Dissertations / Theses on the topic 'Computational docking'

This thesis describes the development and refinement of a number of techniques for molecular docking and ligand database screening, as well as the application of these techniques to predict the structures of several protein-ligand complexes and to discover novel ligands of an important receptor protein. Global energy optimisation by Monte-Carlo minimisation in internal co-ordinates was used to predict bound conformations of eight protein-ligand complexes. Experimental X-ray crystallography structures became available after the predictions were made. Comparison with the X-ray structures showed that the docking procedure placed 30 to 70% of the ligand molecule correctly within 1.5A from the native structure. The discrimination potential for identification of high-affinity ligands was derived and optimised using a large set of available protein-ligand complex structures. A fast boundary-element solvation electrostatic calculation algorithm was implemented to evaluate the solvation component of the discrimination potential. An accelerated docking procedure utilising pre-calculated grid potentials was developed and tested. For 23 receptors and 63 ligands extracted from X-ray structures, the docking and discrimination protocol was capable of correct identification of the majority of native receptor-ligand couples. 51 complexes with known structures were predicted. 35 predictions were within 3A from the native structure, giving correct overall positioning of the ligand, and 26 were within 2A, reproducing a detailed picture of the receptor-ligand interaction. Docking and ligand discrimination potential evaluation was applied to screen the database of more than 150000 commercially available compounds for binding to the fibroblast growth factor receptor tyrosine kinase, the protein implicated in several pathological cell growth aberrations. As expected, a number of compounds selected by the screening protocol turned out to be known inhibitors of the tyrosine kinases. 49 putative novel ligands identified by the screening protocol were experimentally tested and five compounds have shown inhibition of phosphorylation activity of the kinase. These compounds can be used as leads for further drug development.

Each biomolecule in a living organism needs to adopt a specific threedimensional conformation to function properly. Function itself is usually achieved by specific interactions between biomolecular units. Structural knowledge at atomic level of biomolecules and their interaction is important to understand the mechanisms leading to biological response and to develop strategies to interfere with them when necessary. Antibodies are molecules of the immune system playing an ever more prominent role in basic research as well as in the biotechnology and pharmaceutical sectors. Characterizing their region of interaction with other proteins (epitopes) is useful for purposes ranging from molecular biology research to vaccine design. During my PhD studies I used a combination of solution NMR mapping, molecular biology and computational docking to provide a structural and biophysical characterization of new neutralizing antibodies from Dengue virus recovered subjects, comparing the binding of the same antibody to the four Dengue serotypes and the binding of different antibodies to the same serotype. We were able to rationally mutate an antibody to first alter its selectivity for different viral strains and then increase its neutralization by ~40 folds. For the first time, this was achieved without the availability of an x-ray structure. In a second sub-project, I investigated the interaction of the chemokine CXCL12 with the chromatin-associated protein HMGB1, confirming their direct interaction (only proposed but never proved before) and providing a structural explanation for the HMGB1 dependent increase of CXCL12 cellular activity. High profile publications resulted from the two above projects. The above mentioned projects relied heavily on solution NMR spectroscopy, which is ideally suited to the atomic level characterization of intermolecular interfaces and, as a consequence, to antibody epitope discovery. Having provided a residue-level description of a protein-protein interface by NMR, we subsequently used this experimental information to guide and validate computational docking experiments aimed at providing a three dimensional structure of the protein-protein (or antibody-protein) complex of interest. In collaboration with other members of my research group I validated the use of NMR and computational simulations to study antibody-antigen interactions, publishing two reviews in collaboration with other members of my research group.

The docking problem is to start with unbound conformations for the components of a complex, and computationally model a near-native structure for the complex. This thesis describes work in developing computer programs to tackle both protein/protein and protein/DNA docking. Empirical pair potential functions are generated from datasets of residue/residue interactions. A scoring function was parameterised and then used to screen possible complexes, generated by the global search computer algorithm FTDOCK using shape complementarity and electrostatics, for 9 systems. A correct docking (RMSD < 2.5A) is placed within the top 12% of the pair potential score ranked complexes for all systems. The computer software FTDOCK is modified for the docking of proteins to DNA, starting from the unbound protein and DNA coordinates modelled computationally. Complexes are then ranked by protein/DNA pair potentials derived from a database of 20 protein/DNA complexes. A correct docking (at least 65% of correct contacts) was identified at rank < 4 for 3 of the 8 complexes. This improved to 4 out of 8 when the complexes were filtered using experimental data defining the DNA footprint. The FTDOCK program was rewritten, and improved pair potential functions were developed from a set of non-homologous protein/protein interfaces. The algorithms were tested on a non-homologous set of 18 protein/protein complexes, starting with unbound conformations. Us ing cross-validated pair potential functions and the energy rninimisation software MultiDock, a correct docking ( RMSD of CQ interface 25% correct contacts) is found in the top 10 ranks in 6 out of 18 systems. The current best computational docking algorithms are discussed, and strategies for improvement are suggested.

Cette thèse porte sur les méta-heuristiques hiérarchiques parallèles adaptatives pour l'échantillonnage conformationnel. Étant un problème hautement combinatoire et multlmodal, l'échantillonnage conformationnel requière la construction d'approches hybrides à large échelle. Après une analyse dei modèles mathématiques, nécessitant l'examen des différentes formulations du champ de force, nous avons proposé une étude des opérateurs de variation et des méthodes de recherche locale adaptés au problème ainsi que leur hybridation dynamique et adaptative. Cette étude nous a conduit à la proposition de mécanismes d'adaptation des paramètres des algorithmes utilisés en fonction du processus d'évolution. Dans cette thèse, nous proposons également des algorithmes adaptatifs hybndes hiérarchiques distribués, fortement extensibles. L'expérimentation, basée sur l'utilisation de multiples modèles parallèles, démontre la grande efficacité de ces algorithmes. En effet, les résultats obtenus montrent que des RMSD moyens en dessous de 1.0 A peuvent être obtenus sur des instances difficiles des problèmes de prédiction de la structure des protéines et de docking moléculaire. La validation des approches hybrides proposées a été effectuée sur Grid'5000, une grille expérimentale d'échelle nationale composée d'environ 5000 coeurs de calcul. Une image système a été développée en utilisant Globus pour permettre des déploiements distribués à large échelle. L'approche hiérarchique distribuée construite a été ainsi déployée sur plusieurs grappes, avec près de 1000 coeurs de calcul
The thesis proposes an extensive analysis of adaptive hierarchical parallel metaheuristics for ab initio conformational sampling. Standing as an NP, combinatorial, highly multi-modal optimization problem, conformational sampling requires for high-performance large scale hybrid approaches to be constructed. Following an incremental definition, minimum complexity conformational sampling mathematical models are first analyzed, entailing a review of different force field formulations. A comprehensive analysis is conducted on a large set of operators and local search algorithms including adaptive and dynamic mechanisms. As determined by the analysis outcomes, complex a priori and online parameter tuning stages are designed. finally, highly scalable hierarchical hybrid distributed algorithm designs are proposed. Experimentation is carried over multiple parallelization models with afferent cooperation topologies. Expenmentations resulted in unprecedented results to be obtained. Multiple perfect conformational matches have been determined, on highly difficult protein structure prediction and molecular docking benchmarks, with RMSD average values below 1.0A. The validation of the proposed hybrid approaehes was performed on Grid'5000, a French computational grid, with almost 5000 computational cores. A Globus Toolkit hased Grid'SOOO system image has been developed, sustaining large scale distributed deployments. The constructed hierarchical hybrid distributed algorithm has been deployed on multiple clusters, with almost 1000 computing cores. Finally, a parallel AutoDock version was developed using the ParadisEO framework, integrating the developed algorithms

Computing has pushed a paradigm shift in many disciplines, including structural biology and chemistry. This change has been mainly driven by the increase in performance of computers, the capacity of dealing with huge amounts of experimental and analysis data and the development of new algorithms. Thanks to these advances, our understanding on the chemistry that supports life has increased and it is even more sophisticated that we had never imagined before. Proteins play a major role in nature and are often described as the factories of the cell as they are involved in virtually all important function in living organisms. Unfortunately, our understanding of the function of many proteins is still very poor due to the actual limitations in experimental techniques which, at the moment, they can not provide crystal structure for many protein complexes. The development of computational tools as protein-protein docking methods could help to fill this gap. In this thesis, we have presented a new protein-protein docking method, LightDock, which supports the use of different custom scoring functions and it includes anisotropic normal analysis to model backbone flexibility upon binding process. Second, several interesting web-based tools for the scientific community have been developed, including a web server for protein-protein docking, a web tool for the characterization of protein-protein interfaces and a web server for including SAXS experimental data for a better prediction of protein complexes. Moreover, the optimizations made in the pyDock protocol and the increase in th performance helped our group to score in the 5th position among more than 60 participants in the past two CAPRI editions. Finally, we have designed and compiled the Protein-Protein (version 5.0) and Protein-RNA (version 1.0) docking benchmarks, which are important resources for the community to test and to develop new methods against a reference set of curated cases.
Gràcies als recents avenços en computació, el nostre coneixement de la química que suporta la vida ha incrementat enormement i ens ha conduït a comprendre que la química de la vida és més sofisticada del que mai haguéssim pensat. Les proteïnes juguen un paper fonamental en aquesta química i són descrites habitualment com a les fàbriques de les cèl·lules. A més a més, les proteïnes estan involucrades en gairebé tots els processos fonamentals en els éssers vius. Malauradament, el nostre coneixement de la funció de moltes proteïnes és encara escaig degut a les limitacions actuals de molts mètodes experimentals, que encara no són capaços de proporcionar-nos estructures de cristall per a molts complexes proteïna-proteïna. El desenvolupament de tècniques i eines informàtiques d’acoblament proteïna-proteïna pot ésser crucial per a ajudar-nos a reduir aquest forat. En aquesta tesis, hem presentat un nou mètode computacional de predicció d’acoblament proteïna-proteïna, LightDock, que és capaç de fer servir diverses funcions energètiques definides per l’usuari i incloure un model de flexibilitat de la cadena principal mitjançant la anàlisis de modes normals. Segon, diverses eines d’interès per a la comunitat científica i basades en tecnologia web han sigut desenvolupades: un servidor web de predicció d’acoblament proteïna-proteïna, una eina online per a caracteritzar les interfícies d’acoblament proteïna-proteïna i una eina web per a incloure dades experimentals de tipus SAXS. A més a més, les optimitzacions fetes al protocol pyDock i la conseqüent millora en rendiment han propiciat que el nostre grup de recerca obtingués la cinquena posició entre més de 60 grups en les dues darreres avaluacions de l’experiment internacional CAPRI. Finalment, hem dissenyat i compilat els banc de proves d’acoblament proteïna-proteïna (versió 5) i proteïna-ARN (versió 1), molt importants per a la comunitat ja que permeten provar i desenvolupar nous mètodes i analitzar-ne el rendiment en aquest marc de referència comú.

The dynamic character of proteins strongly influences biomolecular recognition mechanisms. With the development of the main models of ligand recognition (lock-and-key, induced fit, conformational selection theories), the role of protein plasticity has become increasingly relevant. In particular, major structural changes concerning large deviations of protein backbones, and slight movements such as side chain rotations are now carefully considered in drug discovery and development. It is of great interest to identify multiple protein conformations as preliminary step in a screening campaign. Protein flexibility has been widely investigated, in terms of both local and global motions, in two diverse biological systems. On one side, Replica Exchange Molecular Dynamics has been exploited as enhanced sampling method to collect multiple conformations of Lactate Dehydrogenase A (LDHA), an emerging anticancer target. The aim of this project was the development of an Ensemble-based Virtual Screening protocol, in order to find novel potent inhibitors. On the other side, a preliminary study concerning the local flexibility of Opioid Receptors has been carried out through ALiBERO approach, an iterative method based on Elastic Network-Normal Mode Analysis and Monte Carlo sampling. Comparison of the Virtual Screening performances by using single or multiple conformations confirmed that the inclusion of protein flexibility in screening protocols has a positive effect on the probability to early recognize novel or known active compounds.

The dynamic character of proteins strongly influences biomolecular recognition mechanisms. With the development of the main models of ligand recognition (lock-and-key, induced fit, conformational selection theories), the role of protein plasticity has become increasingly relevant. In particular, major structural changes concerning large deviations of protein backbones, and slight movements such as side chain rotations are now carefully considered in drug discovery and development. It is of great interest to identify multiple protein conformations as preliminary step in a screening campaign. Protein flexibility has been widely investigated, in terms of both local and global motions, in two diverse biological systems. On one side, Replica Exchange Molecular Dynamics has been exploited as enhanced sampling method to collect multiple conformations of Lactate Dehydrogenase A (LDHA), an emerging anticancer target. The aim of this project was the development of an Ensemble-based Virtual Screening protocol, in order to find novel potent inhibitors. On the other side, a preliminary study concerning the local flexibility of Opioid Receptors has been carried out through ALiBERO approach, an iterative method based on Elastic Network-Normal Mode Analysis and Monte Carlo sampling. Comparison of the Virtual Screening performances by using single or multiple conformations confirmed that the inclusion of protein flexibility in screening protocols has a positive effect on the probability to early recognize novel or known active compounds.

La presente Tesis Doctoral constituye un estudio teórico sobre el proceso de inhibición de la acetilcolinesterasa por moléculas que bloquean el sitio activo de la proteína. Se han estudiado un conjunto de 44 inhibidores, dividido en 8 clases distintas, por método QM/MM, docking molecular, análisis estadístico utilizando los métodos de análisis multivariados de datos, reconocimiento de patrones y de algoritmos de agrupamiento. Se ha estudiado la interacción y se ha calculado la energía libre de enlace (Gbind) utilizando los métodos híbridos QM/MM MD asociados a los métodos de la FEP y del BIE. De esta forma, podemos entender la interacción entre los residuos de la enzima, las moléculas de agua estructurales y los diversos inhibidores, a nivel atómico, que son fundamentales para el diseño de nuevos inhibidores y drogas con aplicaciones directas en la enfermedad de Alzheimer.

