Tez No	İndirme	Tez Künye	Durumu
636315		Sanal tarama ve çok boyutlu moleküler modelleme yöntemleri ile p53-MDM2 potansiyel inhibitörlerinin belirlenmesi / Identification of p53-MDM2 potential inhibitors with virtual screening and multidimensional molecular modeling methods Yazar:GÜLŞAH AYDIN Danışman: PROF. DR. MİNE YURTSEVER ; PROF. DR. SERDAR DURDAĞI Yer Bilgisi: İstanbul Teknik Üniversitesi / Fen Bilimleri Enstitüsü / Kimya Ana Bilim Dalı / Kimya Bilim Dalı Konu:Kimya = Chemistry Dizin:	Onaylandı Doktora Türkçe 2020 206 s.
Tümör baskılayıcı p53 geni kanser vakalarında en fazla mutasyona uğrayan gen olması dolayısıyla kanser araştırmaları için büyük önem taşımaktadır. Sağlıklı hücrelerde MDM2 proteini, p53 geni tarafından kodlanan p53 proteininin transkripsiyonel aktivasyonunu negatif olarak düzenler. Bu düzenleme hücrelerdeki optimum p53 protein seviyelerini ve hücre sağ kalımını koruyarak normal hücre döngüsü ilerlemesini sağlamak için hayati önem taşımaktadır. MDM2 proteininin mutasyona bağlı olarak aşırı ifade edilmesiyle bu denge bozularak tümör oluşumu meydana geldiğinden MDM2 proteini inhibisyonunda etkili olabilecek moleküllerin tasarlanmasına yönelik çalışmalar yeni anti kanser aday moleküllerin geliştirilmesinde yüksek öneme sahiptir. Bilgisayar destekli (in siliko) ilaç tasarımı klasik ilaç tasarımı yaklaşımlarına göre zamandan tasarruf, maliyetin azaltılması gibi birçok avantaja sahiptir. Bilgisayar simülasyonları yardımıyla çalışılan sistemin deneysel ve teorik olarak ispatlanmış tüm bilgileri ilaç keşfinde bir strateji geliştirmek için kullanılabilir. Bu sayede bilinen araştırmalardan faydalanılarak bilinmeyen moleküllerin olasılığı öngörülür ve amaca yönelik senteze yardımcı olunabilir. Simülasyonlar başlangıçtaki durumu bilinen bir sistemin belirli bir zaman sonraki durumunun fiziksel ve matematiksel denklemlere bağlı olarak öngörülmesi konusunda büyük avantaj sağlamaktadır. Simülasyonlar yardımıyla deneysel olarak gerçekleştirebileceğimiz çalışmalar çok daha pratik şekilde öngörülebilirken deneysel olarak ortaya konması mümkün olmayan bir çok çalışma hakkında da önemli fikirler elde edilebilir. Tez çalışması kapsamında anti kanser aday moleküller olarak yeni p53-MDM2 etkileşim inhibitörlerinin tasarlanması için çok sayıda molekül içeren kütüphanelerin pratik şekilde taranmasına olanak sağlayan in siliko ilaç tasarım stratejileri uygulanmıştır. Yeni ilaç adaylarının MDM2 inhibisyon kapasitesini araştırmak için moleküler kenetlenme, Moleküler Dinamik (MD) simülasyonlarını ve ADME/Toksisite özellikleri tahminlerini içeren in siliko yaklaşımlar ve in vitro aktivite testleri gerçekleştirilmiştir. Çalışmalarda MDM2 proteininin N-terminal p53 bağlanma bölgesini hedef alan inhibitör adaylarının geliştirilmesine odaklanılmıştır. Simülasyonlarda kullanılacak kristal yapının seçimi için MDM2 proteininin 40 farklı kristal yapısı PDB'den temin edilmiş ve kristal kompleks içindeki ligant farklı kenetlenme algoritmaları kullanılarak proteine yeniden kenetlenmiştir. Kristal içindeki ligandın ilk konformasyonu ile kenetlenme sonrası konformasyonu üst üste çakıştırılarak RMSD (Root Mean Square Deviation) değerleri hesaplanmıştır. Simülasyonlarda, tüm kenetlenme yöntemleri ile 2 Å altında RMSD değeri verdiği gözlenen PDB:4HBM kristal yapısının kullanılmasına karar verilmiştir. Çalışılacak kenetlenme protokollerinin deneysel değerlerle korele edilebilir sonuçlar vermesi de doğru kristal yapıyla çalışmak kadar önem arz etmektedir. Bu sebeple IC50 (inhibition concentration 50) değeri bilinen 42 standart MDM2 inhibitörünün ∆Gdeneysel skorları ile ∆Gkenetlenme