Glycosylated copper(ii) ionophores as prodrugs for β-glucosidase activation in targeted cancer therapy

Article abstract of DOI:10.1039/c2dt32429f

8-Hydroxyquinoline derivatives are metal-binding compounds that have recently attracted interest as therapeutic agents for cancer therapy. In this scenario, we designed and synthesized three new glucoconjugates, 5,7-dichloro-8-quinolinyl-β-d-glucopyranoside, 5-chloro-8-quinolinyl- β-d-glucopyranoside and 2-methyl-8-quinolinyl-β-d-glucopyranoside and investigated their biological properties in comparison to the parent 8-hydroxyquinoline derivatives in the presence of Cu²⁺. In vitro data show that 2 out of 3 glycosylated compounds possess a pharmacologically- relevant antiproliferative activity against tumor cells, similar to that of their parent compounds; this activity is associated with a relevant triggering of apoptosis. The pharmacological profile of the glucoconjugates depends on the cellular enzymatic β-glucosidase activity, as demonstrated by the inhibition of antiproliferative activity in the presence of the 2,5-dideoxy-2,5-imino-d-mannitol. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c2dt32429f

Dalton
Transactions
View Article Online
View Journal | View Issue
PAPER
Glycosylated copper(II) ionophores as prodrugs for
β-glucosidase activation in targeted cancer therapy†
Cite this: Dalton Trans., 2013, 42, 2023
Valentina Oliveri,a Maurizio Viale,b Giulia Caron,c Cinzia Aiello,b Rosaria Gangemib
and Graziella Vecchio*a
8-Hydroxyquinoline derivatives are metal-binding compounds that have recently attracted interest as
therapeutic agents for cancer therapy. In this scenario, we designed and synthesized three new gluco-
conjugates, 5,7-dichloro-8-quinolinyl-β-D-glucopyranoside, 5-chloro-8-quinolinyl-β-D-glucopyranoside and
2-methyl-8-quinolinyl-β-D-glucopyranoside and investigated their biological properties in comparison to
the parent 8-hydroxyquinoline derivatives in the presence of Cu2+. In vitro data show that 2 out of
3 glycosylated compounds possess a pharmacologically-relevant antiproliferative activity against tumor
cells, similar to that of their parent compounds; this activity is associated with a relevant triggering of
apoptosis. The pharmacological profile of the glucoconjugates depends on the cellular enzymatic
β-glucosidase activity, as demonstrated by the inhibition of antiproliferative activity in the presence of the
2,5-dideoxy-2,5-imino-D-mannitol.
Received 11th October 2012,
Accepted 2nd November 2012
DOI: 10.1039/c2dt32429f
www.rsc.org/dalton
contaminated CQ production. Currently, CQ is in phase I clini-
cal trials in patients with hematologic malignancy, such as
Introduction
Cancer, the uncontrolled and pathological proliferation of leukemia, myelodysplasia, non-Hodgkin’s and Hodgkin’s
abnormal cells, is the second leading cause of human death lymphomas, and multiple myeloma.5
after cardiovascular diseases in developing, as well as
The anticancer mechanism of OHQs is related to the com-
advanced, countries.1,2 Although there are many therapeutic plexation of copper ions.6 This transition metal ion is involved
strategies, including chemotherapy and radiotherapy, high in many different cellular mechanisms: as a cofactor for
systemic toxicity and drug resistance limit the successful out- different Cu-dependent oxidases in respiration processes and
comes in most cases. Therefore, novel treatments are urgently in other proteins.7 Even if copper homeostasis is tightly regu-
needed for cancer therapy.
lated, elevated copper levels have been found in many types of
8-Hydroxyquinoline derivatives (OHQs) are metal-binding cancer. Although its accumulation in cancer cells has not yet
compounds that have recently been viewed with interest as been fully explained, it is conceivable that Cu2+ may help the
therapeutic agents for cancer and Alzheimer’s disease (AD).3 tumor growth and release of metastatic cells. In particular,
Clioquinol (5-chloro-7-iodo-8-hydroxyquinoline, CQ) is the Cu2+ is necessary for angiogenesis,8,9 which is the mechanism
most known member of this family. In the 1950s, it was widely by which the tumor may survive and grow beyond one or two
used as an amebicide for the treatment of diarrhea.4 CQ has millimeters of diameter. The need for Cu2+ ions by tumor cells
widely been investigated as a metal–protein-attenuating com- justified the clinical use of specific Cu2+ chelators as anti-
pound. It also reached clinical trials for the treatment of AD cancer agents.10–12 It is important to note that the use of Cu2+
disease, but clinical studies have been interrupted due to complexes as cancer therapeutics is likely to be dependent
issues with the purity of large-scale CQ production: small more on their cytotoxic action, rather than on modulation of
amounts of di-iodo-8-hydroxyquinoline, a known carcinogen, copper homeostasis.13 Although the mechanism of action is
not completely understood, experimental evidence suggests
that OHQs and CQ act as anticancer agents as 20S proteasome
inhibitors in the presence of Cu2+.14,15 Other mechanisms
have been reported to explain the cytotoxicity of CQ, such as
the induction of cytoplasmatic clearance of X linked inhibition
of pro-apoptotic proteins16 and induction of DNA double
strand-breaks.17
aUniversity of Catania, Dipartimento di Scienze Chimiche, Viale A. Doria, 6,
95125 – Catania, Italy. E-mail: gr.vecchio@unict.it; Fax: +39 095580138;
Tel: +39 0957385064
bIRCCS Azienda Ospedaliera Universitaria San Martino – IST Istituto Nazionale per
la Ricerca sul Cancro, U.O.C. Terapia Immunologia, L.go R. Benzi, 10,
16132 – Genova, Italy
cUniversity of Torino, BMSS, Via Quarello, 11, 10135 – Torino, Italy
†Electronic supplementary information (ESI) available: NMR, UV-vis and ESI-MS
spectra. See DOI: 10.1039/c2dt32429f
The crucial role played by Cu2+ ions in many cellular
mechanisms also explains how an uncontrolled depletion of
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 2023–2034 | 2023

View Article Online
Paper
Dalton Transactions
Cu2+ could be the cause of toxic events. For this reason, un-
selective chelation of essential metal ions due to an anticancer
treatment could have severe side effects and should, therefore,
be avoided. A well-accepted strategy to prevent these side
effects consists in using prodrugs.
In general, prodrugs design aims at overcoming a number
of limitations of drug candidates:18 (a) pharmaceutical pro-
blems, such as poor solubility, insufficient chemical stability,
unacceptable taste or odor, and irritation or pain; (b) pharma-
cokinetic problems, such as insufficient oral absorption,
inadequate blood–brain barrier permeability, marked pre-sys-
temic metabolism, and toxicity; and (c) pharmacodynamic pro-
blems, such as a low therapeutic index and lack of selectivity
at the site of action.
For cancer therapy, glucose-conjugated prodrugs are of
great interest since they can enhance uptake of a drug in
cancer cells, as tumor cells require a significant amount of
glucose. In fact, the efficiency of ATP production in normal
cells does not satisfy the high energy requirements of cancer
cells and, thus, glucose transporters (GLUTs) are overexpressed
in cancer cells to improve glucose uptake.19 The ratio of
GLUTs expression in malignant versus normal cells is about
100–300.20
Once delivered to the target tissue, non-cytotoxic prodrugs
require tumor-specific activation. This may occur by the
hypoxic environment of solid tumors, by the tumor-associated
proteases or by the enzymes selectively present in the tumor.
Prodrug monotherapy is a convenient strategy to activate pro-
drugs using an enzyme already present in the tumor. In par-
ticular, appropriate glycosylated prodrugs are activated by
β-glucosidase overexpressed in tumor cells.21
Fig. 1 Chemical structures of the investigated compounds.
the glycosylated derivatives by demonstrating the role of the
β-glucosidase enzyme. For comparison purposes, the data for
the corresponding 8-hydroxyquinolines were also collected
(Fig. 1).
