Synthesis, bioassay, and molecular field topology analysis of diverse vasodilatory heterocycles

Article abstract of DOI:10.1021/ci400723m

A diverse training set composed of 76 in-house synthesized and 61 collected from the literature was subjected to molecular field topology analysis. This resulted in a high-quality quantitative structure-activity relationships model (R² = 0.932,

Full text of DOI:10.1021/ci400723m

Article
pubs.acs.org/jcim
Synthesis, Bioassay, and Molecular Field Topology Analysis of
Diverse Vasodilatory Heterocycles
Polina V. Oliferenko,¶ Alexander A. Oliferenko,¶ Adel S. Girgis,¶,‡ Dalia O. Saleh,§ Aladdin M. Srour,∥
Riham F. George,⊥ Girinath G. Pillai,¶,# Chandramukhi S. Panda,¶ C. Dennis Hall,*,¶
and Alan R. Katritzky¶,∇,○
¶Department of Chemistry, University of Florida, Gainesville, Florida 32611-7200, United States
§
∥
‡Pesticide Chemistry Department, Pharmacology Department, and Therapeutical Chemistry Department, National Research
Centre, Dokki, Cairo 12622, Egypt
⊥Pharmaceutical Chemistry Department, Faculty of Pharmacy, Cairo University, Cairo, Egypt
#Department of Chemistry, University of Tartu, Ravila 14a, Tartu 50411, Estonia
∇Department of Chemistry, King Abdulaziz University, Jeddah 21589, Saudi Arabia
S
* Supporting Information
ABSTRACT: A diverse training set composed of 76 in-house
synthesized and 61 collected from the literature was subjected
to molecular field topology analysis. This resulted in a high-
quality quantitative structure−activity relationships model (R2
= 0.932, Q2 = 0.809) which was used for the topological
functional core identification and prediction of vasodilatory
activity of 19 novel pyridinecarbonitriles, which turned out to
be active in experimental bioassay.
data on vasorelaxation activities are available for various chemical
families of potent vasodilators.6−12
INTRODUCTION
■
Hypertension is a major risk factor for coronary heart and
vascular diseases, stroke, and renal insufficiency which affects
more than 26% of adults,1,2 including some 65 million individuals
in the United States.3 Since the 1950s, when the first clinically
useful antihypertensive drugs appeared, the strategy of such drug
therapy has remained mainly unchanged, i.e., based on
prevention of vasoconstriction or increased sodium excretion.4
The development of new, more effective and selective
antihypertensive drugs is of crucial importance for the treatment
of hypertension as well as control of its prevalence.
Vasodilators are a large group of antihypertensive drugs which
cause smooth muscle relaxation in blood vessels leading to a drop
of arterial pressure. Vasodilating drugs are classified according to
their primary mechanism of action such as α-adrenoceptor
antagonists (quinazoline derivatives), calcium channel blockers
(1,4-dihydropyridine-3,5-dicarboxylates), potassium channel
openers (6-piperidin-1-ylpyrimidine-2,4-diamine-3-oxide),
phosphodiesterase inhibitors (substituted 2-pyridinones, com-
plex heterocyclic systems), nitrodilators, and others.5 Nitro-
dilators such as the ubiquitous nitroglycerol were the first
vasodilators widely used in clinical practice, but it was found long
ago that nitroglycerol can only be administered for short periods
of time because of rapidly developed tolerance. Experimental
The chemical diversity of possible vasodilatory compounds is
thus enormous. According to the literature, many heterocyclic
families are or can be potent vasodilators, but any experimental,
high-throughput screening is barely feasible because the chemical
range spanned by potentially vasodilatory heterocycles is vast. A
viable alternative to explore this huge chemically diverse space
would be in silico analysis including quantitative structure−
activity relationships (QSAR) and virtual screening. Notable
accounts of computational vasodilatation activity modeling
include: (i) linear discriminant analysis of resveratrol-coumarin
hybrids;13 (ii) pharmacophore mapping of pyridazinone
derivatives;14 (iii) 2D-QSAR modeling of flavonoids;15 and
(iv) 3D-pharmacophore and homology modeling studies of α1-
AR antagonists.6d,e The limited number of previous computa-
tional studies is, at least in part, due to the scarcity of available
experimental data: only a few sources provide data numerous
enough (at least 5−6 experimental data points per descriptor) for
QSAR analysis. In the present work we advantageously use
molecular field topology analysis (MFTA)16 to explore the
chemical space of existing and possible vasodilators and also to
Received: December 6, 2013
Published: February 16, 2014
© 2014 American Chemical Society
1103
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
Table 1. Experimentally measured and MFTA predicted IC50 (mM) values for 138 training set compounds
1104
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
Table 1. continued
1105
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
Table 1. continued
build predictive models accurate enough for further de novo
design. MFTA is known to be a powerful QSAR method
previously successfully applied to model various biological
activities.17−19As a superposition technique, it is designed to
find common fragments and thus to deal with an array of
structurally diverse data sets. Previously published and new data
recently obtained in our laboratories are collated to make up a
combined diverse data set comprised of several heterocyclic
1106
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
Figure 1. Compounds of the training data set: (a) 2-substituted-4,6-diaryl-3-pyridinecarboxylates, (b) 2-substituted-4,6-diaryl-3-pyridinecarbamides,
(c) 4H-pyrano[3,2-c]pyridine-3-carbonitriles, (d) 7,9-dioxa-1,2-diaza-spiro[4,5]dec-2-ene-6,10-diones, (e) 1,2,7,9-tetraaza-spiro[4.5]dec-2-ene-6,8,10-
triones, (f) 2-alkoxy-4-aryl-6-(1H-benzimidazol-2-yl)-3-pyridinecarbonitrile, (g) 1,8-naphthyridines, (h) phenylcoumarins, (i) furocoumarines, (j)
pyridazinones, (k) N-acylhydrazones, (l) organic nitrates.
Figure 2. Some known vasodilator drugs structurally related to the training set compounds from Figure 1: (a) amlodipine, Ca2+ channel blocker; (b)
milrinone, phosphodiesterase III inhibitor; (c) minoxidil, K+ channel opener; (d) prazosin, α1-adrenoreceptor antagonist. Vasoconstricor agents, class of
α1-adrenoreceptor agonists: (e) norepinephrine; (f) phenylephrine.
families possessing different mechanisms of action (pyridine-
carboxamides 42−55 were previously reported by our group
screened for their vasodilation properties and adopted as training
set analogues, Table 1). The QSAR approach employed in this
paper also suggests a pharmacophore hypothesis, in which a
topological analogue of pharmacophore named a functional core
is defined as a molecular subgraph occurring in all training set
structures.
data points collected from the literature, namely 1,8-naphthyr-
idine,7 furocoumarin,11 phenylcoumarin,8,22 4,5-dihydro-3(2H)-
pyridazinone, N-acylhydrazone,10 and organic nitrate deriva-
tives.12
The mechanisms of action of these compounds have not yet
been fully identified, but the possibility exists that they have the
same origin as those of some known structurally related
vasorelaxing agents. For example, nicotinamide can inhibit
directly vascular smooth muscle cell contraction by calmodulin,
myosin light chain kinase, and adenylate cyclase activation23 or
induce vasodilation through binding with specific nicotinic
acetylcholine receptors, which results in release of adrenergic
neurotransmitters and nitric oxide.24,25 A class of structurally
very similar 1,4-dihydropyridine-3,5-dicarboxylate vasosilators
(amlodipine, manidipine, felodipine, etc., see Figure 2) are
calcium channel blockers stimulating emission of nitric oxide as
well.26 Antihypertensive activity of 1-oxa-4,9-diazaspiro[5.5]-
undecan-3-ones structurally related to 7,9-dioxa-1,2-diaza-spiro-
[4,5]dec-2-ene-6,10-diones and 1,2,7,9-tetraaza-spiro[4.5]dec-2-
ene-6,8,10-triones is due predominantly to peripheral α1-
adrenergic blockade.27
As for the compounds collected from the literature (78−138),
their modes of action can be rather diverse. Thus, there is
evidence that various 1,8-naphthyridines can act as guanilate-
cyclase inhibitors, potassium channel openers,7 and α1-
adrenoceptor antagonists.28 Vasorelaxing potency of pyridazi-
none and N-acylhydrazone derivatives is explained by their
ability to inhibit phosphodiesterase types III14 and V,29
respectively. The effect of furocoumarins may be related to
Ca2+ agonist activity and enhancement of prostaglandins release.
METHODS
■
Data set. Experimental data were collected on vasodilatation
activity expressed as concentrations necessary for 50% reduction
of maximal induced contracture, IC50, as the measure most
commonly used in pharmacological studies and thus more
abundantly present in the literature. In all cases experimental data
were obtained according to the same standard protocol: thoracic
aortae of rats were precontracted by treatment with norepi-
nephrine or phenylephrine, solutions of test compounds added,
and relaxation response of aortae measured in a concentration-
dependent manner. If original concentrations were given in units
other than mM, they were recalculated before −log IC50 values
were computed.
General structure patterns of training set compounds are
shown in Figure 1, with numerical data summarized in Table 1.
