Simulacije molekulske dinamike, dizajn, sinteza i in vitro ispitivanje inhibitora epigenetskih proteina kao potencijalnih antineoplastika

Abstract

Među epigenetskim ciljnim mestima lekova, sirtuin 2 (SIRT2) i WD-ponavljajući domen 5 protein (WDR5) se ističu kao dva proteina sa velikim potencijalom za razvoj novih grupa antineoplastika. Uprkos činjenici da su inhibitori ova dva proteina pokazali odlične rezultate u pretkliničkim istraživanjima, nijedan od predstavnika još uvek nije ušao u fazu kliničkih ispitivanja. Razvojem novih protokola za računarsko dizajniranje lekova kroz implementaciju informacija iz velikih baza podataka i/ili molekulskih simulacija, kao i njihovom ekperimentalnom validacijom primenom organskih sinteza i in vitro bioloških ispitivanja, istraživanja sprovedena u okviru ove teze imala su za cilj da postave osnovu za budući razvoj novih hemijskih grupa inhibitora navedenih epigenetskih proteina kao potencijalnih novih antineoplastika. U prvoj fazi istraživanja, razvijen je protokol za poboljšano uzorkovanje dinamike vezivnog mesta SIRT2 koji se oslanja na kombinaciju simulacija molekulske dinamike i metadinamičkih simulacija. Primenom razvijenog protokola simulacija, otkriveno je postojanje skrivenog džepa unutar vezivnog mesta SIRT2. Implementacijom konformacionih stanja sa novootkrivenim skrivenim džepom u protokol virtuelnog skrininga, iz do sada neistraženih delova hemijskog prostora inhibitora izdvojena su dva nova potentna i selektivna SIRT2 inhibitora ND18 (IC50 = 58,7 μM) i ND85 (IC50 = 85,9 μM). Daljim ispitivanjem antitumorskog potencijala hit jedinjenja ND18 na ćelijskoj liniji adenokarcinoma dojke MDA-MB-231, ND18 je okarakterisano kao necitotoksično jedinjenje koji indukuje zaustavljanje ćelija u G2/M fazi ćelijskog ciklusa i povećava ekspresiju Ekadherina. U drugoj fazi istraživanja, uspostavljeni su kvantitativni modeli odnosa strukture i aktivnosti, odnosno selektivnost (SIRT1/3) SIRT2 inhibitora. Modeli su trenirani na velikom setu podataka od ukupno 1797 jedinjenja, primenom pet vodećih algoritama mašinskog učenja i četiri različita molekulska prikaza. Najbolji modeli su upotrebljeni za kreiranje in silico platforme za dizajn novih SIRT2 inhibitora, nazvane SIRT2i_Predictor. U okviru platforme, kreiran je grafički korisnički interfejs uz pomoć koga je moguće efikasno izvesti virtuelni skrining i detaljno ispitati odnose stukture i aktivnosti, odnosno selektivnost nepoznatih jedinjenja. SIRT2i_Predictor je dostupan putem linka: https://github.com/echonemanja/SIRT2i_Predictor. U trećoj fazi istraživanja, izveden je dizajn i sinteza 12 novih SIRT2 inhibitora. Proces dizajna se oslanjao na hipotezu o protein-ligand interakciji lead strukture ND18, kao i na predviđanja SIRT2i_Predictor-a. Šest sintetisanih jedinjenja je pri koncentraciji od 100 μM pokazalo selektivnu inhibitornu aktivnost veću od 50% što ih čini obećavajućim molekulima za dalja istraživanja. Među sintetisanim jedinjenjima se posebno ističe klasa aril-1,2,3-triazola kao potpuno nove hemijske grupe SIRT2 inhibitora. U četvrtoj fazi istraživanja izvedena je detaljna karakterizacija protein-ligand interakcije potpuno nove grupe fotoreaktivnih peptidomimetskih inhibitora WDR5. Proračunima slobodne energije vezivanja postavljen je protokol za dizajn novih predstavnika, dok je metadinamičkim simulacijama detaljno ispitan proces fotoizomerizacije. Dodatno, simulacijama je otkriven do sada nepoznati put alosterne komunikacije unutar WDR5 koji može poslužiti kao osnova daljeg dizajna efikasnijih inhibitora...

Among the epigenetic drug targets, sirtuin 2 (SIRT2) and WD repeat-containing protein 5 (WDR5) stand out as two protein targets with a great potential for the development of novel antineoplastic drugs. Despite the excellent results in preclinical studies, none of the SIRT2 or WDR5 inhibitors have entered clinical trials. By developing new protocols for computer-aided drug design based on information from large databases and/or molecular simulations, the research conducted within this thesis aimed to establish the foundation for future development of new chemical classes of SIRT2 or WDR5 inhibitors proteins as novel antineoplastic drugs. In the first phase of the research, a protocol for enhanced sampling of the binding pocket dynamics of SIRT2 relying on metadynamics simulations was developed. By applying the developed protocol, the existence of a cryptic pocket within the SIRT2 binding site was discovered. By incorporating conformational states with the newly discovered cryptic pocket into the virtual screening protocol, two new potent and selective SIRT2 inhibitors, ND18 (IC50 = 58.7 μM) and ND85 (IC50 = 85.9 μM), were discovered from unexplored regions of the chemical space of inhibitors. Further investigation of the anticancer potential of the lead compound ND18 on the breast tumor cell line MDA-MB-231 revealed that ND18, at a concentration of 20 μM, is a noncytotoxic compound that significantly induces cell cycle arrest at the G2/M phase and increases the expression of E-cadherin. In the second phase of the research, quantitative structure-activity relationship (QSAR) models were established to assess the potency, as well as selectivity (SIRT1/3) of novel SIRT2 inhibitors. The models were trained on a large dataset of 1797 compounds using five machine learning algorithms and four different molecular representations. The best-performing models were utilized to create an in silico platform for designing new SIRT2 inhibitors, named SIRT2i_Predictor. Developed graphical user interface enable performing virtual screening campaigns and detailed investigation of structure-activity and selectivity relationships of unknown compounds. SIRT2i_Predictor is available at: https://github.com/echonemanja/SIRT2i_Predictor. In the third phase of the research, the design and synthesis of 12 new SIRT2 inhibitors were performed. The design process relied on the hypothesis of protein-ligand interaction of ND18, as well as the predictions from SIRT2i_Predictor. Six of the synthesized compounds exhibited selective inhibitory activity greater than 50% at a concentration of 100 μM, making them promising molecules for further studies. Among the synthesized compounds, the aryl-1,2,3-triazole derivatives stood out as a completely new group of SIRT2 inhibitors. In the fourth phase of the research, a detailed characterization of the protein-ligand interaction of a completely new group of light-controlled peptidomimetic inhibitors of WDR5 was performed. The protocol for designing new representatives was established based on binding free energy calculations, while details of photoisomerization were investigated through metadynamics simulations. Simulations revealed a previously unknown pathway of allosteric communication within WDR5, which can serve as a basis for designing more effective inhibitors. In the fifth phase of the research, relying on data on the sensitivity of pancreatic tumor cell lines to bioactive compounds, a bioinformatics protocol for prediction of synergistic combinations of histone deacetylase inhibitors (including sirtuins) and other bioactive compounds was developed...