A doença de Chagas é uma infecção causada pelo protozoário intracelular Trypanosoma cruzi. Atualmente 7 a 8 milhões de pessoas encontram-se infectadas, e há 25 milhões de pessoas em áreas de risco de contaminação. A cada ano ocorrem 56.000 novos casos e, aproximadamente, 12.000 mortes por complicações oriundas da doença. É endêmica em 21 países da América do Sul, e pode ser encontrada, também, na América do Norte e Europa devido a processos migratórios. Somente dois fármacos estão disponíveis para o tratamento da doença de Chagas, o nifurtimox e o benznidazol, que são ativos somente na fase aguda e causam sérios efeitos adversos. Diante deste panorama, é eminente a necessidade de novos antichagásicos. A enzima cruzaína é a principal cisteíno-protease presente no T. cruzi, é importante para a sobrevivência, diferenciação e entrada do parasita no hospedeiro, se apresentando como um excelente alvo biológico na busca de novos quimioterápicos. Derivados de semicarbazona, tais como o nitrofural e o hidroximetilnitrofural demonstraram atividade inibitória da cruzaína, sendo considerados protótipos na busca de antichagásicos. Utilizando estratégias modernas de planejamento de fármacos por meio da integração entre técnicas computacionais, modelagem molecular e docking, e experimentais, síntese e ensaios biológicos, realizou-se neste trabalho o planejamento, síntese e avaliação biológica de bioisósteros do nitrofural como candidatos à antichagásicos. Aplicou-se estudos de modelagem molecular e docking para 10 compostos derivados de aminoguanidina, semi e tiossemicarbazona; observamos nesses estudos que os compostos contendo tiossemicarbazona apresentaram resultados mais favoráveis ao mecanismo de ação proposto, o qual sugere-se um ataque nucleofílico do resíduo de Cys25 presente no sítio catalítico da enzima cruzaína à tiocarbonila presente nesses compostos. Obtiveram-se através da síntese, 6 compostos caracterizados por RMN 1H e 13C. Tais compostos foram submetidos a ensaios de inibição da cruzaína, sendo que os derivados 6, 9 e 10, apresentaram um perfil de inibição favorável em dose de 100 µM, com valores entre 70 e 75% de inibição. Em ensaio de inibição de crescimento celular em formas epimastigotas do T. cruzi o composto 9 apresentou um IC50 de 19,8 µM, sendo o melhor protótipo para desenvolvimento de um novo agente antichagásico. De uma maneira geral os resultados obtidos nos ensaios biológicos corroboram com os dados apresentados na modelagem molecular, uma vez que os compostos contendo a cadeia lateral tiossemicarbazona mostraram melhores resultados em ambos os testes, demonstrando que a integração entre técnicas computacionais e experimentais se apresenta como uma excelente estratégia na busca de novos agentes antichagásicos. / Chagas disease is an infection caused by the intracellular protozoan Trypanosoma cruzi. Currently 7 million to 8 million people are infected, and there are 25 million people in areas at risk of contamination, with 56,000 new cases each year and roughly 12,000 deaths are related to Chagas complications. It is endemic in 21 countries in South America, and can also be found in North America and Europe due to migration processes. Only two drugs are available for treatment of Chagas disease, nifurtimox and benznidazole, which are active only in the acute phase and cause serious adverse effects. Against this background, it is imminent need for new antichagasic. The enzyme cruzain is the major cysteine protease present in the T. cruzi, is important for the survival, differentiation and entry of the parasite in the host, presenting itself as an excellent biological target in the search for new chemotherapeutic agents. Semicarbazone derivatives, such as nitrofurazone and hydroxymethylnitrofurazone showed inhibitory activity cruzain being considered prototypes in search antichagasic. Using modern drug design strategies through the integration of computational techniques, molecular modeling and docking, and experimental synthesis and biological assays. In this work were performed design, synthesis and biological evaluation of the bioisosters nitrofurazone as candidates for antichagasic. Were applied molecular modeling and docking studies for ten derivatives compounds of aminoguanidine, semi and thiosemicarbazone. In these studies thiosemicarbazone derivatives compounds showed more favorable for the mechanism of action proposed, that suggest a nucleophilic attack of the Cys25 residue present in the catalytic site of the enzyme cruzain in the thiocarbonyl group. Six compounds were