The ion channels are important membrane bound proteins and multi-therapeutic target for a number of diseases. There are many scorpion toxins reported to bind with KV1 channels. We have identified a plant toxin. We have studied binding of two different sequence length peptides, 1-47 and 10-44 with KV1.3 and KV1.1 channels. The binding complex structures are obtained by molecular docking and stability of the complexes are checked by MD simulations. The pore inserting residue is predicted as K33. The KV1.1-toxin complexes are found to be unstable, and toxin doesn't block the pore in MD simulations. The predicted binding free energies for both complexes of KV1.3-toxin are within the range of experimental values with picomolar and nanomolar activities, respectively. Experiments have also confirmed that Toxin(1-47) does not block the current with KV1.1 channel as well. The inward-rectifier potassium (Kir) channels play significant roles in several physiological disorders. I have studied binding of honey bee toxin called tertiapin (TPN) with Kir3.x channels. K21 is predicted to be the pore inserting residue. The binding free energies are calculated and validated with experimental value. I have studied TPN complexes with Kir3.1 and Kir 3.3 and explained insensitivity of TPN to these channels. A few crystal structures of bacterial sodium channels have been determined but the crystal structures of mammalian ones have not been resolved yet. So, it is important to understand similarities and differences between the bacterial and mammalian channels. The validated homology modeled NaV1.4 complexed with GIIIA system provides a good model for such comparisons. Here I have studied the binding of GIIIA to the bacterial sodium channels NaVAb and NaVRh by combination of docking and MD simulations then the potential mean forces are constructed. Comparison of the binding mode of GIIIA between mammalian and bacterial channels. Protein aggregation affects both human physiological functions and bioengineered products so finding methods to prevent aggregation will we very useful. It is an important area of investigation which could be facilitated by a molecular-level understanding of dimer formation, which is the first step in aggregation. Here we propose a computational method based on molecular dynamics simulations that will facilitate finding aggregation-prone regions on human lysozyme HL[D67H], noting that the wild type HL doesn't aggregate.

In this dissertation a novel approach to protein-ligand docking is presented. First an existing method to predict putative active sites is employed. These predictions are then used to cut down the search space of an algorithm that uses the fast Fourier transform to calculate the geometrical and electrostatic complementarity between a protein and a small organic ligand. A simplified hydrophobicity score is also calculated for each active site. The docking method could be applied either to dock ligands in a known active site or to rank several putative active sites according to their biological feasibility. The method was evaluated on a set of 310 protein-ligand complexes. The results show that with respect to docking the method with its initial parameter settings is too coarse grained. The results also show that with respect to ranking of putative active sites the method works quite well.

O surgimento de novas substâncias psicoativas (NPS-New Psychoactive Substances) levantou muitas questões no contexto da aplicação da lei e políticas públicas de drogas. De acordo com o Escritório das Nações Unidas sobre Drogas e Crime (UNODC- United Nations Office on Drugs and Crime) como uma alternativa às drogas proibidas. Esses novos compostos foram projetados e formulados para escapar à legislação de controle de drogas, criando um fenômeno que se tornou um problema internacional. No Brasil, essas substâncias são controladas e penalmente puníveis, no pela Lei 11.343/2006, também conhecida como Lei de Drogas. Este trabalho traz estudos relativos às catinonas sintéticas com metodologia in silico para investigar mecanismos de detecção e tendência de atuação no organismo humano. No estudo relacionado à detecção utilizamos a reação dessas drogas com o isotiocianato de fluoresceína (FITC Fluorescein isothiocyanate). Para essa proposta foram feitos estudos de viabilidade de métodos de cálculo, análise conformacional do FITC, avaliação energética da reação com as catinonas e os espectros de emissão. Em relação à viabilidade dos métodos de cálculo temos que a otimização prévia dos compostos envolvidos com o semi-empírico PM6 e posterior refinamento com o método B3LYP/6-31G** foram adequados para os cálculos. A avaliação energética mostrou que a reação é favorável para anfetaminas, aminoácidos e catinonas, e os menores valores foram encontrados no último caso. Nos estudos de emissão obtivemos resultados semelhantes ao perfil energético; no entanto, observamos que os espectros são únicos, representando uma baixa probabilidade de falsos positivos. Avaliações de docking mostraram que as catinonas têm mais afinidade com o receptor dopaminérgico do que suas anfetaminas homólogas, confirmando dados experimentais relatados na literatura. Por fim, os estudos realizados neste trabalho demonstraram a importância e a capacidade dos métodos in silico que apresentam grau potencial na área e que podem ser amplamente utilizados em investigações com diferentes propósitos no campo forense.
The emergence of new psychoactive substances (NPSs) has raised many issues in the context of law enforcement and public drug policies. According to the United Nations Office on Drugs and Crime (UNODC), NPS were created as an alternative to forbidden drugs. These new compounds were designed and formulated to escape the drug control legislation, creating a phenomenon that has become an international problem. In Brazil, these substances are controlled and punishable by Law 11,343 / 2006, also known as the Drug Law. This work presents studies on synthetic cathinones with in silico methodology to investigate mechanisms of detection and tendency of action in the human organism. In the detection-related study, we used the reaction of these drugs with fluorescein isothiocyanate (FITC). For this proposal were made studies regarding to the viability of the calculation methods, FITC conformational analysis, energetic evaluation of the reaction with the cathinones and the emission spectra. In relation to the viability of the calculation methods we have that the previous optimization of the compounds involved with the semi-empirical PM6 and subsequent refinement with the B3LYP / 6-31G ** method were adequate for the calculations. The energetic evaluation showed that the reaction is favorable for amphetamines, amino acids and cathinones, and the lowest values were found in the last case. In the emission studies we obtained similar results to the energy profile; however, we observed that the spectra are unique representing a low probability of false positive. Docking evaluations have shown that cathinones have more affinity to the dopaminergic receptor than their homologous amphetamines, confirming experimental data reported in the literature. Finally, the studies carried out in this work demonstrated the importance and the capacity of the in silico methods that present with potential grade in the area and that can be widely used in investigations with different purposes in the forensic field.

Computer-based techniques have become especially important in molecular biology, since they often represent the only viable way to understand some phenomena at atomic and molecular level. The complexity of biological systems, which usually needs to be analyzed with different levels of accuracy, requires the application of different approaches. Computational methodologies applied to biotechnologies allow a molecular comprehension of biological systems at different levels of depth. Quantum mechanics (QM) ab-initio techniques allow the study of enzymes and organometallic models at sub-atomic levels keeping into account electronic effects on stereochemistry and chemical reactivity. We set to study [FeFe]-hydrogenases, enzymes able to both produce and oxidize H2 at high rate. The study was focused to better elucidate some redox states of the cofactor during catalysis. The principal aim of this work was to take advantage of hydrogenases biomimetic complexes to gain further inside on the catalysis of the enzyme, and pinpoint the structural and stereo-electronic features necessary to improve the efficiency of the synthetic models. H2 is a desirable fuel but the usage of this gas is somehow problematic due to its physical properties, leading to safety concerns and low energy density. A possible way to overcome these problems is to store H2 in safe and valued added chemicals. In this perspective we studied the catalytic mechanism of the first iron-containing synthetic complex able to catayze the chemical storage of H2 and CO2, converting it into HCOOH. QM methodologies were also used in a project in collaboration with the Dept. of Forensic Medicine at the University of Verona, aimed at the use of carbohydrate deficient transferrin (Tf) as marker of alcohol abuse. Tf is the protein deputed for the iron transport in the blood stream. Low glycosylated forms are known to be associated to alcohol abuse. Different spectroscopies were useful tools to discover the binding site of terbium and the best experimental conditions for terbium-Tf binding. To get more information about the active site, we optimized a method that allowed us to determine the molecular structure of the metal environment through QM computational techniques. Docking techniques to study small-ligand protein binding are useful methods to predict the binding mode of a molecule to a receptor, in order to understand its mechanism of action and improve its activity. Here, we focused on different pharmacological targets involved in different pathological mechanisms to understand how the ligand is able to interact with the receptor and exert its pharmacological effect, and how to ameliorate its structure to increase the specificity. Since the evolution of the human species exists a struggle for survival between host and pathogens, with measure and countermeasure to respectively infect and defend against infections. Positively selected sites on protein genes are the result of evolutionary pressure on certain aminoacidic residues that could be fundamental for host and pathogen infections. Protein-protein docking is a useful tool, together with computational stability analysis, to understand how residues variations modify the binding among different proteins in the immune system and how the proteins stability is affected. In conclusion, the choice of the computational methods is what determines the level of the description of the molecular system. The study of biotechnologically relevant system with computational techniques is a powerful tool to gain insight into molecular properties that are otherwise not explorable by experimental techniques.

Abstract Histidine acid phosphatases are characterized by the presence of a conserved RHGXRXP motif. One medically important acid phosphatase is the Prostatic Acid Phosphatase (PAP). PAP has been associated with prostate cancer for a long time and has been used as a marker to stage prostate carcinoma. Yet, there is no clear understanding on the functioning of the enzyme in vivo. This thesis work focuses on the characterization of putative ligands and elucidation of the reaction mechanism of PAP using computational methods. The ligand-enzyme complexes were generated by docking and molecular dynamics simulations. The complexes showed that the conserved arginines of RHGXRXP motif are important for binding the highly negatively charged phosphate group. The complexes also highlighted that the active site aspartate (Asp258) should be neutral in the complex and is involved as a general acid-base in the reaction. The studies support that PAP could dephosphorylate the growth factor receptors EGFR and ErbB-2. The studies also found that the majority of tyrosine phosphorylated peptides from these growth factor receptors could bind to PAP. The affinities were assessed based on theoretical calculations and were further confirmed by experimental measurements in the feasible cases. To clearly understand the mechanism of PAP, quantum mechanical methods were employed. The enzymatic reaction involves two steps. In the first step, the phosphate moiety is transferred from the ligand to the conserved histidine. The calculations on the first step of the reaction involved generating the transition state (TS) structures and estimating the respective barriers. The calculations clearly support that Asp258 becomes neutral by picking up the proton from the monoanionic ligand entering the binding site. The proton from neutral Asp258 is later transferred to the leaving group via a water bridge, restoring the negative state of Asp258. The second step involves the hydrolysis of phosphohistidine enzyme intermediate. Using hybrid quantum mechanics/molecular mechanics calculations, it was found that the Asp258 accepts a proton from the nucleophilic water only after the TS is crossed. This proton is possibly then transferred to the free phosphate while it leaves the binding site, restoring the enzyme to its free state. The study highlights the importance of active site arginines in the binding as well as the stabilization of TS. Further, the analysis of TS structures in both the steps showed an associative mechanism, based on the distance of the nucleophilic and the leaving atoms to the phosphate atom. These distances are much smaller than what has been found in other well studied nonmetallo-phopshatases. Thus, the study finds a novel mechanism of enzymic phosphotransfer in PAP mediated catalysis.

