Pharmacophore Modeling for the Prediction of
Compound Activity on Anti-Targets.
Case Study: The Pregnane X Receptor (PXR)
Daniela Niederreiter and Thierry Langer
Institute of Pharmacy, Department of Pharmaceutical Chemistry,
University of Innsbruck, Innrain 52, A-6020 Innsbruck, Austria
INTRODUCTION
Drug metabolizing enzymes and transporters
are often involved in clinically relevant drugdrug interactions. These functional proteins
can be induced by a wide range of
xenobiotics. A group of receptors known as
orphan nuclear receptors mediates this
effect.
RESULTS:
PHARMACOPHORE MODELS
The examination of the binding pocket
revealed potential sites of interactions which
were then transformed into a pharmacophore
model. Amino acid residues within the
binding pocket were included into the model
as excluded volume spheres.
The pregnane X receptor (PXR) – as a
member of this receptor family – regulates
the expression of multiple Cytochrome P450
families (e.g. CYP 3A and 2B), phase II
enzymes
(e.g.
UDP
glucuronosyl
transferases),
and
transporters
(e.g.
multidrug resistance protein 1). [1]
THE PXR FILTER
CASE STUDY: WDI
WDI: 48405 drugs
Database
Hypothesis A
Rifampicin fitted into the Hypothesis A.
potential
PXR ligands
1900 hits (3.9%)
Recently, the crystal structure of PXR cocrystallized with the co-activator peptide
SRC-1 and the ligand SR12813 was
published. [2] Therefore we used the 3D
coordinates of the PDB [3] entry 1NRL as a
template to develop a pharmacophore model
for PXR ligands.
Hypothesis B
Hyperforin fitted into the Hypothesis B.
potential highly
active PXR
ligands
701 hits (1.4%)
AF-698
1NRL, the crystal structure of PXR co-crystallized with
its co-activator peptide SRC-1 and the ligand
SR12813. [3]
Mycothiazole
Phytolaccagenate
Examples of potential PXR ligands derived by
the PXR filter.
CONCLUSION
AIM OF THE STUDY
Starting from the 3D coordinates of 1NRL [2], a
chemical feature-based pharmacophore model
was constructed and used as a prediction tool
for potential PXR activation.
METHODS
We used the software package CATALYST [4]
for:
• generation of structure models for the test set
• conformational analysis (Monte Carlo)
• manual generation and validation of the
pharmacophore models
• 3D-database search (Derwent WDI [5])
Conversion of binding information from the crystal
structure into a pharmacophore hypothesis
consisting of one hydrogen bond acceptor (green),
six hydrophobic features (turquoise), and 15
excluded volume spheres (grey).
Based on this pharmacophore model, a two
step filter for potential PXR ligands was
created.
Hypothesis A consisting of four features and
15 excluded volume spheres was able to
identify all PXR ligands of our test set.
Hypothesis B consisting of seven features –
three of them defined as “leave-one-out”, 15
excluded volume spheres, and a combined
shape (SR12813 or rifampicin) is able to
identify SR12813, hyperforin, and rifampicin
as highly potent PXR ligands.
LITERATURE
[1] Xie W, Uppal H, Saini SPS, Mu Y, Little JM, Radominska-Pandya A, Zermaitis MA. Orphan nuclear receptor-mediated xenobiotic regulation in drug
We achieved a filter consisting of two
hypotheses to predict PXR activity of
compounds present in a 3D database.
Hypothesis A was able to identify 100% of a
test set consisting of 31 PXR ligands as
active. However, it can only serve as a first
indicator for PXR activity because of many
false positive results.
Hypothesis B was shown to be able to identify
highly active PXR ligands, e.g. hyperforin and
rifampicin. It could be used to highlight
probably highly active PXR ligands out of a
hitlist derived from a database search with
Hypothesis A.
Our model could serve as a useful tool in the
early drug discovery process to identify the
potential of a new compound to activate PXR.
No fitting in Hypothesis A:
PXR activation unlikely
Fitting in Hypothesis B:
PXR activation possible
metabolism. DDT 2004;9(10):442-9
[2] Watkins RE, Davis- Searles PR, Lambert MH, Redinbo MR. Coactivator Binding Promotes the Specific Interaction Between Ligand and the Pregnane X
Receptor. J. Mol. Biol. 2003;331:815-28
Fitting in Hypothesis A and Hypothesis B:
PXR activation likely
[3] PDB – The Protein Data Bank. www.rcsb.org
[4] CATALYST Version 4.9, MSI, San Diego, CA, USA
[5] Derwent World Drug Index, Version 96, Derwent Ltd., London, UK, We thank Dr. Rémy Hoffmann ( Accelrys) for the performance of the search.
[email protected]