Supramolecular synthon pattern in solid clioquinol and cloxiquine (APIs of antibacterial, antifungal, antiaging and antituberculosis drugs) studied by ³⁵Cl NQR, ¹H-¹⁷O and ¹H-¹⁴N NQDR and DFT/QTAIM.

Abstract

The quinolinol derivatives clioquinol (5-chloro-7-iodo-8-quinolinol, Quinoform) and cloxiquine (5-chloro-8-quinolinol) were studied experimentally in the solid state via ³⁵Cl NQR, ¹H-¹⁷O and ¹H-¹⁴N NQDR spectroscopies, and theoretically by density functional theory (DFT). The supramolecular synthon pattern of O-H···N hydrogen bonds linking dimers and π-π stacking interactions were described within the QTAIM (quantum theory of atoms in molecules) /DFT (density functional theory) formalism. Both proton donor and acceptor sites in O-H···N bonds were characterized using ¹H-¹⁷O and ¹H-¹⁴N NQDR spectroscopies and QTAIM. The possibility of the existence of O-H···H-O dihydrogen bonds was excluded. The weak intermolecular interactions in the crystals of clioquinol and cloxiquine were detected and examined. The results obtained in this work suggest that considerable differences in the NQR parameters for the planar and twisted supramolecular synthons permit differentiation between specific polymorphic forms, and indicate that the more planar supramolecular synthons are accompanied by a greater number of weaker hydrogen bonds linking them and stronger π···π stacking interactions.

Full text

Page 1

ORIGINAL PAPER
Supramolecular synthon pattern in solid clioquinol
and cloxiquine (APIs of antibacterial, antifungal, antiaging
and antituberculosis drugs) studied by35Cl NQR,1H-17O
and1H-14N NQDR and DFT/QTAIM
Jolanta Natalia Latosińska & Magdalena Latosińska &
Marzena Agnieszka Tomczak & Janez Seliger &
Veselko Žagar
Received: 18 April 2010 /Accepted: 11 October 2010 /Published online: 16 November 2010
# The Author(s) 2010. This article is published with open access at Springerlink.com
Abstract The quinolinol derivatives clioquinol (5-chloro-
7-iodo-8-quinolinol, Quinoform) and cloxiquine (5-chloro-
8-quinolinol) were studied experimentally in the solid state
via35Cl NQR,1H-17O and1H-14N NQDR spectroscopies,
and theoretically by density functional theory (DFT). The
supramolecular synthon pattern of O–H···N hydrogen
bonds linking dimers and π–π stacking interactions were
described within the QTAIM (quantum theory of atoms in
molecules) /DFT (density functional theory) formalism.
Both proton donor and acceptor sites in O–H···N bonds
were characterized using
spectroscopies and QTAIM. The possibility of the existence
of O–H···H–O dihydrogen bonds was excluded. The weak
intermolecular interactions in the crystals of clioquinol and
cloxiquine were detected and examined. The results
obtained in this work suggest that considerable differences
in the NQR parameters for the planar and twisted
supramolecular synthons permit differentiation between
specific polymorphic forms, and indicate that the more
planar supramolecular synthons are accompanied by a
1H-17O and
1H-14N NQDR
greater number of weaker hydrogen bonds linking them
and stronger π···π stacking interactions.
Keywords Clioquinol.Cloxiquine.Nuclear quadrupole
resonance.Antiaging.Anticancer.Antituberculosis.
Antibacterial.Antifungal.DFT.QTAIM.Intermolecular
interactions.Supramolecular synthon.Polymorphism
Introduction
The quinolinol derivatives clioquinol (5-chloro-7-iodo-8-
quinolinol, Quinoform, Quinambicide, Vioform) and clox-
iquine (5-chloro-8-quinolinol, Chloroxychinolin, Cloxi-
quine, Dermofongin A) Fig. 1 are active pharmaceutical
ingredients (APIs) of a wide spectrum of known potent
antibacterial, antifungal and antiamoebic agents used in the
treatment of dermatoses [1–3] and antiseptic or disinfectant
formulations [4] that have been known for decades. Both
are members of the group of drugs called 8-quinolinols
which inhibit DNA replication and are active against both
viral and protozoal infections [5]. Clioquinol is also used to
treat diarrhea and other gastrointestinal disorders, skin
infections such as eczema, athlete’s foot, jock itch,
ringworm, and some bacteria (Staphylococcus, Streptococ-
cus, E. coli), yeasts (Candida albicans), and some
protozoan parasites (particularly Trichomonas sp. [6, 7]),
and recently tuberculosis [8]. On the other hand, both
clioquinol and cloxiquine are common causes of epigastric
discomfort, contact dermatitis and neuropathy, and both are
considered mutagens. Clioquinol produces not only allergic
reactions, eosinophilia and hyperthyreosis, but it also
exhibits teratogenic and carcinogenic effects, and is the
J. N. Latosińska (*):M. Latosińska:M. A. Tomczak
Faculty of Physics, Adam Mickiewicz University,
Umultowska 85,
61-614 Poznań, Poland
e-mail: Jolanta.Latosinska@amu.edu.pl
J. Seliger:V. Žagar
“Jozef Stefan” Institute,
Jamova 39,
1000 Ljubljana, Slovenia
J. Seliger
Faculty of Mathematics and Physics, University of Ljubljana,
Jadranska 19,
1000 Ljubljana, Slovenia
J Mol Model (2011) 17:1781–1800
DOI 10.1007/s00894-010-0876-4

