Using entropy of drug and protein graphs to predict FDA drug-target network: Theoretic-experimental study of MAO inhibitors and hemoglobin peptides from Fasciola hepatica

European Journal of Medicinal Chemistry 46 (2011) 1074 e 1094

Contents lists available at ScienceDirect

European Journal of Medicinal Chemistry

journal homepage: http://www.elsevier.com/locate/ejmech

Original article

Using entropy of drug and protein graphs to predict FDA drug-target network:

Theoretic-experimental study of MAO inhibitors and hemoglobin peptides from

Fasciola hepatica

a

,

a

*

Francisco Prado-Prado , Xerardo García-Mera , Paula Abeijón , Nerea Alonso , Olga Caamaño ,

b

c

d

e

Matilde Yáñez , Teresa Gárate , Mercedes Mezo , Marta González-Warleta , Laura Muiño ,

Florencio M. Ubeira , Humberto González-Díaz

Department of Organic Chemistry, Faculty of Pharmacy, USC 15782, Spain

Department of Pharmacology, USC 15782, Spain

Parasitology Service, National Center of Microbiology, Instituto de Salud Carlos III, Majadahonda, Madrid 28220, Spain

Centro de Investigaciones Agrarias (INGACAL), Mabegondo, A Coruña 15318, Spain

Department of Microbiology and Parasitology, Faculty of Pharmacy, USC 15782, Spain

e

a

b

c

d

e

a r t i c l e i n f o

a b s t r a c t

Article history:

Received 13 October 2010

Received in revised form

There are many drugs described with very different afﬁnity to a large number of receptors. In this work,

we selected Drug-Target pairs (DTPs/nDTPs) of drugs with high afﬁnity/non-afﬁnity for different targets

like proteins. Quantitative StructureeActivity Relationships (QSAR) models become a very useful tool in

this context to substantially reduce time and resources consuming experiments. Unfortunately, most

QSAR models predict activity against only one protein. To solve this problem, we developed here a multi-

target QSAR (mt-QSAR) classiﬁer using the MARCH-INSIDE technique to calculate structural parameters

of drug and target plus one Artiﬁcial Neuronal Network (ANN) to seek the model. The best ANN model

found is a Multi-Layer Perceptron (MLP) with proﬁle MLP 32:32e15e1:1. This MLP classiﬁes correctly

10 January 2011

Accepted 13 January 2011

Available online 21 January 2011

Keywords:

DrugeProtein interaction complex

networks

Protein structure networks

Multi-target QSAR

Markov model

Rasagiline derivatives

MAO enzymes

6

23 out of 678 DTPs (Sensitivity ¼ 91.89%) and 2995 out of 3234 nDTPs (Speciﬁcity ¼ 92.61%), corre-

sponding to training Accuracy ¼ 92.48%. The validation of the model was carried out by means of

external predicting series. The model classiﬁes correctly 313 out of 338 DTPs (Sensitivity ¼ 92.60%) and

1411 out of 1534 nDTP (Speciﬁcity

¼

91.98%) in validation series, corresponding to total

Accuracy ¼ 92.09% for validation series (Predictability). This model favorably compares with other LDA

and ANN models developed in this work and Machine Learning classiﬁers published before to address

the same problem in different aspects. These mt-QSARs offer also a good opportunity to construct drug

eprotein Complex Networks (CNs) that can be used to explore large and complex drugeprotein receptors

databases. Finally, we illustrated two practical uses of this model with two different experiments. In

experiment 1, we report prediction, synthesis, characterization, and MAO-A and MAO-B pharmacological

assay of 10 rasagiline derivatives promising for anti-Parkinson drug design. In experiment 2, we report

sampling, parasite culture, SEC and 1DE sample preparation, MALDI-TOF MS and MS/MS analysis,

MASCOT search, MM/MD 3D structure modeling, and QSAR prediction for different peptides of hemo-

globin found in the proteome of the human parasite Fasciola hepatica; which is promising for anti-

parasite drug targets discovery.

Fasciola hepatica proteome

ꢀ

1

. Introduction

bioinformatics and proteome research toward drug discovery.

Therefore, there is a strong incentive to develop new methods

capable of detecting these potential drug-target interactions efﬁ-

ciently [1]. In this sense, graphs and complex network theory may

play an important role at different stages of modeling process with

different degrees of organization of matter [2e9]. In a ﬁrst stage, we

can use molecular graphs to represent and calculate structural

parameters for drugs sometimes called Topological Indices (TIs) but

The fast and accurate prediction of interactions between drugs

and target proteins is a keystone piece on the combination of

*

Corresponding author. Fax: þ 34 981 594912.

E-mail address: fenol1@hotmail.com (F. Prado-Prado).

doi:10.1016/j.ejmech.2011.01.023

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1075

also estimate physicochemical parameters based on a graph

method, see our recent reviews [10]. At a higher level we can use

graphs to represent structure of the drug-target proteins and

calculate characteristic TIs and/or physicochemical parameters of

proteins structure or protein interactions networks, see for instance

the works of Giuliani [11e17], or our recent reviews [18,19]. Next,

we can develop a kind of computer program with network topology

called Artiﬁcial Neural Networks (ANNs) that after adequate

training learn predicts target proteins for a given drug. It means,

ANNs are network-like software that may use as inputs TIs and/or

physicochemical parameters calculated in the previous steps to

predict which network-like molecular structures present or not

a desire property, see for instance the works after Caballero and

Fernandez et al. [20e23] with applications to both drugs and

proteins or works of Zbilut et al. [24,25]. In particular, using the

parameters of the drug and the target ANNs may select Drug-Target

pairs of drugs with high afﬁnity (DTPs) out of those pairs for drugs

with none afﬁnity for different targets (nDTPs). In general, this

technique (using or not ANNs) lie within the kind of studies called

Quantitative StructureeActivity Relationships (QSAR) models and

may become a very useful tool in this context to substantially

reduce time and resources consuming experiments. In a last step,

the prediction of all possible DTPs/nDTPs in the global set of rela-

tionships between protein targets and all drugs form the complex

network of drugs and/or targets. For instance, Yildirim and Goh

et al. [26] have built a bipartite graph composed of US Food and

Drug Administration (US FDA) approved drugs and proteins linked

by drug-target binary associations. The resulting network connects

most drugs into a highly interlinked giant component, with strong

local clustering of drugs of similar types according to Anatomical

Therapeutic Chemical classiﬁcation. Topological analyses of this

network quantitatively showed an overabundance of ‘follow-on’

drugs, that is, drugs that target already targeted proteins.

combined QSAR-CN analysis may become of major importance for

the prediction of the activity of new compounds against different

targets or the discovery of new targets. In this sense we reported

two illustrative experiments that combine both experimental and

theoretical studies to show how to use this model in practical

situations. In experiment 1, we report by ﬁrst time mt-QSAR and CN

prediction, synthesis, characterization, and MAO-A and MAO-B

pharmacological assay of 8 rasagiline derivatives. In experiment 2,

we report sampling, parasite culture, sample preparation, 2-DE,

MALDI-TOF and -TOF/TOF MS, MASCOT search, MM/MD 3D struc-

ture modeling, and QSAR prediction of CN for peptidome of

hemoglobin found in parasite Fasciola hepatica. In Fig. 1 we depict

a ﬂowchart with the main steps given in this work to train and

validate the ANN classiﬁer.

2

. Materials and methods

.1. Computational methods

.1.1. MARCH-INSIDE technique

Entropy parameters for drug graphs. The MARCH-INSIDE

approach applied to drugs (D) is based on the calculation of the

different physicochemical molecular properties as an average of

atomic properties for all the molecules or groups of atoms (G)

[10,19,29]. For instance, it is possible to derive average estimations

of information about molecular structure using entropy indices

D

k

q (G) [36,37].

X

^h^kp

i

D

^kp

q

ðGÞ ¼ ꢀ

ðGÞ$log

ðGÞ

(1)

k

j

j˛G

It is possible to consider isolated atoms (k ¼ 0) in the estimation of

_t_h_e_m_o_l_e_c_u_l_a_r_p_r_o_p_e_r_t_i_e_sD

0

0 j

q (G). In this case the probabilities p(q )

In a previous work, our group have reported a QSAR model base

on the MARCH-INSIDE method to predict a large network of DTPs

are determined without considering the formation of chemical

bonds (simple additive scheme). However, it is possible to consider

the gradual effects of the neighboring atoms at different distances

[27]. However, even when this model is useful to predict targets for

many proteins it lack of availability for public research due to it was

not implemented as an online web server. The problem with many

QSAR models is more serious because many of them work only for

one target protein or for a limited family of organic compounds. We

then develop new statistical methods to predict simultaneously on

a large scale unknown DTPs from chemical structure and 3D

structure of target proteins. In principle, we can select between

more than 1600 different molecular descriptors to solve the former

problem [28]. However, not many methods offer one unique soft-

ware platform to calculate parameters for both drugs and protein

structures based on uniﬁed theoretic background more easily to

rationalize. Our group has introduced elsewhere a Markov Chain

Model (MCM) method named MARkov CHains Invariants for

Network SImulation and DEsign (MARCH-INSIDE). The MARCH-

INSIDE approach makes use of the same MCM theoretic formula-

tion to calculate the average values of different molecular TIs and

physicochemical properties from 2D, 3D, and/or sequence chemical

structures including drugs, DNA, RNA, and proteins, see a recent

review [29]. MARCH-INSIDE parameters not only offer these

advantages but also may used as inputs to train ANNs with the

software STATISTICA (e.g.) [30e35].

(

k > 0) in the molecular backbone. In order to reach this goal the

method uses an MM, which determines the absolute probabilities

) with which the atoms placed at different distances k affect the

k j

p (q

contribution of the atom j to the molecular property in question.

Finally, it is interesting to note that one can sum only the atoms

included in a speciﬁc group of atoms (G) rather than all atoms. In

this way we can approach speciﬁc classes of average properties

3

such as average entropy for sp carbon atoms (Csp₃) or average

entropy for heteroatoms (Het). All calculations were performed

using the program MARCH-INSIDE [10,19,29], which was developed

in-house, see recent reviews for details.

Entropy parameters of protein residue networks. In previous

works we have predicted protein function based on different

protein structural parameters derived from a Markov matrix that

account for electrostatic interactions between aminoacid pairs in

the 3D structure of the protein. One of the classes of parameters

_u_s_e_d_w_a_s_c_a_l_l_e_d_t_h_e_S_h_a_n_n_o_n_E_n_t_r_o_p_yT

k

q (R) of the markov matrix.

These values are used here as inputs to describe information about

the structure of the drug-target proteins (T) in order to construct

the mt-QSAR models for DTPs. The detailed explanation has been

published before [30,38e45] and reviewed in detail more recently

In this work, we developed a multi-target QSAR (mt-QSAR)

classiﬁer using the MARCH-INSIDE technique to calculate structural

parameters of drug and target plus one ANN to seek the model. The

validation of the model was carried out by means of external pre-

dicting series. We also compare this model with other ANN models

developed in this work and Machine Learning (ML) classiﬁers

published before to address the same problem. A very good

MARCH-INSIDE-QSAR model was obtained, and the subsequent

₂₉_]_._A_s_f_o_l_l_o_w_s_w_e_g_i_v_e_t_h_e_f_o_r_m_u_l_a_f_o_rT

[

q

k

(R) values and some

general explanations:

X

h

i

T

k

q

_kðRÞ ¼ ꢀ

p_jðRÞ$log p_jðRÞ

(2)

j˛R

Where, ^kp

effect of the electrostatic interaction propagates from the

(R) values are the absolute probabilities with which the

i

1076

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Fig. 1. Flowchart of all steps given in this work to develop the new model.

aminoacid i^t^hto other aminoacids j^t^hnext to it and returns to i^t^h

after k-steps. These probabilities refer to: aminoacids considered

isolated in the space (k ¼ 0), interaction between aminoacids in

direct contact (k ¼ 1) or spatial (k > 1) indirect interactions

between aminoacids placed at a distance equal to k-times the cut-

off distance (rij ¼ k$rcut-off) in the residue network. Euclidean 3D

(types of regions) ꢁ 6(orders considered) ¼ 30

T

k

q (R) indices for

each protein.

Statistical analysis. Let be ^D

q (G) entropy descriptors molecular

that codify information about drug structure and

descriptors that codify information about drug-target proteins; we

attempt to develop a simple mt-QSAR model in the form of a linear

classiﬁer with the general formula:

k

T

k

q (R) entropy

space r

3

¼ (x, y, z) coordinates of the C

a

atoms of aminoacids listed

in protein PDB ﬁles. For calculation, all water molecules and metal

ions were removed [19]. All calculations were carried out with our

in-house software MARCH-INSIDE 2.0 [19]. For the calculation, the

MARCH-INSIDE software always uses the full matrix, never a sub-

matrix, but the last summation term may run either for all ami-

noacids or only for some speciﬁc protein regions (R) denoted as: c

5

X

D

T

SðDTPÞ_p_r_e_d

¼

a_G_;_k

$

q

_kðGÞ þ

b_R_;_k

$

q

_kðRÞ þ c₀

(3)

k ¼ 0

We used Linear Discriminating Analysis (LDA) to ﬁt this discrimi-

nant function. The model deals with the classiﬁcation of

a compound set with or without afﬁnity on different receptors.

A dummy variable Afﬁnity Class (AC) was used as input to codify the

afﬁnity. This variable indicates either high (AC ¼ 1) or low (AC ¼ 0)

for core, i for inner, m for middle, and s for surface regions,

T

respectively). Consequently, we can calculate different

k

q (R) for the

aminoacids contained in the regions (c, i, m, s, or t) and placed at

a topological distance k each other within this orbit (k is the order)

afﬁnity of the drug by the receptor. S(DTP)

pred

or DTP afﬁnity pre-

[32,38,39,46,47]. In this work, we have calculated altogether 5

dicted score is the output of the model and it is a continuous

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1077

dimensionless score that sorts compounds from low to high afﬁnity

to the target coinciding DTPs with higher values of S(DTP)pred and

nDTPs with lowest values. In equation (3), b represents the coefﬁ-

cients of the classiﬁcation function, determined by the LDA module

of the STATISTICA 6.0 software package [48]. We used Forward

Stepwise algorithm for a variable selection. The statistical signiﬁ-

cance of the LDA model was determined calculating the p-level (p)

of error with Chi-square test. We also inspected the Speciﬁcity,

Sensitivity, and total Accuracy to determine the quality-of-ﬁt to

data in training. Cases for training set were selected at random out

of the cases in full dataset. The remnant cases were used to validate

the model. The validation of the model was corroborated with

these external prediction series; these cases were never used to

train the model. The ration between training/validation set was 2/1

approximately. This procedure to select training and validation sets

is largely known and used to train QSAR models [49e55].

Dataset. The dataset was formed by a set of marketed and/or

reported drugs/receptors (proteins) pairs where afﬁnity of drugs

with the receptors was established taking into consideration the

Drug Bank. The dataset was formed to more than 526 drugs with

respectively 323 protein receptors, so we were able to collect

already 5784 cases (drug/protein receptors) instead of 526 ꢁ 323

cases. The dataset were used to perform an LDA model. Overall

model classiﬁcation accuracy was 88.83% (3475/3912 cases) in

training, 88.51% (1657/1872) in validation. In addition the dataset

was used to develop ANN models to performance the model.

Overall ANN model classiﬁcation accuracy was 92%. To know if the

model works well, the model used 8 own compounds rasagiline

derivates to predict their protein receptor. The names or codes for

all compounds are depicted in Table 1SM and Table 2SM of the

supplementary material, due to space constraints, as well as the

references consulted to compile the data in this table.