synthesized and characterized by 1H and 13C NMR. These compounds were tested for inhibition of cruzain, and derivatives 6, 9 and 10 showed favorable enzyme inhibition at single dose of 100 µM, with values between 70 and 75%. In the inhibition assay of cell growth in epimastigotes forms of T. cruzi, the compound 9 showed an IC50 of 19.8 µM, the best prototype for the development of a new antichagasic agent. In general the results obtained by biological assay corroborate the data presented in molecular modeling, since compounds containing side chain thiosemicarbazone showed better results in both tests, showing that the integration of experimental and computational techniques is presented as a excellent strategy in the search for new agents antichagasic

Docking is a computer simulation method used to predict the preferred orientation of two interacting chemical species that has been successfully applied to numerous macromolecules over the years. However, non-traditional targets have inherent difficulties associated with their screening. Large interfaces, lack of obvious binding sites, and transient pockets are some examples. Additionally, most natural ligands of challenging targets are inadequate models for identifying or designing new ligands. Therefore, it is not surprising that customary techniques of structure-based virtual screening are incompatible with these non-traditional targets. We hypothesized that an integrative virtual screening campaign comprised of docking followed by refinement of best receptor–ligand complexes would effectively identify small-molecule ligands of challenging receptors. We targeted the single-stranded DNA (ssDNA) binding groove of the human RAD52, and a cryptic allosteric pocket of the Helicobacter pylori Glutamate Racemase (GR). In this project, we first determined which docking method was more appropriate for each studied non-traditional target, and then examined how good our two-step docking workflow was in finding novel active ligand scaffolds. This research developed a powerful layered virtual screening workflow for the discovery of lead compounds against challenging protein targets. Furthermore, we successfully applied a statistical analysis method, which used receiver operating characteristic (ROC) curves, to validate the selected docking protocol that would be used in the screening campaigns. Using the validated workflow, we identified a natural compound that competes with ssDNA to bind to RAD52. The performed screening campaigns also provided new insights into the studied binding pockets, as well as structure-activity relationships (SAR) and binding determinants of the ligands. Our achievements reinforce the power of the ROC curve analysis approach in directing the search for the most appropriate docking protocol and helping to speed up drug discovery in pharmaceutical research.

Growing resistance of bacteria towards antibiotics resulted in extensive research effort for development and application of new materials and techniques. Due to their unique properties, graphene quantum dots (GQDs) have attracted much attention and are a promising material with potential applications in many fields. One use of GQDs is as a photodynamic therapy agent that generates singlet oxygen. In this work, GQDs synthesized by focusing nanosecond laser pulses into a mixture of benzene and nickel(II) oxide were combined with methylene blue (MB) to eradicate Gram-negative Escherichia coli and Gram-positive Micrococcus luteus. Theoretical calculation of pressure evolution was calculated using the standard finite difference method. Detailed characterizations were performed with transmission electron microscopy (TEM), scanning electron microscopy (SEM), Fourier-transform infrared (FTIR), UV-Visible (UV-Vis), and photoluminescence (PL) spectra. Furthermore, singlet oxygen generation from MB-GQD mixture was investigated by measuring the rate of 9,10-anthracenediyl-bis(methylene) dimalonic acid photobleaching at 400 nm. Combining MB with GQDs caused enhanced singlet oxygen generation, leading to improved bacterial deactivation rate. The (3-(4,5-dimethylthiazol-2- yl)-2,5-diphenyltetrazolium bromide) (MTT) assay was used to determine if GQDs in dark conditions caused human cellular side-effects and affected cancer and noncancer cellular viability. We found that even high concentrations of GQDs do not alter viability under dark conditions. These results suggest that the MB-GQD combination is a promising photodynamic therapy agent that may be useful when antibiotics resistance is present.

