efficient algorithm based on graph theory, that is guaranteed to find the peptides for which the model predicts maximal bioactivity. We also present a second algorithm capable of sorting the peptides of maximal bioactivity. Extensive analyses demonstrate how these algorithms can be part of an iterative combinatorial chemistry procedure to speed up the discovery and the validation of peptide leads. Moreover, the proposed approach does not require the use of known ligands for the target protein since it can leverage recent multi-target machine learning predictors where ligands for similar targets can serve as initial training data. Finally, we validated the proposed approach in vitro with the discovery of new cationic antimicrobial peptides. Source code freely available at http://graal.ift.ulaval.ca/peptide-design/.

Part of the complexity of drug discovery is the sheer chemical diversity to explore combined to all requirements a compound must meet to become a commercial drug. Hence, it makes sense to automate this chemical exploration endeavor in a wise, informed, and efficient fashion. Here, we focused on peptides as they have properties that make them excellent drug starting points. Machine learning techniques may replace expensive in-vitro laboratory experiments by learning an accurate model of it. However, computational models also suffer from the combinatorial explosion due to the enormous chemical diversity. Indeed, applying the model to every peptides would take an astronomical amount of computer time. Therefore, given a model, is it possible to determine, using reasonable computational time, the peptide that has the best properties and chance for success? This exact question is what motivated our work. We focused on recent advances in kernel methods and machine learning to learn a model that already had excellent results. We demonstrate that this class of model has mathematical properties that makes it possible to rapidly identify and sort the best peptides. Finally, in-vitro and in-silico results are provided to support and validate this theoretical discovery.

Differentiating antimicrobial peptides interacting with lipid bilayer: Molecular signatures derived from quartz crystal microbalance with dissipation monitoring

Many antimicrobial peptides (AMPs) kill bacteria by disrupting the lipid bilayer structure of their inner membrane. However, there is only limited quantitative information in the literature to differentiate between AMPs of differing molecular properties, in terms of how they interact with the membrane. In this study, we have used quartz crystal microbalance with dissipation monitoring (QCM-D) to probe the interactions between a supported bilayer membrane of egg phosphatidylcholine (egg PC) and four structurally different AMPs: alamethicin, chrysophsin-3, indolicidin, and sheep myeloid antimicrobial peptide (SMAP-29). Multiple signatures from the QCM-D measurements were extracted, differentiating the AMPs, that provide information on peptide addition to and lipid removal from the membrane, the dynamics of peptide-membrane interactions and the rates at which the peptide actions are initiated. The mechanistic variations in peptide action were related to the fundamental structural properties of the peptides including the hydrophobicity, hydrophobic moment, and the probability of α-helical secondary structures.

QSAR modeling: where have you been? Where are you going to?

Quantitative structure-activity relationship modeling is one of the major computational tools employed in medicinal chemistry. However, throughout its entire history it has drawn both praise and criticism concerning its reliability, limitations, successes, and failures. In this paper, we discuss (i) the development and evolution of QSAR; (ii) the current trends, unsolved problems, and pressing challenges; and (iii) several novel and emerging applications of QSAR modeling. Throughout this discussion, we provide guidelines for QSAR development, validation, and application, which are summarized in best practices for building rigorously validated and externally predictive QSAR models. We hope that this Perspective will help communications between computational and experimental chemists toward collaborative development and use of QSAR models. We also believe that the guidelines presented here will help journal editors and reviewers apply more stringent scientific standards to manuscripts reporting new QSAR studies, as well as encourage the use of high quality, validated QSARs for regulatory decision making.

Methods for building quantitative structure-activity relationship (QSAR) descriptors and predictive models for computer-aided design of antimicrobial peptides

Antimicrobial peptides are ubiquitous in nature where they play important roles in host defense and microbial control. More than 1,000 naturally occurring peptides have been described so far and those considered for pharmaceutical development have all been further optimized by rational approaches. In recent years, high-throughput screening assays have been developed to specifically address optimization of AMPs. In addition to these cell-based in vivo systems, a range of computational in silico systems can be applied in order to predict the biological activity of AMPs for specific bacteria. Among them, quantitative structure-activity relationships (QSARs) method, which attempts to correlate chemical structure to biological measurement, has shown promising results in the optimization and discovery of peptide candidates. Therefore, this chapter is devoted to describe the QSAR method and recent progress applied in AMP.

