In vitro and in vivo anticancer activity of tridentate thiosemicarbazone copper complexes: Unravelling an unexplored pharmacological target

Article abstract of DOI:10.1016/j.ejmech.2020.112266

Certain metal complexes can have a great antitumor activity, as the use of cisplatin in therapy has been demonstrating for the past fifty years. Copper complexes, in particular, have attracted much attention as an example of anticancer compounds based on

Full text of DOI:10.1016/j.ejmech.2020.112266

Journal Pre-proof
In vitro and in vivo anticancer activity of tridentate thiosemicarbazone copper
complexes: Unravelling an unexplored pharmacological target
Mauro Carcelli, Matteo Tegoni, Jennifer Bartoli, Cristina Marzano, Giorgio Pelosi,
Marika Salvalaio, Dominga Rogolino, Valentina Gandin
PII:
S0223-5234(20)30233-6
DOI:
https://doi.org/10.1016/j.ejmech.2020.112266
EJMECH 112266
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 18 December 2019
Revised Date: 16 March 2020
Accepted Date: 20 March 2020
Please cite this article as: M. Carcelli, M. Tegoni, J. Bartoli, C. Marzano, G. Pelosi, M. Salvalaio, D.
Rogolino, V. Gandin, In vitro and in vivo anticancer activity of tridentate thiosemicarbazone copper
complexes: Unravelling an unexplored pharmacological target, European Journal of Medicinal Chemistry
(2020), doi: https://doi.org/10.1016/j.ejmech.2020.112266.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2020 Published by Elsevier Masson SAS.

Graphical Abstract

In vitro and in vivo anticancer activity of tridentate
thiosemicarbazone copper complexes: unravelling an
unexplored pharmacological target
a
a
a
b
a
Mauro Carcelli , Matteo Tegoni , Jennifer Bartoli , Cristina Marzano , Giorgio Pelosi , Marika
b
,a
,b
Salvalaio , Dominga Rogolino* and Valentina Gandin*
a
Dipartimento di Scienze Chimiche, della Vita e della Sostenibilità Ambientale and Consorzio
Interuniversitario di Ricerca in Chimica dei Metalli nei Sistemi Biologici, Università di Parma,
Parco Area delle Scienze 11/A, 43124 Parma, Italy
b
Department of Pharmaceutical and Pharmacological Sciences, University of Padova
via F. Marzolo 5, 35131 Padova - Italy
1

ABSTRACT: Certain metal complexes can have a great antitumor activity, as the use of cisplatin
in therapy has been demonstrating for the past fifty years. Copper complexes, in particular, have
attracted much attention as an example of anticancer compounds based on an endogenous metal. In
this paper we present the synthesis and the activity of a series of copper(II) complexes with
variously substituted salicylaldehyde thiosemicarbazone ligands. The in vitro activity of both
ligands and copper complexes was assessed on a panel of cell lines (HCT-15, LoVo and LoVo
oxaliplatin resistant colon carcinoma, A375 melanoma, BxPC3 and PSN1 pancreatic
adenocarcinoma; BCPAP thyroid carcinoma, 2008 ovarian carcinoma, HEK293 non-transformed
embryonic kidney), highlighting remarkable activity of the metal complexes, in some cases in the
low nanomolar range. The copper(II) complexes 1-6 were also screened, with good results, against
3D spheroids of colon (HCT-15) and pancreatic (PSN1) cancer cells. Detailed investigations on the
mechanism of action of the copper(II) complexes are also reported: they are able to potently inhibit
Protein Disulfide Isomerase, a copper-binding protein, that is recently emerging as a new
therapeutic target for cancer treatment. Good preliminary results obtained in C57BL mice indicate
that this series of metal-based compounds could be a very promising weapon in the fight against
cancer.
Keywords: thiosemicarbazones; copper complexes; anticancer compounds; Protein Disulfide
Isomerase inhibitors; nanomolar activity; X-ray structure
2

1
. INTRODUCTION
Bioinorganic chemistry can offer an innovative approach to many issues in the biomedical arena,
allowing to exploit the characteristics of metal ions (redox activity, interaction with cellular
proteins, alteration of homeostatic equilibria) in synergy with organic ligands [1,2]. The cisplatin
milestone is an epitome in this field. It is in fact one of the most used anticancer agents in several
therapeutic regimens, in combination with other drugs, including topoisomerase II inhibitors (like
doxorubicin or bleomycin), antimetabolites (e.g. gentamicin, 5-fluorouracil, methotrexate), and
taxol [3,4]. However, its strong activity is accompanied by a poor selectivity, with consequent
important side effects such as neurotoxicity, nephrotoxicity and ototoxicity, that limit its efficacy. In
addition, the use of cisplatin can be undermined by innate or acquired drug resistance[5,6].
Therefore, extensive research is on-going in order to develop other metal-based anti-tumor
compounds with improved pharmacological profiles. Researchers have been extensively
investigating the possibility of using endogenous metals, since they could be less toxic and more
selective than platinum[7].
In this scenario, copper has attracted considerable interest, also due to the different response of
tumor cells to the presence of this metal when compared to healthy ones. It has been verified that in
2
+
cancerous tissues the concentration of Cu is much higher than that found in healthy tissues
8,9,10]. This fact has been related to the crucial role of copper in the angiogenesis processes and,
[
consequently, in tumor growth and metastasis formation [11]. The “tumor-specific” high copper
level could represent a key target to develop novel selective anticancer drugs. Two different
approaches [12] have been pursued so far: the use of chelating compounds able to sequester copper
ions and the development of copper(I/II)-based antitumor drugs. If, schematically, the mode of
action of cisplatin is ascribable to its ability to crosslink the purine bases of DNA, causing DNA
damages, and subsequently inducing apoptosis in cancer cells, the questions about the mechanism
of action of anticancer copper complexes are mainly still unanswered [12,13,14].
3