Over the past few decades, amorphous solid dispersions (ASDs) have been of great interest to pharmaceutical scientists to address bioavailability issues associated with poorly water-soluble drugs. ASDs consist of an active pharmaceutical ingredient (API) that is typically dispersed in an inert polymeric matrix. Despite promising advantages, a major concern that has resulted in limited marketed formulations is the physical instability of these complex formulations. Physical instability is often manifested as phase heterogeneity, where the drug and carrier migrate and generate distinct phases, which can be a prelude to recrystallization. One important factor that dictates the physical stability of ASDs is the spatial distribution of API in the polymeric matrix. It is generally agreed that intimate mixing of the drug and polymer is necessary to achieve maximum stabilization, and thus understanding the factors controlling phase mixing and nano-domain structure of ASDs is crucial to rational formulation design. The focus of this thesis work is to better understand the factors involved in phase mixing on the nanometric level and get insights on the role of excipients on overall stabilization of these systems. The central hypothesis of this research is that an intimately mixed ASD will have better physical stability as compared to a partially homogeneous or a non-homogeneous system. Our approach is to probe and correlate phase homogeneity and intermolecular drug-excipient interactions to better understand the physical stability of ASDs primarily using solid-state nuclear magnetic resonance (SSNMR) spectroscopy and other solid-state characterization tools. A detailed investigation was carried out to understand the role of hydrogen bonding on compositional homogeneity on different model systems. A comprehensive characterization of ternary ASDs in terms of molecular interactions and physical stability was studied. Finally, long-term physical stability studies were conducted in order to understand the impact of different grades of a cellulosic polymer on phase homogeneity for two sets of samples prepared via different methods. Overall, through this research an attempt has been made to address some relevant questions pertaining to nano-phase heterogeneity in ASDs and provide a molecular level understanding of these complex systems to enable rational formulation design.

In order to reduce the time associated with and the costs of drug discovery, machine learning is being used to automate much of the work in this process. However the size and complex nature of molecular data makes the application of machine learning especially challenging. Much work must go into the process of engineering features that are then used to train machine learning models, costing considerable amounts of time and requiring the knowledge of domain experts to be most effective. The purpose of this work is to demonstrate data driven approaches to perform the feature selection and extraction steps in order to decrease the amount of expert knowledge required to model interactions between proteins and drug molecules.

The overall aim of this research project was to develop surfactant dry powder formulations and devices for efficient delivery of aerosol formulations to infants using the excipient enhanced growth (EEG) approach. Use of novel formulations and inline delivery devices would allow for more efficient treatment of infants suffering from neonatal respiratory distress syndrome and bronchiolitis. A dry powder aerosol formulation has been developed using the commercial product, Survanta ® (beractant) and EEG technology to produce micrometer-sized hygroscopic particles. Spray drying and formulation parameters were initially determined with dipalmitoylphosphatidylcholine (DPPC, the dominant phospholipid in pulmonary surfactant), which produced primary particles 1 um in size with a mass median aerodynamic diameter of 1-2 um. Investigation of dry powder dispersion enhancers and alcohol concentration on the effect of powder aerosol characteristics were performed with the Survanta-EEG formulation. The optimal formulation consisted of Survanta ® , mannitol and sodium chloride as hygroscopic excipients, and leucine as the dry powder dispersion enhancer, prepared in 20% v/v ethanol/water. The powders produced primary particles of 1 um with >50% of the particles less than 1 um. The presence of surfactant proteins and surface activity were demonstrated with the Survanta-EEG formulation following processing. A novel containment unit dry powder inhaler (DPI) was designed for delivery of the surfactant-EEG formulation using a low volume of dispersion air. Studies explored optimization of air entrainment pathway, inlet hole pattern, delivery tube internal diameter and length. With 3- 10 mg fill masses of spray dried surfactant powder, the DPI enabled delivery of >2 mg using one 3-mL actuation of dispersion air. Overall, it was possible to deliver >85% of the loaded fill mass using three actuations. Nebulized aerosol formulations are characterized with low delivered doses. Using a novel mixer-heater delivery system, the highest estimated percent lung dose achieved during realistic in vitro testing of a Survanta-EEG formulation aerosolized with a commercial mesh nebulizer was when nebulization was synchronized with inhalation of the breathing profile. Design changes to the mixer-heater system eliminated the need for synchronization, achieving an estimated percent lung dose of 31% of the nominal, an improvement compared with existing systems that achieve approximately <2% lung dose.