DrugeProtein (DP) network construction. In order to achieve the

drug and protein afﬁnity with a network approach where one node

represents a drug, target, or drug-target pair and the edges express

relationships between pairs of drugs and/or targets [26]. We built

a bipartite graph composed of Drug Bank and Protein Data Bank-

approved drugs and proteins linked by drug-target binary associ-

ations. The resulting network connects most drugs into a highly

interlinked giant component, with strong local clustering of drugs

of similar types. We build two complex networks, a ﬁrst for the

observed data and the second with the data predicted by the

model. First, using the Excel software in a column we introduce all

the proteins, the drugs used quotation marks in our database. Then

in another column lists all the cases, in total 837 vertices. At the

beginning of this column puts the total number of vertices, there

are currently two columns of the name of drug and protein and

their corresponding number of vertices. At the end of the columns

are placed bows in the ﬁrst column put the number of vertices for

the drug and in another column corresponding to the protein. The

ﬁle was saved as a .txt format ﬁle. After we had renamed the .txt ﬁle

as a .net ﬁle we read it with the CentiBin software [56,57]. Using

CentiBin we can not only represent the network but also highlight

all Drugs and Protein (nodes) connected to a speciﬁc DRP. To

analyze the relationships between drug targets, we measured and

calculating closeness parameter, and the centrality. Lastly, the

protein and drug centralities were used as input in STATISTICA in

order to study the distribution of the network and compare it with

other ideal network distributions including normal, exponential.

geometries was carried out by the Molecular Mechanics Force Field

BIO þ (CHARMM). In setup we keep the options implemented by

default, but allowing a cut-off switching truncation rin ¼ 15 Å and

rout ¼ 17 Å. We refer here to MD in the sense of MD stochastic

simulation by the Monte Carlo (MC) method, although some authors

understand MD as only the MD deterministic search. The Molecular

Dynamics Trajectories (MDTs) or energetic proﬁles of all the starting

structure of peptides were obtained by means of MC method, with

the HyperChem package [59,60]. In this sense, the force ﬁeld AMBER

[61] of molecular mechanics was used with distant-dependent

dielectric constant (scale factor 1), electrostatic and van der Waals

values by default and a cut-off switched function with rin ¼ 15 and

r

out ¼ 17 Å (see Fig. 2). All the components in the force ﬁeld were

included and the atom type was recalculated by maintaining the

current charges. Finally, MD simulation was carried out by use of the

Monte Carlo algorithm in the vacuo at 300 K and 1000 optimization

d

steps, thus obtaining MDTs with 100 potential energy E

j

(j ¼ 1, 2, 3,

.100) values for each. We obtained 22 MDTs for 19 peptides. In

order to obtain realistic MDTs we monitored an additional param-

eter in MD algorithms; this is known as the acceptance ratio (ACCR).

It appears as ACCR on the list of possible selections in the MC

Averages dialog boxof HyperChem (see Fig. 2). The ACCR is a running

average of the ratio of the number of accepted moves to attempted

moves. Varying the step size can have a large effect on the ACCR

value. The step size, Dr, is the maximum allowed atomic displace-

ment used in the generation of trial conﬁgurations. The default value

of r in HyperChem is 0.05 Å [59]. For most organic molecules, this

will result in an ACCR of about 0.5 Å, which means that about 50% of

all moves are accepted. Increasing the size of the trial displacements

may lead to a more complete search of conﬁguration space, but the

acceptance ratio will, in general, decrease. Smaller displacements

generally lead to higher acceptance ratios but result in more limited

sampling. There has been little research to date as regards the

optimum value of the acceptance ratio.

2.2. Experimental methods

2.2.1. Study of rasagiline analogs

Identiﬁcation. Melting points are uncorrected and were deter-

mined in a Reichert Koﬂer Thermopan or in capillary tubes in

a Büchi 510 apparatus. Infrared spectra were recorded in a Per-

1

kineElmer 1640 FTIR spectrophotometer.

H NMR spectra

13

(300 MHz and 500 MHz) and C NMR spectra (75 MHz) were

recorded in a Bruker AMX 300 and DRX 500 spectrometer using

TMS as internal reference (chemical shifts in d values, J in Hz). Mass

spectra were recorded on a HP5988A spectrometer. FABMS were

obtained using MICROMASS AUTOSPEC mass spectrometer.

Microanalyses were performed in a PerkineElmer 240B elemental

analyzer by the Microanalysis Service of the University of Santiago

see Table 3SM. Most of reactions were monitored by TLC on pre-

coated silica gel plates (Merck 60 F254, 0.25 mm). Synthesized

products were puriﬁed by ﬂash column chromatography on silica

gel (Merck 60, 230e240 mesh) and crystallized if necessary.

Solvents were dried by distillation prior use.

Compound 2. (ꢂ)-3-Amino-3-(thiophen-3-yl)propanoic acid. A

solution of thiophen-3-carbaldehyde (2.00 g, 17.83 mmol) in EtOH

(5.5 mL) was added ammonium acetate (2.74 g, 35.58 mmol) and

malonic acid (1.85 g, 17.78 mmol). The reaction mixture was stirred

was reﬂuxed for 7 h. The precipitate formed was ﬁltered and washed

with boiling EtOH (3 ꢁ 10 mL) to give 2 (2.20 g, yield 72%) as a white

ꢃ

2

.1.2. Theoretical study of hemoglobin peptidome of parasite F.

hepatica

MM/MD. Molecular Mechanics (MM) and Molecular Dynamics

MD) study. For MM study we ﬁrst introduced the sequence of the 30

peptides in the HyperChem [58]; the optimization of their

solid. Mp 223e224 C. IR (KBr):

n

¼ 2954, 2152,1532,1403,1101, 992,

ꢀ

1 1

925, 786 cm . H NMR (300 MHz, TFA-d):

exch., OH þ NH ), 6.97e6.91 (m, 2H, 2-Hthiophene and 5-Hthiophene),

6.63e6.61 (m,1H, 4-Hthiophene), 4.58e4.54 (m,1H, 3-H), 2.92 (dd,1H,

d

¼ 11.03 (br s, 3H, D

2

O

2

(

1

3

J ¼ 18.4, 9.6 Hz, 2-HH), 2.72 (dd, 1H, J ¼ 18.4, 4.2 Hz, 2-HH) ppm.

C

1078

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Fig. 2. Snapshot of software HyperChem with MM/MD study for one peptide.

NMR (75 MHz, TFA-d):

d

¼ 175.01 (CO), 131.92 (C2thiophene), 126.76

removed under reduced pressure and the residue so obtained was

partitioned between H O (100 mL) and EtOAc (75 mL). The aqueous

phase was extracted with EtOAc (15 ꢁ 75 mL), and the pooled organic

extracts were dried over Na SO , after which removal of the solvent

under reduced pressure afforded 1.72 g of an oily yellow residue that

was fractionated chromatographically on a column of silica gel using

6/1 hexane/acetone as eluent. From the ﬁrst non-void fractions eluted

the cis-isomer 4a was isolated (0.65 g, 32%). A second group of frac-

tions afforded a mixture of trans/cis-isomers 4a/b (0.40 g, 20%) and the

(

[

(

C4thiophene), 123.48 and 122.45 (C3thiophene and C5thiophene), 47.01

2

þ

CH), 34.12 (CH

2

) ppm. EIMS: m/z (%) ¼ 172 (2) [M þ 1] , 171 (17)

þ

M ],112 (100) [M ꢀ CH

2

COOH],111 (3),110 (17), 85 (54), 84 (4), 83

2

4

þ

3) [M ꢀ C

3

H

6

NO

2

], 70 (5), 58 (13). Anal. calcd. for C

7

H NO

9 2

S

171.21): C 49.10, H 5.30, N 8.18; found C 49.43, H 5.37, N 8.05.

Compound 3. (ꢂ)-2,2,2-Triﬂuoro-N-(5,6-dihydro-6-oxo-4H-

cyclopenta[b]thiophen-4-yl)acetamide. A solution of 2 (9.50 g,

5.49 mmol) in triﬂuoracetic acid (22 mL) was stirred for 30 min at

room temperature under argon and then triﬂuoracetic anhydride

22 mL) was added. The reaction mixture was reﬂuxed for 5 h and

5

last group of fractions afforded the cis-isomer 4b (0.62 g, 31%).

ꢃ

(

Compound (ꢂ)-4a: mp 135e136 C. IR (KBr):

n

¼ 3272, 3093,

H NMR (300 MHz,

O exch., J ¼ 7.5 Hz, NH), 7.52 (d, 1H,

J ¼ 4.9 Hz, 2-H), 6.81 (d, 1H, J ¼ 4.9 Hz, 3-H), 5.55 (d, 1H, D O exch.,

J ¼ 6.2 Hz, OH), 5.11e5.01 (m, 2H, 4-H þ 6-H), 3.15 (dt, 1H, J ¼ 13.3,

ꢀ

1 1

evaporated to dryness and the crude residue (18 g) so obtained was

puriﬁed on a silica gel column using 1:1 hexane/EtOAc as eluent, to

2893, 1697, 1555, 1190, 1041, 995, 939 cm

DMSO-d ):

¼ 9.78 (d, 1H, D

.

6

d

2

ꢃ

giving pure 3 (10.70 g, 77%) as a white solid. Mp 138e141 C [Lit. mp

2

ꢃ

1

41 C]. H NMR (300 MHz, CDCl

.17 (d,1H, J ¼ 4.8 Hz, 3-H), 7.00 (br s,1H, D

J ¼ 7.5 Hz, 4-H), 3.50 and 2.83 (AB part of an ABM system, 2H,

3

):

d

¼ 7.97 (d, 1H, J ¼ 4.9 Hz, 2-H),

13

7

2

O exch., NH), 5.59 (t,1H,

7.4 Hz, 5-HH), 2.14 (dt, 1H, J ¼ 13.3, 5.5 Hz, 5-HH) ppm. C NMR

(75 MHz, DMSO-d ):

¼ 148.42 (CO), 145.45 (C6a), 131.74 (C2),

121.74 (C3), 118.19 (C3a), 114.35 (CF ), 69.11 (C6), 48.62 (C4), 46.87

6

d

13

J ¼ 18.6, 7.1, 2.2 Hz, 5-H

¼ 192.78 (C6), 165.31 (CO), 143.05 (C6a), 142.57 (C2), 123.82 (C3),

17.93 (C3a), 114.12 (CF ), 48.90 (C5), 46.08 (C4) ppm. EIMS: m/z

2

) ppm. C NMR (75 MHz, CDCl

3

):

3

þ

d

(C5) ppm. EIMS: m/z (%) ¼ 251 (3) [M ], 233 (100) [M ꢀ H

2

O], 139

NO], 137 (31), 136 (49),

NO

þ

1

3

(23) [M ꢀ C

2

HF

3

NO], 138 (17) [M ꢀ C

2 2 3

H F

þ

(

[

(

%) ¼ 250 (6) [M þ 1] , 249 (49) [M ], 232 (3) [M ꢀ OH], 180 (17)

121 (20), 110 (43), 97 (10), 69 (31). Anal. calcd. for C

9

H F

8 3

2

S

þ

M

ꢀ CF

3

], 152 (33), 137 (42) [M ꢀ NHCOCF

3

], 136 (100)

(251.22): C 43.03, H 3.21, N 5.58; found C 43.37, H 3.43, N 5.65.

ꢃ

ꢀ NH

2

COCF

3

], 134 (23), 124 (10), 109 (31), 108 (22), 97 (25), 69

Compound (ꢂ)-4b: mp 118e121 C. IR (KBr):

n

¼ 3299, 2929,

):

¼ 9.71 (d,

O exch., J ¼ 7.6 Hz, NH), 7.53 (d, 1H, J ¼ 4.9 Hz, 2-H), 6.83 (d,

1H, J ¼ 4.9 Hz, 3-H), 5.45 (d, 1H, D O exch., J ¼ 6.4 Hz, OH),

5.37e5.30 (m, 1H, 6-H), 5.28e5.24 (m, 1H, 4-H), 2.68e2.52 (m, 2H,

ꢀ1 1

48). Anal. calcd. for C

9

H

6

F NO

3 2

S (249.21): C 43.38, H 2.43, N 5.62;

1695, 1550, 1183 cm . H NMR (300 MHz, DMSO-d

1H, D

6

d

found C 43.67, H 2.65, N 5.49.

2

Compounds 4a and 4b. (ꢂ)-2,2,2-Triﬂuoro-N-(cis-5,6-dihydro-6-

hydroxy-4H-cyclopenta[b]thiophen-4-yl)acetamide and (ꢂ)-2,2,2-

triﬂuoro-N-(trans-5,6-dihydro-6-hydroxy-4H-cyclopenta[b]thiophe-

n-4-yl)acetamide. NaBH

compound 3 (2.00 g, 8.02 mmol) in MeOH (8 mL). The resulting

2

13

5-H

2

) ppm. C NMR (75 MHz, DMSO-d

(C6a), 132.01 (C2), 121.80 (C3), 118.19 (C3a), 114.36 (CF

6

):

d

¼ 148.73 (CO), 146.46

4

(0.60 g,15.86 mmol) is added to a solution of

3

), 69.76 (C6),

þ

49.82 (C4), 47.28 (C5) ppm. EIMS: m/z (%) ¼ 233 (1) [M ꢀ H

2

O],139

þ

mixture is stirred at room temperature for 10 min. The solvent was

(16) [M ꢀ C

2

HF

3

NO], 138 (100) [M ꢀ C

2 2 3

H F NO], 137 (11), 136

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1079

ꢃ

(

11), 121 (8), 110 (18), 69 (22). Anal. calcd. for C

9

H

8

F

3

NO

2

S (251.02):

Compound (ꢂ)-7a: brown, waxy solid, mp 104e108 C. IR (KBr):

¼ 3285, 3269, 3125, 3077, 2970, 2921, 2844, 2824, 1444, 1333,1122,

C 43.03, H 3.21, N 5.58; found C 43.25, H 3.47, N 5.69.

Compound 5. (ꢂ)-cis-4-Amino-5,6-dihydro-4H-cyclopenta[b]

thiophen-6-ol. Compound 4a (0.30 g, 1.19 mmol) was suspended in

n

ꢀ1 1

1077, 1033, 904 cm . H NMR (300 MHz, CDCl

3

):

d

¼ 7.34 (d, 1H,

J ¼ 4.9 Hz, 2-H), 7.00 (d, 1H, J ¼ 4.9 Hz, 3-H), 5.09e5.07 (m, 1H, 6-H),

4.19 (dd, 1H, J ¼ 6.9, 2.8 Hz, 4-H), 3.63e3.50 (m, 4H, 2 ꢁ CH ),

2.97e2.87 (m, 1H, 5-HH), 2.82 (br s, 1H, D O exch., OH), 2,45 (dt, 1H,

J¼ 14.2, 2.7 Hz, 5-HH), 2.26(t, 2H, J ¼ 2.2 Hz,2 ꢁ C^CH) ppm. C NMR

(75 MHz, CDCl ):

a mixture of MeOH/H

2

O (4.5/2.6 mL), and K

2

CO

3

(0.33 g, 2.39 mmol)

2

was added. The mixture was heated to reﬂux with stirring for 1 h.

The reaction mixture was cooled to room temperature and the

solvents was evaporated to dryness and the residue so obtained was

partitioned between H

phase was extracted with EtOAc (4 ꢁ 50 mL). The organic phase was

separated, dried (Na SO ), and evaporated to dryness to give 5a as

a waxy yellow solid (0.18 g, 99%). Mp 101e104 C. IR and H NMR

2

13

3

d

¼ 147.83 (C6a), 146.77 (C3a), 130.95 (C2), 122.88

2

O (30 mL) and EtOAc (50 mL). The aqueous

(C3), 79.47 (2 ꢁ C^CH), 73.50 (2 ꢁ C^CH), 70.27 (C6), 60.86 (C4),

þ

43.90 (C5), 39.85 (2 ꢁ CH

2

C^CH) ppm. EIMS: m/z (%) ¼ 231 (1) [M ],

þ

2

4

230 (4) [M ꢀ 1] , 214 (4) [M ꢀ OH], 213 (3) [M ꢀ H

2

O],194 (9), 193

ꢃ

1

þ

(19),192 (100) [M ꢀ CH

2

C^CH],188 (11),174 (19),173 (11),139 (47),

138 (9),137 (14), 136 (13),135 (7),122 (17),121 (13), 111 (21), 110 (17),

85 (16), 84 (12), 71 (16), 57 (17), 55 (10). Anal. calcd. for C13 13NOS

13

were coincidences with Lit. C NMR (75 MHz, CDCl

3

):

d

¼ 151.25

(

C6a), 146.56 (C3a), 131.75 (C2), 121.52 (C3), 70.85 (C6), 51.50 (C5),

H

þ

5

1.42 (C4) ppm. EIMS: m/z (%) ¼ 155 (2) [M ], 138 (22) [M ꢀ OH],

(231.31): C 67.50, H 5.66, N 6.06; found C 67.69, H 5.88, N, 6.13.