Page 2

most toxic of the antibacterial compounds which can
depress the central nervous system. It was withdrawn from
use in the 1970s due to serious adverse events like blindness,
paralysis or death [9]. The increasing interest in both
compounds stems from the recent finding that cloxiquine
exhibits good antituberculosis activity, even for multidrug
resistant (MDR) isolates [10], whereas clioquinol, which is
known to be extremely neurotoxic (on account of its ability
to chelate metals [11]) in large doses, and is one of the agents
that leads to lethal subacute myelo-optico-neuropathy
(SMON) [12, 13] in small doses, has been found to be
capable of reversing the progression of neurodegenerative
disorders. The latter effect is probably due to the action
directed at the protein called CLK-1 (“clock-1”), and thus
helps to suppress the initiation of Alzheimer’s, Parkinson’s
and Huntington’s diseases by slowing aging [14–17] due to
its free radical scavenging capabilities. Very recently,
clioquinol was found to inhibit proteasomes, display preclin-
ical activity in leukemia and myeloma [18], and to exert
anticancer effects both in vitro and in vivo [19]. Unfortu-
nately, the mechanism of its action has not been identified
yet, although it is believed to be related to the iodine content.
Considering the above discussion, and the differences in
their biological activities, a comparison of the structural and
electronic properties of clioquinol and cloxiquine (which
differs from clioquinol only in its lack of iodine at the 7
position of the quinoline ring) appears to be a very
promising line of research, and therefore deserves to be
the focus of detailed studies. It is known that the biological
activities of compounds are related to their chemical
structures, especially their electron density distributions
and bonding capabilities [20, 21]. Ever since the first
experiments in solid-state nuclear quadrupole resonance
(NQR),thegreat potential ofthismolecule-specific method—
which allows the nondestructive characterization of solid
pharmaceutical products—for the analysis of biological
systems has been recognized [22–25]. The electric field
gradient (EFG) tensor depends on the positions and charges
of the nuclei and electrons about the quadrupolar nucleus,
and thus the quadrupole coupling constant—which is the
largest (in absolute value) principal component of the EFG
tensor multiplied by the nuclear quadrupole moment and
divided by Planck’s constant—reflects the electron distribu-
tion in the vicinity of the quadrupolar nuclei, making it a
very sensitive tool for investigating molecular and crystal
structure in detail. NQR appears to be the optimal method
for studying clioquinol and cloxiquine, because—rather
unusually—both compounds contain three kinds of quad-
rupolar nuclei in their molecules:14N,17O and35Cl (indeed,
clioquinol actually contains four, including127I). In order to
elucidate these details and to aid our understanding of the
differences in the biological activities of both compounds,
we performed a joint study that applied NQR for different
isotopes (14N,17O and35Cl), as well as density functional
theory (DFT), which reveals the local and global electron
density distribution in the molecules. We expect that this
combined study will permit a detailed understanding of the
differences in the structural features of clioquinol and
cloxiquine, and contribute to an explanation of the role of
clioquinol at the molecular level, especially the functional
implications of iodine substitution and O–H···N hydrogen
bonding formation for the recognition and binding of
clioquinol molecules to the mitochondrial enzyme CLK-1
(also known as COQ7).
Experimental
High-purity polycrystalline samples of clioquinol and clox-
iquine (95% and 97%, respectively) were purchased from
Sigma–Aldrich and used without any additional purification.
NQR spectroscopy
The35Cl,17O and14N nuclei have spins I=3/2, 5/2 and 1,
respectively, and therefore, when there is no external
magnetic field,35Cl exhibits two doubly degenerate,17O
exhibits three doubly degenerate, and14N exhibits nonde-
generate nuclear quadrupole energy levels. Their energies
depend on the nuclear quadrupole moment eQ and on the
electric field gradient (EFG) tensor Vik¼ @2V=@xi@xk,
which consists of the second derivatives of the electrostatic
potential V with respect to coordinates originating at the
position of the nucleus. The symmetric traceless second-
rank EFG tensor has three principal values: VZZ= eq, VYY
and VXX VZZ
j
obtain two unique NQR parameters: the nuclear quadrupole
coupling constant (e2Qqh−1) and the asymmetry parameter
(η), which are related to the NQR frequencies (ν) through
the following equations [26]:
jo > VYY
jj ? jVXXjðÞ, which are used to
a) For35Cl, the frequencies do not uniquely depend on the
quadrupole coupling constant e2qQ/h and the asymme-
try parameter η:
nð35ClÞ ¼e2Qq
2h
but for biologically active compounds η << 0.1
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ h2=3
p
;
ð1Þ
Fig. 1 The molecular structures
of 8-quinolinol derivatives
(R=I in clioquinol, R=H
in cloxiquine)
1782J Mol Model (2011) 17:1781–1800

Page 3

b) For17O, the three NQR frequencies, which are usually
termed v5=2?1=2> v5=2?3=2? v3=2?1=2, uniquely depend
on e2qQ/h and η, but can be calculated via the
following secular equation:
x3? 7ð3 þ h2Þx ? 20ð1 ? h2Þ ¼ 0;
where x is a solution of Eq. 2, and the energies of the
NQR levels can be calculated using the formula E ¼
e2qQh?1=20
c) For14N, the three NQR frequencies are usually termed
vþv?? v0and uniquely depend (in a similar manner to
those for
e2qQ/h and the asymmetry parameter η:
ð2Þ
??x
17O) on the quadrupole coupling constant
nþð14NÞ ¼e2Qq
n?ð14NÞ ¼e2Qq
4h
ð3 þ hÞ
4h
ð3 ? hÞ
n0ð14NÞ ¼ nþð14NÞ ? n?ð14NÞ ¼e2Qq
2h
h
ð3Þ
The natural abundance of the
(75.4%), and the NQR frequencies are typically between
30 and 40 MHz, so it is possible to use the pure NQR
experimental technique. The natural abundance of the14N
isotope is very high (99.636%), but the NQR signals are
weak, and the NQR frequencies are typically between 0.5
and 4 MHz, so the use of the1H-14N NQDR technique is
preferred to the use of pure NQR. The natural abundance of
the
frequencies are typically below 5 MHz, so the use of
several1H-17O NQDR techniques is required in this case.
35Cl isotope is high
17O isotope is rather low (0.037%), and the NQR
1H-14N-NQDR
Different double resonance techniques based on magnetic
field cycling were used to detect14N NQR frequencies. The
proton spin system was polarized in B0=0.75 T for 30 s.
Then the sample was pneumatically transferred into another
magnet within 0.1 s, where it was left for 0.3 s. In this other
magnet, the magnetic field can be varied continuously
between zero and 0.1 T. After its stay in this other magnet,
the sample was pneumatically transferred back into the first
magnet within 0.1 s, and the proton NMR signal was
measured immediately after the sample had been stopped in
the first magnet.
We used1H-14N cross-relaxation spectroscopy [27–29]
as a first method. In this method, the sample is left to relax
in a low magnetic field for a fixed time τ (in our case τ =
0.5 s), and the low magnetic field is varied between the
magnetic field cycles in steps of approximately 0.5 mT,
corresponding to a step in the proton Larmor frequency νL
of 20 kHz. A proton Larmor frequency range of between 0
and 4 MHz is usually scanned by this technique. When the
proton Larmor frequency νLmatches a14N NQR frequency
νQ, the proton spin-lattice relaxation time shortens, which
results in a decrease in the proton NMR signal after the
cycle. In some cases, especially at higher proton Larmor
frequencies, a step of 40 kHz can be used. On the other
hand, around νL=νQ, the step is reduced to 10 kHz to
improve the resolution.
In a second step, we used the solid-effect technique [30].
In this method, the low magnetic field was fixed at a value
B, corresponding to the proton Larmor frequency
vL¼ gHB=2p, and the sample was irradiated in the low
magnetic field with a strong rf magnetic field for 0.5 s at
variable frequencies. When the frequency ν of the rf
magnetic field is equal to ν=νQ±νL, simultaneous spin
flips take place in both the1H and the14N spin systems. As
a result, the proton magnetization decreases. The experi-
ment is repeated at a few low magnetic field values to
clarify the spectrum and get rid of signal artefacts caused by
the direct proton absorption of the rf power at multiples of
the proton Larmor frequency and the level crossing signals
produced by the higher harmonics of the rf magnetic field.
We then used the two-frequency irradiation technique as a
final technique combining the three14N NQR frequencies
from a given nitrogen site [31]. Here, the proton Larmor
frequency νLin the low magnetic field is set in resonance
with the lowest14N NQR frequency ν0, and the sample is
irradiated with two rf magnetic fields at the frequencies ν1=ν
and ν2=ν +ν0. When ν1= ν−and v2¼ v?þ v0¼ vþ, the
proton relaxation rate in the low magnetic field increases
and, as a result, the proton NMR signal at the end of the
magnetic field cycle drops to a low value. This technique is
applied when attempting to solve complex14N NQR spectra
in order to help distinguish between triplets corresponding to
various nitrogen positions in the crystal.
1H-17O NQDR
The
measured by the Slusher and Hahn technique [32]. The
sample was pneumatically moved between two magnets
with magnetic fields of B0=0.75 T and zero. The proton
spin system was polarized in the high magnetic field B0for
30 s and then transferred to the zero magnetic field for
0.8 s. Then the sample was transferred back into the first
magnet and the intensity of the proton NMR signal was
measured immediately after the sample stopped in the high
magnetic field. In the static zero magnetic field, the sample
was irradiated with a phase-modulated rf magnetic field of
frequency ν and amplitude ~3 mT. Square-wave 180° phase
17O NQR frequencies in 5-chloroquinol were first
J Mol Model (2011) 17:1781–18001783