Results
In silico physicochemical and ADME profile
Summing-up, glycosylated prodrugs of copper-binding com-
pounds such as OHQs could combine a selective delivery to Lipinski’s Rule of Five (Ro5) is a widely accepted rule of thumb
tumor tissues with a tumor specific mechanism of activation to determine if a chemical compound with a certain pharma-
and specific antitumor activity and, therefore, could represent cological or biological activity has molecular properties (i.e.
a new strategy for the research of anticancer agents. In the molecular weight is over 500 Da, the calculated log P is
addition the presence of the sugar unit prevents copper over 5, the structure includes more than 5 hydrogen-bond
binding and could prevent the systemic undesired complexa- donors and more than 10 hydrogen-bond acceptors) that could
tion. Moreover, glycosylated prodrugs of OHQs are expected to be associated with poor absorption and distribution proper-
be more soluble than the corresponding parent compounds ties. Designed compounds (Fig. 1) were tested by Ro5. The
and, therefore, more in line with drug-likeness criteria.
results revealed that all are within the range set by Ro5 as per
In a previous explorative study we showed for the first time Table 1.
the antiproliferative activity of 5-chloro-7-iodo-8-quinolinyl-β-D-
Properties included in the Ro5 may vary with molecular
glucopyranoside (GluCQ) and for comparison of 8-hydroxyqui- flexibility. To check the extent of the influence of flexibility on
noline glucoconjugate (GluOHQ) as new prodrugs.22 In par- the 3D structure of the glycosylated prodrugs we monitored
ticular, we observed that these compounds show significant lipophilicity, which is widely accepted as the most relevant
antiproliferative activities against different cancer cell lines in molecular property associated with biodistribution.23 In prac-
the presence of Cu2+. We also hypothesized that the glucose tice, we firstly submitted all prodrugs with a conformational
moiety had to be cleaved in cells by the β-glucosidase and then analysis. Then we calculated the log P of any resulting confor-
the aglycone was able to complex Cu2+ ions. In this paper we mer (called virtual log P24), the average log P value, and the
describe the synthesis of three new glycoconjugates of OHQs lipophilicity range (i.e. the difference between the highest and
(Fig. 1) with good physicochemical and ADME profile, the the lowest virtual log Ps). Data in Table 1 show that the influ-
characterization and stability, the chelating properties of ence of molecular flexibility is modest (lipophilicity range
copper ions, the in vitro anticancer biological profile and, lower than 0.5).
finally, some docking experiments to shed light on experimen-
Solubility is a key property in drug discovery. In this study,
tal results. We clarify some aspects of the action mechanism of solubility data of the compounds are expressed as ALOGpS
2024 | Dalton Trans., 2013, 42, 2023–2034
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
(the higher, the more soluble the compound) and reported in
Table 1. They point out that prodrugs are expected to be more
soluble than the corresponding drugs.
Finally, to predict the ADME behaviour, the compounds
were projected on the following pre-calculated models
implemented in the VolSurf+ software (see the Experimental
part for details): plasma protein affinity (PB), distribution
volume (VD), Caco-2 cell permeability (CACO2), blood–brain
barrier permeation (lgBB) and metabolic stability (MetStab).
The ADME profiles of compounds (Table 1) were evaluated
according to the rules proposed by Carosati et al.25 Most
values are in line with the alert threshold values of drug-like-
ness. The low absorption predicted by the CACO2 descriptor
(0.2 the lowest acceptable threshold) has limited relevance
here since absorption is expected to be related to active trans-
port mechanisms.
Chemistry
SYNTHETIC ASPECT. The new compounds GluCl2HQ, GluClHQ
and GluMeHQ were synthesized by a modified Michael pro-
cedure as reported by us22 for similar compounds (Scheme 1).
The desired compounds were achieved through a one-pot
reaction. Briefly, the quinoline compounds were dissolved in a
water–methanol solution of the base K2CO3 and then a sol-
ution of glycosyl bromide in dichloromethane was added
together with a phase transfer catalyst, such as tetrabutylam-
monium salts. This reaction produced the β-anomer of the gly-
cosides. The pure products were isolated by precipitation. All
novel compounds were characterized by 1H NMR and 13C NMR
spectroscopy as well as ESI-MS. NMR spectra confirm the iden-
tity of the products. The ESI spectra of glucoconjugates show
at least three peaks due to proton and sodium adduct ions and
the dimeric sodium adduct ion. Furthermore, the zoom scan
spectra of the halogenated compounds show the typical isoto-
1
pic patterns due to the presence of the halogen atoms. The H
NMR spectra of glucoconjugates are reported in Fig. 2. In all
the spectra the signals due to the quinoline and the sugar
moiety can be seen. The J1,2 values of the glucose ring (about
7.8 Hz) confirm the β configuration for all derivatives. The H-1
of the glucose moiety is shifted downfield because of the
deshielding effect of the aromatic ring as reported for similar
compounds and resonates at about 5.1 ppm.26
The presence of chlorine at the C-7 position in GluCl2HQ
further shifts the H-1 downfield by about 5.4 ppm. The
glucose protons of GluCl2HQ are more shifted upfield in com-
parison to other two glycoconjugates as reported elsewhere for
CQ glycoconjugates. The disposition of the aromatic ring com-
pared to the sugar unit affects the chemical shifts especially of
glucose H-5 and H-6.
CD spectra of the three new compounds (Fig. 3) were also
recorded in methanol. The presence of the glucose moiety
induces chirality in the region of π–π* of the aromatic ring.27
GluClHQ and GluMeHQ have similar CD spectra showing a
slight negative band at 310 nm and a more intense negative
band at 200 nm. The CD spectrum of GluCl2HQ is significantly
different and displays a positive band at 300 nm and two
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 2023–2034 | 2025

View Article Online
Paper
Dalton Transactions
Scheme 1 Reagents and conditions: (a) K2CO3, Bu4NBr, CH2Cl2, 68 h, r.t.
Fig. 3 CD spectra of GluMeHQ (—), GluClHQ (…), GluCl2HQ (–-) in methanol.
Table 2 ESI-MS characterization of the 63Cu2+ complexes [Cu2+] = [L] = 3.0 ×
10−5 M. L is a hydroxyquinolinate derivative
Ligand
MeHQ
Assignment
Calculated (m/z)
Found (m/z)
[LH + H]+
160.1
239.0
380.1
402.0
418.0
600.0
180.0
292.9
419.9
441.9
659.9
214.0
274.9
487.9
509.9
762.8
160.1
238.9
379.9
401.9
418.0
599.8
180.1
292.8
419.9
441.9
659.9
214.0
274.9
487.8
509.9
762.8
[CuL + H2O]+
[CuL2 + H]+
[CuL2 + Na]+
[CuL2 + K]+
[Cu2L3]+
Fig. 2 NMR spectra of the glucoconjugates at 500 MHz in CD3OD. GluCl2HQ
(up), GluClHQ (middle), GluMeHQ (down).
ClHQ
[LH + H]+
[CuL + H2O]+
[CuL2 + H]+
[CuL2 + Na]+
[Cu2L3]+
intense negative bands at 240 and 206 nm, respectively. The
Δε values of the GluCl2HQ spectrum are higher than the other
two compounds, suggesting a different orientation of the aro-
matic ring compared to the sugar moiety in this case. This
could be due to the presence of a 7-substituent (a chlorine
atom) given that a similar trend is observed for GluCQ. The
CD spectra are perfectly consistent with NMR data.