Durante il mio PhD ho utilizzato diverse tecniche computazionali, in particolare metodi QM e MM, per studiare la relazione struttura/attività delle proteine. Il primo progetto consiste nella caratterizzazione strutturale del cofattore metallico FeMo-co presente nel sito attivo delle Mo-nitrogenasi. L'obbiettivo è studiare da un punto di vista strutturale l'intermedio E 4 che è attualmente ritenuto essere lo step nel quale avviene il legame con N 2 . A tale scopo sono state svolte simulazioni DFT sul FeMo-co considerando ogni possibile stato redox dei ferri proposto in letteratura ( [5FeIII:2FeII]; [3FeIII:4FeII]; [1FeIII:6FeII]) e la forma protonata e non protonata dell'omocitrato. Solo con la configurazione [3FeIII:4FeII] e con l'omocitrato protonato la struttura ottimizzata di E4 mostra la presenza di due idruri a ponte disposti in modo ortogonale tra loro e legati entrambi ai medesimi atomi di ferro (Fe6 e Fe7 ) in accordo con i dati EPR. Tramite Broken Symmetry si è studiato anche il corretto allineamento di spin dei vari atomi di ferro con particolare riguardo per i due ferri (Fe6 e Fe7) che legano gli idruri. Un altro aspetto studiato riguardante le Mo-nistrogenasi è stata la caratterizzazione della mutazione Ar g 96→Gln . Tale sostituzione rende possibile il binding dell'acetilene al sito attivo delle nitrogenasi anche durante il resting state E0 . Per tentre di dare una spiegazione a tale fatto, si è calcolata l'energia di binding dell'acetilene sia per sito attivo wild type che per quello mutato tramite simulazioni DFT. L'energia di binding dell'acetilene risulta essere più favorevole per il mutante che per il wild type in linea con i dati sperimentali. Dalle simulazioni effettuate risulta che la sostituzione Arg 96 → Gln altera le proprietà chimico-fisiche della tasca di legame, in particolare tale mutazione ha l'effetto di aumentare le caratteristiche idrofobiche del sito di binding. Considerando la natura apolare dell'acetilene, l'aumento del carattere idrofobico dell'intorno aminoacidico del FeMo-co spiega bene il motivo del binding anche durante il resting state E0 . Il secondo progetto è la caratterizzazione del meccanismo catalitico di un enzima di de novo design tramite simulazioni DFT e di docking molecolare. La proteina presa in considerazione è TRIL9CL23H, un analogo dell'anidrasi carbonica. Il meccanismo catalitico proposto per TRIL9CL23H risulta essere una combinazione dei meccanismi catalitici della α-anidrasi e della β-anidrasi: come accade nel α-anidrasi il sito attivo è costituito da una molecola di acqua coordinata ad un atomo di zinco, a sua volta coordinato con geometria tetraedrica a tre istidine; come accade invece nella β-anhydrase l'accettore del protone perso dalla molecola d'acqua coordinata allo zinco risulta essere un glutammato. I risultati mostrano che a differena dell'anidrasi carbonica, dove il passaggio limitante è l'attivazione della molecola d'acqua, nel caso di TRIL9CL23H il collo di bottiglia di tutto il processo catalitico risulta essere l'uscita dei prodotti dal sito attivo dell'enzima. Nell'ultimo progetto si è studiato l'effetto della sostituzione R10T della subuità Mre11 del complesso MRX. L'obbiettivo di questo studio è quello di caratterizzare l'effetto di questa sostituzione utilizzando tecniche di simulazione di dinamica molecolare (MD). L'analisi delle dinamiche ha mostrato che la sostituzione R10T provoca una diversa disposizione spaziale del capping domain di Mre11 che ha come effetto quello di aumentare l'attività di svolgimento del DNA. Questa aumentata predisposizione di Mre11-R10T ad aprire la doppia elica di DNA garantendo così alle nucleasi un più facile accesso, spiega l'aumentata attività di resection osservata sperimentalmente.
In this work I used different computational techniques, in particular QM and MM methods, to study the relation between structure and activity. The first project was the characterization under structural point of view of the FeMo-cofactor presents in the active site of nitrogenase. My work was focused on the structural characterization of the key cofactor state E4 , that is now accepted as the intermediate that bind N2 in the active site. For this purpose, DFT simulations have been performed on the FeMo-co active site considering all the iron oxidation state assignments proposed in literature ([ 5Fe III :2Fe II ]; [ 3Fe III :4Fe II ]; [ 1Fe III :6Fe II ]) with protonated and not protonated forms of homocitrate. Only in cases of [ 3F e III :4F e II ] with protonated form of homocitrate the optimized E 4 structure shows two hydrides bridged to the same iron atoms (Fe6 and Fe7 ) in an orthogonal fashion, consistently with experimental data. I have also studied the electronic properties of E 4 intermediate in order to evaluate the spin alignment of ferrous and ferric sites, with peculiar focus on the two Fe ions sharing hydride ligands. DFT broken symmetry calculations showed a parallel spin-coupling pattern for the two iron atoms (↑Fe6 : ↑Fe7 ). Another issue investigated is the effect of the substitution of Ar g 96 → Gln on the resting state of the MoFe protein. Experimental evidences show that in presence of this mutation the active site of nitrogenase is able to (non-covalently) bind acetylene even in the resting state (whereas WT protein cannot). To rationalize the observed behaviour, DFT approach has been used to evaluate the binding energy between acetylene and FeMo-co active site in the cases of mutant and wild type systems. In line with experiments, the binding energy results more favourable for the mutant than for the wild type enzyme. An explanation can be found in the different chemico-physical properties of the region available for acetylene binding in cofactor proximity: in presence of the substitution Ar g 96 → Gln the pocket environment becomes more hydrophobic than in the wild type case. Considering the nonpolar nature of acetylene, the mutated MoFe protein has therefore active site features that make it more suitable for substrate binding. The second project was the characterization of the hydrolytic mechanism of a de novo design peptide TRIL9CL23H (an analogue of carbonic anhydrase) trough DFT calculations and identification of a possible binding site for substrates using flexible docking. The mechanism proposed for TRIL9CL23H is a merge of α-anhydrase and β-anhydrase ones: as in α-anhydrase there is a zinc coordinated by three histidine residues and a water molecule, as in β-anhydrase a glutamate residue accepts the water’s proton from active site. Contrary to carbonic anhydrase, results show that the rate determining step is the release of the products from active site and not the activation of water. According to docking results the binding site is located on the side of proteins close to the active site. The last project has investigated the effect of the R10T amino acid change in the Mre11 protein which causes a rapid DSB resection. The wild type and the mutated dimer of Mre11 were studied trough molecular dynamics simulations. Results pointed out that the R10T substitution alters the mobility of the capping domain of Mre11, movement that is implicated in DNA unwinding. This mutation augments the rotation of the capping domain that can lead to a better DNA unwind activity. In this way nucleases that are involved in the DSB resection can have a better access to the DNA.

Virtual screening is a fast, low cost method to identify potential small molecule therapeutics from large chemical databases for the vast amount of target proteins emerging from the life sciences and bioinformatics. In this work, we applied several conventional and newly developed virtual screening approaches to identify novel non-steroidal ligands for the human and zebrafish sex hormone binding globulin (SHBG). The ‘benchmark set of steroids’ is a set of steroids with known affinities for human SHBG that has been widely used for validation in the development of different virtual screening methods. We have updated this data set by including additional steroidal SHBG ligands and by modifying the predicted binding orientations of several benchmark steroids in the SHBG binding site based on the use of an improved docking protocol and information from recent crystallographic data. The new steroid binding orientations and the expanded version of the benchmark set was then used to create new in silico models which were applied in virtual screening to identify high-affinity non-steroidal human SHBG ligands from a large chemical database. Anthropogenic compounds with the capacity to interact with the steroid-binding site of SHBG pose health risks to humans and other vertebrates including fish. We constructed a homology model of SHBG from zebrafish and applied virtual screening to identify ligands for zebrafish SHBG from a set of 80 000 existing commercial substances, many of which can be exposed to the aquatic environment. Six hits from this in silico screen were tested experimentally for zebrafish SHBG binding and three of them, hexestrol, 4-tert-octylcatechol, dihydrobenzo(a)pyren-7(8H)-one demonstrated micromolar binding affinity for the zebrafish SHBG. These findings demonstrate the feasibility of using virtual screening to identify anthropogenic compounds that may disrupt or highjack functionally important protein:ligand interactions. Studies applying this new computational toxicology method could increase the awareness of hazards posed by existing commercial chemicals at relatively low cost.

Recent studies have identified the Class B g-protein coupled receptor (GPCR) pituitary adenylate cyclase activating polypeptide type 1 (PAC1R) as a key component in physiological stress management. Over-activity of neurological stress response systems due to prolonged or extreme exposure to traumatic events has led researchers to investigate PAC1R inhibition as a possible treatment for anxiety disorders such as post-traumatic stress disorder (PTSD). In 2008, Beebe and coworkers identified two such small molecule hydrazide antagonists and a general pharmacaphore for PAC1R inhibition. However, a relative dearth of information about Class B GPCRs in general, and PAC1R in specific, has significantly hindered progress toward the development of small molecule antagonists of PAC1R. The recent crystallization of the homologically similar glucagon receptor (GCGR) by Siu and coworkers in 2013, also a Class B receptor, has provided an experimentally resolved template from which to base computationally derived models of PAC1R. Initially, this research was focused towards synthesizing small molecule antagonists for PAC1R which were to be biologically screened via a qualitative western blot assay followed by a radioisotope binding assay for those hydrazides exhibiting down-stream signaling inhibitory capabilities. However, the resolution of the GCGR crystal structure shifted research objectives towards developing a homology model of PAC1R and evaluating that computationally created model with Beebe's known small molecule antagonists. Created using academic versions of on-line resources including UniProtKB, Swiss-Model and Maestro, a homology model for PAC1R is presented here. The model is validated and evaluated for the presence of conserved Class B GPCR residues and motifs, including expected disulfide bridges, a conserved tyrosine residue, a GWGxP motif, a conserved glutamic acid residue and the extension of the transmembrane helix 1 (TM1) into the extra-cellular domain. Having determined this virtual PAC1R an acceptable model, ligand docking studies of known antagonists to the receptor were undertaken using AutoDock Vina in conjunction with AutoDock Tools and PyMol. Computational docking results were evaluated via comparison of theoretical binding affinity results to Beebe's experimental data. Based on hydrogen bonding capabilities, several residues possibly key to the ligand-receptor binding complex are identified and include ASN 240, TYR 241 and HIST 365. Although the docking software does not identify non-bonding interactions other than hydrogen-bonding, the roles of additional proposed binding pocket residues are discussed in terms of hydrophobic interactions, π-π interactions and halogen bonding. These residues include TYR 161, PHE 196, VAL 203, PHE 204, ILE 209, LEU 210, VAL 237, TRP 297, PHE 362 and LEU 386. Although theoretical in nature, this reported homology modeling and docking exercise details a proposed binding site that may potentially further the development of drugs designed for the treatment of PTSD.

The study of the 3D structural details of protein interactions is essential to understand biomolecular functions at the molecular level. In this context, the limited availability of experimental structures of protein-protein complexes at atomic resolution is propelling the development of computational docking methods that aim to complement the current structural coverage of protein interactions. One of these docking approaches is pyDock, which uses van der Waals, electrostatics, and desolvation energy to score docking poses generated by a variety of sampling methods, typically FTDock or ZDOCK. The method has shown a consistently good prediction performance in community-wide assessment experiments like CAPRI or CASP, and has provided biological insights and insightful interpretation of experiments by modeling many biomolecular interactions of biomedical and biotechnological interest. Here, we describe our approach using pyDock for the structural modeling of protein assemblies and the application of its modules to different biomolecular recognition phenomena, such as modeling of binding mode, interface, and hot-spot prediction, use of restraints based on experimental data, the inclusion of low-resolution structural data, binding affinity estimation, or modeling of homo- and hetero-oligomeric assemblies. The integration of template-based and ab initio docking approaches is emerging as the optimal strategy for modeling protein complexes and multi-molecular assemblies. We will review the new methodological advances on ab initio docking and integrative modeling. The seventh CAPRI edition imposed new challenges to the modeling of protein-protein complexes, such as multimeric oligomerization, protein-peptide, and protein-oligosaccharide interactions. Many of the proposed targets needed the efficient integration of rigid-body docking, template-based modeling, flexible optimization, multi-parametric scoring, and experimental restraints. This was especially relevant for the multi-molecular assemblies proposed in the CASP13-CAPRI46 joint rounds. We will present the results for the 7th CAPRI edition and CAPRI Round 46, the third joint CASP-CAPRI protein assembly prediction challenge. One of the known potential effects of disease-causing amino acid substitutions in proteins is to modulate protein-protein interactions (PPIs). To interpret such variants at the molecular level and to obtain useful information for prediction purposes, it is important to determine whether they are located at protein-protein interfaces, which are composed of two main regions, core and rim, with different evolutionary conservation and physicochemical properties. Here we have performed a structural, energetics and computational analysis of interactions between proteins hosting mutations related to diseases detected in newborn screening. Interface residues were classified as core or rim, showing that the core residues contribute the most to the binding free energy of the PPI. Disease-causing variants are more likely to occur at the interface core region rather than at the interface rim (p < 0.0001). In contrast, neutral variants are more often found at the interface rim or at the non-interacting surface rather than at the interface core region. We also found that arginine, tryptophan, and tyrosine are over-represented among mutated residues leading to disease. These results can enhance our understanding of disease at the molecular level and thus contribute towards personalized medicine by helping clinicians to provide adequate diagnosis and treatments. The phenotypic effects of non-synonymous genetic variations leading or predisposing to disease can be rationalized on the basis of the functional and structural impact in the mutated protein, including the perturbation of the interaction network and molecular pathways in which such protein is involved. Therefore, understanding these effects at the molecular level is essential to build accurate disease models and to achieve higher precision in diagnosis and therapeutic intervention. In this context, we can computationally characterize the effect of pathological mutations on specific protein-protein interactions ("edgetic"), based on their protein structure, if available, or on docking models. Protein-protein interactions that are clearly stabilized or destabilized by these mutations can be potential targets for therapeutic intervention. We have analyzed the predicted energetical effect of mutations on PPIs by applying a variety of computing methods to model the mutation and compute the change in binding affinity (FoldX, mCSM, pyDock combined to SCWRL3). We validate the predictive energetical impact through experimental mutations contained in SKEMPI 2.0 and apply these approaches in pathological and neutral single amino acid variants (SAVs) afterward (from ClinVar/Humsavar and gnomAD). Based on this, we have identified pathological mutations that clearly affect the analyzed interactions by stabilizing or destabilizing them. As discussed above, protein-protein interactions are important for biological processes and pathological situations and are attractive targets for drug discovery. However, rational drug design targeting protein-protein interactions is still highly challenging. Hot-spot residues are seen as the best option to target such interactions, but their identification requires detailed structural and energetic characterization, which is only available for a tiny fraction of protein interactions. This thesis covers a variety of computational methods that have been reported for the energetic analysis of protein-protein interfaces in search of hot-spots, and the structural modeling of protein-protein complexes by docking. This can help to rationalize the discovery of small-molecule inhibitors of protein-protein interfaces of therapeutic interest. Computational analysis and docking can help to locate the interface, molecular dynamics can be used to find suitable cavities, and hot-spot predictions can focus the search for inhibitors of protein-protein interactions. A major difficulty for applying rational drug design methods to protein-protein interactions is that in the majority of cases the complex structure is not available. Fortunately, computational docking can complement experimental data. An interesting aspect to explore in the future is the integration of these strategies for targeting PPIs with large-scale mutational analysis.
El estudio de los detalles estructurales en 3D de las interacciones de proteínas es esencial para comprender las funciones biomoleculares a nivel molecular. En este contexto, la disponibilidad limitada de estructuras experimentales de complejos proteína-proteína en resolución atómica está impulsando el desarrollo de métodos de acoplamiento computacional que apuntan a complementar la cobertura estructural actual de interacciones de proteínas. Uno de estos enfoques de acoplamiento es pyDock, que utiliza van der Waals, electrostática y energía de desolvatación para puntuar las poses de acoplamiento generadas por una variedad de métodos de muestreo, generalmente FTDock o ZDOCK. El método ha demostrado un rendimiento de predicción consistentemente bueno en experimentos de evaluación de toda la comunidad como CAPRI o CASP, y ha proporcionado conocimientos biológicos e interpretación profunda de experimentos al modelar muchas interacciones biomoleculares de interés biomédico y biotecnológico. Aquí, describimos nuestro enfoque utilizando pyDock para el modelado estructural de ensamblajes de proteínas y la aplicación de sus módulos a diferentes fenómenos de reconocimiento biomolecular, como el modelado del modo de unión, la interfície y la predicción de puntos calientes, el uso de restricciones basadas en datos experimentales, la inclusión de datos estructurales de baja resolución, estimación de afinidad de unión o modelado de ensamblajes homo- y hetero- oligoméricos. La integración de enfoques de acoplamiento ab initio y basados en plantillas está emergiendo como la estrategia óptima para modelar complejos de proteínas y ensamblajes multimoleculares. Revisaremos los nuevos avances metodológicos sobre el acoplamiento ab initio y el modelado integrativo. La séptima edición de CAPRI impuso nuevos desafíos al modelado de complejos proteína-proteína, como la oligomerización multimérica, las interacciones proteína-péptido y proteína-oligosacárido. Muchos de los objetivos propuestos necesitaban la integración eficiente de acoplamiento de cuerpo rígido, modelado basado en plantillas, optimización flexible, puntuación multiparamétrica y restricciones experimentales. Esto fue especialmente relevante para los conjuntos multimoleculares propuestos en las rondas conjuntas CASP13-CAPRI46. Presentaremos los resultados para la séptima edición de CAPRI y la Ronda 46 de CAPRI, el tercer desafío de predicción del ensamblaje de proteínas CASP-CAPRI conjunto. Uno de los efectos potenciales conocidos de las sustituciones de aminoácidos que causan enfermedades en las proteínas es modular las interacciones proteína-proteína (PPI). Para interpretar tales variantes a nivel molecular y obtener información útil con fines de predicción, es importante determinar si están ubicadas en interfaces proteína-proteína, que se componen de dos regiones principales, núcleo y borde, con diferente conservación evolutiva y diferentes propiedades fisicoquímicas. En la tesis, hemos realizado un análisis estructural, energético y computacional de interacciones entre proteínas que albergan mutaciones relacionadas con enfermedades detectadas en el cribado neonatal. Los residuos de interfície se clasificaron como núcleo o borde, lo que demuestra que los residuos del núcleo son los que más contribuyen a la energía libre de unión del PPI. Es más probable que las variantes que causan enfermedades se produzcan en la región del núcleo de la interfície que en el borde de la interfície de la proteína (p <0,0001). Por el contrario, las variantes neutrales se encuentran más a menudo en el borde de la interfície o en la superficie que no interactúa en lugar de en la región del núcleo de la interfície. También encontramos que la arginina, el triptófano y la tirosina están sobrerrepresentados entre los residuos mutados que conducen a la enfermedad. Estos resultados pueden mejorar nuestra comprensión de las enfermedades a nivel molecular y, por lo tanto, contribuir a la medicina personalizada al ayudar a los médicos a proporcionar diagnósticos y tratamientos adecuados. Los efectos fenotípicos de variaciones genéticas no-sinónimas que conducen o predisponen a la enfermedad pueden racionalizarse sobre la base del impacto funcional y estructural en la proteína mutada, incluida la perturbación de la red de interacción y las vías moleculares en las que participa dicha proteína. Por lo tanto, comprender estos efectos a nivel molecular es esencial para construir modelos de enfermedad precisos y lograr una mayor precisión en el diagnóstico y la intervención terapéutica. En este contexto, podemos caracterizar computacionalmente el efecto de mutaciones patológicas en interacciones proteína-proteína específicas ("edgetic"), en base a su estructura proteica, si está disponible, o en modelos de acoplamiento. Las interacciones proteína-proteína que están claramente estabilizadas o desestabilizadas por estas mutaciones pueden ser objetivos potenciales para la intervención terapéutica. Hemos analizado el efecto energético predicho de las mutaciones en los PPI aplicando una variedad de métodos informáticos para modelar la mutación y calcular el cambio en la afinidad de unión (FoldX, mCSM, pyDock combinados con SCWRL3). Validamos el impacto energético predictivo a través de mutaciones experimentales contenidas en SKEMPI 2.0 y aplicamos estos enfoques en variantes patológicas y neutrales de un solo aminoácido (SAV en inglés) posteriormente (de ClinVar / Humsavar y gnomAD). En base a esto, hemos identificado mutaciones patológicas que inciden claramente en las interacciones analizadas estabilizándolas o desestabilizándolas. Como se ha discutido, las interacciones proteína-proteína son importantes para procesos biológicos y situaciones patológicas y son objetivos atractivos para el descubrimiento de fármacos. Sin embargo, el diseño racional de fármacos dirigidos a las interacciones proteína-proteína sigue siendo un gran desafío. Los residuos de puntos calientes se consideran la mejor opción para apuntar a tales interacciones, pero su identificación requiere una caracterización estructural y energética detallada, que solo está disponible para una pequeña fracción de interacciones de proteínas. Esta tesis doctoral cubre una variedad de métodos computacionales que han sido reportados para el análisis energético de interfícies proteína-proteína en la búsqueda de puntos calientes y el modelado estructural de complejos proteína-proteína por acoplamiento. Esto puede ayudar a racionalizar el descubrimiento de inhibidores de moléculas pequeñas de interfícies proteína-proteína de interés terapéutico. El análisis computacional y el acoplamiento pueden ayudar a localizar la interfície, la dinámica molecular se puede utilizar para encontrar cavidades adecuadas y las predicciones de puntos calientes pueden enfocar la búsqueda de inhibidores de interacciones proteína-proteína. Una dificultad importante para aplicar métodos racionales de diseño de fármacos a las interacciones proteína-proteína es que en la mayoría de los casos no se dispone de la estructura compleja. Afortunadamente, el acoplamiento computacional puede complementar los datos experimentales. Un aspecto interesante para explorar en el futuro es la integración de estas estrategias para apuntar a las interacciones proteína-proteína con análisis mutacionales a gran escala.

