Structural variations on antitumour agents derived from bisacylimidoselenocarbamate. A proposal for structure-activity relationships based on the analysis of conformational behaviour

Article abstract of DOI:10.1016/j.ejmech.2013.06.001

A molecular modelling study has been carried out on a previously reported series of symmetrically substituted bisacylimidoselenocarbamate (BSeC) derivatives that show remarkable antitumour activity in vitro against a panel of human tumour cell lines. These derivatives can be considered as a central scaffold constructed around a methyl carbamimidoselenoate nucleus in which two heteroarylacyl fragments are located on the scaffold nitrogen atoms, thus forming the different BSeCs. The results reveal that the nature of the selected heteroaryl ring has a marked influence on the antiproliferative activity of the compounds and this can be related, as a first approximation, to the ability to release methylselenol (MeSeH), a compound that, according to our initial hypothesis, is ultimately responsible for the antitumour activity of the compounds under investigation. The release of MeSeH from the active BSeCs has been confirmed by means of Head Space Gas Chromatography Mass Spectrometry techniques. The data that support this connection include the topography of the molecules, the conformational behaviour of the compounds, which influences the accessibility of the hydrolysis point, the interaction map obtained for an O2H type probe, and the location and energy of the HOMO/LUMO orbitals.

Full text of DOI:10.1016/j.ejmech.2013.06.001

European Journal of Medicinal Chemistry 66 (2013) 489e498
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Structural variations on antitumour agents derived from
bisacylimidoselenocarbamate. A proposal for structureeactivity
relationships based on the analysis of conformational behaviour
,
b
c
c
c
María Font a , Elena Lizarraga , Elena Ibáñez , Daniel Plano , Carmen Sanmartín ,
*
Juan A. Palop c
a Sección de Modelización Molecular, Facultad de Farmacia, Departamento de Química Orgánica y Farmacéutica, Universidad de Navarra,
Irunlarrea, 1, E-31008 Pamplona, Spain
b Sección de técnicas instrumentales, Departamento de Química Orgánica y Farmacéutica, Universidad de Navarra, Irunlarrea, 1, E-31008 Pamplona, Spain
c Sección de síntesis, Departamento de Química Orgánica y Farmacéutica, Universidad de Navarra, Irunlarrea, 1, E-31008 Pamplona, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
A molecular modelling study has been carried out on a previously reported series of symmetrically
substituted bisacylimidoselenocarbamate (BSeC) derivatives that show remarkable antitumour activity
in vitro against a panel of human tumour cell lines. These derivatives can be considered as a central
scaffold constructed around a methyl carbamimidoselenoate nucleus in which two heteroarylacyl frag-
ments are located on the scaffold nitrogen atoms, thus forming the different BSeCs. The results reveal
that the nature of the selected heteroaryl ring has a marked influence on the antiproliferative activity of
the compounds and this can be related, as a first approximation, to the ability to release methylselenol
(MeSeH), a compound that, according to our initial hypothesis, is ultimately responsible for the anti-
tumour activity of the compounds under investigation. The release of MeSeH from the active BSeCs has
been confirmed by means of Head Space Gas Chromatography Mass Spectrometry techniques. The data
that support this connection include the topography of the molecules, the conformational behaviour of
the compounds, which influences the accessibility of the hydrolysis point, the interaction map obtained
for an O2H type probe, and the location and energy of the HOMO/LUMO orbitals.
Received 25 March 2013
Received in revised form
31 May 2013
Accepted 1 June 2013
Available online 18 June 2013
Keywords:
Antiproliferative
Bisacylimidoselenocarbamate
Methylselenol release
Ó 2013 Elsevier Masson SAS. All rights reserved.
1. Introduction
metabolic products, their doses and the possible involvement of
different molecular targets and apparent mechanisms. However, it is
Among the different strategies pursued in recent years to study
the factors involved in the development of tumour processes, it is
worth highlighting the studies related to the true role played by
some essential trace elements such as selenium (which is essential
for the proper functioning of an organism but is toxic in high doses
[1,2]) in cancer prevention and the possible involvement of this
element in the control of processes that trigger tumour develop-
ment. In fact, a substantial body of persuasive evidence indicates
that selenium (Se) plays a role in cancer prevention [3e6] despite
the discrepancies observed in different epidemiological nutritional
studies [7e9]. These discrepancies could be related to the different
chemical forms in which Se was administered to subjects involved
in the studies [10]. The mechanisms underlying the influence of Se
chemical forms are not yet fully elucidated e including their
commonly accepted that the three most important forms of Se in
cancer prevention are sodium selenite (SS), -selenomethionine
L
(SeMet) and Se-methylselenocysteine (MeSeCys) [11e13]. Further-
more, there is evidence that the various forms of selenium have
different capacities to induce cell death in various tumour cell lines
and that this behaviour is related to Se metabolites [14e16].
As an example, methylselenol (MeSeH) has been hypothesized
to be a critical Se metabolite for anticancer activity in vivo [17e19].
The biological specificity of MeSeH generation in cell culture media
has been well established [20,21]. In this respect, several recent
studies have shown that, in different culture media and cell lines,
MeSeH generated by incubating methioninase (METase) with
SeMet is responsible for anticarcinogenic effects, including regu-
lation of gene expression [22,23]. It can be hypothesized that
different signalling pathways are involved in MeSeH-induced
regulation of the cell cycle and apoptosis and that these factors
play a significant role in the inhibition of the invasive potential of
tumour cells.
* Corresponding author. Tel.: þ34 948 425 600; fax: þ34 948 425 649.
E-mail address: mfont@unav.es (M. Font).