The role of the ligands in the activity of the metal complexes is obviously crucial, since they can
modulate important aspects such as lipophilicity, solubility, stability versus sequestration by serum
or cellular proteins. A very interesting class of ligands studied so far for their anticancer activity is
constituted by thiosemicarbazones (TSCs) [15,16,17]. The most outstanding representative of this
class of compounds is Triapine (3-AP, Fig. 1), that has already entered a number of clinical trials
[18,19]. Moreover, very recently, other promising TSCs, like di-2-pyridylketone 4-cyclohexyl-4-
methyl-3-thiosemicarbazone (DpC) and (E)-N′-(6,7-dihydroquinolin-8(5H)-ylidene)-4-(pyridin-2-
yl)piperazine-1-carbothiohydrazide (COTI-2) (Fig. 1), have also entered clinical trials [20,21].
NH2
H
N
S
N
N
NH2
3
-AP
N
S
N
N
N
N
N
N
H
N
N
S
HN
N
COTI-2
Fig. 1. Schematic representation of thiosemicarbazones in clinical trial.
DpC
TSCs represent a very diversified class, in which structural variations can have different
mechanisms of action and modulate different pathways, also as a function of their coordinating
properties, since the role of metal chelation seems to be crucial in relation to their anticancer
4

activity [22,23,24]. TSCs can form a great variety of coordination compounds with biologically-
relevant transition metal ions, like iron(II/III), copper (I/II) and zinc(II) [16,25,26]. Interestingly,
metal complexes very often show enhanced cytotoxic profiles and alternative modes of action when
compared with the parent ligands. Many investigations highlighted the ability of TSCs metal
complexes to inhibit enzymatic pathways related to DNA synthesis and polymerization (i.e. to
inhibit ribonucleotide reductase or DNA polymerase) [27,28,29]. TSCs can also chelate intracellular
iron and, by establishing redox cycling, produce oxygen reactive species (ROS) within the
cytoplasm [15,30,31]. The chelation properties of TSCs can be exploited also to target copper(II)
and its homeostasis [12,32]. Copper, as well as iron, can be involved in the Haber-Weiss reaction
and in the production of ROS, the activation of redox cycles and the reduction of GSH, inducing
oxidative stress [12,33]; in addition, other pathways are probably also to be taken into account
[
12,34,35].
With these considerations in mind and looking at encouraging previous results that we obtained
with salicylaldehyde-TSCs [36,37,38], we focused our attention on the 2,3-dihydroxy- and 2-
1
6
hydroxy-3-methoxy-benzaldehyde thiosemicarbazone derivatives HL -HL , with different
substituents at the N4 nitrogen (Fig. 2). Salicylaldehyde thiosemicarbazones behave essentially as
tridentate ligands, but they can also be involved in the formation of polynuclear species [39]. The
1
6
copper(II) complexes 1-6 (Fig. 2) with HL -HL were characterized in solution by means of UV-
visible spectrophotometric titrations, and in the solid state, also by means of X-ray diffraction
1
6
analysis. The antitumor properties of HL -HL and 1-6 were assayed in vitro on a panel of cell lines
HCT-15, LoVo and LoVo oxaliplatin resistant colon carcinoma, A375 melanoma, BxPC3 and
(
PSN1 pancreatic adenocarcinoma, BCPAP thyroid carcinoma, 2008 ovarian carcinoma, HEK293
non-transformed embryonic kidney). The in vivo antitumor activity of 1, the most promising
copper(II) complex, was finally evaluated. Detailed investigations on the mechanism of action of
the copper(II) complexes 1-6 are also reported, unveiling new possible biological targets so far
unexplored.
5

R
R’
OCH3
OH
4
H
1
2
3
Ph
Et
5
6
1
6
Fig. 2. The ligands HL -HL and the corresponding copper(II) complexes 1-6.
2
. RESULTS AND DISCUSSION
1
6
2
.1 Synthesis. The thiosemicarbazone ligands HL -HL (Fig. 2) were synthesized in high yields by
condensation of thiosemicarbazide or 4-N-substituted-3-thiosemicarbazide with 2-hydroxy-3-
methoxybenzaldehyde or 2,3-dihydroxybenzaldehyde. They were satisfactorily characterized by
standard spectroscopic and spectrometric techniques and experimental data are in accordance with
the literature data [36,57]. The ligands are in the E form in DMSO-d solution at room temperature,
6
6

1
as evidenced by the chemical shift values of the HC=N and NH protons in the H-NMR spectrum
1
6
[
57]. HL -HL can give rise to thione-thiol tautomerism, but the absence, in the IR spectrum, of the
–
1
S–H stretching band near 2600 cm and the presence of a N–H stretching band in the range 3130-
–
1
3
170 cm are indicative of the presence of the thione form in the solid state [40,12].
1
6
HL -HL are versatile ligands: they can coordinate to transition metal ions via the O, N and S donor
atoms, acting as bidentate or tridentate. Some examples are reported in the literature of copper(II)
complexes with salicylaldehyde TSCs, with different stoichiometry and coordination geometry
depending on the experimental conditions used [41,39]. The metal complexes 1-6 (Fig. 2) were
obtained by reacting CuCl with the ligands in methanol in basic conditions, in order to force the
2
deprotonation of the ligand. Thiosemicarbazones can undergo intramolecular oxidative cyclisation
in presence of bases, oxidants or redox-active metals, giving rise to 1,2,4-triazole-3-thione
derivatives, 1,3,4-thiadiazoline-2-amine derivatives or to desulfurization processes [42]. Therefore,
to avoid possible oxidative cyclization reactions, synthesis was carried out under nitrogen by using
degassed solvents. In the IR spectra of 1-6, the iminic bond stretching vibrations, assigned to bands
–
1
–1
around 1522-1608 cm in the ligands, underwent up-shifts of 15-60 cm upon complexation,
indicating the involvement of this group in coordination. Elemental analysis suggested the isolation
.
of complexes of general formula [CuLCl] nH O, with a 1:1 metal to ligand ratio, and various grades
2
of hydration. In the ESI mass spectra of 1-6, recorded in methanol in positive ion mode, there are
+
the peaks relative to the species [CuL] , corresponding to the loss of a chloride ion.
Overall, the data suggested the formation of a 1:1 metal chelate for all the complexes 1-6, with the
mono-deprotonated ligand behaving as O,N,S tridentate, and a chloride ion completing the
coordination sphere of the Cu(II) ion (Fig. 2).
1
.
2
.2 X-ray discussion. The crystal structure of [CuL Cl] H O had already been reported [43],but
2
since it was collected at 123K, we have decided to recollect the data at room temperature to see if
there are significant structural differences. The structure is obviously substantially the same, but we
reached a better R value and noticed a few details which are worth to be reported. The geometry
7