Compounds 6b, 7b. (ꢂ)-trans-5,6-Dihydro-4-(prop-2-ynyla-

mino)-4H-cyclopenta[b]thiophen-6-ol (6b) and (ꢂ)-trans-5,6-

dihydro-4-(diprop-2-ynylamino)-4H-cyclopenta[b]thiophen-6-ol

(7b). Compound 6b and 7b were obtained in the same way as 6a and

7a from 5b (0.75 g, 4.84 mmol) but under slightly different reaction

conditions [reﬂux 4.5 h]. The yellow oil obtained (0.8 g) was frac-

tionated chromatographically on a column of silica gel using 3/1 and

1/2 hexane/EtOAc as successive mixtures of eluents. From the ﬁrst

non-void fractions eluted the compound 7b (0.49 g, 53%) was iso-

lated. A second group of fractions afforded the compound 6b (0.14 g,

þ

1

8

5

37 (15) [M ꢀ H

2

O], 136 (36), 112 (100), 111 (16), 110 (46), 109 (13),

5 (19), 69 (11), 58 (10). Anal. calcd. for C NOS (155.22): C 54.17, H

.84, N 9.02; found C 54.40, H 5.70, N 9.19.

Compound 5b. (ꢂ)-trans-4-Amino-5,6-dihydro-4H-cyclopenta

b]thiophen-6-ol. Compound 5b was obtained, as the sole product,

7 9

H

[

as an orange waxy solid in the same way as 5a, from 4b under

slightly different reaction conditions (reﬂux 6 h). The residue

obtained was fractionated chromatographically on a column of

silica gel using 2/1 CH

fractions eluted the trans-isomer 5b, as a orange waxy solid (75%).

2 2

Cl /MeOH as eluent. From the non-void

1

IR concordat with Lit. H NMR (300 MHz, CDCl

J ¼ 4.9 Hz, 2-H), 6.91 (d, 1H, J ¼ 4.9 Hz, 3-H), 5.16e5.14 (m, 1H, 6-H),

.33 (t, 1H, J ¼ 5.8 Hz, 4-H), 3.98e3.23 (br s, 3H, D O exch.,

), 2.46e2.42 (m,1H, 5-HH), 2.32e1.97 (m,1H, 5-HH) ppm.

3

):

d

¼ 7.44 (d, 1H,

13%).

ꢃ

Compound (ꢂ)-6b: mp 71e74 C. IR (KBr):

n

¼ 3268, 3254, 2927,

ꢀ

1 1

4

2

2853, 1486, 1432, 1336, 1301, 1193, 1115, 1097, 1034 cm

(300 MHz, CDCl ):

¼ 7.34 (d, 1H, J ¼ 4.9 Hz, 2-H), 6.92 (d, 1H,

J ¼ 4.9 Hz, 3-H), 5.41e5.38 (m,1H, 6-H), 4.55e4.51 (m,1H, 4-H), 3.46

(d, 2H, J ¼ 2.3 Hz, CH ), 2.61e2.56 (m, 2H, 5-H ), 2.26 (t,1H, J ¼ 2.3 Hz,

C^CH), 2.11 (br s, 2H, D

O exch., OH þ NH) ppm. C NMR (75 MHz,

CDCl ):

¼ 149.11 (C6a), 146.44 (C3a), 131.41 (C2), 121.70 (C3), 81.72

(C^CH), 71.86 (C^CH), 71.26 (C6), 56.52 (C4), 49.07 (C5), 36.11

H NMR

OH þ NH

2

3

d

1

3

C NMR (75 MHz, CDCl

3

):

d

¼ 152.66 (C6a), 146.07 (C3a), 130.93

(

[

(

C2), 122.12 (C3), 69.68 (C6), 51.65 (C5), 51.54 (C4) ppm. EIMS: m/z

2

þ

13

%) ¼ 155 (1) [M ], 139 (27), 138 (33) [M ꢀ OH], 137 (13)

2

þ

M

ꢀ H

2

O],136 (33),123 (3),121 (8),113 (8),112 (100),110 (57),109

3

d

7 9

16), 85 (26), 69 (26). HRMS m/z calcd. for C H NOS, 155.0405;

þ

found, 155.0421.

(CH

(2) [M ꢀ H

2

) ppm. EIMS: m/z (%) ¼ 192 (3) [M ꢀ 1] ,176 (6) [M ꢀ OH],175

þ

Compounds 6a, 7a. (ꢂ)-cis-5,6-Dihydro-4-(prop-2-ynylamino)-

2

O], 174 (10), 155 (9), 154 (100) [M ꢀ CH

2

C^CH], 150

4

H-cyclopenta[b]thiophen-6-ol (6a) and (ꢂ)-cis-5,6-dihydro-4-

(15), 139 (11),138 (7), 137 (12), 136 (21),135 (3),111 (8),110 (13), 109

(7), 85 (4), 77 (5), 65 (6). Anal. calcd. for C10H11NOS (193.26): C 62.15,

(

diprop-2-ynylamino)-4H-cyclopenta[b]thiophen-6-ol (7a). To

a solution of 5a (0.64 g, 4.13 mmol) and K

2

CO

3

(0.57 g, 4.13 mmol) in

H 5.74, N 7.25; found C 62.32, H 5.86, N, 7.19.

ꢃ

MeCN (13 mL) under argon was added dropwise a solution of

propargyl bromide (0.46 mL, 4.13 mmol). The resulting suspension

was heated to reﬂux with stirring for 18 h; whereafter most of

volatiles were partitioned between EtOAc (50 mL) and 2 N NaOH

Compound-(ꢂ)-7b: mp 57e59 C. IR (KBr):

n

¼ 3287, 3203,

1 1

ꢀ

2949, 2829, 1440, 1419, 1393, 1365, 1126, 1115, 1031 cm

(300 MHz, CDCl ):

¼ 7.37 (d, 1H, J ¼ 4.7 Hz, 2-H), 7.00 (d, 1H,

J ¼ 4.7 Hz, 3-H), 5.49e5.40 (m, 1H, 6-H), 4.64e4.61 (m, 1H, 4-H),

3.55e3.43 (m, 4H, 2 ꢁ CH ), 2.97 (ddd, 1H, J ¼ 14.5, 6.7, 3.7 Hz, 5-

HH), 2.45 (ddd, 1H, J ¼ 14.5, 7.2, 3.4 Hz, 5-HH), 2.26e2e0.24 (m, 2H,

H NMR

3

d

(

50 mL), the organic phase was extracted with 2 N NaOH

SO ), and

2

2 ꢁ 50 mL). The organic phase was separated, dried (Na

2

4

1

3

evaporated to dryness. The residue (0.64 g) was ﬂash chromato-

graphed on silica gel, using 3/1 and 1/2 hexane/EtOAc as successive

mixtures of eluents. From the ﬁrst non-void fractions eluted the

compound 7a (0.34 g, 36%) was isolated. A second group of frac-

2 ꢁ C^CH), 1.95 (br s, 1H, D

2

O exch., OH) ppm. C NMR (75 MHz,

¼ 147.47 (C6a), 147.13 (C3a), 131.27 (C2), 122.93 (C3), 79.65

CDCl ):

3

d

(2 ꢁ C^CH), 73.13 (2 ꢁ C^CH), 71.41 (C6), 62.08 (C4), 44.25 (C5),

þ

39.51 (2 ꢁ CH

2

) ppm. EIMS: m/z (%) ¼ 232 (1) [M þ 1] , 231 (7)

þ

tions afforded the compound 6a (0.19 g, 24%).

[M ], 230 (7) [M ꢀ 1] , 214 (6) [M ꢀ OH], 213 (2) [M ꢀ H

2

O], 194

ꢃ

þ

Compound (ꢂ)-6a: white solid, mp 59e63 C. IR (KBr):

n

¼ 3290,

(18), 193 (39), 192 (100) [M ꢀ CH

2

C^CH], 188 (22), 174 (24), 173

3

1

220, 2922, 2850, 1599, 1494, 1429, 1339, 1183, 1113, 1086, 1060,

(11), 162 (11),149 (10),148 (15),147 (11),140 (10),139 (69),138 (22),

137 (28), 136 (27), 135 (13), 134 (11), 122 (24), 121 (25), 111 (36), 110

(40), 109 (18), 85 (10), 84 (7), 78 (14), 77 (23), 69 (10), 67 (12), 66

ꢀ

1 1

012, 954, 762 cm . H NMR (300 MHz, CDCl

J ¼ 4.8 Hz, 2-H), 6.92 (d, 1H, J ¼ 5.0 Hz, 3-H), 5.06 (dd, 1H, J ¼ 6.7,

.0 Hz, 6-H), 4.23 (dd,1H, J ¼ 6.7, 2.0 Hz, 4-H), 3.44 (d, 2H, J ¼ 2.3 Hz,

CH ), 3.15 (br. s, 2H, D

O exch., OH þ NH), 3.05e2.95 (m, 1H, 5-HH),

.27 (t, 1H, J ¼ 2.2 Hz, C^CH), 2.09 (dt, 1H, J ¼ 14.5, 2.0 Hz, 5-HH)

3

):

d

¼ 7.33 (d, 1H,

2

(19), 65 (21), 55 (13). Anal. calcd. for C13H13NOS (231.31): C 67.50, H

5.66, N 6.06; found C 67.77, H 5.81, N, 6.19.

2

Compound 8a. (ꢂ)-cis-4-[(Diprop-2-ynyl)amino]-5,6-dihydro-

1

3

ppm. C NMR (75 MHz, CDCl

3

):

C2), 121.62 (C3), 81.33 (C^CH), 72.20 (C^CH), 70.60 (C6), 55.38

C4), 47.62 (C5), 35.89 (CH

d

¼ 148.27 (C6a),147.41 (C3a),131.19

4H-cyclopenta[b]thiophen-6-yl acetate. The compound 7a (50 mg,

(

[

(

2 3

0.216 mmol) was stirred with Ac O (0.5 mL) and dry Et N (0.5 mL)

for 5.5 h under argon at room temperature, the resulting mixture

was concentrated to dryness, and the solid residue was dissolved in

2

C^CH) ppm. EIMS: m/z (%) ¼ 193 (3)

þ

M ], 192 (16) [M ꢀ 1] , 176 (5) [M ꢀ OH], 175 (5) [M ꢀ H

2

O], 154

þ

37) [M ꢀ CH

2

C^CH], 140 (11), 139 (100), 138 (20), 137 (19), 136

CH

with saturated NaHCO

(Na SO ), and concentrated under reduced pressure. The resulting

yellow oil (60 mg) was chromatographed on silica gel with 10/1

2

Cl

2

(30 mL) and the organic phase was washed successively

19),135 (4),122 (16),121 (13),111 (18),110 (16),109 (12), 85 (11), 77

11), 69 (10), 65 (10), 57 (12), 55 (12). Anal. calcd. for C10 11NOS

193.26): C 62.15, H 5.74, N 7.25; found C 62.36, H 5.92, N, 7.41.

3

(3 ꢁ 20 mL) and H O (3 ꢁ 20 mL), dried

2

H

2

4

1080

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

hexane/EtOAc as eluent, affording 8a (51 mg, 86%) as a white solid,

186 (5), 176 (5), 175 (13), 174 (48), 173 (11), 148 (7), 147 (8), 139 (4),

138 (2), 137 (10), 136 (12), 135 (7), 134 (5), 123 (16), 122 (100), 121

(52), 105 (60), 92 (34), 77 (37). HRMS m/z calcd. for C20H17NO S,

2

ꢃ

mp 59e60 C. IR (KBr):

n

¼ 3281, 3262, 2922, 1721, 1428, 1367, 1305,

ꢀ

1 1

1245 cm

H NMR (300 MHz, CDCl ):

3

d

¼ 7.38 (d, 1H, J ¼ 5.0 Hz, 2-

H), 6.97 (d, 1H, J ¼ 5.0 Hz, 3-H), 5.83 (dd, 1H, J ¼ 7.6, 2.9 Hz, 6-H),

335.0980; found, 335.0996.

4

2

3

.42 (dd,1H, J ¼ 7.9, 3.2 Hz, 4-H), 3.60e3.44 (AB system, 2H, J ¼ 17.0,

.3 Hz, CH ),

), 3.59e3.43 (AB system, 2H, J ¼ 17.0, 2.3 Hz, CH

.11e3.01 (m, 1H, 5-HH), 2.57 (dt, 1H, J ¼ 15.2, 3.2 Hz, 5-HH), 2.24

Compound 9b. (ꢂ)-trans-4-[(Diprop-2-ynyl)amino]-5,6-dihiy-

dro-4H-cyclopenta[b]thiophen-6-yl benzoate. The compound 9b

was obtained in the same way as 9a from 7b (53 mg, 0.23 mmol)

but under slightly different reaction conditions [room temperature

21 h]. The obtained residue (62 mg) was fractionated chromato-

graphically on a column of silica gel using 20/1 hexane/EtOAc as

eluent. From the non-void fractions eluted the compound 9b

2

13

(

t, 2H, J ¼ 2.3 Hz, 2 ꢁ C^CH), 2.08 (s, 3H, CH

75 MHz, CDCl ):

¼ 171.22 (CO), 148.58 (C6a), 142.62 (C3a), 132.69

C2), 122.31 (C3), 80.03 (2 ꢁ C^CH), 72.98 and 72.51 (2 ꢁ C^CH),

2.47 (C6), 61.18 (C4), 39.33 (C5), 38.63 (2 ꢁ CH ), 21.12 (C7H ) ppm.

EIMS: m/z (%) ¼ 273 (2) [M ], 234 (12) [M ꢀ CH C^CH], 230 (62),

15 (15), 214 (79), 213 (98), 212 (100), 200 (11),199 (9),198 (22),192

16), 186 (11), 176 (13), 175 (28), 174 (94), 173 (26), 148 (14), 147 (16),

39 (48), 138 (13), 137 (23), 136 (24), 135 (13), 134 (10), 123 (21), 122

93), 121 (90), 111 (18), 110 (9), 109 (11), 92 (28), 77 (12), 66 (15), 65

11). Anal. calcd. for C15 S (273.35): C 65.91, H 5.53, N 5.12;

3

) ppm. C NMR

3

d

7

2

3

þ

2

(72 mg, 99%) as a yellow oil. IR (ﬁlm):

1600, 1450, 1266, 1107, 1069, 991 cm . H NMR (300 MHz, CDCl ):

3

n

¼ 3290, 2923, 2819, 1707,

ꢀ

1 1

2

(

1

(

d

¼ 8.02e7.99 (m, 2H, 2-Hbencene þ 6-Hbencene), 7.55e7.50 (m, 1H,

4-Hbencene), 7.42e7.38 (m, 3H, 3-Hbencene þ 5-Hbencene þ 2-H), 6.99

(d,1H, J ¼ 5.1 Hz, 3-H), 6.32 (d, J ¼ 6.6 Hz,1H, 6-H), 4.75 (ddd, J ¼ 7.4,

H

15NO

2

4.3,1.2 Hz 1H, 4-H), 3.56e3.47 (AB system, 2H, J ¼ 16.8, 2.3 Hz, CH

2

),

), 3.09 (ddd, 1H,

J ¼ 14.6, 7.2, 4.3 Hz, 5-HH), 2.76 (ddd, 1H, J ¼ 14.6, 7.2, 2.7 Hz, 5-HH),

found C 66.05, H 5.70, N, 5.08.