0223-5234/$ e see front matter Ó 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2013.06.001

490
M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
In recent years, our group has developed a series of bisacyli-
highly dependent on the nature of the lateral rings and the type,
size and location of substituents on these rings and correlated with
their proven pro-apoptotic activity [25,26].
midoselenocarbamate (BSeC) derivatives that show interesting
antitumour activity [24,25] and this activity has recently been
related to the ability to release MeSeH [26,27]. According to our
starting hypothesis, the possibility of controlled delivery of MeSeH,
achieved by means of structural variations carried out on a lead
compound, makes this approach an interesting tool to develop new
antitumour agents and to provide new insights into the mecha-
nisms of the anticancer properties of Se.
In our investigations into the mechanism of action of these
derivatives we have demonstrated that several of them release
MeSeH under hypoxic conditions. In our previous work, N,N’-
diacetyl-2-methylisoselenourea (I, Fig. 1) was selected as a lead
compound and this is the simplest derivative from a structural
viewpoint. This compound showed the most intense and rapid
release of MeSeH of the tested BSeCs and we subsequently
designed new derivatives in which the molecular architecture
became progressively more complicated as a consequence of the
replacement of the lateral methyl groups by aromatic or hetero-
aromatic rings (II, Fig. 1). The aim of these modifications was to
modulate the speed and intensity of the MeSeH release, taking into
account the kinetics of this liberation, which were found to be
Taking into account the data previously obtained in the first
design cycle (Fig. 1a), in the work described here we carried out a
molecular modelling study on the new BSeCs corresponding to the
second design cycle (Fig. 1b). The compounds can be considered to
consist of a central scaffold constructed around a methyl carba-
mimidoselenoate in which two heteroarylacyl fragments are
located on the scaffold nitrogen atoms. The lateral heteroaromatic
rings (monocyclic, bicyclic and tricyclic rings) were chosen in order
to cover a wide range of electronic properties (pi-deficient, pi and
pi-excessive systems), surface and volume. The aim was to evaluate
the ability of these features to modulate the release of MeSeH. In
this way, the whole imidoselenocarbamate is considered as a pro-
drug and the heteroaromatic moieties as the release regulator
elements. According to the proposed mechanism (Fig. 2), hydrolysis
can be mediated by the nucleophilic attack of a water molecule on
the carbon that supports the methylselene fragment.
The cytotoxic activities of the analysed compounds were
reported previously [24,26]. The GI50 value (i.e. the dose that in-
hibits 50% of cell growth) was considered as a cytotoxic parameter
Fig. 1. (a) Structural variations for the first design cycle. Solid arrow indicates the suggested hydrolysis point. (b) Structural overview of the analysed compounds corresponding to
the second design cycle.

M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
491
Fig. 2. Proposed mechanism for the MeSeH release.
in colon (HT-29) and breast adenocarcinoma (MCF-7) cell lines and
the results are given in Table 1.
The molecular modelling calculations were performed on a Dell
Precision 380 workstation provided with the software package
MOE [31] and on an SGI Virtu VS100 workstation provided with the
MOPAC2009 software package [32].
In order to facilitate the understanding of the results concerning
the effect of the structural modifications on the target activity, the
previously reported analytical and spectroscopic data as well as the
activity data and a description of the biological methods are
included as supplementary material. Some representative mass and
1H and 13C NMR spectra are also included in the supplementary on
line material.
The work described here had three main goals. Firstly, the
release of the proposed active agent, MeSeH, was verified by
applying a screening method supported by analytical techniques,
i.e., Head Space Gas Chromatography Mass Spectrometry (HS-GCe
MS) [28]. This analysis provided direct identification of the highly
volatile, unstable and reactive key compound MeSeH and its
recombination products, which include dimethyldiselenide
(DiMeDiSe) and dimethylselenide (DiMeSe) according to our pre-
vious study [27]. The second goal was to study the conformational
behaviour of the molecules in order to evaluate the influence of this
factor on the accessibility of the hydrolysis point. The third goal was
the application of molecular modelling approaches in order to
obtain descriptors, preferably topological and quantum descriptors
[26], that would allow us to correlate the molecular structural
variations with the hydrolysis of the MeSeH moiety and subse-
quently with the cytotoxic activity.
The HS-GCeMS data were obtained on an Agilent 6890 GC
system coupled to a 5973N quadrupole mass spectrometer (Agilent
Technologies, Santa Clara, CA, USA). The compounds were dissolved
in 5% dimethylsulfoxide in H2O (5% DMSO/H2O) at a concentration
of 60 mM. The samples were kept at room temperature in sealed
HS-GCeMS vials and samples for analysis were taken from the vial
head space at 8, 16 and 24 h for each of the studied BSeCs. Evidence
for the positive release of MeSeH was obtained by direct detection
of this compound as well through the detection of the secondary
recombination compounds DiMeDiSe and DiMeSe [29,30].
2. Results and discussion
The architecture of the analysed compounds can be considered
as consisting of a methyl carbamimidoselenoate scaffold decorated
with a series of different monocyclic, bicyclic and tricyclic aryl or
heteroaryl planar rings. These units are connected through a
carbonyl fragment on the nitrogen atoms of the scaffold; the
structural variations were selected with the aim of evaluating their
ability to modulate the release of MeSeH.
The high volatility and reactivity of MeSeH means that its
effective release from the analysed BSeCs must be established
through the detection of either the compound itself or of any sec-
ondary products (DiMeDiSe and DiMeSe) that may appear due to
oxidation of the initially released MeSeH and subsequent recom-
bination of the oxidized sub-products. Therefore, in order to stan-
dardise the analytical procedure, we first determined the HS-GCe
MS behaviour of MeSeH generated in situ by reductive hydrolysis
with NaBH4 of a solution of commercial DiMeDiSe in methanol.
The data obtained show that hydrolysis is caused by NaBH4 and
that a significant amount of MeSeH is detected. This compound
subsequently undergoes a progressive oxidative process, which
results in the regeneration of the initial compound DiMeDiSe [28].