Compounds comprising the data set include: (i) 76 in-house
synthesized nicotinate esters and amides,6a−c 2-amino-8a-
methoxy-4H-pyrano[3,2-c]pyridine-3-carbonitriles,6d 7,9-dioxa-
1,2-diaza-spiro[4,5]dec-2-ene-6,10-diones,6e 1,2,7,9-tetraaza-
spiro[4.5]dec-2-ene-6,8,10-triones,20 and 2-alkoxy-4-aryl-6-
(1H-benzimidazol-2-yl)-3-pyridinecarbonitriles21 and (ii) 61
1107
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
Figure 3. Characteristics of the MFTA model obtained for the overall set of 138 compounds: (a) factor dynamics plot for correlation coefficient R and
cross-validation Q10%2 and (b) “observed vs predicted” fitting plot (right). Seven chemotypes of vasodilators are marked in different colors.
Figure 4. MSG built for the training data set of 138 structures with representative superimposed molecules. The numbers correspond to the numbers of
the molecules in Table 1.
Finally, a group of nitrodilators, organic nitrates, is suggested to
undergo an enzymatic processing to release NO, an important
endothelial-derived smooth muscle relaxing agent with multiplex
vascular functions.
Molecular Field Topology Analysis. MFTA is an analytical
tool for the construction of quantitative structure−activity
relationships of structurally related compounds. It considers
structural similarity both locally (when specific subgraphs within
the two molecular graphs are compared) and globally (when two
molecular graphs are compared). MFTA represents molecular
structures in the form of molecular graphs, which are
subsequently superimposed onto the emerging molecular
supergraph. The ultimate shape of the two-dimensional
molecular supergraph (MSG) takes into account the common-
ality of distinct common scaffolds as well as of disjoint parts of the
molecules. The MSG vertices and edges correspond to the atoms
and bonds, respectively, and are assigned the values of local
descriptors, which are then processed further by partial least-
squares regression to obtain a structure−activity model. The
descriptor pool utilized includes charge, electronegativity,
1108
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
hydrogen bonding, van der Waals radius, lipophilicity, and
electrotopological descriptors. The predictive quality of a model
is characterized by the squared correlation coefficient, R2, and the
cross-validation parameter Qn , where n is a user-defined
parameter for the number of structures in a test set.
Interpretation of MFTA Models. A distinguishing feature
of MFTA and similar 2D and 3D-QSAR methods is that they
employ local characteristics versus whole-molecule descriptors
used in multiple-regression-based QSAR/QSPR approaches. As
a result, a target activity is interpreted in terms of vectors rather
than constants, i.e., it can be regulated by increasing or decreasing
particular atomic parameters by replacing one atom by another
or changing nearest neighborhood.
2
RESULTS AND DISCUSSION
■
MFTA Modeling. The MFTA structure−activity model,
generated for 138 compounds of the training set, includes five
molecular fields or descriptors: Gasteiger−Marsili atomic
charges, van der Waals radii, local atomic lipophilicities, and
hydrogen-bond acceptor and donor scales.
The impacts of molecular fields or descriptors on the general
model are reflected by two-colored images displayed in Figure 5.
The factor dynamics and “observed vs predicted” fitting plots
are shown in Figure 3. As can be seen in Figure 3a, both the
correlation coefficient R and the squared cross-validation
coefficient Q2 grow almost in parallel until the number of
components reaches six: above this, Q2 starts to exhibit a decrease
as the number of the factor increases. When six factors were taken
into account, the squared correlation coefficient was R2 = 0.932
and the predictive ability characterized by a 10%-out cross-
validation, Q10%2, was 0.809, with the mean unsigned error equal
to 0.187.
The data points on the correlation plot shown in Figure 3b are
marked in different colors to distinguish between the seven
chemotypes. As can be seen, chemical classes fall roughly into
two subgroups: a more compact arrangement of less active
compounds and less numerous points with higher pIC50 values.
One can clearly see that naphthiridines cover the range of more
than 2.5 log units and that each of the seven chemotypes overlap
with at least two others, except for organic nitrates, which overlap
only with the pyridazinone series. Taken together, the data form
a smooth and evenly distributed outline spanning over almost 5.0
logarithmic units.
The molecular supergraph, constructed by the superimposi-
tion of all the training set structures, is shown in Figure 4 along
with examples of superimposed structures. As expected, the MSG
is complex because of the high diversity of the constituent
molecules. As can be seen, the molecular subgraphs of all
chemical groups except organic nitrates have at least one
fragment shared with two or more other subgraphs, which attests
to the overall consistency of the MSG. Such an even character of
subgraphs superimposition leads to an equivalently even
manifestation of the features thereof. This opens up an
opportunity to suggest new prospective vasodilators by the
analysis of the MSG’s different fragments along with the
descriptor contributions (see Interpretation of MFTA Models
section).
Figure 5. Relative impacts of the molecular fields on the title activity for
the model built on 138 data points: atomic charge (Q), atomic van der
Waals radius (vdW), hydrogen-bond acceptor (HBa), hydrogen-bond
donor (HBb), and atomic lipophilicity (La). Red and blue colors
indicate, respectively, increase or decrease of local descriptor values that
enhance the target activity. Positions discussed in the text are marked in
green in MSG. The red circle indicates the position of the functional
core.
As MFTA was applied for a heterogeneous data set composed
of different chemotypes with presumably diverse mechanisms of
action, one can find it highly relevant to compare the prediction
quality of the general MFTA model with that of the models built
for each individual chemotype. Thus, a separate MFTA model
was generated for a subset of 54 pyridine derivatives (#1−#19,
#42−#76) including pyridinecarboxylates, pyridinecarbamides,
and pyridinecarbonitriles. The statistical quality was the best with
Red color indicates that an increase of a local descriptor value
leads to an enhancement of activity, whereas blue color denotes
that magnification of activity can be achieved by a decrease of the
local descriptor value. As was mentioned earlier, merging of local
influences which pertain at least three different structural
fragments significantly complicates analysis of individual
influences, but it is a beneficial factor for the development of
novel, potentially active agents.
2
two factors: R2 = 0.675, Q10% = 0.600; mean unsigned error
equal to 0.117. Another two-factor model was built for 22
naphthiridines: R2 = 0.778, Q10%2 = 0.619; mean unsigned error
0.259. So, the general MFTA model statistically outperforms the
individual chemotype models, which supports validity of the
choice to build a single general, heterogeneous MFTA model
rather than several focused ones.
Visual analysis of the MSG images shown in Figure 5 reveals a
mutual correlation between descriptor impacts at certain
1109
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
positions. For example, concurrent rendering of position 11 in
blue for electrostatic (Q) and steric (vdW) fields and in red for
lipophilicity (La) field indicate that a more negative charge, a
smaller volume and higher atomic lipophilicity are preferable at
this location. Taking into account that position 11 is occupied by
carbon and nitrogen atoms and comparing respective descriptor
values in Table 2 suggest that an sp3-hybridized heterocyclic
one can hypothesize an efficiency of a bulky substituent in
pyridine-3-carboxylates and pyridine-3-carboxamides, which
contain a strong negatively charged fragment such as nitro or
nitrate group.
Identification of Topological Functional Core (Phar-
macophore). All classes of compounds comprising the training
set presumably have different modes of action, which do not have
much in common at least from the topological point of view.
Based on the MFTA models described above, one can deduce a
functional core center from the MSG structures. Here we can
speak about a topological analog of a pharmacophore, because
MFTA superimposes structures in the topological or graph-
theoretical way. The notion of topological pharmacophore can
be formulated differently. The very first definition probably goes
back to the early work of Franke et al.,30 where a “topological
pattern” or “topological pharmacophore” was regarded as “an
ensemble of structural features that is characteristic of a group of
compounds possessing a desired biological property”. Another
definition is based on pairwise topological distances proposed by
Schneider31 to account for isofunctionality of structurally distant
compounds. Bonachera et al. applied fuzzy logic to define
tricentric pharmacophore descriptors.32 The field of topological
pharmacophore and its application in drug design was
comprehensively reviewed by Horvath in Chapter 2 of
“Chemoinformatics Approaches to Virtual Screening”.33 We
define a topological functional core as a subgraph (not
necessarily smallest) of the MFTA molecular supergraph,
which is common in all training set structures. This definition
is related to the clique or a “maximal completely connected
graph”, but because of heuristic assumptions is not as rigorous as
the definition of clique.
Table 2. Descriptor Values for the Atom Types Occupying
Position 11 of the MSG
atom type
Q
vdW
La
2
Csp (benzyl ring)
0.0614; −0.0010; −0.015
0.1577
1.7
1.7
−0.1378; −0.0548
−0.1703
2
Csp (CO)
Nsp (heterocycle ring)
3
−0.0923
−0.0623
1.55
1.55
0.1414
2
Nsp
0.1757
nitrogen atom is more favorable than an sp2-hybridized carbon
atom. High activities of pyridazinones (str. 119−123) with the
3
Nsp atom in position 11 (see structure 120 in Figure 4) serve as a
good validation of this conclusion.
