Towards the automatic design of synthetically accessible protein ligands: peptides, amides and peptidomimetics

Non-Canonical Amino Acids in Analyses of Protease Structure and Function

All known organisms encode 20 canonical amino acids by base triplets in the genetic code. The cellular translational machinery produces proteins consisting mainly of these amino acids. Several hundred natural amino acids serve important functions in metabolism, as scaffold molecules, and in signal transduction. New side chains are generated mainly by post-translational modifications, while others have altered backbones, such as the β- or γ-amino acids, or they undergo stereochemical inversion, e.g., in the case of D-amino acids. In addition, the number of non-canonical amino acids has further increased by chemical syntheses. Since many of these non-canonical amino acids confer resistance to proteolytic degradation, they are potential protease inhibitors and tools for specificity profiling studies in substrate optimization and enzyme inhibition. Other applications include in vitro and in vivo studies of enzyme kinetics, molecular interactions and bioimaging, to name a few. Amino acids with bio-orthogonal labels are particularly attractive, enabling various cross-link and click reactions for structure-functional studies. Here, we cover the latest developments in protease research with non-canonical amino acids, which opens up a great potential, e.g., for novel prodrugs activated by proteases or for other pharmaceutical compounds, some of which have already reached the clinical trial stage.

Application of Machine Learning Techniques to Predict Binding Affinity for Drug Targets: A Study of Cyclin-Dependent Kinase 2

BACKGROUND

The elucidation of the structure of cyclin-dependent kinase 2 (CDK2) made it possible to develop targeted scoring functions for virtual screening aimed to identify new inhibitors for this enzyme. CDK2 is a protein target for the development of drugs intended to modulate cellcycle progression and control. Such drugs have potential anticancer activities.

OBJECTIVE

Our goal here is to review recent applications of machine learning methods to predict ligand- binding affinity for protein targets. To assess the predictive performance of classical scoring functions and targeted scoring functions, we focused our analysis on CDK2 structures.

METHODS

We have experimental structural data for hundreds of binary complexes of CDK2 with different ligands, many of them with inhibition constant information. We investigate here computational methods to calculate the binding affinity of CDK2 through classical scoring functions and machine- learning models.

RESULTS

Analysis of the predictive performance of classical scoring functions available in docking programs such as Molegro Virtual Docker, AutoDock4, and Autodock Vina indicated that these methods failed to predict binding affinity with significant correlation with experimental data. Targeted scoring functions developed through supervised machine learning techniques showed a significant correlation with experimental data.

CONCLUSION

Here, we described the application of supervised machine learning techniques to generate a scoring function to predict binding affinity. Machine learning models showed superior predictive performance when compared with classical scoring functions. Analysis of the computational models obtained through machine learning could capture essential structural features responsible for binding affinity against CDK2.

How Docking Programs Work

Protein-ligand docking simulations are of central interest for computer-aided drug design. Docking is also of pivotal importance to understand the structural basis for protein-ligand binding affinity. In the last decades, we have seen an explosion in the number of three-dimensional structures of protein-ligand complexes available at the Protein Data Bank. These structures gave further support for the development and validation of in silico approaches to address the binding of small molecules to proteins. As a result, we have now dozens of open source programs and web servers to carry out molecular docking simulations. The development of the docking programs and the success of such simulations called the attention of a broad spectrum of researchers not necessarily familiar with computer simulations. In this scenario, it is essential for those involved in experimental studies of protein-ligand interactions and biophysical techniques to have a glimpse of the basics of the protein-ligand docking simulations. Applications of protein-ligand docking simulations to drug development and discovery were able to identify hits, inhibitors, and even drugs. In the present chapter, we cover the fundamental ideas behind protein-ligand docking programs for non-specialists, which may benefit from such knowledge when studying molecular recognition mechanism.

Electrostatic Energy in Protein-Ligand Complexes

Computational analysis of protein-ligand interactions is of pivotal importance for drug design. Assessment of ligand binding energy allows us to have a glimpse of the potential of a small organic molecule as a ligand to the binding site of a protein target. Considering scoring functions available in docking programs such as AutoDock4, AutoDock Vina, and Molegro Virtual Docker, we could say that they all rely on equations that sum each type of protein-ligand interactions to model the binding affinity. Most of the scoring functions consider electrostatic interactions involving the protein and the ligand. In this chapter, we present the main physics concepts necessary to understand electrostatics interactions relevant to molecular recognition of a ligand by the binding pocket of a protein target. Moreover, we analyze the electrostatic potential energy for an ensemble of structures to highlight the main features related to the importance of this interaction for binding affinity.

