Please use this identifier to cite or link to this item: https://hdl.handle.net/10321/1751
Title: Computational and micro-analytical techniques to study the in vitro and in silico models of novel therapeutic drugs
Authors: Gumede, Njabulo Joyfull 
Issue Date: 2016
Abstract: 
In drug discovery and development projects, metabolism of new chemical entities (NCEs) is a major contributing factor for the withdrawal of drug candidates, a major concern for other chemical industries where chemical-biological interactions are involved. NCEs interact with a target macro-molecule to stimulate a pharmacological or toxic response, known as pharmacodynamics (PD) effect or through the Adsorption, Distribution, Metabolism, and Excretion (ADME) process, triggered when a bio-macromolecule interacts with a therapeutic drug. Therefore, the drug discovery process is important because 75% of diseases known to human kind are not all cured by therapeutics currently available in the market. This is attributed to the lack of knowledge of the function of targets and their therapeutic use in order to design therapeutics that would trigger their pharmacological responses.

Accordingly, the focus of this work is to develop cost saving strategies for medicinal chemists involved with drug discovery projects. Therefore, studying the synergy between in silico and in vitro approaches maybe useful in the discovery of novel therapeutic compounds and their biological activities. In this work, in silico methods such as structure-based and ligand-based approaches were used in the design of the pharmacophore model, database screening and flexible docking methods. Specifically, this work is presented by the following case studies:

The first involved molecular docking studies to predict the binding modes of catechin enantiomer to human serum albumin (HSA) interaction; the second involved the use of docking methods to predict the binding affinities and enantioselectivity of the interaction of warfarin enantiomers to HSA. the third case study involved a combined computational strategy in order to generate information on a diverse set of steroidal and non-steroidal CYP17A1 inhibitors obtained from literature with known experimental IC50 values. Finally, the fourth case study involved the prediction of the site of metabolisms (SOMs) of probe substrates to Cytochrome P450 metabolic enzymes CYP 3A4, 2D6, and 2C9 making use of P450 module from Schrödinger suite for ADME/Tox prediction.

The results of case study I were promising as they were able to provide clues to the factors that drive the synergy between experimental kinetic parameters and computational thermodynamics parameters to explain the interaction between drug enantiomers and thetarget protein. These parameters were correlated/converted and used to estimate the pseudo enantioselectivity of catechin enantiomer to HSA. This approach of combining docking methodology with docking post-processing methods such as MM-GBSA proved to be vital in estimating the correct pseudo binding affinities of a protein-ligand complexes. The enantioselectivity for enantiomers of catechin to HSA were 1,60 and 1,25 for site I and site II respectively.

The results of case study II validates and verifies the preparation of ligands and accounting for tautomers at physiological pH, as well as conformational changes prior to and during docking with a flexible protein. The log KS = 5.43 and log KR = 5.34 for warfarin enantiomer-HSA interaction and the enantioselectivity (ES = KS/KR) of 1.23 were close to the experimental results and hence referred to as experimental-like affinity constants which validated and verified their applicability to predict protein-ligand binding affinities.

In case study III, a 3D-QSAR pharmacophore model was developed by using 98 known CYP17A1 inhibitors from the literature with known experimental IC50 values. The starting compounds were diverse which included steroidal and non-steroidal inhibitors. The resulting pharmacophore models were trained with 69 molecules and 19 test set ligands. The best pharmacophore models were selected based on the regression coefficient for a best fit model with R2 (ranging from 0.85-0.99) & Q2 (ranging from 0.80-0.99) for both the training and test sets respectively, using Partial Least Squares (PLS) regression. On the other hand, the best pharmacophore model selected was further used for a database screening of novel inhibitors and the prediction of their CYP17A1 inhibition. The hits obtained from the database searches were further subjected to a virtual screening workflow docked to CYP17A1 enzyme in order to predict the binding mode and their binding affinities. The resulting poses from the virtual screening workflow were subjected to Induced Fit Docking workflow to account for protein flexibility during docking. The resulting docking poses were examined and ranked ordered according to the docking scores (a measure of affinity).

Finally, the resulting hits designed from an updated model from case study III were further synthesized in an external organic chemistry laboratory and the synthetic protocols as well as spectroscopic data for structure elucidation forms part of the provisional patent specification. A provisional patent specification has been filed (RSA Pat. Appln. 2015/ 07849). The case studies performed in this thesis have enabled the discovery of non-steroidal CYP17A1 inhibitors.
Description: 
Submitted in fulfillment of the requirements for the Doctor of Philosophy degree in Chemistry, Durban University of Technology, Durban, South Africa, 2016.
URI: http://hdl.handle.net/10321/1751
DOI: https://doi.org/10.51415/10321/1751
Appears in Collections:Theses and dissertations (Applied Sciences)

Files in This Item:
File Description SizeFormat
GUMEDE_2016.pdf1.42 MBAdobe PDFThumbnail
View/Open
Show full item record

Page view(s)

617
checked on Jul 23, 2024

Download(s)

383
checked on Jul 23, 2024

Google ScholarTM

Check

Altmetric

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.