Copper binding agents acting as copper ionophores lead to caspase inhibition and paraptotic cell death in human cancer cells

Article abstract of DOI:10.1021/ja109413c

We report a quantitative structure-activity relationship study of a new class of pyrazole-pyridine copper complexes that establishes a clear correlation between the ability to promote copper accumulation and cytotoxicity. Intracellular metal accumulation is maximized when ligand lipophilicity allows the complex to rapidly cross the membrane. Copper and ligand follow different uptake kinetics and reach different intracellular equilibrium concentrations. These results support a model in which the ligand acts as an ionophore for the metal ion, cycling between intra- and extracellular compartments as dissociated or complexed entities. When treating cancer cells with structurally unrelated disulfiram and pyrazole-pyridine copper complexes, as well as with inorganic copper, the same morphological and molecular changes were reproduced, indicating that copper overload is responsible for the cytotoxic effects. Copper-based treatments drive sensitive cancer cells toward paraptotic cell death, a process hallmarked by endoplasmic reticulum stress and massive vacuolization in the absence of apoptotic features. A lack of caspase activation, as observed in copper-treated dying cells, is a consequence of metal-mediated inhibition of caspase-3. Thus, copper acts simultaneously as an endoplasmic reticulum (ER) stress inducer and a caspase-3 inhibitor, forcing the cell into caspase-independent paraptotic death. The establishment of a mechanism of action common to different copper binding agents provides a rationale for the exploitation of copper toxicity as an anticancer tool.

Full text of DOI:10.1021/ja109413c

ARTICLE
pubs.acs.org/JACS
Copper Binding Agents Acting as Copper Ionophores Lead to Caspase
Inhibition and Paraptotic Cell Death in Human Cancer Cells
Saverio Tardito,† Irene Bassanetti,‡ Chiara Bignardi,‡ Lisa Elviri,‡ Matteo Tegoni,‡ Claudio Mucchino,‡
Ovidio Bussolati,† Renata Franchi-Gazzola,† and Luciano Marchiꢀo*,‡
†Dipartimento di Medicina Sperimentale, Sezione di Patologia Generale e Clinica, Via Volturno 39, 43100 Parma, Italy
‡Dipartimento di Chimica Generale ed Inorganica, Chimica Analitica, Chimica Fisica, Universitꢀa degli Studi di Parma, Viale G.P. Usberti
17/A, 43100 Parma, Italy
S Supporting Information
b
ABSTRACT: We report a quantitative structureꢀactivity re-
lationship study of a new class of pyrazole-pyridine copper
complexes that establishes a clear correlation between the
ability to promote copper accumulation and cytotoxicity. Intra-
cellular metal accumulation is maximized when ligand lipophi-
licity allows the complex to rapidly cross the membrane. Copper
and ligand follow different uptake kinetics and reach different
intracellular equilibrium concentrations. These results support a
model in which the ligand acts as an ionophore for the metal ion,
cycling between intra- and extracellular compartments as dis-
sociated or complexed entities. When treating cancer cells with
structurally unrelated disulfiram and pyrazole-pyridine copper complexes, as well as with inorganic copper, the same morphological
and molecular changes were reproduced, indicating that copper overload is responsible for the cytotoxic effects. Copper-based
treatments drive sensitive cancer cells toward paraptotic cell death, a process hallmarked by endoplasmic reticulum stress and
massive vacuolization in the absence of apoptotic features. A lack of caspase activation, as observed in copper-treated dying cells, is a
consequence of metal-mediated inhibition of caspase-3. Thus, copper acts simultaneously as an endoplasmic reticulum (ER) stress
inducer and a caspase-3 inhibitor, forcing the cell into caspase-independent paraptotic death. The establishment of a mechanism of
action common to different copper binding agents provides a rationale for the exploitation of copper toxicity as an anticancer tool.
’ INTRODUCTION
to modulate metal ion reactivity and toxicity may contribute to a
successful outcome.
The use of metal ions in medicine can be traced back to the
early stages of civilization. Their (multi)positive charges and
stereoelectronic properties make them particularly apt to alter
the structure and function of important biological targets. As
such, metal ions are fundamental components of the biochemical
machinery. Notably, 47% of enzymes contain metals (41% in
their catalytic centers).1 For this reason, the transport and
homeostasis of metal ions are strictly controlled with specific
chaperones that have been evolutionarily designed.