Cl2HQ
[LH + H]+
[CuL + H2O]+
[CuL2 + H]+
[CuL2 + Na]+
[Cu2L3]+
COPPER COMPLEXES. Copper complexes of some hydroxyquino-
line derivatives have been investigated by different tech-
niques.28 We investigated the coordination properties of decomposition was detected after 24 hours under these con-
MeHQ, ClHQ and Cl2HQ by ESI-MS. A picture of the metal– ditions. In particular, incubation of prodrugs at 37 °C in PBS
ligand complexes formed in solution can be obtained by the (phosphate-buffered saline, pH 7.4) revealed that the prodrugs
mass spectrometric data. The ESI-MS spectra in the positive were stable for at least 1 week.
ion mode of the hydroxyquinoline derivative solutions contain-
In order to evaluate the ability of prodrugs to release the
ing CuSO4 were recorded. The experiments were carried out at active moiety in the presence of the specific enzyme, hydrolysis
different pH, from 4.0 to 10.0. The formation of complex assays were carried out with β-glucosidase and monitored by
species was observed at pH > 5. In Table 2 the pseudo-molecular several techniques, especially UV-vis spectroscopy. Almond
ions of the observed species complexes are reported. The β-glucosidase, commercially available, was used as the model
assignments of the peaks in ESI-MS spectra were done by enzyme. The UV-vis spectra were recorded at different time
comparing the experimental isotopic patterns with the corres- points to follow the hydrolysis at pH 7.4. When the glycoconju-
ponding simulated profiles. As expected, the three compounds gates are hydrolyzed, the OHQ moiety is released and new
can form different complex species with Cu2+. These data are bands due to the free oxine moiety appear. Analogously, the
in agreement with those previously reported in the literature same assay was carried out for the prodrugs in the absence of
for CQ and other similar compounds.28
glucosidase and no modification of the bands was observed.
CHEMICAL AND ENZYMATIC STABILITY. Chemical stabilities of the The glucoconjugates showed different behaviour in the pres-
compounds (Fig. 1) were monitored by several techniques ence of β-glucosidase (Fig. 4). The glycosidic bond of
such as mass spectrometry, UV-vis spectroscopy and chromato- GluCl2HQ and GluClHQ was rapidly cleaved (within 40 min),
graphy.
showing that the glucoside trigger is readily accessible by the
All glucoconjugates are very stable in the range of pH values enzyme, despite the presence of bulky aromatic moieties.
(pH 4–11) and temperatures (25–37 °C) investigated. No Oxine was released significantly faster from GluCl2HQ and
2026 | Dalton Trans., 2013, 42, 2023–2034
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
Table 4 Summary of IC50s (μM) of glycosides and their parent compounds on
A2780, A549 and MDA-MB-231 cells
Cl2HQ
GluCl2HQ
Cu2+
Cu2+
A2780
A549
24.64 2.59a 2.53 0.33 25.18 2.22
23.87 2.09 1.75 0.49 31.97 7.44
3.12 0.27
6.03 1.37
MDA-MB-231 24.98 3.15
4.97 1.29 35.53 3.09 28.12 0.86
ClHQ
GluClHQ
Cu2+
Cu2+
A2780
A549
5.12 0.94 0.47 0.03
5.93 0.92 0.41 0.20
4.10 0.47 0.36 0.07
5.32 1.27 0.56 0.03
MDA-MB-231 18.67 5.89 0.59 0.07 55.34 4.98 5.20 0.23
MeHQ
GluMeHQ
Cu2+
Fig. 4 Hydrolysis of prodrugs in the presence of almond β-glucosidase (1.0 ×
Cu2+
10−6 M) at 37 °C (pH 7). GluClHQ ( ), GluCl2HQ (▼), GluMeHQ ( ), GluCQ (◆),
GluOHQ (▲).
●
■
A2780
A549
MDA-MB-231
8.46 2.21
25.5 3.12
50.65 3.41
4.09 0.23
4.66 0.34
5.33 0.15
>100
>100
>100
>100
>100
48.55 2.18
Table 3 Half-lives of prodrugs in the presence of almond β-glucosidase (1.0 ×
10−6 M) at 37 °C (pH 7.0)
a The numbers express the mean SD of 4–8 data.
GluCl2HQ
9.5
GluClHQ
26.5
GluMeHQ
>600
GluCQ
130.5
GluOHQ
145
t1/2 (min)
GluClHQ than from GluCQ and GluOHQ. In contrast,
GluMeHQ did not show any significant cleavage within
2 hours. However, GluMeHQ was partially cleaved in longer
periods of times. Half-lives of prodrugs in the presence of
β-glucosidase are reported in Table 3.
Biological activities
ANTIPROLIFERATIVE ACTIVITY. When analyzed for their antiproli-
ferative activity against human tumor A2780, A549 and
MDA-MB-231 cells, in the absence of Cu2+ ions, the glycosides
displayed moderate IC50s that ranged from 4.1 μM to 55.3 μM
(mean SD, 26.2 17.8 μM), the exception being the glycoside
GluMeHQ that displayed no activity (IC50 > 100 μM) also in the
presence of Cu2+ ions and, above all, compared to its non-gly-
cosylated counterpart (MeHQ, Table 4).
Fig. 5 Histograms represent mean IC50s of glucosylated and non-glucosylated
hydroxyquinolines in the presence or absence of the β-glucosidase inhibitor
DMDP at 100 μM concentration.
In the presence of 20 μM Cu2+ we observed a deep reduction
of IC50s for both the active compounds GluClHQ (mean
reduction, 90
7%) and GluCl2HQ (mean reduction, 63
inhibitor 2,5-dideoxy-2,5-imino-D-mannitol (DMDP) signifi-
30%) (Table 4), with compound GluClHQ that displayed a cantly increased the IC50 of both glycosylated compounds
higher mean antiproliferative activity compared to GluCl2HQ GluClHQ and GluCl2HQ. As negative control we also evaluated
(2.04
2.23 μM vs. 12.4
11.2 μM). A similar increase of the antiproliferative activity of non-glycosylated ClHQ and
activity in the presence of the Cu2+ ion and a similar ratio Cl2HQ compounds in the presence of DMDP. As expected in
between the mean antiproliferative activity were also observed this case, no variation of the antiproliferative activity was
for the parent compounds ClHQ and Cl2HQ (Table 4).
observed demonstrating that glycosylated prodrugs may be
As regards the target cell lines, the breast cancer cell line activated only in the presence of an efficient β-glycosidase
MDA-MB-231 was the less sensitive target for both glycosylated activity that releases the active 8-hydroxyquinoline moiety.
and non-glycosylated compounds in most culture conditions.
IDENTIFICATION OF APOPTOTIC CELLS BY DAPI AND ANNEXIN V/PI STAININGS.
EFFECT OF THE INHIBITION OF GLYCOSIDASE ACTIVITY. As shown in Due to their pharmacologically significant antiproliferative
Fig. 5, the presence in culture conditions of the β-glycosidase activity both GluClHQ and GluCl2HQ and their non-
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 2023–2034 | 2027

View Article Online
Paper
Dalton Transactions
glycosylated parent compounds ClHQ and Cl2HQ were ana- pocket. The active site is shaped as an oval or a slit-like pocket
lyzed by DAPI staining for the triggering of apoptosis. In with a cluster of hydrophobic amino acid sidechains (in yellow
general, the compounds showed a pharmacologically relevant in Fig. 7B) lining the walls of the aglycone binding slot,
apoptotic activity that was higher when the Cu2+ ions were whereas hydrophobic, polar and charged residues are present
present in culture. Compound Cl2HQ represented an exception in the glycone binding moiety.
to this general rule in fact, at both equitoxic concentrations,
we did not observe significant differences in terms of apopto- ing molecular docking studies. Investigated compounds were,
tic cells in the presence or absence of Cu2+ ions (Table 5).
therefore, docked to the binding cavity of 2JFE to evaluate
In order to confirm the previous data the GluClHQ and their binding mode and affinity (kcal mol−1).