Central infusion of the hexapeptide angiotensin IV (Ang IV) and its analogs have been demonstrated to markedly enhance memory retention and retrieval in rats using a range of learning and memory paradigms. This effect is mediated by the binding of the peptide to the specific binding site previously described as the AT4 receptor. The AT4 receptor has been isolated and identified as insulin-regulated aminopeptidase (IRAP), a type II transmembrane protein belonging to the M1 family of zinc-dependent aminopeptidases. Subsequently, AT4 receptor ligands, including Ang IV and its analogues and the unrelated peptide LVV-hemorphin-7, were demonstrated to be peptide inhibitors of IRAP. These findings suggest that AT4 ligands may exert their cognitive effects by inhibiting the catalytic activity of IRAP in the brain. Therefore, IRAP is an important target for the development of a new class of therapeutic agents for the treatment of memory loss. / To characterize IRAP at the molecular level and identify non-peptide inhibitors of IRAP for drug development, the aims of this study were to: 1) determine whether IRAP exists as a homodimer; 2) identify cysteine residue(s) involved in IRAP dimerization; 3) investigate the roles of the conserved residues of the HEXXH(X)18E Zn2+-binding motif and the GAMEN motif in substrate/inhibitor binding using site-directed mutagenesis; 4) use a molecular model of the catalytic domain of IRAP based on the crystal structure of a related M1 family metallopeptidase to: (i) identify key residues required for substrate/inhibitor binding; (ii) identify and characterize non-peptide IRAP inhibitors from a compound database by in silico virtual screening based on the homology model of IRAP. / Co-immunoprecipitation followed by Western blotting of IRAP under reducing and non-reducing conditions showed IRAP exists both as covalently- and non-covalently-bound homodimers. Serine scanning of cysteine residues potentially involved in forming inter-molecule disulfide-bonds was performed. Mutational analyses indicated that covalent homodimerization of IRAP is due to more than one cysteine residue. Limited trypsin digestion followed by co-immunoprecipitation suggests that non-covalent homodimerization of IRAP involves residues/regions within the last 130 amino acids of the protein. / The catalytic site of IRAP contains two consensus motifs, the H464EXXH468(X)18E487 Zn2+-binding motif and the G428AMEN432 motif. The role of conserved residues with these motifs was investigated using site-directed mutagenesis and pharmacological analyses. The conserved His and Glu residues of the Zn2+-binding motif were shown to be essential for IRAP catalytic activity. This was also observed for the Met and Glu residues of the GAMEN motif, while Asn mutant retained some catalytic activity. Residues important for substrate or inhibitor binding were identified as Gly, Ala and Asn. / A molecular model of the catalytic domain of IRAP based on the crystal structure of a homologous M1 metallopeptidase, leukotriene A4 hydrolase (LTA4H) was used to compare the catalytic sites of IRAP and LTA4H, and identified two amino acids at the putative substrate-binding pocket: Ala427 and Leu483 in IRAP, and the corresponding residues Tyr267 and Phe314 in LTA4H. A mutational analysis involving substitution of Ala427 and Leu483 with the corresponding residues revealed Ala427 and Leu483 characterize the enzyme S1 subsite, influencing the affinity and placement of substrates and peptide inhibitors in the catalytic site. / The molecular model of IRAP was also used for virtual screening of compound databases to identify novel non-peptide inhibitors. After two rounds of in silico screening, a family of compounds was identified and shown to be specific and competitive inhibitors of IRAP. Preliminary results suggest that one of these inhibitors, referred to as HFI 142, may possess memory-enhancing properties. The identification of non-peptide IRAP inhibitors will assist in pharmacological studies aimed at understanding the molecular mechanisms of IRAP aminopeptidase activity and physiological role of IRAP. In addition, the new inhibitors have the potential to form the basis for the development of a novel class of drugs useful for treating memory disorders.