In fact, two significant peaks were detected when the HS-GC-MS is
obtained in SCAN mode: the first peak, with a retention time (tR) of
1.5 min, corresponds the in situ generated MeSeH, whereas the
second peak, with a tR of 2.9 min, corresponds to the regenerated
DiMeDiSe. The mass spectrum corresponding to each of the peaks
are consistent with the expected compounds MeSeH and DiMeDiSe
(See representative figures in the supplementary on line material).
Once the mass spectra had been obtained in SCAN mode, the
mass detector was operated in SIM mode. The most abundant
fragment ions were selected for each standard, according to the
characteristic Se isotope patterns of compounds containing one
or two Se atoms, and their relative intensities. As a result, m/z
values of 96, 95, 93 and 80 for MeSeH and 190, 188, 186, 175 and 160
for DiMeDiSe were monitored (See representative figure in the
supplementary on line material).
Table 1
Average GI50a values (
mM) for analysed compounds.
Ref.
HT-29b
MCF-7c
1b
2b
3b
4b
5b
6b
5 ꢀ 10ꢁ5
3.4 ꢀ 10ꢁ4
0.0049
6.3 ꢀ 10ꢁ4
7.9 ꢀ 10ꢁ4
5.62
0.03
0.0034
0.30
7.6 ꢀ 10ꢁ4
9 ꢀ 10ꢁ5
5.07
7b
0.53
0.99
8b
0.20
0.66
9b
0.46
0.49
10b
7.63
0.20
0.05
31.62
0.003
7.95
0.04
73.92
0.02
0.01
19.95
0.003
3.16
Doxorubicind
Camptothecind
Etoposided
Taxold
Cisplatind
Methotrexated
0.06
a
Concentration that inhibits 50% of cell growth.
Colon carcinoma.
Breast adenocarcinoma.
b
c
d
NCI data (http://dtp.nci.nih.gov).

492
Table 2
M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
In order to gain an insight into the structureeactivity relation-
ships that would allow us to explain the results, two different ap-
proaches were used to establish a preliminary SAR. Firstly, we
analysed the possible influence that structural variations carried
out on the BSeC derivatives have on the conformational behaviour.
This was achieved by determining the preferred conformation
(lowest energy conformation, considered as the global minimum)
and subsequent evaluation of the accessibility to the proposed
hydrolysis point, i.e. the carbon attached to the selenomethyl
fragment (Fig. 1a). This evaluation was carried out using an O2H
probe interaction map and by considering the hydrogen bond
acceptor/donor map around the hydrolysis point. The second goal
was to determine quantitative parameters, preferably topological
and quantum parameters, which would allow us to establish a new
data set to aid the further design of new molecular entities with
improved activity profiles.
DMeDSe amount (tR (min) ¼ 2.897) obtained after 16 and 24 h, for repre-
sentative compounds (expressed as area).
Ref.
16 h
24 h
1b
2b
8b
10b
23584
16164
15036
12254
29333
23196
23982
18425
The BSeC samples were dissolved in 5% DMSO/H2O and were
stored at room temperature in sealed vials. These samples show
interesting behaviour in the same analytical procedure. The data
corresponding to the first sampling, taken after 8 h of incubation
(data not shown for the sake of brevity), show that compounds 1b
and 3b undergo the expected hydrolysis, with two intense peaks
due to DiMeDiSe and MeSeH. Compound 4b only gave rise to the
MeSeH peak as a trace component. Compounds 2b and 5b showed
only the DiMeDiSe peak and this had a medium intensity compared
to those of 1b and 3b. Peaks were not detected for the other
compounds tested.
To further study the release kinetics, data were obtained after
incubation for 16 and 24 h for four of the compounds,1b, 2b, 8b and
10b, which were selected as being representative of the different
behaviours in the biological activity.
The areas obtained for the DiMeDiMe generated after incuba-
tion for the indicated times (16 and 24 h) are shown in Table 2. The
results indicate a greater MeSeH release rate for 1b. Thus, after an
incubation period of 16 h, the amount of DiMeSeMe released by
compound 1b was 45.9% greater than that from compound 2b,
56.8% greater than that from compound 8b and 92.46% greater than
that from compound 10b. After 24 h the release rate was similar for
all compounds under investigation (See representative figure in the
supplementary on line material).
On the basis of the data obtained from this analysis, an initial
correlation between the structure of the compounds and the
MeSeH release kinetics can be established. In the case of BSeC de-
rivatives that contain monocyclic rings on the scaffold, the release
of the active agent is more rapid and has a higher intensity than for
the bi- or tricyclic derivatives. This behaviour correlates with bio-
logical activity (Table 1) as the most active compounds, especially
in the HT-29 cell line (pGI50 ¼ 7.3010, 6.4685, 6.2007, 6.1024 and
5.3098 for compounds 1b, 2b, 4b, 5b and 3b, respectively), showed
greater ability to liberate the selenomethyl moiety. Similar behav-
iour was found for the MCF-7 cell line, for which BSeCs 2b, 4b and
5b showed the highest activity.
The activity data allowed us to propose that the cytotoxic ca-
pacity could also be related to the different sensitivities of cell lines
to the active agent MeSeH. The release rate may also have an in-
fluence, as a slow release rate would, in principle, allow the
released MeSeH to undergo a more gradual oxidation, thus leading
to a higher proportion of DiMeDiSe, which has no direct activity as a
cytotoxic agent. A rapid and intense release would lead to a more
direct and immediate cytotoxic activity.
The three-dimensional models of the studied compounds were
constructed according to three initial basic conformations (Fig. 3aec)
in the vacuum phase (see experimental section for details) and, after
a preliminary optimization, the models were subjected to a confor-
mational search (systematic search strategy). This process gave the
conformational trajectory (maximum number of lowest energy
conformations ¼ 100) for each analysed compound (Fig. 3d).
The final conformations obtained for each compound were
distributed into two conformational families named ‘folded’ and
‘extended’. The ‘folded’ family includes the conformations that have
a more globular architecture, frequently with the lateral rings
involved in piepi stacking interactions (preferably T-shaped or
sandwich), whereas the ‘extended’ family includes the conforma-
tions that fit an extended molecular architecture. Several repre-
sentative conformations are shown in Fig. 4 and these illustrate the
results of the conformational analysis.