2JFE provides useful 3D structural information for perform-
GluCl2HQ compounds were also analyzed for their ability to
To do that among the plethora of docking softwares, Auto-
trigger apoptosis by the annexin V/PI staining. Once used at Dock Vina30 was selected as the main computational tool since
their IC50s these compounds confirmed the previous results it was recently used to model hydroxyquinolines in docking
obtained by the microscopy observation of nuclear mor- studies.31 The performances of AutoDock Vina were evaluated
phology after DAPI staining. In particular, these results by reproducing the crystal structure of the complex of the
showed that a higher early and/or late apoptotic activity was maize β-glucosidase (ZMGlu1) with the non-hydrolysable
developed in the presence of Cu2+ ions (Fig. 6).
inhibitor p-nitrophenylbeta-D-thioglucoside (PDB code: 1E1F).
In this validation step we paid particular attention to check
the position of the glycone moiety at the bottom of the active
Docking studies
The crystallographic structure of human cytosolic β-glucosi- site close to the two catalytic residues (Glu165 and Glu373
dase29 (hCBG, PDB code 2JFE, Fig. 7A) shows that the active shown in Fig. 7A for hCBG). The orientation of the glycone
site of the enzyme is located on the bottom of a deep pocket moiety of the substrate was, in fact, used as a filter to
(Fig. 7B). It can be divided into two parts; the recognition site evaluate hCBG docking results. In particular, we retained
for the glucose moiety located at the bottom of the active site the best poses of any ligand (= the most stable) provided that
and the binding site for the aglycone at the entrance of the their glycone moiety was located at the bottom of the active
Table 5 Triggering of apoptosis by glucoconjugate compounds, as evaluated by nuclear morphological analysis after DAPI staining
a
IC50
Cl2HQ
GluCl2HQ
ClHQ
GluClHQ
Cu2+
Cu2+
Cu2+
Cu2+
A2780
A549
MDA-MB-231
69.5 26.1b
10.2 6.2
2.8 1.0
59.8 16.2
7.3 3.2
2.1 1.8
14.5 2.1
15.5 4.2
2.3 1.5
24.5 0.7
38.4 15.0
34.0 12.2
5.8 2.9
4.8 3.9
3.0 2.6
33.0 13.9
14.2 6.5
65.0 7.0
7.0 4.6
4.0 2.0
3.0 2.0
27.0 13.8
22.6 3.6
67.0 4.7
a Each cell line was treated with the specific equitoxic concentrations IC50, as calculated by the MTT assay. b The means SD (4–9 data) express
the percentage of apoptotic cells.
Fig. 6 Scatter plots of A2780 and A549 cells labeled with Annexin-V/PI after three days culture in the presence of GluClHQ and GluCl2HQ with or without Cu2+
ions.
2028 | Dalton Trans., 2013, 42, 2023–2034
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
entirely new since previously other authors had identified this
mechanism as essential for the activation of their glycosylated
compounds.39–41 Taken together, all these works clearly
demonstrate that the expression of the β-glucosidase or
β-galactosidase activities is responsible for the bioactivation of
the novel glycosylated prodrugs with the formation of the
active aglycone moieties. Of course the level of expression of
these enzymes into cancer cells remains an essential point dis-
tinguishing the pharmacological from the toxic effect of these
compounds. On the whole a good balance between the β-gluco-
sidase activity and the transmembrane glucose transporters
may ensure the right and effective activation of glycosylated
prodrugs.42 In practice, high cellular β-glycosidase levels might
be effective only when transport proteins are expressed as well.
An important future goal for the anticancer research is the
oral administration of antitumor drugs, which is expected to
be accompanied by much greater patient tolerability and
reduced associated cost. To achieve this aim the application in
the early stages of the drug discovery process of knowledge of
the compound’s absorption, distribution, metabolism and
excretion (ADME) profile is mandatory. In fact, the optimi-
zation of the ADME profile is expected to significantly improve
the candidates’ ‘druggability’43 and in fact, few of the anti-
cancer drugs currently in the clinic have optimized ADME fea-
tures.44 Also solubility issues should be taken into
consideration. Here we designed compounds that are pre-
dicted to show drug-like physicochemical properties since they
are all within the range set by Lipinski’s Ro5. Moreover, pro-
drugs are predicted to be significantly more soluble than
parent compounds. Finally, ADME in silico data predict a good
pharmacokinetic profile at least for distribution and meta-
bolism events.
Fig. 7 (A) The crystallographic structure of human cytosolic β-glucosidase
(hCBG, PDB code 2JFE); the two catalytic residues (Glu165 and Glu373) are evi-
denced (B) the active site of hCBG; (C) the best poses of the five docked struc-
tures as obtained with AutoDock Vina; (D) the best poses of the five docked
structures in the active site (hydrophobic residues in yellow; hydrophilic residues
in light blue).
site. This was verified for the three best poses of all
compounds.
Results (Fig. 7C and D) indicated that GluMeHQ has
the best affinity for the binding site (8.4 kcal mol−1 for
the best pose), whereas GluCl2HQ shows the lowest affinity
(7.04 kcal mol−1 for the best pose). The three remaining com-
pounds show intermediate values. FLAP32 analysis confirmed
this trend (data not shown).
GluClHQ, GluCl2HQ and GluMeHQ were synthesized in
good yields through a straightforward procedure. The syn-
thesis was optimized in order to obtain the pure β-anomer in a
one-pot reaction. The β-anomers were the compounds of inter-
est in this study because the tumor cells generally have a high
Discussion
The main purpose of the study was the design of new com- expression of β-glucosidase. The desired products have been
pounds able to selectively target tumor cells exploiting some of fully characterized by NMR, UV-vis, CD, and ESI-MS. The
their specific biochemical features.
purity of compounds has been confirmed by different tech-
Firstly, as known and described previously, tumor cells niques. The compounds were obtained pure without the use of
have the unique characteristic of elevated copper concen- elaborate processes of purification. This aspect is very impor-
trations.33,34
tant, considering one of the main reasons of CQ withdrawal
Secondly, tumor cells express an altered carbohydrate from the market was the presence of di-iodo impurities. Fur-
metabolism that is characterized by an elevated glycolysis, thermore, O-glycosylation of compounds masks the metal
which in turn justifies the glucose avidity described in the binding site, thereby preventing systemic chelation. This is the
1950s by Otto Warburg and generated the idea of the so-called reason why the glucosides are not able to complex metal ions.
“Warburg effect”.35 This concept implies that tumor cells, due The pyridine ring can complex Cu2+ only with a log K1 about
to their glucose avidity, possess an increased glucose transport 2.6,45 thus the glucose moiety has to be cleaved before efficient
caused by an over-expression of the glucose transporters.36,37
Finally, Arafa38 demonstrated that in the cancer cells used
metal chelation of the quinoline ring.
ClHQ, Cl2HQ and MeHQ are able to complex Cu2+ with
for his study the different β-glucosidase activity profiles well good stability constants,45 forming CuL and CuL2 as main
correlated with the cytotoxicity rank order of the glucosides species. The amount of each complex species depends on the
object of his research. This demonstration underlines the concentration of Cu2+ and the ligands.
important role that these enzyme activities could play in the
bioactivation of glycosylated prodrugs. This concept was not makes glycosylated compounds more active in terms of
We have already described22 that the presence of Cu2+ ions
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 2023–2034 | 2029

View Article Online
Paper
Dalton Transactions
antiproliferative activity than under culture conditions charac- we also performed antiproliferative experiments in the pres-
terized by lower Cu2+ concentration. This difference is con- ence of Cu2+ ions and the known β-glucosidase inhibitor
firmed by present data where the presence of a high Cu2+ DMDP. Interestingly, the IC50s of the most active glycosylated
concentration improved the activity of the derivatives by about hydroxyquinolines GluClHQ and GluCl2HQ, after treatment
ten times. The presence of a discrete activity in the absence of with DMDP, significantly increased by several times, while
added Cu2+ ions may be ascribed to the possible high Cu2+ DMDP did not affect the activities of the parent compounds.
content of tumor cells utilized for our tests in vitro. It is also These data unequivocally confirm the importance of the pres-
worth noting that ClHQ is the most active hydroxyquinoline of ence of β-glucosidase activity to allow the release of the active
the series here studied, with IC50 ranging between 5.12
0.94 μM and 18.7 5.9 μM.
quinoline moiety in order to develop the antiproliferative
activity after the binding of the Cu2+ ions.