3.55e3.46 (AB system, 2H, J ¼ 16.8, 2.4 Hz, CH

2

Compound 8b. (ꢂ)-trans-4-[(Diprop-2-ynyl)amino]-5,6-dihiy-

dro-4H-cyclopenta[b]thiophen-6-yl acetate. The compound 8b was

obtained in the same way as 8a from 7b (53 mg, 0.23 mmol) but

under slightly different reaction conditions [room temperature

13

2.25 (t, 2H, J ¼ 2.3 Hz, 2 ꢁ C^CH) ppm. C NMR (75 MHz, CDCl

¼ 166.68 (CO), 149.15 (C6a), 142.78 (C3a), 133.06 (C4bencene),

132.75 (C2), 129.95 (C1bencene), 129.65 and 128.30

3

):

d

1

5 h]. The obtained residue (62 mg) was fractionated chromato-

[C2bencene þ C3bencene þ C5bencene þ C6bencene], 122.29 (C3), 79.72

graphically on a column of silica gel using 10/1 hexane/EtOAc as

eluent. From the non-void fractions eluted the compound 8b (46 mg,

(2 ꢁ C^CH), 74.57 (C6), 73.10 (2 ꢁ C^CH), 62.37 (C4), 39.71 (C5),

þ

39.34 (2 ꢁ CH

2

) ppm. EIMS: m/z (%) ¼ 334 (1) [M ꢀ 1] , 296 (1)

þ

7

1

4%) as a yellow oil. IR (ﬁlm):

n

¼ 3286, 1726, 1427, 1371, 1235,

015 cm . H NMR (300 MHz, CDCl ):

¼ 7.39 (d, 1H, J ¼ 5.0 Hz, 2-

H), 6.98 (d, 1H, J ¼ 5.0 Hz, 3-H), 6.09e6.07 (m, 1H, 6-H), 4.68 (ddd,

J ¼ 7.2, 4.7,1.5 Hz,1H, 4-H), 3.55e3.42 (m, 4H, 2 ꢁ CH ), 2.97 (ddd,1H,

J ¼ 14.6, 7.3, 4.4 Hz, 5-HH), 2,60 (ddd, 1H, J ¼ 14.6, 7.3, 2.9 Hz, 5-HH),

[M -CH

2

C^CH], 230 (37), 215 (8), 214 (36), 213 (100), 212 (54), 200

ꢀ1 1

3

d

(4), 199 (5), 198 (15), 186 (6), 176 (7), 175 (16), 174 (53), 173 (17), 148

(9), 147 (13), 139 (4), 138 (3), 137 (14), 136 (15), 135 (10), 134 (7), 123

(15), 122 (65), 121 (57), 105 (90), 92 (10), 78 (10), 77 (67). HRMS m/z

2

calcd. for C20H17NO S, 335.0980; found, 335.0992.

13

2

(

6

(

.25 (t, 2H, J ¼ 2.3 Hz, 2 ꢁ C^CH), 2.04 (s, 3H, CH

75 MHz, CDCl ):

¼ 171.21 (CO), 149.09 (C6a), 142.72 (C3a), 132.65

C2), 122.30 (C3), 79.74 (2 ꢁ C^CH), 73.95 (C6), 73.11 (2 ꢁ C^CH),

2.32 (C4), 39.59 (C5), 39.32 (2 ꢁ CH ), 21.09 (CH ) ppm. EIMS: m/z

%) ¼ 272 (1) [M ꢀ 1] , 234 (5) [M ꢀ CH C^CH], 230 (34), 215 (10),

14 (49), 213 (100), 212 (58), 199 (6), 198 (15), 192 (10), 186 (7), 176

3

) ppm. C NMR

MAO Inhibition Assay of rasagiline analogs. The potential effects

of the test drugs on hMAO activity were investigated by measuring

their effects on the production of hydrogen peroxide from p-tyra-

mine (a common substrate for both hMAO-A and hMAO-B), using

the 10-acetyl-3,7-dihydroxyphenoxazine as reagent and micro-

somal MAO isoforms prepared from insect cells (BTI-TN-5B1-4)

infected with recombinant baculovirus containing cDNA inserts for

hMAO-A or hMAO-B [36,37]. The production of H2O2 catalyzed by

MAO isoforms can be detected using the previously mentioned

reagent, a nonﬂuorescent, highly sensitive, and stable probe that

3

d

2

3

þ

2

(

8),175 (19),174 (63),173 (18),148 (11),147 (12),139 (31),138 (9),137

18), 136 (18), 135 (10), 134 (8), 123 (13), 122 (42), 121 (66), 111 (12),

110 (6), 109 (9), 92 (10), 77 (9), 66 (14), 65 (10). HRMS m/z calcd. for

C

15

H

15NO

2

S, 273.0823; found, 273.0841.

Compound 9a. (ꢂ)-cis-4-[(Diprop-2-ynyl)amino]-5,6-dihiydro-

H-cyclopenta[b]thiophen-6-yl benzoate. The compound 7a

L, 0.324 mmol),

L, 0.432 mmol) and a catalytic amount of DMAP in

2 2

reacts with H O in the presence of horseradish peroxidase to

4

produce a ﬂuorescent product, resoruﬁn. In this study, hMAO

activity was evaluated using the above method following the

general procedure described previously by us. The tested drugs

(new compounds and reference inhibitors) inhibited the control

enzymatic MAO activities and the inhibition was concentration

dependent. The corresponding IC50 values and MAO-B selectivity

ratios [IC50 (MAO-A)]/[IC50 (MAO-B)] are shown in Table 5. The

assayed compounds themselves do not react directly with the 10-

acetyl-3,7-dihydroxyphenoxazine, which indicates that these drugs

do not interfere with the measurements. In our experiments and

under our experimental conditions, hMAO-A displayed a Michaelis

(

50 mg, 0.216 mmol) was stirred with BzCl (37.61 m

3

dry Et N (60.21 m

MeCN (2.16 mL) for 23 h under argon at room temperature. The

resulting mixture was concentrated to dryness, and the solid

residue was dissolved in EtOAc (30 mL), the solid formed was

collected by ﬁltration and the ﬁltrate was concentrated under

reduced pressure and the resulting orange oil (130 mg) was chro-

matographed on silica gel with 20/1 hexane/EtOAc as eluent,

affording 9a (58 mg, 80%) as a yellow oil. IR (ﬁlm):

818, 1707, 1600, 1584, 1265, 1107, 1069, 1024, 1010, 710 cm

NMR (300 MHz, CDCl ):

¼ 8.04 (d, 2H, J ¼ 7.8 Hz, 2-

bencene þ 6-Hbencene), 7.54 (t, 1H, J ¼ 7.3 Hz, 4-Hbencene), 7.44e7.38

n

¼ 3290, 2922,

ꢀ

1 1

2

H

3

d

constant (SEM) of 457.17 (38.62

mM) and a maximum reaction

H

velocity (SD) of 185.67 (12.06 nmol/min/mg protein), whereas

(

6

3

m, 3H, 3-Hbencene þ 5-Hbencene þ 2-H), 6.98 (d, 1H, J ¼ 4.9 Hz, 3-H),

hMAO-B showed a SEM of 220.33 (32.80

1.97 nmol/min/mg protein (n ¼ 5)).

mM and a SD of 24.32,

.09 (dd, J ¼ 7.7, 2.1 Hz, 1H, 6-H), 4.50 (dd, 1H, J ¼ 7.8, 3.1 Hz, 4-H),

.63e3.49 (AB system, 2H, J ¼ 17.0, 2.6 Hz, CH ), 3.62e3.48 (AB

), 3.20e3.12 (m, 1H, 5-HH), 2.77 (m,

2

system, 2H, J ¼ 17.0, 2.6 Hz, CH

2

2.2.2. Experimental study Fasciola protein ﬁngerprints

13

1

CDCl

H, 5-HH), 2.23 (br s, 2H, 2 ꢁ C^CH) ppm. C NMR (75 MHz,

2.2.2.1. Experimental methods. Obtaining the ExcretoryeSecretory

Antigens (ESAs) of F. hepatica. ESAs of F. hepatica were obtained as

previously described by Mezo et al. [62]. Brieﬂy, liver adult ﬂukes

collected from bile ducts of naturally infected cows were washed

twice, ﬁrst in sterile saline solution containing antibiotics (100 IU/

0

3

):

d

¼ 166.61 (CO), 148.80 (C6a), 142.61 (C3a), 133.10 (C4 ),

0

1

32.86 (C2), 129.94 (C1 ), 129.67 and 128.37 (C2 þ C6 and

0

C3 þ C5 ), 122.28 (C3), 80.07 (2 ꢁ C^CH), 73.04 (C6), 72.97

(

2 ꢁ C^CH), 61.32 (C4), 39.32 (2 ꢁ CH

2

), 38.59 (C5) ppm. EIMS: m/z

þ

%) ¼ 335 (1) [M ], 334 (1) [M ꢀ 1] , 296 (1) [M ꢀ CH

2

C^CH], 230

mL penicillin and 100

38 C, and then in RPMI 1640 medium, supplemented with 20 mM

m

g streptomycin) and glucose (2 mg/mL) at

ꢃ

45), 215 (5), 214 (29), 213 (40), 212 (45), 200 (3), 199 (4), 198 (10),

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1081

HEPES, 0.3 g/L

L

-glutamine, 2 g/L sodium bicarbonate and antibi-

trypsin. The MALDI-TOF/TOF mass spectrometry analysis produces

peptide mass ﬁngerprints. Those observed with a signal-to-noise

greater than 10 were collated and represented as a list of mono-

isotopic molecular weights. Proteins ambiguosly identiﬁed by

peptide mass ﬁngerprints, were subjected to MS/MS sequencing

analyses using the 4800 Proteomics Analyzer (Applied Biosystems,

Framingham, MA). The suitable MS spectra precursors for MS/MS

sequencing analyses were selected, and fragmentation was carried

out using the acquisition method in the 1 kV ion reﬂector mode,

collision induced dissociation (CID) on, and precursor mass

window ꢂ5 Da. The plate model & default calibration were opti-

mized for the MSeMS spectra processing. For protein identiﬁcation,

the non-redundant NCBI database or SwissProt was searched using

MASCOT 2.1 (matrixscience.com) through the Global Protein Server

v3.5 from Applied Biosystems. The Mascot Search Parameters were:

i) Taxonomy: metazoa (animals); ii) Database: SwissProt and

NCBInr; iii) Enzyme: Trypsin, allow up 1 missed trypsin cleavage

site; iv) Modiﬁcations: carbamidomethyl cystein as ﬁxed modiﬁ-

cation and oxidized methionine as variable modiﬁcation; v) Peptide

mass tolerance: 50 ppm (PMF) ꢀ100 ppm (MSeMS or Combined

search); vi) Peptide charge state: þ1 MSeMS; vii) Fragments

tolerance: 0.3 Da. The parameters for the combined search (Peptide

mass ﬁngerprint and MSeMS spectra) were the same described

above. In all protein identiﬁcation, the probability scores were

greater than the score ﬁxed by MASCOT as signiﬁcant with

a p-value <0.05. Protein identiﬁcation by de novo sequencing from

fragmentation spectra of peptides was performed using de Novo

tool software (Applied Biosystems). Tentative sequences were

manually checked and validated, and the homology search for these

sequences was obtained using the BLAST tool at (http://www.ncbi.

nlm.nih.gov/BLAST).

ꢃ

otics at 38 C under 5% CO

2

in air. Flukes were then transferred to

2

a 75 cm tissue culture ﬂask (Iwaki, Sciteck Div. Asahi Techno Glass,

Funabashi City, Chiba, Japan) and mantained in culture medium

ꢃ

(

3 mL/ﬂuke) at 38 C under 5% CO

2

in air. After 24 h of incubation,

the medium containing ESAs was removed, the protease activity

inhibited with a protease inhibitor cocktail (SigmaFASTꢁ Protease

Cocktail Tablet, SigmaeAldrich, Madrid, Spain), and centrifuged at

ꢃ

10,000 g for 20 min at 4 C. Then, the supernatant was passed

through a 0.45 mm pore ﬁlter disc, and submitted to a process of

washing/concentration with distilled water using Microsep

Microconcentrators (3000 molecular weight cut-off; Filtron Tech-

nology Corporation, Northborough, Massachusetts) and the

concentrate dried using a vacuum concentrator (SpeedVac, Thermo

Scientiﬁc, Barcelona, Spain) and stored at 4 C until use.

Size-Exclusion Chromatography (SEC). Samples of ESAs of

F. hepatica were dissolved in PBS at a ﬁnal protein concentration of

ꢃ

3

.5 mg/mL and fractionated by size-exclusion chromatography

using an FPLC system (ÄKTA Basic 10, Amersham Biosciences

Europe BmbH, Barcelona, Spain) equipped with a column of

Superdex 75 HR 10/30 column (Amersham Biosciences) as

described previously [62]. Brieﬂy, 0.5 mL samples were eluted at

a ﬂow rate of 0.3 mL/min and the protein concentration was

monitored at 280 nm. The molecular weight of the eluted proteins

was estimated using a mixture of proteins of known molecular

weight (Gel Filtration LMW Calibration Kit, Amersham Biosci-

ences). Finally, the protein concentration of each peak obtained was

measured by Pirogallol Red Method (Sigma) and stored in aliquots

ꢃ

at ꢀ80 C until further analysis.

Polyacrylamide gel electrophoresis (SDS-PAGE). The proteins from

an aliquot of puriﬁed F. hepatica ESAs corresponding to peak IV (see

Results Section), obtained as described above by exclusion chro-

matography, were resolved by SDS-PAGE under reducing condi-

3. Results

tions (2.5

mg of protein per lane) using a mini-vertical gel

electrophoresis unit (Mighty Small II, Hoefer Inc., Holliston, MA).

One-dimensional electrophoresis (1DE) was performed at 200 V

constant current at 20 C for 1 h using the Laemmli (1970) buffer

3

.1. DTPs classiﬁcation models and complex network assembly

ꢃ

LDA model. The present is the ﬁrst mt-QSAR model for the

system with discontinuous slab gels made up of a 5% poly-

acrylamide stacking gel and a 10e20% linear-gradient poly-

acrylamide resolving gel, of 1 mm of thickness. 1D-SDS-PAGE

molecular weight standards (BioRad) were used to calibrate gels.

Mass Spectrometry (MS). Protein bands in the polyacrylamide

gels were stained with Imperial Protein Stain (Pierce) overnight and

the dye excess removed with distilled water. Then the selected

bands (band #2 in this study; see Results Section) were excised

using a scalpel, washed twice with water, shrunk 15 min with 100%

acetonitrile and dried in a Savant SpeedVac for 30 min. Then, the

samples were reduced with 10 mM dithioerytritol in 25 mM

probability of binding organic compounds to very large diversity of

receptors based only on the molecular connectivity of the drug and

the protein receptor. One application for the present model is

predict the protein receptor or active place with a speciﬁc drug and

vice versa, predict drugs with theirs proteins. In both cases,

receptor susceptibility identiﬁcation is imperative. Detailed infor-

mation on the compounds, predicted classiﬁcation, and probability

of afﬁnity on different receptors of the drugs used to seek the model

appears in Table 1SM of the supplementary material. Using this

model we can predict the different relationships between the

drugeprotein connectivity same physicochemical property [32].

Common physicochemical properties have been demonstrated to

be useful on protein QSAR [63,64]. This work introduces for the ﬁrst

time a single linear mt-QSAR equation model to classify drugs with

your respective protein receptor. The best model found was:

ꢃ

ammonium bicarbonate for 30 min at 56 C and subsequently

alkylated with 55 mM iodoacetamide in 25 mM ammonium

bicarbonate for 15 min in the dark. Finally, samples were digested

with 12.5 ng/ml sequencing grade trypsin (Roche Molecular

Biochemicals) in 25 mM ammonium bicarbonate (pH 8.5) over-

^S^ð^D^T^P^Þpred ¼ ꢀ8:51$^D

q

D

ꢃ

ðTÞ þ 8:23$

4^ð^T^Þ^þ²^:⁰¹^$0^ð^X^Þ

q q

night at 37 C. After digestion, the supernatant was collected and

2

1

ml was spotted onto a MALDI target plate and allowed to air-dry at

D

þ 0:19$

þ 3:78$

q

ðHetÞ þ 21:74$

0^ð^c^Þ^þ⁰^:⁵¹^$

q

2

ðH ꢀ HetÞ ꢀ 20:16$

0^ð^c^Þ^þ⁰^:³⁴^$

q

₅ðH ꢀ HetÞ

0

room temperature. Then, 0.4 l of a 3 mg/mL of -cyano-4-hydroxy-

m

a

T

D

T

q

0^ðⁱ^Þ^þ¹^:²⁵^$

q

ðiÞ

q

transcinnamic acid matrix (SigmaeAldrich) in 50% acetonitrile

were added to the dried peptide digest spots and allowed again to

air-dry at room temperature.