Page 4

modulation was used with a frequency of 1.7 kHz. The
frequency ν was changed between the repetitive magnetic
field cycles in steps of 20 kHz. The frequency range
between 1.0 and 5.0 MHz was scanned by the frequency ν.
Two dips corresponding to the17O NQR frequencies ν3/2-1/2
and ν5/2-3/2 are usually observed by this technique. The
third dip at the highest17O NQR frequency ν = ν5/2-1/2has
(as shown in [33]) a much lower intensity than the other
two dips, and is usually not observed by the Slusher and
Hahn technique.
During the second part of the experiment we used the
two-frequency irradiation technique [34] to determine the
dipolar structure of the
magnetic fields of frequencies ν1and ν2were applied in
repetitive pulses at frequencies ν1, ν2, ν1, ν2... The duration
of a pulse was 1 ms. The amplitude of the rf magnetic field
during the measurement of the dipolar structure of the
NQDR lines at ν=ν3/2-1/2and ν=ν5/2-3/2was reduced to
0.3 mT. The dipolar structure of the NQDR line with the
highest frequency was measured with an rf magnetic field
of amplitude 3 mT, due to the lower transition probability
per unit time.
The dipolar structure of a NQDR line was determined in
two experiments. The first of the two frequencies, say ν1,
was fixed at the lower edge of the NQDR line, and the line
was scanned at the second frequency ν2in steps of 5 kHz.
No drop in the proton NMR signal was observed when ν1=
ν2. The strongest drop in the proton NMR signal was
observed when ν2was in the upper part of the NQDR line.
To determine the dipolar structure of the lower part of the
NQDR line, the experiment was repeated with the frequen-
cy ν1fixed at the upper edge of the NQDR line. The dipolar
structures of the three NQDR lines determined by two-
frequency irradiation were analyzed according to [35].
1H-17O NQDR lines. Two rf
35Cl-NQR
The35Cl-NQR spectra of clioquinol and cloxiquine were
taken at 77 K. The NQR signals assigned to Cl nuclei were
weak (S/N=3 after 1000 accumulations), and the resonance
line was wide [full width at half maximum (FWHM) was
27 kHz], so the classical Hahn sequence π- τ-2π was
applied. In the NQR of a powdered sample, both excitation
and signal reception depend on the relative orientations of
the crystallites with respect to the coil axis, so the pulse
sequence was optimized; the optimized pulse length was 5
μs and the interval between the pulses was 90 μs. The NQR
lineshape was obtained from the fast Fourier transform
(FFT) of both half-echo signals after 1000 accumulations
for the desired signal-to-noise ratio. The repetition time of
the scans was 200 ms. The accuracy of the
frequency determination was ~10 kHz.
35Cl-NQR
DFT calculations
Quantum chemical calculations were carried out within the
GAUSSIAN03™ code [36], which was run on the CRAY
supercomputer at the Poznan Supercomputer and Network
Centre (PCSS) in Poznan, Poland. All calculations were
performed within the density functional theory (DFT) with
a hybrid exchange-correlation functional, B3LYP (the
three-parameter exchange functional of Becke B3 [37]
combined with the Lee–Yang–Parr correlation functional
LYP [38]), using the extended basis set 6-311++G** with
polarization and diffuse functions. The calculations were
carried out under the assumption of the crystallographic as
well as the partially optimized geometry, where only the
positions of the hydrogen atoms were allowed to relax
during optimization (as performed using the Berny algo-
rithm), while those of all other atoms remained frozen. The
NQR parameters—quadrupole coupling constants, asym-
metry parameters and frequencies at all nitrogen atoms—
were calculated assuming different polymorphic forms,
which differed in the molecular aggregations formed as a
result of intermolecular interactions.
Theoretical analysis of the intermolecular interactions
was performed according to the quantum theory of atoms in
molecules theory (QTAIM) [39], and the topological
parameters were calculated, including the bond critical
points (BCPs), ring critical points (RCPs), Laplacian of the
electron density (Δρ), ellipiticity of the bond (ε), the total
electron energy density at the BCP (HBCP) and its
components: the local kinetic energy density (GBCP) and
the local potential energy density (VBCP). Because of the
dependence of the basis on atomic position (i.e., the basis
set superposition error, BSSE), small interaction energies
are often overestimated, so the interaction energies were
corrected for BSSE by the standard counterpoise (CP)
method [40]. Another correction was made for the zero
point vibrational energies (ZPVEs). However, this approach
is not useful for intramolecular or multiple hydrogen bonds,
so the energy of the interactions was calculated according
to Espinosa [41].
Results and discussion
The
presented in Fig. 2. Only one resonance line is observed in
each, at 35.170 MHz for clioquinol and at 34.787 MHz for
cloxiquine at 77K. All of the molecules are thus crystallo-
graphically equivalent in each of these compounds, which
is in agreement with the X-ray data [42–45].
The1H-14N NQDR spectra of clioquinol and cloxiquine
as obtained by the solid-effect technique at T=295 K are
presented in Fig. 3. Seven NQDR lines are unambiguously
35Cl-NQR spectra of clioquinol and cloxiquine are
1784J Mol Model (2011) 17:1781–1800

Page 5

resolved in each spectrum. However, as shown in [29], the
intensities of the NQDR lines differ strongly when the
population of the energy levels of the nitrogen spin system
approaches the Boltzmann population more slowly than
that of the energy levels of the proton spin system. A triplet
is usually observed at ν = ν+−νL, ν+and ν++νL. The solid-
effect lines ν = ν0−νLand ν = ν−−νLare usually missing.
With this in mind, and utilizing the previously recorded
cross-relaxation spectra, we can determine the
frequencies as 3.330 MHz, 2.830 MHz and 0.500 MHz for
cloxiquine and 3.410 MHz, 2.830 MHz and 0.580 MHz for
clioquinol. The accuracy is within ±10–20 kHz. A more
precise determination of the
accuracy of ±2 kHz was obtained using the two-frequency
irradiation technique. The
3.328 MHz, 2.828 MHz and 0.500 MHz for cloxiquine
and 3.407 MHz, 2.832 MHz and 0.575 MHz for clioquinol
are not far from the frequencies determined from the solid-
14N NQR
14N NQR frequencies to an
14N NQR frequencies of
Fig. 2 The35Cl-NQR spectra of clioquinol and cloxiquine at T=77 K
Fig. 3 The1H-14N solid-effect double-resonance spectra of clioquinol
and cloxiquine at T=295 K. The proton Larmor frequency is νL=
100 kHz
Fig. 4a–b The1H-17O double resonance spectra of cloxiquine at T=
213 K, as measured by the Slusher and Hahn technique (a) and by the
two-frequency irradiation technique (b)
J Mol Model (2011) 17:1781–1800 1785