An opposite situation is observed for position 24, which
demands a concurrent increase in atomic charge and a decrease
in atomic volume and lipophilicity, while being populated by a
variety of atomic types (see Table 3 for details). At the same time
Table 3. Descriptors for the Atom Types Occupying Position
24 of the MSG
atom type
NH2
Q
vdW
La
HBd
−0.139
−0.132
−0.198
−0.109
−0.204
−0.0873
−0.058
−0.025
1.55
1.55
1.52
1.5
0.141
0.141
0.6
3
Nsp (cycle)
−2.00
−2.00
−2.00
2.10
Analysis of vasodilators in Figure 2 and structures in the data
set together with their superimpositions on the MSG exemplified
in Figure 4 enables one to define the key structural elements
common to all molecules. One of them is the structural motif
shown in blue in Figure 6a. This motif is present in pyridazinones
3
Osp (−OCH3)
0.341
2
Osp (CO)
−0.173
0.521
OH
1.52
1.75
1.8
Cl
0.697
−2.00
−2.00
−2.00
SH
1.015
CH3
1.7
−0.633
the presence of hydrogen-bond donors such as hydroxyl and
amino groups are not beneficial, as indicated by the strong blue
hue of position 24 on the hydrogen-bond donor field. According
to the data in Table 3, the CH3 group has the optimum
combination of all four descriptor values.
Other informative parts in the MSG descriptor images are the
conglomerates of adjacent five- and six-membered rings marked
with a and b for simplicity. Ring a is constructed by
superimposition of aryl fragments of 3-pyridinecarboxylates, 3-
pyridinecarbamides, and 3-pyridinecarbonitriles as well as thienyl
and furanyl rings in spiro compounds. Predominance of the blue
hue for steric and lipophilic fields attests that less bulky and less
lipophilic systems are more beneficial compared to aromatic
hydrocarbon cores. Similar features are observed in color
rendering of the b ring occupied particularly by a thienyl
fragment in N-acylhydrazones. Analysis reveals that aliphatic
carbon atoms are preferred compared to the benzene and five-
membered rings, because they are smaller and lipophilic. The
overall analysis of steric and lipophilic characteristics of the
fragment c suggests that an aliphatic C5−C6 linear moiety ending
up with a bulky electronegative group could contribute better to
the target activity. Such a moiety is present in organic nitrates
which, as shown in Figure 4 (str. 137), are superimposed onto
this particular section of the MSG. Based on these conclusions,
Figure 6. (a,b) Structural fragments common for molecules of the
training data set. (c) Topological functional core proposed on the basis
of MFTA molecular supergraph and descriptor field’s analysis.
and N-acylhydrazones; a very similar motif can be found in
diazaspiro and tetraazospiro compounds. The second commonly
met and partially overlapping with the first motif is the
combination shown in Figure 6b; it is incorporated in the
molecular core of 3-pyridinecarboxamides, 3-pyridinecarboxy-
lates, 3-pyridinecarbonitriles, 4H-pyrano[3,2-c]pyridine-3-car-
bonitriles, 1,8-naphthyridines, and furocoumarines. All these
fragments concurrently superimposed on the MSG within the
same topological area are marked on Figure 5 with a red circle.
1110
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
We define this fragment as a common topological pattern or
topological functional core.
experimental ones at higher than 0.1 mM (compounds 148,
149, and 165 reveal error in their IC50 values 0.255, 0.137, and
0.181, respectively). The correlation itself between experimental
and predicted values for all 19 compounds turned out to be low,
but upon removal of two major outliers (compounds 147 and
152), the correlation coefficient significantly improved to 0.620.
As all the synthesized compounds in the test set belong to the
same chemotype, it was suggested highly relevant to predict their
activities using the model built on a smaller training set of 54
pyridine derivatives (#1−#19, #42−#76), including pyridine-
carboxylates, pyridinecarbamides, and pyridinecarbonitriles. The
statistical quality of this model was the highest when two factors
were taken into account: R2 = 0.675, Q2 = 0.600, and the mean
unsigned error was equal to 0.117. Prediction of activities for 19
thienyl-nicotinonitrile derivatives using the model for 54
pyridine compounds turned out to be much inferior to the
general model shown in Figure 3, even after outliers 147 and 152
were excluded.
The topological functional core shown in Figure 6c embraces
the whole general data set of 138 compounds. It was derived from
analysis of the above supergraph as well as the chemical
structures themselves. This core consists basically of a fused
bicyclic fragment enriched with heteroatoms. It also includes
functional groups in the immediate vicinity of the bicyclic
fragment, which imparts not only topological but also chemical
similarity. On this graph R stands for alkyl substituents and Ar1 is
the aryl or other bulky substituent which contains an alkyl chain
with a strong electronegative group at the end, such as nitro or
carboxyl group. This functional core may or may not be the only
possible one, but it demonstrates a good quality QSAR, as
represented by R2 = 0.932 and Q2 = 0.809. The goal of deriving
such a functional core is not prediction per se but rather a starting
point for de novo design by combining a common topological
pattern with favorable structural features inherited from various
chemotypes.
Prediction of Vasodilatory Activity with the MFTA
Model. MFTA model was applied to the prediction of
vasodilatory activity of 19 newly synthesized thienyl-nicotinoni-
trile derivatives. Experimental and predicted values are given in
Table 4. Three most active compounds, 147, 153 and 154, have
vasodilatory activities with the IC50 values equal to 0.066, 0.084,
and 0.045 mM, respectively. As can be seen, the predicted values
match quite well the experimental data, with the exception of
compounds 147 and 152 which have errors more than 0.290
mM. Four out of 19 predicted values deviated from the
As a result, the use of a large diverse data set gives a better
predictive model for the test set compounds compared to using
the focused, chemotype-specific training sets. This brings
evidence that the proposed MFTA model can be reliably applied
to the prediction of vasodilatory activity of novel chemical
structures.
SYNTHESIS
■
Compounds 31−4120 and 42−586b were prepared according to
previously reported procedures, and their vasodilation activities
were determined by standard techniques.6a,c−e The targeted 6-
(2-thienyl)-3-pyridinecarbonitriles 147−165 were synthesized
according to the synthetic procedures shown in Schemes 1 and 2.
Table 4. Predicted Vasodilatory Activity IC50 (mM) for 17
Thienyl-Nicotinonitrile Derivatives
Scheme 1. Synthetic Route Towards 2-Bromo-3-
pyridinecarbonitriles
A mixture of equimolar amounts of 1,3-diaryl-2-propen-1-ones
139−142 with malononitrile in ethanol in the presence of
morpholine as a basic catalyst at room temperature afforded the
corresponding (1,3-diaryl-3-oxo-propyl)propanedinitriles 143−
146 in good yields (81−89%). The structure of 143−146 was
established through spectroscopic (IR, 1H NMR) and elemental
analysis. The IR spectra of 143, as a representative example of
this family, revealed the presence of a ketonic carbonyl function
at ν = 1653 cm−1 plus a nitrile stretching vibration band at ν =
2260 cm−1 confirming isolation of the open-chain Michael
1
adduct product. H NMR spectrum of 143 showed the propyl
methylene H2C-2 as a diastereotropic protons (two sets of
double of doublet signals at δH = 3.56, 3.80) coupled with each
other and in turn with the vicinal propyl methine HC-1
(multiplet at δH = 4.07−4.14). The latter coupled with the
1111
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
13C NMR spectrum of 153 exhibited the piperidinyl C-4, C-3/5,
and C-2/6 at δC = 24.5, 25.9, and 50.1, respectively. The 13C
NMR spectrum of 155 revealed the morpholinyl CH2N-3/5 and
CH2O-2/6 at δC = 49.3 and 66.4, respectively. The 13C NMR
spectrum of 159 revealed the piperazinyl C-2/6 and C-3/5 at δC =
48.7 and 54.8, respectively. For additional spectral data, see the
Experimental Section.
Scheme 2. Reactions of 2-Bromo-3-pyridinecarbonitriles with
Amines
Reaction of 150 with p-anisidine in refluxing pyridine afforded
the expected 4,6-bis(2-thienyl)-2-[(4-methoxyphenyl)amino]-3-
pyridinecarbonitrile (164). However, reaction of 150 with p-
toluidine under the same conditions gave 2-amino-4,6-bis(2-
thienyl)-3-pyridinecarbonitrile (165). Although formation of the
latter product is unexpected however, similar observations
describing this behavior were reported6a−c,34−36 (Scheme 2).
VASODILATION PROPERTIES
■
Vasodilation properties of 3-pyridinecarbonitriles 147−165 were
investigated in vitro using isolated thoracic aortic rings of rats
precontracted with norepinephrine hydrochloride standard
reported procedure6a,c−e,21 and compared with prazosin hydro-
chloride (highly selective α1-adrenoceptor antagonist), which
was used as a reference standard. The observed data (Table 4 and
Figures 1 and 2 of Supporting Information) reveal that all the
synthesized compounds exhibited significant vasodilation
properties compatible with the used standard reference.
Additionally, compounds 147, 153, and 154 exhibit remarkable
activity (IC50, concentration necessary for 50% reduction of
maximal norepinephrine hydrochloride induced contracture =
0.066, 0.084, and 0.045 mM, respectively), compared to prazosin
hydrochloride, the used reference standard in the present study
revealed a potency, IC50 = 0.487 mM.37
It was noticed that in most cases (compound 159 is an
exception) substituting the pyridinecarbonitriles position-4 with
a 4-(chlorophenyl) function enhanced the pharmacological
properties relative to the 4-(fluorophenyl) analogue. In most
cases (compound 152 is an exception) substitution at position-2
of the pyridinecarbonitrile nucleus with a piperidinyl function,
enhanced the vasodilation activity compared to a morpholinyl
residue.
vicinal propanedinitrile methine proton that appeared as a
doublet at δH = 5.26 (Scheme 1).