MALDI-TOF MS analyses were performed in a 4800 Proteomics

Analyzer MALDI-TOF/TOF mass spectrometer (Applied Biosystems,

Framingham, MA) at the Genomics and Proteomics Center, Uni-

versidad Complutense de Madrid, operated in positive reﬂector

mode, with an accelerating voltage of 20,000 V. All mass spectra

were calibrated internally using peptides from the auto digestion of

3

T

ꢀ

¼

0:45$

0:68$^T

₅ðiÞ þ 1:97$

ðmÞ þ 0:19$

q

₁ðmÞ ꢀ 4:01$

₀ðsÞ ꢀ 1:05$

¼ 2241:061 p < 0:001

q

₂ðmÞ þ 3:74$

₂ðsÞ ꢀ 1:99N

₃ðmÞ

T

q

5

2

5784

c

(

4)

In this model the N is the number of cases (DTPs) used to train the

2

model and Chi-square (

c ) is the statistic used to test the

1082

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Table 1

Detailed list of the symbols and description for all parameters present in the model.

Molecule

Symbol

Atom group

Descriptor name

D

Drug

Molecule

Target protein

q

ðTÞ

All atoms

Halogens

Heteroatoms

Hydrogens bonded to heteroatoms

Protein region

Core

Inner

Middle

Surface

Entropy of drug for all atoms at distance k ꢄ 2

D

ðTÞ

Entropy of drug for all atoms at distance k ꢄ 4

ðXÞ

Entropy of drug for halogens at distance k ¼ 0

ðHetÞ

ðH ꢀ HetÞ

Entropy of drug for heteroatoms at distance k ¼ 0

Entropy of drug for hydrogens-heteroatoms and all the neighbors at distance k ꢄ 2

Entropy of drug for hydrogens-heteroatoms and all the neighbors at distance k ꢄ 5

Descriptor name

Symbol

T

q

ðcÞ

ðiÞ

Entropy of all aminoacids in the core of the protein at distance k ¼ 0

Entropy of all aminoacids in the core of the protein at distance k ꢄ 5

Entropy of all aminoacids placed in the inner region at distance k ¼ 0

Entropy of all aminoacids placed in the inner region and all the neighbors at distance k ꢄ 3

Entropy of all aminoacids placed in the inner region and all the neighbors at distance k ꢄ 5

Entropy of all aminoacids placed in the middle region and all the neighbors at distance k ꢄ 1

Entropy of all aminoacids placed in the middle region and all the neighbors at distance k ꢄ 2

Entropy of all aminoacids placed in the middle region and all the neighbors at distance k ꢄ 3

Entropy of all aminoacids placed in the middle region and all the neighbors at distance k ꢄ 5

Entropy of all aminoacids placed in the surface region at distance k ¼ 0

Entropy of all aminoacids placed in the surface region and all the neighbors at distance k ꢄ 2

T

ðiÞ

ðmÞ

ðsÞ

signiﬁcance of DTPs/nDTP discrimination with p < 0.001 (level of

error). Signiﬁcant entropy parameters were calculated for the

totality (T) of the atoms in the molecule or for speciﬁc collections of

atoms. These collections are atoms with a common characteristic as

for instance are: halogens (X), heteroatom (Het), hydrogen atom

bonded to one heteroatom (H-Het) or protein region (protein

region). In Table 1 we report a detailed list of the symbols and

description for all parameters present in the model.

This model, with 18 variables, classiﬁes correctly 597 out of 678

DTPs (Sensitivity of 88.05%) and 2878 out of 3234 nDTP (DrugePro-

tein Pair for compounds with low connectivity) (Speciﬁcity of 89%).

Overall training Accuracy was 88.83%. The validation of the model

was carried out by means of external predicting series. The model

classiﬁes correctly 298 out of 338 DTPs (88.17%) and 1359 out of 1534

nDTPs (88.59%) in validation series. Accuracy for validation series

under curve higher than 0.5. The model presented an area greater

than 0.97. The vitality of this type of procedures developing ANN-

QSAR models has been demonstrated before; see, for instance, the

works of Fernandez and Caballero [67,68]. The same is true about

the other kinds of ANNs tested.

Complex networks assembly. Two possible applications for the

present model are the bio-molecular screening of drug afﬁnity to

different proteins and the construction of multi-protein afﬁnity

proﬁle networks for drugs. In order to recall the capacity of the mt-

QSAR to predict new CNs we selected a database of recently assayed

drugs with their respectively Proteins. With these goals in mind, we

constructed ﬁrst a new observed DrugeProtein (DP) DP-CN,

obtaining a CN with 855 vertices and 1016 DP (edges) an average

distance equal to 6.66. The same as before, we constructed a new

predicted DP-CN obtaining an average distance equal to 5.47 and

1298 DP (edges) see Fig. 6. In this, we illustrate visually both

observed DP-CN and predicted DP-CN. The numeric labels of the

nodes identify the different inputs (DRPs) used in the analysis.

In Table 3 we show the results of calculating the functions of

normal and exponential distribution for proteins and drugs

observed and predicted presented in dataset. It is seen in that the

protein has a difference (d) biggest on normal function to the

(

predictability) was 88.51% (1399 out of 1872 DTPs). These results

Table 2) indicate that we developed an accurate model according to

previous reports on the use of LDA in QSAR [65,66].

ANN models. The present model shows good results with

a relatively small number of parameters (18 parameters) and

a linear equation. To show how important is this result, we

compared the present model with other models used to address the

same problem. We processed our data with different Artiﬁcial

Neural Networks (ANNs) looking for a better model. Four types of

ANNs were used, namely, Probabilistic Neural Network (PNN),

Radial Basic Function (RBF), Three Layers Perceptron (MLP-3), and

Four Layer Perceptron (MLP-4). The Fig. 3 depicts the networks

maps for some of the ANN models tested. In general, at least one

ANN of every type tested was statically signiﬁcant. However, one

must note that the proﬁles of each network indicate that these are

highly nonlinear and complicated models.

Table 2

Comparison of LDA and different ANNs classiﬁcation models.

Model proﬁle

Class Train

DTPs nDTPs

DTPs 2878 356

Stat. Par. Validation

%

DTPs nDTPs

LDA

17:17e1:1

89.0 Sn

88.1 Sp

88.8 Ac

89.4 Sn

88.6 1359 175

88.2 40

88.5

88.5 299 39

87.4 193 1341

87.6

nDTPs 81

Total

597

298

LNN

DTPs 606 72

nDTPs 359 2875 88.9 Sp

Total

One network found was MLP and it showed training perfor-

mance higher than 92%. We compare different types of networks to

obtain a better model; Table 2 shows the classiﬁcation matrix of the

different networks. Was taken as the main network (MLP

64:64e1:1

89

Ac

PNN

DTPs 26

652

3.8 Sn

3.6 12

326

65:65e3912e2e2:1 nDTPs

0

3234 100 Sp

83.3 Ac

100

82.6

0

1534

Total

3

2:32e15e1:1) because it presents a wider range of variables,

MLP

DTPs 602 76

88.8 Sn

88.8 300 38

88.9 171 1363

88.8

presents 32 inputs in the ﬁrst layer and 32 neurons in second layer,

two sets of cases (Training and Validation). Another tested

networks found were MLP 39:39e20e21e1:1, RBF 1:1e1e1:1

presents the same type of variables; Linear 64:64e1:1 present

many variables and PNN 65:65e3912e2e2:1 has a very low

percentage of DTPs leading to possible errors in the model although

your accuracy is very well, see Table 2. We depict the ROC-curve for

MLP 32:32e15e1:1 to show how reliable was the network model

developed, see Fig. 4. Notably, almost the model presented was

39:39e20e21e1:1 nDTPs 354 2880 89.1 Sp

Total

89

Ac

RBF

DTPs 369 309

nDTPs 1502 1732 53.6 Sp

Total

DTPs 623 55

54.4 Sn

52.4 177 161

52.6 727 807

52.6

92.6 313 25

92.0 123 1411

92.1

1:1e1e1:1

53.7 Ac

91.9 Sn

MLP

MLP 32:32e15e1:1 nDTPs 239 2995 92.6 Sp

Total 92.5 Ac

DTPs: Drug-target pairs for compounds with high afﬁnity; nDTPs: Drug-target pair

for compounds with non-afﬁnity.

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1083

Fig. 3. Topology of some ANN models trained in this work.

exponential function, so that we can conclude that our CN tends to

be an exponential distribution. To illustrate these results a node

degree distribution for both observed and predicted DPs were

performed, see Fig. 7. In this ﬁgure we can see that the drugs follow

a normal distribution, while proteins follow an exponential

distribution.

In Table 4 we show the results of closeness centrality and the

number of node degree for proteins and drugs used in the database.

Fig. 4. ROC curve for classiﬁer.

1084

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Centrality closeness measures how many steps are required to

access every other given vertex from a vertex. An interesting result

that can be seen in the table is the protein with a closeness 1HOF

equal to 20 and a higher degree node equal to 51. This means that

many drugs in the CN have interaction and the node degree is high,

but these drugs are selective for this protein and that within the CN

have the lowest closeness. This is a very interesting result; the

protein 1HOF is a G protein [69]. G proteins are important signal

transducing molecules in cells. In fact, diseases such as diabetes,

blindness, allergies, depression, cardiovascular defects, and certain

forms of cancer, among other pathologies, are thought to arise due

to derangement of G protein signaling [70,71]. The drugs that

interact with 1HOF can be, in according to these results can be more

selective for a particular disease and cause fewer adverse reactions

Fig. 5. Synthesis of rasagiline derivates obtained in this work.

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1085

Fig. 6. Observed vs. predicted drug-target complex networks.

1086

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Table 3

against illness before descript. On the contrary this protein 1A8M

Study of network distribution functions.

closeness to present a very low node degree with a low closeness,

which means that drugs that interact with 1A8M interacts with

other proteins and which mean the drugs that interact with 1A8M

are not selective, in this case there are a variety of drugs that can

interact with protein see Table 4.

In the case of drugs as an example is the Moexipril. This drug is

a ACE inhibitor that can be for the treatment of hypertension [72].

Moexipril presents a low closeness equal to 29 with a high node

degree 37, this means that the drug is selective for proteins to

which is connected to the CN. These results are consistent with the

literature [72e74]; Moexipril is a long-acting, non-sulfhydryl

angiotensine-converting enzyme inhibitor (ACE). On the other

Sub-network

Proteins

Network

Observed

Distribution

d

p

Fit

Best ﬁt

Normal

Exponential

Normal

Exponential

Normal

Exponential

Normal

0.46

0.30

0.39

0.27

0.35

0.40

0.34

0.40

0.01

No

Yes

No

Yes

No

Yes

No

Predicted

Observed

Predicted

Drugs

Exponential

Fig. 7. Distribution functions for complex networks sub-sets.

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1087

hand Acetazolamide, that is a carbonic anhydrase inhibitor that is

used to treat glaucoma, epileptic seizures, benign intracranial

hypertension (pseudotumor cerebri), altitude sickness, cystinuria,

and dural ectasia [75,76], acetozolamide submit a node degree

equal ¼ 5 with a closeness ¼ 22, which indicates that this drug

interacts with ﬁve proteins these proteins only interact with

Acetazolamide; there are speciﬁc to this drug.

chromatography achieving clean separation of these mixtures: 6a

(24%) and 7a (36%), and 6b (13%) and 7b (753%). Finally the

compounds 7a and 7b were converted in goods yields, to the cor-

responding acetyl esters 8a and 8b, and benzoyl esters 9a and 9b,

3

treatment with acetic anhydride in Et N [83] and with benzoyl

chloride in Et N and acetonitrile [84] respectively.

3

Prediction of rasagiline analogs vs. MAO proteins. In this in silico

experiment we used the 3D structure MAO-A and B proteins with

PDB ID 2Z5X and 2VZ2 and the model MLP 32:32e15e1:1 to

predict the scores. We also generated the SMILE codes for these

compounds and predicted their propensity to form DTPs with

MAO-A and MAO-B using the model. In Table 5 we confront the

results obtained using this model and the outcomes of the phar-

macological assay. The compound Rasagiline a known selective

inhibitor for MAO-B was used as control. We consider the observed

3

.2. Illustrative experiments

.2.1. Study of rasagiline analogs (experiment 1)

Synthesis, characterization, and assay. The preparation of (ꢂ)-cis-

and trans-amino alcohols 5a,5b and their N-propargyl derivatives

6

a,6b and N,N-dipropargyl derivatives 7a,7b has been reported see

Fig. 5. In short, the amino alcohols 5a,5b were prepared from the

class for active compounds OC ¼ 1 if compound IC50 < 50

cut-off is in the similar range than other used in previous works

36,37]. As we can see in this table the only one active compound in

mM this

2

4

,2,2-triﬂuoro-N-(6-oxo-5,6-dihydro-4H-cyclopenta[b]thiophen-

-yl)acetamide, 3 that were synthesized by intramolecular cycli-

[

zation of 3-amino-3-(thiophen-3-yl)propanoic acid, 2 following

previously described pathways [77]; thus the -aminoacid 2 were

the pharmacological assay (OC ¼ 1) was compound 8b predicted

also as active with PC ¼ 1 and high score S(DTP)pred ¼ 0.64. Two

other compounds (5b and 6b) that are inactive (both as MAO A and

MAO B inhibitors) in pharmacological assays (OC ¼ 0) were also

predicted as inactive against MAO A (PC ¼ 0) but predicted as active

for MAO B (PC ¼ 1). In this case the model fails, however the

prediction have a very low activity scores S(DTP)pred ¼ 0.15.

b

obtained by treatment of thiophen-3-carbaldehide with ammo-

nium acetate and malonic acid in reﬂuxing ethanol [78]. The

treatment of the carboxylic acid 2 with a mixture of boiling tri-

ﬂuoroacetic acid and anhydride gave in one step the oxotri-

ﬂuoroacetamide derivative 3, whose melting point and

spectrophotometry infrared has concordat with the literature

Prediction of rasagiline analogs vs. US FDA drug-target proteins. An

additional use of the model was to predict the activity of the new

compounds with respect to all other targets previously approved by

US FDA [85]. At the same time we can use this model to predict the

selectivity of the new rasagiline derivative 8a as MAO B inhibitor

with respect to all FDA drugs targets and predict possible toxico-

logical effects depending on the other targets predicted for these

compounds. This type of experiment is of the major importance

due to the cost in terms of animal sacriﬁce, time, materials and

human resources of the experimental assay of all compounds

against all these targets, see recent reviews by Duardo-Sanchez

et al. [86e89]. In fact, over a decade, the US FDA has been engaged

in the applied research, development, and evaluation of computa-

tional toxicology methods used to support the safety evaluation of

a diverse set of regulated products. The basis for evaluating

computational toxicology methods is multi-factorial, including the

potential for increased efﬁciency, reduction in the numbers of

animals used, lower costs, and the need to explore emerging

[

4

77,79]. The reaction of the oxoamide 3 with NaBH

days at room temperature (molar ratio 3/NaBH

4

in MeOH during

, 1/5) [80,81]

4

produced, after chromatographic puriﬁcation, an 86% yield of

a mixture of cis/trans-5a/5b epimers, however attempts of sepa-

ration of the mixture were unsuccessful, leading only to partial

resolution of epimers. Alternatively, the compound 3 were con-

verted to the mixture of cis/trans-4a/4b epimers (83% yield) by

treatment with NaBH

temperature (molar ratio 3/NaBH

lution of those by ﬂash column chromatography was efﬁcient

4a:32%, 4b:31%), so the needed separation of the corresponding

isomers was performed at this stage of the synthetic route. The

triﬂuoroacetamides hydrolysis of these compounds with K CO in

MeOH/H O gave the corresponding amino alcohols 5a and 5b with

yields of the 99% and 75% respectively. Propargylation of these

compounds with propargyl bromide and K CO in acetonitrile [82]

afforded the corresponding mixtures cis and trans of mono- and

4

in methanol during 10 min at room

4

, 1/2) and the subsequent reso-

(

2

3

2

3

dipropargyl derivatives

6a/7a

and

6b/7b.

