Automatic design of decision-tree induction algorithms tailored to flexible-receptor docking data

Background: This paper addresses the prediction of the free energy of binding of a drug candidate with enzyme InhA associated with Mycobacterium tuberculosis. This problem is found within rational drug design, where interactions between drug candidates and target proteins are verified through molecular docking simulations. In this application, it is important not only to correctly predict the free energy of binding, but also to provide a comprehensible model that could be validated by a domain specialist. Decision-tree induction algorithms have been successfully used in drug-design related applications, specially considering that decision trees are simple to understand, interpret, and validate. There are several decision-tree induction algorithms available for general-use, but each one has a bias that makes it more suitable for a particular data distribution. In this article, we propose and investigate the automatic design of decision-tree induction algorithms tailored to particular drug-enzyme binding data sets. We investigate the performance of our new method for evaluating binding conformations of different drug candidates to InhA, and we analyze our findings with respect to decision tree accuracy, comprehensibility, and biological relevance. Results: The empirical analysis indicates that our method is capable of automatically generating decision-tree induction algorithms that significantly outperform the traditional C4.5 algorithm with respect to both accuracy and comprehensibility. In addition, we provide the biological interpretation of the rules generated by our approach, reinforcing the importance of comprehensible predictive models in this particular bioinformatics application. Conclusions: We conclude that automatically designing a decision-tree algorithm tailored to molecular docking data is a promising alternative for the prediction of the free energy from the binding of a drug candidate with a flexible-receptor.

A Combined Study Using Ligand-Based Design, Synthesis, and Pharmacological Evaluation of Analogues of the Acetaminophen Ortho-Regioisomer with Potent Analgesic Activity

A ligand-based drug design study was performed to acetaminophen regioisomers as analgesic candidates employing quantum chemical calculations at the DFT/B3LYP level of theory and the 6-31G* basis set. To do so, many molecular descriptors were used such as highest occupied molecular orbital, ionization potential, HO bond dissociation energies, and spin densities, which might be related to quench reactivity of the tyrosyl radical to give N-acetyl-p-benzosemiquinone-imine through an initial electron withdrawing or hydrogen atom abstraction. Based on this in silico work, the most promising molecule, orthobenzamol, was synthesized and tested. The results expected from the theoretical prediction were confirmed in vivo using mouse models of nociception such as writhing, paw licking, and hot plate tests. All biological results suggested an antinociceptive activity mediated by opioid receptors. Furthermore, at 90 and 120 min, this new compound had an effect that was comparable to morphine, the standard drug for this test. Finally, the pharmacophore model is discussed according to the electronic properties derived from quantum chemistry calculations.

Pharmacokinetic properties and in silico ADME modeling in drug discovery

The discovery and development of a new drug are time-consuming, difficult and expensive. This complex process has evolved from classical methods into an integration of modern technologies and innovative strategies addressed to the design of new chemical entities to treat a variety of diseases. The development of new drug candidates is often limited by initial compounds lacking reasonable chemical and biological properties for further lead optimization. Huge libraries of compounds are frequently selected for biological screening using a variety of techniques and standard models to assess potency, affinity and selectivity. In this context, it is very important to study the pharmacokinetic profile of the compounds under investigation. Recent advances have been made in the collection of data and the development of models to assess and predict pharmacokinetic properties (ADME - absorption, distribution, metabolism and excretion) of bioactive compounds in the early stages of drug discovery projects. This paper provides a brief perspective on the evolution of in silico ADME tools, addressing challenges, limitations, and opportunities in medicinal chemistry.

Advanced Protein Modeling Method: Benchmarking and Applications in Computer-Aided Drug Discovery

Development of homology modeling methods will remain an area of active research. These methods aim to develop and model increasingly accurate three-dimensional structures of yet uncrystallized therapeutically relevant proteins e.g. Class A G-Protein Coupled Receptors. Incorporating protein flexibility is one way to achieve this goal. Here, I will discuss the enhancement and validation of the ligand-steered modeling, originally developed by Dr. Claudio Cavasotto, via cross modeling of the newly crystallized GPCR structures. This method uses known ligands and known experimental information to optimize relevant protein binding sites by incorporating protein flexibility. The ligand-steered models were able to model, reasonably reproduce binding sites and the co-crystallized native ligand poses of the β2 adrenergic and Adenosine 2A receptors using a single template structure. They also performed better than the choice of template, and crude models in a small scale high-throughput docking experiments and compound selectivity studies. Next, the application of this method to develop high-quality homology models of Cannabinoid Receptor 2, an emerging non-psychotic pain management target, is discussed. These models were validated by their ability to rationalize structure activity relationship data of two, inverse agonist and agonist, series of compounds. The method was also applied to improve the virtual screening performance of the β2 adrenergic crystal structure by optimizing the binding site using β2 specific compounds. These results show the feasibility of optimizing only the pharmacologically relevant protein binding sites and applicability to structure-based drug design projects.