An important result of the study is finding that glycosylated
The differences in sensitivity between the cell lines, in par-
compounds show an antiproliferative activity nearly similar to ticular the lower sensitivity of the breast cancer cell line
that of their parent compounds except for the methyl deriva- MDA-MB-231, could be interpreted on the basis of the absence
tive that showed a very low activity in all three cell lines used of a sufficient number of glucose-transporters since the level
for our in vitro tests.
of β-glucosidase (data not shown) was similar to those
These data are consistent with the glucosidase cleavage observed in the other two cell lines A2780 and A549. Suitable
assay in vitro and docking studies. Experimental cleavage dosages of the concentration of membrane glucose-transpor-
studies suggest that the glucoconjugates are not hydrolyzed in ters could better explain our results.
the same way by glucosidase. In fact, GluMeHQ, which results
A final point to be highlighted concerns the mechanism of
inactive in the antiproliferative assay, is the compound cleaved action of OHQs as anticancer agents. In the last few years the
in longer times in comparison to the other glycosides. These ubiquitin–proteasome system (UPS) has emerged as a possible
experimental results are supported by docking studies that target for new anticancer compounds due to its involvement in
show that prodrugs with high t1/2 are more tightly bound to the regulation and control of important protein in turn
the active site than those with low t1/2. This trend is verified by involved in vital cell mechanisms, such as proliferation, signal
using two docking algorithms showing very different theoreti- transduction, gene transcription, DNA repair and apoptosis.50
cal approaches and could suggest that compounds that are too The UPS acts in the turnover of proteins thanks to the involve-
tightly bound to the enzyme are slowly hydrolysed. We are also ment of tens of genes51,52 some of which code enzymes
aware that this relationship between experimental and compu- working in processes, such as the protein ubiquitination and
tational data should be considered with caution for at least deubiquitination with very specific mechanisms of targeting.53
two reasons. Firstly, it is known that the active site of β-glucosi- At present the bulk work on the UPS has led to the synthesis
dases is well-conserved, but here experimental and compu- and clinical use of Bortezomib, an anticancer drug specific for
tational data were obtained using two different enzymes, the chymotrypsin-like activity of proteasome whose inhibition
almond glucosidase and hCBG, respectively. Secondly, it must seems to be directly involved in the induction of apoptosis in
be kept in mind that docking calculations can simulate only tumors,54 which is mainly used in clinics for the treatment of
the recognition phase between enzyme and substrates and give myeloma.55 Other hydroxyquinolines and their copper com-
no information on the whole enzymatic phenomenon.
plexes have as the main mechanism of action the ability to
As for the apoptosis, the opinion, initially quite contro- interact with the UPS, in particular with the chymotrypsin-like
versial, that this death mechanism plays an essential role in activity of proteasome, causing the arrest of the cell cycle and
the response to the chemotherapeutics46,47 has now been con- the triggering of apoptosis. The glycosylated compounds,
sidered a landmark for anticancer drug research and for new object of this study, could have the same mechanism of action
chemotherapy-based procedures. It is well known that apopto- once deprived of the glucose moiety thanks to the action of
sis represents, together with necrosis, an important pathway of β-glucosidase.
cell death after chemotherapy. Thus, the evaluation of the trig-
gering of apoptosis by the new compounds with anticancer
activity becomes necessary both in in vitro and in vivo
Conclusions
studies.48,49 In the case of our glucoconjugates it is worth
noting that their antiproliferative activity was always developed We designed and synthesized new glucoconjugates of OHQ
together with a significant triggering of this important and derivatives so as to exploit three common features of cancer
essential death event for chemotherapeutic agents and that cells, such as the high Cu2+ concentration and high expression
this was in general higher when Cu2+ ions were added to the and content of β-glycosidase and the high membrane concen-
culture medium under equitoxic conditions of antiproliferative tration of glucose transporters linked to the “natural” glucose
activity. The only exception to this rule is the use of Cl2HQ avidity of tumors. We show that these features can be synergis-
where the presence of Cu2+ ions does not imply an increased tically combined to obtain new compounds highly specific for
apoptosis.
neoplastic cells. Physicochemical and ADME considerations
In order to verify that the glucosidase enzymatic activation were also included in the design phase of the project. The
of glucoconjugates is crucial for their antiproliferative activity exploitation of these characteristics, all contained in a single
2030 | Dalton Trans., 2013, 42, 2023–2034
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
molecule, should not only allow for a better selection of target 5-Chloro-8-quinolinyl-β-D-glucopyranoside (GluClHQ)
tumor cells, but also minimize the chemotherapeutic side
effects.
Yield: 68%. TLC: Rf = 0.51 (PrOH–AcOEt–H2O–NH3 4 : 4 : 1 : 1).
HPLC purity 98%.
ESI-MS: m/z = 341.8 [GluClHQ + H]+, 363.9 [GluClHQ +
Na]+, 704.6 [2GluClHQ + Na]+.
1H NMR (500 MHz, CD3OD) δ (ppm): 8.97 (dd, J2,3 = 4.3,
Experimental section
Materials
J2,4 = 1.6 Hz, 1H, H-2), 8.71 (dd, J4,3 = 8.6, J4,2 = 1.6 Hz, 1H,
H-4), 7.75 (dd, J3,4 = 8.6, J3,2 = 4.3 Hz, 1H, H-3), 7.67 (d, J6,7
8.5 Hz, 1H, H-6), 7.51 (d, J7,6 = 8.5 Hz, 1H, H-7), 5.09 (d, J1,2
=
=
Commercially available reagents were used directly, unless
otherwise noted. 5-Chloro-8-quinolinol (ClHQ), 5,7-dichloro-
8-quinolinol (Cl2HQ), 2-methyl-8-quinolinol (MeHQ) were pur-
chased by Sigma-Aldrich. TLC was carried out on silica gel
plates (Merck 60-F254). β-Glucosidase from almonds was
obtained from Sigma-Aldrich.
7.8 Hz, 1H, H-1 of Glu), 3.97 (dd, J6a,6b = 12.1, J6a,5 = 2.2 Hz,
1H, H-6a of Glu), 3.81–3.64 (m, 2H, H-2 and H-6b of Glu),
3.60–3.54 (m, 2H, H-3, H-5), 3.45 (dd, J = 9.6, 9.0 Hz, 1H, H-4).
13C NMR (125 MHz, CD3OD) δ (ppm): 153.2 (C-8), 151.1
(C-2), 140.9 (C-9), 135.1 (C-4), 128.4 (C-6, C-5), 125.0 (C-10),
124.1 (C-3), 114.5 (C-7), 103.3 (C-1 of Glu), 78.6 (C-5 of Glu),
77.4 (C-3), 74.7 (C-2 of Glu), 71.6 (C-4 of Glu), 63.0 (C-6 of Glu).
UV-vis: (CH3OH) λ/nm (ε): 243 (25 600), 317 (3360).
NMR spectroscopy
1H and 13C NMR spectra were recorded in CD3OD at 25 °C
with a Varian UNITY PLUS-500 spectrometer at 499.889 and at
125.7 MHz respectively. The NMR spectra were obtained by
using standard pulse programs from Varian library. The 2D
experiments (COSY, TOCSY, gHSQCAD, gHMBC) were acquired
using 1 K data points, 256 increments and a relaxation delay of
1.2 s. DSS was used as an external standard.
