Skip to main content
Login | Suomeksi | På svenska | In English

Browsing by Subject "virtual screening"

Sort by: Order: Results:

  • Vuorinen, Anna (2010)
    11β-hydroxysteroid dehydrogenase/reductase (11β-HSD) enzymes 1 and 2 regulate the amount of cortisone and cortisol in human tissues. Since overexpression of 11β-HSD1 especially in the adipose tissue causes symptoms of metabolic syndrome, selective inhibition of 11β-HSD1 provides a way to treat this syndrome and type II diabetes. Inhibition of 11β-HSD2 causes cortisol-dependent mineralocorticoid activation, which leads to hypertensive side effects. There are several reported 11β-HSD1 inhibitors, for selective 11β-HSD2 inhibitition, only a few compounds have been developed. The difference between 11β-HSD1 and 2 ligand binding sites is unknown, which complicates the search of selective inhibitors to both of the enzymes. This study was done with two aims: (1) to identify the difference between the two isozymes, (2) to create pharmacophore models for selective 11β-HSD2 inhibitiors. These tasks were approached with computational methods: homology modeling, docking, ligand-based pharmacophore modeling and virtual screening. The homology model of 11β-HSD2 was constructed using SwissModeler and it showed satisfying superimposition both with is template 17β-HSD1 and 11β-HSD1. The difference between the enzymes could not be identified by visual inspections Therefore, seven compounds, of which six are 11β-HSD2 -selective, were docked both to 11β-HSD1 and 11β-HSD2 ligand binding sites using the program GOLD. The docking results revealed that the compounds orientate differently in the enzymes. To 11β-HSD1, the compounds were anchored similar than unselective compound carbenoxolone, whereas in 11β-HDS2, they adopted a flipped binding mode. The flipped binding mode in 11β-HDS2 enables hydrogen bonds to Ser310 and to Asn171, both residues that are only present in 11β-HSD2. Pharmacophore modeling and virtual screening were done using the program LigandScout3.0. The ligand-based pharmacophores were based on the six 11β-HSD2 selective compounds, which were also used for the docking studies. Both of the models consisted of six features (hydrogen bond acceptors, hydrogen bond donor and hydrophobic feature) besides the exclusion volumes. The most important features considering the 11β-HSD2 selectivity seem to be the hydrogen bond acceptor feature that could interact with the Ser310 and the hydrogen bond donor feature next to it. The interaction pair for this hydrogen bond donor feature was not observed in the homology model. However, a possibility of water molecule as an interaction pair was evaluated and it seems to be a possible solution to the problem. Since both of the models were able to find the selective 11β-HSD2 inhibitors and exclude the unselective ones from the test set database, they were employed for the screening of the database that consists of 2700 compounds stored at the University of Innsbruck. From the hits of these screenings ten compounds were selected and sent to biological testing. The results of the biological tests will decide how well the models represent the theory of the 11β-HSD2 selectivity.
  • Turku, Ainoleena (2010)
    The aims of this work were (1) to compare the three dimensional structures of different S- adenosylmethionine (SAM)-dependent methyltrasferases and (2) to screen in silico a commercial library for potential methyltransferase inhibitors. In this work we decided to focus on DNA methyltransferase-like enzyme (DNMT2) and catechol-O-methyltransferase(COMT). There were two different parts in my work. The first part was to analyze the 3Dstructures of DNMT2 and COMT in relation with their amino acid sequences. The structures of DNMT2 and COMT were compared together by means of superimposition with Sybyl 8. The ligand binding properties were studied by manual and automatic docking of known inhibitors in order to understand the binding specificity of these methyltransferases. The softwares I used for docking were Autodock 4.2 and Gold 4.0. The sequence alignments and superimposition of the known crystal structures showed that the structures of DNMT2 and COMT share a similar fold. Furthermore the main similarities between the structures of these enzymes are in the co-enzyme binding sites. The only significant difference in the binding sites is the place of one tyrosine residue, which causes a slight change in the conformation of the bound co-enzyme. Unlike co- enzyme binding sites, the substrate binding sites of DNMT2 and COMT are different. There is indeed a bound magnesium ion in the substrate binding site of COMT but not in the substrate binding site of DNMT2. Because the substrate binding sites are more different than the co-enzyme binding sites, we decided to screen the potential active ligands only at the substrate binding sites. The second part of the work was virtual screening. I used a subset of 20.000 molecules of ChemBridge DIVER Set that can be purchased commercially. The softwares I used for library preparation were CONCORD and Balloon, from which Balloon created more reasonable 3D structures for the docking. I did two parallel screenings to the crystal structure of COMT (PDB code 3BWM) with docking program GOLD 4.0, which is the only program that can take account metal coordination. To DNMT2 I did two sets of screenings, one with GOLD 4.0 and another with Autodock 4.2. I used known COMT inhibitors as control in the COMT run and known DNA methyltransferase inhibitors as control in DNMT2 run. Before docking to the three dimensional structure of DNMT2, one loop near the substrate binding site had to be modeled. I used Swiss-Modeler and Modeller softwares for that. Docking to COMT was successful according to the rank of the known COMT inhibitors compared to the subset of the FIMM library that was screened. I created the hitlist of 60 compounds based on the scores of these compounds, pharmacophore search and visual examination. 30 of these compounds were purchased and are currently being tested. The results of the DNMT2 run were not as reliable as the results of COMT run mentioned before, since the DNMT2 run was unable to retrieve known inhibitors better than random. The reason for that can be the quality of the model of the missing loop or the chosen controls. Furthermore only one of the ten small molecules that we used as controls is proved to be DNMT2 inhibitor, the others are DNMT1 and DNMT3 inhibitors and while the binding sites of DNMT1, DNMT2 and DNMT3 are very similar, they are, however, not completely identical.
  • Niemelä, Akseli (2022)
    Lecithin:cholesterol acyltransferase (LCAT), a key enzyme in maturating high-density lipoprotein (HDL) particles, has been targeted to promote the efficiency of reverse cholesterol transport by small molecular positive allosteric modulators (PAM) of Daiichi Sankyo. For a set of these compounds their Vmax and EC50 values and binding site in the membrane-binding domain (MBD) of LCAT have been determined. Through molecular dynamics (MD) simulations we previously found a metric that qualitatively described which compounds were active, so in this study we aimed to improve it by finding a quantitative metric. This led to the discovery of the Cα distance between CYS50 and ASN65, which correlates with this set’s Vmax values and which can be utilized to predict the Vmax values of novel compounds. Additional simulations were performed to discover whether this metric is changed by a lipid interface present, and to reveal a likely entry pathway PAMs take. As LCAT activation is likely a benign and potentially overlooked effect, we performed a virtual screen of FDA-approved compounds and secondary metabolites associated with LCAT. From secondary metabolites, a key finding was that flavonoids were overwhelmingly associated with LCAT and had a high binding potential to the MBD in docking simulations. The best binding compounds were subjected to MD simulations to discover their Vmax values using the discovered metric. This provided us with a set of compounds, which can be used to validate our in silico model in vitro. Should this model be validated, it can be used in optimising and discovering novel PAMs of LCAT, and it would bring evidence to the benefit of MD in drug discovery processes in general. Furthermore, if our discovered compounds can activate LCAT in vitro, they may be used as precursors for novel PAMs or as therapies by themselves not only for LCAT deficiencies, but perhaps for atherosclerotic cardiovascular diseases as well.