Page 6

effect spectra. All nitrogen positions are crystallographically
equivalent in cloxiquine as well as in clioquinol.
The
cloxiquine. Due to the low natural abundance of
(0.037%), the sensitivity of the NQDR technique strongly
depends on the proton spin-lattice relaxation time in zero
magnetic field, which should be about 1 s or more to
observe the NQDR dips. For cloxiquine we obtained a
sufficiently long proton spin-lattice relaxation time in zero
magnetic field by reducing the temperature of the sample to
213 K. For clioquinol we also varied the temperature of the
sample, but within the working range of the spectrometer
(130–400 K) we did not observe a sufficiently long proton
T1in zero magnetic field. This is presumably the effect of
the contribution of127I to the spin-lattice relaxation of the
proton dipolar system.
The1H-17O NQDR spectrum of cloxiquine as obtained
by the Slusher and Hahn technique is presented in Fig. 4a.
Two NQDR dips are observed at frequencies of 2.425 and
1.750 MHz. These two dips correspond to the17O NQR
frequencies ν5/2-3/2and ν3/2-1/2, respectively. The third dip
expected at the frequency v5=2?1=2¼ v5=2?3=2þ v3=2?1=2¼
4:175MHz is too weak to be observed by the Slusher and
Hahn technique.
The dipolar structure of the three
measured by the two-frequency irradiation technique, is
presented in Fig. 4b. The 5/2–1/2 transition is also observed
by this technique. The dipolar structure of the three
NQR lines is not well resolved due to the dipolar proton–
proton interaction. Nevertheless, it is still possible to
determine the proton–oxygen distance plus the polar angle
θ and the azimuthal angle ϕ describing the orientation of
the O–H bond in the principal-axis frame of the EFG tensor
1H-17O NQDR spectrum was only measured for
17O
17O NQR lines, as
17O
at the position of the oxygen nucleus from the widths of
the dipolar split
follows: the O–H distance is 0.99±0.01 Å; the O–H bond
lies in the xz plane of the EFG tensor (’=0°) and makes an
angle θ=50°5° with the principal axis z of the EFG tensor.
The experimental values of the14N,17O and35Cl NQR
frequencies, together with the values of e2Qqh−1and η
calculated using these frequencies, are collected in Table 1
for both compounds. As follows from these results, all of
the clioquinol as well as the cloxiquine molecules in the
elementary cell are equivalent within the resolution of the
experiment, which is in good agreement with the X-ray
data, according to which clioquinol crystallizes in the
monoclinic system P2/c or P2/a [42, 43] while cloxiquine
crystallizes in the orthorhombic system Fdd2 (form I) or the
monoclinic system P2/c (form II) [44, 45].
Parameters of the elementary cells of form 1 of
cloxiquine at 90 K [45] and at RT [44], as determined by
crystallographic methods, differ insignificantly (by 92.5 Å3
in volume; V(90 K)=3032.5 and V(RT)=3126 Å3), which
suggests that a phase transition does not take place over this
range of temperatures. Because there is only one type of
each site (nitrogen, oxygen, chlorine or iodine) in clioqui-
nol and cloxiquine, assigning the NQR frequencies to
particular sites is a straightforward task. The differences in
the reproduced NQR parameters for forms I and II of
cloxiquine (see Table 2) are large enough to allow the
unambiguous identification of the present polymorphic
form as form I. This conclusion is based on two factors:
significant differences between e2Qq/h and η, especially at
the
importantly, the much better correlation between the
experimental and DFT-calculated
17O NQR lines [34]. The results are as
17O site for both forms (see Table 2), and, more
14N and
17O NQR
Table 1 The experimental NQR parameters for clioquinol and
cloxiquine (35Cl NQR frequency,14N NQR frequencies ν+, ν−and
ν0,17O NQR frequencies ν3/2-1/2, ν3/2-1/2and ν3/2-1/2, the widths δν3/2-
1/2, δν3/2-1/2and δν3/2-1/2of the NQDR lines, the quadrupole coupling
constants e2qQ/h, and the asymmetry parameters η of the EFG tensor,
the proton–oxygen distance R(O–H), the angle θ between the O–H
bond and the principal axis z of the EFG tensor, and the angle ϕ
between the projection of the O–H bond on the xy plane and the
principal axis x of the EFG tensor)
CompoundSite
ν in MHz (δν in kHz)e2Qqh−1(MHz)
η
R(O–H) (Å)
θϕ
T (K)
CloxiquineN3.328
2.828
0.500
4.175 (120)
2.425 (120)
1.750 (110)
34.787 (21)
3.407
2.831
0.575
35.170 (22)
4.1040.244
–––
295
O8.610 0.6220.9950° 0°213
Cl
N
69.574a
4.159
0
0.276
–
–
–
–
–
–
77
295Clioquinol
Cl71.420a
0
–––
77
aCalculated under the assumption η=0
1786J Mol Model (2011) 17:1781–1800

Page 7

frequencies (for form I, the correlation coefficient and curve
fit standard errors are 0.997 and 0.025 MHz at RT, and
0.9956 and 0.036 MHz at 90 K, but for form II they are
only 0.986 and 0.105 MHz); see Fig. 5a. It is worth noting
that the NQR frequency at the
Although it is more sensitive to the influence of tempera-
ture than those at the17O and14N sites, the35Cl site is
35Cl site is less useful.
distinct and thus less sensitive to structural changes in the
stacked dimers.
Surprisingly, the NQR parameters for clioquinol are
better reproduced when assuming the X-ray structure from
[43], denoted form Ia, rather than that from [42], denoted
form Ib (see Tables 3 and 4), which suggests that the first
structure is better resolved. However, the residual factor
Table 2 The NQR parameters calculated at the B3LYP/6-311++G(d,p) level of theory for different polymorphic forms of clioquinol and
cloxiquine
Compound FormSite Monomer DimerStacked dimers
ν+,ν−, ν0
ν5/2-1/2,ν5/2-3/2,ν3/2-1/2
ν(MHz)
e2Qqh−1
(MHz)
ην+,ν−, ν0
ν5/2-1/2,ν5/2-3/2,ν3/2-1/2
ν(MHz)
e2Qqh−1
(MHz)
ην+,ν−, ν0
ν5/2-1/2,ν5/2-3/2,ν3/2-1/2
ν(MHz)
e2Qqh−1
(MHz)
η
Cloxiquine Form I RT [44] opta
N 3.969
3.285
0.684
5.116
2.907
2.208
33.634
4.007
3.310
0.698
4.838
2.730
2.108
34.005
3.884
3.202
0.683
4.802
2.547
2.256
34.038
4.06
3.41
0.64
5.189
2.866
2.424
34.372
889.136
585.596
303.540
4.06
3.53
0.52
5.033
2.834
2.199
35.335
897.710
596.647
301.062
4.8360.283 3.6074.457 0.237 3.6104.4600.238
3.079
0.528
3.080
0.530
O10.424 0.680 4.717 9.7760.599 4.732 9.7660.614
2.764
1.953
2.757
1.974
Cl
N
67.074
4.878
0.093 33.663
0.286 3.860
67.145
4.741
0.090 33.881
0.257 3.881
67.611
4.758
0.082
0.263
Form I 90 K [45] opta
3.251
0.609
3.256
0.626
O 9.817 0.699 4.6219.4530.662 4.6359.452 0.676
2.644
1.977
2.639
1.996
Cl
N
67.859
4.724
0.082 33.987
0.289 3.666
67.826
4.500
0.081 34.224
0.259 3.706
68.330
4.537
0.072
0.267
Form II 90 K [45] opta
3.084
0.583
3.100
0.606
O 9.4020.855 4.5688.975 0.840 4.5288.9810.834
2.437
2.131
2.405
2.123
Cl
N
67.928
4.979
0.081 34.030
0.259 3.71
67.918
4.621
0.079 34.195
0.210 3.69
68.273
4.602
0.072
0.206
Clioquinol Form I [43] opta
3.22
0.49
3.21
0.47
O 10.4030.754 4.830 9.915 0.646 4.8369.8860.665
2.781
2.049
2.765
2.072
Cl
I
68.566
1971.039 0.104 886.267
0.088 34.41368.657
1964.244 0.109 894.756
0.086 34.589 69.012
1985.258 0.082
0.085
587.800
298.467
594.783
299.972
Form I [42]opta
N5.062 0.207 3.704.6920.154 3.694.680 0.152
3.34
0.36
3.33
0.36
O10.2020.704 4.6749.673 0.605 4.686 9.6540.628
2.732
1.941
2.716
1.969
Cl
I
70.551
1991.812 0.084 895.961
0.071 35.382 70.655
1987.492 0.089 903.91
0.068 35.539 70.975
2007.117 0.060
0.066
597.043
300.708
601.23
302.673
aPartially optimized geometry
J Mol Model (2011) 17:1781–18001787