Integrins are an important class of transmembrane receptors that relay signals bidirectionally across the plasma membrane, regulating several cell functions and playing a key role in diverse pathological processes. Specifically, integrin subtype αIIbβ3 is involved in thrombosis and stroke, while subtypes αvβ3 and α5β1 play an important role in angiogenesis and tumor progression. They therefore emerged as attractive pharmacological targets. In the past decades several peptides and peptidomimetics targeting these proteins and based on the integrin recognition motif RGD (Arg-Gly-Asp) have been developed, whereby their affinity and selectivity for a specific integrin subtype have been fine-tuned by modulation of RGD flanking residues, by cyclization or by introduction of chemical modifications. Thus far, the design and development of RGD-based cyclopeptides have been mainly based on empirical approaches, requiring expensive and time-consuming synthesis campaigns. In this field, the employment of computational tools, that could be valuable to accelerate the drug design and optimization process, has been limited by the inherent difficulties to predict in silico the three-dimensional structure and the inhibitory activity of cyclopeptides. However, recent improvements in both computational resources and in docking and modeling techniques are expected to open new perspectives in the development of cyclopeptides as modulators of protein-protein interactions and, particularly, as integrin inhibitors. Within this PhD project, I have investigated the applicability of computational techniques in predicting and rationalizing how the environment of the recognition-motif in cyclopeptides (i.e. flanking residues and introduction of chemical modification) could influence their integrin affinity and selectivity. These features can regulate integrin affinity both by favoring direct interactions with the receptor and/or by modulating the three-dimensional conformation properties of the recognition motif. To take into account both these aspects, I have proposed and optimized a multi-stage computational protocol in which an exhaustive conformational sampling of the investigated cyclopeptides is followed by docking calculations and re-scoring techniques. Specifically: i) the exhaustive sampling could be achieved by using Metadynamics in its Bias Exchange variant (BE-META), an enhanced sampling technique which represents a valuable methodology for the acceleration of rare events, allowing to cross the high free energy barriers characteristic of cyclopeptides and providing reliable estimations of the populations of the accessible conformers. ii) The docking calculations, complemented with the re-scoring technique MM-GB/SA (Molecular Mechanics Generalized Born Surface Area) and the cluster analysis of the decoy poses, allow to evaluate the ability of each peptide to engage interactions with the receptors and to rank the docking poses according to their binding ability; iii) a joint analysis of the previous outcomes results in a reliable ranking of cyclopeptides based on their binding affinity and in the rationalization of their structure-activity relationship. This computational protocol has been exploited in two different applications, illustrated within the thesis. In the first application the protocol has been applied to rationalize how the introduction of chemical modifications, specifically backbone N-methylation, impacts on the equilibrium conformation and consequently on the integrin affinity of five RGD containing cyclic hexapeptides, which were previously generated by the group of professor Kessler to modulate their selectivity for αIIbβ3 integrin. The study revealed that backbone N-methylation affects the preferences of the φ dihedral angle of the methylated residue, specifically favoring the adoption of additional conformations, characterized by a 180° twist of the peptide bond plane preceding the methylated residue. These twists of dihedral angles were found to have relevant consequences on the cyclopeptides conformation, influencing the formation of intra-molecular hydrogen bonds as well as some structural features which are known to be fundamental in integrin binding. Both structural analysis and docking calculations allowed to identify the “bioactive” conformation (i.e. an extended RGD conformation able to recapitulate the canonical electrostatic and the additional stabilizing hydrophobic interactions). Of note, the cyclopeptides that are pre-organized, already in their free state, in this bioactive conformation are the ones displaying the best αIIbβ3 binding affinity in terms of IC50 values, confirming that pre-organization of cyclopeptides in solution can strongly affect their binding strength to the receptor and demonstrating that the knowledge of their conformational equilibrium is fundamental to provide reliable affinity predictions. In the second application, I have focused my attention on cyclopeptides harboring a recently discovered integrin recognition motif: isoDGR (isoAsp-Gly-Arg), deriving from the spontaneous deamidation of NGR (Asp-Gly-Arg) sequence present in integrin natural ligands. As a preliminary step, I have systematically tested the accuracy of eight Molecular Mechanics force fields in reproducing the equilibrium properties of isoDGR-based cyclopeptides, for which NMR experiments have been acquired. The comparison between simulated and NMR-derived data (i.e. chemical shifts and J scalar couplings) revealed that, while most of the investigated force fields can properly reproduce the equilibrium conformational properties of cyclic peptides, only two of them (i.e. the AMBER force fields ff99sb-ildn and ff99sb*-ildn) are able to recover the NMR observables characteristics of the non-standard residue isoAspartate with an accuracy close to the systematic uncertainty. Overall, these results suggest that the transferability of force field parameters to non standard amino acids is not straightforward. However, two force fields allowed to obtain a satisfactory accuracy and have been therefore employed for the subsequent investigation. I thus applied the computational protocol to rationalize the diverse selectivity and affinity profiles for integrins αvβ3 and α5β1, both related to cancer, displayed by three isoDGR-based cyclic hexapeptides. These molecules differ in the residues flanking the isoDGR motif and show appealing tumor-homing properties; specifically it has been shown that one of these, c(CGisoDGRG), can be coupled with human serum albumin through a chemical linker to be used as a drug delivery agent for functionalized gold nanoparticles. Herein, I investigated the role of the chemical linker in improving affinity and selectivity of c(CGisoDGRG) for αvβ3. The application of the multi-stage protocol allowed to propose an explanation for the different selectivity profiles displayed by these molecules, where the direct interactions engaged by the flanking residues and/or their steric hindrance seem to be largely responsible for the observed different affinities. As a last result, through the combination of MD and NMR techniques, I demonstrated that the chemical linker improved the αvβ3 affinity of c(CGisoDGRG) by engaging direct interactions with the receptor and I proposed two possible complex models, which well-reproduce data from Saturation Transfer Difference experiments. Overall, in this PhD work I have shown that the combination of different computational techniques, BE-META, docking and MM-GB/SA re-scoring, could be a reliable approach to perform structure-activity relationship studies in cyclopeptides. Specifically, the proposed protocol is able to predict the influence of the recognition motif environment (i.e. chemical modification and flanking residues) on integrin affinities. These two features regulate integrin affinity differently: the first one by conformational modulation of the recognition motif, the second by engaging direct interactions with the receptor. Of note, the approach can deal with both these mechanisms of affinity modulation. We expect that the protocol herein described could be used in future to screen novel peptides library or to complement biochemical experiments during the drug optimization stages, assisting organic chemists in the design of more effective integrin-targeting peptides.

In modern drug discovery, lead discovery is a term used to describe the overall process from hit discovery to lead optimisation, with the goal being to identify drug candidates. This can be greatly facilitated by the use of computer-aided (or in silico) techniques, which can reduce experimentation costs along the drug discovery pipeline. The range of relevant techniques include: molecular modelling to obtain structural information, molecular dynamics (which will be covered in Chapter 2), activity or property prediction by means of quantitative structure activity/property models (QSAR/QSPR), where machine learning techniques are introduced (to be covered in Chapter 1) and quantum chemistry, used to explain chemical structure, properties and reactivity. This thesis is divided into five parts. Chapter 1 starts with an outline of the early stages of drug discovery; introducing the use of virtual screening for hit and lead identification. Such approaches may roughly be divided into structure-based (docking, by far the most often referred to) and ligand-based, leading to a set of promising compounds for further evaluation. Then, the use of machine learning techniques, the issue of which will be frequently encountered, followed by a brief review of the "no free lunch" theorem, that describes how no learning algorithm can perform optimally on all problems. This implies that validation of predictive accuracy in multiple models is required for optimal model selection. As the dimensionality of the feature space increases, the issue referred to as "the curse of dimensionality" becomes a challenge. In closing, the last sections focus on supervised classification Random Forests. Computer-based analyses are an integral part of drug discovery. Chapter 2 begins with discussions of molecular docking; including strategies incorporating protein flexibility at global and local levels, then a specific focus on an automated docking program – AutoDock, which uses a Lamarckian genetic algorithm and empirical binding free energy function. In the second part of the chapter, a brief introduction of molecular dynamics will be given. Chapter 3 describes how we constructed a dataset of known binding sites with co-crystallised ligands, used to extract features characterising the structural and chemical properties of the binding pocket. A machine learning algorithm was adopted to create a three-way predictive model, capable of assigning each case to one of the classes (regular, orthosteric and allosteric) for in silico selection of allosteric sites, and by a feature selection algorithm (Gini) to rationalize the selection of important descriptors, most influential in classifying the binding pockets. In Chapter 4, we made use of structure-based virtual screening, and we focused on docking a fluorescent sensor to a non-canonical DNA quadruplex structure. The preferred binding poses, binding site, and the interactions are scored, followed by application of an ONIOM model to re-score the binding poses of some DNA-ligand complexes, focusing on only the best pose (with the lowest binding energy) from AutoDock. The use of a pre-generated conformational ensemble using MD to account for the receptors' flexibility followed by docking methods are termed “relaxed complex” schemes. Chapter 5 concerns the BLUF domain photocycle. We will be focused on conformational preference of some critical residues in the flavin binding site after a charge redistribution has been introduced. This work provides another activation model to address controversial features of the BLUF domain.