Bromination of 143−146 in glacial acetic acid at 60−70 °C
afforded ethyl 2-bromo-4,6-diaryl-3-pyridinecarbonitriles 147−
150 in good yields (70−78%). The structures of 147−150 were
1
inferred from spectroscopic (IR, H NMR, 13C NMR) and
elemental analysis data. The IR spectra of 147, as a representative
example of this family, lacked any band assignable to a carbonyl
function. However, a strong nitrile stretching vibration band was
observed (ν = 2220 cm−1). 1H NMR spectra of 147 showed the
characteristic pyridinyl H-5 as a singlet signal at δH = 8.20. 13C
NMR spectrum of 147 showed the pyridinyl quaternary C-3 and
methine HC-5 at δC = 109.2, 116.9, respectively. Additionally, the
nitrile carbon was observed at δC = 118.4 (Scheme 1).
Reaction of 2-bromo-3-pyridinecarbonitriles 147−150 with
secondary amines (piperidine, morpholine, 1-methylpiperazine,
and 1-ethylpiperazine) in refluxing tetrahydrofuran led to
aromatic nucleophilic substitution affording 2-(alicyclic-
amino)-3-pyridinecarbonitriles 151−163 in good yields (79−
97%), whose structures were deduced through spectroscopic
(IR, 1H NMR, 13C NMR) and elemental analysis data (Scheme
2). The 1H NMR spectrum of 151, as a representative example of
this family, revealed the piperidinyl function as a multiplet signal
at δH = 1.70−1.78 (piperidinyl H2C-3, H2C-4 and H2C-5) beside
a triplet signal at δH = 3.78 (piperidinyl H2C-2 and H2C-6). The
morpholinyl function of 155 appeared in its 1H NMR spectrum
as two triplet signals at δH = 3.80, 3.90 assigned to morpholinyl
CH2N-3/5 and CH2O-2/6, respectively. However, the 1-(4-
methylpiperazinyl) residue of 159 was observed in its 1H NMR
spectrum as two triplet signals at δH = 2.69 and 3.88 beside a
singlet signal at δH = 2.43 due to the piperazinyl methyl group.
EXPERIMENTAL SECTION
■
General Information Concerning Synthetic Methods.
Melting points were determined on an Electrothermal Stuart
SMP3 melting point apparatus. IR spectra (KBr) were recorded
1
on a Shimadzu FT-IR 8400S spectrophotometer. H NMR
spectra were recorded on a Varian MERCURY 300 (1H: 300
MHz) spectrometer. 13C NMR spectra were recorded on a JEOL
AS 500 (13C: 125 MHz) spectrometer. The starting compounds
139−14240,38,39 were prepared according to previously reported
procedures.
Synthesis of (1,3-Diaryl-3-oxo-propyl)propanedinitriles
143−146 (General Procedure). A mixture of equimolar
amounts of appropriate 139−142 and malononitrile (10
mmol) in absolute ethanol (20 mL) containing morpholine
(3−5 drops) was stirred at room temperature (20−25 °C) for 6
h. The solid which separated after storing the reaction mixture
overnight at room temperature was collected and crystallized
from a suitable solvent to afford 143−146.
[3-(4-Chlorophenyl-1-(2-thienyl)-3-oxo-propyl)-
propanedinitrile (143). Colorless microcrystals from n-butanol.
Mp 152−154 °C (1.40 g, 89% yield). IR: ν 2260, 1653, 1516,
1491. 1H NMR (DMSO-d6): δ 3.56 (dd, J = 17.6, 5.6 Hz, 1H),
1112
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
3.80 (dd, J = 17.4, 9.0 Hz, 1H), 4.07−4.14 (m, 1H), 5.26 (d, J =
6.6 Hz, 1H), 7.26 (dd, J = 5.1, 3.9 Hz, 1H), 7.46 (d, J = 9.0 Hz,
2H), 7.51 (d, J = 8.7 Hz, 2H), 8.02 (dd, J = 5.0, 1.1 Hz, 1H), 8.08
(dd, J = 3.9, 1.2 Hz, 1H). Anal. calcd for C16H11ClN2OS: C,
61.05; H, 3.52; N, 8.90. Found: C, 61.18; H, 3.59; N, 9.13.
[3-(4-Fluorophenyl-3-oxo-1-(2-thienyl)-propyl)-
propanedinitrile (144). Colorless microcrystals from methanol.
Mp 160−162 °C (1.20 g, 81% yield). IR: ν 2261, 1651, 1605,
1510. 1H NMR (DMSO-d6): δ 3.56 (dd, J = 17.4, 5.7 Hz, 1H),
3.79 (dd, J = 17.4, 8.7 Hz, 1H), 4.07−4.14 (m, 1H), 5.25 (d, J =
6.3 Hz, 1H), 7.19−7.27 (m, 3H), 7.53 (dd, J = 8.6, 5.6 Hz, 2H),
8.02 (dd, J = 4.8, 0.9 Hz, 1H), 8.08 (dd, J = 3.9, 1.2 Hz, 1H). Anal.
Calcd for C16H11FN2OS: C, 64.42; H, 3.72; N, 9.39. Found: C,
64.58; H, 3.76; N, 9.48.
[3-(4-Methylphenyl-3-oxo-1-(2-thienyl)-propyl)-
propanedinitrile (145). Colorless microcrystals from n-butanol.
Mp 154−156 °C (1.30 g, 88% yield). IR: ν 2261, 1655, 1518,
1410. 1H NMR (CDCl3): δ 2.38 (s, 3H), 3.56 (dd, J = 18.0, 5.7
Hz, 1H), 3.64 (dd, J = 18.2, 8.6 Hz, 1H), 3.88−3.94 (m, 1H),
4.62 (d, J = 5.1 Hz, 1H), 7.17 (dd, J = 5.0, 3.8 Hz, 1H), 7.24 (d, J =
8.4 Hz, 2H), 7.34 (d, J = 8.4 Hz, 2H), 7.72 (dd, J = 5.0, 1.1 Hz,
1H), 7.79 (dd, J = 3.8, 1.1 Hz, 1H). Anal. Calcd for C17H14N2OS:
C, 69.36; H, 4.79; N, 9.52. Found: C, 69.41; H, 4.83; N, 9.61.
[1,3-Bis(2-thienyl)-3-oxo-propyl)propanedinitrile (146).
Colorless microcrystals from benzene−light petroleum (60−80
°C) as 2:1 v/v. Mp 123−125 °C39 (1.25 g, 87% yield).
7.17−7.27 (m, 2H), 7.57−7.63 (m, 2H), 7.72 (s, 1H), 7.79 (d, J =
3.9 Hz, 1H), 7.95 (d, J = 3.9 Hz, 1H). Anal. Calcd for
C14H7BrN2S2: C, 48.42; H, 2.03; N, 8.07. Found: C, 48.53; H,
2.01; N, 8.19.
Reaction of 147−150 with Secondary Amines (General
Procedure). A mixture of 147−150 (5 mmol) and the
corresponding secondary amine (10 mmol) in tetrahydrofuran
(20 mL) was heated under reflux for the specified times. The
reaction mixture was evaporated to dryness, and the residue was
triturated material with methanol (5 mL), collected, and
crystallized from a suitable solvent to afford 151−163.
4-(4-Chlorophenyl)-2-(1-piperidinyl)-6-(2-thienyl)-3-pyridi-
necarbonitrile (151). Reaction time 7 h. Yellow microcrystals
from n-butanol. Mp 152−154 °C (1.85 g, 97% yield). IR: ν 2201,
1593, 1566. 1H NMR (CDCl3): δ 1.68−1.80 (m, 6H), 3.78 (t, J =
4.8 Hz, 4H), 7.06 (s, 1H), 7.13 (dd, J = 4.8, 3.9 Hz, 1H), 7.45−
7.57 (m, 5H), 7.66 (d, J = 3.9 Hz, 1H). Anal. Calcd for
C21H18ClN3S: C, 66.39; H, 4.78; N, 11.06. Found: C, 66.47; H,
4.86; N, 11.18.
4-(4-Fluorophenyl)-2-(1-piperidinyl)-6-(2-thienyl)-3-pyridi-
necarbonitrile (152). Reaction time 6 h. Pale-yellow micro-
crystals from ethanol. Mp 125−127 °C (1.60 g, 88% yield). IR: ν
2208, 1605, 1572. 1H NMR (CDCl3): δ 1.64−1.82 (m, 6H), 3.77
(t, J = 5.1 Hz, 4H), 7.07 (s, 1H), 7.13 (dd, J = 5.1, 3.9 Hz, 1H),
7.20 (t, J = 8.6 Hz, 2H), 7.48 (d, J = 5.1 Hz, 1H), 7.58 (dd, J = 8.4,
5.4 Hz, 2H), 7.67 (dd, J = 3.6, 0.9 Hz, 1H). Anal. Calcd for
C21H18FN3S: C, 69.40; H, 4.99; N, 11.56. Found: C, 69.52; H,
5.04; N, 11.69.