around the copper atom (Fig. 3) is in both structures slightly distorted from the ideal square planar
with the bond angles ranging between 86.30(8) and 92.39(7) [86.54(8) and 92.23(6)] (in square
parentheses the values reported in Sen’s paper). The coordination distances remain very similar:
Cu–S 2.263(1) Å [2.262(9) Å, this distance reported in Sen’s description is actually wrong being
2.662 Å], Cu–N 1.952(3) Å [1.951(2) Å], Cu–O 1.919(2) Å [1.918(2) Å], and Cu–Cl 2.259(1) Å
[
2.266(1) Å].
1
.
Fig. 3. ORTEP plot of the crystal structure of [CuL Cl] H O (probability 50%).
2
Noteworthy is that, at room temperature, the C=S bond is shorter (1.697(4) Å) than at lower
temperature, 1.708(3) Å, and that the ligand is more twisted, with an angle between the average
plane of the aromatic ring and that of the TSC fragment of 4.70° (4.58° at low temperature).
3
The crystal structure of {Na[CuL Cl] Cl} is characterized by the presence of a sodium ion that
2
3
plays a key role in the packing by coordinating the oxygen and the chlorine atoms of two [CuL Cl]
units (Fig. 4).
8

3
Fig. 4. ORTEP plot of the crystal structure of {Na[CuL Cl] Cl} (probability 50%).
2
The two units taken singularly present structures very similar to the one reported above for
1
.
[
CuL Cl] H O. Apart from the presence of the ethyl fragment on the TSC terminal amino group, the
2
rest of the molecule presents approximately the same geometrical parameters. The sodium ion is
surrounded by four oxygens belonging to the ligands, two chlorine atoms that are coordinated by
the copper and a third chloride which acts as a counterion to neutralize the sodium positive charge.
1
.
3
Bond distances (Å) and angles (°) for [CuL Cl] H O and {Na[CuL Cl] Cl}are reported in Table S1
2
2
and Table S2, respectively.
.3 Studies in solution. We have studied the complexes formed by Cu(II) ions with the ligands
2
1
4
HL and HL in solution by means of spectrophotometric techniques. The study of the
1
complexation equilibria for the system Cu(II)/HL required the use of a competitive Zn(II)/Cu(II)
titration strategy; for these reasons, the stability of the complexes with Zn(II) was also examined.
All the metal:ligand systems were investigated by means of spectrophotometric titrations of the
ligand solutions with the metal ions in methanol:water 9:1 (v:v) HEPES buffer at pH 7.4. This
buffer has been reported to have a weak and still competing role in metal ions complexation
[44,45,46]. For this reason, the determined formation constants are apparent rather than
9

conditional, i.e. their values reflect the stability of the complexes species under these specific
medium conditions and in the presence of this buffer dye. Perhaps more important, we should also
note that the buffered conditions did not allow to establish the protonation state of the ligands in
solution (either free or coordinated). Therefore, hereafter, in the study of the complexation
equilibria, we will use L to represent the coordinated ligand irrespective of its protonation state,
omitting consequently the charge of the complex.
-1
-1
1
The ligands have strong absorption bands at ca. 300-325 nm (ε ca. 32000 M cm for HL , 19000
-1
-1
4
M cm for HL , see Fig. S7 and S18, Supporting information). As a general behavior, upon
addition of the metal ions, a band at ca. 370-400 nm appeared in all titrations. This band, associated
to the formation of complex species, is due to intra-ligand or, for copper(II), to ligand to metal
charge transitions. At these dilution levels (ca. 40 µM), ligand-field transitions for the Cu(II)
species are not expected to be observable. UV visible spectra, titration curves and Job’s plots for the
studied systems are reported in Fig. S7-S22.
1
We first examined the system Cu(II)/HL . This ligand forms the 1:1 Cu(II):ligand species that has
been characterized in the solid state and is expected to be present also in solution in significant
1
amount. Actually, the spectral dataset for the titration of HL with Cu(II) presents two bands, one at
ca. 320 nm and one at ca. 390 nm which decrease and increase respectively upon addition of the
titrant (Fig. 5).
1
0

1.2
1.0
0.8
0.6
0.4
0.2
0.0
250
300
350
400
450
500
Wavelength (nm)
1
Fig. 5. UV-visible spectra for the titration of HL with Cu(II) in methanol:water 9:1 (v/v) at pH 7.4
1
(
25 mM HEPES buffer. CHL1 = 41 µM, HL :Cu(II) = 1:0-2.1).
The evaluation of the absorbance values at 388 nm (Fig. S8) and the Job’s plot (Fig. S9) strongly
suggests the formation of a predominant Cu(II)/ligand 1:1 species. Unfortunately, the titration
endpoint (Fig. S8) is very pronounced and this situation prevents the determination of the formation
constant.
We decided therefore to adopt a strategy which involves the use of Zn(II) as competing metal ion
1
for the complexation of HL . Through this strategy, it is possible to study equilibria of the form
2
+
2+
[
ZnL] + Cu = [CuL] + Zn (charge of the complexes omitted), and to determine the formation
constants of the Cu(II) species, given the formation constants of the Zn(II) ones. Firstly, we carried
1
out the titration of HL with Zn(II): the spectral dataset (Fig. S10) shows a behavior similar to that
observed with Cu(II). The results of the data treatment suggested the formation of a [ZnL] species,
in agreement with the Job’s plot data (Fig. S12). The formation constant is reported in Table 1.
1
Lastly, we carried out the titration with Cu(II) of a solution containing HL and Zn(II) in 1:2 ratio.
The spectral dataset (Fig. S14) is consistent with the interconversion of one predominant species
([ZnL]) into a second one ([CuL]). This hypothesis is supported by the decrease of a band at ca. 375
nm (Zn(II) complex) with increase at ca. 390 nm (Cu(II) species). We have treated this spectral
1
1

