Probing function of unknown proteins by using pharmacophore searching and biophysical techniques

Ibrahim, Musadiq (2013) Probing function of unknown proteins by using pharmacophore searching and biophysical techniques. PhD thesis, University of Glasgow.

Full text available as:
Download (33MB) | Preview
Printed Thesis Information:


The number of protein structures deposited in the Protein data bank is increasing almost exponentially and among these structures many of the proteins are novel with unknown function. Like Docking, Pharmacophore searching is an In-silico technique which is widely used for drug discovery. In pharmacophore searching the main focus is on the hydrogen bond interactions between the ligand and the target protein. The pharmacophore models are generated either by using the already known actives as templates or by utilizing the significant chemical features of the active site.
In this thesis the pharmacophore searching has been used to find potential ligands/substrates for unknown proteins and then ligand binding is confirmed by using different biophysical techniques. In the initial phases the pharmacophore models were generated by using Cerius2 and Weblab Viewer pro programs. While in later stages more sophisticated searches were carried out by using DSV (Discovery studio visualizer, Accelrys®). Procedures were optimized for model building by using DSV, which enabled the pharmacophore searching via both the Vector and the Query atom methods. To validate the technique, it was first used on known enzymes with established function e.g. xylose reductase and shikimate kinase. The optimized pharmacophore model when search through the database successfully identified the true substrates for these enzymes among other ligands thereby demonstrating the attainment.

The pharmacophore searching technique has been used to find potential ligands for proteins with unknown function on three test cases e.g. TdcF, HutD and PARI. Of the potential pharmacophore hits obtained through database search, a number of compounds were either purchased or synthesised to be tested for binding affinity. Different biophysical techniques like DSC, ITC, CD and NMR were used for this purpose. Among these techniques NMR proved to be the most sensitive technique to differentiate binders from non-binders and to further detect weak and strong bonding in terms of Kd values. For TdcF among other binders the best binder was 2-ketobutyrate with a Kd value of 200µM. In case of HutD, formyl glutamate (Kd = 92µM) and formimino glutamate (Kd = 500µM) came out to be the best binders and could be the true ligands of the protein at physiological concentration. For PARI L-glutamate appeared to be a potential ligand for the protein as confirmed through the NMR experiments. Pharmacophore modelling has been successful in identifying potential interactions provided by the protein active site which in turns specifies the required features to be present in a ligand and later on the successful binding studies further confirm its applicability.

In addition, protein structures from the protein data bank (PDB) with unknown ligands (UNK) were identified and manually screened to find examples that could be used to test the applicability of pharmacophore searching methods. The diversity of structures showed that the definition of an unknown ligand is completely inconsistent with many examples where any non spherical density was labelled as unknown ligand and in most cases a single atom is labelled as an unknown ligand, which most likely can be an ion or a water molecule. It appeared that some compounds like glycerol, phosphate and citrate which co-crystallized with the protein due to their presence in the crystallization conditions were also mistakenly assigned as UNK. The pharmacophore method worked successfully in finding suitable ligand (s) for the protein.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: pharmacophore searching, Unknown proteins, function, Kd values, NMR titrations, docking, drug discovery, Hydrogen bond interactions, potential ligands, Cerius2, weblab viewer pro, Discovery studio visualizer (DSV), Accelrys, Vector method, query atom method, DSC, ITC, CD, Unknown ligands (UNK)
Subjects: Q Science > QD Chemistry
Q Science > QH Natural history > QH345 Biochemistry
Colleges/Schools: College of Science and Engineering > School of Chemistry
Supervisor's Name: Lapthorn, Dr. Adrian
Date of Award: 2013
Depositing User: Mr Musadiq Ibrahim
Unique ID: glathesis:2013-3924
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 01 Feb 2013 11:37
Last Modified: 04 Feb 2016 10:57

Actions (login required)

View Item View Item


Downloads per month over past year