Mes recherches ont eu pour objectif de concilier deux aspects complémentaires de la bioinformatique structurale : la modélisation de la structure 3D des protéines et la modélisation des petites molécules modulatrices des premières. La connaissance de la structure tridimensionnelle des protéines est un élément déterminant pour la compréhension fine de leur mécanisme d'action et indispensable pour le développement d'approches thérapeutiques rationnelles. Ainsi, l'identification et l'analyse structurale des sites de fixation de leurs ligands (protéine ou petite molécule) permettent d'envisager la modulation de leur fonction biologique. Les interactions protéine-protéine ou protéine-ligand peuvent être prédites, par exemple, par des programmes d'amarrage (ou ‘docking').<br /> La modélisation par homologie permet d'obtenir un modèle tridimensionnel d'une protéine lorsque sa structure n'a pas été déterminée expérimentalement. Ma contribution dans ce domaine fut la réalisation du serveur @TOME avec le soutien de la GENOPOLE Languedoc-Roussillon (accessible à l'adresse http://bioserver.cbs.cnrs.fr). Ce serveur était le premier de ce type à avoir été développé en France. Le serveur @TOME rassemble et traite d'une manière automatique toutes les étapes nécessaires à la construction d'un modèle 3D d'une protéine. Cela inclut la reconnaissance du repliement, la construction des modèles protéiques et leur évaluation. Les résultats du CASP5 en 2005 (session internationale d'évaluation des méthodes de prédiction de la structure des protéines ; http://predictioncenter.llnl.gov/) ont montré que notre serveur utilisé en mode automatique propose des modèles très proches de la structure expérimentale lorsque l'identité de séquence avec la structure support est supérieure à 30%. Le serveur a été classé 26ième sur 187 groupes inscrits.<br /> Dans un second temps, mes recherches m'ont permis de réaliser une base de données de complexes protéiques co-cristallisés, base fondatrice du projet DOCKGROUND. Ce projet de grande envergure, soutenu par le NIH depuis 2005, vise à établir un système intégré et dynamique de bases de données dédié à l'étude et à la prédiction des interactions entre protéines et permettre ainsi d'améliorer nos connaissances des interactions et de développer des outils de prédiction plus fiables. Ce travail a été effectué au sein de l'équipe du Pr. Ilya Vakser à l'Université de Stony Brook, NY, USA. Dans la réalisation de cette première base de données, un ensemble de programmes collectent, classent et annotent les complexes protéiques qui ont été co-cristallisés (données sur la séquence, la fonction, le repliement 3D, les particularités telles qu'une fixation à de l'ADN, ...). Ensuite, j'ai mis en œuvre une sélection dynamique des représentants des complexes contenus dans cette base. Les représentants sont essentiels pour éviter une surreprésentation de certaines familles de protéines. Cette base de donnée est accessible par Internet et est régulièrement mise à jour (http://dockground.bioinformatics.ku.edu). Le projet DOCKGROUND va être poursuivi par la réalisation de 3 autres bases de données qui s'ancreront sur la présente appelée ‘Bound-Bound'.<br /> L'objectif principal de mes travaux est d'identifier de nouveaux composés bio-actifs afin de comprendre le fonctionnement de leur cible dans un contexte biologique. Les méthodes que j'utilise se basent sur la chémoinformatique, le criblage virtuel et le de novo ‘drug design'. Dans le cadre de ce dernier, j'ai mis au point un programme propriétaire LEA3D (‘Ligand by Evolutionary Algorithm' 3D). Le programme génère des petites molécules à partir de la combinaison de fragments moléculaires issus de drogues et de molécules ‘bio' (substrats ou produits de réactions enzymatiques). Le criblage virtuel basé sur la structure protéique et le de novo ‘drug design' par LEA3D, ont été appliqués avec succès à la thymidine monophosphate kinase (TMPK) de Mycobacterium tuberculosis dans le cadre d'une collaboration avec une équipe de chimistes et de biologistes de l'Institut Pasteur. De nouvelles familles d'inhibiteurs ont été identifiées dont un inhibiteur synthétique trois fois plus affin que le substrat naturel. Plusieurs publications et une demande de brevet couvrent les résultats de ces recherches. Dans la continuité de ces travaux, je m'intéresse maintenant, plus particulièrement, à développer des stratégies de criblages de fragments (molécules de petit poids moléculaire). Il a été montré que de petites chimiothèques contenant des petites molécules polaires sont plus efficaces pour identifier des touches. Ce travail doit être réalisé conjointement avec des criblages structuraux expérimentaux comme la RMN ou la diffraction des rayons X. Ces derniers se posent comme une alternative aux tests in vitro avec pour avantage de donner une information détaillée, au niveau atomique, des interactions entre le ligand et sa cible. S'ensuit une étape d'optimisation/maturation des touches en ligands plus élaborés et plus affins par l'utilisation d'outils de chémoinformatique.