4-(4-Methylphenyl)-2-(1-piperidinyl)-6-(2-thienyl)-3-pyridi-
necarbonitrile (153). Reaction time 7 h. Pale-yellow micro-
crystals from n-butanol. Mp 139−141 °C (1.55 g, 86% yield). IR:
ν 2201, 1566, 1533. 1H NMR (CDCl3): δ 1.70−1.81 (m, 6H),
2.44 (s, 3H), 3.76 (t, J = 5.1 Hz, 4H), 7.11 (s, 1H), 7.13 (dd, J =
5.1, 3.6 Hz, 1H), 7.32 (d, J = 7.8 Hz, 2H), 7.46 (dd, J = 5.0, 0.8
Hz, 1H), 7.50 (d, J = 8.1 Hz, 2H), 7.67 (dd, J = 3.6, 0.9 Hz, 1H).
13C NMR (DMSO-d6): δ 21.4, 24.5, 25.9, 50.1, 91.5, 109.9,
118.5, 128.4, 129.3, 129.8, 131.1, 134.5, 140.1, 144.2, 153.2,
157.1, 162.7. Anal. Calcd for C22H21N3S: C, 73.50; H, 5.89; N,
11.69. Found: C, 73.58; H, 5.91; N, 11.77.
4,6-Bis(2-thienyl)-2-(1-piperidinyl)-3-pyridinecarbonitrile
(154). Reaction time 7 h. Pale-yellow microcrystals from n-
butanol. Mp 106−108 °C (1.60 g, 91% yield). IR: ν 2207, 1574,
1539. 1H NMR (CDCl3): δ 1.60−1.90 (m, 6H), 3.75 (t, J = 5.0
Hz, 4H), 7.14 (dd, J = 5.0, 3.8 Hz, 1H), 7.20 (dd, J = 5.0, 3.8 Hz,
1H), 7.21 (s, 1H), 7.48 (dd, J = 5.0, 1.1 Hz, 1H), 7.52 (dd, J = 5.0,
1.1 Hz, 1H), 7.69 (dd, J = 3.8, 1.1 Hz, 1H), 7.78 (dd, J = 3.6, 1.2
Hz, 1H). 13C NMR (DMSO-d6): δ 24.5, 25.9, 50.2, 89.7, 108.9,
118.7, 128.6, 128.9, 129.3, 130.5, 130.6, 131.4, 138.0, 143.9,
148.7, 153.4, 163.2. Anal. Calcd for C19H17N3S2: C, 64.93; H,
4.88; N, 11.95. Found: C, 64.98; H, 4.93; N, 12.11.
4-(4-Chlorophenyl)-2-(4-morpholinyl)-6-(2-thienyl)-3-pyri-
dinecarbonitrile (155). Reaction time 5 h. Yellow microcrystals
from n-butanol. Mp 185−187 °C (1.75 g, 92% yield). IR: ν 2203,
1595, 1572. 1H NMR (CDCl3): δ 3.80 (t, J = 4.4 Hz, 4H), 3.90 (t,
J = 4.5 Hz, 4H), 7.13−7.16 (m, 2H), 7.48−7.55 (m, 5H), 7.68 (d,
J = 3.6 Hz, 1H). 13C NMR (DMSO-d6): δ 49.3, 66.4, 91.8, 110.7,
118.1, 128.8, 129.25, 129.3, 131.2, 131.4, 135.4, 135.9, 143.8,
153.4, 155.8, 162.1. Anal. Calcd for C20H16ClN3OS: C, 62.90; H,
4.22; N, 11.00. Found: C, 62.98; H, 4.20; N, 11.12.
4-(4-Fluorophenyl)-2-(4-morpholinyl)-6-(2-thienyl)-3-pyri-
dinecarbonitrile (156). Reaction time 5 h. Yellow microcrystals
from n-butanol. Mp 191−193 °C (1.65 g, 90% yield). IR: ν 2199,
1603, 1560. 1H NMR (CDCl3): δ 3.80 (t, J = 4.2 Hz, 4H), 3.91 (t,
Synthesis of 2-Bromo-4,6-diaryl-3-pyridinecarbonitriles
147−150 (General Procedure). To a solution of 143−146 (5
mmol) in glacial acetic acid (10 mL) at 60−70 °C, a solution of
bromine (5.5 mmol) in glacial acetic acid (5 mL) was added
dropwise with stirring over a period of 10 min. After complete
addition, the reaction was kept at the same temperature for 3 h
and stored at room temperature (20−25 °C) overnight. The
separated solid was collected, washed with water, and crystallized
from a suitable solvent to afford 147−150.
2-Bromo-4-(4-chlorophenyl)-6-(2-thienyl)-3-pyridinecar-
bonitrile (147). Pale-yellow microcrystals from n-butanol. Mp
1
222−224 °C (1.35 g, 72% yield). IR: ν 2220, 1576, 1566. H
NMR (DMSO-d6): δ 7.26 (t, J = 4.5 Hz, 1H), 7.69 (d, J = 8.7 Hz,
2H), 7.78 (d, J = 8.4 Hz, 2H), 7.90 (d, J = 4.8 Hz, 1H), 8.16 (d, J =
4.5 Hz, 1H), 8.20 (s, 1H). 13C NMR (DMSO-d6): δ 109.2, 116.9,
118.4, 129.4, 129.8, 130.6, 131.2, 133.2, 134.4, 136.0, 141.5,
144.7, 155.0, 155.4. Anal. Calcd for C16H8BrClN2S: C, 51.16; H,
2.15; N, 7.46. Found: C, 51.21; H, 2.19; N, 7.58.
2-Bromo-4-(4-fluorophenyl)-6-(2-thienyl)-3-pyridinecarbo-
nitrile (148). Yellow microcrystals from acetic acid. Mp 212−214
1
°C (1.40 g, 78% yield). IR: ν 2226, 1580, 1564. H NMR
(DMSO-d6): δ 7.26 (dd, J = 5.0, 3.8 Hz, 1H), 7.46 (t, J = 9.0 Hz,
2H), 7.82 (dd, J = 9.0, 5.4 Hz, 2H), 7.90 (dd, J = 5.1, 1.2 Hz, 1H),
8.16 (dd, J = 3.9, 1.2 Hz, 1H), 8.20 (s, 1H). Anal. Calcd for
C16H8BrFN2S: C, 53.50; H, 2.24; N, 7.80. Found: C, 53.62; H,
2.27; N, 7.78.
2-Bromo-4-(4-methylphenyl)-6-(2-thienyl)-3-pyridinecar-
bonitrile (149). Almost colorless microcrystals from acetic acid.
Mp 228−230 °C (1.25 g, 70% yield). IR: ν 2220, 1584, 1516; 1H
NMR (DMSO-d6): δ 2.42 (s, 3H), 7.25 (t, J = 5.1 Hz, 1H), 7.41
(d, J = 7.8 Hz, 2H), 7.64 (d, J = 8.1 Hz, 2H), 7.89 (dd, J = 5.0, 1.1
Hz, 1H), 8.15 (s, 1H), 8.16 (dd, J = 5.0, 1.4 Hz, 1H). Anal. Calcd
for C17H11BrN2S: C, 57.48; H, 3.12; N, 7.89. Found: C, 57.39; H,
3.18; N, 7.97.
2-Bromo-4,6-bis(2-thienyl)-3-pyridinecarbonitrile (150).
Colorless microcrystals from n-butanol. Mp 181−183 °C (1.30
1
g, 75% yield). IR: ν 2222, 1574, 1501. H NMR (CDCl3): δ
1113
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
J = 4.2 Hz, 4H), 7.15 (dd, J = 5.1, 4.2 Hz, 1H), 7.16 (s, 1H), 7.22
(t, J = 8.6 Hz, 2H), 7.50 (dd, J = 5.1, 0.6 Hz, 1H), 7.56−7.61 (m,
2H), 7.68 (dd, J = 3.6, 0.9 Hz, 1H). 13C NMR (DMSO-d6): δ
49.3, 66.4, 92.0, 110.8, 116.1, 116.3, 118.2, 128.8, 129.3, 131.4,
131.7, 131.8, 133.5, 143.8, 153.4, 156.0, 162.2, 162.6, 164.6. Anal.
Calcd for C20H16FN3OS: C, 65.74; H, 4.41; N, 11.50. Found: C,
65.79; H, 4.47; N, 11.62.
4-(4-Methylphenyl)-2-(4-morpholinyl)-6-(2-thienyl)-3-pyri-
dinecarbonitrile (157). Reaction time 6 h. Yellow microcrystals
from n-butanol. Mp 198−200 °C (1.65 g, 91% yield). IR: ν 2197,
1560, 1533. 1H NMR (CDCl3): δ 2.45 (s, 3H), 3.79 (t, J = 4.4 Hz,
4H), 3.91 (t, J = 4.5 Hz, 4H), 7.14 (t, J = 4.4 Hz, 1H), 7.19 (s,
1H), 7.33 (d, J = 7.8 Hz, 2H), 7.46−7.52 (m, 3H), 7.67 (d, J = 3.6
Hz, 1H). 13C NMR (DMSO-d6): δ 21.4, 49.4, 66.5, 92.1, 110.7,
118.3, 128.6, 129.3, 129.8, 131.3, 134.2, 140.2, 143.9, 153.3,
157.1, 162.4. Anal. Calcd for C21H19N3OS: C, 69.78; H, 5.30; N,
11.62. Found: C, 69.85; H, 5.34; N, 11.68.
4,6-Bis(2-thienyl)-2-(4-morpholinyl)-3-pyridinecarbonitrile
(158). Reaction time 7 h. Yellow microcrystals from n-butanol.