dataset by using the formation constant and the molar spectrum of the [ZnL] species of HL as fixed
1
parameters, along with the molar spectrum of the HL ligand. The best fitting of the experimental
data was indeed obtained with a [CuL] species, leading to the calculation of the formation constant
reported in Table 1.
4
The spectra dataset for the titration of HL with Cu(II) (Fig. 6) was more complicated compared to
1
that of HL , and it suggests the formation of more than one complex species.
1.0
0.8
0.6
0.4
0.2
0.0
250
300
350
400
450
500
Wavelength (nm)
4
Fig. 6. UV-visible spectra for the titration of HL with Cu(II) in methanol:water 9:1 (v/v) at pH 7.4
4
(
25 mM HEPES buffer. CHL4 = 41 µM, HL :Cu(II) = 1:0-2.6).
Actually, for the addition of the first equivalent of Cu(II), the band at 310 nm decreases in intensity,
with concomitant increase of the band at 390 nm. This behavior is similar to that observed for the
1
titration of HL , and was provisionally associated to the formation of a predominant 1:1 [CuL]
4
4
complex also for HL . Upon addition of a second equivalent of Cu(II) to HL , the band at 390 nm
decreases with increase of the shoulder at ca. 355-360 nm, and the consequent appearance of an
isosbestic point at 374 nm. No further significant changes in the spectra occurred after the addition
of 2 equivalents of Cu(II). The formation of two predominant species seems plausible: a
Cu(II):ligand 1:1 species for the addition of the first equivalent of copper(II), and a 2:1
Cu(II):ligand species upon addition of the second equivalent of metal. We then treated the entire
1
2

dataset, obtaining the best fitting of the experimental data by considering three species: two
mononuclear ([CuL] and [CuL ]) species, and a bimetallic ([Cu L]) one (Figure S16); their
2
2
formation constants are reported in Table 1. Remarkably, the bimetallic species has been found for
4
1
4
HL but not for HL : the 2,3-dihydroxyphenyl group in HL may act as another binding site for
Cu(II), with consequent formation of a bimetallic complex in excess of metal. For the sake of
4
completeness, we have also studied the Zn(II)/HL system (Fig. S19). In the treatment of these
4
spectral data we have considered for HL the species [ZnL] and [ZnL ], as suggested by the Job’s
2
plot at 376 nm, which has its maximum for χ > 0.5 (Fig. S21). The stability constants of these
Zn(II) species are reported in Table 1.
Table 1. Logarithms of the apparent formation constants of the Cu(II) and Zn(II) complexes with
1
4
HL and HL in HEPES buffer (25 mM, pH 7.4) in methanol:water 9:1 (v/v), at 298.2 K. Charges
are omitted for clarity. Standard deviations are reported in parentheses.
1
4
HL
HL
[
Cu(L)]
7.90(3)
7.32(18)
13.0(2)
[
[
Cu(L)2]
-
-
Cu (L)]
12.80(2)
2
[
Zn(L)]
5.97(2)
-
6.91(11)
12.0(2)
[
Zn(L)2]
2.4 Effects on cell viability. The copper(II) thiosemicarbazone complexes 1-6 as well as the
1
6
corresponding ligands HL -HL were tested for their cytotoxic activity by means of the MTT assay,
as reported in the Experimental section. The in-house human cancer cell line panel contains
examples of ovarian (2008), colon (HCT-15), pancreatic (PSN-1 and BxPC3), and thyroid
(BCPAP) cancers as well as of melanoma (A375). Cisplatin was used as reference compound and
1
3

was tested under the same experimental conditions. The cytotoxicity parameters, in terms of IC50
obtained after 72 h of exposure, are listed in Table 2.
3
−1
Table 2. Cytotoxicity assays. Cells (3-8×10 ·mL ) were treated for 72 h with increasing
concentrations of the tested compounds. Cytotoxicity was assessed by MTT test. IC values were
5
0
calculated by 4-PL model (P < 0.05). S.D.= standard deviation. ND = non detectable. CDDP =
cisplatin.
IC50 (µM) ± S.D.
2
008
HCT-15
PSN-1
A375
BxPC3
BCPAP
1
2
3
4
5
6
HL
0.78±0.21
0.013±0.008
4.85±1.12
0.035±0.011
1.97±0.66
0.010±0.001
ND
0.51±0.14
0.004±0.001
3.97±0.38
0.017±0.006
2.22±0.76
0.036±0.012
ND
0.51±0.12
0.008±0.002
6.82±1.08
0.32±0.10
3.13±1.11
0.021±0.090
ND
0.10±0.030
0.003±0.0004
2.11±0.4
0.005±0.001
0.03±0.01
0.98±0.25
0.002±0.0005
1.11±0.76
0.015±0.006
ND
0.51±0.09
0.033±0.01
3.15±0.62
0.84±0.16
0.51±0.090
0.041±0.011
ND
1
HL
2
0.009±0.001
2.23±0.78
0.019±0.004
ND
HL
3
HL
4
0.098±0.004
ND
0.091±0.06
ND
0.36±0.010
ND
0.029±0.006
ND
0.011±0.004
ND
0.35±0.09
ND
HL
5
0.71±0.15
ND
0.82±0.14
ND
0.51±0.12
ND
0.28±0.09
ND
0.11±0.09
ND
0.83±0.11
ND
HL
6
0.16±0.04
2.17 ± 1.37
0.23±0.09
13.92±1.68
0.029±0.008
12.10±2.87
0.069±0.020
3.11±0.98
0.029±0.011
13.98±1.23
0.072±0.012
6.65±2.85
CDDP
1
3
4
6
TSCs HL -HL showed IC values in the low-micromolar range, whereas for HL -HL , maybe
5
0
due to their low solubility, non-reproducible and non-reliable MTT results were obtained, thus
precluding the calculation of IC values. This behavior could be tentatively attributed to the scarce
5
0
1
4