Positive strand RNA viruses, which include hepatitis C virus (HCV), human immunodeficiency virus (HIV, and Bovine Viral Diarrhea Virus (BVDV), are known to create havoc for humans and animal health alike. Although vaccines have helped to control several of the most important viral pathogens, there is currently little prospect of an effective vaccine for either HCV or HIV. These pathogens infect ~170 million and ~40 million people worldwide, respectively, hastening the need for effective antiviral drugs. Likewise BVDV infects domesticated livestock causing significant economic losses worldwide. The development of new, effective antiviral compounds for combating these debilitating human (HIV and HCV) and animal pathogen (BVDV) is therefore of paramount importance, and is the focus of this thesis. Herein, polymerases of three positive strand RNA viruses, viz HCV, BVDV and HIV have been targeted with the goal of improving the efficacy of antivirals against wide range of resistant mutations. Lack of effective therapies for these viral infections as most of the established treatments are not always effective or well tolerated, highlights an urgent need for further refinement and development of antiviral drugs. It is not only the specific need that has inspired this work but also the idea to test and develop protocols that might enable a more rational structurebased drug design to be performed by keeping a tradeoff among rapidity, accuracy, and efficacy. Traditional methods for general drug discovery typically include evaluating random compound libraries for activity in relevant cellfree or cellbased assays. Success in antiviral development has emerged from the discovery of more focused libraries that provide clues about structureactivity relationships. Combining these with more recent approaches including structural biology and computational modeling can work efficiently to hasten discovery of active molecules. The ability to design drugs interfering with the progression of infection of virus comes with i)the knowledge of pathological, cellular and molecular mechanism involved in the disease; and ii)the identification of macromolecule (i.e possible drug target) involved in pathological pathways, their 3D structures and their functions. The biological activity of drug molecules is dependent on the threedimensional arrangement of its functional groups, which specifically bind to their target. Consequently, the structural information of the target protein is essential in drug development. Proteins are dynamic molecules and often undergo conformational change upon ligand binding. The flexible loop regions and in general the flexibility of the structure have a critical functional role in enzymes, but those features and their connection with the functionality of protein are hard to retrieve from xray, NMR techniques and cryoEM techniques. Being aware of the importance of the relationship structurefunction and structureactivity at large, i.e., including dynamics and interactions with solvent, in our work we are trying to address some of the relevant problems of drug development; basic key determinants in proteinligand stability, mechanism of inhibition, why and how, flexibility and collective motion of the protein is essential part in improvement of rational drug design, how mutation renders the protein resistant again potent drugs; the effect of resistance mutation on the flexibility and stability of protein, what is the mechanism of drug resistance, change in energetics consequences, affecting the conformation in wild and mutant systems. Various biophysical techniques of the computational arsenal we have applied have provided huge wealth of information related to protein dynamics and proteinligand recognition. These methods have grown in their effectiveness not only by offering a deeper understanding of the basic science, the biological events and molecular interactions that define a target for therapeutic intervention, but also because of advances in algorithms, representations, and mathematical procedures for studying such processes. This work represents the application of several computational techniques, such as docking, molecular dynamics, algorithms to calculate free energy of binding of ligands into the binding pocket (ex MMPBSA) and algorithms to study rare events (for ex. binding and unbinding of ligand from the binding site, Metadynamics) to explore, at microscopic level, the key pattern of interaction between protein and ligand, to understand the effect of mutations, to get an insight of the full docking and undocking path and to calculate binding energetics. 4

Topoisomerase I (TopoI) is an essential DNA-targeting enzyme which alters the supercoiling of DNA through a concerted process of breaking and rejoining of a DNA strand, thereby controlling the DNA topology required for replication and transcription. TopoI mediates relaxation of supercoiled DNA by creating a transient single-strand break in the DNA duplex that originates from a transesterification reaction involving a nucleophilic attack by the active-site tyrosine (Tyr723) hydroxyl group on a DNA phosphodiester bond situated at the site of cleavage. The resulting 3'-phosphotyrosine enzyme-DNA complex (“covalent binary complex” or “cleavable complex”) is then reversed, after DNA relaxation through strand passage, when the released 5'-OH of the nicked strand reattacks the phosphotyrosine intermediate in a second transesterification reaction. These events result in the relaxation of the DNA structure, which is required during transcription or replication. TopoI is a specific target for the pentacyclic alkaloid Camptothecin (CPT), which was first isolated in 1966 from extracts of Camptotheca Acuminata, and its derivatives, known as TopoI poisons. These molecules block DNA religation, thus converting TopoI into a DNA-damaging agent. In the presence of a TopoI poison a ternary complex between DNA, an intercalator and topoisomerase is formed. Such a ternary complex is more stable than the DNA-TopoI associate, which may lead to an enhanced lifetime of the initially cleaved DNA. As a consequence of the misalignment of the free 5-hydroxyl group and the scissile tyrosine-DNA bond due to the presence of the drug, the religation of the broken strand cannot take place, i.e. the strand break persists. This activates a complex sequence of intacellular responses, that ultimately lead to cell death by apoptosis. Therefore, molecules which form such stabilized ternary complexes with DNA and TopoI exhibit a high potential as DNA-targeting anticancer drugs. Camptothecin was early shown to be clinically problematic because, in addition to its negligible water solubility, its active “ring-closed” ?-hydroxylactone form is rapidly converted under physiological conditions to the open carboxylate form, which is inactive and readily binds to human serum albumin, making it inaccessible for cellular uptake. To date, only two semisynthetic analogs of Camptothecin (Topotecan and Irinotecan) have been approved by FDA for the clinical treatment of the ovarian, small cell-lung and colon cancers. Solving crystal structures of Top1 in complex both with Camptothecin and Topotecan and with structurally different molecules (indolocarbazoles and indoloquinolines) has significantly increased the amount of structural information about the interaction between the Top1-DNA binary complex and the poison molecule. This encouraged the application of structure-based drug design to investigate and rationalize the activity of Top1 poisons and to rationally design new potential anticancer drugs. Thus, we considered the possibility to find a new pharmacophore exploiting all the available crystal structures to find the set of structural features which are in common to all of the five TopoI poisons. In order to obtain new derivatives easier to be synthesized while maintaining the pharmacophoric features required for the formation of a stable TopoI-DNA-poison ternary complex, the synthesis of simplified CPT derivatives was proposed. It is possible to suppose that a pentacyclic scaffold is not necessarily required for a molecule to effectively bind the TopoI-DNA binary complex, providing that it maintains an aromatic planar system for the intercalation and appropriate functional groups to interact with specific residues of TopoI. On the basis of this assumption, several scaffolds were designed and docking studies with different search algorithms were performed to predict the ability of simplified analogs to form stable ternary complex with TopoI and DNA. Moreover, we performed a molecular docking analysis on a series of new 5-substituted CPT derivatives to achieve more insight into the interactions of the new compounds with the binary TopoI-DNA complex. The introduction of substituents in position 5 of the CPT scaffold results in derivatives with reduced cytotoxic potency compared with the reference drugs. In general, the presence of a small lipophilic substituent is well tolerated while the modification with hydrophilic groups results in reduced affinity for the binding site. The presence of bulky groups is detrimental for cytotoxic potency because of the loss of important stabilizing interactions. Computational results indicated as a general feature that the 5-?-epimer is better tolerated in the binding site compared to the 5-?-epimer, resulting in higher biological activity. Recently, a new series of 16a-thio-CPT derivatives has been synthesized. Biological assays for this class of compounds revealed that 16a-thio-CPT derivatives have higher anticancer activity compared to their oxo-analogs, both in vitro and in vivo, and are potent TopoI inhibitors. In order to find a rational explanation to their remarkable activity, a computational analysis was performed, both for predicting elements of their pharmacokinetic behavior and for describing their binding mode through a docking analysis. Docking results did not reveal striking differences between thio- and oxo-derivatives in terms of binding mode and docking score. On the contrary, the predicted values for features such as lipophilicity, water solubility and cell permeability indicated that these properties may be responsible for a significantly different pharmacokinetic profile for the two classes of compounds. For the docking analysis of 5-substituted and 16a-thio CPT derivatives we used the quantum mechanics (QM)-polarized ligand docking (QPLD) protocol (Schrödinger software suite) with an aim to improve accuracy in docking calculations. Traditional docking methods employ an approximated physical chemistry-based representation of protein-ligand interactions, obtaining charges from a molecular mechanics force-field. The QPLD protocol uses an ab initio methodology to calculate ligand charges within the protein environment, thus taking into account the charge polarization induced by the protein environment. The work on 5-substituted and 16a-thio CPT derivative was sponsored by the pharmaceutical company Indena S.p.A., Milano, Italy. The work was carried out in collaboration with the group of Professor Arturo Battaglia (Centro Nazionale Richerche Bologna, Italy) for the synthesis of the compounds, and with the group of Professor Franco Zunino (Istituto Nazionale per lo Studio e la Cura dei Tumori, Milano, Italy), that was concerned with biological assays.
La Topoisomerasi I eucariotica (TopoI) è un enzima essenziale che modifica il grado di superavvolgimento del DNA mediante un processo di rottura e successiva ricongiunzione di uno dei due filamenti del DNA, regolandone lo stato topologico richiesto nei processi di replicazione e di trascrizione. Il rilassamento del DNA superavvolto indotto dalla TopoI avviene attraverso la rottura transitoria di un filamento per mezzo di una reazione di transesterificazione. Quest'ultima coinvolge l'attacco nucleo lo da parte del gruppo ossidrilico di un residuo di tirosina (Tyr723) del sito attivo dell'enzima ad un legame fosfodiesterico del DNA a livello del sito di taglio, con formazione di un complesso covalente 3'-fosfotirosina TopoI-DNA (noto anche come cleavable complex). La rotazione del filamento tagliato intorno a quello intatto rende possibile il rilassamento della molecola di DNA. Il complesso binario covalente subisce in seguito un attacco nucleofilo a livello della fosfotirosina da parte dell'estremità libera 5'-OH del lamento rotto, che permette la riformazione del legame fosfodiesterico. La TopoI è il target specifico dell'alcaloide pentaciclico Camptotecina (CPT), isolato per la prima volta nel 1966 dalla corteccia di Camptotheca Acuminata, e dei suoi derivati, noti come veleni di TopoI. Queste molecole bloccano il processo di ricongiunzione del filamento di DNA tagliato, convertendo la TopoI in un agente dannoso per il DNA stesso. Queste molecole si intercalano tra le basi del DNA a livello del sito di taglio dando luogo alla formazione di un complesso ternario con la TopoI e il DNA. La maggiore stabilità del complesso ternario rispetto al complesso binario TopoI-DNA risulta in un aumento del tempo di vita del DNA tagliato. Poiché la presenza del veleno induce il disallineamento dell'estremità libera 5'-OH del filamento rotto con il legame fosfotirosinico, il processo di ricongiunzione non può avvenire. Questo porta all'attivazione di una serie complessa di segnali intracellulari, il cui risultato ultimo è la morte della cellula per apoptosi. Di conseguenza, le molecole in grado di formare complessi ternari stabili hanno attività antitumorale. L'utilizzo clinico della CPT è ostacolato sia dalla sua scarsissima idrosolubilità, sia perchè la sua forma attiva lattonica “chiusa” è rapidamente convertita, in condizioni fisiologiche, nella forma carbossilica “aperta”, che è inattiva e si lega in elevata percentuale all'albumina serica. Attualmente solo due derivati semisintetici della CPT (Topotecan ed Irinotecan) sono utilizzati nella pratica clinica per il trattamento di cancro ovarico, polmonare e colo-rettale. La recente risoluzione, mediante cristallografia a raggi X, della struttura tridimensionale di TopoI in complesso sia con la Camptotecina e il Topotecan che con derivati strutturalmente diversi (indolocarbazolici ed isochinolinici), ha fornito chiarimenti sulle caratteristiche dell'interazione enzima-inibitore. Tali informazioni costituiscono elementi molto utili nello studio dei veleni di TopoI con un approccio di tipo computazionale il cui scopo è comprendere in che modo queste molecole interagiscono con il loro target e progettare nuovi nuovi derivati come potenziali farmaci ad attività antitumorale. È stato quindi possibile individuare un nuovo farmacoforo sfruttando le cinque strutture cristallogra che disponibili, al fine di rintracciare un set di caratteristiche strutturali comuni a tutti i veleni di TopoI. Con lo scopo di disporre di nuovi derivati più semplici da ottenere dal punto di vista della sintesi ma che mantengano tutte le caratteristiche farmacoforiche richieste per la formazione di complessi ternari stabili con TopoI e DNA, è stata proposta la sintesi di analoghi “semplificati” della CPT. Si può infatti ipotizzare che una molecola non debba necessariamente possedere uno scaffold pentaciclico per legarsi al complesso binario TopoI-DNA, purchè mantenga un sistema planare aromatico per intercalare il DNA e opportuni gruppi funzionali in grado di interagire con specifici aminoacidi dell'enzima. Sulla base di questa assunzione, diversi possibili scaffold sono stati disegnati e sottoposti a studi di docking con diversi protocolli di ricerca per predire la loro capacità di formare stabili complessi ternari con TopoI e DNA. È stato inoltre condotto uno studio di docking molecolare per una serie di nuovi derivati della CPT sostituiti in posizione 5 (5-derivati) per descrivere la loro modalità di interazione con il complesso TopoI-DNA. In generale, l'introduzione di sostituenti in posizione 5 ha come risultato una diminuzione dell'attività citotossica rispetto alla CPT. La presenza di sostituenti lipofili di piccole dimensioni sembra essere tollerata all'interno del sito di binding, mentre l'introduzione di gruppi idro lici dà luogo ad una diminuzione dell'affinità. L'introduzione di gruppi stericamente ingombranti provoca la perdita di importanti interazioni al'interno del sito di legame. I risultati computazionali indicano come caratteristica generale che l'epimero ? dei 5-derivati ha un'affinità maggiore per il sito di binding rispetto all'epimero ?. Recentemente sono stati sintetizzati una serie di 16a-tio-CPT analoghi. I saggi biologici condotti su questi composti hanno rivelato che i tio-derivati sono potenti inibitori della TopoI ed hanno attività antitumorale maggiore rispetto ai loro oxo-analoghi, sia in vitro che in vivo. Al fine di razionalizzare la maggiore attività dei tio-analoghi rispetto ai derivati tradizionali, è stato eseguito uno studio computazionale volto sia a predire proprietá chimico-fisiche rilevanti per la descrizione del possibile profilo farmacocinetico delle nuove molecole, sia a descriverne la modalità di legame nel sito del complesso binario TopoI-DNA tramite studi di docking. I risultati di docking non hanno messo in evidenza differenze di rilievo tra i tio- e gli oxo-derivati in termini di modalità di binding e di docking score. Al contrario, i valori predetti per proprietà come la lipofilicità, la solubilità in acqua e la permeabilità cellulare indicano che il profilo farmacocinetico delle due classi di composti potrebbe essere significativamente diverso. Per le simulazioni di docking condotte sui 5-CPT-derivati e sui tio-CPT-derivati è stato utilizzato il protocollo quantum mechanics (QM)-polarized ligand docking (QPLD) (implementato nella suite Schrödinger) con lo scopo di aumentare l'accuratezza dei calcoli di docking. I metodi tradizionali utilizzati per il docking molecolare utilizzano una rappresentazione chimico-fisica approssimata delle interazioni proteina-ligando, dove le cariche atomiche sono calcolate in base al campo di forza. Il protocollo QPLD utilizza invece una metodologia ab initio per il calcolo delle cariche del ligando all'interno della cavità della proteina, tenendo in considerazione la polarizzazione di carica indotta dagli aminoacidi del sito di legame sul ligando. Il lavoro relativo ai 5-CPT-derivati e ai tio-CPT-derivati à stato supportato dall'azienda farmaceutica Indena S.p.A., Milano, Italia. Il lavoro è stato condotto in collaborazione con il gruppo del Professor Arturo Battaglia (Centro Nazionale Richerche Bologna) per la sintesi dei composti e con il gruppo del Professor Franco Zunino (Istituto Nazionale per lo Studio e la Cura dei Tumori, Milano) per i saggi biologici.