Mp 203−205 °C (1.55 g, 88% yield). IR: ν 2199, 1560, 1541. 1H
NMR (CDCl3): δ 3.66 (t, J = 4.5 Hz, 4H), 3.78 (t, J = 4.5 Hz,
4H), 7.22 (dd, J = 5.0, 3.8 Hz, 1H), 7.30 (dd, J = 5.0, 3.8 Hz, 1H),
7.60 (s, 1H), 7.78 (dd, J = 5.0, 1.1 Hz, 1H), 7.85 (dd, J = 3.6, 1.2
Hz, 1H), 7.91 (dd, J = 5.3, 1.4 Hz, 1H), 8.05 (dd, J = 3.8, 1.1 Hz,
1H). Anal. Calcd for C18H15N3OS2: C, 61.17; H, 4.28; N, 11.89.
Found: C, 61.30; H, 4.34; N, 12.03.
4-(4-Chlorophenyl)-2-[1-(4-methylpiperazinyl)]-6-(2-thien-
yl)-3-pyridinecarbonitrile (159). Reaction time 10 h. Pale-
yellow microcrystals from n-butanol. Mp 184−186 °C (1.60 g,
81% yield). IR: ν 2208, 1595, 1574. 1HNMR (CDCl3): δ 2.43 (s,
3H), 2.69 (t, J = 4.5 Hz, 4H), 3.88 (t, J = 5.0 Hz, 4H), 7.12 (s,
1H), 7.14 (t, J = 4.4 Hz, 1H), 7.40−7.55 (m, 5H), 7.67 (dd, J =
3.8, 1.1 Hz, 1H). 13CNMR (DMSO-d6): δ 46.2, 48.7, 54.8, 91.6,
110.3, 118.2, 128.7, 129.25, 129.3, 131.3, 131.4, 135.3, 136.0,
143.9, 153.4, 155.8, 162.1. Anal. Calcd for C21H19ClN4S: C,
63.87; H, 4.85; N, 14.19. Found: C, 64.02; H, 4.93; N, 14.27.
4-(4-Chlorophenyl)-2-[1-(4-ethylpiperazinyl)]-6-(2-thien-
yl)-3-pyridinecarbonitrile (160). Reaction time 12 h. Yellow
microcrystals from n-butanol. Mp 167−169 °C (1.75 g, 86%
yield). IR: ν 2212, 1597, 1574. 1H NMR (CDCl3): δ 1.22 (t, J =
7.2 Hz, 3H), 2.62 (q, J = 7.2 Hz, 2H), 2.62−2.82 (m, 4H), 3.91 (t,
J = 4.7 Hz, 4H), 7.12 (s, 1H), 7.14 (t, J = 4.8 Hz, 1H), 7.48−7.55
(m, 5H), 7.67 (d, J = 3.9 Hz, 1H). 13C NMR (DMSO-d6): δ 12.4,
48.8, 52.1, 52.6, 91.5, 110.3, 118.2, 128.8, 129.3, 129.4, 131.3,
131.4, 135.3, 136.1, 143.9, 153.4, 155.9, 162.0. Anal. Calcd for
C22H21ClN4S: C, 64.61; H, 5.18; N, 13.70. Found: C, 64.73; H,
5.16; N, 13.79.
(DMSO-d6): δ 21.4, 46.2, 48.8, 54.9, 91.8, 110.4, 118.4, 128.4,
128.6, 128.7, 128.9, 129.3, 129.8, 131.2, 134.3, 140.2, 144.0,
153.2, 157.1, 162.4. Anal. Calcd for C22H22N4S: C, 70.56; H,
5.92; N, 14.96. Found: C, 70.71; H, 5.96; N, 14.88.
4,6-Bis(2-thienyl)-2-[1-(4-methylpiperazinyl)]-3-pyridine-
carbonitrile (163). Reaction time 12 h. Yellow microcrystals
from methanol. Mp 128−130 °C (1.50 g, 82% yield). IR: ν 2205,
1574, 1543. 1H NMR (CDCl3): δ 2.4 (s, 3H), 2.65 (t, J = 5.0 Hz,
4H), 3.83 (t, J = 5.0 Hz, 4H), 7.14 (dd, J = 5.1, 3.6 Hz, 1H), 7.20
(dd, J = 5.1, 3.6 Hz, 1H), 7.25 (s, 1H), 7.48 (dd, J = 5.1, 1.2 Hz,
1H), 7.52 (dd, J = 5.1, 1.2 Hz, 1H), 7.68 (dd, J = 3.8, 1.1 Hz, 1H),
7.77 (dd, J = 3.8, 1.1 Hz, 1H). 13C NMR (DMSO-d6): δ 46.2,
48.9, 54.9, 90.0, 109.4, 118.5, 128.7, 128.9, 129.3, 130.5, 130.6,
131.4, 137.9, 143.7, 148.7, 153.4, 162.9. Anal. Calcd for
C19H18N4S2: C, 62.27; H, 4.95; N, 15.29. Found: C, 62.34; H,
4.98; N, 15.37.
Reaction of 150 with Primary Amines (General Procedure).
A mixture of 150 (5 mmol) and the corresponding primary
amine (5.5 mmol) in pyridine (20 mL) was heated under reflux
for the appropriate time. The solid that separated on pouring the
reaction mixture into ice-cold water (200 mL) and acidification
with dil. HCl (5%) was collected, washed with water, and purified
by silica gel TLC (F254) affording 164 and 165.
4,6-Bis(2-thienyl)-2-[(4-methoxyphenyl)amino]-3-pyridine-
carbonitrile (164). Obtained via reaction of 150 and p-anisidine.
Reaction time 15 h. Yellow microcrystals purified by silica gel
TLC (F254) using methylene chloride−light petroleum (60−80
°C) as 1:2 v/v for elution. Mp 197−199 °C (1.10 g, 57% yield).
IR: ν 3325, 2210, 1601, 1578. 1H NMR (CDCl3): δ 3.86 (s, 3H),
6.97 (d, J = 9.3 Hz, 2H), 7.15 (t, J = 4.5 Hz, 1H), 7.22 (d, J = 5.1
Hz, 1H), 7.24 (s, 1H), 7.50 (d, J = 5.1 Hz, 1H), 7.56 (d, J = 5.1
Hz, 1H), 7.61 (d, J = 8.7 Hz, 2H), 7.73 (br s, 1H), 7.86 (d, J = 4.5
Hz, 1H). 13C NMR (DMSO-d6): δ 55.7, 87.0, 108.3, 114.0,
117.6, 123.8, 128.5, 128.9, 129.3, 130.2, 130.6, 131.5, 133.0,
137.9, 144.0, 147.6, 153.8, 155.8, 157.6. Anal. Calcd for
C21H15N3OS2: C, 64.76; H, 3.88; N, 10.79. Found: C, 64.88;
H, 3.94; N, 10.92.
2-Amino-4,6-bis(2-thienyl)-3-pyridinecarbonitrile (165).
Obtained via reaction of 150 and p-toluidine. Reaction time 18
h. Pale-yellow microcrystals purified by silica gel TLC (F254
)
using methylene chloride−light petroleum (60−80 °C) as 1:2 v/
v for elution. Mp 177−179 °C (0.75 g, 53% yield). IR: ν 3478,
3370, 2210, 1626, 1572. 1H NMR (DMSO-d6): δ 6.99 (br s, 2H),
7.19 (dd, J = 5.3, 3.8 Hz, 1H), 7.28 (dd, J = 5.1, 3.6 Hz, 1H), 7.36
(s, 1H), 7.74 (dd, J = 5.0, 1.1 Hz, 1H), 7.84 (dd, J = 3.8, 1.1 Hz,
1H), 7.87 (dd, J = 5.1, 1.2 Hz, 1H), 7.98 (dd, J = 3.8, 1.1 Hz, 1H).
13C NMR (DMSO-d6): δ 84.5, 106.8, 117.9, 128.4, 128.9, 129.2,
129.8, 130.2, 131.0, 138.1, 143.8, 146.8, 154.8, 161.7. Anal. Calcd
for C14H9N3S2: C, 59.34; H, 3.20; N, 14.83. Found: C, 59.51; H,
3.18; N, 15.04.
Vasodilation Activity Screening. The vasodilation activity
screening procedure was carried out according to the standard
reported in vitro bioassay technique6a,c−e,21 by testing the effects
of the synthesized compounds on isolated thoracic aortic rings of
male Wistar rats (250−350 g) precontracted with norepinephr-
ine hydrochloride. After light ether anesthesia, the rats were
sacrificed by cervical dislocation. The aortae were immediately
excised, freed of extraneous tissues, and prepared for isometric
tension recording. Aorta was cut into (3−5 mm width) rings and
each ring was placed in a vertical chamber “10 mL jacketed
automatic multi-chamber organ bath system (Model no.
ML870B6/C, Panlab, Spain)” filled with Krebs solution
composed of (in mM): NaCl, 118.0; KCl, 4.7; NaHCO3, 25.0;
4-(4-Fluorophenyl)-2-[1-(4-methylpiperazinyl)]-6-(2-thien-
yl)-3-pyridinecarbonitrile (161). Reaction time 10 h. Yellow
microcrystals from ethanol. Mp 158−159 °C (1.50 g, 79% yield).