Subsequent

Table 4

Results of node degree and closeness centrality for top-20 proteins and drugs.

Rank

PDB

Function

-2A adrenergic receptor

DNA

C

deg

C

clo

Drugs

Activity

C

deg

C

clo

1

2

3

4

5

6

7

8

9

1HOF

1BNA

1HA2

1QYX

1UWJ

1EMI

1CZM

1MLD

1P49

1BYW

1T40

1F8U

1N7D

1SLM

1EXX

1E3G

1VJB

a

51

49

47

31

22

20

19

18

16

15

14

13

12

11

10

20.3

Moexipril

Sertraline

Levamisole

Adenine

ACE inhibitor

SSRI class

Anthelminthic

Biochemistry metabolism

Atrial ﬁbrilation

37

18

16

15

11

10

8

7

6

5

29.4

34.5

27.1

27.6

32.3

31.6

24.9

27.9

27.3

21.9

23.9

25.4

21.8

29.5

21.9

31.2

20.3

25.3

22.8

21.9

33.49

35.29

27.89

26.64

23.67

26.77

31.47

26.98

27.62

25.2

23.33

22.7

21.1

25.5

27.3

Serum albumin

Estradiol 17-

b

-dehydrogenase 1

B-Raf proto-oncogene

23S rRNA

Carbonic anhydrase 1

Malate dehydrogenase

Digoxin

Atorvastatin

Pyrazinamide

ADP

Mibefradil

Alitretinoin

L-Aspartic acid

Megestrol

Vinblastine

Acitretin

Inhibitor of HMG-coa reductase

FAS inhibitor

AMP/IMP catalysis

2þ

Steryl-sulfatase

Ca channel blocker

Antineoplasic

þ

10

11

12

13

14

15

16

17

18

19

20

K

voltage-gated H member 2

Aldose reductase

Stimulates NMDA receptors

Antineoplasic

Psoriasis inhibitor

Analgesic

Antimalaria

Vasodilatation

Catalyst

Carbonic anhydrase inhibitor

Anti-acne

Acetylcholinesterase

Low-density lipoprotein receptor

Stromelysin-1

g-1 Retinoic acid receptor

Etorphine

Halofantrine

Androgen receptor

Estrogen-related receptor

Glutamate carboxypeptidase 2

Tumor necrosis factor

Cytochrome c oxidase subunit 1

g

23.4

24.7

30.9

26

L

-Arginine

1Z8L

1A8M

1OCZ

L

-Ornithine

Acetazolamide

Adapalene

1088

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Table 5

Prediction of rasagiline analogues with the new model.

MAO-B

Compound

Structure

MAO-A

IC50

Drug^a

IC50^b

OC

PC

S(DTP)pred

0.70

OC

0

PC

0

S(DTP)pred

0.97

NH₂

5a

>100

0

>100

S

OH

NH₂

5

b

>100

0

1

0

0.15

0.70

0

0.88

0.97

S

OH

HN

6a

>100

S

OH

HN

6b

>100

0

1

0.15

>100

0

0.88

S

OH

N

7

8

a

b

a

b

>100

46.25

0

1

0

1

0.94

0.92

0.64

>100

0

0.99

0.98

0.96

S

OH

N

S

OH

OAc

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1089

Table 5 (continued ).

MAO-B

IC50^b

Compound

Structure

MAO-A

IC50

Drug^a

OC

0

PC

0

S(DTP)pred

0.94

OC

0

PC

0

S(DTP)pred

N

9

a

>100

0.99

S

OBz

N

b

>100

0

1

0

1

0.92

0.14

>100

0

1

0

0.98

0.87

S

Rasagiline

0.412 ꢂ 0.04

0.0443 ꢂ 0.009

HN

a

Rasagiline was used as positive control.

b

>100 ¼ compound inactive at 100

m

M (highest concentration tested), OC ¼ observed class and PC ¼ Predicted class, OC ¼ 1 if compound IC50 < 50 M and PC ¼ 1 if the DTP

m

probability predicted for pair drug-MAO-i enzyme p(MAO-i) > 0.5 (2Z5X and 2VZ2 are PDB ID of MAO-A and B used to predict p-values).

technologies that support the goals of the US FDA’s Critical Path

Initiative (e.g. to make decision support information available early

in the drug review process). The US FDA’s efforts have been facili-

tated by agency-approved data-sharing agreements between

government and commercial software developers [90].

For this we used the model to calculate DTPs scores for our

compounds (rasagiline derivatives) vs. FDA approved targets. We

depict in Table 4SM, all proteins in FDA dataset predicted vs. the 10

rasagiline derivatives. We found that overall the 10 derivatives

were predicted negative against almost all proteins in the FDA

database. In Table 6 we depict some results, which show that our

rasagiline derivative 8b is predicted as selective MAO B inhibitor,

because it have only interaction with MAO-B, and have not pre-

dicted interaction with the proteins used in the database.

Using these results (depicted in Table 4SM), we constructed

a DP-CN for rasagiline derivatives and the FDA dataset (see Fig. 8).

As a result we obtained a CN with 669 nodes (FDA drugs, proteins,

or rasagiline derivatives) and 839 DP (edges, DTPs). As In this

network we can see that protein 2BK3 (MAO-B) is predicted to

interact with compounds 5b, 6b and 8b, this protein is a known

Fig. 8. Drug-target sub-network obtained after coupling the real network for FDA drugs and proteins with new rasagiline derivates (based on QSAR prediction with MLP model).

1090

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

Fig. 9. ESAs of F. hepatica: (A) Size-exclusion chromatogram obtained by FPLC separation of ESAs from F. hepatica. The arrow indicates peak number IV of the chromatogram

according to Mezo et al. [62]; (B) SDS-PAGE analysis of proteins contained in peak number IV (Lane b). The proteins present in band 2 were analyzed by MS and MS/MS spec-

trometry. The relative molecular masses of standard markers (lane a) from top to bottom were: 170, 130, 95, 72, 55, 43, 34, 26, 17 and 10 kDa.

rasagiline target [94,95]. These results are satisfactory because they

agree with the experimental results presented in this paper where

the compound 8b show MAO-B activity. The use of such complex

networks can help us ﬁnd and predict new drugseprotein inter-

actions, and therefore ﬁnd new drugs with improved biological

activity and fewer side effects, especially in parasite disease.

The selected band (band number 2 in this study) was excised

and the peptides were analyzed by MALDI-TOF MS and MS/MS.

Once we obtained the data from MALDI-TOF MS analysis of the

query band, the most relevant MS signals were introduced into the

MASCOT search engine [91,92]. We obtained 20 hits (template

proteins) for this protein with MASCOT scores (M

s

) higher than 81

(p < 0.05), the threshold value for signiﬁcant match, see Table 7. The

3

.2.2. Study of peptidome for Fasciola hemoglobin protein

maximum scores obtained were 286 and 283, which corresponded

to proteins gij196049684 (16 550 Da) and gij159461074 (16 681 Da)

of F. hepatica, both are fHb proteins annotated as: Fasciola Chain

A hemoglobin (Hb2) and Fasciola hemoglobin F2, respectively.

We provide in Table 8, detailed information on the results of the

MS and MASCOT search engine for band number 2. This table

includes the 9 most interesting peptides matching with the fHb

sequence. We found an excellent match between the mass of the

peptide detected and the mass of the template peptide recorded in

MASCOT database with known sequence for all these peptides.

After that, we decided to investigate the structureefunction rela-

tionships for all sequences of the 9 fHb peptides found on the PMF

(

experiment 2)

SEC, 1DE, PMF and MS/MS study of peptides found on Fasciola

hemoglobin (fHb). In this section we present an example of the

practical use of the QSAR model to predict enzyme scores for

peptides found in the PMF and MS/MS study of a new query protein.

We illustrate an overall view of SEC and 1D electrophoresis study of

F. hepatica proteome carried out in this work, see Fig. 9. In this ﬁgure,

we label the bands obtained after 1DE including band number 2.

Table 6

Some illustrative scores obtained in experiments 1 and 2.

Table 7

Prediction of rasagiline analogues (experiment 1)

Targets^a

Top-20 template proteins in F. hepatica found by MASCOT search.

Drugs codes

PDB

PC

Score

Protein Accession

Score Function

8

5

9

5

9

5

7

b

a

b

a

b

a

b

a

b

2VZ2

1SD2

1E18

1A8M

1AGS

1BWC

1BXS

1D3H

1

0

0.64

0.74

1.00

0.93

1.00

MAO-B

MTA phosphorylase

1

gij196049684

16,550 286

Chain A, hemoglobin (Hb2)

Fasciola hepatica

Hemoglobin F2 Fasciola hepatica

Fatty acid-binding protein type 3

D,D-heptose 1,7-bisphosphate

2

3

4

gij159461074

gij47116941

gij209964147

16,681 283

14,671

20742

79

69

phosphatase 1

5

6

7

8

gij73539911

gij53713210

gij162149431

gij156844632

gij51245092

gij188590974

69,948

34,431

56,150

52,531

32,033

69,946

68

64

62

61

60

59

Thiamine biosynthesis protein thic

Putative transcription regulator

Hypothetical protein GDI3669

Hypothetical protein Kpol-1058p57

Cysteine synthase A

Thiamin biosynthesis protein

Hypothetical protein NCU00551

Myosin heavy chain

Hypothetical protein HNE_2630

Hypothetical protein gll2634

COG0673 dehydrogenases

Cytochrome P450 52A10

Peptidase M16 domain-containing

protein

Neutrophil cytosolic factor 4

Hypothetical protein OTT_1626

Competence protein F

Prediction of peptides (experiment 2)

fHb^b

9

Drugs name

Amphetamine

Aprotinin

Peptide

p1

p6

p16

p14

p5

p22

p29

p20

p5

PC

1

Score

0.82

0.81

0.78

0.57

0.83

0.58

0.89

0.58

0.55

10

11

12

13

14

15

16

17

Yes

gij164428423 274,657

gij58268372

gij114800067

gij37522203

gij46203356

gij3913328

88,612

52,676

19,619

41,788

59,940

46,771

Cisplatin

Debrisoquin

Hexachlorophene

Pentamidine

Phentermine

Propofol

gij146278964

Rasagiline

1

18

1

2

gij6679022

38,881

29,079

29,137

59

58

9

0

gij189184533

gij148284486

a

We give only the function of positive targets.

fHb ¼ yes means that peptide may be present in hemoglobin of F. hepatica (fHb).

b

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1091

Table 8

MASCOT study of hemoglobin peptidome in Fasciola hepatica.

Table 9

MM/MD study of hemoglobin peptidome in Fasciola hepatica.

Pept Sequences

fHb^aObserved Mr(expt) Mr(calc) Delta

Input

Pept.

MM/MD

ACCR

p1

p2

p3

p4

p5

p6

p7

p8

p9

kaasnpsvleeri

kaasnpsvleerivqgakd

karpvtkdqftgaapiﬁkf

kavnnyhkv

kcpentthvvre

kdnvgqsegiry

kdqftgaapiﬁkf

kdsdskisqvqkc

kffqgllkkq

Yes 1172.59 1171.58 1171.58 ꢀ0.002

Yes 1768.80 1767.79 1767.95 ꢀ0.1577

Yes 1960.03 1959.02 1959.09 ꢀ0.0745

_f_H_ba

sequences

EPOT

DEPOT

p1

p2

p3

p4

p5

p6

p7

p8

kaasnpsvleeri

Yes

0.62

0.60

0.47

0.67

0.65

0.63

0.61

0.63

0.62

0.65

0.61

0.62

0.64

0.60

0.63

0.64

0.66

0.64

0.62

0.65

0.63

0.64

0.62

0.63

0.60

0.64

0.59

0.60

ꢀ224.29

ꢀ237.99

288.65

24.38

26.91

35.00

61.95

34.27

24.15

21.92

27.43

36.88

34.29

25.75

15.87

23.68

25.73

33.40

30.24

26.48

19.80

24.13

25.78

35.22

38.69

19.04

35.68

37.46

33.44

18.94

28.09

34.40

25.37

22.92

kaasnpsvleerivqgakd

karpvtkdqftgaapiﬁkf

kavnnyhkv

kcpentthvvre

kdnvgqsegiry

kdqftgaapiﬁkf

kdsdskisqvqkc

kffqgllkkq

kﬂlhvmqaiaakm

kiaahaadlakg

944.58

943.58

943.49

0.0883

1212.56 1211.56 1211.57 ꢀ0.0154

Yes 1074.56 1073.55 1073.51 0.0414

18.15

Yes 1307.68 1306.68 1306.69 ꢀ0.0168

1234.56 1233.56 1233.62 ꢀ0.0631

Yes

ꢀ26.93

55.90

Yes

980.60

979.60

979.59

0.0119

ꢀ152.86

56.88

p10 kﬂlhvmqaiaakm

p11 kiaahaadlakg

p12 kiahfcsmcgpkf

Yes 1504.70 1503.70 1503.79 ꢀ0.0967

p9

Yes

980.60

979.60

979.55

0.0523

0.0778

p10

p11

p12

p13

p14

p15

p16

p17

p18

p19

p20

p21

p22

p23

p24

p25

p26

p27

p28

p29

p30

ꢀ118.13

ꢀ34.73

ꢀ139.91

ꢀ130.95

ꢀ5.93

1307.65 1306.64 1306.56

p13 kleqsenmdavlqkl

p14 klitsskpeitftlegnkm

p15 kllndhgyfvfvvtnqsgvarg

p16 klqgltkdnvgqsegiry

p17 kmiatvtvgdvka

p18 kmiatvtvgdvkavnnyhkv

p19 ktlfaahpeyisyfskl

p20 kttvisftfgeefkeetadgrt

p21 kwclahhke

1504.70 1503.70 1503.72 ꢀ0.0264

kiahfcsmcgpkf

1877.92 1876.91 1877.01 ꢀ0.0996

kleqsenmdavlqkl

klitsskpeitftlegnkm

kllndhgyfvfvvtnqsgvarg

klqgltkdnvgqsegiry

kmiatvtvgdvka

kmiatvtvgdvkavnnyhkv

ktlfaahpeyisyfskl

kttvisftfgeefkeetadgrt

kwclahhke

2236.24 2235.23 2235.14

0.0853

Yes 1714.81 1713.80 1713.90 ꢀ0.099

Yes 1133.61 1132.60 1132.62 ꢀ0.0113

1960.03 1959.02 1959.02 ꢀ0.0051

1773.79 1772.78 1772.88 ꢀ0.0987

2264.07 2263.06 2263.06 ꢀ0.0009

ꢀ69.69

ꢀ203.86

49.97

Yes

86.49

ꢀ66.69

951.55

950.54

50.45

0.0886

0.0125

19.60

p22 rcrkpepgmlldlcdrw

p23 rdawrgaaﬂdrd

p24 kwclahhke

1859.90 1858.90 1858.89

ꢀ145.51

ꢀ55.35

ꢀ38.58

ꢀ133.69

ꢀ209.57

76.47

ꢀ208.79

ꢀ168.13

ꢀ362.27

ꢀ253.95

1277.61 1276.61 1276.63 ꢀ0.0246

rcrkpepgmlldlcdrw

rdawrgaaﬂdrd

kwclahhke

1114.61 1113.61 1113.58

1114.61 1113.605 1113.57

2225.17 2224.16 2224.11

0.0278

0.0456

p25 rdlqaaeaagirg

p26 rgaaﬂdrdgvlnidhgyvhrr

p27 rghlftggdlsefvgallar

p28 rklitsskpeitftlegnkm

p29 rkpepgmlldlcdrwpvdr

p30 rredvewiqgaitavkl

rdlqaaeaagirg

1960.03 1959.02 1959.02 ꢀ0.0017

2006.10 2005.10 2005.11 ꢀ0.0135

2196.99 2195.99 2196.08 ꢀ0.0965

1714.81 1713.80 1713.90 ꢀ0.103

rgaaﬂdrdgvlnidhgyvhrr

rghlftggdlsefvgallar

rklitsskpeitftlegnkm

rkpepgmlldlcdrwpvdr

rredvewiqgaitavkl

a

fHb ¼ yes means that peptide may be present in hemoglobin of F. hepatica (fHb).

a

fHb ¼ yes means that peptide may be present in hemoglobin of F. hepatica (fHb).

of the new protein and other peptides reported by MASCOT as well

(

30 peptides in total).

