Ligand-based in silico modeling for the prediction and the identification of Tpl2 inhibitors
Georgia Melagraki, Varnavas Mouchlis and George Papadatos, NovaMechanics Ltd, Cyprus
Haralambos Sarimveis, National Technical University of Athens, Greece
Antreas Afantitis, NovaMechanics Ltd, Cyrpus
Tpl2 is a serine/threonine kinase in the MAP3K family that is upstream
of MEK in the ERK pathway. Recent studies using Tpl2 knock-out mice
indicate an important role for Tpl2 in the LPS-induced production of
TNF-a and other pro-inflammatory cytokines. Tpl2 is also required for
TNF-a signaling (rheumatoid arthritis target), and thus an inhibitor of
Tpl2 would have the double benefit of blocking both TNF-a production and
signaling. Furthermore, the unique features of Tpl2 presumably increase
the potential for discovering a selective Tpl2 inhibitor. [1-3].
In
this work, we have selected from the literature a large database of
small molecules which were recently evaluated as inhibitors of Tpl2
[1-3]. The first major result is the development of an accurate and
reliable QSAR model involving physicochemical and structural descriptors
that are able to predict successfully Tpl2 inhibition. The accuracy [4]
of the proposed in silico model is illustrated using the following
evaluation techniques: cross-validation, validation through an external
test set and Y-randomization. Furthermore, the domain of applicability
which indicates the area of reliable predictions is defined. A virtual
screening procedure could be based on the proposed QSAR model. The
method can also be used to screen existing databases or virtual
combinations to identify derivatives with desired activity. In this
scenario, the in silico model will be used to screen out compounds,
while the applicability domain will serve as a valuable tool to filter
out “dissimilar” combinations. The molecular descriptors used in the in
silico model encode information about the structure, branching,
electronic effects, and polarity of the modules and thus implicitly
account for cooperative effects between functional groups. The proposed
model aims to help researchers to design novel chemistry driven
molecules with desired biological activity.
References
1. L.A Gavrin et al., Bioorganic & Medicinal Chemistry Letters 15 (2005), 5288–5292
2. Y. Hu et al., Bioorganic & Medicinal Chemistry Letters 16 (2006), 6067–6072
3. N Green et al., J. Med. Chem. 50 (2007), 4728-4745
4. A. Afantitis, G. Melagraki et al., Eur. J. Med. Chem. 46 (2011), 497-508
Acknowledgments
This
work (ΔΕΣΜΗ 2008/ΕΠΙΧΕΙΡΗΣΕΙΣ/ΕΦΑΡΜ/0380/20) is co-funded by Republic
of Cyprus (Research Promotion Foundation) & European Regional
Development Fund (ERDF)
(presenting author: Antreas Afantitis)