Metal complexes have been investigated in recent decades as
potential anticancer compounds, largely due to the success of
cisplatin. This compound, one of the most widely employed
chemotherapeutic agents, has had a major impact on the treat-
ment of several tumor types, especially for testicular and ovarian
cancers.2 The efforts of synthetic chemistry to provide novel
anticancer agents have focused on low-toxicity platinum-based
and nonplatinum-based drugs with alternative mechanisms of
action. Regarding the latter, ruthenium,3 gold,4 and titanium5
have been investigated as candidates for the production of
anticancer compounds. In all of these cases, ligands that are able
Copper is a fundamental component of metalloproteins
because it exerts both catalytic and structural functions and its
distribution is strictly regulated at the levels of the cell, organ, and
body.6ꢀ8 Mutations in the genes coding for the copper trans-
porters ATP7A and ATP7B impair copper handling and result in
Menkes and Wilson diseases, respectively.9 Copper also plays
important roles in neoplastic diseases.10 In particular, copper is
crucial to the angiogenic process that sustains tumor and
metastasis development, and its sequestration underlies an antic-
ancer strategy aimed at preventing establishment of the tumor
blood supply.11 Alternatively, copper complexes have shown
anticancer activity, whereby the organic component (the ligand)
is responsible for directing the metal to different molecular
targets.12
Although the majority of anticancer compounds trigger apop-
tosis, other types of programmed cell death (PCD) occur in
response to common treatment regimens.13,14 Among the
Received: November 2, 2010
Published: March 31, 2011
r
2011 American Chemical Society
6235
dx.doi.org/10.1021/ja109413c J. Am. Chem. Soc. 2011, 133, 6235–6242
|

Journal of the American Chemical Society
ARTICLE
Scheme 1. Synthesis of Pyrazole-Pyridine Ligands L1ꢀL10
structure of the ligand fragment and the cytotoxic activity of
the resulting Cu(II) complexes, specific structural modifications
were applied to bis-pyrazole and pyrazole-pyridine, two molec-
ular backbones endowed with N,N copper-coordinating abilities.
The synthesis of the pyrazole-pyridine derivatives is described in
Scheme 1. Ligands L1ꢀL6 were prepared by reacting the electro-
philic chloroethyl-dimethylpyridine and pyrazole, in biphasic solvent
(toluene/water) via phase transfer catalysis (n-Bu4NOH).20
The preparation of L7ꢀL10 ligands is preformed by treating the
pyrazole-pyridine parent compounds with n-BuLi and an elec-
trophile bearing a thioether function (ClCH2SCH3).21,22
The bis-pyrazole ligands L11ꢀL14 were prepared by refluxing
deprotonated pyrazoles in dichloromethane, whereas ligands
L15ꢀL20 were prepared by treating bis(pyrazolyl)ketones with
aldehydes bearing the desired sulfur donor groups that were to be
attached to the bispyrazole scaffold (Scheme 2).23ꢀ25
Copper complexes were then prepared by mixing equimolar
Scheme 2. Synthesis of Bispyrazole Ligands L11ꢀL20
amount of CuCl2 2H2O and L1ꢀL20 ligands in methanol. In
3
most of the cases, the product precipitated out from the green
solution, from which it was recovered.
According to the molecular backbone and the nature of a
thioether third arm on the carbon center linking the two rings,
the ligands could be divided into five classes (aꢀe, Figure 1).
Within each class, peripheral residues on the pyrazole ring
were modified to vary the hydrophobicities or coordinative
abilities of ligands. Specifically, hydrophobicity was modulated
by substituting the H-atom of the parent compound with Me and
i-Pr as well as t-Bu and Ph when synthetically feasible. To
influence coordinative ability, a thioether group was introduced
on the pyrazole ring of compounds 6 and 10. According to the
X-ray structures reported in Figure 2 and in the Supporting
Information, the invariant feature of the Cu(II) complexes is an
N,N-chelation motif, whereas sulfur binding strongly depends on
the types of thioether fragments introduced. In fact, the sulfur
atom participates in metal coordination only when it is intro-
duced in a favorable steric position (compounds 6 and 10,
Figures 1 and 2). When the thioether was attached as a flexible
(compounds 7ꢀ9 and 18ꢀ20) or rigid (15ꢀ17) central arm, it
influenced the hydrophobicity of the ligand, but not its coordi-
native properties. The molecular structures can be grouped in
two categories: mononuclear and dinuclear. In the first case, the
metal geometry is flattened tetrahedral (compounds 4, 5, 13, 17,
19, and 20), whereas for the dinuclear case, the metal geometry is
square pyramidal, with the apex of the pyramid occupied by a
bridging chloride anion (compounds 2, 3, 7, 8, 9, 15, and 16).
Exceptions are represented by compounds 6 and 10 that exhibit a
trigonal bypyramidal geometry by virtue of the presence of the N,
N,S ligand donor set, and by compound 11 that forms chains
constituted by the tetrahedral [CuCl4]2ꢀ anion and the octahe-
dral [Cu(L11)2]2þ cation. Crystals suitable for X-ray data
collection could not be obtained for compounds 1, 12, 14, and
18. Nevertheless, on the basis of the molecular structures of
homologous compounds, it may be expected that the ligands
behave as N,N donors, 1/18 present dinuclear structures,
whereas compounds 12 and 14 exhibit mononuclear structures.
This tentative hypothesis can be justified by the presence or
absence of steric hindrance on the pyrazole rings of compounds
1/18 and 12/14, respectively.
various types of PCD, paraptosis, characterized by the occur-
rence of large cytoplasmic vacuoles derived from the endoplas-
mic reticulum (ER),15 was observed in cancer cells treated with
different copper complexes.16,17 A comparison between the
transcriptional responses elicited by cisplatin or a thioxotriazole
copper complex, showed activation of distinct pathways.18
Whereas cisplatin triggers canonical p53-dependent apoptosis,
the copper complex causes irreversible ER stress, ER vacuoliza-
tion and, eventually, paraptotic cell death.
