A Network-Based Multi-Target Computational Estimation
Scheme for Anticoagulant Activities of Compounds
Qian Li1,2,4., Xudong Li1., Canghai Li3, Lirong Chen1*, Jun Song3, Yalin Tang2*, Xiaojie Xu1*
1Beijing National Laboratory for Molecular Sciences, State Key Lab of Rare Earth Material Chemistry and Applications, College of Chemistry and Molecular Engineering,
Peking University, Beijing, People’s Republic of China, 2Beijing National Laboratory for Molecular Sciences, Center for Molecular Sciences, State Key Laboratory for
Structural Chemistry of Unstable and Stable Species, Institute of Chemistry Chinese Academy of Sciences, Beijing, People’s Republic of China, 3Experimental Research
Center, China Academy of Chinese Medical Sciences, Beijing, People’s Republic of China, 4Graduate University of Chinese Academy of Sciences, Beijing, People’s Republic
Background: Traditional virtual screening method pays more attention on predicted binding affinity between drug
molecule and target related to a certain disease instead of phenotypic data of drug molecule against disease system, as is
often less effective on discovery of the drug which is used to treat many types of complex diseases. Virtual screening against
a complex disease by general network estimation has become feasible with the development of network biology and
system biology. More effective methods of computational estimation for the whole efficacy of a compound in a complex
disease system are needed, given the distinct weightiness of the different target in a biological process and the standpoint
that partial inhibition of several targets can be more efficient than the complete inhibition of a single target.
Methodology: We developed a novel approach by integrating the affinity predictions from multi-target docking studies
with biological network efficiency analysis to estimate the anticoagulant activities of compounds. From results of network
efficiency calculation for human clotting cascade, factor Xa and thrombin were identified as the two most fragile enzymes,
while the catalytic reaction mediated by complex IXa:VIIIa and the formation of the complex VIIIa:IXa were recognized as the
two most fragile biological matter in the human clotting cascade system. Furthermore, the method which combined
network efficiency with molecular docking scores was applied to estimate the anticoagulant activities of a serial of
argatroban intermediates and eight natural products respectively. The better correlation (r=0.671) between the
experimental data and the decrease of the network deficiency suggests that the approach could be a promising
computational systems biology tool to aid identification of anticoagulant activities of compounds in drug discovery.
Conclusions: This article proposes a network-based multi-target computational estimation method for anticoagulant
activities of compounds by combining network efficiency analysis with scoring function from molecular docking.
Citation: Li Q, Li X, Li C, Chen L, Song J, et al. (2011) A Network-Based Multi-Target Computational Estimation Scheme for Anticoagulant Activities of
Compounds. PLoS ONE 6(3): e14774. doi:10.1371/journal.pone.0014774
Editor: Jo ¨rg Langowski, German Cancer Research Center, Germany
Received November 18, 2009; Accepted February 19, 2011; Published March 22, 2011
Copyright: ? 2011 Li et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This project was supported by the National Science and Technology Major Project 2008ZX09401-006 and 2008ZX09202-007 (http://www.most.gov.cn/).
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com (LC); firstname.lastname@example.org (YT); email@example.com (XX)
. These authors contributed equally to this work.
The formation of a fibrin clot at the site of an injury to the wall
of a blood vessel is an essential part in stop blood loss after vascular
injury. The reactions that lead to the formation of fibrin clots are
commonly described as the clotting cascade, in which the product
of each step is an enzyme or cofactors necessary for the following
reactions to proceed effectively. The clotting cascade can be
divided into three parts, the extrinsic pathway, the intrinsic and
the common pathway. The extrinsic pathway begins with the
release of tissue factor at the site of vascular damage and leads to
the activation of factor X. The route provides an alternative
mechanism to activate factor X, from the activation of factor XII.
The common pathway is composed of steps linking the activation
of factor X to the formation of a multimeric, cross-linked fibrin
clot. Each of these processes includes not only a cascade of events
that generate the necessary catalyst for the formation of clots, but
also many positive and negative regulatory events.
As a result of advances of computational techniques and
hardware solutions, virtual screening has dramatically speeded up
modern lead identification and lead optimization. Ligand-based
and structure-based virtual screening are two most important
methods used in current computer aided drug design. Ligand-
based methods such as chemical similarity analysis and
pharmacophore modeling mainly focused on the features of
the active ligands structure. With high performance output,
ligand-based virtual screening was widely used to screen large
compound database. However, the fundamental problem of the
methods is that definition of what constitutes an active scaffold is
highly subjective. Synergized with X-ray crystallography, NMR
spectroscopy and isothermal titration calorimetry (ITC), structure-
based virtual screening has been used to complement experimental
PLoS ONE | www.plosone.org1 March 2011 | Volume 6 | Issue 3 | e14774
high-throughput screening (HTS) methods to improve the
efficiency and efficacy of discovering lead inhibitors[7–11].
Structure-based screens typically the molecular docking to fit
small organic molecules into targets of known structure, evaluate
them for structural and chemical complementary. In last few
years, investigators have also turned to predict new substrates for
enzymes or receptors of unknown function (such as the membrane
proteins) and to predicting potent small molecules based on multi-
With emergences of new paradigms in multi-target drug
discovery for several complex diseases, multi-target virtual
screening has been presented and executed to discover the
regimen which could target many different proteins and could
be of low cost, efficacy and better tolerance. However, the
importance and role of target in many complex disease systems
were not explicitly considered in the reported literatures about
multi-target virtual screening. Moreover, as most traditional
virtual screening method, more attention was paid on binding
affinity between drug molecule and target instead of phenotypic
data of drug molecule against disease system.