Machine Learning to Predict Binding Affinity

Recent progress in the development of scientific libraries with machine-learning techniques paved the way for the implementation of integrated computational tools to predict ligand-binding affinity. The prediction of binding affinity uses the atomic coordinates of protein-ligand complexes. These new computational tools made application of a broad spectrum of machine-learning techniques to study protein-ligand interactions possible. The essential aspect of these machine-learning approaches is to train a new computational model by using technologies such as supervised machine-learning techniques, convolutional neural network, and random forest to mention the most commonly applied methods. In this chapter, we focus on supervised machine-learning techniques and their applications in the development of protein-targeted scoring functions for the prediction of binding affinity. We discuss the development of the program SAnDReS and its application to the creation of machine-learning models to predict inhibition of cyclin-dependent kinase and HIV-1 protease. Moreover, we describe the scoring function space, and how to use it to explain the development of targeted scoring functions.

Computer-aided design of amino acid-based therapeutics: a review

During the last two decades, the pharmaceutical industry has progressed from detecting small molecules to designing biologic-based therapeutics. Amino acid-based drugs are a group of biologic-based therapeutics that can effectively combat the diseases caused by drug resistance or molecular deficiency. Computational techniques play a key role to design and develop the amino acid-based therapeutics such as proteins, peptides and peptidomimetics. In this study, it was attempted to discuss the various elements for computational design of amino acid-based therapeutics. Protein design seeks to identify the properties of amino acid sequences that fold to predetermined structures with desirable structural and functional characteristics. Peptide drugs occupy a middle space between proteins and small molecules and it is hoped that they can target "undruggable" intracellular protein-protein interactions. Peptidomimetics, the compounds that mimic the biologic characteristics of peptides, present refined pharmacokinetic properties compared to the original peptides. Here, the elaborated techniques that are developed to characterize the amino acid sequences consistent with a specific structure and allow protein design are discussed. Moreover, the key principles and recent advances in currently introduced computational techniques for rational peptide design are spotlighted. The most advanced computational techniques developed to design novel peptidomimetics are also summarized.

De novo design: balancing novelty and confined chemical space

IMPORTANCE OF THE FIELD

De novo drug design serves as a tool for the discovery of new ligands for macromolecular targets as well as optimization of known ligands. Recently developed tools aim to address the multi-objective nature of drug design in an unprecedented manner.

AREAS COVERED IN THIS REVIEW

This article discusses recent advances in de novo drug design programs and accessory programs used to evaluate compounds post-generation.

WHAT THE READER WILL GAIN

The reader is introduced to the challenges inherent in de novo drug design and will become familiar with current trends in de novo design. Furthermore, the reader will be better prepared to assess the value of a tool, and be equipped to design more elegant tools in the future.

TAKE HOME MESSAGE

De novo drug design can assist in the efficient discovery of new compounds with a high affinity for a given target. The inclusion of existing chemoinformatic methods with current structure-based de novo design tools provides a means of enhancing the therapeutic value of these generated compounds.

Computer-aided analysis of the interactions of glutamine synthetase with its inhibitors

Mechanism of inhibition of glutamine synthetase (EC 6.3.1.2; GS) by phosphinothricin and its analogues was studied in some detail using molecular modeling methods. Among three possible conformations of phosphinothricin in the active site of GS, this compatible with binding mode of methionine sulfoximine, determined recently by crystallography, was found to be energetically favored. Basing on these results eleven inhibitors of GS were docked into its active site. Taking into consideration that phosphinothricin acts as suicide inhibitor, which is due to phosphorylation by the enzyme, seven of studied analogues were additionally analyzed in their phosphorylated forms. All the inhibitor-enzyme complexes were evaluated quantitatively by using eight scoring functions implemented in Insight and Sybyl program packages and significant correlation between the obtained scores and experimental pK(i) values was achieved. Computed surface charge distribution for five selected inhibitors in both free and phosphorylated forms and their comparison with electronic structure of enzymatic reaction transition state allowed us to determine important electronic features required to construct potent inhibitors of glutamine synthetase.