To understand the structural basis of these effects, we pre-
pared and screened a series of pyrazole-pyridine and bis-pyrazole
copper complexes for cytotoxicity. The effects of the most active
compound were compared with those of inorganic copper and
disulfiram, a copper-interacting drug with promising anticancer
properties.19 The results showed that (a) the copper complexes
induce paraptosis as a function of their ability to cause intracel-
lular copper accumulation, (b) the metal itself is sufficient to
produce paraptotic features in cancer cells (provided that high
doses are used), and (c) during copper-dependent lethal ER
stress, caspase-3 inhibition by the metal accounts for the switch
from apoptotic to paraptotic cell death.
The anticancer activities of the complexes were evaluated in
human fibrosarcoma HT1080 cells by means of a resazurin-based
viability assay. Given that some compounds exhibited dinuclear
structures and some others mononuclear structures, the dose
’ RESULTS
Structureꢀactivity Relationship of Novel Cytotoxic Cop-
per Complexes. To establish a relationship between the
6236
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242

Journal of the American Chemical Society
ARTICLE
Figure 1. (A) Pyrazole-pyridine-based (rows a, b) and bis-pyrazole-
based (rows c, d, and e) copper(II) complexes. The coordinated chloride
anions are omitted for clarity. The table shows IC50 values for complexes
that are able to decrease HT1080 cell viability by more than 50% with
respect to the control. C.I. 95%: confidence interval at 95% for the IC50
values. Cells were treated with complexes in the 1ꢀ30 μM range for
48 h. (B) Relationship between cytotoxic activity and the cellular copper
concentration obtained by treating HT1080 cells for 6 h with 13.2 μM of
the indicated complexes. Cellular copper concentration was determined
by ICP-AES. The points are means ( SEM of three independent
experiments, with six and two replicates for viability and copper
determination, respectively.
Figure 2. X-ray molecular structures of representative Cu(II) com-
plexes. Solid thermal ellipsoids are reported at the 30% probability level.
Hydrogen atoms are omitted for clarity: C (gray), N (blue), Cl (green),
O (red), S (yellow), Cu (brown).
of the ligand and the cytotoxic activity of the respective complex
suggests that, for each class of compounds, there is an optimal
logP value (between 4 and 6) at which activity is maximal
(Supporting Information Figure S9).
response curves were based on the concentration of copper ion.
The IC50 values (Figure 1A), extrapolated from the dose
response curves reported in Supporting Information Figure S8,
allowed us to identify the following structureꢀactivity relation-
ships: (i) substitution of the H-atom on the pyrazole ring is
essential for the activity of the complex and (ii) both the presence
of the pyridine in place of pyrazole (compare classes a with c) and
the presence of the central thioether arm (compare classes a with
b and c with d and e) enhance activity. The highest activity was
observed for classes with flexible thioether groups (compare
classes b with d and e). Among the 20 compounds, 9 was the
most active (IC50 = 3 μM). The rank of the IC50 values within
each class points to the fundamental role of ligand hydrophobi-
city in the biological response of the corresponding Cu(II)
complex. The relationship between the calculated logP value26
The stability of the Cu(II)-ligand complexes was assessed by
evaluating the equilibrium formation constant as derived from
the following equation: Cu(II) þ L h Cu(II)L (β =
[Cu(II)L]/([Cu(II)] [L])). The logβ value was determined
3
for five complexes; it varied between 3.34 ( 0.15 and 4.69 ( 0.08
(Supporting Information Figure S7A). As expected, the highest
stability was found for compound 10, which employs three donor
atoms (N,N,S) for copper binding instead of the bidentate N,N
coordination seen in all of the other complexes. Higher stability
was not associated with increased cytotoxicity. Speciation calcu-
lations (Supporting Information Figure S7B) showed that, when
6237
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242

Journal of the American Chemical Society
ARTICLE
the ligand of compound 9 and copper were present at 5 μM each,
the expected concentration of the complex was less than 50 nM, a
value that would decrease further if copper-binding competitors
were taken into account. It may then be expected that for all the
complexes the metal should be largely displaced from the ligands by
the copper-binding competitors present in the cell culture medium.
Cell copper content was determined for the eleven active
complexes (IC50 < 30 μM) by means of ICP-AES. When cells
were incubated with the complexes at constant concentrations
corresponding to the mean IC50 (13.2 μM), a strong correlation
emerged between cytotoxic activity and cell copper content
(nonparametric Spearman r = 1). Figure 1B describes the
relationship between these two parameters and indicates that
the ability of a complex to promote copper accumulation in the
cell depends on the ligand structure and is predictive of its
cytotoxic activity.
Characterization of Ligands and Copper Uptake in Cells
Treated with Pyrazole-Pyridine Copper Complexes. To cor-
relate cytotoxic activity and complex bioavailability, we assessed
ligand uptake of the most active compound 9 and of the inactive
compound 7. The cell extracts were analyzed by means of
Reversed-Phase Liquid Chromatography (RPLC) combined
with an electrospray ionization linear ion trap (ESI-LIT) mass
spectrometry. As shown in Figure 3A, the accumulation of
9-ligand was 30-fold higher than that observed for the ligand of
the inactive (and less lipophilic) compound 7. This evidence
strongly suggests that passive diffusion accounts for membrane
crossing for this class of complexes. To verify this hypothesis, a
dose-dependent uptake experiment was performed in live cells.
