3.9 TASSFUN (MCM)
Thrombo-embolic
diseases are quite common in humans. They are often affected by enzymes
of the
blood coagulation cascade, like thrombin, factor Xa (fXa) and urokinase
type
plasminogen activator (uPA), so the specific inhibition of these
enzymes is one
of the major goals in drug design. In this project Comparative binding
Energy
(COMBINE) analysis should be developed for generating target-specific
scoring
functions (TASSFUN), which will be applied for virtual screening and
for
predicting the bioactivity of new inhibitors.
The COMBINE analysis
based on a training set of experimental determined
protein-inhibitor-complex crystal
structures as well as on bioactivity values, e.g. inhibitor constants.
For these
complexes the electrostatic and van der Waals interaction energies will
be
calculated between the inhibitor and each protein residue. In addition,
electrostatic desolvation energy terms are computed by solving the
Poisson-Boltzmann equation for the protein and the inhibitor. The
decomposed
interaction energies and the electrostatic desolvation energy terms
formed
Xvariables in a matrix and were correlated by Partial Least
Squares (PLS)
to bioactivity or binding free energy values given as Yvariables.
Subsequently, the correlation or scoring function will be used to
predict the
bioactivity of inhibitors where no experimental values available.
In the first year
of the two-year-project, we focused on acquisition of published data
(e.g.,
structure and kinetic information) of blood coagulation cascade-related
enzymes, as well as on programming and implementing necessary tools for
handling a large number of protein-ligand-complexes. The programs have
reached
a stage, which makes it possible to build COMBINE models based on a
training
set of structures of complexes. These COMBINE models were applied to a
large
test set of ligands with unknown crystal structures but docked ligand
conformations.
For testing the
procedure, ligands with known binding conformation and inhibitor
constants were
minimized by molecular mechanics together with a urokinase receptor
model.
After calculating and decomposing the interaction energy terms and the
correlation to inhibitor constants, further ligands were docked by a
commercial
program into the receptor model. A comparison between predicted and
experimental inhibitor constants showed promising results.
In the next part
of the project we will focus on building target-specific scoring
functions for
different enzymes to use these after validation for virtual screening
of large
compound libraries with the aim of finding target-specific inhibitors.
Figure: Visualization of interaction
and desolvation energy terms. Electrostatic (left) and van der Waals
(middle) interaction
values between the receptor model of urokinase and ligands of the
training set
were mapped as colours onto receptor surface (in red stabilization and
in blue
destabilization areas for complex formation). In the right panel the
back bone
of urokinase (green) is shown together with desolvation energy term of
a
ligand.
6
Professional Activities
6.3
Presentations (incl: Talks, Posters, Demos)