Page 8

given by the authors for this structure is only 10%, whereas
it is 8.4% for the second structure. The NQR parameters
e2Qqh−1and η and the frequencies at all quadrupolar nuclei
were calculated at the B3LYP/6-311++G(d,p) level using
X-ray data for clioquinol and cloxiquine (for both poly-
morphic forms) and assuming different molecular aggrega-
tions (monomer, dimer, stacked dimers) are formed by the
intermolecular interactions. The results are collected in
Table 2. The orientations of the principal axes of the EFG
tensors at each site are shown in Fig. 6a–c. Good
reproduction accuracy of e2Qqh−1and η and the NQR
frequencies at all quadrupolar nuclei (14N,17O and35Cl) at
the chosen DFT level (correlation coefficients as high as
0.999, and curve fit standard errors as low as 0.38 MHz)
requires the assumption of the optimized proton positions
and consideration of the intermolecular bonding (see
Tables 2, 3 and 4, and Fig. 5b), which is in a good
agreement with the results of our previous studies of
purines [23, 24] or polyhalogenobenzimidazoles [25].
Besides e2Qqh−1and η, another criterion for checking the
quality of reproduction of the EFG tensor by DFT is to
compare the orientations of its principal axes as deduced
from DFT as well as the O–H bond length with those
obtained from the experimental17O spectrum. The orien-
tation of the z-axis of the EFG tensor determined from17O
for cloxiquine is in good agreement with the results of the
DFT calculations. DFT suggests that the O–H bond, which
lies in the xz plane of the EFG tensor according to the
NQDR results, deviates from the xz plane by only 3°, and
makes an angle of 44±1° with the z-axis of the EFG tensor,
while the experiment suggests θ = 50 ±5°. The source of
this slight discrepancy could be the neglect of other
interactions aside from the hydrogen bonding. The O–H
bond length, R(O–H), as determined by
cloxiquine, is R(O–H) = (0.99±0.02) Å (i.e., in good
agreement with the X-ray data at 90K [45] and the DFT
partial optimization result for the monomer, 0.987 Å, and
dimer, 1.011 Å). The O–H bond length for clioquinol
obtained from partial DFT optimization, R(O–H)=0.992 Å,
is also much higher than those given in [42], i.e., 0.728 Å.
The change in the ∠OHN angle describing the linearity of
the O–H···N bond influences the NQR parameters consid-
erably, even more than the changes in its length, which
explains why the NQR parameters for the structure with the
optimized proton positions is reproduced more accurately.
17O NQDR for
Structural pattern
Supramolecular synthon
According to the crystallographic data, the hydrogen
bonding patterns in solid clioquinol and cloxiquine (form
I) [42–45] are isostructural to those in the parent 8-
quiniolinol [46]. The hydroxyl hydrogens are capable of
forming multicenter bonds, i.e., bifurcated O–H···N hydro-
gen bonds, one intramolecular and the other intermolecular,
which simultaneously lead to the formation of five-
membered hydrogen-bonded chelate rings [N, C(9), C(8),
O, H(8)] and to the dimerization of the molecules,
respectively, as indicated in Fig. 7a–c. Such patterns,
usually termed supramolecular synthons [47], are, accord-
ing to the X-ray data, independent of the polymorphic form.
However, in the dimeric structures of form I of clioquinol
and forms Ia and Ib of cloxiquine, the paired molecules in
the units are twisted, while in cloxiquine (form II) they are
not. Due to this subtle difference in the planarity of the
dimeric structures which consist of the paired molecules
linked by bifurcated hydrogen bonds in the units (twisted in
form I, and planar in form II) [44, 45], accompanied by a
change in the hydrogen bond lengths, the structural units in
Fig. 5a–b The correlation between the experimental and calculated
NQR frequencies. a Cloxiquine (form I at RT and 90 K, form II). b
Cloxiquine (form I), clioquinol (form Ia)
1788 J Mol Model (2011) 17:1781–1800

Page 9

Table 3 Topological parameters of ρ for the supramolecular synthons of cloxiquine (the electron density at the BCP and RCP, ρ; its Laplacian, Δ(ρ); the potential electron energy density, VBCP;
the kinetic electron energy density, GBCP; the total electron energy density (HBCP), and the energy of interactions (EBSSEor EE) calculated at the B3LYP/6-311++G(d,p) level of theory
Compound
Form
Interaction/critical point type
EBSSE
(kJ/mol)
R (X–H···Y)b
(Å)
R(Y···H)c
(Å)
ρ
(a.u.)
Δ(ρ)
(a.u.)
ε
GBCP
(a.u.)
VBCP
(a.u.)
HBCP
(a.u.)
EE
(kJ/mol)
Strength
Cloxiquine
Form I RT
[R44] R=4.9%
O–H···N intermolecular
–
2.858
1.934 (0.770)
0.0172
0.0711
0.0459
0.0149
−0.0119
0.0029
−15.68
Weak
N···N
–
3.373
–
0.0072
0.0202
0.9228
0.0045
−0.0039
0.0006
−5.07
–
RCP CCOHNCCOHN
intermolecular
–
–
–
0.0070
0.0210
–
0.0046
−0.0039
0.0007
–
–
RCP benzene ring
–
–
–
0.0209
0.1519
–
0.0306
−0.0232
0.0074
–
–
RCP heterocyclic ring
–
–
–
0.0240
0.1727
–
0.0356
−0.0281
0.0076
–
–
Form I RT [44] opta
O–H···N intermolecular
−35.75
2.859
1.934 (1.017)
0.0301
0.083
0.038
0.021
−0.021
−0.0002
−27.90
Moderate
RCP CCOHNCCOHN
intermolecular
–
–
–
0.0090
0.026
–
0.035
−0.005
0.0306
–
–
RCP benzene ring
–
–
–
0.0207
0.151
–
0.030
−0.023
0.0074
–
–
RCP heterocyclic ring
–
–
–
0.0239
0.172
–
0.035
−0.028
0.0076
–
–
Form I 90 K [45]
O–H···N intermolecular
–9.42
2.811
2.063 (0.911)
0.0230
0.079
0.046
0.018
−0.016
0.0019
−21.07
Weak
RCP CCOHNCCOHN
intermolecular
–
–
–
0.0086
0.025
–
0.005
−0.005
0.0007
–
–
RCP benzene ring
–
–
–
0.0208
0.152
–
0.030
−0.023
0.0074
–
–
RCP heterocyclic ring
–
–
–
0.0233
0.167
–
0.034
−0.027
0.0074
–
–
Form II 90 K [45]
O–H···N inter
−14.55
2.833
2.131 (0.982)
0.0203
0.068
0.021
0.015
−0.013
0.0019
−17.41
Weak
C–H···O inter
2.999
2.471 (0.949)
0.0089
0.041
0.609
0.008
−0.006
0.0019
−8.33
Weak
O–H···N intra
–
2.742
2.181 (0.982)
0.0198
0.077
0.928
0.017
−0.015
0.0022
−19.50
Weak
RCP O–H···N intramolecular
NHOCC
–
–
–
0.0195
0.098
–
0.021
−0.017
0.0038
−
–
RCP intramolecular (CH···O)
NCHOH
–
–
–
0.0086
0.046
–
0.009
−0.007
0.0022
–
–
RCP CCOHNCCOHN
O–H···N intermolecular
–
–
–
0.0072
0.020
–
0.004
−0.004
0.0008
–
–
RCP benzene ring
–
–
–
0.0210
0.154
–
0.031
−0.024
0.0075
–
–
RCP heterocyclic ring
–
–
–
0.0235
0.169
–
0.035
−0.028
0.0074
–
–
aPartially optimized geometry
bR(X···Y): donor (X) to acceptor (Y) distance
cR(Y···H): acceptor (Y) to proton (H) distance
J Mol Model (2011) 17:1781–18001789