IR: ν 2210, 1574, 1545. 1H NMR (CDCl3): δ 2.41 (s, 3H), 2.67
(t, J = 5.0 Hz, 4H), 3.87 (t, J = 5.1 Hz, 4H), 7.12 (s, 1H), 7.14 (dd,
J = 5.1, 3.9 Hz, 1H), 7.18−7.24 (m, 2H), 7.49 (dd, J = 5.1, 1.2 Hz,
1H), 7.55−7.61 (m, 2H), 7.67 (dd, J = 3.8, 1.1 Hz, 1H). Anal.
Calcd for C21H19FN4S: C, 66.64; H, 5.06; N, 14.80. Found: C,
66.79; H, 5.13; N, 15.02.
4-(4-Methylphenyl)-2-[1-(4-methylpiperazinyl)]-6-(2-thien-
yl)-3-pyridinecarbonitrile (162). Reaction time 10 h. Pale-
yellow microcrystals from methanol. Mp 151−153 °C (1.60 g,
86% yield). IR: ν 2212, 1575, 1535. 1H NMR (CDCl3): δ 2.43 (s,
3H), 2.44 (s, 3H), 2.70 (t, J = 4.8 Hz, 4H), 3.87 (t, J = 4.8 Hz,
4H), 7.13 (t, J = 4.4 Hz, 1H), 7.16 (s, 1H), 7.33 (d, J = 8.1 Hz,
2H), 7.47−7.51 (m, 3H), 7.66 (d, J = 3.9 Hz, 1H). 13C NMR
1114
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
(2) Burt, V. L.; Cutler, J. A.; Higgins, M.; Horan, M. J.; Labarthe, D.;
Whelton, P.; Brown, C.; Roccella, E. J. Trends in the Prevalence,
Awareness, Treatment, and Control of Hypertension in the Adult US
Population: Data from the Health Examination Surveys, 1960 to 1991.
Hypertension 1995, 26, 60−69.
CaCl2, 1.8; NaH2PO4, 1.2; MgSO4, 1.2; glucose, 11.0 and
oxygenated with carbogen gas (95% O2/5% CO2) at 37 0.5 °C.
Each aortic ring was mounted between two stainless steel hooks
passed through its lumen. The lower hook was fixed between two
plates, while the upper one was attached to a force displacement
transducer (Model no. MLT0201, Panlab, Spain) connected to
an amplifier (PowerLab, AD Instruments Pty. Ltd.), which was
connected to a computer. The Chart for windows (v 3.4)
software was used to record and elaborate data.
(3) Egan, B. M.; Zhao, Y.; Axon, R. N. US Trends in Prevalence,
Awareness, Treatment, and Control of Hypertension, 1988−2008.
JAMA, J. Am. Med. Assoc. 2010, 303, 2043−2050.
(4) McInnes, G. T. Clinical Pharmacology and Therapeutics of
Hypertension; Elsevier: Amsterdam, The Netherlands, 2008.
Preparations were stabilized under 2 g resting tension during 2
h, and then the contracture response to norepinephrine
hydrochloride (10−6 M) was measured before and after exposure
to increasing concentrations of the tested synthesized com-
pounds. The tested compounds were dissolved in dimethylsulf-
oxide (DMSO) as stock solution (10 mL of 0.005 M). Control
experiments were performed in the presence of DMSO alone, at
the same concentrations as those used with the derivatives tested,
which demonstrated that the solvent did not affect the contractile
response of isolated aorta. The observed vasodilation activity
screening data for the newly synthesized compounds are
expressed as IC50, concentration necessary for 50% reduction
of maximal norepinephrine hydrochloride induced contracture.
(5) Klabunde, R. E. Cardiovascular Physiology Concepts; Lippincott
Williams & Wilkins: Philadelphia, PA, 2004.
(6) (a) Girgis, A. S.; Kalmouch, A.; Ellithey, M. Synthesis of Novel
Vasodilatory Active Nicotinate Esters with Amino Acid Function.
Bioorg. Med. Chem. 2006, 14, 8488−8494. (b) Girgis, A. S.; Hosni, H.
M.; Barsoum, F. F. Novel Synthesis of Nicotinamide Derivatieves of
Cytotoxic Properties. Bioorg. Med. Chem. 2006, 14, 4466−4476.
(c) Girgis, A. S.; Mishriky, N.; Farag, A. M.; El-Eraky, W. I.; Farag, H.
Synthesis of New 3-Pyridinecarboxylates of Potential Vasodilation
Properties. Eur. J. Med. Chem. 2008, 43, 1818−1827. (d) Girgis, A. S.;
Ismail, N. S. M.; Farag, H. Facile Synthesis, Vasorelaxant Properties and
Molecular Modeling Studies of 2-Amino-8a-methoxy-4H-pyrano[3,2-
c]pyridine-3-carbonitriles. Eur. J. Med. Chem. 2011, 46, 2397−2407.
(e) Girgis, A. S.; Ismail, N. S. M.; Farag, H.; El-Eraky, W. I.; Saleh, D. O.;
Tala, S. R.; Katritzky, A. R. Regioselective Synthesis and Molecular
Modelling Study of Vasorelaxant Active 7,9-Dioxa-1,2-diaza-spiro[4,5]-
dec-2-ene-6,10-diones. Eur. J. Med. Chem. 2010, 45, 4229−4238.
(7) Badawneh, M.; Ferrarini, P. L.; Calderone, V.; Manera, C.;
Martinotti, E.; Mori, C.; Saccomanni, G.; Testai, L. Synthesis and
Evaluation of Antihypertensive Activity of 1,8-Naphthyridine Deriva-
tives. Part X. Eur. J. Med. Chem. 2001, 36, 925−934.
CONCLUSIONS
■
A highly representative and diverse data set of seven vasodilatory
chemical classes analyzed with molecular field topology analysis
resulted in a statistically significant relationship between the
structure and the title activity. Based on this model, predictions
for 19 newly synthesized compounds were made, which were
subsequently biotested and demonstrated high vasodilatory
potency. This QSAR model was also used for topological
functional core identification. Such a functional core can be a
valuable tool in the design of antihypertensive drugs with more
than one mechanism of action.
(8) Quezada, E.; Delogu, G.; Picciau, C.; Santana, L.; Podda, G.;
Borges, F.; García-Morales, V.; Vina, D.; Orallo, F. Synthesis and
̃
Vasorelaxant and Platelet Antiaggregatory Activities of a New Series of
6-Halo-3-phenylcoumarins. Molecules 2010, 15, 270−279.
(9) Bansal, R.; Kumar, D.; Carron, R.; de la Calle, C. Synthesis and
Vasodilatory Activity of Some Amide Derivatives of 6-(4-Carboxyme-
thyloxyphenyl)-4,5-dihydro-3(2H)-pyridazinone. Eur. J. Med. Chem.
2009, 44, 4441−4447.
ASSOCIATED CONTENT
* Supporting Information
(10) Silva, A. G.; Zapata-Sudo, G.; Kummerle, A. E.; Fraga, C. A. M.;
Barreiro, E. J.; Sudo, R. T. Synthesis and Vasodilatory Activity of New N-
Acetylhydrazone Derivatives, Designed as LASSBio-294 Analogues.
Bioorg. Med. Chem. 2005, 13, 3431−3437.
(11) Campos-Toimil, M.; Orallo, F.; Santana, L.; Uriarte, E. Synthesis
and Vasorelaxant Activity of New Coumarin and Furocoumarin
Derivatives. Bioorg. Med. Chem. Lett. 2002, 12, 783−786.
■
S
Charts represent the effect of synthesized compounds 147−165
on contracture induced by norepinephrine hydrochloride (NE·
HCl) in rat thoracic aortic rings and the potency. This material is
available free of charge via the Internet at http://pubs.acs.org
(12) Chegaev, K.; Lazzarato, L.; Marcarino, P.; Di Stilo, A.; Fruttero,
R.; Vanthuyne, N.; Roussel, C.; Gasco, A. Synthesis of Some Novel
Organic Nitrates and Comparative in vitro Study of their Vasodilator
Profile. J. Med. Chem. 2009, 52, 4020−4025.
(13) Vilar, S.; Quezada, E.; Alcaide, C.; Orallo, F.; Santana, L.; Uriarte,
E. Quantitative Structure Vasodilatory Activity Relationship − Synthesis
and “in silico” and “in vitro” Evaluation of Resveratrol-Coumarin
Hybrids. QSAR Comb. Sci. 2007, 26, 317−332.
(14) Abouzid, K.; Hakeem, M. A.; Khalil, O.; Maklad, Y. Pyridazinone
Derivatives: Design, Synthesis, and in vitro Vasorelaxant Activity. Bioorg.
Med. Chem. 2008, 16, 382−389.
(15) Dong, X.; Wang, Y.; Liu, T.; Wu, P.; Gao, J.; Xu, J.; Yang, B.; Hu,
Y. Flavonoids as Vasorelaxant Agents: Synthesis, Biological Evaluation
and Quantitative Structure Activities Relationship (QSAR) Studies.
Molecules 2011, 16, 8257−8272.