"De novo" design of peptides with specific lipid-binding properties

In this study, we describe an in silico method to design peptides that can be made of non-natural amino acids and elicit specific membrane-interacting properties. The originality of the method holds in the capacities developed to design peptides from any non-natural amino acids as easily as from natural ones, and to test the structure stability by an angular dynamics rather than the currently-used molecular dynamics. The goal of this study was to design a non-natural tilted peptide. Tilted peptides are short protein fragments able to destabilize lipid membranes and characterized by an asymmetric distribution of hydrophobic residues along their helix structure axis. The method is based on the random generation of peptides and their selection on three main criteria: mean hydrophobicity and the presence of at least one polar residue; tilted insertion at the level of the acyl chains of lipids of a membrane; and conformational stability in that hydrophobic phase. From 10,000,000 randomly-generated peptides, four met all the criteria. One was synthesized and tested for its lipid-destabilizing properties. Biophysical assays showed that the "de novo" peptide made of non-natural amino acids is helical either in solution or into lipids as tested by Fourier transform infrared spectroscopy and is able to induce liposome fusion. These results are in agreement with the calculations and validate the theoretical approach.

The process of structure-based drug design

The field of structure-based drug design is a rapidly growing area in which many successes have occurred in recent years. The explosion of genomic, proteomic, and structural information has provided hundreds of new targets and opportunities for future drug lead discovery. This review summarizes the process of structure-based drug design and includes, primarily, the choice of a target, the evaluation of a structure of that target, the pivotal questions to consider in choosing a method for drug lead discovery, and evaluation of the drug leads. Key principles in the field of structure-based drug design will be illustrated through a case study that explores drug design for AmpC beta-lactamase.

Ligand-supported homology modelling of protein binding-sites using knowledge-based potentials

A new approach, MOBILE, is presented that models protein binding-sites including bound ligand molecules as restraints. Initially generated, homology models of the target protein are refined iteratively by including information about bioactive ligands as spatial restraints and optimising the mutual interactions between the ligands and the binding-sites. Thus optimised models can be used for structure-based drug design and virtual screening. In a first step, ligands are docked into an averaged ensemble of crude homology models of the target protein. In the next step, improved homology models are generated, considering explicitly the previously placed ligands by defining restraints between protein and ligand atoms. These restraints are expressed in terms of knowledge-based distance-dependent pair potentials, which were compiled from crystallographically determined protein-ligand complexes. Subsequently, the most favourable models are selected by ranking the interactions between the ligands and the generated pockets using these potentials. Final models are obtained by selecting the best-ranked side-chain conformers from various models, followed by an energy optimisation of the entire complex using a common force-field. Application of the knowledge-based pair potentials proved efficient to restrain the homology modelling process and to score and optimise the modelled protein-ligand complexes. For a test set of 46 protein-ligand complexes, taken from the Protein Data Bank (PDB), the success rate of producing near-native binding-site geometries (rmsd<2.0A) with MODELLER is 70% when the ligand restrains the homology modelling process in its native orientation. Scoring these complexes with the knowledge-based potentials, in 66% of the cases a pose with rmsd <2.0A is found on rank 1. Finally, MOBILE has been applied to two case studies modelling factor Xa based on trypsin and aldose reductase based on aldehyde reductase.

A structure-based design approach for the identification of novel inhibitors: application to an alanine racemase

We report a new structure-based strategy for the identification of novel inhibitors. This approach has been applied to Bacillus stearothermophilus alanine racemase (AlaR), an enzyme implicated in the biosynthesis of the bacterial cell wall. The enzyme catalyzes the racemization of L- and D-alanine using pyridoxal 5'-phosphate (PLP) as a cofactor. The restriction of AlaR to bacteria and some fungi and the absolute requirement for D-alanine in peptidoglycan biosynthesis make alanine racemase a suitable target for drug design. Unfortunately, known inhibitors of alanine racemase are not specific and inhibit the activity of other PLP-dependent enzymes, leading to neurological and other side effects. This article describes the development of a receptor-based pharmacophore model for AllaR, taking into account receptor flexibility (i.e. a 'dynamic' pharmacophore model). In order to accomplish this, molecular dynamics (MD) simulations were performed on the full AlaR dimer from Bacillus stearothermophilus (PDB entry, 1 sft) with a D-alanine molecule in one active site and the non-covalent inhibitor, propionate, in the second active site of this homodimer. The basic strategy followed in this study was to utilize conformations of the protein obtained during MD simulations to generate a dynamic pharmacophore model using the property mapping capability of the LigBuilder program. Compounds from the Available Chemicals Directory that fit the pharmacophore model were identified and have been submitted for experimental testing. The approach described here can be used as a valuable tool for the design of novel inhibitors of other biomolecular targets.