Figure 3B shows the linear relationship between the intracellular
concentration of 9-ligand and the extracellular concentration of
9. These results indicate that, in the range of concentrations
tested, ligand transport is not saturated and follows the Fick law
for passive diffusion. The accumulation of 9-ligand reached a
plateau by approximately 30 min (Figure 3C). On the contrary,
under the same experimental conditions, copper uptake showed
much slower kinetics, approaching a steady state at 360 min.
However, at this time, in cells incubated with 5 μM of complex 9,
the intracellular copper content was approximately 4-fold higher
than the intracellular ligand content and more than 20-fold
higher than the copper level observed upon incubation with
5 μM inorganic copper. These results suggest that the ligand
carries copper across the membrane bilayer by virtue of its
lipophilicity and favors its accumulation in the intracellular
compartment, acting as an ionophore for the metal.
Relationship between Copper Content and Biological
Responses of Tumor and Normal Cells. Given the positive
correlation between the cytotoxic activity of 1ꢀ20 and the cell
copper content obtained with the same complexes (see
Figure 1B), we investigated whether inorganic copper (CuCl2)
was sufficient to produce similar effects. The cytotoxic response
to CuCl2 was assessed in three human tumor cell lines (HT1080,
HeLa, and SW872) and normal fibroblasts (HF) and compared
with the effects of the most active compound 9. The dose
response curve showed that CuCl2 was toxic above 200 μM,
while compound 9 was active at concentrations less than 10 μM
(Supporting Information Figure S10A and B). When CuCl2 was
used at a toxic concentration, the morphological changes induced
were identical to those caused by compound 9 (see Figure 5A).
These alterations were similar to those described for a thioxo-
triazole copper(II) complex that induces paraptosis.16,18 The
correlation between cell copper accumulation and cytotoxicity
Figure 3. (A) Cellular concentrations of 7- and 9-ligands were
determined by RPLC ESI-MS. HT1080 cells were incubated for 6 h
with 5 μM of the indicated complexes. (B) Uptake of 9-ligand in
HT1080 cells incubated for 2 min with the indicated concentrations of
9 (5ꢀ200 μM). (C) Uptake kinetics of 9-ligand (solid line) and copper
(dotted line). HT1080 cells were incubated with 9 (5 μM) for the
indicated time periods (0.5ꢀ360 min). The copper content of cells
incubated with 5 μM CuCl2 for 360 min is reported for comparison
(4). The inset is an enlargement of the first 30 min of uptake. All data
are means ( SEM of two independent experiments with three
replicates each.
6238
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242

Journal of the American Chemical Society
ARTICLE
Figure 5. (A) Phase contrast images representative of HT1080 or HeLa
cells incubated for 15 h with 9 (5 μM), CuCl2 (750 μM), or CuDSF
(1.5 μM). Vacuolated cells are evident in treated cultures (original magni-
fication: 400ꢁ). (B) Immunoblot analysis of UPR markers. HT1080 and
HeLa cells were treated as in A, and extracts were analyzed to detect
polyubiquitinylated proteins (poly-Ub), eIF2R phosphorylated in S51 (p-
eIF2R), and GADD34. Ponceau staining of the poly-Ub blotted mem-
brane (total prot) and total eIF2R are shown as the loading control.
induced by the various complexes (Figure 1B) holds true for
compound 9 and CuCl2 in HT1080, HeLa, SW872 2008, A549,
and MCF7 cells (nonparametric Spearman r = 0.86, Figure 4A).
Therefore, copper itself appears to be responsible for the biological
alterations induced in tumor celllines, independently of cancer cell
type and of the presence of the ligand. An exception was
represented by normal human fibroblasts (HF), where high
intracellular copper concentration did not translate into cyto-
toxicity, as should be expected from the curve in Figure 4A. The
results obtained in HF treated with complex 9 (Figure 4B and
Supporting Information Figure S10) prove that these cells are
able to withstand an intracellular copper concentration
(7.5 nmol/mg prot) that is toxic for cancer cells. Notably, when
Figure 4. (A) Tumor cell lines were treated with compound 9 or
CuCl2 at the indicated concentrations for 6 h. Untreated cells were
used as the control. Cytotoxicity (48 h) is plotted vs cell copper
content (6 h), and the data are fitted with a nonlinear regression curve
(r2 = 0.8). Data are means of three determinations of a representative
experiment. (B, left) Cell content of copper (ICP-AES). (right) Cell
content of 9-ligand (RPLC ESI-MS). Cells were incubated for 6 h with
5 μM of 9. Data are means ( SD of three independent experiments
with three replicates each. (C) qPCR analysis of gene expression
induced by 9. Cells were left untreated (white bars) or incubated for 9 h
with 5 μM of 9 (black bars). The expression levels of the indicated
genes are normalized for the amount of RPL15 mRNA and are
reported as arbitrary units. Data are means ( SD of two independent
experiments with three replicates each.