CD (CH3OH) λ/nm (Δε): 197.8 (−4.24), 311.4 (−0.70).
Elemental analysis for C15H16NClO6·H2O: calc. C 50.06;
H 5.05; N 3.89; found C, 50.00; H, 4.99; N, 3.94.
5,7-Dichloro-8-quinolinyl-β-D-glucopyranoside (GluCl2HQ)
Yield: 61%. TLC: Rf = 0.52 (PrOH–AcOEt–H2O–NH3 4 : 4 : 1 : 1).
HPLC purity 98%. ESI-MS: m/z = 375.8 [GluCl2HQ + H]+, 397.9
[GluCl2HQ + Na]+, 772.5 [2GluCl2HQ + Na]+.
Mass spectrometry, UV-vis and circular dichroism
spectroscopy
1H NMR (500 MHz, CD3OD) δ (ppm): 8.98 (dd, J2,3 = 4.3 Hz,
J2,4 = 1.5 Hz, 1H, H-2), 8.68 (dd, J4,3 = 8.6, J4,2 = 1.6 Hz, 1H,
H-4), 7.82 (s, 1H, H-6), 7.72 (dd, J3,4 = 8.6, J3,2 = 4.3 Hz, 1H,
The ESI-MS measurements were performed using a Finnigan
LCQ-Duo ion trap electrospray mass spectrometer. The UV
spectra were recorded using an Agilent 8452A diode array spec-
trophotometer. CD measurements were performed under a
constant flow of nitrogen using a JASCO model J-810 spectro-
polarimeter. The spectra represent the average of 10 scans and
were recorded at 25 °C, on freshly prepared methanol
solutions.
H-3), 5.34 (d, J1,2 = 7.8 Hz, 1H, H-1 of Glu), 3.77 (dd, J6a,6b
=
11.9, J6a,5 = 2.4 Hz, 1H, H-6a), 3.72 (dd, J2,3 = 8.8, J2,1 = 7.8 Hz,
1H, H-2), 3.66 (dd, J6b,6a = 11.9, J6b,5 = 5.3 Hz, 1H, H-6b),
3.50–3.42 (m, 2H, H-4, H-3), 3.23 (ddd, J5,4 = 9.4, J5,6b = 5.3,
J5,6a = 2.3 Hz, 1H, H-5).
13C NMR (125 MHz, CD3OD) δ (ppm): 152.3 (C-2), 143.4
(C-8), 144.1 (C-9), 135.6 (C-4), 130.0 (C-6), 129.2 (C-7), 128.5
(C-5), 127.8 (C-10), 124.2 (C-3), 107.3 (C-1 of Glu), 78.6 (C-5 of
Glu), 78.4 (C-3), 72.2 (C-2 of Glu), 71.3 (C-4 of Glu), 62.7 (C-6 of
Glu).
General procedure for the synthesis of the glucoconjugates
8-Quinolinol derivative (0.75 mmol) and K2CO3 (7.5 mmol)
were added to water (50 mL) and methanol (40 mL). Dichloro-
methane (100 mL) was added to this aqueous solution,
followed by acetobromoglucose (1.88 mmol) and tetrabutyl-
ammonium bromide (0.75 mmol). The resulting mixture was
vigorously stirred for 68 h. As for ClHQ and Cl2HQ, a solid pre-
cipitated, and the deacetylated products were collected by
filtration, washed with cool methanol and dried. As for MeHQ,
the two-phase system was separated, and the aqueous phase
UV-vis (CH3OH) λ/nm (ε): 207 (20 900), 242 (24 800), 302
(3020).
CD (CH3OH) λ/nm (Δε): 206.6 (−5.10), 240.2 (−10.77), 318.0
(0.73).
Elemental analysis for C15H15NCl2O6·H2O: calc. C 45.68; H
4.35; N 3.55; found C, 45.59; H, 4.30; N, 3.58.
2-Methyl-8-quinolinyl-β-D-glucopyranoside (GluMeHQ)
was washed repeatedly with dichloromethane. The aqueous Yield: 57%. TLC: Rf = 0.44 (PrOH–AcOEt–H2O–NH3 4 : 4 : 1 : 1).
phase was partially evaporated under vacuum to remove HPLC purity 98%. ESI-MS: m/z = 321.9 [GluMeHQ + H]+,
residual organic solvents and then a white solid precipitated. 344.00 [GluMeHQ + Na]+, 359.9 [GluMeHQ + K]+, 664.6
The product was collected by filtration, washed with cool [2GluMeHQ + Na]+, 680.8 [2GluMeHQ + K]+.
methanol and dried.
1H NMR (500 MHz, CD3OD) δ (ppm): 8.24 (d, J4,3 = 8.4 Hz,
The purity was assayed by HPLC (Varian Prepstar instru- 1H, H-4), 7.56 (dd, J5,6 = 6.8, J5,7 = 2.7 Hz, 1H, H-5), 7.48 (m,
ment equipped with a Prostar 330 photodiode array detector). 3H, other aromatic H), 5.05 (d, 1H, J1,2 = 7.8. Hz, H-1 of Glu),
Column: Econosphere ODS (Alltech) column (5 μm, 2.0 × 3.96 (dd, 1H, J6a,6b = 12.1, J6a,5 = 2.2 Hz, H-6a of Glu), 3.75 (dd,
250 nm), eluent: (linear gradient water → CH3CN).
1H, J6a,6b = 12.1, J6b,5 = 15.0 Hz, H-6b of Glu), 3.73 (t, 1H, H-2
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 2023–2034 | 2031

View Article Online
Paper
Dalton Transactions
of Glu), 3.56 (m, 2H, H-5 and H-3 of Glu), 3.46 (t, 1H, J4,3 = J4,5 8452A diode array spectrophotometer at 37 °C. All assays were
= 18 Hz, H-4 of Glu), 2.77 (s, 3H, CH3).
performed in triplicate.
13C NMR (125 MHz, CD3OD) δ (ppm): 154.5 (C-2), 148.0
(C-8), 133.2 (C-9), 134.7 (C-4), 123.6 (C-6, C-10), 120.3 (C-3),
118.9 (C-5), 111.5 (C-7), 105.3 (C-1 of Glu), 77.0 (C-3 and C-5 of
Glu), 74.7 (C-2 of Glu), 69.8 (C-4 of Glu), 61.2 (C-6 of Glu), 20.8
(CH3).
UV-vis (CH3OH) λ/nm (ε): 204 (31 700), 240 (32 200), 297
(3110).
CD (CH3OH) λ/nm (Δε): 196.8 (−2.68), 312.6 (−0.34).
Elemental analysis for C15H15NO6·H2O: calc. C 56.62;
H 6.24; N 4.13; found C, 56.65; H, 6.21; N, 4.09.
Dilution of glycosides and parent compounds for biological
determinations
All glycosides and parent compounds were firstly dissolved in
100% dimethylsulfoxide (DMSO) at a concentration of 100 mM
and then further diluted in fetal calf serum (FCS, final concen-
tration DMSO 0.2–0.4%).
Determination of antiproliferative activity by the MTT assay
Human cell lines A2780 (ovary, adenocarcinoma), A549 (lung,
carcinoma) and MDA-MB-231 (breast, carcinoma) were plated
at 105 mL−1, 0.89 × 105 mL−1 and 2 × 105 mL−1, respectively, in
a 180 μL complete medium into flat-bottomed 96-well micro-
titer plates. After 6–8 hours cells were treated with 20 μL con-
taining five 1 : 10 fold concentrations of our compounds
diluted in FCS containing 2% or 4% DMSO in the presence or
absence of CuCl2 salt (20 μM). After 72 hours plates were pro-
cessed and IC50s values calculated as described elsewhere.56
Each experiment was repeated 4–8 times.