De novo generation of antimicrobial LK peptides with a single tryptophan at the critical amphipathic interface

De novo design of amphipathic model peptides has been successful for generating many antimicrobial peptides with various lengths and amino acid compositions. Here, we suggest a very simple strategy to design antimicrobial peptides with a short length and a simple amino acid composition. Amphipathic helical properties were conferred by using only leucines and lysines and a single tryptophan was positioned at the critical amphipathic interface between the hydrophilic ending side and the hydrophobic starting side, in the helical wheel projection. According to this rule, the model peptides with 7 to 13 residues exhibited antimicrobial activity. Among them, the most potent activity against both Gram-positive and Gram-negative bacteria, covering all of the nine bacterial strains tested in this study, was found for the 11-mer sequences having a 1:1 (L(5)K(5)W(6)) or a 3:2 (L(6)K(4)W(6)) ratio of leucines to lysines. In particular, the former peptide L(5)K(5)W(6) could be evaluated as the most useful agent, as it showed no significant hemolytic activity with a broad-spectrum of antimicrobial activity.

Enzyme improvement in the absence of structural knowledge: a novel statistical approach

Most existing methods for improving protein activity are laborious and costly, as they either require knowledge of protein structure or involve expression and screening of a vast number of protein mutants. We describe here a successful first application of a novel approach, which requires no structural knowledge and is shown to significantly reduce the number of mutants that need to be screened. In the first phase of this study, around 7000 mutants were screened through standard directed evolution, yielding a 230-fold improvement in activity relative to the wild type. Using sequence analysis and site-directed mutagenesis, an additional single mutant was then produced, with 500-fold improved activity. In the second phase, a novel statistical method for protein improvement was used; building on data from the first phase, only 11 targeted additional mutants were produced through site-directed mutagenesis, and the best among them achieved a >1500-fold improvement in activity over the wild type. Thus, the statistical model underlying the experiment was validated, and its predictions were shown to reduce laboratory labor and resources.

Better library design: data-driven protein engineering

Data-driven protein engineering is increasingly used as an alternative to rational design and combinatorial engineering because it uses available knowledge to limit library size, while still allowing for the identification of unpredictable substitutions that lead to large effects. Recent advances in computational modeling and bioinformatics, as well as an increasing databank of experiments on functional variants, have led to new strategies to choose particular amino acid residues to vary in order to increase the chances of obtaining a variant protein with the desired property. Strategies for limiting diversity at each position, design of small sub-libraries, and the performance of scouting experiments, have also been developed or even automated, further reducing the library size.

Engineering proteinase K using machine learning and synthetic genes

Background

Altering a protein's function by changing its sequence allows natural proteins to be converted into useful molecular tools. Current protein engineering methods are limited by a lack of high throughput physical or computational tests that can accurately predict protein activity under conditions relevant to its final application. Here we describe a new synthetic biology approach to protein engineering that avoids these limitations by combining high throughput gene synthesis with machine learning-based design algorithms.

Results

We selected 24 amino acid substitutions to make in proteinase K from alignments of homologous sequences. We then designed and synthesized 59 specific proteinase K variants containing different combinations of the selected substitutions. The 59 variants were tested for their ability to hydrolyze a tetrapeptide substrate after the enzyme was first heated to 68°C for 5 minutes. Sequence and activity data was analyzed using machine learning algorithms. This analysis was used to design a new set of variants predicted to have increased activity over the training set, that were then synthesized and tested. By performing two cycles of machine learning analysis and variant design we obtained 20-fold improved proteinase K variants while only testing a total of 95 variant enzymes.

