Lysyl hydroxylases (LH) is a family of proteins that catalyse the formation of hydroxylysine residues in collagens, an important post-translational modification which makes them able to form collagen cross-links, thus contributing to stabilization of the extracellular matrix. These residues are further glycosylated through the intervention of two specific enzyme activities, the hydroxylysyl galactosyltransferase (GT) and the galactosylhydroxylysyl glucosyltransferase (GGT), converting collagen lysines into glucosylgalactosyl-hydroxylysines (HyK). Among LH enzymes, LH3 is the only one with a multifunctional role, as it is able to hydroxylate lysines and also to catalyse its subsequent glycosylation. Several studies demonstrate that overexpression of this enzyme can be correlated with increased metastatic progression of a variety of tumors, suggesting that it is a valid target for development of targeted therapies for cancer. The aim of this thesis was the design and development of potential new hit compounds capable of inhibiting the GT-activity of the enzyme LH3. Using as a starting point the crystal structure of LH3 (PDB: 6TEC) in complex with a weak inhibitor (uridine diphosphate xylose-UDX) in the GT-domain, a series of mimics of this inhibitor have been designed and tested in silico, with fragment-based methodologies, by replacing the sugar moiety with new fragments capable of establishing more efficient interactions with the active site. First, a docking protocol able to re-dock the ligand present in the crystal structure into its binding pocket has been defined, in order to validate the method and to use it for subsequent calculations. Then, the same docking protocol was used to screen a library of fragments, provided from collaborators, therefore readily available for in vitro testing, for substituting the sugar moiety of UDX. A series of fragments able to establish additional interactions with the active site compared to the known inhibitor has been identified. Linking strategies for connecting these fragments to the uridine diphosphate have been evaluated and tested. Afterward, a pharmacophore model recapitulating the potential interactions that the catalytic pocket can establish with a ligand has been generated. Interestingly, this pharmacophore was able to distinguish LH3 binders from a series of GT-inhibitors of glycosyltransferases similar to LH3, which, however, could not bind the enzyme. This model allowed us to define the structural determinants of productive interactions with the binding pocket and to rationally explain why these ligands cannot bind LH3. Using the information derived from this pharmacophore model, a second virtual screening has been carried out on a larger fragments’ library, to explore a wider chemical space. Compounds from the filtered library were ranked according to a fit value that indicates how well the features in the pharmacophore map with the chemical features of the compound. These fragments were then linked to the uridine diphosphate, in order to generate a large variety of UDX-mimics compounds. These new compounds were docked on the binding site of LH3, and ranked through the docking score, which is an approximation of the binding free energy of each ligand, to determine the most promising molecules. Finally, to assess the drug-likeness of the new compounds, an ADMET analysis has been performed.
Le lisil-idrossilasi (LH) costituiscono una famiglia di proteine che catalizzano la formazione di idrossilisine sul collagene, effettuando un’importante modifica post-traslazionale responsabile della formazione dei legami cross-linkati nel collagene, contribuendo alla stabilizzazione della matrice extracellulare. Questi residui idrossilati vengono successivamente glicosilati ad opera di due enzimi ad attività specifica, le idrossilisil galattosiltrasferasi (GT) e le galattoidrossilisil glucosiltrasferasi (GGT) aventi il compito di convertire l’idrossilisina in glucosilgalattosilidrossilisina (HyK). All’interno della famiglia LH, LH3 è l’unico enzima ad avere un ruolo multifunzione, essendo in grado di idrossilare le lisine ed catalizzarne la successiva glicosilazione. Diversi studi dimostrano come la sovraespressione di questo enzima sia direttamente correlata con la progressione di forme tumorali, suggerendo come quest’ultimo possa essere un buon bersaglio per lo sviluppo di terapie anticancro. Lo scopo di questa tesi è stato quello di progettare e sviluppare nuovi potenziali composti hit, capaci di inibire l’attività GT dell’enzima LH3. Usando come punto di partenza la struttura cristallografica di LH3 (PDB: 6TEC) in complesso con un debole inibitore (xilosio-uridindifosfato – UDX) presente nel sito GT, una serie di mimici di questo inibitore sono stati progettati e testati con metodologie in silico, sfruttando un approccio fragment-based pensato per sostituire la porzione zuccherina dell’inibitore con diversi frammenti capaci di formare nuove e migliori interazioni con il sito di legame. Per prima cosa è stato definito un protocollo di docking in grado di re-dockare il ligando presente nella struttura cristallografica, in modo da avere un modello validato da usare per tutti i successivi calcoli. Successivamente il medesimo protocollo di docking è stato usato per filtrare una libreria di frammenti fornita da collaboratori e quindi pronti per essere testati in vitro, con lo scopo di