Molecular modelling assisted design and synthesis of cyclothialidine derivatives as DNA gyrase inhibitors

Since cyclothialidine was discovered as the most active DNA gyrase inhibitor in 1994, enormous efforts have been devoted to make it into a commercial medicine by a number of pharmaceutical companies and research groups worldwide. However, no serious breakthrough has been made up to now. An essential problem involved with cyclothialidine is that though it demonstrated the potent inhibition of DNA gyrase, it showed little activity against bacteria. This probably is attributable to its inability to penetrate bacterial cell walls and membranes. We applied the TSAR programme to generate a QSAR equation to the gram-negative organisms. In that equation, LogP is profoundly indicated as the key factor influencing the cyclothialidine activity against bacteria. However, the synthesized new analogues have failed to prove that. In the structure based drug design stage, we designed a group of open chain cyclothialidine derivatives by applying the SPROUT programme and completed the syntheses. Improved activity is found in a few analogues and a 3D pharmacophore of the DNA gyrase B is proposed to lead to synthesis of the new derivatives for development of potent antibiotics.

Design and synthesis of potential antimicrobial agents

Tuberculosis is one of the most devastating diseases in the world primarily due to several decades of neglect and an emergence of multidrug-resitance strains (MDR) of M. tuberculosis together with the increased incidence of disseminated infections produced by other mycobacterium in AIDS patients. This has prompted the search for new antimycobacterial drugs. A series of pyridine-2-, pyridine-3-, pyridine-4-, pyrazine and quinoline-2-carboxamidrazone derivatives and new classes of carboxamidrazone were prepared in an automated fashion and by traditional synthesis. Over nine hundred synthesized compounds were screened for their anti mycobacterial activity against M. fortutium (NGTG 10394) as a surrogate for M. tuberculosis. The new classes of amidrazones were also screened against tuberculosis H37 Rv and antimicrobial activities against various bacteria. Fifteen tested compounds were found to provide 90-100% inhibition of mycobacterium growth of M. tuberculosis H37 Rv in the primary screen at 6.25 μg mL-1. The most active compound in the carboxamidrazone amide series had an MIG value of 0.1-2 μg mL-1 against M. fortutium. The enzyme dihydrofolate reductase (DHFR) has been a drug-design target for decades. Blocking of the enzymatic activity of DHFR is a key element in the treatment of many diseases, including cancer, bacterial and protozoal infection. The x-ray structure of DHFR from M. tuberculosis and human DHFR were found to have differences in substrate binding site. The presence of glycerol molecule in the Xray structure from M. tuberculosis DHFR provided opportunity to design new antifolates. The new antifolates described herein were designed to retain the pharmcophore of pyrimethamine (2,4- diamino-5(4-chlorophenyl)-6-ethylpyrimidine), but encompassing a range of polar groups that might interact with the M. tuberculosis DHFR glycerol binding pockets. Finally, the research described in this thesis contributes to the preparation of molecularly imprinted polymers for the recognition of 2,4-diaminopyrimidine for the binding the target. The formation of hydrogen bonding between the model functional monomer 5-(4-tert-butyl-benzylidene)-pyrimidine-2,4,6-trione and 2,4-diaminopyrimidine in the pre-polymerisation stage was verified by 1H-NMR studies. Having proven that 2,4-diaminopyrimidine interacts strongly with the model 5-(4-tert-butylbenzylidene)- pyrimidine-2,4,6-trione, 2,4-diaminopyrimidine-imprinted polymers were prepared using a novel cyclobarbital derived functional monomer, acrylic acid 4-(2,4,6-trioxo-tetrahydro-pyrimidin-5- ylidenemethyl)phenyl ester, capable of multiple hydrogen bond formation with the 2,4- diaminopyrimidine. The recognition property of the respective polymers toward the template and other test compounds was evaluated by fluorescence. The results demonstrate that the polymers showed dose dependent enhancement of fluorescence emissions. In addition, the results also indicate that synthesized MIPs have higher 2,4-diaminopyrimidine binding ability as compared with corresponding non-imprinting polymers.