skorlarının mutlak farkı alınarak kenetlenme protokollerinin deneysel değerlerden ortalama sapması hesaplanmıştır. En düşük mutlak sapmayı veren Glide/SP kenetlenme protokolü deneysel verilerle daha uyumlu sonuçlar verdiğinden MD simülasyonlarda başlangıç pozu olarak Glide/SP pozları tercih edilmiştir. Kristal yapının uygunluğuna karar verildikten sonra EnamineStore veri tabanından alınan 500.000 ligant ve MDM2 kristali pH 7.4'te hazırlanarak hiyerarşik sanal tarama protokolleri yardımıyla kenetlenme çalışmaları gerçekleştirilmiştir. Farklı algoritmalara dayanan altı kenetlenme yönteminden elde edilen pozların bağlanma modları ve bağlanma serbest enerjilerinin tahminleri birlikte değerlendirilerek 100 ligant filtrelenmiştir. Filtrelenen ligantların terapötik aday olabilmeleri için ADME/Toksisite profillerinin uygunluğu da çok önemlidir. Bundan dolayı 100 aday MetaCore ve MetaDrug (MC/MD) araçları yardımıyla topoloji temelli ikili QSAR modelleri analizine tabi tutulmuştur. Önerilen 100 ligantın kanser terapötik aktiviteleri; Lipinski'nin 5 kuralı, Prot-Bind, GlogP, Wsol gibi ADME özellikleri; AMES, sitotoksisite, MRTD, kanserojenite, kardiyotoksisite, genotoksisite, hepatotoksisite, nefrotoksisite, nörotoksisite, karaciğer toksisitesi ve böbrek toksisitesi gibi 26 farklı toksisite profili tahminleri değerlendirilmiştir. QSAR model analizine dayalı filtreleme sonucu, 19 aday molekülün kanser için terapötik aktivite gösterme potansiyeli taşıyabileceği ve herhangi bir toksisite göstermeyeceği öngörülmüştür. Önerilen 19 adayın 16'sının, antitümör aktivitesi literatürde sıklıkla vurgulanan indol iskeleti içeren peptidomimetik bileşiklerden oluştuğunun ve bunlardan 4 çiftin birbirlerinin geometrik izomeri olduğunun gözlenmesi umut verici bir parametre olarak kaydedilmiştir. Sanal tarama çalışmaları, çok sayıda ligant içeren kütüphanelerin elenmesi için önemli fikirler sunmakta fakat protein-ligant komplekslerinin dinamik etkileşim süreçleri hakkında herhangi bir fikir vermemektedir. Bundan dolayı, filtrelenen 19 aday arasından seçilen 11 tanesine, pozitif kontrole, negatif kontrole ve protein apo formuna yapısal ve dinamik profillerini incelemek için 1'er µs MD simülasyonları uygulanmıştır. MD simülasyonlarında protein bağlanma bölgesiyle etkileşim halindeki ligantların dinamik davranışı hakkında önemli bilgiler sağlanmıştır. Daha sonra MD simülasyonları sonucu elde edilen yörüngelerden alınan 500'er poza MM/GBSA (Molecular Mechanics Generalized Born Surface Area) uygulanarak bağlanma serbest enerjileri tahmin edilmiştir. Ayrıca EnamineStore veri tabanından temin edilen 11 molekül için HUVEC vasküler endotel, kolon kanseri ve meme kanseri hücre hatlarına karşı in vitro aktivite testleri gerçekleştirilmiştir. Test sonuçlarına göre meme kanseri hücre hattında 6 ligantın (E1, E2, E5, E6, E9 ve E11) ve kolon kanseri hücre hattında 6 ligantın (E1, E2, E5, E6, E8 and E10) µM-seviyede inhibisyona sebep olduğu, ayrıca ligantların proliferasyonu inhibe ederek apoptoza yol açtığı gözlenmiştir. E11, 13 µM IC50 değeri ile en iyi inhibisyon gösteren ligant olmuştur. Önerilen adayların yapısal benzerlikleri de dikkate alınarak tüm sonuçlar birlikte değerlendirildiğinde, adayların ve/veya türevlerinin MDM2 inhibe edici etkisinin kaçınılmaz olabileceği öngörülmüştür. Çalışılan hücrelerin MDM2 protein ekspresyon seviyeleri farklılık gösterdiği için ileri çalışmalarda tüm adayların MDM2 proteini spesifik inhibisyonuna yönelik in vitro testlerinin gerçekleştirilmesinin p53 ve/veya doğrudan MDM2 aracılıklı kanser vakalarının tedavisi için umut vaat edebileceği öngörülmüştür.