It can be deduced from the data collected in Table 3 that the
BSeCs with lateral monocyclic rings showed a preference for
extended conformations, especially for compounds that were
unsubstituted at the rings. For example, compound 1b only gave
rise to extended-type conformations; similar behaviour was found
for compounds 2b and 5b, both of which showed a significant
preference for the extended conformation. Although compounds
3b and 4b also show a clear preference for the extended form,
appreciable amounts of folded forms were evident due to the in-
teractions established by the substituents on the rings. It can also
be observed that as the molecular architecture becomes more
complex, i.e. on increasing the size of the lateral rings from
monocyclic to bi- and tricyclic, the proportion of folded confor-
mations progressively increases and the aforementioned in-
teractions are detected between the rings.
These data allow us to establish a clear relationship between the
preferred conformation shown by the analysed BSeCs and their
ability to release MeSeH. As can be observed, the compounds for
which the determined release was faster and more intense (e.g.
compound 1b) preferentially adopt an extended conformation,
whereas compounds such as 10b, for which the release of MeSeH
was very slow and minor, the folded conformation was preferred.
Fig. 3. Initial conformations taken as a starting point for the conformational analysis. Solid line, selected rotatable bonds.

M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
493
Fig. 4. Representative folded and extended conformations. Dotted lines indicate the detected clashes.
Compounds with an intermediate release, such as 8b, showed
conformational behaviour that was distributed between the folded
and extended families.
The lowest energy conformations for compounds 1b, 2b, 3b, 4b,
5b and 8b are included in the extended conformational family,
while those for 6b, 7b, 9b and 10b are included in the folded family
(see Fig. 4 for representative examples).
It can be concluded from this initial study that in the extended
conformations the hydrolysis point is more exposed to attack by the
nucleophile than in the folded forms, a situation that results in an
increased release of MeSeH.
In order to confirm the above conclusion, the interaction map of
the molecular surface was also calculated for each compound by
employing an O2H probe (0.00 charge, with an interaction energy
of ꢁ5.5 kcal/mol). The graphical representation showing where this
chemical probe has favourable interactions with the molecular
surface, calculated according to Goodford’s GRID approach [33],
confirms that the more favourable interactions were shown by
compounds 1b, 2b, 4b and 5b (see Figs. 5 and 6 for representative
examples). In these cases the interaction map is located close to the
C that bears the selenomethyl moiety, i.e., the proposed hydrolysis
point.
Table 3
Conformational distribution for the analysed compounds.
It can be observed in Fig. 6 that the folded compounds, such as
9b or 10b, have an interaction map that is located far from the
hydrolysis point.
With respect to the second molecular modelling approach, we
calculated a significant number of 2D and 3D molecular descriptors
(see supplementary material for details) with the aim of estab-
lishing the parameters that could aid the design of new series’ of
compounds with the target activity.
Data obtained in the first design cycle were used to select the
calculated descriptors. These descriptors were subsequently
applied in the design of the BSeCs analysed here according to the
bond order between Se and the carrier carbon (Fig. 1). The bond
order (b.o.) should be as low as possible and the LUMO and HOMO
orbitals should preferably be located on the scaffold atoms. The data
obtained for the currently analysed BSeCs, which concern the
accessibility of the hydrolysis point and the interaction between the
O2H probe and acceptor/donor maps, were also considered when
Ref.
% Folded
% Extended
1b
2b
3b
4b
5b
6b
7b
8b
9b
10b
0
8
100
92
72
61
84
68
58
55
5
28
39
16
32
42
45
95
98
2

494
M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
Fig. 5. Left; interaction map (O2H probe, ꢁ5.5 kcal/mol contour value); right, acceptor/donor interaction map (ꢁ2 kcal/mol contour value) for 1b and 3b, as examples of extended
compounds showing a faster and more intense MeSeH release, and 8b, as an example of a folded compound with medium release rate (calculated on the lowest energy
conformations).
selecting which parameters to estimate. Thus, as an initial approach
the calculated descriptors can be grouped as follows: (a) Topological
information indices (Adjacency and Distance Matrix Descriptors):
BalabanJ [34] (as a topostructural descriptor) and BCUT_PEOE
[35] (as a topochemical descriptor). (b) Quantum_chemical de-
scriptors: PM3-QSe (Se partial charge); PM3-QC1 (C1 partial charge);
b.o. (bond order C1eSe); PM3-EHOMO (HOMO orbital energy and
location); PM3-ELUMO (LUMO orbital energy and location);
Fig. 6. Left; interaction map (O2H probe, ꢁ5.5 kcal/mol contour value); right, acceptor/donor interaction map (ꢁ2 kcal/mol contour value) for compounds 9b and 10b, as examples
of folded compounds with low or null release of MeSeH (calculated on the lowest energy conformations).

M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
495
PM3-dipole. (c) Physical properties and hydrophobicity descriptors:
logP (o/w) and SlogP (log of the octanol/water partition), logS (log of
the aqueous solubility), vdw_vol and vdw_area (van der Waals
volume and area calculated using a connection approximation). (d)
Surface Area, volume and shape descriptors: ASA (water accessible
surface area); vol and VSA (van der Waals volume and area calcu-
lated using a grid approximation); vsuf_V and vsurf_S (interaction
field volume and area). (e) Conformation dependent charge de-
scriptors and partial charge descriptors: ASAþ (water accessible
surface area of all atoms with positive partial charge); ASAꢁ (Water
accessible surface area of all atoms with negative partial charge);
ASA_H (water accessible surface area of all hydrophobic atoms);
ASA_P (water accessible surface area of all polar atoms);
PEOE_VSA_NEG (Total negative van der Waals surface area);
PEOE_VSA_POS (Total positive van der Waals surface area) [36]. In
this way it was intended to assess topological, hydrophobic and
electrostatic aspects. The final values used for the conformationally
dependent parameters, as well as for the quantum descriptors, were
the mean values obtained for the conformational trajectory (see
supplementary material).
included in the Mopac software. As a general trend, higher values
for the HOMO orbital were obtained for the inactive compounds 6b
and 10b. The preferred location for this orbital in the active com-
pounds is in the central molecular fragment, whereas for the
inactive compounds the location is displaced towards one of the
lateral rings. Regarding the LUMO orbital, as a general trend this
frontier orbital is also located on the central fragment of the
molecule in the active compounds, with a significant contribution
from the C atom proposed as the hydrolysis point; a lower contri-
bution from this atom to the LUMO orbital was observed for the
inactive BSeCs (see Fig. 7 for representative examples).