(16) (a) Palyulin, V. A.; Radchenko, E. V.; Zefirov, N. S. Molecular
Field Topology Analysis Method in QSAR Studies of Organic
Compounds. J. Chem. Inf. Comput. Sci. 2000, 40, 659−667. (b) Palyulin,
V. A.; Zefirov, N. S. Generation of Molecular Graphs for QSAR Studies.
Dokl. Chem. 2005, 402, 81−85.
(17) (a) Zhan, W.; Liang, Z.; Zhu, A.; Kurtkaya, S.; Shim, H.; Snyder, J.
P.; Liotta, D. C. Discovery of Small Molecule CXCR4 Antagonists. J.
Med. Chem. 2007, 50, 5655−5664. (b) Berhanu, W. M.; Pillai, G. G.;
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: charlesdennishall@gmail.com.
Notes
The authors declare no competing financial interest.
○Deceased
ACKNOWLEDGMENTS
■
This study was supported financially by the Science and
Technology Development Fund (STDF), Egypt, grant no. 1357.
ABBREVIATIONS
■
MFTA, molecular field topology analysis; MSG, molecular super
graph
REFERENCES
■
(1) Kearney, P. M.; Whelton, M.; Reynolds, K.; Muntner, P.; Whelton,
P. K.; He, J. Global Burden of Hypertension: Analysis of Worldwide
Data. Lancet 2005, 365, 217−223.
1115
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116

Journal of Chemical Information and Modeling
Article
Oliferenko, A. A.; Katrtitzky, A. R. Quantitative Structure−Activity/
Property Relationships: The Ubiquitous Links Between Cause and
Effect. ChemPlusChem 2012, 77, 507−517.
(18) Radchenko, E. V.; Makhaeva, G. F.; Malygin, V. V.; Sokolov, V. B.;
Palyulin, V. A.; Zefirov, N. S. Modelling of the Relationships Between
the Structure of o-Phosphorylated Oximes and their Anticholinesterase
Activity and Selectivity Using Molecular Field Topology Analysis
(MFTA). Dokl. Biochem. Biophys. 2008, 418, 47−51.
(19) Chupakhin, V. I.; Bobrov, S. V.; Radchenko, E. V.; Palyulin, V. A.;
Zefirov, N. S. Computer-Aided Ddesign of Selective Ligands of the
Benzodiazepine-Binding Site of the GABAA Receptor. Dokl. Chem.
2008, 422, 227−230.
(20) Girgis, A. S.; Farag, H.; Ismail, N. S. M.; George, R. F. Synthesis,
Hypnotic Properties and Molecular Modelling Studies of 1,2,7,9-
Tetraaza-spiro[4.5]dec-2-ene-6,8,10-triones. Eur. J. Med. Chem. 2011,
46, 4964−4969.
(34) Girgis, A. S.; Hosni, H. M.; Barsoum, F. F.; Amer, A. M. M.; Farag,
I. S. A. Novel Nicotinate Esters of Vasodilatation Activity. Boll. Chim.
Farm. 2004, 143, 365−375.
(35) Girgis, A. S.; Kalmouch, A.; Hosni, H. M. Synthesis of Novel 3-
Pyridinecarbonitriles with Amino Acid Function and their Fluorescence
Properties. Amino Acids 2004, 26, 139−146.
(36) Girgis, A. S.; Tala, S. R.; Oliferenko, P. V.; Oliferenko, A. A.;
Katritzky, A. R. Computer-Assisted Rational Design, Synthesis, and
Bioassay of Nonsteroidal Anti-inflammatory Agents. Eur. J. Med. Chem.
2012, 50, 1−8.
(37) Abou-Seri, S. M.; Abouzid, K.; Abou El Ella, D. A. Molecular
Modeling Study and Synthesis of Quinazolinone-arylpiperazine
Derivatives as α1-Adrenoreceptor Antagonists. Eur. J. Med. Chem.
2011, 46, 647−658.
(38) Ramesh, B.; Rao, B. S. Synthesis, Spectral Studies and Anti-
inflammatory Activity of 2-Acetyl Thiophene. E-J. Chem. 2010, 7, 433−
436.
(21) Nofal, Z. M.; Srour, A. M.; El-Eraky, W. I.; Saleh, D. O.; Girgis, A.
S. Rational Design, Synthesis and QSAR Study of Vasorelaxant Active 3-
Pyridinecarbonitriles Incorporating 1H-Benzimidazol-2-yl Function.
Eur. J. Med. Chem. 2013, 63, 14−21.
̈
(39) Turan-Zitouni, G.; Ozdemir, A.; Guven, K. Synthesis of Some 1-
̈
[(N, N-Disubstituted thiocar bamoylthio)acetyl]-3-(2-thienyl)-5-aryl-2-
pyrazoline Derivatives and Investigation of their Antibacterial and
Antifungal Activities. Arch. Pharm. (Weinheim, Ger.) 2005, 338, 96−104.
(40) Greiner-Bechert, L.; Otto, H. H. 1,4-Pentandien-3-ones, XXXII:
Reaction of 2-Acetylthiophene and 2-Acetylfuran with Malononitrile
and Aldehydes, and Synthesis and Properties of Phenylene-bis[(thienyl/
furyl)nicotinonitrile] Derivative. Arch. Pharm. (Weinheim, Ger.) 1991,
324, 563−572.
́
(22) Vilar, S.; Quezada, E.; Santana, L.; Uriarte, E.; Yanez, M.; Fraiz, N.;
Alcaide, C.; Cano, E.; Orallo, F. Design, Synthesis, and Vasorelaxant and
Platelet Antiaggregatory Activities of Coumarin-Resveratrol Hybrids.
Bioorg. Med. Chem. Lett. 2006, 16, 257−261.
(23) Burns, D. M.; Ruddock, M. W.; Walker, M. D.; Allen, J. M.;
Kennovin, G. D.; Hirst, D. G. Nicotinamide-Inhibited Vasoconstriction:
Lack of Dependence on Agonist Signalling Pathways. Eur. J. Pharmacol.
1999, 374, 213−220.
(24) Eguchi, S.; Miyashita, S.; Kitamura, Y.; Kawasaki, H. α3β4-
Nicotinic Receptors Mediate Adrenergic Nerve- and Peptidergic
(CGRP) Nerve-Dependent Vasodilation Induced by Nicotine in Rat
Mesentergic Arteries. Br. J. Pharmacol. 2007, 151, 1216−1223.
(25) Si, M.; Lee, T. J. F. Presynaptic α7-Nicotinic Acetylcholine
Receptors Mediate Nicotine-Induced Nitric Oxidergic Neurogenic
Vasodilation in Porcine Basilar Arteries. J. Pharmacol. Exp. Ther. 2001,
298, 122−128.
(26) Toba, H.; Nakagawa, Y.; Miki, S.; Shimizu, T.; Yoshimura, A.;
Inoue, R.; Asayama, J.; Kobara, M.; Nakata, T. Calcium Channel
Blockades Exhibit Anti-inflammatory and Antioxidative Effects by
Augmentation of Endothelial Nitric Oxide Synthase and the Inhibition
of Angiotensin Converting Enzyme in the NG-Nitro-L-arginine methyl
ester - Induced Hypertensive Rat Aorta: Vasoprotective Effects Beyond
the Blood Pressure-Lowering Effects of Amlodipine and Manidipine.
Hypertens. Res. 2005, 28, 689−700.
(27) Clark, R. D.; Caroon, J. M.; Repke, D. B.; Strosberg, A. M.; Bitter,
S. M.; Okada, M. D.; Michel, A. D.; Whiting, R. L. Antihypertensive 9-
Substituted 1-Oxa-4,9-diazaspiro[5.5]undecan-3-ones. J. Med. Chem.
1983, 26, 855−861.
(28) Ferrarini, P. L.; Mori, C.; Calderone, V.; Calzolari, L.; Nieri, P.;
Saccomanni, G.; Martinotti, E. Synthesis of 1,8-Naphthyridine
Derivatives: Potential Antihypertensive Agents − Part VIII. Eur. J.
Med. Chem. 1999, 34, 505−513.
(29) Silva, C. L. M.; Noel, F.; Barreiro, E. J. Cyclic GMP-Dependent
̈
Vasodilatory Properties of LASSBio-294 in Rat Aorta. Br. J. Pharmacol.
2002, 135, 293−298.
(30) Franke, R.; Huebel, S.; Streich, W. J. Substructural QSAR
Approaches and Topological Pharmacophores. Environ. Health Persp.
1985, 61, 239−255.
(31) Schneider, G.; Neidhart, W.; Giller, T.; Schmid, G. Scaffold-
Hopping” by Topological Pharmacophore Search: a Contribution to
Virtual Screening. Angew. Chem., Int. Ed. 1999, 38, 2894−2896.
(32) Bonachera, F.; Parent, B.; Barbosa, F.; Froloff, N.; Horvath, D.
Fuzzy Tricentric Pharmacophore Fingerprints. 1. Topological Fuzzy
Pharmacophore Triplets and Adapted Molecular Similarity Scoring
Schemes. J. Chem. Inf. Model. 2006, 46, 2457−2477.
(33) Horvath, D. Topological pharmacophores, Ch. 2 in Chemo-
informatics Approaches to Virtual Screening; Varnek, A., Tropsha, A., Eds.;
RSC: Cambridge, U.K., 2008.
1116
dx.doi.org/10.1021/ci400723m | J. Chem. Inf. Model. 2014, 54, 1103−1116