Page 10

cloxiquine can be differently packed to yield two poly-
morphs. As follows from X-ray data, the intramolecular
hydrogen bonds O–H···N in clioquinol (RO-H···N=2.753 [42]
or 2.818 Å [43]) are longer than those reported in
cloxiquine (RO-H···N=2.811 [44] and 2.747 Å [45]), both
of which are close to those in 8-quinolinol (RO-H···N =
2.753 Å [46]). The intermolecular hydrogen bonds O–H···N
in form Ia of clioquinol (RO-H···N=2.792) are shorter than
those reported in cloxiquine (RO-H···N=2.859 and 2.833 Å),
but similar to those in 8-quinolinol (RO-H···N=2.793 Å),
while in form Ib of clioquinol they (RO-H···N=2.850) are
close to those reported in cloxiquine and much longer than
those reported in 8-quinolinol (RO-H···N=2.793 Å). The
intramolecular hydrogen bonds in clioquinol and cloxiquine
are much more nonlinear [<OHN=87° (form Ia) or 95°
(form Ib) and 108° (form I) or 115° (form II) for clioquinol
and cloxiquine, respectively, versus 109° in 8-quinolinol]
than the intermolecular ones (<OHN=151.5° and 151.8°
versus 146.1° and 127.5° versus 143.0° in 8-quinolinol).
The hydrogen bonds were characterized within the Bader
QTAIM theory, with their molecular topology described in
terms of the BCP and RCP. The topological parameters (bond
length r, electron density ρ, its Laplacian Δρ, ellipiticity ε,
BCP and RCP) are collected in Tables 3 and 4. They
describe molecular stability and characterize the internuclear
pathways, which can be classified as shared or closed-shell.
All of the expected BCPs associated with the standard
covalent bonds and RCPs at the centroids of all benzenoid
rings were found in monomers and dimers of both
compounds. In addition, in the monomer, one extra BCP
was assigned to a weak intramolecular interaction, which in
turn generated one more RCP ring, but only in cloxiquine
form II. The presence of the RCP and BCP in the monomer
of form II of cloxiquine confirms the existence of intramo-
lecular O–H···N hydrogen bonding, since the topological
criteria proposed by Koch and Popelier [48] are fulfilled.
Surprisingly, in contrast to form II of cloxiquine, the lack of
a BCP between the putative donor and the acceptor in the
monomer of form I of cloxiquine or form I of cloquinol
implies no O–H···N hydrogen bond. This suggests that, if
there is this interaction at all, it is very weak and repulsive
rather than attractive. The presence of RCPs in five-
membered hydrogen-bonded chelate rings and the presence
of BCPs in both kinds of hydrogen bonds in the supramo-
lecular synthon of form II of cloxiquine, see Tables 3 and 4,
confirms the existence of different kinds of H-bonds:
intramolecular O–H···N and intermolecular O–H···N and
C–H···O in form II of cloxiquine, in contrast to the
occurrence of only intermolecular O–H···N in the structures
of cloxiquine and clioquinol form I.
For the supramolecular synthons of cloxiquine or
clioquinol form I, in a similar manner to the monomer,
there is no evidence of intramolecular H-bond O–H···N,
which suggests that the geometric criteria are insufficient to
determine the existence of of hydrogen bonding. Addition-
ally, very weak intermolecular halogen contacts Cl···I (Cl···I
=3.710 Å) and hydrogen bonds C–H···I (RC-H···I=4.026 Å)
were revealed in the structure of clioquinol, but only in
form Ib. The QTAIM calculations yielded electron densities
of 0.019 a.u. for intramolecular bonds and 0.020–0.034 a.u.
for intermolecular ones (the electron density falls within a
certain range of values, typically between 0.001 and
0.035 a.u.), which are markedly lower than those for the
covalent bonds. The corresponding Laplacian values, Δρ,
are positive and amount to 0.08 a.u. and 0.04–0.09 a.u.
(typically between 0.006 and 0.130 a.u.), which is
indicative of the closed-shell interaction. The relief maps
of the Laplacian of electron density for clioquinol and
cloxiquine in the plane of the intermolecular H-bond O–
H···N, see Fig. 8a–c, exhibit a maximum in the negative
Laplacian on either side of the oxygen and nitrogen atoms,
corresponding to the lone pair model. Moreover, they show
the polarization of nitrogen’s lone-pair electrons toward
hydrogen and differences in the polarization of the oxygen
and nitrogen lone pairs caused by changes in planarity as
well as iodine substitution.
To get further insight into the nature of intermolecular
and intramolecular interactions, the electronic energy
density HBCPand its components—the local one-electron
kinetic energy density (GBCP) and the local potential energy
density (VBCP)—were calculated for the charge distribution
at the BCP. The hydrogen and halogen bond energies were
calculated using CP as well as the Espinosa method; see
Tables 3 and 4. According to Roza’s [49] criterion, the
intramolecular O–H···N bonds in form II of cloxiquine are
weak but slightly stronger than the intermolecular O–H···N
bonds (see Tables 3 and 4), but they are generally weaker
than typical and mainly electrostatic, while the intermolec-
ular O–H···N bonds in clioquinol and cloxiquine form I at
RT are moderate and partially covalent in nature. The
estimated H-bond energies for the systems studied lie
within the range 8.33–34.39 kJ mol−1; see Tables 3 and 4.
Partitioning the DFT energy of the hydrogen bond into
classical components showed that about 75% was electro-
static (coulombic), and less than 5% came from polarization
and charge transfer.
The large differences in the strengths of the corresponding
H-bonds in both forms of cloxiquine suggest that there is
interplay between the different H-bonds in adjacent mole-
cules that lead to dimers. It is worth noting that the energies
of the H-bonds in form I of cloxiquine estimated by the
Espinosa method seem significantly higher at 90 K than at
RT (about 5.6 kJ mol−1) when the nonoptimized structures
are taken into account. Moreover, additional weak intermo-
lecular contact between nitrogen atoms is detected, but only
when an extra short OH bond taken from the X-ray data at
1790 J Mol Model (2011) 17:1781–1800