A clear relationship can be established between the topology,
evaluated through the BalabanJ index, and the cytotoxic activity. In
fact, the most active compounds have a BalabanJ value in the range
1.7556e1.8230, whereas the inactive compounds (e.g. 9b and 10b)
have a value below 1.200.
With respect to the data obtained for the HT-29 cell line, it can
be observed that the ASA descriptor (water accessible surface area
ꢀ
calculated using a radius of 1.4 A for the water molecule) shows a
significant relationship with the cytotoxic activity, with the most
active compounds having the lowest ASA values. These data can be
misleading as one would expect that a larger surface exposed to
water, the nucleophile responsible for hydrolysis in the proposed
mechanism, would lead to greater accessibility to the attack point.
However, analysis of the data, together with the O2H probe inter-
action map described above, led us to propose that the exposed
surface is smaller for the most active compounds but is located
close to the aforementioned attack point. The ASA þ descriptor,
which represents the water accessible surface area of all atoms with
positive partial charge (strictly greater than 0), showed a similar
general trend.
The van der Waals volume (vdw_vol) and area (VSA) values can
also be related to the activity. The active compounds 1b, 2b, 3b, 4b
and 5b have the smallest volumes and areas, with values of less
than 406 Å3 for the van der Waals volume (calculated using a
connection table approach) and 352 Å2 for the VSA descriptor.
With respect to the data obtained for the MCF-7 cell line, it can
be observed that the ASA-H descriptor (water accessible surface
area of all hydrophobic atoms) shows a clear relationship with the
cytotoxic activity. In fact, the most active compounds in this cell
line (1b, 2b, 3b, 4b and 5b) have significantly lower values than the
inactive compounds.
The small number of compounds that comprise this series and
the high number of analysed parameters required us to perform an
initial Principal Components Analysis and a subsequent contin-
gency and correlation analysis [37] (data not shown for the sake of
brevity). This process allowed us to detect those descriptors that are
likely to be more important in establishing structure-activity
relationships.
The selected descriptors, i.e., those for which we obtained the
best correlation values, are shown in Table 4.
The data concerning the atomic charges for Se and C scaffold
atoms and the SeeC b.o. are included for comparative purposes
[26]. It is necessary to bear in mind that the structural modifica-
tions proposed in the design cycle that gave rise to the present
derivatives were carried out using a selection of rings chosen prior
to assessing the later parameters. Structures were chosen that had a
b.o. between 0.8320 and 0.9061, with the exceptions being com-
pound 5b, with a mean b.o. of 0.9449, and compound 6b, with a
mean b.o. of 0.9539. However, 5b was active in both cell lines
whereas 6b, as expected, was inactive. These results, along with the
MeSeH release data discussed previously, allowed us to confirm our
starting hypothesis, according to which a lower b.o. value implies a
more probable and effective release of the active agent.
The energy and location of the frontier orbitals (HOMO 0 and
LUMO 0) were calculated by applying the PM3 Hamiltonian
In a similar way, the PEOE_VSA_HYD descriptor, which quan-
tifies the total hydrophobic van der Waals surface area based on the
Table 4
Selected descriptors.
ASAc
ASA_Hc
PEOE_VSA_HYDc
vdw_vold
VSAc
a
a
b
Ref
pGI50 HT-29
pGI50 MCF-7
QSe
QC1
b.o.
EHOMO
BalabanJ
1b
2b
3b
4b
5b
6b
7b
8b
9b
10b
R2e
R2f
7.3010
6.4685
5.3098
6.2007
6.1024
2.2503
3.2757
3.6990
3.3372
2.1175
4.5229
5.4685
4.4318
6.1192
6.0458
2.2950
3.0044
3.1805
3.3098
1.1312
0.089
0.093
0.092
0.132
0.106
0.093
0.171
0.057
0.0520
0.1220
ꢁ0.117
ꢁ0.110
ꢁ0.108
ꢁ0.116
ꢁ0.128
ꢁ0.114
ꢁ0.104
ꢁ0.128
ꢁ0.126
ꢁ0.113
0.84854
0.83242
0.83248
0.84612
0.94485
0.95391
0.89594
0.90659
0.87491
0.85379
ꢁ9.1990
ꢁ9.1283
ꢁ9.0205
ꢁ9.7555
ꢁ9.5459
ꢁ8.9056
ꢁ9.1083
ꢁ9.2277
ꢁ9.03321
ꢁ8.68594
0.4372
1.7556
1.7556
1.8230
1.7836
1.7556
1.2516
1.2467
1.2429
1.0986
1.1499
0.8308
0.7377
498.4686
518.7032
546.5300
578.4913
482.0561
645.5012
607.7132
660.4120
857.7306
730.5948
0.5763
375.7461
305.7778
346.8099
196.4080
392.7048
481.5653
481.1737
575.5024
789.5089
665.9265
0.5497
210.3316
232.7572
267.9200
164.2993
178.6266
307.6725
278.6299
311.2517
452.2481
386.1669
0.6074
334.4484
357.8914
389.8033
405.6955
314.0371
501.9259
462.7507
524.9222
754.8700
668.9567
0.6168
285.5584
303.7578
334.6382
351.2402
273.8548
398.3111
368.6017
406.3987
564.9666
495.2197
0.5877
0.7177
0.4528
0.5679
0.6210
0.5365
0.4841
a
Coulson type charges.
b
c
d
e
f
in eV, average value for the lowest energy conformers for each analysed compound.