Per glaucoma si intende un gruppo di neuropatie oculari caratterizzate da una perdita progressiva e irreversibile della vista. Attualmente le terapie farmacologiche mirano a far diminuire la pressione intraoculare (IOP), un fattore di rischio modificabile al fine di ridurre la progressione del glaucoma. Come ligando naturale, la melatonina ha breve tempo di emivita e bassa biodisponibilità, in questa ricerca, le tecniche in silico permettono di cercare composti bioattivi con alta affinità per i recettori della melatonina (MTs) e bassa tossicità che potrebbero trovare impiego immediato nel trattamento del glaucoma. Per raggiungere tale obiettivo, a partire da agonisti noti, sono state studiate le basi molecolari del riconoscimento del ligando e dell'attività agonistica per gli MT umani (hMT). L'agomelatina, bioisostere della melatonina, è stata scelta per la sua elevata stabilità e attività melatonergica. Inoltre, questo composto ha un'elevata affinità e attività antagonista per il recettore della serotonina umano 5HT2C, la cui inibizione sembra essere coinvolta nel controllo della IOP. Partendo da questi risultati, siamo andati oltre lo screening in silico di enormi librerie di composti (farmaci sia naturali, psicoattivi che nootropi) verso i target macromolecolari hMTs e h5HT2C, selezionando quelli più promettenti che sono stati poi testati in vivo su modello animale di ratto. Nuove librerie molecolari, basate sulle caratteristiche di agonisti noti della melatonina, sono state realizzate mediante ricerca farmacoforica. Infine, con l'obiettivo di studiare una formulazione liposomiale in grado di aumentare la stabilità della melatonina in soluzione, sono stati effettuati ulteriori studi in silico per valutare l'entità dell'assorbimento della melatonina all'interno del doppio strato lipidico in soluzioni saline a diverse concentrazioni già utilizzate in campo oftalmologico.
Glaucoma is a group of ocular neuropathies characterized by a progressive and irreversible loss of vision. Currently, pharmacological therapies aim to decrease intraocular pressure (IOP), a modifiable risk factor to lower glaucoma progression. As the natural ligand melatonin has a short half-life and low bioavailability, in this research, in silico techniques were used to detect bioactive compounds with high affinity for melatonin receptors (MTs) and low toxicity that could be immediately used for the treatment of glaucoma. To reach that aim, starting from known agonistic compounds, it was investigated the molecular basis of the ligand recognition and agonistic activity for hMTs. The melatonin bioisostere agomelatine have been chosen because of its high stability and melatonergic activity. In addition, this compound has high affinity and antagonistic activity for the human serotonin receptor 5HT2C, whose inhibition seems to be involved in the IOP control. Starting from these results, we went further to in silico screening huge libraries of compounds (both natural, psychoactive and nootropic drugs) towards the macromolecular targets hMTs and h5HT2C, selecting the most promising ones that were then tested in vivo on animal model of rat. Pharmacophoric search was realized to build new molecular libraries based on the features of well-known melatonergic agonists. Finally, in order to study a liposomal formulation able to increase the melatonin stability in solution, further in silico studies were carried out in order to evaluate the extent of melatonin absorption inside lipid bilayer in different saline solutions and concentrations just used in ophthalmologic field.

Over a billion individuals worldwide suffer from neglected diseases. This equates to approximately one-sixth of the human population. These infections are often endemic in remote tropical regions of impoverished populations where vectors can flourish and infected individuals cannot be effectively treated due to a lack of hospitals, medical equipment, drugs, and trained personnel. The few drugs that have been approved for the treatments of such illnesses are not widely used because they are riddled with inadequate implications of cost, safety, drug availability, administration, and resistance. Hence, there exists an eminent need for the design and development of improved new therapeutics. Influential world-renowned scientists in the Consortium for Parasitic Drug Development (CPDD) have preformed extensive biological testing for compounds active against parasites that cause neglected diseases. These data were acquired through several collaborations and found applicable to computational studies that examine quantitative structure-activity relationships through the development of predictive models and explore structural relationships through docking. Both of these in silico tools can contribute to an understanding of compound structural importance for specific targets. The compilation of manuscripts presented in this dissertation focus on three neglected diseases: trypanosomiasis, Chagas disease, and leishmaniasis. These diseases are caused by kinetoplastid parasites Trypanosoma brucei, Trypanosoma cruzi, and Leishmania spp., respectively. Statistically significant predictive devices were developed for the inhibition of the: (1) T. brucei P2 nucleoside transporter, (2) T. cruzi parasite at two temperatures, and (3) two species of Leishmania. From these studies compound structural importance was assessed for the targeting of each parasitic system. Since these three parasites are all from the Order Kinetoplastida and the kinetoplast DNA has been determined a viable target, compound interactions with DNA were explored to gain insight into binding modes of known and novel compounds.

This thesis contains three main sections. In the first section, we examine methodologies to discriminate Type II protein kinase inhibitors from the Type I inhibitors. We have studied the structure of 55 Type II kinase inhibitors and have notice specific descriptive geometric features. Using this information we have developed a pharmacophore and a shape based screening approach. We have found that these methods did not effectively discriminate between the two inhibitor types used independently, but when combined in a consecutive way – pharmacophore search first, then shape based screening, we have found a method that successfully filtered out all Type I molecules. The effect of protonation states and using different conformer generators were studied as well. This method was then tested on a freely available database of decoy molecules and again shown to be discriminative. In the second section of the thesis, we implement and assess swarm-based docking methods. We implement a repulsive particle swarm optimization (RPSO) based conformational search approach into Autodock 3.05. The performance of this approach with different parameters was then tested on a set of 51 protein ligand complexes. The effect of using different factoring for the cognitive, social and repulsive terms and the importance of the inertia weight were explored. We found that the RPSO method gives similar performance to the particle swarm optimization method. Compared to the genetic algorithm approach used in Autodock 3.05, our RPSO method gives better results in terms of finding lower energy conformations. In the final, third section we have implemented a Monte Carlo (MC) based conformer searching approach into Gaussian03. This enables high level quantum mechanics/molecular mechanics (QM/MM) potentials to be used in docking molecules in a protein active site. This program was tested on two Zn2+ ion-containing complexes, carbonic anhydrase II and cytidine deaminase. The effects of different QM region definitions were explored in both systems. A consecutive and a parallel docking approach were used to study the volume of the active site explored by the MC search algorithm. In case of the carbonic anhydrase II complex, we have used 1,2-difluorobenzene as a ligand to explore the favourable interactions within the binding site. With the cytidine deaminase complex, we have evaluated the ability of the approach to discriminate the native pose from other higher energy conformations during the exploration of the active site of the protein. We find from our initial calculations, that our program is able to perform a conformational search in both cases, and the effect of QM region definition is noticeable, especially in the description of the hydrophobic interactions within the carbonic anhydrase II system. Our approach is also able to find poses of the cytidine deaminase ligand within 1 Å of the native pose.

Structure-based drug design and virtual screening are areas of computational medicinal chemistry that use 3D models of target proteins. It is important to develop better methods in this field with the aim of increasing the speed and quality of early stage drug discovery. The first part of this thesis focuses on the application of structure-based drug design in the search for inhibitors for the protein 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR), one of the enzymes in the DOXP/MEP synthetic pathway. This pathway is found in many bacteria (such as Mycobacterium tuberculosis) and in the parasite Plasmodium falciparum. In order to evaluate and improve current virtual screening methods, a benchmarking data set was constructed using publically available high-throughput screening data. The exercise highlighted a number of problems with current data sets as well as with the use of publically available high-throughput screening data. We hope this work will help guide further development of well designed benchmarking data sets for virtual screening methods. Conformal prediction is a new method in the computer-aided drug design toolbox that gives the prediction range at a specified level of confidence for each compound. To demonstrate the versatility and applicability of this method we derived models of skin permeability using two different machine learning methods; random forest and support vector machines.

Computational methods have been extensively developed at various levels of approximation in recent years to model biomolecular interactions and for rational drug design. This research work aims to explore the feasibility of using quantum mechanical (QM) methods within the two broad categories of in silico ligand-based and structure-based drug design. First, density functional theory at the M06L level of theory was employed to examine structure-activity relationships of boron-based heterocyclic compounds, anti-inflammatory inhibitors targetting the interleukin-1β (IL-1β) cytokine. Our findings from computed energies and shapes of the molecular orbitals provide understanding of electronic effects associated with the inhibitory activity. We also found that the boron atom, specifically its electrostatic polarity, appears to be essential for the anti-IL-1β activity as evidenced by the biological assay of non-boron analogues selected from the ligand-based virtual screening results. Secondly, we aimed to dock ligands at the active sites of zinc-containing metalloproteins with reasonable computational cost and with accuracy. Therefore, an in-house docking scheme based on a Monte Carlo sampling algorithm using the semiempirical PM6/AMBER force field scoring function was compiled for the first time within the Gaussian 09 program. It was applied to four test cases, docking to cytidine deaminase and human carbonic anhydrase II proteins. The docking results show the method’s promise in resolving false-positive docking poses and improving the predicted binding modes over a conventional docking scheme. Finally, semiempirical QM methods which include dispersion and hydrogen-bond corrections were assessed for modelling conformations of β-cyclodextrin (βCD) and their adsorption on graphene. The closed in vacuo βCD cccw conformer was found to be in the lowest energy, in good agreement with previous ab initio QM studies. DFTB3, PM6-DH2 and PM7 methods were applied to model the intermolecular interactions of large βCD/graphene complexes, over a thousand atoms in size. We found that the binding preference of βCD on graphene is in a closed conformation via its C2C3 rim, agreeing with reported experimental and computational findings.

The voltage activated potassium channel Kv1.3 is an important therapeutic target due to its vital role in the treatment of autoimmune diseases. A family of plant toxins identified recently but their binding affinities for Kv1 channels have not been characterized. Here we have studied the binding of four plant toxins EgK1, SmK1, JrK1, and CcK1 with Kv1 channels using molecular docking and molecular dynamics (MD) simulations. The EgK1 toxin has been found as a potent blocker of Kv1.3 and highly selective for Kv1.3 over Kv1.1. Umbrella sampling MD simulations are performed for the Kv1.3–EgK1 complex to calculate binding free energy. The binding modes of all toxins are compared to design an EgK1 analog with higher affinity for Kv1.3 which could be a potential therapeutic lead for the treatment of autoimmune diseases. Herceptin is used in the treatment of breast cancer for patients whose tumors excessively express the HER2 protein. Here we have performed a computational study of HER2–herceptin-Fab complex and identified three mutations on herceptin to increase its binding affinity for HER2. Using MD simulations and Free energy perturbation method, D28R mutation found as the most promising one for improving the binding affinity of herceptin for HER2. Aggregation of protein is an undesired phenomena which reduces the antibody activity, so it is vitally necessary to understand the mechanism of aggregation at a molecular level for designing aggregation resistant versions of therapeutic antibodies. Here, we use higher temperature MD simulations to identify the aggregation prone regions in an antibody with a crystal structure (1HZH). The role of glycosylation in 1HZH is found to increase the overall stability of the antibody. The identified aggregation prone regions are modified via mutations to increase the aggregation resistance of the antibody.