4
6
solubility of HL -HL in physiological conditions. On the other hand, copper(II) 1-6 compounds
showed a significant in vitro antitumor activity, with IC values even in the nanomolar range.
5
0
As a general consideration, copper(II) complexes 1-3, with the 2-hydroxy-3-methoxyphenyl group,
were, on average, more cytotoxic than 4-6, the corresponding copper(II) complexes with the 2,3-
dihydroxyphenyl group. As an example, the 2-hydroxy-3-methoxybenzaldehyde thiosemicarbazone
Cu(II) complex 3 elicited average IC values over the six employed cancer cell lines of roughly
5
0
0.024 µM, compared with average IC50 of 0.098 µM obtained with the corresponding 2,3-
dihydroxybenzaldehyde thiosemicarbazone complex 6.
More importantly, all the tested copper(II) complexes 1-6 were significantly more cytotoxic than
the reference compound cisplatin, being their IC values up to three order of magnitude lower than
5
0
those calculated with the reference metal-based drug. Among all, 1 emerged as the most cytotoxic
derivative with mean IC values against all tested cancer cell lines about 580 times lower than
5
0
those calculated for cisplatin (average IC values of 0.015 and 8.7 µM, respectively). In particular,
5
0
against human pancreatic (BxPC3 and PSN1) and colon (HCT-15) cancer cells, the copper(II)
complex 1 was about 466, 1510 and 3480 times more effective than cisplatin, respectively.
Compound 5 was, on the contrary, the one endowed with the weakest in vitro antiproliferative
effect, with average IC values over the six cancer cell lines of 0.54 µM. However, its cytotoxic
5
0
activity was significantly superior (16 times) than cisplatin.
The in vitro antitumor activity of the complexes 1-6 was evaluated on a human colon cancer cell
line pair (LoVo/LoVo-OXP) which was selected for sensitivity/resistance to oxaliplatin, the key
drug in FOLFOX (folinic acid, 5-fluorouracil, and oxaliplatin) chemotherapeutic regimen for the
management of colorectal cancers [47]. Similarly to cisplatin, the clinical efficacy of oxaliplatin is
seriously limited by the development of cancer cell resistance and, presently, no other drugs in
advanced clinical development for the treatment of patients with oxaliplatin-refractory colorectal
cancer are available. Cytotoxicity on sensitive and resistant cells was assessed after 72 h of drug
treatment by MTT test. Table 3 shows the cytotoxicity parameters, in terms of IC and resistance
5
0
1
5

factor (RF), the latter defined as the ratio between IC values calculated for resistant cells and those
5
0
obtained with sensitive ones.
3
−1
Table 3. Cross-resistance profiles. Cells (3×10 ·mL ) were treated for 72 h with increasing
concentrations of the tested compounds. Cytotoxicity was assessed by MTT test. IC values were
5
0
calculated by 4-PL model (P < 0.05). RF = IC resistant/IC parental cell lines. S.D.= standard
5
0
50
deviation.
IC (µM) ± S.D.
5
0
LoVo
LoVo-OXP
RF
1
0.031±0.001
0.029±0.008
0.036±0.009
0.020±0.001
0.21±0.08
0.004±0.001
0.030±0.010
0.008±0.002
0.020±0.001
0.09±0.01
0.13
1.03
0.22
1
2
3
4
5
0.43
0.67
6.41
6
0.030±0.001
2.17 ± 1.37
0.02±0.01
oxaliplatin
13.92±1.68
All copper(II) complexes exhibited activity levels very similar on both oxaliplatin-sensitive (LoVo)
and -resistant (LoVo-OXP) cell lines. Actually, the RFs were from 6- to 50-fold lower than that of
oxaliplatin, indicating the absence of cross-resistance phenomena.
As one of the main drawbacks of chemotherapeutic drugs are the possible toxic effects toward non-
cancerous cells, we measured the cytotoxicity of 1-6 against non-cancerous cells (HEK293) and
calculated the selectivity index (SI, defined as the ratio of the IC s in non-cancerous cells to those
5
0
in cancer cells). The cytotoxicity data for HEK293 cells and the selectivity indices are reported in
Table 4. It is worthy of note that the copper(II) complexes 2 and 3 were less cytotoxic against
HEK293 non-cancerous cells, eliciting SI values even better than that calculated with cisplatin, thus
1
6

indicating a preferential cytotoxicity against tumor cells. On the contrary, for complexes 1 and 4-6
the SI values clearly suggest the absence of a preferential activity against cancer cells.
3
-1
Table 4. Cytotoxicity in HEK293 human non-cancerous cells. Cells (5×10 mL ) were treated for
72 h with increasing concentrations of tested compounds. The cytotoxicity was assessed by the
MTT test. IC50 values were calculated by a four parameter logistic model (P < 0.05). S.D. =
standard deviation. Selectivity index (SI) is defined as IC non-tumor/tumor cell lines (LoVo).
5
0
CDDP = cisplatin.
IC50 (µM) ± S.D.
HEK293
SI
1
2
3
4
5
6
0.009±0.002
0.093±0.011
0.11±0.080
0.007±0.001
0.021±0.010
0.034±0.001
24.38±3.45
0.3
3.2
3.05
0.3
0.1
1.1
2.1
CDDP
The copper(II) compounds 1-6 were also screened against 3D spheroids of colon (HCT-15) and
pancreatic (PSN1) cancer cells. 3D cell cultures, comprising cancer cells in various cell growth
stages, possess several features that more closely mimic the heterogeneity and complexity of in vivo
tumors, being potentially more predictive for in vivo results than conventional 2D cell cultures [48].
The cancer spheroids were treated with copper(II) complexes or cisplatin for 72 h and the cell
viability was assessed by means of the acid phosphatase (APH) assay (Table 5). 1-6 were extremely
effective, being roughly up to 60 times more active than cisplatin against HCT-15 and PSN1 cells.
These latter data, attesting the ability of TSCc copper(II) complexes to penetrate in the core area of
spheroids, strongly confirm the antitumor potential of these copper(II) complexes.
1
7

3
Table 5. Activity in 3D cell cultures. Spheroids (2.5 × 10 cells/well) were treated for 72 h with
increasing concentrations of the tested compounds. The growth inhibitory effect was evaluated by
means of APH test. IC values were calculated from the dose-survival curves by 4-PL model (P <
5
0
0.05). S.D. = standard deviation. CDDP = cisplatin.
IC (µM)± S.D.
50
HCT-15
PSN-1
1
2
3
4
5
6
1.08±0.38
3.56±1.67
1.25±0.98
1.17±0.62
1.69±0.45
1.28±0.62
68.2±4.57
0.90±0.02
1.17±0.11
0.90±0.3
0.94±0.27
1.18±0.23
0.91±0.01
52.6±3.78
CDDP
2.5 Cellular uptake and distribution. In the attempt to correlate cytotoxicity potency with the
ability of the tested complexes to enter cancer cells and to distribute into cell compartments, we
performed cellular uptake and distribution studies in LoVo human colon cancer cells (Fig. 7A and
6
B). The results expressed as ppb of Cu per 10 cells are summarized in Fig. 7.
1
8