With the progress of system biology and bionetwork, we know
that the biological potency of an ideal drug may not merely
determined by the inhibition of a single target, but rather by the
rebalancing of several proteins or events, which contribute to the
etiology, pathogeneses, and progression of a complex disease [13–
26]. The available methodologies of in silico screening based on a
single target seem not effective in studying ligands’ effects on
biological process comprehensively for some cases[27,28]. In the
current work, a novel approach was developed by integrating the
predictions based on multi-target docking studies through
biological network efficiency analysis to estimate the biological
potency[26,29–31]. The work flow was shown in Figure 1. The
satisfactory predictions of our model were validated by the
experiments. Similar model to predict the biological potency of
drugs quantitatively by combining the multi-target virtual
screening and biological network calculation together have not
been yet reported in the past references. This novel model could
be a powerful tool for combinatorial drug discovery and the
development of multi-target drugs.
1 Constructions of the docking library
The docking library for multi-target virtual screening against
clotting cascade comprises 1177 compounds from 24 Traditional
Chinese Medicines (TMCs) that were widely used as components
of recipes against cardiac system diseases. These TCMs include 23
original plants and 1 original animal (their information can be
found in the Supporting Information S1). All compounds
identified in these TCMs were collected from Chinese Herbal
Drug Database developed in our group and other litera-
tures[5,33–35]. In addition, some active synthetic compounds
against coagulation cascade available to our laboratory were
included in the docking library, for example, seven argatroban
intermediates. The structures of these compounds were construct-
ed and minimized with the MMFF force field in Discovery
Studio molecular simulation system (DS, Accelrys Inc.). In
minimization, the threshold of root mean square deviation
(RMSD) of potential energy was set to 0.001 kcal?A˚-1?mol-1.
The optimized structures of all compounds were saved as sdf and
mol2 formats, respectively, for further docking study and were
included in the Supporting Information S1.
2 Network-Based dual-step hierarchical Computational
Fourteen proteins authorized as drug targets by US Food and
Drug Administration (FDA)wereused inthe virtualscreening based
on docking simulations. These targets include coagulation factor
Xa, thrombin, coagulation factor IXa, tissue factor:coagulation
factor VIIa complex, coagulation factor VIIa, fibrin, kallikrein,
tissue factor, prothrombin, von Willebrand factor, coagultaion
factor VIII, coagulation factor XI, fibrinogen, and coagulation
factor XIII. To reduce computational cost while not degrade the
calculation accuracy, two docking approaches, including Ligandfit
and Autodock, were successively employed to dock candidates to
the binding sites of these receptors in accordance with the order of
their docking simulation accuracies in network-based dual-step
hierarchical virtual screening. Top ten percent of hits from the
previous step were used for the next step. In every steps of serial
virtual screening, one candidate was estimated and ranked based on
its influence on the network efficiency of clotting cascade network
instead of the scoring functions of these binding poses on one target
as used in conventional virtual screening methods.
(a) Docking and scoring with Ligandfit.
structures of fourteen targets were retrieved from the Protein Data
Bank (PDB entries: 1FJS, 1TA2, 1RFN, 1W0Y,
1YGC, 2HLO, 2ANW, 1TFH, 1K22, 1AUQ,
3CDZ, 2F83, 1FZG and 1GGT). Hetero atoms
were removed from the receptors, and then hydrogen atoms were
added and wrong valence shells were corrected using Discovery
Studio. For receptor/ligand complex with crystal structure, the
binding site was defined as the grid points around the ligand which
were unoccupied by receptor atoms, whereas for a receptor without
crystal complex structure, potential binding sites were found based
on the shape of the receptor. Ligandfit protocol in Discovery Studio
was used to dock ligands into the specified site by the following steps:
(1). conformational search of candidate ligand for docking, (2).
ligand/site shape matching, (3). positioning the selected ligand
conformation into the binding site, and (4) rigid body energy
minimization of the candidate ligand pose/conformation using the
DockScore energy function and updating the saved list of ligands
with the candidate pose. Except maximum poses retained was set to
1, and default values were adopted for the other parameters. The
Piecewise Linear Potential 1 (PLP1) was selected for subsequent
Figure 1. The work flow of our virtual screening approach.
A New Virtual Screening Scheme
PLoS ONE | www.plosone.org2March 2011 | Volume 6 | Issue 3 | e14774
63. Shi CS, Huang HC, Wu HL, Kuo CH, Chang BI, et al. (2007) Salvianolic acid
B modulates hemostasis properties of human umbilical vein endothelial cells.
Thromb Res 119: 769–775.
64. Jiang R, Lau K, Hon P, Mak T, Woo K, et al. (2005) Chemistry and biological
activities of caffeic acid derivatives from Salvia miltiorrhiza. Curr Med Chem 12:
65. Mitsos A, Melas IN, Siminelakis P, Chairakaki AD, Saez-Rodriguez J, et al.
(2009) Identifying drug effects via pathway alterations using an integer linear
programming optimization formulation on phosphoproteomic data. PLoS
Comput Biol 5: e1000591.
A New Virtual Screening Scheme
PLoS ONE | www.plosone.org9 March 2011 | Volume 6 | Issue 3 | e14774