Page 11

RT is assumed. On the other hand, the hydrogen bond
energies for clioquinol forms Ia and Ib differ by as much
as 6.5 kJ mol−1, but both structures were determined at
RT, and the difference in the relevant R factors is only
1.6%. It should be mentioned that the H-bond energy
should only be estimated using the Espinosa method when
comparing the strengths of H-bonds; it should not be used
quantitatively.
According to the QTAIM results, the intermolecular
hydrogen bonds O–H···N in cloxiquine and clioquinol
(form I) are the strongest interactions, which is in
agreement with previous suggestions based on differences
in hydrogen acceptor distances [50] as well as IR spectra in
solution [51]. In clioquinol, an additional electron-
withdrawing substituent (iodine) ortho to the hydroxyl
group compeats with chlorine at the para position, resulting
in the formation of stronger hydrogen bonds than those in
cloxiquine, which is consistent with the IR spectra for the
solid (KBr). The direction of the spectral shifts of the O–H
stretch mode in the mid-infrared, often used to infer the
strength of a hydrogen bond, is in a good agreement with
the DFT results; see Tables 3 and 4. Proton donor and
acceptor sites in hydrogen bonds in supramolecular
synthons can be reliably characterized for the solid using
1H-17O and
noted that the calculations reveal that the presence of the
intermolecular O–H···N hydrogen bonding influences NQR
parameters differently at different sites: only slightly for
35Cl (i.e., at the site distant from the hydrogen bond), while
significantly at the17O and14N participating in this bond; a
comparison of the experimental data and the results of
calculations for the monomer and stacked dimers (see
Tables 3 and 4) provides a drastic example.
The presence of the electron-withdrawing iodine substit-
uent at position 7 of 5-chloro-8-quinolinol results in a
decrease in the proton affinity of the nitrogen atom and an
increase in the hydrogen bond strength, which induces
changes in the values of e2Qqh−1and η; see Tables 1 and 2.
As a result of this substitution, e2Qqh−1increases on
35Cl and14N by 0.01 and 2%, respectively, while on17O
this parameter is predicted to decrease by 0.05%. The
predicted changes in η on35Cl are negligibly small, on17O
this parameter decreases by 3%, but on14N it increases by
2%. The direction of the changes in the NQR parameters
observed upon switching from cloxiquine to clioquinol is in
agreement with Seliger’s [52] observation that when
shortening and hence also strengthening the hydrogen
bond, the asymmetry parameter η increases. A considerable
increase in e2Qqh−1on the chlorine atom, which implies a
decrease in the symmetry of the charge distribution at this
atom (i.e., nonspherical symmetry), is also observed mainly
as a consequence of the presence of the iodine electron-
withdrawing substituent meta to the chlorine. A comparison
1H-14N NQDR spectroscopies. It should be
Table 4 Topological parameters of ρ for the supramolecular synthons of clioquinol [the electron density at the BCP and RCP (ρ), its Laplacian Δ(ρ), the potential electron energy density (VBCP),
the kinetic electron energy density (GBCP), and the total electron energy density (HBCP)] and energy of interactions (EBSSEor EE) calculated at the B3LYP/6-311++G(d,p) level of theory
Compound
Form
Interaction/critical point type
EBSSE
(kJ/mol)
R (X–H···Y)a
(Å)
R(Y···H)b(Å)
ρ
(a.u.)
Δ(ρ)
(a.u.)
ε
GBCP
(a.u.)
VBCP
(a.u.)
HBCP
(a.u.)
EE
(kJ/mol)
Strength
Clioquinol
Form Ia RT
[43] R=10%
BCP O–H···N intermolecular
−27.86
2.792
1.883 (0.986)
0.0340
0.094
0.041
0.025
−0.026
−0.0013
−34.39
Moderate
RCP O–H···N intermolecular
–
–
–
0.0102
0.030
–
0.007
−0.006
0.0009
–
–
RCP benzene ring
–
–
–
0.0213
0.154
–
0.031
−0.024
0.0075
–
–
RCP heterocyclic ring
–
–
–
0.0234
0.169
–
0.035
−0.027
0.0075
–
–
Form Ib RT
[42] R=8.4%
O–H···N intermolecular
−35.28
2.850
1.942 (0.988)
0.0297
0.084
0.045
0.021
−0.021
0.0001
−27.68
Weak
I···Cl
−1.88
3.710
–
0.0066
0.021
0.106
0.004
−0.0029
0.0012
−3.89
Weak
I···H-C
4.026
3.373 (0.653)
0.0045
0.013
0.161
0.003
−0.0019
0.0007
−2.49
weak
RCP CCOHNCCOHN intermolecular
–
–
–
0.0089
0.026
–
0.006
−0.005
0.0008
–
–
RCP CCCClIH
–
–
–
0.0037
0.012
–
0.002
−0.002
0.0001
–
–
RCP benzene ring
–
–
–
0.0184
0.131
–
0.026
−0.019
0.0068
–
–
RCP heterocyclic ring
–
–
–
0.0215
0.152
–
0.031
−0.028
0.0031
–
–
R is the reliability factor (R factor)
aR(X···Y) donor (X) to acceptor (Y) distance
bR(Y···H) acceptor to proton (H) distance
J Mol Model (2011) 17:1781–18001791

Page 12

Fig. 6 The orientations of the
EFG tensor axes at all
quadrupolar sites: cloxiquine
(form I), first column;
cloxiquine (form II), second
column; clioquinol (form Ia),
third column. In each case,
one of the axes is perpendicular
to the screen
Fig. 7 Molecular graphs of 8-
quinolinol derivatives:
cloxiquine (form I), supramo-
lecular synthon; cloxiquine
(form II), supramolecular
synthon; clioquinol (form Ia),
supramolecular synthon. Dashed
lines indicate the interactions,
large circles correspond to
attractors, and small circles
to critical points (red RCP, green
BCP)
1792J Mol Model (2011) 17:1781–1800

Page 13

of the results obtained for clusters of different sizes (see
Table 2) shows that the value of e2Qqh−1depends strongly
on the presence of hydrogen bonding, and that η can be
used as an indicator of the strength of the hydrogen
bonding.
Because of the relatively short distances to the protons
(2.182 and 2.622 Å) in the structures of cloxiquine and
clioquinol, O–H···H–O dihydrogen bonds could be
expected as well as HB intermolecular interactions. An
H···H distance of less than 2.4 Å [i.e., twice the van der
Waals radius of the hydrogen atom (1.2 Å)] is the
geometrical criterion that is most widely used to identify
the formation of this type of hydrogen bond. However,
because the dihydrogen bonds are electrostatic in nature,
and such interactions act beyond this distance, this van der
Waals criterion has been strongly criticized [53]. Additional
criteria taken into account include the interaction energy
(which falls within the same range that are typical of
hydrogen bonds: 12–41 kJ mol−1), the linearity of the
dihydrogen bonds, and the difference between the charges
on both of the electronegative atoms [54, 55]. QTAIM does
not detect the critical points that indicate the presence of an
O–H···H–O dihydrogen bond for any form of clioquinol or
cloxiquine, which appears to be due to both oxygen atoms
having the same polarization, as a consequence of the
symmetric dimeric structure.
Stacked supermolecular synthon
An important type of interaction that is specific to the solid
state and commonly seen in aromatic systems is the vertical
stacking of parallel supramolecular synthons; see Fig. 9a–c.
As mentioned earlier, in the dimeric structures of clioquinol
and cloxiquine (form I), the paired molecules in the units
are twisted and the units expand into columns linked by
π–π stacking interactions with distinct layers separated by
only 3.814 Å in cloxiquine [45] (i.e., close to the 3.811 Å
observed in 8-quinolinol [46]), and as large as 4.141 or
4.161 Å in clioquinol [42, 43], which suggests that there is
a great deal of steric repulsion from I–I, forcing the large
spacing between layers. The difference between forms I and
II of cloxiquine (Fig. 9a, b) is that the twisted and
Fig. 8 Relief map of the Lap-
lacian of electron density in the
OH···N plane: cloxiquine (form
I), supramolecular synthon;
cloxiquine (form II), supramo-
lecular synthon; clioquinol
(form Ia), supramolecular
synthon
J Mol Model (2011) 17:1781–1800 1793