Fig. 7. Cellular uptake (A, B) and distribution studies (C) in cancer cells. (A) LoVo cancer cells
were incubated with 1 µM of copper complexes for 24 or 36 h, and cellular copper content was
detected by GF-AAS analysis. (B) LoVo cancer cells were incubated with 0.1, 0.5 or 1 µM of
copper complexes for 24 h, and cellular copper content was detected by GF-AAS analysis. (C)
LoVo cancer cells were incubated with 1 µM of copper complexes for 24 h and samples were
processed for sub-fractions preparation as reported in the Experimental Section. The sub-fraction
copper content was detected by GF-AAS analysis. Error bars indicate the standard deviation.
1
9

Copper(II) complexes 1-3 derived from 2-hydroxy-3-methoxybenzaldehyde thiosemicarbazones
were able to accumulate in a time- and dose-dependent manner, and significantly greater with
respect to the corresponding copper(II) complexes with 2,3-dihydroxybenzaldehyde
thiosemicarbazones 4-6, which poorly permeated cancer cells, independently of exposure time (Fig.
7
, A and B). The 2-hydroxy-3-methoxybenzaldehyde derivative 1 was the most effective in entering
cancer cells; after 24 or 36 h, in 1-treated an intracellular copper content roughly 10 times higher
than in 4-6-treated cells was detected. However, by comparing uptake and cytotoxicity data in
LoVo human colon cancer cells, a direct correlation between cellular accumulation and cytotoxic
potency does not result.
Concerning cellular content, nuclear and mitochondrial levels of copper(II) complexes in LoVo
cancer cells, it is evident that 2-hydroxy-3-methoxyphenyl compounds 1-3 accumulated
substantially in the mitochondria fraction and, to a lesser extent, in cytosolic fractions (Fig. 7 C).
Conversely, very low levels of copper were found in the nuclei of treated cancer cells. A similar
cellular distribution was found for 2,3-dihydroxyphenyl copper(II) compounds 4-6, but detected
copper amounts were about one order of magnitude lower.
2.6 Mechanistic studies. Since the ‘60s copper has been found to possess high DNA binding
affinity, depending on copper complex size, electron affinity, and geometry of the formed adduct.
DNA has been referenced as the main molecular target for various copper(II) complexes in many
studies [12]. On this basis, we thought of interest to evaluate the ability of copper(II) complexes 1-6
to damage DNA. LoVo cancer cells were treated with 1-6 for 6 and 12 h, and DNA fragmentation
was estimated by using the alkaline single cell gel electrophoresis (comet assay). The results were
compared with those obtained after treatment of LoVo cells with equitoxic concentrations of
dichloro(1,10-phenanthroline)copper(II) (Cu(phen)) a well-known copper complex with nuclease
activity (Fig. 8 A and B). LoVo cancer cells treated with 1-6 showed no increase in electrophoretic
migration of DNA fragments, and comet tail lengths of analyzed single cells were below 1 µm. On
2
0

the contrary, Cu(phen)-treated LoVo cells displayed a statistically significant increase in DNA
fragmentation.
Fig. 8. Mechanistic studies: DNA damage (A and B) and proteasome inhibition (C and D). (A)
LoVo cells were treated for different incubation times (6 or 12 h) with 1 µM of the tested
complexes and then processed for comet assay as reported in the Experimental Section. Comet tail
length was calculated from the center of the cell and measured in micrometers with CellF software.
The error bars indicate the standard deviation. (B) Representative images of LoVo cells after 12 h
incubation with compound 1 or Cu(phen). (C) Inhibition of CT-L, T-L and C-L activities of purified
rabbit 26S proteasome after 60 min of incubation with increasing concentrations of compounds 1-6
or Lactacystin. Proteasome catalytic activity was estimated by means of specific fluorogenic
substrates. IC50 values were calculated by 4-PL model (P < 0.05). The error bars indicate the
standard deviation. (D) LoVo cells were treated for 24 h with increasing concentrations of
2
1

complexes 1 and 4. Proteasome catalytic activities were estimated fluorometrically as reported in C.
The error bars indicate the standard deviation.
It has been previously demonstrated that some metal-based compounds exert their antiproliferative
activity by inhibiting the proteolytic proteasome activities. In particular, various classes of copper
complexes were reported to behave as proteasome inhibitors [12, 49]. On these bases, the ability of
1-6 to hamper the functioning of chymotriptic-like (CT-L) activity was first assessed in vitro, on
purified rabbit 26S proteasome after incubation with increasing concentrations of the tested
compounds or Lactacystin, an irreversible nonpeptidomimetic proteasome inhibitor used as a
positive control. In addition, to further evaluate the anti-proteasome activity of 1-6, LoVo cells were
treated with increasing concentrations of 1 and 4 for 24 h, and then the functioning of each
individual proteasome active site, chymotriptic-like (CT-L), trypsin-like (T-L), and caspase-like (C-
L) activities were assessed. All copper(II) complexes were barely effective in reducing the CT-L
activity of 26S proteasome in vitro, eliciting IC values up to 25 times higher than those calculated
5
0
for the reference proteasome inhibitor Lactacystin (Fig. 8 C). In intact cancer cells, on the contrary,
copper complexes proved to be very effective, being able to inhibit 50% of all the three catalytic
activities even at low micromolar concentrations. IC values calculated for CT-L activity were very
5
0
similar to those of Lactacystin, whereas copper(II) complexes 1 and 4 were significantly more
effective than the reference compound in inhibiting T-L and C-L proteasome activities (Fig. 8 D).
Copper complexes have been regarded as redox active compounds and redox modulators [50]
Actually, copper complexes may catalyze hydrogen peroxide in the form of Fenton-like reactions
inside the cell to produce ROS, thus altering cellular redox homeostasis and driving cells towards
oxidative stress [51]. On these bases, and tacking into consideration biodistribution studies
highlighting the ability of our Cu(II) thiosemicarbazone complexes to accumulate into
mitochondria, key organelles involved in cellular redox balance, we also evaluated the ability of 1-6
2
2