Subsite specificity of trypanosomal cathepsin L-like cysteine proteases. Probing the S2 pocket with phenylalanine-derived amino acids

The S2 subsite of mammalian cysteine proteinases of the papain family is essential for specificity. Among natural amino acids, all these enzymes prefer bulky hydrophobic residues such as phenylalanine at P2. This holds true for their trypanosomal counterparts: cruzain from Trypanosoma cruzi and congopain from T. congolense. A detailed analysis of the S2 specificity of parasitic proteases was performed to gain information that might be of interest for the design of more selective pseudopeptidyl inhibitors. Nonproteogenic phenylalanyl analogs (Xaa) have been introduced into position P2 of fluorogenic substrates dansyl-Xaa-Arg-Ala-Pro-Trp, and their kinetic constants (Km, kcat/Km) have been determined with congopain and cruzain, and related host cathepsins B and L. Trypanosomal cysteine proteases are poorly stereoselective towards D/L-Phe, the inversion of chirality modifying the efficiency of the reaction but not the Km. Congopain binds cyclohexylalanine better than aromatic Phe derivatives. Another characteristic feature of congopain compared to cruzain and cathepsins B and L was that it could accomodate a phenylglycyl residue (kcat/Km = 1300 mM-1.s-1), while lengthening of the side chain by a methylene group only slightly impaired the specificity constant towards trypanosomal cysteine proteases. Mono- and di-halogenation or nitration of Phe did not affect Km for cathepsin L-like enzymes, but the presence of constrained Phe derivatives prevented a correct fitting into the S2 subsite. A model of congopain has been built to study the fit of Phe analogs within the S2 pocket. Phe analogs adopted a positioning within the S2 pocket similar to that of the Tyr of the cruzain/Z-Tyr-Ala-fluoromethylketone complex. However, cyclohexylalanine has an energetically favorable chair-like conformation and can penetrate deeper into the subsite. Fitting of modeled Phe analogs were in good agreement with kinetic parameters. Furthermore, a linear relationship could be established with logP, supporting the suggestion that fitting into the S2 pocket of trypanosomal cysteine proteases depends on the hydrophobicity of Phe analogs.

DREAM++: flexible docking program for virtual combinatorial libraries

We present a set of programs, DREAM+2 (Docking and Reaction programs using Efficient seArch Methods written in C++), for docking computationally generated ligands into macromolecular binding sites. DREAM++ is composed of three programs: ORIENT++, REACT++ and SEARCH++. The program ORIENT++ positions molecules in a binding site with the DOCK algorithm. Its output can be used as input to REACT++ and SEARCH+2. The program REACT++ performs user-specific chemical reactions on a docked molecule, so that reaction products can be evaluated for three dimensional complementarity with the macromolecular site. The program SEARCH++ performs an efficient conformation search on the reaction products using a hybrid backtrack and incremental construction algorithm. We have applied the programs to HIV protease-inhibitor complexes as test systems. We found that we can differentiate high-affinity ligands based on several measures: interaction energies, occupancy of protein subsites and the number of successfully docked conformations for each product. Encouraged by the results in the test case, we applied the programs to propose novel inhibitors of HIV protease. These inhibitors can be generated by organic reactions using commercially available reagents. They are alternatives to the inhibitors synthesized by Glaxo.

The discovery of steroids and other novel FKBP inhibitors using a molecular docking program

The molecular docking computer program SANDOCK was used to screen small molecule three-dimensional databases in the hunt for novel FKBP inhibitors. Spectroscopic measurements confirmed binding of over 20 compounds to the target protein, some with dissociation constants in the low micromolar range. The discovery that FK506 binding protein is a steroid binding protein may be of wider biological significance. Two-dimensional NMR was used to determine the steroid binding mode and confirmed the interactions predicted by the docking program.