6239
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242

Journal of the American Chemical Society
ARTICLE
Promotion of Massive Cytoplasmic Vacuolization and the
Unfolded Protein Response by Inorganic and Complexed
Copper. The anticancer activity of disulfiram (DSF), an inhibitor
of aldehyde dehydrogenase used in ethilism therapy, is markedly
enhanced by copper28,29 and a clinical trial with DSF in combina-
tion with copper against liver cancer is in the recruitment phase
(ClinicalTrials.gov Identifier: NCT00742911). When equimolar
amounts of CuCl2 and DSF were administered to HT1080 and
HeLa cells (Figure 5A), massive cytoplasmic vacuolization in the
absence of apoptotic features was observed. These morphologi-
cal changes coincided with the paraptotic morphology15,16
caused by complex 9 or CuCl2 and observed with [Cu(8-
hydroxiquinoline)2] and [Cu(clioquinole)2], which promote
copper transport into cells30 (Supporting Information Figure
S11). Consistent with previous data,28,29,31,32 the metal shifted
the IC50 value of DSF from 5.2 to 0.4 μM and from 15.7 to 0.4
μM in HT1080 and HeLa cells, respectively.
ER-derived paraptotic vacuolization is part of the unfolded
protein response (UPR) that is sustained by unresolved ER
stress.18,33 Consistently, in HT1080 and HeLa cells, complex 9,
CuCl2, and CuDSF all increased the amount of polyubiquitiny-
lated proteins, the causal event of the UPR (Figure 5B). More-
over, in both cell models, complex 9, CuCl2, and CuDSF
increased the ratio between phosphorylated and total eukaryotic
initiation factor 2 alpha (eIF2R), the phosphorylation of which
promotes the UPR.34 GADD34, an ATF4-dependent UPR target
gene downstream of eIF2R, was barely detectable in untreated
cells but was clearly induced in both cell lines treated with
compound 9, CuCl2, or CuDSF.
Caspase Inhibition by Inorganic and Complexed Copper.
The activation of caspases, the apoptotic execution machinery,
was assessed during treatment with inorganic (CuCl2) or com-
plexed copper (9 and CuDSF). Supporting Information Figure
S12A shows images of HT1080 cultures undergoing cell death
upon treatment with cisplatin, complex 9, CuCl2, or CuDSF. In a
significant fraction of cells, cisplatin clearly induced an apoptotic
morphology, including nuclear fragmentation (DAPI staining)
and caspase activation, as visualized with a fluorogenic pan-
caspase substrate. On the contrary, vacuolated cells, treated with
either inorganic or complexed copper, were negative for caspase
and exhibited neither chromatin condensation nor nuclear
fragmentation in apoptotic bodies. In both HT1080 and HeLa
cells, differential caspase-3 behavior in cisplatin-treated and
copper-treated cells was confirmed by Western blot, with the
cleaved form much more evident in cisplatin-treated than in
copper-treated cells (Figure 6A), where caspase activation was
only detectable after prolonged film exposure (Supporting
Information Figure S12B). The expression of a known pro-death
ER stress-related protein, CHOP, was clearly induced by all of
the copper-based treatments, whereas it remained undetectable
or slightly induced in cisplatin treated HT1080 and HeLa cells,
respectively (Figures 6A and Supporting Information Figure
S12B).
Figure 6. (A) Immunoblot analysis of caspase-3 and CHOP. HT1080
and HeLa cells were treated for 15 h with 9 (5 μM), CuCl2 (750 μM),
CuDSF (1.5 μM), or cisplatin (cisPt, 20 μM) and extracts were analyzed
to detect both the inactive (procasp-3) and active (cleaved casp-3) forms
of caspase-3. The same membrane was blotted for the ER stress-related,
pro-death protein CHOP. The * symbols corresponds to a nonspecific
band. β-Tubulin is shown as a loading control, and cisplatin (cisPt) is
used as a positive control for apoptosis. (B) Effects of inorganic or
complexed copper on caspase-3 activity. (left) Cell extracts used for the
immunoblots shown in A analyzed for caspase-3 activity. Data are means
( SD of six replicates of a representative experiment. (right) HT1080
cells treated with cisplatin as in A. Cell extracts (circles) or a solution of
purified human recombinant caspase-3 (triangles) were incubated for
1 h with CuCl2 (1ꢀ30 μM) and then analyzed for caspase-3 activity. The
results are reported as the percentage of the maximal activity obtained in
the absence of CuCl2. Data are means ( SD of three independent
experiments with three replicates each.
HT1080, HeLa, SW872, and HF were treated for 6 h with 5 μM
of compound 9, cell copper contents were not related to the cell
contents of 9-ligand (Figure 4B). In fact, copper always reached
significantly higher cell levels than the ligand, with a copper/
ligand ratio that varied from 2.3 to 4.4. These results confirm that
ligand and copper establish different transmembrane gradients.