The scope of this work focuses on computationally modeling compounds with protein structures. While the impetus of drug discovery is the innovation of new therapeutic molecules, it also involves distinguishing molecules that would not be an effective drug. This can be achieved by inventing new tools or by refining old tools. Virtual screening (VS, also called docking), the computational modeling of a molecule in a receptor structure, is a staple in predicting a molecule's affinity for an intended target. In our Virtual Target Screening system (also called inverse-docking), VS is used to find high-affinity targets, which can potentially explain absorption, distribution, metabolism, and excretion (ADME) of a molecule of interest in the human body. The next project, low-mode docking (LD), attempts to improve VS by incorporating protein flexibility into traditional docking where a static receptor structure has potential to produce poor results due to incorrectly predicted ligand poses. Finally, VS, performed mostly on small molecules, is scaled up to cyclic peptides by employing Monte Carlo simulations and molecular dynamics to mimic the steps of small molecule VS. The first project discussed is Virtual Target Screening (also called inverse-docking) where a small molecule is virtually screened against a library of protein structures. Predicting receptors to which a synthesized compound may bind would give insights to drug repurposing, metabolism, toxicity, and lead optimization. Our protocol calibrates each protein entry with a diverse set of small molecule structures, the NCI Diversity Set I. Our test set, 20 kinase inhibitors, was predicted to have a high percentage of kinase "hits" among approximately 1500 protein structures. Further, approved drugs within the test set generally had better rates of kinase hits. Next, normal mode analysis (NMA), which can computationally describe the fundamental motions of a receptor structure, is utilized to approach the rigid body bias problem in traditional docking techniques. Traditional docking involves the selection of a static receptor structure for VS; however, protein structures are dynamic. Simulation of the induced fit effect in protein-ligand binding events is modeled by full articulation of the approximated large-scale low-frequency normal modes of vibration, or "low-modes," coupled with the docking of a ligand structure. Low-mode dockings of 40 cyclin dependent 2 (CDK2) inhibitors into 54 low-modes of CDK2 yielded minimum root-mean-square deviation (RMSD) values of 1.82 – 1.20 Å when compared to known coordinate data. The choice of pose is currently limited to docking score, however, with ligand pose RMSD values of 3.87 – 2.07 Å. When compared to corresponding traditional dockings with RMSD values of 5.89 – 2.33 Å, low-mode docking was more accurate. The last discussion involves the rational docking of a cyclic peptide to the murine double minute 2 (MDM2) oncoprotein. The affinity for a cyclic peptide (synthesized by Priyesh Jain, McLaughin Lab, University of South Florida), PJ-8-73, in MDM2 was found to be within an order of magnitude of a cyclic peptide from the Robinson Lab at the University of Zurich in Switzerland. Both are Β-hairpin cyclic peptides with IC50 values of 650 nm and 140 nm, respectively. Using the co-crystalized structure of the Robinson peptide (PDB 2AXI), we modeled the McLaughlin peptide based on an important interaction of the 6-chloro-tryptophan residue of the Robinson peptide occupying the same pocket in MDM2 as the tryptophan residue by the native p53 transactivation helical domain. By preserving this interaction in initial cyclic peptide poses, the resulting pose of PJ-8-73 structure in MDM2 possessed comparable active site residue contacts and surface area. These protocols will aid medical research by using computer technology to reduce cost and time. VTS utilizes a unique structural and statistical calibration to virtually assay thousands of protein structures to predict high affinity binding. Determining unintended protein targets aids in creating more effective drugs. In low-mode docking, the accuracy of virtual screening was increased by including the fundamental motions of proteins. This newfound accuracy can decrease false negative results common in virtual screening. Lastly, docking techniques, usually for small molecules, were applied to larger peptide molecules. These modifications allow for the prediction of peptide therapeutics in protein-protein interaction modulation, a growing interest in medicine. Impactful in their own ways, these procedures contribute to the discovery of drugs, whether they are small molecules or cyclic peptides.

Kaposi’s sarcoma-associated herpesvirus, is a double-stranded DNA γ - herpesvirus and the main causative agent of Kaposi’s sarcoma (KS). γ - herpesviruses undergo both lytic and latent replication cycles; and encode proteins that modulate host transcription at the RNA level, by inducing decay of certain mRNAs. Here we describe a mechanism that allows the viral endo-/exonuclease SOX to recognise mRNA targets on the basis of an RNA motif and fold. To induce rapid RNA degradation by subverting the main host mRNA degradation pathway SOX was shown to directly bind Xrn1. This may shed light as to how some viruses evade the host antiviral response and how mRNA degradation processes in the eukaryotic cell are involves in this.

The rationalisation of drug potency using three-dimensional structures of protein-ligand complexes is a central paradigm in medicinal research. For over two decades, a major goal has been to find the rules that accurately relate the structure of any protein-ligand complex to its affinity. Addressing this problem is of great concern to the pharmaceutical industry, which uses virtual screens to computationally assay up to many millions of compounds against a protein target. A fast and trustworthy affinity estimator could potentially streamline the drug discovery process, reducing reliance on expensive wet lab experiments, speeding up the discovery of new hits and aiding lead optimization. Water plays a critical role in drug-protein interactions. To address the often ambiguous nature of water in binding sites, a water placement method was developed and found to be in good agreement with X-ray crystallography, neutron diffraction data and molecular dynamics simulations. The method is fast and has facilitated a large scale study of the statistics of water in ligand binding sites, as well as the creation of models pertaining to water binding free energies and displacement propensities, which are of particular interest to medicinal chemistry. Structure-based scoring functions employing the explicit water models were developed. Surprisingly, these attempts were no more accurate than the current state of the art, and the models suffered from the same inadequacies which have plagued all previous scoring functions. This suggests a unifying cause behind scoring function inaccuracy. Accordingly, mathematical analyses on the fundamental uncertainties in structure-based modelling were conducted. Using statistical learning theory and information theory, the existence of inherent errors in empirical scoring functions was proven. Among other results, it was found that even the very best generalised structure-based model is significantly limited in its accuracy, and protein-specific models are always likely to be better. The theoretical framework developed herein hints at modelling strategies that operate at the leading edge of achievable accuracy.

Al fine di identificare nuovi farmaci anti-HIV capaci di superare i problemi legati alla resistenza, è stato condotto uno studio teorico combinando l’analisi strutturale sui modelli cristallografici della trascrittasi inversa (RT), i dati clinici relativi ai residui conservati dell’RT ed un’innovativa metodica computazionale basata sulle mappe di GRID. Tale analisi ha permesso di riprodurre i risultati clinici e di evidenziare le conseguenze delle mutazioni nella fase di ricognizione. Inoltre l’approccio computazionale ha portato all’identificazione di un modello farmacoforico utile per la progettazione di nuovi inibitori dell’RT. E’ stato riscontrato che la presenza del polimorfismo I135T nei pazienti NNRTI-naïve correlasse in modo significativo con la mutazione K103N nei casi di fallimento agli NNRTI, suggerendo così che la sostituzione I135T rappresenti un punto cruciale per l’evoluzione della resistenza agli NNRTI. Le simulazioni di dinamica molecolare (MD) hanno mostrato che la mutazione I135T contribuisce alla stabilizzazione della chiusura della tasca di legame degli NNRTI indotta dalla K103N in seguito alla riduzione della distanza ed all’aumento del numero di legami idrogeno tra l’Asn103 e la Tyr188. Inoltre è stata valutata l’influenza di due mutazioni associate a resistenza, L33F e L76V, presenti a livello della proteasi (PR) di HIV-1 rispetto alla ricognizione molecolare del Lopinavir (LPV) e dell’Atazanavir (ATV). L’analisi delle energie di interazione ottenute in seguito alla MD ha rivelato che la mutazione L33F determina una riduzione delle interazioni tra il ligando ed il recettore, dell’affinità di legame e della stabilità del dimero per entrambi gli inibitori della PR. In presenza della mutazione L76V, il LPV ha mostrato una minore affinità di legame ed un ridotto network di legami idrogeno, mentre i complessi con l’ATV hanno rivelato una migliore affinità, un effetto stabilizzante a livello dell’interfaccia del dimero e più efficaci interazioni ligando-recettore, in accordo con i dati di ipersuscettibilità. Al fine di valutare la stabilità del 6-helix bundle, sono state studiate le proprietà conformazionali della glicoproteina gp41 in presenza delle mutazioni associate a resistenza all’enfuvirtide V38A ed N140I. Le simulazioni di MD hanno mostrato che la copresenza delle mutazioni V38A+N140I è in grado di abolire l’interazione stabilita tra i residui 38 e 145, che risulta fondamentale per la stabilizzazione del 6-helix bundle.
In order to discover novel selective anti-HIV resistance-evading drugs, a theoretical study was carried out combining structural analysis of RT crystallographic models, clinical data about RT conserved residues and an innovative computational method based on GRID maps. Such analysis allowed to reproduce clinical results and to highlight the consequences of the mutations in the recognition step. Moreover the computational approach generated a pharmacophore model useful for the design of novel RT inhibitors. The presence of the I135T polymorphism in NNRTI-naive patients significantly correlated with the appearance of K103N in cases of NNRTI failure, suggesting that I135T may represent a crucial determinant of NNRTI resistance evolution. Molecular Dynamics simulations (MD) showed that I135T can contribute to the stabilization of the K103N-induced closure of the NNRTI binding pocket by reducing the distance and increasing the number of hydrogen bonds between 103N and 188Y. In addition the influence of two drug resistance-associated mutations, L33F and L76V, of HIV-1 PR has been evaluated with respect to lopinavir (LPV) and atazanavir (ATV) molecular recognition. The evaluation of the interaction energies after the MD revealed that L33F substitution is related to reduced host/guest interactions, decreased affinity and to a dimer destabilizing effect for both PR inhibitors. In presence of L76V mutation, LPV showed a lowered binding affinity and a reduced hydrogen bonding network, while ATV complexes revealed a more productive binding affinity, increased host/guest interactions and dimer stabilizing effects, in agreement with hyper susceptibility data. With the aim to estimate the stability of its 6-helix bundle, the gp41 conformational properties were investigated in presence of V38A and N140I, known enfuvirtide resistance-associated mutations. MD showed that the co-presence of V38A+N140I abolished the interaction between residue 38 and 145 important for the 6-helix-bundle stabilization.

Flies are a widely distributed pest insect that poses a significant threat to food security. Flight is essential for the dispersal of the adult flies to find new food sources and ideal breeding spots. The supply of metabolic fuel to power the flight muscles of insects is regulated by adipokinetic hormones (AKHs). The fruit fly, Drosophila melanogaster, the flesh fly, Sarcophaga crassipalpis, and the oriental fruit fly, Bactrocera dorsalis all have the same AKH that is present in the blowfly, Phormia terraenovae; this AKH has the code-name Phote-HrTH. Binding of the AKH to the extracellular binding site of a G protein-coupled receptor causes its activation. In this thesis, the structure of Phote-HrTH in SDS micelle solution was determined using NMR restrained molecular dynamics. The peptide was found to bind to the micelle and be reasonably rigid, with an S 2 order parameter of 0.96. The translated protein sequence of the AKH receptor from the fruit fly, Drosophila melanogaster, the flesh fly, Sarcophaga crassipalpis, and the oriental fruit fly, Bactrocera dorsalis were used to construct two models for each receptor: Drome-AKHR, Sarcr-AKHR, and Bacdo-AKHR. It is proposed that these two models represent the active and inactive state of the receptor. The models based on the crystal structure of the β-2 adrenergic receptor were found to bind Phote-HrTH with a predicted binding free energy of –107 kJ mol–1 for Drome-AKHR, –102 kJ mol–1 for Sarcr-AKHR and –102 kJ mol–1 for Bacdo-AKHR. Under molecular dynamics simulation, in a POPC membrane, the β-2AR receptor-like complexes transformed to rhodopsin-like. The identification and characterisation of the ligand-binding site of each receptor provide novel information on ligand-receptor interactions, which could lead to the development of species-specific control substances to use discriminately against these pest flies.

In this thesis, molecular dynamics simulations, molecular docking, and homology modeling methods have been used in combination to design possible inhibitors as well as to study the structural changes and function of target proteins related to diseases that today are in the spotlight of drug discovery. The inwardly rectifying potassium (Kir) channels constitute the first target in this study; they are involved in cardiac problems. On the other hand, tensin, a promising target in cancer research, is the second target studied here. The first chapter includes a brief update on computational methods and the current proposal of the combination of MD simulations and docking techniques, a procedure that is applied for the engineering of a new blocker for Kir2.1 ion channels and for the design of possible inhibitors for Tensin. Chapter two focuses in Kir ion channels that belong to the family of potassium-selective ion channels which have a wide range of physiological activity. The resolved crystal structure of a eukaryotic Kir channel was used as a secondary structure template to build the Kir-channels whose crystallographic structures are unavailable. Tertiapin (TPN), a 21 a.a. peptide toxin found in honey bee venom that blocks a type of Kir channels with high affinity was also used to design new Kir channel blockers. The computational methods homology modeling and protein-protein docking were employed to yield Kir channel-TPN complexes that showed good binding affinity scores for TPN-sensitive Kir channels, and less favorable for Kir channels insensitive to TPN block. The binding pocket of the insensitive Kir-channels was studied to engineer novel TPN-based peptides that show favorable binding scores via thermodynamic mutant-cycle analysis. Chapter three is focused on the building of homology models for Tensin 1, 2 and 3 domains C2 and PTP using the PTEN X-ray crystallographic structure as a secondary structure template. Molecular docking was employed for the screening of druggable small molecules and molecular dynamics simulations were also used to study the tensin structure and function in order to give some new insights of structural data for experimental binding and enzymatic assays. Chapter four describes the conformational changes of FixL, a protein of bradyrhizobia japonicum. FixL is a dimer known as oxygen sensor that is involved in the nitrogen fixation process of root plants regulating the expression of genes. Ligand behavior has been investigated after the dissociation event, also the structural changes that are involved in the relaxation to the deoxy state. Molecular dynamics simulations of the CO-bound and CO-unbound bjFixL heme domain were performed during 10 ns in crystal and solution environments then analyzed using Principal Component Analysis (PCA). Our results show that the diffusion of the ligand is influenced by internal motions of the bound structure of the protein before CO dissociation, implying an important role for Arg220. In turn, the location of the ligand after dissociation affects the conformational changes within the protein. The study suggests the presence of a cavity close to the methine bridge C of the heme group in agreement with spectroscopic probes and that Arg220 acts as a gate of the heme cavity.

