Case Study


DNA-Gyrase

HSP-90

Pyruvate Kinase Dehydrogenase

Histidine Kinase


All these proteins have a related, druggable domain that retains almost no sequence identity.


All four proteins have been considered as drug targets. We discuss the identification process and highlight key structural features.

ebisu group

chemogenomix © 2012

Read more below

Overview

This case study shows the structural similarity between the ATP binding domains of four proteins that have all been considered as potential drug discovery targets, including DNA gyrase B which is the target for novobiocin. The study shows how collective consideration of all four structures can assist drug discovery. For more detailed information see our original Nature publication: Tanaka et al., 1998, 396, 88-92.

Structural Similarity

Irrespective of whether a relationship to a larger family is predicted on the basis of sequence or identified using structure comparison, the objective is the same; to transfer useful information from the well annotated protein(s) to those without productive annotation.

In our historical case, we were interested in the function of histidine kinases and related proteins which form roles such as osmolarity sensing in bacteria and had been considered as potential antimicrobial drug targets.

So at that time, it was with considerable excitement that we found it was related to DNA Gyrase, the target of the drug novobiocin (see image to right)
and Heat Shock Protein 90 which had just started to be considered seriously as a potential anti-cancer target (now of course HSP90 is viewed as a strong prospect, with many projects in clinical trials)

The alignment below adapted from our original publication emphasises two key elements of distant relationships. Firstly the low overall sequence identity and secondly the clustering of the few identities in specific locations. Subsequent analysis show these key residues to be involved with function, in this case the ATP binding site.


The conserved colours of beta strands emphasises the structural relationship and this is also presented as a toplogical digram with similar colouring. The DxGxG motif is clearly highly conserved and th
is lies at a key point for ATP binding.

Those with a particular interest in protein structure may have already noticed a left-handed connection from the yellow to orange strands in EnvZ. Surprisingly the DNA binding domain of DNA gyrase B also contains one of these rarely seen connections (same colour but striped). So it seems possible, even likely, that a deletion event during evolution led to the formation of the EnvZ protein.


Additional Relationships

Subsequent to this work being published, other proteins that were being considered as potential drug targets for different diseases came to light. One was pyruvate dehydrogenase kinase, a human protein associated with hyperglycaemia. This too has the same ATP binding domain. In the future it is possible that other human proteins related to pyruvate dehydrogenase kinase and HSP90 might be considered for different therapeutic treatments. If so, the wealth of chemistry and function data already available for this family could be transferred to the new protein.

The disease relevance of pyruvate dehydrogenase kinase is summarised simply in wikipedia from which this paragraph is adapted: Cells lacking insulin or are insensitive to insulin overexpress pyruvate dehydrogenase kinase4 and as a result, the pyruvate formed from glycolysis cannot be oxidized. This leads to hyperglycaemia, due to the fact that glucose in the blood cannot be used efficiently. It was therefore consider that PDK4 inhibitors might form a treatment for type II diabetes. In other experiments, PDK1 was shown to have increased activity in hypoxic cancer cells due to the presence of HIF-1. PDK1 shunts pyruvate away from the citric acid cycle and keeps the hypoxic cell alive so inhibition might serve as part of an antitumor therapy. Similarly, PDK3 has been shown to be overexpressed in colon cancer cell lines. Proposed inhibitors such as AZD7545 that bind to PDK1 are still progressing and Radicicol binds to PDK3.


Fragment Based Drug Discovery

The aspartate of the conserved DxGxG motif previously referred to also occurs in other publications. In publications on FBDD by two companies; Vernalis and Astex the Asp is specifically highlighted used essentially as an anchor point for many of the fragments. The following image shows four fragments from these publications interacting with the Asp (shown to the fore in blue). All the interactions are shown in a common orientation using Ligplot+. Vernalis structures are: (a) 2wi1 and (b) 2wi7; and Astex structures are: (c) 2xdl and (d) 2xab.

It would seem clear that not all residue positions in a binding site are equal and alignments of distant homologues can help to pinpoint these key regions. Such observations should have positive implications for pharmacophore screening and docking.

For more details see our publication and references therein. LigPlot+: Multiple Ligand-Protein Interaction Diagrams for Drug Discovery, Roman A. Laskowski and Mark B. Swindells. Journal of Chemical Information and Modelling. 2011, 51, 2778–2786