Copper complex studies
The copper complexes were prepared by adding a solution of
CuSO4 to a ligand solution in a 1 : 1 ratio. ClHQ and Cl2HQ are
poorly soluble in water and their metal complexes were exam-
ined in a methanol–water mixture (80 : 20 w/w).
All the ESI-MS measurements were carried out by using a
Finnigan LCQ DECA XP PLUS ion trap spectrometer operating
in the positive ion mode and equipped with an orthogonal ESI
source (Thermo Electron Corporation, USA). Sample solutions
were injected into the ion source without the addition of any
other solvent at a flow rate of 5 μL min−1. For electrospray ion-
ization the drying gas (N2) was heated at 270 °C. The capillary
exit and skimmer voltage were varying in order to optimize the
signal responses. Scanning was performed from m/z = 100 to
2000 and no fragmentation processes were observed under our
experimental conditions. Xcalibur software was used for the
elaboration of mass spectra. Each species is indicated with the
m/z value of the first peak of its isotopic cluster.
Inhibition of glucosidase activity
The evaluation of antiproliferative activity was performed
using A2780 and A549 cells in the presence of Cu2+ ions and
100 μM of the β-glycosidase inhibitor 2,5-dideoxy-2,5-imino-D-
mannitol (DMDP). The same experiments were carried out
without DMDP as a control.
Microscopy visualization of apoptotic cells after 4′-6-
diamidino-2-phenylindole (DAPI) staining
In vitro stabilities of glucoconjugates
The three final compounds were incubated at 37 °C in PBS to
monitor their stabilities by TLC, ESI-MS and UV-vis spectro-
scopy. Furthermore, the compound stability was monitored at
pH 4 and 11.
A2780, A549 and MDA-MB-231 were plated in 24-well plates at
their optimal densities per well (see above) in a 1 mL complete
medium for 6–8 h. ClHQ, Cl2HQ and their glycosylated
counterparts were then added to obtain their specific final
IC50, as determined by the MTT assay. After 72 hours floating
and adherent cells were harvested, washed twice with cold
PBS, fixed with 60 μL of 70% ethanol in PBS and maintained
at 4 °C until analysis. For microscopy examination 5 μL of
10 μg mL−1 DAPI in water were added to the samples, and the
percentage of apoptotic segmented nuclei per cells evaluated.
Enzymatic cleavage assay
Glycoconjugates were incubated at 37 °C for 2 h. TLC and
ESI-MS monitoring of the glucosidase reactions were used to
determine whether the clioquinol glycosides showed any sign
of being cleaved by the enzyme. Enzymatic cleavage was also
monitored by UV-vis spectroscopy. Unless stated otherwise,
β-glucosidase (1.0 × 10−6 M) was incubated in the assay solu-
tion in the presence of the glycoconjugates (5.0 × 10−5 M) at
Evaluation of apoptosis by annexin-V/PI staining
37 °C. Samples (2 mL) were removed at selected time points The triggering of apoptosis was also evaluated for the glyco-
over 4 h and were immediately added with a solution of NaOH sylated compounds GluClHQ and GluCl2HQ in A2780 and
until the pH value of the solution reached 11.0 in order to stop A549 cells. Cells were treated with the specific IC50 of each
enzymatic activity. Furthermore at this pH the released drug compound in the presence or absence of Cu2+ salt (20 μM).
shows a shift of the bands because of the deprotonation of the Three days after floating and adherent cells were harvested,
phenolic group and so it is possible to quantify the released washed with cold PBS and early and late apoptotic cells deter-
drug through UV-vis spectroscopy. Calibration curves of the mined by double staining with FITC-annexin-V/PI (rh Annexin-
free ligands for absorbance against their concentrations were V/FITC Kit, Bender MedSystem GmbH, Vienna, Austria). The
obtained at pH 11.0. The UV-vis spectra were recorded in subsequent analysis was performed by flow cytometry as
1.0 cm path length matched quartz cuvettes with an Agilent described elsewhere.57
2032 | Dalton Trans., 2013, 42, 2023–2034
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
Physicochemical and ADME profile
5 http://clinicaltrials.gov/ct2/show/study/NCT00963495
6 S. Zhai, L. Yang, Q. C. Cui, D. Sun, Q. P. Ying and B. Yan,
J. Biol. Inorg. Chem., 2010, 15, 259.
7 L. Banci, I. Bertini, F. Cantini and S. Ciofi-Baffoni, Cell.
Mol. Life Sci., 2010, 67, 2563.
Solubility (expressed as ALOGpS, the higher, the more soluble
the compound) was calculated with VCCLAB, Virtual Compu-
tational Chemistry Laboratory, http://www.vcclab.org, 2005.58
The conformational hypersurface of compounds was
explored with the standard conformational search module
implemented in MOE2009.10. Virtual log P calculations were
performed with VEGA ZZ 3.0.0 (http://www.vegazz.net).
The 3D structures (see below) were submitted to VolSurf+
(version 1.0.4, Molecular Discovery Ltd. Pinner, Middlesex, UK,
2009, http://www.moldiscovery.com)59 using default settings
and four probes (OH2, DRY N1 and O probes that mimic
respectively water, hydrophobic, hydrogen bond acceptor and
hydrogen bond donor interaction of the compounds with the
environment). The molecules were projected on the pre-
calculated models implemented in VolSurf+ to calculate Ro5
and ADME parameters.
8 S. Brem, Cancer Control, 1999, 6, 436.
9 G. J. Brewer, Exp. Biol. Med., 2001, 226, 665.
10 H. I. Pass, G. J. Brewer, R. Dick, M. Carbone and
S. Merajver, Ann. Thorac. Surg., 2008, 86, 383.
11 N. L. Henry, R. Dunn, S. Merjaver, Q. Pan, K. L. Pienta,
G. Brewer and D. C. Smith, Oncology, 2006, 71, 168.
12 G. J. Brewer, F. Askari, R. B. Dick, J. Sitterly, J. K. Fink,
M. Carlson, K. J. Kluin and M. T. Lorincz, Transl. Res.,
2009, 154, 70.
13 C. Duncan and A. R. White, Metallomics, 2012, 4, 127–138.
14 A. D. Schimmer, Curr. Cancer Drug Targets, 2011, 11, 325.
15 V. Milacic, P. Jiao, B. Zhang, B. Yan and Q. P. B. Dou,
Int. J. Oncol., 2009, 35, 1481.
16 M. A. Cater and Y. Haupt, Biochem. J., 2011, 436, 481.
17 M. Katsuyama, K. Iwata, M. Ibi, K. Matsuno, M. Matsumoto
and C. Yale-Nishimura, Toxicology, 2012, 299, 55.
18 M. Palombo, P. J. Sinko and Y. Singh, Curr. Med. Chem.,
2008, 15, 1802.
19 T. Nishioka, Y. Oda, Y. Seino, T. Yamamoto, N. Inagaki,
H. Yano, H. Imura, R. Shigemoto and H. Kikuchi, Cancer
Res., 1992, 52, 3972.
20 K. C. Carvalho, I. W. Cunha, R. M. Rocha, F. R. Ayala,
M. M. Cajaíba, M. D. Begnami, R. S. Vilela, G. R. Paiva,
R. G. Andrade and F. A. Soares, Clinics (Sao Paulo), 2011,
66, 965.
21 H. Cheng, X. Cao, M. Xian, L. Fang, T. V. Cai, J. J. Ji,
J. B. Tunac, D. Sun and P. G. Wang, J. Med. Chem., 2005,
48, 645.
Molecular docking
The crystal structure of human cytosolic β-glucosidase was
obtained from the RCSB Protein Data Bank, code 2JFE. The 3D
structure was visualized and modified with Pymol v1.1.r1 as
follows. Water and NAG molecules were removed, hydrogen
atoms were added and the resulting structure was saved in the
pdb format. Ligands were built using CORINA60 and saved in
the mol2 format. AutoDock Vina30 and FLAP UI version
0.43.032 were used to perform docking studies. The active site
in AutoDock was defined by the coordinates of Cα of the cata-
lytic residue Glu165 (33 (X), 51 (Y) and 38 (Z)).29 To ensure a
correct docking process, the search space was defined to
encompass all flexible residues as a box with dimensions of
30.0 × 30.0 × 30.0 Å.