To verify if the paraptotic “transcriptional fingerprint” induced
in HT1080 cells by a cytotoxic thioxotriazole copper(II) com-
plex18 was reproduced by complex 9, we assessed the expression
of markers involved in metal-induced response (MT1X), oxida-
tive stress (HMOX), and endoplasmic reticulum (ER) stress
(CHOP, GADD34, sXBP1) experiments. In HT1080 cells,
compound 9 markedly induced all tested genes (Figure 4C),
confirming that copper complexes with unrelated ligand struc-
tures induce the same molecular events associated with para-
ptosis. The ability of compound 9 to induce this transcriptional
response in other cell models was evaluated. As expected, MT1X
mRNA was increased as a result of an increase in cellular copper
concentration in all cell types tested. On the contrary, the
induction of HMOX, a known indicator of oxidative stress27
was considerably higher in the two sensitive cell lines (48- and
23-fold induction, normalized to the control value, for HT1080
and HeLa cells, respectively) compared to the less sensitive
SW872 and HF cells (2.5- and 1.8-fold induction, respectively).
The genes encoding ER stress executors were upregulated
exclusively in the sensitive HT1080 and HeLa cells.
The activity of caspase-3 was also evaluated in extracts of
cells incubated with complex 9, CuCl2, CuDSF, or cisplatin
(Figure 6B). Copper-based compounds increased caspase activ-
ity by 21% and 53% in HT1080 and HeLa cells, respectively,
whereas cisplatin induced increases of 216% and 142%, respe-
ctively.
The effects of copper (CuCl2) on caspase-3 activity were
directly assessed in either extracts of HT1080 cells pretreated
with cisplatin or human recombinant caspase-3 in solution
6240
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242

Journal of the American Chemical Society
ARTICLE
spaces. It is worth mentioning that, despite its high stability, the
[Cu(histidine)2] complex exhibits negligible cytotoxic activity
(Supporting Information Figure S11).
In the cell, high concentrations of high-affinity Cu-binding
proteins and ligands are present (i.e., glutathione may reach
concentrations of approximately 10 mM), entrapping the metal,
which cannot freely pass the plasma membrane; conversely,
neutral ligands with optimal lipophilicity (such as that of complex
9) freely move across the three compartments. The kinetics of
complex dissociation may also play a role in influencing the
amount of complex that can cross the membrane. Steric encum-
brance at the metal center, as seen for ligands with a mobile
thioether group, can slow substitution kinetics by hindering the
interaction with potential biological chelators. This fact may
explain why complexes 9 and 19 are more active than complexes 3
and 16. Provided that it is used at sufficiently high concentrations,
inorganic copper causes the same effects observed for copper
complexes, thus yielding further evidence that the complex is
dissociated inside cells.
This study demonstrates that there is a correlation between
observed cytotoxicity and the intracellular copper content. How
can high levels of intracellular copper cause cytotoxicity? Given
that no free copper ions are available inside eukaryotic cells,35 the
metal is largely coordinated by reduced glutathione (GSH), the
major redox buffer system of the cell, hence perturbing the GSH/
GSSG ratio.36,37 Because correct protein folding needs a con-
trolled redox environment, a lack of copper-dependent oxidative
balance can markedly affect protein conformation.38 Further-
more, when the copper-handling capacity of the cell is exceeded,
the metal binds to nonspecific targets and may perturb their
structure and function, as is the case of copper-induced protea-
some inhibition that leads to the accumulation of misfolded
proteins.18,31,39 These combined events cause unsolvable ER
stress, manifested as massive cytoplasmic vacuolization, and
activation of the UPR (eIF2R phosphorylation, XBP1 splicing,
induction of GADD34 and CHOP).34 Therefore, a different
response to misfolded protein accumulation and ER stress (i.e.,
eIF2R phosphorylation) may translate into different sensitivities
to copper overload, explaining the variability observed between
the cell models used in this study.
Although UPR may link ER stress to the apoptotic pathway
through the induction of pro-death proteins such as CHOP,
apoptosis is not observed in the death process triggered by high
intracellular copper levels because the metal itself inhibits
caspase-3. Despite the lack of evidence for a direct interaction
with the enzyme, it is likely that copper interferes with catalytic
cysteine residues in the active site. The different inhibition
profiles of caspase-3, obtained with purified recombinant enzyme
or cell lysates, may be reasonably explained by taking into
account the presence of several metal-binding species in crude
extracts that impair the copper-caspase interaction. This phe-
nomenon was previously described in a study that assessed the
role of Zn(II) in caspase inhibition.40
Figure 7. Proposed model for a copper-binding agent mechanism of
action.
(Figure 6B). The two doseꢀresponse curves show that copper
inhibits caspase with an IC50 of 11.5 μM and a maximum
inhibition of 70% at 30 μM. Nevertheless, the two responses
are distinct, as only the activity of recombinant caspase-3 was
significantly inhibited at 1 μM CuCl2.