The design of novel antifolates against pneumocystis carinii

The pneumonia caused by Pneumocystis carinii is ultimately responsible for the death of many acquired immunodeficiency syndrome (AIDS) patients. Large doses of trimethoprim and pyrimethamine in combination with a sulphonamide and/or pentamidine suppress the infection but produce serious side-effects and seldom prevent recurrence after treatment withdrawal. However, the partial success of the aforementioned antifolates, and also trimetrexate used alone, does suggest dihydrofolate reductase (DHFR) as a target for the development of antipneumocystis agents. From the DHFR inhibitory activities of 3'-substituted pyrimethamine analogues it was suggested that the 3'-(3'',3''-dimethyltriazen-1''-yl) substituent may be responsible for the greater activity for the P.carinii over the mammalian enzyme. Crystallographic and molecular modeling studies revealed considerable geometrical and electronic differences between the triazene and the chemically related formamidine functions that may account for the differences in DHFR inhibitory profiles. Structural and electronic parameters calculated for a series of 3'-(3'',3''-disubstitutedtriazen-1''-yl) pyrimethamine analogues did not correlate with the DHFR inhibitory activities. However, the in vitro screening against P.carinii DHFR revealed that the 3''-hydroxyethyl-3''-benzyl analogue was the most active and selective. Models of the active sites of human and P.carinii DHFRs were constructed using DHFR sequence and structural homology data which had identified key residues involved in substrate and cofactor binding. Low energy conformations of the 3'',3''-dimethyl and 3''-hydroxyethyl-3''-benzyle analogues, determined from nuclear magnetic resonance studies and theoretical calculations, were docked by superimposing the diaminopyrimidine fragment onto a previously docked pyrimethamine analogue. Enzyme kinetic data supported the 3''-hydroxyethyl-3''-benzyl moiety being located in the NADPH binding groove. The 3''-benzyl substituent was able to locate to within 3 AA of a valine residue in the active site of P.carinii DHFR thereby producing a hydrophobic contact. The equivalent residue in human DHFR is threonine, more hydrophilic and less likely to be involved in such a contact. This difference may account for the greater inhibitory activity this analogue has for P.carinii DHFR and provide a basis for future drug design. From an in vivo model of PCP in immunosuppressed rats it was established that the 3"-hydroxyethyl-3"-benzyl analogue was able to reduce the.P.carinii burden more effectively with increasing doses, without causmg any visible signs of toxicity. However, equivalent doses were not as effective as pentamidine, a current treatment of choice for Pneumocystis carinii pneumonia.

Relieving the first bottleneck in the drug discovery pipeline:using array technologies to rationalize membrane protein production

The slow down in the drug discovery pipeline is, in part, owing to a lack of structural and functional information available for new drug targets. Membrane proteins, the targets of well over 50% of marketed pharmaceuticals, present a particular challenge. As they are not naturally abundant, they must be produced recombinantly for the structural biology that is a prerequisite to structure-based drug design. Unfortunately, however, obtaining high yields of functional, recombinant membrane proteins remains a major bottleneck in contemporary bioscience. While repeated rounds of trial-and-error optimization have not (and cannot) reveal mechanistic details of the biology of recombinant protein production, examination of the host response has provided new insights. To this end, we published an early transcriptome analysis that identified genes implicated in high-yielding yeast cell factories, which has enabled the engineering of improved production strains. These advances offer hope that the bottleneck of membrane protein production can be relieved rationally.

Improving drug discovery using hybrid softcomputing methods

Virtual screening (VS) methods can considerably aid clinical research, predicting how ligands interact with drug targets. Most VS methods suppose a unique binding site for the target, but it has been demonstrated that diverse ligands interact with unrelated parts of the target and many VS methods do not take into account this relevant fact. This problem is circumvented by a novel VS methodology named BINDSURF that scans the whole protein surface in order to find new hotspots, where ligands might potentially interact with, and which is implemented in last generation massively parallel GPU hardware, allowing fast processing of large ligand databases. BINDSURF can thus be used in drug discovery, drug design, drug repurposing and therefore helps considerably in clinical research. However, the accuracy of most VS methods and concretely BINDSURF is constrained by limitations in the scoring function that describes biomolecular interactions, and even nowadays these uncertainties are not completely understood. In order to improve accuracy of the scoring functions used in BINDSURF we propose a hybrid novel approach where neural networks (NNET) and support vector machines (SVM) methods are trained with databases of known active (drugs) and inactive compounds, being this information exploited afterwards to improve BINDSURF VS predictions.

Improving drug discovery using a neural networks based parallel scoring function

Virtual Screening (VS) methods can considerably aid clinical research, predicting how ligands interact with drug targets. Most VS methods suppose a unique binding site for the target, but it has been demonstrated that diverse ligands interact with unrelated parts of the target and many VS methods do not take into account this relevant fact. This problem is circumvented by a novel VS methodology named BINDSURF that scans the whole protein surface to find new hotspots, where ligands might potentially interact with, and which is implemented in massively parallel Graphics Processing Units, allowing fast processing of large ligand databases. BINDSURF can thus be used in drug discovery, drug design, drug repurposing and therefore helps considerably in clinical research. However, the accuracy of most VS methods is constrained by limitations in the scoring function that describes biomolecular interactions, and even nowadays these uncertainties are not completely understood. In order to solve this problem, we propose a novel approach where neural networks are trained with databases of known active (drugs) and inactive compounds, and later used to improve VS predictions.