sostituirli alla porzione zuccherina dell’UDX. In questo modo è stata trovata una serie di frammenti capaci di stabilire interazioni aggiuntive con il sito attivo. Per definire i potenziali inibitori dell’enzima risultanti da questo screening, sono state esplorate e testate diverse strategie di connessione tra l’uridinadifosfato e tali frammenti. In seguito, è stato generato un modello farmacoforico per identificare le interazioni che il sito attivo può potenzialmente stabilire con un ligando. È stato interessante notare come questo modello farmafocorico sia in grado di distinguere molecole effettivamente in grado di legare LH3 da una serie di inibitori attivi su enzimi GT ad esso simili, che però non legano LH3. Questo modello ci ha quindi permesso di definire i determinanti strutturali necessari per stabilire interazioni efficaci con la tasca di legame, oltre a fornire una spiegazione razionale del perché gli inibitori delle GT non leghino LH3. Usando le informazioni derivanti da questo modello farmacoforico, è stato eseguito un secondo virtual screening utilizzando una libreria di frammenti molto più numerosa, in modo da esplorare uno spazio chimico più ampio. I frammenti così filtrati sono stati ordinati secondo un valore di fitting che indica quanto bene le caratteristiche chimiche dei composti si sovrappongono spazialmente ai punti farmacoforici del modello. Anche questi frammenti sono stati connessi all’uridinadifosfato per generare una maggiore varietà di mimetici dell’UDX. Questi nuovi composti sono stati dockati nel sito attivo di LH3 ed ordinati in base al docking score, un’approssimazione dell’energia libera di legame di ogni ligando, per determinare quali molecole fossero i leganti più promettenti. Infine, per saggiare quanto i composti fossero drug-like, è stata condotta un’analisi ADMET.
Progettazione in silico di inibitori della lisil-idrossilasi-3 umana con un approccio fragment-based
SCIVA, CRISTIANO
2021/2022
Abstract
Lysyl hydroxylases (LH) is a family of proteins that catalyse the formation of hydroxylysine residues in collagens, an important post-translational modification which makes them able to form collagen cross-links, thus contributing to stabilization of the extracellular matrix. These residues are further glycosylated through the intervention of two specific enzyme activities, the hydroxylysyl galactosyltransferase (GT) and the galactosylhydroxylysyl glucosyltransferase (GGT), converting collagen lysines into glucosylgalactosyl-hydroxylysines (HyK). Among LH enzymes, LH3 is the only one with a multifunctional role, as it is able to hydroxylate lysines and also to catalyse its subsequent glycosylation. Several studies demonstrate that overexpression of this enzyme can be correlated with increased metastatic progression of a variety of tumors, suggesting that it is a valid target for development of targeted therapies for cancer. The aim of this thesis was the design and development of potential new hit compounds capable of inhibiting the GT-activity of the enzyme LH3. Using as a starting point the crystal structure of LH3 (PDB: 6TEC) in complex with a weak inhibitor (uridine diphosphate xylose-UDX) in the GT-domain, a series of mimics of this inhibitor have been designed and tested in silico, with fragment-based methodologies, by replacing the sugar moiety with new fragments capable of establishing more efficient interactions with the active site. First, a docking protocol able to re-dock the ligand present in the crystal structure into its binding pocket has been defined, in order to validate the method and to use it for subsequent calculations. Then, the same docking protocol was used to screen a library of fragments, provided from collaborators, therefore readily available for in vitro testing, for substituting the sugar moiety of UDX. A series of fragments able to establish additional interactions with the active site compared to the known inhibitor has been identified. Linking strategies for connecting these fragments to the uridine diphosphate have been evaluated and tested. Afterward, a pharmacophore model recapitulating the potential interactions that the catalytic pocket can establish with a ligand has been generated. Interestingly, this pharmacophore was able to distinguish LH3 binders from a series of GT-inhibitors of glycosyltransferases similar to LH3, which, however, could not bind the enzyme. This model allowed us to define the structural determinants of productive interactions with the binding pocket and to rationally explain why these ligands cannot bind LH3. Using the information derived from this pharmacophore model, a second virtual screening has been carried out on a larger fragments’ library, to explore a wider chemical space. Compounds from the filtered library were ranked according to a fit value that indicates how well the features in the pharmacophore map with the chemical features of the compound. These fragments were then linked to the uridine diphosphate, in order to generate a large variety of UDX-mimics compounds. These new compounds were docked on the binding site of LH3, and ranked through the docking score, which is an approximation of the binding free energy of each ligand, to determine the most promising molecules. Finally, to assess the drug-likeness of the new compounds, an ADMET analysis has been performed.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/15819