’ DISCUSSION
This study demonstrates that lipophilicity is a relevant deter-
minant of the cytotoxic activity of a copper complex. The highest
activity is reached with an optimal ligand hydrophobicity that
allows rapid equilibrium of the complex between the lipid (cell
membranes) and aqueous compartments (intra- and extra-
cellular). For lower logP values, the Cu(II)-ligand ensemble
may be too hydrophilic to cross the cell membrane efficiently,
while for higher logP values, the complex may prefer to stay in the
membrane fraction. Thus, ligands contribute to copper toxicity
by facilitating metal release into the cell, according to the
mechanism of action model proposed in Figure 7. However,
the fact that the correlation between logP value and cytotoxicity
does not obey the same law for the five ligand classes (Supporting
Information Figure S9) indicates that other parameters influence
cytotoxic activity.
According to the stability constant, the ligand is mostly
dissociated from the metal in the extracellular environment,
where copper is sequestered by binding to competitors such as
amino acids and serum proteins. For instance, the stability of the
[Cu(histidine)2] complex is several orders of magnitude higher
than the stabilities of the complexes studied here.12 Nevertheless,
the portion of copper that remains complexed, following the diff-
usion law, equilibrates between the extracellular and intracellular
Our findings point to a mechanism of action for copper
binding agents that can also apply to disulfiram (DSF), a
copper-interacting drug repeatedly described as an apoptotic
inducer in several in vitro and in vivo cancer models.31,32 In our
models, DSF concentrates copper in the cell (Supporting
Information Figure S11B), reproducing the morphological and
molecular changes observed with other copper complexes,
indicative of paraptotic cell death (Figures 5 and 6). Only in a
very limited fraction of treated cells CuDSF causes apoptotic
6241
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242

Journal of the American Chemical Society
ARTICLE
changes, which is consistent with the marginal increase in the
caspase-3 activity observed under these conditions.
(16) Tardito, S.; Bussolati, O.; Gaccioli, F.; Gatti, R.; Guizzardi, S.;
Uggeri, J.; Marchiꢀo, L.; Lanfranchi, M.; Franchi-Gazzola, R. Histochem.
Cell Biol. 2006, 126, 473.
For the copper complexes examined in this study, as well as for
other previously described compounds,17,37 the biological effects
consistently observed are the intracellular accumulation of high
amounts of copper, the appearance of cytoplasmic vacuolization,
and the activation of caspase-independent cell death. These
observations can be reasonably explained if the ligand behaves
as a copper ionophore that exacerbates the intrinsic toxicity of the
metal by increasing its intracellular concentration. If a ligand has
autonomous biological activities, as in the case of DSF, the two
effects might be synergistic, leading to a potent cytotoxic
behavior. Since DSF associated with copper is currently under
clinical investigation as an anticancer agent, these findings may be
useful for designing an effective therapeutic combination with
pro-apoptotic drugs.
(17) Marzano, C.; Pellei, M.; Colavito, D.; Alidori, S.; Gioia-Lobbia,
G.; Gandin, V.; Tisato, F.; Santini, C. J. Med. Chem. 2006, 49, 7317.
(18) Tardito, S.; Isella, C.; Medico, E.; Marchiꢀo, L.; Bevilacqua, E.;
Hatzoglou, M.; Bussolati, O.; Franchi-Gazzola, R. J. Biol. Chem. 2009,
284, 24306.
(19) Cvek, B.; Dvorak, Z. Drug Discov. Today 2008, 13, 716.
(20) Gennari, M.; Tegoni, M.; Lanfranchi, M.; Pellinghelli, M. A.;
Marchiꢀo, L. Inorg. Chem. 2007, 46, 3367.
(21) Gennari, M.; Lanfranchi, M.; Marchiꢀo, L. Inorg. Chim. Acta
2009, 362, 4430.
(22) Gennari, M.; Tegoni, M.; Lanfranchi, M.; Pellinghelli, M. A.;
Giannetto, M.; Marchiꢀo, L. Inorg. Chem. 2008, 47, 2223.
(23) Higgs, T. C.; Ji, D.; Czernuszewicz, R. S.; Matzanke, B. F.;
Schunemann, V.; Trautwein, A. X.; Helliwell, M.; Ramirez, W.; Carrano,
C. J. Inorg. Chem. 1998, 37, 2383.
(24) The, K. I.; Peterson, L. K.; Kiehlmann, E. Can. J. Chem. 1973,
51, 2448.
(25) Gennari, M.; Bassanetti, I.; Marchiꢀo, L. Polyhedron 2010,
29, 361.
’ ASSOCIATED CONTENT
S
Supporting Information. Synthesis of ligands and cop-
b
(26) The octanol-water partition coefficient P is the common
quantitative descriptor of lipophilicity. P is defined as the ratio of the
concentrations of a compound in two immiscible phases (octanol and
water) under equilibrium conditions. It is mostly used in its logarithmic
form, logP. Experimental procedures to measure logP are described in
detail in: Sangster, J. Octanol-water partition coefficients: Fundamentals
and physical chemistry; John Wiley & Sons Ltd.: Chichester, 1997. In the
present work, the calculation of the logP values was performed with
fragmental methods implemented in the ALOGPS program:VCCLAB,
Virtual Computational Chemistry Laboratory; http://www.vcclab.org,
2005.
per complexes, stability constants determination, X-ray struc-
tures of 2ꢀ9, 11ꢀ13, 15ꢀ17, and 19ꢀ20 complexes (CCDC
792828ꢀ792842), cell cultures, Western blots, cell viability
assays, real time PCR, copper and ligand cellular uptake assays,
caspase-3 activity assays, and supporting figures. This material is
available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
marchio@unipr.it
(27) Gozzelino, R.; Jeney, V.; Soares, M. P. Annu. Rev. Pharmacol.