Conclusion

The number of protein variants that must be tested to obtain significant functional improvements determines the type of tests that can be performed. Protein engineers wishing to modify the property of a protein to shrink tumours or catalyze chemical reactions under industrial conditions have until now been forced to accept high throughput surrogate screens to measure protein properties that they hope will correlate with the functionalities that they intend to modify. By reducing the number of variants that must be tested to fewer than 100, machine learning algorithms make it possible to use more complex and expensive tests so that only protein properties that are directly relevant to the desired application need to be measured. Protein design algorithms that only require the testing of a small number of variants represent a significant step towards a generic, resource-optimized protein engineering process.

Improving catalytic function by ProSAR-driven enzyme evolution

We describe a directed evolution approach that should find broad application in generating enzymes that meet predefined process-design criteria. It augments recombination-based directed evolution by incorporating a strategy for statistical analysis of protein sequence activity relationships (ProSAR). This combination facilitates mutation-oriented enzyme optimization by permitting the capture of additional information contained in the sequence-activity data. The method thus enables identification of beneficial mutations even in variants with reduced function. We use this hybrid approach to evolve a bacterial halohydrin dehalogenase that improves the volumetric productivity of a cyanation process approximately 4,000-fold. This improvement was required to meet the practical design criteria for a commercially relevant biocatalytic process involved in the synthesis of a cholesterol-lowering drug, atorvastatin (Lipitor), and was obtained by variants that had at least 35 mutations.

Directed molecular evolution by machine learning and the influence of nonlinear interactions

Alternative search strategies for the directed evolution of proteins are presented and compared with each other. In particular, two different machine learning strategies based on partial least-squares regression are developed: the first contains only linear terms that represent a given residue's independent contribution to fitness, the second contains additional nonlinear terms to account for potential epistatic coupling between residues. The nonlinear modeling strategy is further divided into two types, one that contains all possible nonlinear terms and another that makes use of a genetic algorithm to select a subset of important interaction terms. The performance of each modeling type as a function of training set size is analysed. Simulated molecular evolution on a synthetic protein landscape shows the use of machine learning techniques to guide library design can be a powerful addition to library generation methods such as DNA shuffling.

Application of 'inductive' QSAR descriptors for quantification of antibacterial activity of cationic polypeptides

On the basis of the inductive QSAR descriptors we have created a neural network-based solution enabling quantification of antibacterial activity in the series of 101 synthetic cationic polypeptides (CAMEL-s). The developed QSAR model allowed 80% correct categorical classification of antibacterial potencies of the CAMEL-s both in the training and the validation sets. The accuracy of the activity predictions demonstrates that a narrow set of 3D sensitive 'inductive' descriptors can adequately describe the aspects of intra- and intermolecular interactions that are relevant for antibacterial activity of the cationic polypeptides. The developed approach can be further expanded for the larger sets of biologically active peptides and can serve as a useful quantitative tool for rational antibiotic design and discovery.

Putting engineering back into protein engineering: bioinformatic approaches to catalyst design

Complex multivariate engineering problems are commonplace and not unique to protein engineering. Mathematical and data-mining tools developed in other fields of engineering have now been applied to analyze sequence-activity relationships of peptides and proteins and to assist in the design of proteins and peptides with specified properties. Decreasing costs of DNA sequencing in conjunction with methods to quickly synthesize statistically representative sets of proteins allow modern heuristic statistics to be applied to protein engineering. This provides an alternative approach to expensive assays or unreliable high-throughput surrogate screens.

Dynorphin A inhibits nociceptin-converting enzyme from the rat spinal cord

Cysteine proteinase found in the spinal cord of rat, called nociceptin-converting enzyme (NCE), is competitively inhibited by dynorphin A and its fragment des-[Tyr(1)]-DYN A. This proteinase converts orphanin FQ/nociceptin (OFQ/N) to two major fragments: OFQ/N(1-11) and further OFQ/N(1-6) with analgesic properties. Dynorphin A at the concentration of 10 microM increases K(M) from 15.0 to 55.9 microM. The calculated K(i) for this interaction was estimated at 3.7 microM. This observation may suggest an interaction between opioid and nociceptive systems which may be affected by the balance between opioid and antiopioid systems. This balance between particular OFQ/N sequences that are derived from the same precursor and regulated by proteinases may play an important role in pain. Interestingly, dynorphin B does not reveal a similar action on the NCE.