Tumor suppressor p53 gene, which is of great importance for cancer research, is the most mutated gene in the case of cancer. The p53 protein, which is encoded by the p53 gene, is effective in stopping the cell cycle and initiating apoptosis in response to DNA damage. p53 mutations are frequently seen in sarcomas, leukemia, lymphomas, tumors of the nervous system and cancer of organs such as colon, breast, lung. If the p53 protein does not show the activity, apoptosis cannot occur in cells with DNA damage and tumor formation is observed. In healthy cells, the MDM2 protein negatively regulates the transcriptional activation of p53. This regulation is vital to provide normal cell cycle progression while maintaining optimum p53 protein levels in cells and cell survival. When the MDM2 protein is overexpressed in most tumor types due to mutation, this equilibrium is disrupted and tumor formation occurs. When MDM2 protein is inhibited by the appropriate antagonist, tumor growth is stopped. Cancer therapeutic effect is achieved by inhibiting both p53-dependent and p53-independent functions of MDM2. Although there are many MDM2 inhibitors designed for this purpose, the discovery of hit inhibitors with low side effects and high efficacy is hot in cancer research. Therefore, it is of high importance to design novel anti-cancer hit molecules that may be effective in inhibiting MDM2 protein. Computer-aided (in silico) drug design allows effective drug development by providing insight into the interaction of a molecule with the binding site of the protein. In silico drug design has many advantages compared to classical drug design approaches such as saving time and reducing costs. The main objective is to investigate whether the hit molecule binds to the target site and has a pharmacological effect. In siliko drug design has a wide range of applications including determination of active conformation of protein-ligand complex by docking simulations, investigation of time-dependent changes by Molecular Dynamics (MD) simulations, pharmacophore development and quantitative structure activity relationship (QSAR). All the experimental and theoretically proven information of the studied system can be used to develop a strategy in drug discovery with the help of computer simulations. In this way, the possibility of unknown molecules is predicted by making use of known researches and the aim-oriented synthesis can be realized. Simulations provide a great advantage in predicting the next state of a system whose initial state is known based on physical and mathematical equations. While the simulations allow us to predict the studies that we can perform experimentally in a much more practical way, important ideas can be obtained about many studies that cannot be empirically demonstrated. Within scope of the thesis, in siliko drug design strategies which allow the practical screening of chemical libraries containing a large number of molecules have been applied to design new p53-MDM2 interaction inhibitors as anti cancer hits. To investigate the MDM2 inhibition capacity of novel hits, in siliko approaches including molecular docking, molecular dynamics (MD) simulations and binnary QSAR model analyzes for the prediction of ADME/Toxicity properties, and then in vitro activity tests were performed. Studies have focused on the development of inhibitory hits targeting the N-terminal p53 binding region of the MDM2 protein. In the selection of the MDM2 crystal structure to be used in the simulations, it is necessary to pay attention to the parameters such as the resolution of the structure and the type of organism from which it is taken. In order to test the compatibility of in siliko simulation studies with experimental data, it is important that the selected crystal structure gives the most accurate pose with studied docking protocols. 40 MDM2 crystals were selected from PDB according to the parameters mentioned. The ligands in the N-terminal binding region were re-docked to the crystal structures using docking protocols and the RMSD values were calculated by overlapping the initial poses and the post-docking poses. It was decided to use PDB: 4HBM crystal structure which showed RMSD values under 2 Å in all methods in the simulations. As well as working with the accurate crystal structure, it is important that the docking protocols to be worked out give the accurate results. For this reason, 42 standard MDM2 inhibitors with known IC50 values were docked to the protein by six different docking algorithms and ∆Gdocking scores were calculated. The docking score was correlated with the ∆Gdocking values calculated using IC50 values, and the compatibility of the docking protocols with the experimental values was tested. By taking the absolute difference of ∆Gexperimental scores and ∆Gdocking scores, the mean deviation of the docking protocols from the experimental values was calculated. It was estimated that the Glide/SP docking protocol yielding the lowest absolute deviation. Thus, Glide/SP poses were preferred as the initial poses of MD simulations. After determining the suitability of the crystal structure, MDM2 crystals and 500.000 ligands from the EnamineStore database were prepared under physiological conditions (pH: 7.4). The ligands were docked to the protein with hierarchical virtual screening protocols using 5 different algorithms of the Schrödinger Maestro Molecular Modeling software and GOLD software. 