22 V. Oliveri, M. L. Giuffrida, G. Vecchio, C. Aiello and
M. Viale, Dalton Trans., 2012, 41, 4530.
23 H. Han van de Waterbeemd, Chem. Biodiversity, 2009, 6,
1760.
Statistical analysis
The Mann–Whitney test for non-parametric data was used for
the statistical analysis of data.
24 P. Gaillard, P. A. Carrupt, B. Testa and A. Boudon,
J. Comput. Aided Mol. Des., 1994, 8, 83.
25 E. Carosati, R. Budriesi, P. Ioan, M. P. Ugenti, M. Frosini,
F. Fusi, G. Corda, B. Cosimelli, D. Spinelli, A. Chiarini and
G. Cruciani, J. Med. Chem., 2008, 51, 5552.
26 Y. Mimaki, K. Watanabe, Y. Ando, C. Sakuma, Y. Sashida,
S. Furuya and H. Sakagami, J. Nat. Prod., 2001, 64, 17.
27 A. Rodger and B. Norden, Circular Dichroism and Linear
Dichroism, Oxford University Press, New York, 1997.
28 A. Budimir, N. Humbert, M. Elhabiri, I. Osinska, M. Birus
and A. M. Albrecht-Gary, J. Inorg. Biochem., 2011, 105, 490.
Acknowledgements
We thank MIUR [2008R23Z7K, PRIN 2008F5A3AF,
FIRB2010_RBAP114AMK and RBNB08HWLZ (Merit)] for the
financial support. We thank Tiziana Campagna for her techni-
cal assistance.
Notes and references
1 A. Jemal, R. Siegel, J. Xu and E. Ward, CA Cancer J. Clin., 29 S. Tribolo, J. G. Berrin, P. A. Kroon, M. Czjzek and N. Juge,
2010, 60, 277. J. Mol. Biol., 2007, 370, 964–975.
2 A. Jemal, F. Bray, M. M. Center, J. Ferlay, E. Ward and 30 O. Trott and A. J. Olson, J. Comput. Chem., 2010, 31, 455.
D. Forman, CA Cancer J. Clin., 2011, 61, 69.
3 L. W. H. Hung and J. K. Barnham, Future Med. Chem., 2012,
4, 955.
31 C. Bruyère, S. Madonna, G. Van Goietsenoven, V. Mathieu,
J. Dessolin, J. L. Kraus, F. Lefranc and F. Kiss, Transl.
Oncol., 2011, 4, 126.
4 M. S. Yassin, J. Ekblom, M. Xilimas and C. G. Gottfries, 32 S. Cross, M. Baroni, E. Carosati, P. Benedetti and
J. Neurol. Sci., 2000, 173, 40.
S. Clementi, J. Chem. Inf. Model., 2010, 50, 1442.
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 2023–2034 | 2033

View Article Online
Paper
Dalton Transactions
33 K. Geraki, M. J. Farguharson and D. A. Bradley, Phys. Med. 48 J. M. Lerma-Díaz, G. Hernández-Flores, J. R. Domínguez-
Biol., 2002, 47, 2327.
Rodríguez, P. C. Ortíz-Lazareno, P. Gómez-Contreras,
R. Cervantes-Munguía, D. Scott-Algara, A. Aguilar-Lemar-
roy, L. F. Jave-Suárez and A. Bravo-Cuellar, Immunol. Lett.,
2006, 103, 149.
34 M. Diez, M. Arroyo, F. J. Cerdan, M. Munoz, M. A. Martin
and J. L. Balibrea, Oncology, 1989, 46, 230.
35 C. D. Young and S. M. Anderson, Breast Cancer Res., 2008,
10, 202.
49 Y. Yoshimoto, M. Kawada, D. Ikeda and M. Ishizuka, Int.
Immunopharmacol., 2005, 5, 281.
36 T. Nishioka, Y. Oda, Y. Seino, T. Yamamoto, N. Inagaki,
H. Yano, H. Imura, R. Shigemoto and H. Kikuchi, Cancer 50 Y. Yang, J. Kitagaki, H. Wang, D. Hou and A. O. Perantoni,
Res., 1992, 52, 3972. Cancer Sci., 2009, 100, 24.
37 T. Kawamura, T. Kusakabe, T. Sugino, K. Watanabe, 51 S. S. Vembar and J. L. Brodsky, Nat. Rev. Mol. Cell Biol.,
T. Fukuda, A. Nashimoto, K. Honma and T. Suzuki, Cancer,
2001, 92, 634.
38 H. M. M. Arafa, Eur. J. Pharmacol., 2009, 616, 58.
2008, 9, 944.
52 D. Ron and P. Walter, Nat. Rev. Mol. Cell Biol., 2007, 8, 519.
53 Y. Liu and Y. Ye, Cell Res., 2011, 21, 867.
39 K. N. Syrigos and G. Rowlinson-Busza, Int. J. Cancer, 1998, 54 B. An, R. H. Goldfarb, R. Siman and Q. P. Dou, Cell Death
78, 712. Differ., 1998, 5, 1062.
40 H. Seker, B. Bertram, H. Troster, R. Fisher, 55 J. P. Leonard, R. R. Furman and M. Coleman, Int. J. Cancer,
M. Trendelenburg and M. Wiessler, Annu. Meet. Am. Assoc.
Cancer Res., 2000, 41, 602.
41 M. De Graaf, H. M. Pinedo, R. Quadir, H. J. Haisma and
E. Boven, Biochem. Pharmacol., 2003, 65, 1875.
2006, 119, 971.
56 S. Cafaggi, E. Russo, R. Stefani, R. Leardi, G. Caviglioli,
B. Parodi, G. Bignardi, D. De Totero, C. Aiello and M. Viale,
J. Controlled Release, 2007, 121, 110.
42 M. Nomoto, K. Yamada, M. Haga and M. Hayashi, 57 M. Viale, C. Cordazzo, D. de Totero, R. Budriesi, C. Rosano,
J. Pharm. Sci., 1998, 87, 326.
A. Leoni, P. Ioan, C. Aiello, M. Croce, A. Andreani,
M. Rambaldi, P. Russo, A. Chiarini and D. Spinelli, Invest.
New Drugs, 2011, 29, 98.
43 J. Owens, Nat. Rev. Drug Discovery, 2007, 6, 187.
44 P. Daisy and S. Suveena, Asian J. Pharm. Clin. Res., 2012, 5,
149.
45 A. E. Martell and R. M. Smith, Critical Stability Constants,
Plenum, New York, 1977.
46 J. J. Fuster, S. M. Sanz-González, U. M. Moll and V. Andrés,
Trends Mol. Med., 2007, 13, 192.
47 A. Gadducci, S. Cosio, S. Muraca and A. R. Genazzani,
Eur. J. Gynaecol. Oncol., 2002, 23, 390.
58 I. V. Tetko, J. Gasteiger, R. Todeschini, A. Mauri,
D. Livingstone, P. Ertl, V. A. Palyulin, E. V. Radchenko,
N. S. Zefirov, A. S. Makarenko, V. Y. Tanchuk and
V. V. Prokopenko, J. Comput. Aided Mol. Des., 2005, 19, 453.
59 G. Cruciani, P. Crivori, P. Carrupt and B. Testa, J. Mol.
Struct. (THEOCHEM), 2000, 503, 17.
60 http://www.molecular-networks.com/node/1
2034 | Dalton Trans., 2013, 42, 2023–2034
This journal is © The Royal Society of Chemistry 2013