MM/MC study of peptides found on new protein PMF. Our main

contribution to the ligand interaction. This may allow us to select

peptides for drug design and/or obtain information for drug-target

interest on the study of the peptides in the PMF of the new

discovery. We therefore have to calculate the

k

q for all peptides and

unknown proteins is to ﬁnd which of them make a positive

substitute these values in the QSAR model to predict ligand

Fig. 10. Drug-target sub-network for FDA drugs vs. 30 peptides of different proteins including 9 peptides of fHb.

1092

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

interaction score for one ligand (levulinic acid). For this, we ﬁrst

need the 3D structures of the peptides in order to calculate the

University of Santiago de Compostela from Angeles Alvariño, Xunta

de Galicia. N. Alonso thanks sponsorships for research position at

the University of Santiago de Compostela from FPU program, Xunta

de Galicia. The present study was partially supported by grants

FAU2006-00021-C03-00 and AGL2010-22290-C03-01 (Ministerio

de Ciencia e Innovación, Spain), 07CSA008203PR (Xunta de Galicia,

Spain) and by the European Fund for Regional Development

(FEDER). The authors also would like to dedicate the present work

to Prof. F. Orallo (in memoriam) by his kind attention, support, and

friendship.

q

k

values. For this study we used the same 30 peptides found by PMF

of the new protein (9 peptides from fHb and the rest from other

proteins). Unfortunately, we only have the sequences of the

peptides but not the 3D structures. We therefore ﬁrst obtained the

optimal 3D folded structures by use of an MM geometry optimi-

zation for the 30 peptides (see Fig. 2). We complemented the MM

by MC search in order to explore alternative geometrical structures

for the peptides. We summarized the results of MC simulation of

these peptides in Table 9. In this table we reported the initial energy

(

E

0

) based on the starting structure constructed with standard

parameters for -helixes (bond distances, angles, and dihedral

angles) set as default on the sequence editor of Hyperchem [59,60].

We also reported the (E ) obtained after optimization of the

structure with AMBER force ﬁeld obtained by MC method using

000 steps for 30 peptides. Lastly, we report the ACCR values for the

Appendix. Supplementary material

a

Supplementary data related to this article can be found online at

doi:10.1016/j.ejmech.2011.01.023.

1

MDTof the 30 peptides in Table 9. In the MD study most researchers

tend to try for an average ACCR value around 0.5; smaller values

may be appropriate when longer runs are acceptable and more

extensive sampling is necessary. In the present study all the ACCR

values were between 0.47 and 0.66 because MC simulation has

been realized by 1000 steps; in consequence, we can accept the MD

results as valid [59,60].

Assemble of drug-target binding site network. We use our best

model to predict whether the FDA drugs used in the database have

interaction or afﬁnity with the 30 peptides, but we put special

interest on the 9 peptides from fHb. The reason for this interest in

fHb peptides is due to the work of Dewilde et al. [93]; which

described for the ﬁrst time fHb, a potential immunogen, in the

search for an effective vaccine. Here mt-QSAR models in combi-

nation with complex networks may be helpful to predict possible

drugs that interact with fHb effectively with fewer adverse reac-

tions. In this sense, we assembled drug-target binding site network

with the predicted probabilities of binding of organic compounds

to 30 peptides of different proteins found with MASCOT (including

References

[1] Y. Yamanishi, M. Araki, A. Gutteridge, W. Honda, M. Kanehisa, Prediction of

drug-target interaction networks from the integration of chemical and

genomic spaces, Bioinformatics 24 (2008) i232e240.

[

2] A. Giuliani, Collective motions and speciﬁc effectors: a statistical mechanics

perspective on biological regulation, BMC Genomics 11 (Suppl. 1) (2010) S2.

3] P.K. Dhar, A. Giuliani, Laws of biology: why so few? Syst. Biol. Synth. 4 (2010)

7

e13.

4] S. Bornholdt, H.G. Schuster, Handbook of Graphs and Complex Networks:

From the Genome to the Internet. WILEY-VCH GmbH & CO. KGa., Wheinheim,

2003.

[

5] E. Estrada, Virtual identiﬁcation of essential proteins within the protein

interaction network of yeast, Proteomics 6 (2006) 35e40.

6] E. Estrada, Protein bipartivity and essentiality in the yeast proteineprotein

interaction network, J. Proteome Res. 5 (2006) 2177e2184.

7] A. Réka, A.-L. Barabasi, Statistical mechanics of complex networks, Rev. Mod.

Phys. 74 (2002) 47e97.

8] A.L. Barabasi, Z.N. Oltvai, Network biology: understanding the cell’s functional

organization, Nat. Rev. Genet. 5 (2004) 101e113.

9] A.L. Barabasi, Network medicineefrom obesity to the “diseasome”, N. Engl. J.

Med. 357 (2007) 404e407.

[

10] H. González-Díaz, S. Vilar, L. Santana, E. Uriarte, Medicinal chemistry and

bioinformatics e current trends in drugs discovery with networks topological

indices, Curr. Top. Med. Chem. 7 (2007) 1025e1039.

9

peptides of fHb). In so doing, we used the MARCH-INSIDE 2.0

11] A. Giuliani, L. Di Paola, R. Setola, Proteins as networks: a mesoscopic approach

software to calculate the values for the 30 peptides. We

q

k

using haemoglobin molecule as case study, Curr. Proteomics

6 (2009)

substituted these values in the model and predicted the probability

with the 30 peptides interact with FDA drugs. In bottom of Table 6

235e245.

[12] A. Krishnan, J.P. Zbilut, M. Tomita, A. Giuliani, Proteins as networks: usefulness

of graph theory in protein science, Curr. Protein Pept. Sci. 9 (2008) 28e38.

(

experiment 2), we show some results used to assemble the drug-

[

13] A. Krishnan, A. Giuliani, J.P. Zbilut, M. Tomita, Implications from a network-

based topological analysis of ubiquitin unfolding simulations, PLoS ONE 3

(2008) e2149.

14] M.C. Palumbo, A. Colosimo, A. Giuliani, L. Farina, Essentiality is an emergent

property of metabolic network wiring, FEBS Lett. 581 (2007) 2485e2489.

15] A. Krishnan, A. Giuliani, J.P. Zbilut, M. Tomita, Network scaling invariants help

to elucidate basic topological principles of proteins, J. Proteome Res. 6 (2007)

target binding site network. We constructed a new observed

DrugeProtein DP-CN, obtaining a CN with 70 vertices (peptides or

drugs) and 321 DP (edges, peptideedrug pairs) see Fig. 10.

[

4

. Conclusions

3924e3934.

[

16] A. Krishnan, A. Giuliani, M. Tomita, Indeterminacy of reverse engineering of

gene regulatory networks: the curse of gene elasticity, PLoS ONE 2 (2007)

e562.

Combining entropy parameters, calculated with MARCH-INSIDE

approach to codify information of drug and target structure, and

ANNs is possible to seek one mt-QSAR classiﬁer to predict the

probability of drugs to bind more than 500 different drug-target

proteins approved by FDA of USA with Accuracy >90%. This ANN

classiﬁer is useful to carry out Data Mining of PDB in order to

discovery new drug targets for any drug. We can also predict the

interactions with speciﬁc drugs of peptides or protein 3D structures

generated by MM/MD modeling.

[

17] K. Tun, P.K. Dhar, M.C. Palumbo, A. Giuliani, Metabolic pathways variability

and sequence/networks comparisons, BMC Bioinf. 7 (2006) 24.

18] L.G. Pérez-Montoto, F. Prado-Prado, F.M. Ubeira, H. González-Díaz, Study of

parasitic infections, cancer, and other diseases with mass-spectrometry and

quantitative proteome-disease relationships, Curr. Proteomics

46e261.

6 (2009)

2

[19] H. González-Díaz, Y. González-Díaz, L. Santana, F.M. Ubeira, E. Uriarte, Pro-

teomics, networks and connectivity indices, Proteomics 8 (2008) 750e778.

[20] J. Caballero, M. Fernandez, Artiﬁcial neural networks from MATLAB in

medicinal chemistry. Bayesian-regularized genetic neural networks (BRGNN):

application to the prediction of the antagonistic activity against human

Acknowledgments

platelet thrombin receptor (PAR-1), Curr. Top. Med. Chem.

580e1605.

8 (2008)

1

[

21] L. Fernández, J. Caballero, J.I. Abreu, M. Fernández, Aminoacid sequence

autocorrelation vectors and Bayesian-regularized genetic neural networks for

modeling protein conformational stability: gene v protein mutants, Proteins

67 (2007) 834e852.

22] M. Fernández, J. Caballero, L. Fernández, J.I. Abreu, M. Garriga, Protein radial

distribution function (P-RDF) and Bayesian-regularized genetic neural

networks for modeling protein conformational stability: chymotrypsin

inhibitor 2 mutants, J. Mol. Graph. Model. 26 (2007) 748e759.

The authors appreciate the technical assistance and advice of

Lola Gutiérrez, from the Genomics and Proteomics Center, Uni-

versidad Complutense de Madrid. H. González-Díaz thanks spon-

sorships for a research position at the University of Santiago de

Compostela from the Isidro Parga Pondal Program, Xunta de Galicia.

F. Prado-Prado thanks sponsorships for research position at the

[

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

1093

[

23] M. Fernández, F. Caballero, L. Fernández, J.I. Abreu, G. Acosta, Classiﬁcation of

conformational stability of protein mutants from 3D pseudo-folding graph

representation of protein sequences using support vector machines, Proteins

quantitative proteomeedisease relationships (QPDRs) and predicting prostate

cancer, Biochem. Biophys. Res. Commun. 372 (2008) 320e325.

[45] H. Gonzalez-Diaz, M.A. Dea-Ayuela, L.G. Perez-Montoto, F.J. Prado-Prado,

G. Aguero-Chapin, F. Bolas-Fernandez, R.I. Vazquez-Padron, F.M. Ubeira, QSAR

for RNases and theoretic-experimental study of molecular diversity on

peptide mass ﬁngerprints of a new Leishmania infantum protein, Mol. Divers.

(2009).

[46] H. Gonzalez-Diaz, R. Molina, E. Uriarte, Recognition of stable protein mutants

with 3D stochastic average electrostatic potentials, FEBS Lett. 579 (2005)

4297e4301.

[47] R. Concu, G. Podda, E. Uriarte, H. Gonzalez-Diaz, Computational chemistry

study of 3D-structureefunction relationships for enzymes based on Markov

models for protein electrostatic, HINT, and van der Waals potentials,

J. Comput. Chem. 30 (2009) 1510e1520.

[48] StatSoft.Inc., STATISTICA (Data Analysis Software System), Version 6.0. Statsoft

Inc., 2002.www.statsoft.com.

[49] G.M. Casanola-Martin, Y. Marrero-Ponce, M.T. Khan, S.B. Khan, F. Torrens,

F. Perez-Jimenez, A. Rescigno, C. Abad, Bond-based 2D quadratic ﬁngerprints

in QSAR studies: virtual and in vitro tyrosinase inhibitory activity elucidation,

Chem. Biol. Drug Des. 76 (2010) 538e545.

[50] J.A. Castillo-Garit, M.C. Vega, M. Rolon, Y. Marrero-Ponce, V.V. Kouznetsov,

D.F. Torres, A. Gomez-Barrio, A.A. Bello, A. Montero, F. Torrens, F. Perez-

Gimenez, Computational discovery of novel trypanosomicidal drug-like

chemicals by using bond-based non-stochastic and stochastic quadratic maps

and linear discriminant analysis, Eur. J. Pharm. Sci. 39 (2010) 30e36.

[51] R. Gozalbes, F. Barbosa, E. Nicolai, D. Horvath, N. Froloff, Development and

validation of a pharmacophore-based QSAR model for the prediction of CNS

activity, ChemMedChem 4 (2009) 204e209.

70 (2008) 167e175.

24] J.P. Zbilut, A. Giuliani, A. Colosimo, J.C. Mitchell, M. Colafranceschi, N. Marwan,

C.L. Webber Jr., V.N. Uversky, Charge and hydrophobicity patterning along the

sequence predicts the folding mechanism and aggregation of proteins:

a computational approach, J. Proteome Res. 3 (2004) 1243e1253.

25] J.P. Zbilut, A. Colosimo, F. Conti, M. Colafranceschi, C. Manetti, M. Valerio,

C.L. Webber Jr., A. Giuliani, Protein aggregation/folding: the role of deter-

ministic singularities of sequence hydrophobicity as determined by nonlinear

signal analysis of acylphosphatase and abeta(1e40), Biophys. J. 85 (2003)

3544e3557.

[

26] M.A. Yildirim, K.I. Goh, M.E. Cusick, A.L. Barabasi, M. Vidal, Drug-target

network, Nat. Biotechnol. 25 (2007) 1119e1126.

27] D. Viña, E. Uriarte, F. Orallo, H. Gonzalez-Diaz, Alignment-free prediction of

a drug-target complex network based on parameters of drug connectivity and

protein sequence of receptors, Mol. Pharmacol. 6 (2009) 825e835.

28] R. Todeschini, V. Consonni, Handbook of Molecular Descriptors. Wiley-VCH,

[

2002.

29] H. Gonzalez-Diaz, F. Prado-Prado, F.M. Ubeira, Predicting antimicrobial drugs

and targets with the MARCH-INSIDE approach, Curr. Top. Med. Chem. 8

(

2008) 1676e1690.

[

30] G. Aguero-Chapin, J. Varona-Santos, G.A. de la Riva, A. Antunes, T. Gonzalez-

Villa, E. Uriarte, H. Gonzalez-Diaz, Alignment-free prediction of poly-

galacturonases with pseudofolding topological indices: experimental isolation

from coffea arabica and prediction of a new sequence, J. Proteome Res. 8 (2009)

2122e2128.

[

31] R. Concu, M.A. Dea-Ayuela, L.G. Perez-Montoto, F. Bolas-Fernandez, F.J. Prado-

Prado, G. Podda, E. Uriarte, F.M. Ubeira, H. Gonzalez-Diaz, Prediction of

enzyme classes from 3D structure: a general model and examples of experi-

mental-theoretic scoring of peptide mass ﬁngerprints of Leishmania proteins,

J. Proteome Res. 8 (2009) 4372e4382.

[52] Y. Marrero-Ponce, A. Meneses-Marcel, O.M. Rivera-Borroto, R. Garcia-Dome-

nech, J.V. De Julian-Ortiz, A. Montero, J.A. Escario, A.G. Barrio, D.M. Pereira,

J.J. Nogal, R. Grau, F. Torrens, C. Vogel, V.J. Aran, Bond-based linear indices in

QSAR: computational discovery of novel anti-trichomonal compounds,

J. Comput. Aided Mol. Des 22 (2008) 523e540.

[

32] H. Gonzalez-Diaz, L. Saiz-Urra, R. Molina, L. Santana, E. Uriarte, A model for the

recognition of protein kinases based on the entropy of 3D van der Waals

interactions, J. Proteome Res. 6 (2007) 904e908.

[53] S.J. Patankar, P.C. Jurs, Classiﬁcation of inhibitors of protein tyrosine phos-

phatase 1B using molecular structure based descriptors, J. Chem. Inf. Comput.