to alter cellular redox homeostasis. The effects induced by the tested complexes on total cellular
sulfhydryl content, total glutathione (GSH + GSSG) and oxidized glutathione (GSSG) levels were
assayed in LoVo human colon cancer cells (Fig. 9 A and B).
Fig. 9. Mechanistic studies: cellular redox state (A and B), TEM analysis (C) and PDI inhibition
(
D). (A) Sulfhydryl content in LoVo cancer cells treated for 24 h with IC concentrations of the
50
tested complexes 1-6. The sulfhydryl group amount was determined by the DTNB assay. (B) Total
and oxidized (insert) glutathione levels in LoVo cancer cells treated with compounds 1-6 at IC50
concentrations. Cells were collected and processed as indicated in the Experimental Section. (C)
TEM analysis of LoVo cells: a) and b) control cells; c), d) and e) LoVo cells treated for 24 h with
IC concentrations of 1. (D) PDI inhibition induced by compounds 1-6 was measured by Proteostat
5
0
PDI assay kit. The PDI inhibitor Bacitracin (0.5 mM) was used as a positive control.
2
3

All the tested complexes 1-6 were ineffective in modulating cellular sulfhydryl content and total
glutathione content. Interestingly, LoVo cancer cells treated with compounds 1-6 showed higher
levels of oxidized glutathione compared to control cells, and this effect was much more pronounced
with complexes 4-6 (Fig. 9 B, insert). Accordingly, ROS evaluation confirmed that TSC complexes
did not provoke a substantial increase of cellular ROS levels (Fig. S23).
In order to characterize the cellular morphological changes induced by the copper(II)
thiosemicarbazone compounds, we observed LoVo cancer cells treated for 24 h with IC50
concentrations of the most representative compound 1 by using transmission electron microscopy
(TEM). Morphological analysis revealed that 1 induced an intense swelling of the mitochondria
associated with decreased electron density of the inner membrane and matrix regions (Fig. 9 C,
panels c, d, and e). In addition, treatment with 1 induced a massive swelling of the ER membrane,
which is a clear sign of ER stress.
It is well known that one of the most important protein which acts as an effective crosstalk between
ER and mitochondria is Protein Disulfide Isomerase (PDI). PDI has several functions in the ER as
catalyst of redox transfer, disulfide isomerization and oxidative protein folding, and as molecular
chaperone. Although the impact of binding on the enzymatic activity is not fully characterized,
there are reports that PDI is able to bind and reduce copper [52]. Interestingly, PDI expression
contributes to copper resistance in various organisms [53]. On these basis, we hypothesized that
copper complexes 1-6 interfere with PDI activity, possibly inhibiting its disulfide bond catalytic
activity. By monitoring PDI activity in treated LoVo cancer cells, all the 1-6 complexes were found
able to potently inhibit the enzyme, being significantly more effective than bacitracin, a well-known
inhibitor of PDI activity (Fig. 9 D). IC values calculated for derivatives 1-6 were in the low
5
0
micromolar range, up to three orders of magnitude lower than those recorded with bacitracin.
.7 In vivo preliminary studies. Owing to the high potential demonstrated by 1-6, we decided to
2
move a step forward towards preclinical translational studies by performing preliminary in vivo
anticancer investigations. For these studies, we selected the most promising derivative, the 2-
2
4

hydroxy-3-methoxybenzaldehyde thiosemicarbazone Cu(II) complex 1, which showed a very high
in vitro antitumor activity and a preferential cytotoxicity toward cancer cells. The in vivo antitumor
activity of 1 was evaluated in a model of solid tumor, the highly aggressive syngeneic murine Lewis
Lung Carcinoma (LLC). Tumor growth inhibition induced by 1 was compared with that promoted
by cisplatin. Seven days after tumor inoculation, tumor-bearing mice were randomized into vehicle
control and treatment groups. Control mice received the vehicle solution (20% Cremophor EL (v/v),
-
1
2
0% PEG400 (v/v) and 60% saline solution (v/v)) or cisplatin (1.5 mg kg in saline solution). The
tumor growth was estimated at day 15 and the results are reported in Table 6. As an estimation of
the adverse side effects, changes in the body weights of tumor-bearing mice were monitored every
two days starting from day 7 (Fig. S24).
Table 6. In vivo antitumor activity. Lewis lung carcinoma (LLC) was implanted i.m. in C57BL
mice. After 24 h from tumor implantation, mice were randomly divided into groups of 8 animals
(10 controls). Chemotherapy was delayed until the tumor was visible (day 7). Day 15: animals were
sacrificed and tumor growth inhibition was determined as the difference in weight of the tumor-
a
bearing leg and the healthy leg (%), referred to control animals. vehicle (20% Cremophor EL (v/v),
20% PEG400 (v/v) and 60% saline solution (v/v)). CDDP = cisplatin.
-
1
Daily dose (mg·kg )
Average tumor weight
Inhibition of tumor growth
(
mean±S.D., g)
0.459±0.13
0.239±0.08
0.118±0.09
0.114±0.08
(%)
-
48.0
74.3
75.2
a
control
1
1
-
3
6
CDDP
1.5
−
1
Noteworthy, administration of 1 at 3 mg kg reduced average tumor growth by half compared to
−
1
control group whereas at 6 mg kg it induced about 75% reduction of the tumor mass, similarly to
2
5

−
1
cisplatin dosed at 1.5 mg kg . However, the time course of body weight changes attested that
cisplatin induced elevated anorexia (Figure S24). Conversely, treatment with 1 did not induce a
substantial body weight loss (<10%) throughout the therapeutic experimentation.
3
CONCLUSIONS
In the last years, the “old story” of TSCs as anticancer compounds seems to find new perspectives
15]. TSCs and, in particular, their metal complexes could have cellular targets different from
[
cisplatin, opening the expectation to widen the spectrum of action and obtain better selectivity.
Until now, TSCs with the 2-pyridil group (Fig. 1) have attracted more attention, but here we would
like to suggest that also salicylaldehyde derivatives can offer interesting opportunities. In fact, if the
1
6
activity of HL -HL is not significantly different from that of other similar TSCs [54], the activity
of the corresponding copper(II) complexes 1-6 is remarkable and in some cases in the low
nanomolar range. The most effective copper(II) complex 1 was about 466, 1510 and 3480 times
more effective than cisplatin against human pancreatic (BxPC3 and PSN1) and colon (HCT-15)
cancer cells, and 60 times more active than cisplatin in 3D spheroids of HCT-15 and PSN1 cancer
cells. The effect in the resistant spheroid model makes complex 1 a promising drug candidate for
further in vivo evaluation, since three-dimensional cell cultures are more similar to real tumors in
terms of tissue morphology, biology and gene expression. Worth of note is also that subtle
variations in the structure of the ligands can lead to significant differences in activity. In particular,
the 3-methoxy 2-hydroxyphenyl thiosemicarbazones form copper(II) complexes (1-3) on average
more active than the 2,3-dihydroxyphenyl derivatives. Speciation studies confirm that the two sets
of ligands have different behavior in solution. The biochemical data here reported demonstrate that
1-6 are preferentially localized in mitochondria, and that they possess a cellular target that until now
has received little attention [55]: Protein Disulfide Isomerase, a copper-binding protein, that is
emerging as a new therapeutic target for cancer treatment [56]. The anticancer activity of iron and
copper complexes of TSCs has been linked to oxidative stress induced by ROS production [30,57]
2
6