Proteins are involved in almost all cell processes, with physical interaction between them being key to their function and dictated by its 3D structure. Hence, the study of protein-protein interactions and protein-protein binding is crucial to fully understand biological systems. In this thesis, we present V-D2OCK, a fast and accurate data-driven docking tool for high throughput prediction of the structure of protein complexes. We have also studied the conformational space of potential encounter complexes by means of non-specific decoys obtained by docking in order to develop BADock, an accurate binding affinity predictor from the unbound individual structures. Finally, we have published online an integrated and centralized resource (InteractoMIX) that allows to the research community an easy access to a compendium of bioinformatic web applications to study protein-protein interactions.
Les proteïnes estan implicades en gairebé tots els processos cel·lulars, amb la interacció física entre elles clau per la seva funció i dictada per la seva estructura 3D. Per tant, l’estudi de la unió i les interaccions proteïna-proteïna és crucial per entendre completament els sistemes biològics. En aquesta tesi, es presenta V-D2OCK, una eina de “docking” dirigit ràpida i precisa per predir l’estructura de complexes de proteïnes a gran escala. També hem estudiat l’espai conformacional de possibles complexes transitoris per mitjà de resultats de “docking” no específics per tal de desenvolupar BADock, un predictor d’energia d’unió a partir de les estructures individuals per separat. Finalment, hem publicat online un recurs integrat i centralitzat (InteractoMIX) que permet a la comunitat investigadora l’accés fàcil a un conjunt de aplicacions web de bioinformàtica per l’estudi de interaccions proteïna-proteïna.

Philosophiae Doctor - PhD
Given the significant disease burden caused by human coronaviruses, the discovery of an effective antiviral strategy is paramount, however there is still no effective therapy to combat infection. This thesis details the in silica exploration of ligand libraries to identify candidate lead compounds that, based on multiple criteria, have a high probability of inhibiting the 3 chymotrypsin-like protease (3CUro) of human coronaviruses. Atomistic models of the 3CUro were obtained from the Protein Data Bank or theoretical models were successfully generated by homology modelling. These structures served the basis of both structure- and ligand-based drug design studies. Consensus molecular docking and pharmacophore modelling protocols were adapted to explore the ZINC Drugs-Now dataset in a high throughput virtual screening strategy to identify ligands which computationally bound to the active site of the 3CUro . Molecular dynamics was further utilized to confirm the binding mode and interactions observed in the static structure- and ligand-based techniques were correct via analysis of various parameters in a IOns simulation. Molecular docking and pharmacophore models identified a total of 19 ligands which displayed the potential to computationally bind to all 3CUro included in the study. Strategies employed to identify these lead compounds also indicated that a known inhibitor of the SARS-Co V 3CUro also has potential as a broad spectrum lead compound. Further analysis by molecular dynamic simulations largely confirmed the binding mode and ligand orientations identified by the former techniques. The comprehensive approach used in this study improves the probability of identifying experimental actives and represents a cost effective pipeline for the often expensive and time consuming process of lead discovery. These identified lead compounds represent an ideal starting point for assays to confirm in vitro activity, where experimentally confirmed actives will be proceeded to subsequent studies on lead optimization.

In recent years, the potential benefits of high-throughput virtual screening to the drug discovery community have been recognized, bringing an increase in the number of tools developed for this purpose. These programs have to process large quantities of data, searching for an optimal solution in a vast combinatorial range. This is particularly the case for protein-ligand docking, since proteins are sophisticated structures with complicated interactions for which either molecule might reshape itself. Even the very limited flexibility model to be considered here, using ligand conformation ensembles, requires six dimensions of exploration - three translations and three rotations - per rigid conformation. The functions for evaluating pose suitability can also be complex to calculate. Consequently, the programs being written for these biochemical simulations are extremely resource-intensive. This work introduces a pure computer science approach to the field, developing techniques to improve the effectiveness of such tools. Their architecture is generalized to an abstract pattern of nested layers for discussion, covering scoring functions, search methods, and screening overall. Based on this, new stratagems for molecular docking software design are described, including lazy or partial evaluation, geometric analysis, and parallel processing implementation. In addition, a range of novel algorithms are presented for applications such as active site detection with linear complexity (PIES) and small molecule shape description (PASTRY) for pre-alignment of ligands. The various stratagems are assessed individually and in combination, using several modified versions of an existing docking program, to demonstrate their benefit to virtual screening in practical contexts. In particular, the importance of appropriate precision in calculations is highlighted.

Non-steroidal anti-inflammatory drugs (NSAIDs) are one of the most commonly used classes of drugs. They target the cyclooxygenases (COX) 1 and 2 to reduce the physiological responses of pain, fever, and inflammation. Due to their role in inducing angiogenesis, COX proteins have also been identified as targets in cancer therapies. In this thesis, I describe computational protocols of molecular docking, molecular dynamics simulations and free energy calculations. These methods were used in this thesis to determine structure-activity relationships of a diverse set of NSAIDs in binding to their target proteins COX-1 and 2. Binding affinities were calculated and used to predict the binding modes. Based on combinations of molecular dynamics simulations and free energy calculations, binding mechanisms of sub-classes of NSAIDs were also proposed. Two stable conformations of COX were probed to understand how they affect inhibitor affinities. Finally, a brief discussion on selectivity towards either COX isoform is discussed. These results will be useful in future de novo design and testing of third-generation NSAIDs.

L'objectiu d'aquesta tesi és desenvolupar i implementar un conjunt d'eines de mineria de dades per ajudar en la reconstrucció de circuits biològics a través de l'anàlisi i la integració de grans conjunts de dades biològiques. Aquests circuits són importants perquè regulen tots els processos que controlen la vida i la salut dels organismes. El treball principal de la tesis es centra en l'anàlisi de les dades bibliòmiques desenvolupant-se dues eines, Biblio-MetReS per la reconstrucció de xarxes de PPIs i la identificació dels processos en què intervenen aquestes xarxes, i CheNER per la identificació de noms de compostos químics. L'eina final desenvolupada es centra en la integració de mètodes per a l'anàlisi estructural i modelització de proteïnes amb mètodes d'acoblament per a la predicció de complexos físics de proteïna-proteïna.
El objetivo de esta tesis es desarrollar e implementar un conjunto de herramientas de minería de datos para ayudar en la reconstrucción de circuitos biológicos a través del análisis y la integración de grandes conjuntos de datos biológicos. Estos circuitos son importantes porque regulan todos los procesos que controlan la vida y la salud de los organismos. El trabajo principal de la tesis se centra en el análisis de los datos bibliómicos, desarrollándose con este fin dos herramientas diferentes, Biblio-MetReS para la reconstrucción de redes PPIs y la identificación de los procesos en que intervienen estas redes, y CheNER para la identificación de nombres de compuestos químicos. La herramienta final que he desarrollado se centra en la integración de métodos para el análisis estructural y modelado de proteínas con métodos de acoplamiento para la predicción de complejos físicos de proteína-proteína.
The aim of this thesis is the development and implementation of a set of data mining tools to aid in the reconstruction of biological circuits through analysis and integration of large biological datasets. These circuits are important because they regulate and maintain life and health in organisms. The main part of the thesis is focused on analyzing bibliomic data for which I develop two tools, Biblio-MetReS for the reconstruction of PPIs networks and to identify the processes in which the networks are involved, and CheNER for the identification of chemical compounds names. The final tool developed focuses on the integration of methods for structural analysis and modeling of proteins with docking methods for prediction of native protein-protein physical complexes.

Although the tremendous catalytic power of enzymes is widely recognized, their exact mechanisms of action are still a source of debate. In order to elucidate the origin of their power, it is necessary to look at individual residues and atoms, and establish their contribution to ligand binding, activation, and reaction. Given the present limitations of experimental techniques, only computational tools allow for such detailed analysis. During my PhD studies I have applied a variety of computational methods, reviewed in Chapter 2, to the study of two enzymes: DfrB dihydrofolate reductase (DHFR) and methyltetrahydrofolate: corrinoid/iron-sulfur protein methyltransferase (MeTr). ¶ The DfrB enzyme has intrigued microbiologists since it was discovered thirty years ago, because of its simple structure, enzymatic inefficiency, and its insensitivity to trimethoprim. This bacterial enzyme shows neither structural nor sequence similarity with its chromosomal counterpart, despite both catalysing the reduction of dihydrofolate (DHF) using NADPH as a cofactor. As numerous attempts to obtain experimental structures of an enzyme ternary complex have been unsuccessful, I combined docking studies and molecular dynamics simulations to produce a reliable model of the reactive DfrB•DHF•NADPH complex. These results, combined with published empirical data, showed that multiple binding modes of the ligands are possible within DfrB. ¶ Comprehensive sequence and structural analysis provided further insight into the DfrB family. The presence of the dfrB genes within integrons and their level of sequence conservation suggest that they are old structures that had been diverging well before the introduction of trimethoprim. Each monomer of the tetrameric active enzyme presents an SH3-fold domain; this is a eukaryotic auxiliary domain never found before as the sole domain of a protein, let alone as the catalytic one. Overall, DfrB DHFR seems to be a poorly adapted catalyst, a ‘minimalistic’ enzyme that promotes the reaction by facilitating the approach of the ligands rather than by using specific catalytic residues. ¶ MeTr initiates the Wood-Ljungdahl pathway of anaerobic CO2 fixation. It catalyses the transfer of the N5-methyl group from N5-methyltetrahydrofolate (CH3THF) to the cobalt centre of a corrinoid/iron-sulfur protein. For the reaction to occur, the N5 position of CH3THF is expected to be activated by protonation. As experimental studies have led to conflicting suggestions, computational approaches were used to address the activation mechanism. ¶ Initially, I tested the accuracy of quantum mechanical (QM) methods to predict protonation positions and pKas of pterin, folate, and their analogues. Then, different protonation states of CH3THF and active-site aspartic residues were analysed. Fragment QM calculations suggested that the pKa of N5 in CH3THF is likely to increase upon protein binding. Further, ONIOM calculations which accounted for the complete protein structure indicated that active-site aspartic residues are likely to be protonated before the ligand. Finally, solvation and binding free energies of several protonated forms of CH3THF were compared using the thermodynamic integration approach. Taken together, these preliminary results suggest that further work with particular emphasis on the protonation state of active-site aspartic residues is needed in order to elucidate the protonation and activation mechanism of CH3THF within MeTr.

In living systems, proteins usually team up into “molecular machinery” implementing several protein-to-protein physical contacts – or protein-protein interactions (PPIs) – to exert biological effects at both cellular and systems levels. Deregulations of protein-protein contacts have been associated with a huge number of diseases in a wide range of medical areas, such as oncology, cancer immunotherapy, infectious diseases, neurological disorders, heart failure, inflammation and oxidative stress. PPIs are very complex and usually characterised by specific shape, size and complementarity. The protein interfaces are generally large, broad and shallow, and frequently protein-protein contacts are established between non-continuous epitopes, that conversely are dislocated across the protein interfaces. For this reason, in the past two decades, PPIs were thought to be “undruggable” targets by the scientific research community with scarce or no chance of success. However, in recent years the Medicinal Chemistry frontiers have been changing and PPIs have gained popularity amongst the research groups due to their key roles in such a huge number of diseases. Until recently, PPIs were determined by experimental evidence through techniques specifically developed to target a small group of interactions. However, these methods present several limitations in terms of high costs and labour- and time-wasting. Nowadays, a large number of computational methods have been successfully applied to evaluate, validate, and deeply analyse the experimentally determined protein interactomes. In this context, a high number of computational tools and techniques have been developed, such as methods designed to construct interaction databases, quantum mechanics and molecular mechanics (QM/MM) to study the electronic properties, simulate chemical reactions, and calculate spectra, and all-atom molecular dynamics simulations to simulate temporal and spatial scales of inter- and intramolecular interactions. These techniques have allowed to explore PPI networks and predict the related functional features. In this PhD work, an extensive use of computational techniques has been reported as valuable tool to explore protein-protein interfaces, identify their hot spot residues, select small molecules and design peptides with the aim of inhibiting six different studied PPIs. Indeed, in this thesis, a success story of in silico approaches to PPI study has been described, where MD simulations, docking and pharmacophore screenings led to the identification of a set of PPI modulators. Among these, two molecules, RIM430 and RIM442, registered good inhibitory activity with IC50 values even within the nanomolar range against the interaction between MUC1 and CIN85 proteins in cancer disease. Furthermore, computational alanine scanning, all-atom molecular dynamics simulations, docking and pharmacophore screening were exploited to (1) rationally predict three potential interaction models of NLRP3PYD-ASCPYD complex involved in inflammatory and autoimmune diseases; (2) identify a potentially druggable region on the surface of SARS-CoV-2 Spike protein interface and select putative inhibitors of the interaction between Spike protein and the host ACE2 receptor against COVID-19 (CoronaVIrus Disease 2019); (3) investigate intramolecular modifications as a consequence of a point mutation on C3b protein (R102G) associated with the age-related macular degeneration (AMD) disease; (4) design non-standard peptides to inhibit transcriptional events associated with HOX-PBX complex involved in cancer diseases; and (5) to optimise a patented peptide sequence by designing helix-shaped peptides embedded with the hydrogen bond surrogate approach to tackle cocaine abuse relapses associated with Ras-RasGRF1 interaction. Although all the herein exploited techniques are based on predictive calculations and need experimental evidence to confirm the findings, the results and molecular insights retrieved and collected show the potential of the computer-aided drug design applied to the Medicinal Chemistry, guaranteeing labour- and time-saving to the research groups. On the other hand, computing ability, improved algorithms and fast-growing data sets are rapidly fostering advances in multiscale molecular modelling, providing a powerful emerging paradigm for drug discovery. It means that more and more research efforts will be done to invest in novel and more precise computational techniques and fine-tune the currently employed methodologies.