2
ꢀ
in A .
3
ꢀ
in A .
R
R
2 correlation coefficient for HT-29 cytotoxicity assay.
2 correlation coefficient for MCF-7 cytotoxicity assay.

496
M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
Fig. 7. Orbital distribution (left: HOMO 0; right: LUMO 0) for representative active and inactive compounds (representative lowest energy conformation).
Partial Equalization of Orbital Electronegativities (PEOE) [37]
method to calculate the atomic partial charges, can be related to
the activity in both cell lines. In this case, the most active com-
pounds (1b, 2b, 4b and 5b) have lower values than the inactive
ones.
means of HS-GCeMS techniques, their ability to release the pro-
posed active cytotoxic agent MeSeH. Furthermore, a common
preference for the extended-type conformation has been high-
lighted and this structural arrangement facilitates access of the
nucleophilic agent, which causes the release of the selenomethyl
moiety. An O2H probe interaction map supports this conclusion.
The molecular topology, evaluated through the BalabanJ index,
has a significant relationship with the target activity. For the most
active compounds the BalabanJ index has values between 1.7556
and 1.8230, whereas the inactive compounds have a significantly
lower BalabanJ value.
3. Conclusions
The data obtained in this study confirmed our initial hypothesis
and allowed us to establish a preliminary correlation between the
structure of the compounds, their conformational behaviour, the
MeSeH release kinetics and the biological activity. In fact, for active
compounds such as 1b, 2b, 4b or 5b we have demonstrated, by
Another structural characteristic is the preferred location of the
frontier HOMO and LUMO orbitals and this can also be related to

M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
497
the cytotoxic activity. In the active compounds the orbital is located
over the scaffold moiety whereas in the inactive compounds the
orbital is displaced to one of the lateral heteroaryl rings.
out by applying the implemented Energy Minimize general strat-
2
ꢀ
egy with a root mean square gradient, rms of 0.001 kcal/mol/A as
completion criterion. Restraints and constraints were not applied.
The first minimized conformations obtained were considered as
the starting conformations for the conformational analysis.
The conformational search was carried out through a systematic
The data presented in this paper are currently being used in a
new design cycle for the development of novel compounds that can
modulate the release of MeSeH through successive structural var-
iations. The structural changes concern the molecular environment
surrounding the hydrolysis site and encourage a bias towards
extended conformations and a topology consistent with the data
obtained in the present study. In this approach the whole molecule
is considered as a pro-drug and the selected structural variations
represent the release regulator elements.
2
ꢀ
search strategy with an rms gradient of 0.001 kcal/mol/A , an en-
ergy window of 5 kcal and a maximum conformation number of
100. The resulting conformations were then optimized by applying
the MOPAC2009 engine, using the PM3 semi-empirical approach,
with the geometry optimized using an eigenvector following al-
2
ꢀ
gorithm (rms gradient of 0.001 kcal/mol/A ).
The resulting conformations for each compound were distrib-
uted into two families (‘folded’ and ‘extended’), with the distribu-
tion criterion taken as the maximum distance between the
geometric centres of the extreme rings.
4. Experimental
4.1. Head Space Gas Chromatography Mass Spectrometry (HS-GCe
MS) analysis
The Interaction Potentials map that predicts the preferred
location of an O2H probe was calculated for the lowest energy
conformation by applying the calculation method implemented in
the MOE2012.10 software suite, which is based on a force field
that incorporates van der Waals, charge and hydrogen bonding
terms. The contour map represents the interaction energy equal
to ꢁ5.5 kcal/mol.
HS-GCeMS was performed as described in the literature [28].
DiMeDiSe and MeSeH were used as standards. DiMeDiSe (Sigma, St.
Louis, MO, USA) was dissolved in methanol as a stock solution
(10 mM). MeSeH was generated in situ by reductive hydrolysis
induced by NaBH4 (Sigma, St. Louis, MO, USA) in HS-GCeMS vials:
2.4
mL of DiMeDiSe was dissolved in 1 mL of ethanol and 20 mg
The Electrostatic Feature Maps that predict the electrostatically
preferred locations of H-bond acceptor and H-bond donor sites
from the solutions of the PoissoneBoltzmann equation were
calculated over the representative lowest conformation selected for
each analysed BSeC by applying the calculation method imple-
mented in the MOE2012.10 software suite. The contour map rep-
resents the interaction energy equal to ꢁ2 kcal/mol.
NaBH4 was added. The vial was capped using a rubber septum with
a crimp cap.
The analysed compounds were prepared in the same vials by
dissolving a sample in 5% DMSO/H2O at a final concentration of
60 mM. The vial was capped as before and the mixture was kept at
room temperature. The analytical samples were taken from the
head space at 8, 16 and 24 h.
Chromatographic analyses were carried out on an Agilent 6890
gas chromatograph coupled to a 5973N quadrupole mass spec-
trometer (Agilent Technologies, Santa Clara, CA, USA). An HP-5MS
The calculated 2D and 3D descriptors (see supplementary
material for details) were selected from those available in the
MOE2012.10 suite, which is intended to cover a wide range of both
physical and topological properties as well as electrostatic and
quantum characteristics.
The value collected for the conformationally dependent de-
scriptors represents the average value calculated for all conforma-
tions of each compound.