Exploration of sequence space for protein engineering

The process of protein engineering is currently evolving towards a heuristic understanding of the sequence-function relationship. Improved DNA sequencing capacity, efficient protein function characterization and improved quality of data points in conjunction with well-established statistical tools from other industries are changing the protein engineering field. Algorithms capturing the heuristic sequence-function relationships will have a drastic impact on the field of protein engineering. In this review, several alternative approaches to quantitatively assess sequence space are discussed and the relatively few examples of wet-lab validation of statistical sequence-function characterization/correlation are described.

Amphipathic, alpha-helical antimicrobial peptides

Gene-encoded antimicrobial peptides are an important component of host defense in animals ranging from insects to mammals. They do not target specific molecular receptors on the microbial surface, but rather assume amphipathic structures that allow them to interact directly with microbial membranes, which they can rapidly permeabilize. They are thus perceived to be one promising solution to the growing problem of microbial resistance to conventional antibiotics. A particularly abundant and widespread class of antimicrobial peptides are those with amphipathic, alpha-helical domains. Due to their relatively small size and synthetic accessibility, these peptides have been extensively studied and have generated a substantial amount of structure-activity relationship (SAR) data. In this review, alpha-helical antimicrobial peptides are considered from the point of view of six interrelated structural and physicochemical parameters that modulate their activity and specificity: sequence, size, structuring, charge, amphipathicity, and hydrophobicity. It begins by providing an overview of how these vary in peptides from different natural sources. It then analyzes how they relate to the currently accepted model for the mode of action of alpha-helical peptides, and discusses what the numerous SAR studies that have been carried out on these compounds and their analogues can tell us. A comparative analysis of the many alpha-helical, antimicrobial peptide sequences that are now available then provides further information on how these parameters are distributed and interrelated. Finally, the systematic variation of parameters in short model peptides is used to throw light on their role in antimicrobial potency and specificity. The review concludes with some considerations on the potentials and limitations for the development of alpha-helical, antimicrobial peptides as antiinfective agents.

Activities of synthetic hybrid peptides against anaerobic bacteria: aspects of methodology and stability

The increasing problem of antibiotic resistance among pathogenic bacteria requires development of new antimicrobial agents. One line of investigation is the synthesis of antimicrobial hybrid peptides. The aim of the present investigation was to determine the in vitro activities of 16 cecropin-melittin hybrid peptides (CAMEL analogues) against 60 anaerobic bacterial strains, to compare their activities with those of seven clinically used antimicrobial agents, and to compare different methods for anaerobic susceptibility testing of these peptides. The stability of one of the peptides, temporin B, with different stereoisomeric configurations was investigated in a fecal milieu. The CAMEL analogues showed antimicrobial activity against the anaerobic bacteria, with MICs ranging from 0.125 to 32 microg/ml. The overall activities (the MICs at which 90% of isolates are inhibited) of the CAMEL analogues against anaerobic bacteria were mainly inferior to those of imipenem, clindamycin, and piperacillin but were equal to or superior to those of metronidazole, cefoxitin, ciprofloxacin, and chloramphenicol. The agarose dilution method was found to be an accurate method for the testing of large numbers of bacterial strains. The D isomer of temporin B was inactivated more slowly in feces than the L isomer. This study shows that the CAMEL analogues are potential agents for the treatment of anaerobic infections.