Design and validation of a novel generic platform for the production of tetravalent IgG1-like bispecific antibodies

International audience

Design and Synthesis of Novel Kinase Inhibitors for the Treatment of Chronic Respiratory Diseases

In 2017, Chronic Respiratory Diseases accounted for almost four million deaths worldwide. Unfortunately, current treatments are not definitive for such diseases. This unmet medical need forces the scientific community to increase efforts in the identification of new therapeutic solutions. PI3K delta plays a key role in mechanisms that promote airway chronic inflammation underlying Asthma and COPD. The first part of this project was dedicated to the identification of novel PI3K delta inhibitors. A first SAR expansion of a Hit, previously identified by a HTS campaign, was carried out. A library of 43 analogues was synthesised taking advantage of an efficient synthetic approach. This allowed the identification of an improved Hit of nanomolar enzymatic potency and moderate selectivity for PI3K delta over other PI3K isoforms. However, this compound exhibited low potency in cell-based assays. Low cellular potency was related to sub optimal phys-chem and ADME properties. The analysis of the X-ray crystal structure of this compound in human PI3K delta guided a second tailored SAR expansion that led to improved cellular potency and solubility. The second part of the thesis was focused on the rational design and synthesis of new macrocyclic Rho-associated protein kinases (ROCKs) inhibitors. Inhibition of these kinases has been associated with vasodilating effects. Therefore, ROCKs could represent attractive targets for the treatment of pulmonary arterial hypertension (PAH). Known ROCK inhibitors suffer from low selectivity across the kinome. The design of macrocyclic inhibitors was considered a promising strategy to obtain improved selectivity. Known inhibitors from literature were evaluated for opportunities of macrocyclization using a knowledge-based approach supported by Computer Aided Drug Design (CADD). The identification of a macrocyclic ROCK inhibitor with enzymatic activity in the low micro molar range against ROCK II represented a promising result that validated this innovative approach in the design of new ROCKs inhibitors.

Applications of network-based approaches in drug research

Recent research trends in computer-aided drug design have shown an increasing interest towards the implementation of advanced approaches able to deal with large amount of data. This demand arose from the awareness of the complexity of biological systems and from the availability of data provided by high-throughput technologies. As a consequence, drug research has embraced this paradigm shift exploiting approaches such as that based on networks. Indeed, the process of drug discovery can benefit from the implementation of network-based methods at different steps from target identification to drug repurposing. From this broad range of opportunities, this thesis is focused on three main topics: (i) chemical space networks (CSNs), which are designed to represent and characterize bioactive compound data sets; (ii) drug-target interactions (DTIs) prediction through a network-based algorithm that predicts missing links; (iii) COVID-19 drug research which was explored implementing COVIDrugNet, a network-based tool for COVID-19 related drugs. The main highlight emerged from this thesis is that network-based approaches can be considered useful methodologies to tackle different issues in drug research. In detail, CSNs are valuable coordinate-free, graphically accessible representations of structure-activity relationships of bioactive compounds data sets especially for medium-large libraries of molecules. DTIs prediction through the random walk with restart algorithm on heterogeneous networks can be a helpful method for target identification. COVIDrugNet is an example of the usefulness of network-based approaches for studying drugs related to a specific condition, i.e., COVID-19, and the same ‘systems-based’ approaches can be used for other diseases. To conclude, network-based tools are proving to be suitable in many applications in drug research and provide the opportunity to model and analyze diverse drug-related data sets, even large ones, also integrating different multi-domain information.

Structural and chemical basis for anticancer activity of a series of 'beta'-tubulin ligands: molecular modeling and 3D QSAR studies

An important approach to cancer therapy is the design of small molecule modulators that interfere with microtubule dynamics through their specific binding to the ²-subunit of tubulin. In the present work, comparative molecular field analysis (CoMFA) studies were conducted on a series of discodermolide analogs with antimitotic properties. Significant correlation coefficients were obtained (CoMFA(i), q² =0.68, r²=0.94; CoMFA(ii), q² = 0.63, r²= 0.91), indicating the good internal and external consistency of the models generated using two independent structural alignment strategies. The models were externally validated employing a test set, and the predicted values were in good agreement with the experimental results. The final QSAR models and the 3D contour maps provided important insights into the chemical and structural basis involved in the molecular recognition process of this family of discodermolide analogs, and should be useful for the design of new specific ²-tubulin modulators with potent anticancer activity.