In order to reduce the dimensionality of the set of selected
molecular descriptors, a Principal Components Analysis (PCA) was
applied, with a component limit equal to 3 and 80% of minimum
variance. A further contingence analysis was carried out on the
aforementioned descriptor set.
capillary column (30 m ꢀ 0.25 mm I.D.) coated with a 0.25
mm
film of stationary phase (5% phenyl methylsiloxane) was used. The
flow rate of the high purity helium carrier gas was 1 mL/min. The
split/split less inlet was operated in split mode, with a 50:1 split
ratio. The injector was held at 250 ꢂC and the transfer line to the
detector at 300 ꢂC.
The GC oven temperature was 60 ꢂC, isothermal for 1 min,
50 ꢂC ꢀ minꢁ1 to 140 ꢂC, then 1 min isothermal. The MS was
operated in electron impact mode (EI, 70 eV).
The GC/MS was first operated in SCAN mode to determine the
selected ion monitoring (SIM) parameters for the expected volatile
species DiMeDiSe and MeSeH; once these parameters had been
established, the GC/MS was operated in SIM mode, establishing the
conditions through the selection of the most abundant fragment
ions of standards and the relative intensity of each one, according
the characteristic Se isotope patterns of compounds containing one
or two Se isotopes. Thus, the SIM conditions were established for
MeSeH (m/z 80, 93, 95 and 96) and for DiMeDiSe (m/z 190, 188, 186,
175 and 160).
The atomic charges (Coulson type charges [39]) and the bond
orders were calculated by applying the PM3 semi-empirical
approach. The data collected represent the average values for the
conformations obtained. The locations of HOMO and LUMO orbitals
were calculated on the lowest energy conformation.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.ejmech.2013.06.001.
4.2. Molecular modelling methods
References
The calculations were performed on a Dell Precision 380 work-
station, provided with the software package MOE [31], and on an
SGI Virtu VS100 workstation, provided with MOPAC2009 [32]
software package.
The three-dimensional models of the studied compounds were
constructed according to three initial basic conformations in
the vacuum phase, using atoms and structural fragments from
the Builder module (MOE2012.10) and using the implemented
MMFF94x [38] force field. A preliminary optimization was carried
[1] P.D. Whanger, Selenium and its relationship to cancer: an update, Br. J. Nutr.
91 (2004) 11e28.
[2] M.P. Rayman, Food-chain selenium and human health: emphasis on intake, Br.
J. Nutr. 100 (2008) 254e268.
[3] M.P. Rayman, Selenium in cancer prevention: a review of the evidence and
mechanism of action, Proc. Nutr. Soc. 64 (2005) 527e542.
[4] H. Zeng, G.F. Combs Jr., Selenium as an anticancer nutrient: roles in cell
proliferation and tumor cell invasion, J. Nutr. Biochem. 19 (2008) 1e7.
[5] J. Lü, C. Jiang, Selenium and cancer chemoprevention: hypotheses integrating
the actions of selenoproteins and selenium metabolites in epithelial and non-
epithelial target cells, Antioxid. Redox Signal 7 (2005) 1715e1727.

498
M. Font et al. / European Journal of Medicinal Chemistry 66 (2013) 489e498
kinase 1/2 pathway and other cancer signaling genes, J. Nutr. 139 (2009)
[6] A.F. Amaral, K.P. Cantor, D.T. Silverman, N. Malats, Selenium and bladder
cancer risk: a meta-analysis, Cancer Epidemiol. Biomarkers Prev. 19 (2010)
1613e1618.
2407e2415.
[22] H. Zeng, M. Briske-Anderson, J.P. Idso, C.D. Hunt, The selenium metabolite
methylselenol inhibits the migration and invasion potential of HT1080 tumor
cells, J. Nutr. 136 (2006) 1528e1532.
[23] R. Zhao, F.E. Domann, W. Zhong, Apoptosis induced by selenomethionine and
methioninase is superoxide mediated and p53 dependent in human prostate
cancer cells, Mol. Cancer Ther. 5 (2006) 3275e3284.
[24] E. Ibáñez, D. Plano, M. Font, A. Calvo, C. Prior, J.A. Palop, C. Sanmartín, Syn-
thesis and antiproliferative activity of novel symmetrical alkylthio- and
alkylseleno-imidocarbamates, Eur. J. Med. Chem. 46 (2011) 265e274.
[25] E. Ibáñez, A. Agliano, C. Prior, P. Nguewa, M. Redrado, I. Gonzalez-Zubeldia,
D. Plano, J.A. Palop, C. Sanmartín, A. Calvo, The quinoline imidoselenocarba-
mate EI201 blocks the AKT/mTOR pathway and targets cancer stem cells
leading to a strong antitumor activity, Curr. Med. Chem. 19 (2012) 3031e
3043.
[26] M. Font, A. Zuazo, E. Ansó, D. Plano, C. Sanmartín, J.A. Palop, J.J. Martínez-Irujo,
Novel structural insights for imidoselenocarbamates with antitumoral activity
related to their ability to generate methylselenol, Bioorg. Med. Chem. 20
(2012) 5110e5116.
[27] A. Zuazo, D. Plano, E. Anso, E. Lizarraga, M. Font, J.J. Martínez-Irujo, Cyto-
toxic and proapototic activities of imidoselenocarbamate derivatives are
dependent on the release of methylselenol, Chem. Res. Toxicol. 25 (2012)
2479e2489.
[28] C. Gabel-Jensen, K. Lunoe, B. Gammelgaard, Formation of methylselenol,
dimethylselenide and dimethyldiselenide in in vitro metabolism models
determined by headspace GC-MS, Metallomics 2 (2010) 167e173.
[29] C.M. Weekley, J.B. Aitken, S. Vogt, L.A. Finney, D.J. Paterson, M.D. de Jonge,
D.L. Howard, I.F. Musgrave, H.H. Harris, Uptake, distribution, and speciation of
selenoamino acids by human cancer cells: X-ray absorption and fluorescence
methods, Biochemistry-US 50 (2011) 1641e1650.