Structural features of helical antimicrobial peptides: their potential to modulate activity on model membranes and biological cells

Antibacterial, membrane-lytic peptides belong to the innate immune system and host defense mechanism of a multitude of animals and plants. The largest group of peptide antibiotics comprises peptides which fold into an amphipathic alpha-helical conformation when interacting with the target. The activity of these peptides is thought to be determined by global structural parameters rather than by the specific amino acid sequence. This review is concerned with the influence of structural parameters, such as peptide helicity, hydrophobicity, hydrophobic moment, peptide charge and the size of the hydrophobic/hydrophilic domain, on membrane activity and selectivity. The potential of these parameters to increase the antibacterial activity and to improve the prokaryotic selectivity of natural and model peptides is assessed. Furthermore, biophysical studies are summarized which elucidated the molecular basis for activity and selectivity modulations on the level of model membranes. Finally, the knowledge about the role of peptide structural parameters is applied to understand the different activity spectra of natural membrane-lytic peptides.

Statistical molecular design of peptoid libraries

Statistical experimental design provides an efficient approach for selecting the building blocks to span the structural space and increase the information content in a combinatorial library. A set of renin-inhibitors, hexapeptoids, is used to illustrate the approach. Multivariate quantitative structure-activity relationships (MQSARs) were developed relating renin inhibition to the peptoid sequences variation, parametrized by the z-scales. By using the information from the models, the number of building block sets could be reduced from six to three. Using a statistical molecular design (SMD) reduces the number of compounds from more than 100,000 down to 90. A second SMD was used for comparison, based on less prior knowledge. This gave a reduction from over 2 billion to 120 compounds.

A validation study of molecular descriptors for the rational design of peptide libraries

Important molecular descriptors used for establishing quantitative structure-activity relationships are investigated to classify similar versus dissimilar peptides. When searching new lead structures, synthesizing and testing compounds which are too similar wastes time and resources. In contrast, any lead optimization program requires the investigation of similar compounds to that lead. Thus, it is important to maximize or minimize the structural diversity of peptides to design useful compound libraries for lead finding or lead refinement projects. If a molecular descriptor is a useful measure of similarity for the design of peptide libraries, small differences in this descriptor for a pair of molecules should only translate into small biological differences. Using this paradigm as a basis for descriptor validation, it was possible to rank different molecular descriptors. Those physicochemical descriptors are 2D fingerprints and five experimentally or theoretically derived principal property scales. Some theoretically derived metrics are obtained by computing interaction energies or similarity indices on predefined 3D grid points using canonical conformations for individual amino acids. The resulting 3D data matrices are analyzed using a principal component analysis leading to three principal properties for CoMFA (Comparative Molecular Field Analysis) or CoMSIA (Comparative Molecular Similarity Index Analysis) derived molecular fields. The descriptor validation results reveal the applicability of design tools on peptide data sets. Experimentally derived descriptors, in general, are more acceptable than computationally derived metrics, while the latter provide a statistically valid alternative to characterize novel building blocks. The CoMSIA metrics perform slightly better than the CoMFA-based principal properties, while GRID-based descriptors are always less acceptable.

Antianaerobic activity of a cecropin---melittin peptide

OBJECTIVE: Several small, 15-residue peptides that contain portions of the amino acid sequences of both cecropin A and melittin have previously been shown to have broad-spectrum antibacterial activities against aerobic microorganisms, with no undesirable hemolytic properties. It would also be useful to know what effect these hybrid peptides have on anaerobic bacteria. METHODS: The minimum inhibitory concentrations of one hybrid, CA(1--7)M(2--9)NH2, were compared with those of seven other antimicrobial agents against 111 clinical anaerobic strains; Bacteroides fragilis, 24 strains; other Bacteroides fragilis group, 14 strains; other Bacteroides species, 13 strains; Fusobacterium nucleatum, six strains; Clostridium difficile, 22 strains; Clostridium perfringens, 10 strains, Propionibacterium spp., nine strains; and anaerobic cocci, 13 strains. RESULTS: Ninety per cent of strains belonging to the B. fragilis group, fusobacteria, propionibacteria and peptostreptococci were inhibited by 4 mg/L CA(1--7)M(2--9)NH2, and the antimicrobial activity was approximately in the same range as that of chloramphenicol. CONCLUSION: This investigation showed that the antimicrobial spectrum of this cecropin---melittin hybrid also includes anaerobic organisms.