Toxicol. 2010, 50, 323.
(28) Zhang, H. J.; Chen, D.; Ringler, J.; Chen, W.; Cui, Q. C.; Ethier,
S. P.; Dou, Q. P.; Wu, G. J. Cancer Res. 2010, 70, 3996.
(29) Iljin, K.; Ketola, K.; Vainio, P.; Halonen, P.; Kohonen, P.; Fey,
V.; Grafstrom, R. C.; Perala, M.; Kallioniemi, O. Clin. Cancer Res. 2009,
15, 6070.
’ ACKNOWLEDGMENT
This work has been funded by a MIUR grant to L.M. (PRIN
2007 20078EWK9B_003).
(30) Zhai, S. M.; Yang, L.; Cui, Q. C.; Sun, Y.; Dou, Q. P.; Yan, B.
J. Biol. Inorg. Chem. 2010, 15, 259.
(31) Chen, D.; Cui, Q. Z. C.; Yang, H. J.; Dou, Q. P. Cancer Res.
2006, 66, 10425.
(32) Cen, D. Z.; Brayton, D.; Shahandeh, B.; Meyskens, F. L.;
Farmer, P. J. J. Med. Chem. 2004, 47, 6914.
(33) Kar, R.; Singha, P. K.; Venkatachalam, M. A.; Saikumar, P.
Oncogene 2009, 28, 2556.
(34) Szegezdi, E.; Logue, S. E.; Gorman, A. M.; Samali, A. Embo Rep.
2006, 7, 880.
(35) Rae, T. D.; Schmidt, P. J.; Pufahl, R. A.; Culotta, V. C.;
O’Halloran, T. V. Science 1999, 284, 805.
(36) Valko, M.; Rhodes, C. J.; Moncol, J.; Izakovic, M.; Mazur, M.
Chem. Biol. Inter. 2006, 160, 1.
(37) Tardito, S.; Bussolati, O.; Maffini, M.; Tegoni, M.; Giannetto,
M.; Dall’Asta, V.; Franchi-Gazzola, R.; Lanfranchi, M.; Pellinghelli,
M. A.; Mucchino, C.; Mori, G.; Marchiꢀo, L. J. Med. Chem. 2007, 50, 1916.
(38) Malhotra, J. D.; Kaufman, R. J. Antioxid. Redox Signal 2007,
9, 2277.
(39) Daniel, K. G.; Chen, D.; Yan, B.; Dou, Q. P. Front. Biosci. 2007,
12, 135.
(40) Perry, D. K.; Smyth, M. J.; Stennicke, H. R.; Salvesen, G. S.;
Duriez, P.; Poirier, G. G.; Hannun, Y. A. J. Biol. Chem. 1997, 272, 18530.
’ REFERENCES
(1) Guengerich, F. P. J. Biol. Chem. 2009, 284, 709.
(2) Kelland, L. Nat. Rev. Cancer 2007, 7, 573.
(3) Antonarakis, E. S.; Emadi, A. Cancer Chemother. Pharmacol.
2010, 66, 1.
(4) Nobili, S.; Mini, E.; Landini, I.; Gabbiani, C.; Casini, A.; Messori,
L. Med. Res. Rev. 2010, 30, 550.
(5) Kostova, I. Anticancer Agents Med. Chem. 2009, 9, 827.
(6) Kim, B. E.; Turski, M. L.; Nose, Y.; Casad, M.; Rockman, H. A.;
Thiele, D. J. Cell Metab. 2010, 11, 353.
(7) Banci, L.; Bertini, I.; Ciofi-Baffoni, S.; Kozyreva, T.; Zovo, K.;
Palumaa, P. Nature 2010, 465, 645.
(8) Collins, J. F.; Prohaska, J. R.; Knutson, M. D. Nutr. Rev. 2010,
68, 133.
(9) de Bie, P.; Muller, P.; Wijmenga, C.; Klomp, L. W. J. J. Med.
Genet. 2007, 44, 673.
(10) Turski, M. L.; Thiele, D. J. J. Biol. Chem. 2009, 284, 717.
(11) Finney, L.; Vogt, S.; Fukai, T.; Glesne, D. Clin. Exp. Pharmacol.
Physiol. 2009, 36, 88.
(12) Tardito, S.; Marchiꢀo, L. Curr. Med. Chem. 2009, 16, 1325.
(13) Broker, L. E.; Kruyt, F. A. E.; Giaccone, G. Clin. Cancer Res.
2005, 11, 3155.
(14) Brown, J. M.; Attardi, L. D. Nat. Rev. Cancer 2005, 5, 231.
(15) Sperandio, S.; de Belle, I.; Bredesen, D. E. Proc. Natl. Acad. Sci.
U.S.A. 2000, 97, 14376.
6242
dx.doi.org/10.1021/ja109413c |J. Am. Chem. Soc. 2011, 133, 6235–6242