[7] B.K. Dunn, E.S. Richmond, L.M. Minasian, A.M. Ryan, L.G. Ford, A nutrient
approach to prostate cancer prevention: the Selenium and Vitamin E Cancer
Prevention Trial (SELECT), Nutr. Cancer 62 (2010) 896e918.
[8] M.S. Stratton, A.M. Algotar, J. Ranger-Moore, S.P. Stratton, E.H. Slate, C.-H. Hsu,
P.A. Thompson, L.C. Clark, F.R. Ahmann, Oral selenium supplementation has no
effect on prostate-specific antigen velocity in men undergoing active sur-
veillance for localized prostate cancer, Cancer Prev. Res. 3 (2010) 1035e1043.
[9] A.M. Algotar, M.S. Stratton, F.R. Ahmann, J. Ranger-Moore, R.B. Nagle,
P.A. Thompson, E. Slate, C.H. Hsu, B.L. Dalkin, P. Sindhwani, M.A. Holmes,
J.A. Tuckey, D.L. Graham, H.L. Parnes, L.C. Clark, S.P. Stratton, Phase 3 clinical
trial investigating the effect of selenium supplementation in men at high-risk
for prostate cancer, Prostate 73 (2013) 328e335.
[10] K. El-Bayoumy, The negative results of the select study do not necessarily
discredit the selenium-cancer prevention hypothesis, Nutr. Cancer 61 (2009)
285e286.
[11] J.L. Barger, T. Kayo, T.D. Pugh, J.A. Vann, R. Power, K. Dawson, R. Weindruch,
T.A. Prolla, Gene expression profiling reveals differential effects of sodium
selenite, selenomethionine, and yeast-derived selenium in the mouse, Genes
Nutr. 7 (2012) 155e165.
[12] M. Suzuki, M. Endo, F. Shinohara, S. Echigo, H. Rikiishi, Differential apoptotic
response of human cancer cells to organoselenium compounds, Cancer Che-
mother. Pharmacol. 66 (2010) 475e484.
[13] A. Bhattacharya, Methylselenocysteine: a promising antiangiogenic agent for
overcoming drug delivery barriers in solid malignancies for therapeutic syn-
ergy with anticancer drugs, Expert Opin. Drug Del. 8 (2011) 749e763.
[14] K. Lunoe, C. Gabel-Jensen, S. Sturup, L. Andresen, S. Skov, B. Gammelgaard,
Investigation of the selenium metabolism in cancer cell lines, Metallomics 3
(2011) 162e168.
[15] H. Infante, S.P. Joel, E. Warburton, C. Hopley, R. Hearn, S. Jüliger, Investigation
of the selenium species distribution in a human B-cell lymphoma line by
HPLC- and GC-ICP-MS in combination with HPLCESIMS/MS and GC-TOFMS
after incubation with methylseleninic acid, J. Anal. At. Spectrom. 22 (2007)
888e896.
[16] S. Cuello, S. Ramos, R. Mateos, M.A. Martin, Y. Madrid, C. Camara, L. Bravo,
L. Goya, Selenium methylselenocysteine protects human hepatoma HepG2
cells against oxidative stress induced by tert-butyl hydroperoxide, Anal.
Bioanal. Chem. 389 (2007) 2167e2178.
[30] C.M. Weekley, J.B. Aitken, I.F. Musgrave, H.H. Harris, Methylselenocysteine
treatment leads to diselenide formation in human cancer cells: evidence from
X-ray absorption spectroscopy studies, Biochemistry-US 51 (2012) 736e738.
[31] Molecular Operating Environment (MOE), 2012.10; Chemical Computing Group
Inc., 1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7,
2012
[32] MOPAC MOPAC2009 J.P. Stewart, Stewart Computational Chemistry, Colorado
Springs, Co, USA, 2009, http://OpenMOPAC.net.
[33] P.J. Goodford,
A computational procedure for determining energetically
[17] H. Zeng, M. Briske-Anderson, M. Wu, M.P. Moyer, Methylselenol, a selenium
metabolite, plays common and different roles in cancerous colon HCT116 cell
and noncancerous NCM460 colon cell proliferation, Nutr. Cancer 64 (2012)
128e135.
favorable binding sites on biologically important macromolecules, J. Med.
Chem. 28 (1985) 849e857.
[34] A.T. Balaban, Highly discriminating distance-based topological index, Chem.
Phys. Lett. 89 (1982) 399e404.
[18] J. Lu, C. Jiang, Selenium and cancer chemoprevention, Antioxid. Redox Signal 7
(2005) 1715e1727.
[35] R.S. Pearlman, K.M. Smith, Novel software tools for chemical diversity, Persp.
Drug Discov. (1998) 339e353, 9/10/11.
[36] J. Gasteiger, M. Marsili, Iterative partial equalization of orbital electronega-
tivity e a rapid access to atomic charges, Tetrahedron 36 (1980) 3219e3228.
[37] R.V. Hogg, E.A. Tanis, Probability and Statistical Inference, Macmillan Pub-
lishing, New York, 1993.
[38] T.A. Halgren, MMFF VI. MMFF94s option for energy minimization studies,
J. Comput. Chem. 20 (1999) 720e729.
́
[19] C. Ip, H. Thompson, Z. Zhu, H. Ganther, In vitro and in vivo studies of meth-
ylseleninic acid: evidence that a monomethylated selenium metabolite is
critical for cancer chemoprevention, Cancer Res. 60 (2000) 2882e2886.
[20] Z. Wang, C. Jiang, J. Lu, Induction of caspase-mediated apoptosis and cell-cycle
G1 arrest by selenium metabolite methylselenol, Mol. Carcinog. 34 (2002)
113e120.
[21] H. Zeng, M. Wu, J.H. Botnen, Methylselenol, a selenium metabolite, induces
cell cycle arrest in G1 phase and apoptosis via the extracellular-regulated
[39] J. Meister, W.H.E. Schwarz, Principal components of ionicity, J. Phys. Chem.-US
98 (1994) 8245e8252.