Abstract—‘OMICS’ techniques have deeply changed the
drug discovery process. The availability of many different
potential druggable genes, generated by these new techniques,
have exploited the complexity of new lead compounds
screening. ‘Virtual screening’, based on the integration of
different analytical tools on high performance hardware
platforms, has speeded up the search for new chemical entities
suitable for experimental validation. Docking is a key step in
the screening process. The aim of this paper is the evaluation of
binding differences due to solvation. We have compared two
commonly used software, one of which takes into account
solvation, on a set of small molecules (Morpholines, flavonoids
and imidazoles) which are able to target the RAC1 protein - a
cardiovascular target. We have evaluated the degree of
agreement between the two different programs using a machine
learning approach combined with statistical test. Our analysis,
on a sample of small molecules, has pointed out that 35% of the
molecules seem to be sensitive to solvation. This result, even
though quite preliminary, stresses the need to combine
different algorithms to obtain a more reliable filtered set of
ligands.
I. INTRODUCTION
HE investigation of the molecular mechanism of the
diseases (Molecular Medicine) is one of the major
challenges in the post-genome era. From a
pharmacological point of view the identification of genes,
involved in a specific pathology, (markers) is an essential
step. In addition, is it also important to elucidate the
functional interactions among them. From a molecular point
of view a disease is characterized by altered genomic and
proteomic profiles [1]. Statistical analysis allows to
determine the set of critical gene or proteins in a panel of
samples. ‘Omics’ has heavily impacted the drug discovery
process. The high throughput screening method, developed
during the sequencing projects, has increased the capability
to screen, in a parallel way, thousand potential targets.
Despite the availability of all these new technologies, the
productivity, as new potential drugs, of pharmaceutical
companies is declined.
The identification of new chemical compounds suitable for
high throughput pharmacological screening has become
computationally intensive. The chemoinformatic screening
is also limited by data accessibility because many
Manuscript received April 15, 2011. Patrizio Arrigo is with the CNR
Institute of Macromolecular Studies (ISMAC), Section of Genoa, Via De
Marini 6, 16149 Genoa, Italy (E-mail: arrigo@ge.ismac.cnr.it). All other
authors are with the Department of Communication, Computer and System
Sciences, Nanobiotechnology and Medical Informatics Laboratory
University of Genoa, Via all’Opera Pia, 13, 16145 Genoa, Italy
(*corresponding author, phone: +390103532991; fax: +390103532154; E-
mail: carmel@dist.unige.it).
compounds are under copyright and there not information
about them. In the last decade the Computer Aided Drug
Design (CADD) has increase its complexity including
genetic information in the screening process
(pharmacogenetics). In the past the role of genetic variability
was not considered, now the effect of genetic variability on
drug efficacy is a well-established knowledge. In many cases
different proteins forms, associated to a slight variation at
genomic level, can be targeted by a drug in a more or less
efficient way. On the other site a drug can induce
differential gene expressions. The evaluation of genomic
modification, triggered by drugs, is the aim of
pharmacogenomics. It is quite obvious that, for virtual
screening, pharmacogenetic knowledge has a critical
influence. The identification of new drug targets requires the
integration of bioinformatics and chemoinformatics with
experimental results. These data are the knowledge
background for the QSAR (Quantitative Structure-activity
Relationship) modelling. At higher level the final step of the
discovery process is related to the capability to evaluate the
efficacy by the ADME (Adsorbtion Diffusion Metabolism
and Excretion) simulation. In order to improve the
performance of target identification and design of new
drugs, bioinformatics and chemoinformatics tools have to be
integrated.
The use of data mining techniques related to gene expression
level during FP6 Cardioworkbench project
(www.cardioworkbench.eu) has highlighted the role of Rac1
protein in cardiovascular diseases [2]. In particular Rac1
protein is implicated in several events of atherosclerotic
plaque development and represents a new potential
pharmacological target for cardiovascular diseases. Rac1 is
member of the Rho family. This protein family serve as
conformational switches in a wide variety of signal
transduction pathways. It is a key regulators of angiogenesis,
modulating a diversity of cellular processes, including
vascular permeability, extracellular matrix remodeling,
migration, proliferation, morphogenesis, and survival [3].
Rac1 plays a central role in endothelial cell migration,
tubulogenesis, adhesion, and permeability in response to
vascular endothelial growth factor (VEGF) and sphingosine-
1-phosphate (S1P), which is likely due to the inability of
Rac1-deficient endothelial cells to form lamellipodial
structures and focal adhesions, and to remodel their cell-cell
contacts [4]. It is been demonstrate that the activation of
Rac1, but not RhoA, in human aortic smooth muscle cells
(SMCs) through the engagement of Į2ȕ1 integrin by type I
collagen induces the expression of matrix metalloproteinase
1 (MMP1) and MMP2, an event that may contribute to
atherosclerotic plaque rupture [2].
Drug Design For Cardiovascular Disease: The Effect Of Solvation
Energy On Rac1-Ligand Interactions
Norbert Maggi, Student Member, IEEE, Patrizio Arrigo, and Carmelina Ruggiero, Member, IEEE
T
978-1-4244-4122-8/11/$26.00 ©2011 IEEE 3237
33rd Annual International Conference of the IEEE EMBS
Boston, Massachusetts USA, August 30 - September 3, 2011