Combinatorial docking and combinatorial chemistry: design of potent non-peptide thrombin inhibitors

A computational algorithm was used to design automatically novel thrombin inhibitors that are available from a single-step chemical reaction. The compounds do not contain amide bonds, are achiral and have a molecular weight below 400. Of the 10 compounds that were synthesized, five bind to thrombin with a Ki in the nanomolar range. Subsequent X-ray structure determination of the thrombin-inhibitor complex for the best compound (Ki = 95 nM) confirms the predicted binding mode. The novel algorithm is applicable to a broad range of chemical reactions.

Chance favors the prepared mind--from serendipity to rational drug design

Accidental discoveries always played an important role in science, especially in the search for new drugs. Several examples of serendipitous findings, leading to therapeutically useful drugs, are presented and discussed. Captopril, an antihypertensive Angiotensin-converting enzyme inhibitor, was the first drug that could be derived from a structural model of a protein. Dorzolamide, a Carboanhydrase inhibitor for the treatment of glaucoma, and the HIV protease inhibitors Saquinavir, Indinavir, Ritonavir, and Nelfinavir are further examples of therapeutically used drugs from structure-based design. More enzyme inhibitors, e.g. the anti-influenza drugs Zanamivir and GS 4104, are in clinical development. In the absence of a protein 3D structure, the 3D structures of certain ligands may be used for rational design. This approach is exemplified by the design of specifically acting integrin receptor antagonists. In the last years, combinatorial and computational approaches became important methods for rational drug design. SAR by NMR searches for low-affinity ligands that bind to proximal subsites of an enzyme; linkage with an appropriate tether produces nanomolar inhibitors. The de novo design program LUDI and the docking program FlexX are tools for the computer-aided design of protein ligands. Work is in progress to combine such approaches to strategies for combinatorial drug design.

Computer-Aided Design of Thrombin Inhibitors

Computer-aided ligand design is an active, challenging, and multidisciplinary research field that blends knowledge of biochemistry, physics, and computer sciences. Whenever it is possible to experimentally determine or to model the three-dimensional structure of a pharmacologically relevant enzyme or receptor, computational approaches can be used to design specific high-affinity ligands. This article describes methods, applications, and perspectives of computer-assisted ligand design.

An example of a protein ligand found by database mining: description of the docking method and its verification by a 2.3 A X-ray structure of a thrombin-ligand complex

A computer program (SANDOCK) has been developed for the automated docking of small ligands to a target protein. It uses a guided matching algorithm to fit ligand atoms into the protein binding pocket. The protein is described by a modified Lee-Richard's dotted surface with each dot coded by chemical property and accessibility. Orientations of the ligand in the active site are generated such that a chemical and a shape complementary between the ligand and the active site cavity have to be fulfilled. The generated fits are evaluated with scoring functions which account for van der Waals, hydrophobic and hydrogen bonding interactions. This newly developed docking program can efficiently screen very large databases in a reasonable time and has been used to successfully identify novel ligands. The X-ray structure of a thrombin-ligand complex predicted by SANDOCK is described. The ligand binds to thrombin with a Kd of 65 microM and has an rmsd of 0.7 A for all ligand atoms from the predicted binding mode by SANDOCK.

Can drugs be designed?

Many examples are now emerging of the successful use of rational, structure-based methods in drug discovery. Of particular note is the development of imaginative NMR-based methods for rapid routes to ligand design. Our understanding of the chemistry underlying protein-ligand interactions, however, remains relatively poor and a major limitation in our ability to truly design drugs.

Modern computational chemistry and drug discovery: structure generating programs

During 1996 and 1997, the first reports were disclosed of active enzyme inhibitors based entirely on novel structures created by de novo methods. De novo methods have also been used to modify and significantly improve the binding affinity of an HIV protease inhibitor. Work continues in the improvement of methods for the de novo design of compounds which fit and chemically complement a binding site. De novo algorithms that generate only synthetically feasible structures have also been reported. In addition, methods are being developed for the automatic computer generation of virtual molecular libraries which can be searched to identify molecules to match a pharmacophore or fit into a binding site.