or to the inhibition of the enzyme TopoIIα [58]. In a different way, the copper complexes 1-6 do
not induce significant oxidative stress, but they are able to potently inhibit PDI (Fig. 9 D);
remarkably, their IC50 values were up to three orders of magnitude lower compared to that of
Bacitracin, considered as standard.
Even if PDI is primarily an ER protein, PDI expression has been documented on the cell surface as
well as in the mitochondria, nucleus, and cytosol [59]. In addition, it was recently proposed that PDI
can translocate from ER to the nucleus or mitochondria in particular conditions [60]. On these
bases, we can speculate that our compounds inhibit mitochondrial resident PDI, thus inducing a
non-oxidative stress mediated cancer cell death. However, further studies will be performed to
better clarify the precise mechanism of action of these very promising TSC complexes, hopefully
by physico-chemical characterization of their mode of interaction with PDI.
Finally, the good results obtained in in vivo tests deserve a particular mention. A lot of work has to
be done, in particular to improve the solubility of these complexes, but preliminary data are for sure
encouraging and clearly indicate that this class of metal-based compounds could be a very
promising weapon in the fight against cancer.
4
4
EXPERIMENTAL
.1 Materials and methods. Chemistry. Commercial reagents were purchased from Sigma-
Aldrich. The purity of the synthesized compounds was determined by elemental analysis and
1
verified to be ≥ 95%. H-NMR spectra were recorded at 25 °C on a Bruker Avance 400 FT
spectrophotometer. The ATR-IR spectra were recorded by means of a Nicolet-Nexus (Thermo
-1
Fisher) spectrophotometer by using a diamond crystal plate in the range of 4000-400 cm .
Elemental analyses were performed by using a FlashEA 1112 series CHNS/O analyzer (Thermo
Fisher) with gas-chromatographic separation. Electrospray mass spectral analyses (ESI-MS) were
performed with an electrospray ionization (ESI) time-of-flight Micromass 4LCZ spectrometer.
2
7

Samples were dissolved in methanol. MS spectra were acquired with a DSQII Thermo Fisher
apparatus, equipped with a single quadrupole analyzer in positive EI mode, by means of a DEP-
probe (Direct Exposure Probe) equipped with a Re-filament. The UV-vis spectra were collected
using a Thermo Evolution 260 Bio spectrophotometer provided with a thermostatting Peltier device,
and quartz cuvettes with 1 cm path length.
1
6
4
.2 General synthesis of HL -HL . The thiosemicarbazone ligands were synthesized by
condensation of the proper thiosemicarbazide with 2-hydroxy-3-methoxybenzaldehyde or 2,3-
1
4
dihydroxybenzaldehyde according to procedures already reported for HL and HL [61]. Briefly, to
a solution of the aldehyde in absolute ethanol, an equimolar amount of thiosemicarbazide dissolved
in the same solvent was added. Few drops of glacial acetic acid were added and the reaction mixture
was heated under reflux for 4 h. A precipitate was obtained, which was subsequently filtered,
washed with ethanol, and dried under vacuum.
1
4
The characterization of HL and HL is reported in ref. 57.
2
2
-hydroxy-3-methoxybenzaldehyde-4-phenyl-3-thiosemicarbazone (HL ). White powder. Yield =
1
8
3%. H-NMR (DMSO-d , 25°C), δ: 3.82 (s, 3H, OCH ); 6.79 (t, 1H, ArH); 6.98 (d, 1H, ArH,
6
3
J =7.4); 7.20 (t, 1H, ArH, J =7.3); 7.36 (t, 2H, ArH, J =7.8); 7.56 (d, 2H, ArH, J =7.7); 7.69
HH
HH
HH
HH
(
d, 1H, ArH, J =7.6); 8.51 (s, 1H, CH=N); 9.26 (s, 1H, OH); 10.03 (s,1H, NH); 11.80 (s, 1H, NH).
HH
+
-1
MS (EI, 70 eV, positive ions) m/z (%) = 301.0 ([M] , 100). IR (cm ): ν = 3299; νC=N = 1594,
NH
1
5
2
8
580; νC=S= 1066, 781. Anal. Calcd. for C H N O S: C 59.78, H 5.02, N 13.94, S 10.64. Found: C
15 15 3 2
9.36, H 4.82, N 13.74, S 11.03.
3
-hydroxy-3-methoxybenzaldehyde-4-ethyl-3-thiosemicarbazone (HL ). White powder. Yield =
1
5%. H-NMR (DMSO-d , 25°C), δ: 1.13 (t, 3H, CH CH ); 3.58 (q, 2H, CH CH ); 3.81 (s, 3H,
6
2
3
2
3
OCH ); 6.78 (t, 1H, ArH, J =7.9); 6.96 (d, 1H, ArH, J =6.6); 7.54 (d, 1H, ArH, J =7.3); 8.39 (s,
3
HH
HH
HH
1H, CH=N); 8.44 (s, br, 1H, NH); 9.20 (s, br, 1H, OH); 11.38 (s, 1H, NH). MS (EI, 70 eV, positive
+
-1
ions) m/z (%) = 253.1 ([M] , 100). IR (cm ): ν = 3305; νC=N = 1522; νC=S= 1063, 784. Anal.
NH
2
8