Page 14

nontwisted adjacent supramolecular synthons form columns
along different crystallographic axes (c in form I and b in
form II) [45]. The π–π stacking interactions between
adjacent supramolecular synthons (stacked dimers) enhance
the stability of the crystal structure in both clioquinol and
cloxiquine, irrespective of the specific polymorph; however,
their strengths are expected to differ because of large differ-
ences in the polarizations ofthe adjacent molecules, so dimers
are also responsible for the specific arrangements seen in
parallel columns.
The dipole moments calculated by the DFT method for a
continuous distribution of electron density provide essential
information on the overall polarity of the charge system.
The monomer of clioquinol has a higher dipole moment
than that of form I of cloxiquine (Ia: 3.15; Ib: 3.20 versus
1.71 D for form I, RT), while the dimer and stacked dimers
have lower dipole moments (for form Ia of clioquinol: 3.97;
form Ib: 3.94 versus 1.97 D for form I of cloxiquine at RT,
and form Ia of clioquinol: 3.07; form Ib: 3.03 versus 4.14 D
for form I of cloxiquine at RT). Moreover, in form II of
cloxiquine, which is predicted to be energetically more
stable than form I (by 13.6 kJ mol−1), the monomer has a
slightly higher dipole moment (1.82 D), while the dimer
and stacked dimers are nonpolar (0.02 and 0.005 D,
respectively), which suggest large differences in the polar-
izations of the adjacent molecules, so that the dimers are
closely connected with the specific arrangement. The
stacking found in both clioquinol and cloxiquine (form I)
involves similar overlapping, albeit much smaller than
observed in cloxiquine (form II). The nucleus-independent
chemical shift (NICS) calculated at the geometrical center,
and the modified form of this parameter, NICS(1),
calculated at 1 Å above the plane of the ring, both reflect
π effects and are both negative, which means that the
criterion of Schleyer et al. [56] for aromaticity is fulfilled
for both compounds irrespective of the form. The standard
method of estimating the energy of stacking π···π inter-
actions (CP) appears to be ineffective for such a compli-
cated system. The roughly estimated energies of π–π
stacking interactions are as high as 16.3 and 14.5 kJ
mol−1for clioquinol forms Ia and Ib, respectively, and 19.8
and 23.6 kJ mol−1for cloxiquine forms I and II,
respectively. A comparison of these energies suggests that
the strength of the stacking π···π interaction depends on the
planarity (i.e., it is much weaker in form I than in form II of
cloxiquine), and that the lower energy of π–π stacking for
clioquinol in comparison to cloxiquine is due to the greater
spacing between layers.
Fig. 9 Molecular graphs of 8-
quinolinol derivatives: cloxi-
quine (form I), stacked supra-
molecular synthon; cloxiquine
(form II), stacked supramolecu-
lar synthon; clioquinol (form Ia),
stacked supramolecular synthon;
dashed lines indicate interac-
tions, large circles correspond to
attractors, small circles to criti-
cal points (green BCP; RCPs are
omitted for clarity)
1794 J Mol Model (2011) 17:1781–1800

Page 15

Table 5 Topological parameters of ρ for the stacked supramolecular synthons of cloxiquine (the electron density at the BCP and RCP ρ, its Laplacian Δ(ρ), the potential electron energy density
VBCP, the kinetic electron energy density GBCP, the total electron energy density HBCP, and the energy of interaction EE, calculated at the B3LYP/6-311++G(d,p) level of theory)
Compound
Form
Interaction
R(X–H···Y)b
(Å)
R(Y···H)c
(Å)
ρ
(a.u.)
Δ(ρ)
(a.u.)
ε
GBCP
(a.u.)
VBCP
(a.u.)
HBCP
(a.u.)
EE
(kJ/mol)
Strength
Cloxiquine
Form I RT [44] R=4.9%
O–H···N intermolecular
2.858
2.163
0.0172
0.0711
0.0455
0.0149
−0.0018
0.0131
−15.70
Weak
N(1)···N(1″)
3.373
–
0.0071
0.0203
1.0940
0.0045
−0.0038
0.0006
−5.01
Weak
C–H···O (interlayer)
3.475
2.809
0.0049
0.0170
0.1647
0.0036
−0.0029
0.0007
−3.77
Weak
Cl(5)···Cl(5′)
3.814
–
0.0043
0.0140
0.0809
0.0027
−0.0018
0.0008
−2.37
Weak
C(6)···C(5′)
3.814
–
0.0055
0.0150
0.1686
0.0031
−0.0025
0.0007
−3.23
Weak
C(10)···C(4′)
3.451
–
0.0052
0.0143
0.8493
0.0029
−0.0023
0.0007
−2.97
Weak
C(8)···C(9′)
3.487
–
0.0050
0.0134
0.4528
0.0028
−0.0022
0.0006
−2.82
Weak
N(1)-C(2′)
3.455
–
0.0045
0.0134
0.4039
0.0028
−0.0022
0.0006
−2.85
Weak
Form I RT [44] opta
O–H···N intermolecular
2.858
1.934
0.0302
0.0829
0.0369
0.0211
−0.0214
−0.0003
−28.09
Weak
C-H···O (interlayer)
3.475
2.782
0.0050
0.0176
0.1599
0.0037
−0.0030
0.0007
−3.92
Weak
Cl(5)···Cl(5′)
3.814
–
0.0043
0.0140
0.0819
0.0027
−0.0018
0.0008
−2.39
Weak
C(6)···C(5′)
3.814
–
0.0053
0.0149
0.2917
0.0031
−0.0024
0.0006
−3.21
Weak
C(10)···C(4′)
3.451
–
0.0053
0.0145
0.8311
0.0029
−0.0023
0.0007
−2.99
Weak
C(8)···C(9′)
3.487
–
0.0050
0.0135
0.5245
0.0028
−0.0021
0.0006
−2.81
Weak
N(1)-C(2′)
3.455
–
0.0045
0.0135
0.4405
0.0028
−0.0022
0.0006
−2.89
Weak
Form I 90 K [45]
R=6.69%
O–H···N intermolecular
2.811
2.063
0.0230
0.0794
0.0453
0.0180
−0.0161
0.0019
−21.14
Weak
C–H···O (interlayer)
3.392
2.791
0.0048
0.0180
0.1983
0.0037
−0.0029
0.0008
−3.82
Weak
Cl(5)···Cl(5′)
3.745
–
0.0049
0.0163
0.0222
0.0031
−0.0022
0.0010
−2.82
Weak
C(6)···C(5′)
3.395
–
0.0058
0.0165
0.2639
0.0034
−0.0027
0.0007
−3.58
Weak
C(10)···C(4′)
3.388
–
0.0058
0.0158
0.7103
0.0032
−0.0025
0.0007
−3.30
Weak
C(8)···C(9′)
3.407
–
0.0056
0.0155
0.6187
0.0032
−0.0025
0.0007
−3.21
Weak
N(1)-C(2′)
3.388
–
0.0050
0.0149
0.1343
0.0031
−0.0025
0.0006
−3.23
Weak
Form II 90 K [45]
R=5.48% opta
O–H···N intermolecular
2.833
2.131
0.0203
0.0681
0.0239
0.0152
−0.0133
0.0019
−17.44
Weak
C–H···O intermolecular
2.999
2.471
0.0089
0.0406
0.6320
0.0082
−0.0063
0.0019
−8.32
Weak
O–H···N intramolecular
2.742
2.181
0.0198
0.0774
0.9893
0.0171
−0.0149
0.0022
−19.61
Weak
Cl(5)···Cl(5′)
3.763
–
0.0047
0.0157
0.0244
0.0030
−0.0021
0.0009
−2.70
Weak
C(6)···C(5′)
3.763
–
0.0061
0.0173
0.3776
0.0036
−0.0028
0.0008
−3.73
Weak
C(10)···C(4′)
3.405
–
0.0057
0.0158
0.9877
0.0032
−0.0025
0.0007
−3.28
Weak
C(8)···C(9′)
3.373
–
0.0060
0.0165
0.8761
0.0034
−0.0026
0.0008
−3.43
Weak
N(1)-C(2′)
3.413
–
0.0048
0.0145
1.1233
0.0030
−0.0023
0.0007
−3.06
Weak
aPartially optimized geometry, R is the reliability factor (R factor)
bR(X···Y) donor (X) to acceptor (Y) distance
cR(Y···H) acceptor (Y) to proton (H) distance
J Mol Model (2011) 17:1781–18001795