Nucleic acids have proven to be viable targets for small molecule drugs. While many examples of such drugs are detailed in the literature, only a select few have found practical use in a clinical setting. These currently employed nucleic acid targeting therapies suffer from either debilitating off-target side effects or succumb to a resistance mechanism of the target. The need for new small molecules that target nucleic acids is evident. However, designing a novel drug to bind to DNA or RNA requires a detailed understanding of exactly what binding environments each nucleic acid presents. In an effort to broaden this knowledge, the work presented in this thesis details the binding location and affinity of known and novel nucleic acid binding small molecules with targets ranging from simple RNA secondary structure all the way to the complex structure of ribosomal RNA. Specifically, it is shown that the anthracycline class of antineoplastics prefer to bind at or near mismatch base pairs in both physiologically relevant iron responsive element RNA hairpin constructs as well as DNA hairpin constructs presenting mismatched base pairs. Also characterized in this thesis is a novel class of topoisomerase II / histone deacetylase inhibitor conjugates that display a unique affinity for DNA over RNA. Finally, the novel class of macrolide-peptide conjugates, known as peptolides, are shown to retain potent translation inhibition of the prokaryotic ribosome. The binding pocket of the peptolides, including a crevice previously unreachable by macrolides that extends away from the peptidyl transferase center toward the subunit interface, is confirmed in detail via chemical footprinting of the 70S ribosome. Overall, the identification of a novel binding site for the anthracycline class of drugs and the characterization of the two novel drug designs presented in this thesis will undoubtedly aid in the effort to design and discover new molecules that aim for nucleic acid targets. For example, the anthracycline derivative topoisomerase II / histone deacetylase inhibitor conjugates, with their differential mode of nucleic acid binding, may prove to have a unique side effect profile in a therapeutic application. The peptolide compounds also have the potential to be applied as novel antibiotics as they bind to an area of the prokaryotic ribosome unrelated to known macrolide resistance mutations. Furthermore, as a result of the observation of this thesis work that some peptolides also posses eukaryotic translation inhibition capabilities, they could prove to be useful in preventing the growth of rapidly proliferating eukaryotic cells such as plasmodium, leishmania, or tumor cells. Additionally, different head groups could be utilized in creating new peptolides; for example, an oxazolidinone antibiotic could be employed to sample a different binding area of the ribosome.

Histone deacetylase (HDAC) inhibitors (HDACi) have demonstrated a wealth of biological effects, including anti-proliferative, anti-inflammatory, anti-parasitic, and cognition-enhancing activities. The recent FDA approvals of the inhibitors SAHA and FK-228 have validated HDACi clinical use in cutaneous T cell lymphoma, while numerous clinical trials are currently ongoing using HDACi against a variety of disease states. While the future of the HDAC field looks increasingly promising, there are lingering issues hindering broader use. Recent data point to dysregulation of specific HDAC isoforms in many disease states. However, most current HDACi are pan-inhibitors, lacking the specificity to target individual isoforms. Adding to this, there are currently 18 identified HDAC isoforms, and most lack a defined crystal structure, further complicating the task of designing isoform-specific inhibitors. Most importantly, HDACi have demonstrated a lack of efficacy against solid tumors in the clinic, a major obstacle to broader use in cancer therapy. Several of these issues could more fully be addressed through specific targeting of HDACi, and could bring HDACi into wider and more efficacious pharmaceutical use. Targeting the specific tissue or organelle where HDAC dysregulation occurs could confer greater efficacy in vivo. To this end, we have created four classes of compounds: (1) aryltriazolyl HDACi that potently inhibit HDAC activity and prostate cancer cell growth, (2) dual-targeted inhibitors of Topoisomerase II and HDAC and (3) dual-targeted inhibitors of Topoisomerase I and HDAC, both of which have potent inhibition against both target enzymes as well as cancer cell lines, and finally (4) macrocyclic HDACi that potently inhibit the growth of lung cancer cell lines and preferentially target lung tissue in vivo.