Sci. 43 (2003) 885e899.

[

33] C.R. Munteanu, J.M. Vazquez, J. Dorado, A.P. Sierra, A. Sanchez-Gonzalez,

F.J. Prado-Prado, H. Gonzalez-Diaz, Complex network spectral moments for

ATCUN motif DNA cleavage: ﬁrst predictive study on proteins of human

pathogen parasites, J. Proteome Res. 8 (2009) 5219e5228.

[54] M. Murcia-Soler, F. Perez-Gimenez, F.J. Garcia-March, M.T. Salabert-Salvador,

W. Diaz-Villanueva, P. Medina-Casamayor, Discrimination and selection of

new potential antibacterial compounds using simple topological descriptors,

J. Mol. Graph. Model. 21 (2003) 375e390.

[

34] Y. Rodriguez-Soca, C.R. Munteanu, F.J. Prado-Prado, J. Dorado, A. Pazos Sierra,

H. Gonzalez-Diaz, Trypano-PPI: a web server for prediction of unique targets

in trypanosome proteome by using electrostatic parameters of protein-

protein interactions, J. Proteome Res. (2009). doi:10.1021/pr900827b.

35] R. Concu, M.A. Dea-Ayuela, L.G. Perez-Montoto, F.J. Prado-Prado, E. Uriarte,

F. Bolas-Fernandez, G. Podda, A. Pazos, C.R. Munteanu, F.M. Ubeira,

H. Gonzalez-Diaz, 3D entropy and moments prediction of enzyme classes and

experimental-theoretic study of peptide ﬁngerprints in Leishmania parasites,

Biochim. Biophys. Acta 1794 (2009) 1784e1794.

[55] R.A. Cercos-del-Pozo, F. Perez-Gimenez, M.T. Salabert-Salvador, F.J. Garcia-

March, Discrimination and molecular design of new theoretical hypolipaemic

agents using the molecular connectivity functions, J. Chem. Inf. Comput. Sci.

40 (2000) 178e184.

[

[56] B.H. Junker, D. Koschutzki, F. Schreiber, Exploration of biological network

centralities with CentiBiN, BMC Bioinf. 7 (2006) 219.

[57] D. Koschützki, pp. CentiBiN Version 1.4.2, centralities in biological networks^ꢀ

2004e2006 Dirk Koschützki research group network analysis. IPK Gate-

rsleben, Germany, 2006.

[

36] L. Santana, H. Gonzalez-Diaz, E. Quezada, E. Uriarte, M. Yanez, D. Vina,

F. Orallo, Quantitative structure-activity relationship and complex network

approach to monoamine oxidase A and B inhibitors, J. Med. Chem. 51 (2008)

[58] Hypercube Inc., Hyperchem Software. Release 7.5 for Windows, Molecular

Modeling System. Hypercube Inc., Gainesville, FL, USA, 2002.

[59] M. Froimowitz, HyperChem: a software package for computational chemistry

and molecular modeling, BioTechniques 14 (1993) 1010e1013.

[60] I. Hypercube, Hyperchem Inc, Gainesville, FL, USA, 2002.

[61] Y. Liu, D.L. Beveridge, Exploratory studies of ab initio protein structure

prediction: multiple copy simulated annealing, AMBER energy functions, and

a generalized born/solvent accessibility solvation model, Proteins 46 (2002)

128e146.

[62] M. Mezo, M. Gonzalez-Warleta, F.M. Ubeira, Optimized serodiagnosis of sheep

fascioliasis by Fast-D protein liquid chromatography fractionation of Fasciola

hepatica excretoryesecretory antigens, J. Parasitol. 89 (2003) 843e849.

[63] O. Ivanciuc, N. Oezguen, V.S. Mathura, C.H. Schein, Y. Xu, W. Braun, Using

property based sequence motifs and 3D modeling to determine structure and

functional regions of proteins, Curr. Med. Chem. 11 (2004) 583e593.

[64] C.H. Schein, O. Ivanciuc, W. Braun, Common physicalechemical properties

correlate with similar structure of the IgE epitopes of peanut allergens,

J. Agric. Food Chem. 53 (2005) 8752e8759.

6740e6751.

37] L. Santana, E. Uriarte, H. González-Díaz, G. Zagotto, R. Soto-Otero, E. Mendez-

Alvarez, A QSAR model for in silico screening of MAO-A inhibitors. Prediction,

synthesis, and biological assay of novel coumarins, J. Med. Chem. 49 (2006)

1149e1156.

38] H. González-Díaz, Y. Pérez-Castillo, G. Podda, E. Uriarte, Computational

chemistry comparison of stable/nonstable protein mutants classiﬁcation

models based on 3D and topological indices, J. Comput. Chem. 28 (2007)

1990e1995.

39] H. Gonzalez-Diaz, L. Saiz-Urra, R. Molina, Y. Gonzalez-Diaz, A. Sanchez-Gon-

zalez, Computational chemistry approach to protein kinase recognition using

3

1

D stochastic van der Waals spectral moments, J. Comput. Chem. 28 (2007)

042e1048.

[

40] G. Aguero-Chapin, A. Antunes, F.M. Ubeira, K.C. Chou, H. Gonzalez-Diaz,

Comparative study of topological indices of macro/supramolecular RNA

complex networks, J. Chem. Inf. Model. 48 (2008) 2265e2277.

[65] Y.M. Alvarez-Ginarte, Y. Marrero-Ponce, J.A. Ruiz-Garcia, L.A. Montero-Cabrera,

J.M. Vega, P. Noheda Marin, R. Crespo-Otero, F.T. Zaragoza, R. Garcia-Domenech,

Applying pattern recognition methods plus quantum and physico-chemical

molecular descriptors to analyze the anabolic activity of structurally diverse

steroids, J. Comput. Chem. (2007).

[

41] M. Cruz-Monteagudo, C.R. Munteanu, F. Borges, M.N.D.S. Cordeiro,

E. Uriarte, K.-C. Chou, H. González-Díaz, Stochastic molecular descriptors

for polymers. 4. Study of complex mixtures with topological indices of

mass spectra spiral and star networks: the blood proteome case, Polymer

4

9 (2008) 5575e5587.

[66] A.H. Morales, J.E. Rodríguez-Borges, X. García-Mera, F. Fernández, M.N. Dias-

Sueiro-Cordeiro, Probing the anticancer activity of nucleoside analogs: a QSAR

model approach using an internally consistent training set, J. Med. Chem. 50

(2007) 1537e1545.

[67] M. Fernandez, J. Caballero, A. Tundidor-Camba, Linear and nonlinear QSAR

study of N-hydroxy-2-[(phenylsulfonyl)amino]acetamide derivatives as

matrix metalloproteinase inhibitors, Bioorg. Med. Chem. 14 (2006)

4137e4150.

[68] J. Caballero, M. Fernandez, Linear and nonlinear modeling of antifungal

activity of some heterocyclic ring derivatives using multiple linear regression

and Bayesian-regularized neural networks, J. Mol. Model. 12 (2006) 168e181.

[69] D.A. Chung, E.R. Zuiderweg, C.B. Fowler, O.S. Soyer, H.I. Mosberg, R.R. Neubig,

NMR structure of the second intracellular loop of the alpha 2A adrenergic

[

42] M.A. Dea-Ayuela, Y. Perez-Castillo, A. Meneses-Marcel, F.M. Ubeira, F. Bolas-

Fernandez, K.C. Chou, H. Gonzalez-Diaz, HP-Lattice QSAR for dynein proteins:

experimental proteomics (2D-electrophoresis, mass spectrometry) and

theoretic study of a Leishmania infantum sequence, Bioorg. Med. Chem. 16

(

2008) 7770e7776.

43] G. Aguero-Chapin, H. Gonzalez-Diaz, G. de la Riva, E. Rodriguez, A. Sanchez-

Rodriguez, G. Podda, R.I. Vazquez-Padron, MMM-QSAR recognition of ribo-

nucleases without alignment: comparison with an HMM model and isolation

from Schizosaccharomyces pombe, prediction, and experimental assay of a new

sequence, J. Chem. Inf. Model. 48 (2008) 434e448.

44] G. Ferino, H. Gonzalez-Diaz, G. Delogu, G. Podda, E. Uriarte, Using spectral

moments of spiral networks based on PSA/mass spectra outcomes to derive

1094

F. Prado-Prado et al. / European Journal of Medicinal Chemistry 46 (2011) 1074e1094

0

receptor: evidence for a novel cytoplasmic helix, Biochemistry 41 (2002)

596e3604.

70] M. Katragadda, M.W. Maciejewski, P.L. Yeagle, Structural studies of the

the synthesis of 2 -deoxy-2 -C-ethynylnucleosides, J. Org. Chem. 68 (2003)

3465e3475.

3

[

[85] N.T. Nguyen, D.M. Cook, L.A. Bero, The decision-making process of US food and

drug administration advisory committees on switches from prescription to

over-the-counter status: a comparative case study, Clin. Ther. 28 (2006)

1231e1243.

[86] A. Duardo-Sanchez, G. Patlewicz, A. Lopez-Diaz, Current topics on software

use in medicinal chemistry: intellectual property, taxes, and regulatory issues,

Curr. Top. Med. Chem. 8 (2008) 1666e1675.

[87] H. González-Díaz, F. Prado-Prado, L.G. Pérez-Montoto, A. Duardo-Sánchez,

A. López-Díaz, QSAR models for proteins of parasitic organisms, plants and

human guests: theory, applications, legal protection, taxes, and regulatory

issues, Curr. Proteomics 6 (2009) 214e227.

[88] H. González-Díaz, A. Duardo-Sanchez, F.M. Ubeira, F. Prado-Prado, L.G. Pérez-

Montoto, R. Concu, G. Podda, B. Shen, Review of MARCH-INSIDE & complex

networks prediction of drugs: aDMET, anti-parasite activity, metabolizing

enzymes and cardiotoxicity proteome Biomarkers, Curr. Drug Metab. 11

(2010) 379e406.

putative helix 8 in the human beta(2) adrenergic receptor: an NMR study,

Biochim. Biophys. Acta 1663 (2004) 74e81.

[

71] K. Khaﬁzov, G. Lattanzi, P. Carloni, G protein inactive and active forms

investigated by simulation methods, Proteins 75 (2009) 919e930.

72] H. Kalasz, G. Petroianu, K. Tekes, I. Klebovich, K. Ludanyi, Z. Gulyas, Metabo-

lism of moexipril to moexiprilat: determination of in vitro metabolism using

HPLC-ES-MS, Med. Chem. 3 (2007) 101e106.

[

73] G.S. Chrysant, P.K. Nguyen, Moexipril and left ventricular hypertrophy, Vasc.

Health Risk Manage. 3 (2007) 23e30.

74] F. Sayegh, J. Topouchian, M. Hlawaty, M. Olzewska, R. Asmar, Regression of left

ventricular hypertrophy with moexipril, an angiotensin-converting enzyme

inhibitor, in hypertensive patients, Am. J. Therapeut. 12 (2005) 3e8.

75] A. Innocenti, R.A. Hall, C. Schlicker, A. Scozzafava, C. Steegborn,

F.A. Muhlschlegel, C.T. Supuran, Carbonic anhydrase inhibitors. Inhibition and

homology modeling studies of the fungal beta-carbonic anhydrase from

Candida albicans with sulfonamides, Bioorg. Med. Chem. 17 (2009)

[

[89] H. Gonzalez-Diaz, F. Romaris, A. Duardo-Sanchez, L.G. Perez-Montoto,

F. Prado-Prado, G. Patlewicz, F.M. Ubeira, Predicting drugs and proteins in

parasite infections with topological indices of complex networks: theoretical

backgrounds, applications, and legal issues, Curr. Pharm. Des. 16 (2010)

2737e2764.

4503e4509.

76] C. Temperini, A. Cecchi, A. Scozzafava, C.T. Supuran, Carbonic anhydrase

inhibitors. Comparison of chlorthalidone, indapamide, trichloromethiazide,

and furosemide X-ray crystal structures in adducts with isozyme II, when

several water molecules make the difference, Bioorg. Med. Chem. 17 (2009)

[90] C. Yang, L.G. Valerio Jr., K.B. Arvidson, Computational toxicology approaches at

the US food and drug administration, Altern. Lab. Anim. 37 (2009) 523e531.

[91] Z. Lei, A.M. Elmer, B.S. Watson, R.A. Dixon, P.J. Mendes, L.W. Sumner, A two-

dimensional electrophoresis proteomic reference map and systematic iden-

tiﬁcation of 1367 proteins from a cell suspension culture of the model legume

Medicago truncatula, Mol. Cell Proteomics 4 (2005) 1812e1825.

[92] M.C. Giddings, A.A. Shah, R. Gesteland, B. Moore, Genome-based peptide

ﬁngerprint scanning, Proc. Natl. Acad. Sci. U. S. A. 100 (2003) 20e25.

[93] S. Dewilde, A.I. Ioanitescu, L. Kiger, K. Gilany, M.C. Marden, S. Van Doorslaer,

J. Vercruysse, A. Pesce, M. Nardini, M. Bolognesi, L. Moens, The hemoglobins of

the trematodes Fasciola hepatica and Paramphistomum epiclitum: a molecular

biological, physico-chemical, kinetic, and vaccination study, Protein Sci. 17

(2008) 1653e1662.

1214e1221.

77] P. Dallemagne, S. Rault, M. Cugnon de Sévricourt, K.M. Hassan, M. Robba,

Tetrahedron Lett. 23 (1986) 2607e2610.

78] S. Rault, P. Dallemagne, M. Robba, Bull. Soc. Chim. Fr. 6 (1987) 1079e1083.

79] M.A. Quermonne, P. Dallemagne, J. Louchahi-Raould, J.C. Pilo, S. Rault,

M. Robba, Eur. J. Med. Chem. 27 (1992) 961e965.

[

80] P. Dallemagne, S. Rault, J.C. Pilo, M.P. Foloppe, M. Robba, Tetrahedron Lett. 32

(

1991) 6327e6328.

81] P. Dallemagne, S. Rault, M. Gordaliza, M. Robba, Heterocycles 26 (1987)

233e3237.

3

82] J. Guillon, G. Hebert, P. Dallemagne, J.M. Leger, C. Vidaillac, C. The, V. Lisowski,

S. Rault, J. Demotes-Mainard, C. Jarry, Synthesis and initial results for MAO-B

inhibition by new N-propargyl-3-pyrrol-1-ylindanamine derivatives, analogs

of rasagiline, J. Enzyme Inhib. Med. Chem. 18 (2003) 147e153.

83] P. Abeijón, J.M. Blanco, F. Fernández, M.D. García, O. Caamaño, Synthesis of

two precursors of heterocarbocyclic nucleoside analogues, Eur. J. Med. Chem.

[94] F. Hubalek, C. Binda, A. Khalil, M. Li, A. Mattevi, N. Castagnoli, D.E. Edmondson,

Demonstration of isoleucine 199 as a structural determinant for the selective

inhibition of human monoamine oxidase B by speciﬁc reversible inhibitors,

J. Biol. Chem. 280 (2005) 15761e15766.

[

(

2003) 759e764.

[95] C. Binda, F. Hubalek, M. Li, N. Castagnoli, D.E. Edmondson, A. Mattevi, Struc-

ture of the human mitochondrial monoamine oxidase B: new chemical

implications for neuroprotectant drug design, Neurology 67 (2006) S5eS7.

[

84] M. Sukeda, S. Ichikawa, A. Matsuda, S. Shuto, The ﬁrst radical method for the

introduction of an ethynyl group using a silicon tether and its application to

Article Doi

Using entropy of drug and protein graphs to predict FDA drug-target network: Theoretic-experimental study of MAO inhibitors and hemoglobin peptides from Fasciola hepatica

DOI: 10.1016/j.ejmech.2011.01.023

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.ejmech.2011.01.023

Full text of DOI:10.1016/j.ejmech.2011.01.023

Products guided by the article

R&D Labs maybe for 3310-99-4

Relevant to this article

Hot Product