500.000 ligands were docked to the protein using the Glide/HTVS method and the structures giving the highest binding score likely to exhibit inhibitory activity were roughly filtered. Thus, 100.000 ligands filtered from Glide/HTVS were used in Glide/SP and 20.000 top-docking scored molecules were used in Glide/XP protocol. IFD (InducedFit Docking), GOLD and QPLD (Quantum Polarized Ligand Docking) algorithms were then used for further protein-ligand interaction analysis of top-100 ligands selected based on Glide/XP docking scores. When the docking score results of the proposed small molecules and the ligand interaction maps are evaluated, it can be seen that they can be considered as potent candidate inhibitors for the MDM2 protein. However, their ADME/Toxicity profiles are also important for being candidate therapeutic hit ligands. For this reason, the topology-based binary QSAR models were used with the help of MetaCore/MetaDrug (MC/MD) tools to estimate the cancer therapeutic activity, pharmacokinetic properties as well as 26 different toxicity profiles including AMES, cytotoxicity, MRTD, carcinogenicity, cardiotoxicity, genotoxicity, hepatotoxicity, nephrotoxicity, neurotoxicity, liver toxicity and kidney toxicity of 100 hits. MC/MD is a commercial software that saves time in literature research by providing predictions by combining QSAR and system biology approaches with the help of statistical procedures. The library of QSAR models created by the MC/MD tools provides useful and reliable information based on the statistical predictions on the aforementioned drug parameters. As a result of filtering based on QSAR model analysis, it was predicted that 19 hit molecules would have the potential to be therapeutic for cancer and would show no toxicity. It has been observed that 16 of the 19 hits are peptidomimetic compounds containing indole skeleton, whose antitumor activity is frequently emphasized in the literature, and that 4 pairs are geometric isomers of each other. Most indole derivatives isolated from natural products or synthetic that are FDA approved and are listed in the WHO list of essential drugs are used in cancer treatment. In addition, direct p53-MDM2 interaction and MDM2 ubiquitin ligase activity is inhibited by indole derivatives are available in studies. Since indole derivative molecules are used in cancer treatment, it is promising that selective prominence of indole derivatives in our virtual screening studies. Virtual screening studies provide important ideas for screening large number of ligands-containing libraries, but do not provide any insight into the dynamic interaction processes of protein-ligand complexes. The predictions of virtual screening studies cannot be considered realistic enough to provide a commentary on therapeutic activity and toxic effects. Therefore, 1 µs MD simulations were applied to examine the structural and dynamic profiles of positive control, negative control, protein apo form and 11 hits including two pairs of geometric isomers selected from 19 filtered hits. In MD simulations important information is provided about the dynamic behavior of ligands interacting with the protein binding site. Many important parameters such as RMSD values, RMSF values, %SASA values, PSA values, protein-ligand contact diagrams, protein-ligand interaction maps and torsion analysis of protein- ligand interactions was evaluated to understand the strengths and types of interactions of the hit ligands with the protein binding site through MD simulations. Then, the binding free energies were estimated by applying MM/GBSA to 500 pose from the trajectory obtained from the second half of MD simulations. In vitro activity tests against HUVEC vascular endothelial, colon cancer and breast cancer cell lines were performed on 11 hits obtained from EnamineStore database. In in vitro tests, two pairs of enantiomers (E1/E2 and E3/E4) were applied as racemic mixtures. According to the test results, 6 ligands (E1, E2, E5, E6, E9 and E11) in the breast cancer cell line and 6 ligands (E1, E2, E5, E6, E8 and E10) in the colon cancer cell line cause µM-level inhibition and also it was observed that the ligands inhibited proliferation, leading to apoptosis. As a result, it has been proposed that 19 hits, most indole derivatives, may be effective in inhibiting p53-MDM2 interaction. Interactions of 11 derivatives with MDM2 protein were monitored dynamically with the help of MD simulations and important information was obtained about interactions. When the MD simulations of the proposed hits were analyzed, it was predicted that, in addition to the three hydrophobic packages critical for p53-MDM2 interaction, all of the hit ligands that could strongly interact with the His96 residue in the protein binding site. In particular, it is envisaged that ligands E1/E2, E3/E4 and E10 may be strong hits with high degree of potential for MDM2 inhibition. When 11 hits MD simulation results were compared, it was estimated that ligands that might strongly interact with His96 residue may be highly effective in inhibiting MDM2. Analysis of MD simulations clearly demonstrated that indole ring-containing ligands may be of great importance in the formation of hydrophobic interactions exhibited with His96. The presence of interactions with critical residues such as Thr15, Thr16, Ser17, Gln18 and Ile19, which emphasized their importance in the development of new MDM2 inhibitors in the literature, was noted in most of the interaction maps of protein complexes formed by hit ligands. It has also been clearly demonstrated in vitro tests that the majority of the hit ligands cause inhibition at µM level in the tested cells and/or induce apoptosis. E11 was the ligand with the best inhibition with 13 µM IC50 value. When all the results were evaluated considering the structural similarities of the proposed hits, it was predicted that the effects of the proposed hits and/or derivatives may be inevitable in inhibiting MDM2. Further studies have suggested that performing in vitro tests for MDM2 protein specific inhibition of all hits may be promising for the treatment of p53 and/or direct MDM2 mediated cancer cases.