Synthesis and characterization of the anticancer and metal binding properties of novel pyrimidinylhydrazone derivatives

Article abstract of DOI:10.1016/j.jinorgbio.2014.12.015

Three novel pyrimidinylhydrazones substituted at either the aromatic moiety or at the imine carbon atom were synthesized and characterized by standard analytical methods. All compounds were found to be toxic in the micro- to submicromolar range against a diverse panel of cancer cell lines including multidrug resistant (MDR) derivatives expressing P-glycoprotein (Pgp). UV-visible spectrophotometry experiments demonstrated that the most active compound (3) forms highly stable complexes with iron(III) and copper(II) in a wide pH range with a stronger preference towards iron(III). The redox activity of the iron and copper complexes of ligand 3 was investigated using cyclic voltammetry and was tested with cellular reductants. The impact of reactive oxygen species (ROS) on the mechanism of toxicity was assessed using the ROS-sensitive cell permeable dye 2′,7′-dichlorofluorescin diacetate (DCFDA). Our results demonstrate that the studied pyrimidinylhydrazones form redox-active iron and copper complexes that are capable of producing intracellular ROS, which might lead to cellular damage and cell death in cancer cells regardless of their resistance status.

Full text of DOI:10.1016/j.jinorgbio.2014.12.015

Synthesis and characterization of the anticancer and metal binding properties
of novel pyrimidinylhydrazone derivatives
Veronika F.S. Pape, D o´ ra T u¨ rk, P a´ l Szab o´ , Michael Wiese, Eva A. Enyedy,
Gergely Szak a´ cs
PII:
DOI:
Reference:
S0162-0134(14)00308-0
doi: 10.1016/j.jinorgbio.2014.12.015
JIB 9638
To appear in:
Journal of Inorganic Biochemistry
Received date:
Revised date:
Accepted date:
22 August 2014
12 December 2014
12 December 2014
Please cite this article as: Veronika F.S. Pape, D ´o ra T u¨ rk, P ´a l Szab o´ , Michael Wiese,
Eva A. Enyedy, Gergely Szak a´ cs, Synthesis and characterization of the anticancer and
metal binding properties of novel pyrimidinylhydrazone derivatives, Journal of Inorganic
Biochemistry (2014), doi: 10.1016/j.jinorgbio.2014.12.015
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. 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.

ACCEPTED MANUSCRIPT
Synthesis and characterization of the anticancer and metal binding
properties of novel pyrimidinylhydrazone derivatives
a,b
a
c
b
d,*
Veronika F.S. Pape , Dóra Türk , Pál Szabó , Michael Wiese , Eva A. Enyedy ,
a,*
Gergely Szakács
a
Institute of Enzymology, Research Centre for Natural Sciences, Hungarian Academy of Sciences, Magyar
Tudósok körútja 2, H-1117 Budapest, Hungary
b
Department of Pharmaceutical Chemistry, University of Bonn, An der Immenburg 4, D-53121 Bonn, Germany
c
Institute of Organic Chemistry, Research Centre for Natural Sciences, Hungarian Academy of Sciences,
Magyar Tudósok körútja 2, H-1117 Budapest, Hungary
d
Department of Inorganic and Analytical Chemistry, University of Szeged, Dóm tér 7. H-6720 Szeged, Hungary
Keywords: Stability Constants ∙ Metal Complexes ∙ Arylhydrazones ∙ Multidrug Resistance ∙ Reactive
Oxygen Species
*
Corresponding authors: enyedy@chem.u-szeged.hu (E. A. Enyedy), szakacs.gergely@ttk.mta.hu
(
G. Szakács)
1

ACCEPTED MANUSCRIPT
Abstract
Three novel pyrimidinylhydrazones substituted at either the aromatic moiety or at the
imine carbon atom were synthesized and characterized by standard analytical methods. All
compounds were found to be toxic in the micro- to submicromolar range against a diverse
panel of cancer cell lines including multidrug resistant (MDR) derivatives expressing P-
glycoprotein (Pgp). Using UV-visible spectrophotometry experiments demonstrated that the
most active compound (3) forms highly stable complexes with iron(III) and copper(II) in a
wide pH range with a stronger preference towards iron(III). The redox activity of the iron and
copper complexes of ligand 3 was investigated using cyclic voltammetry and was tested with
cellular reductants. The impact of reactive oxygen species (ROS) in the mechanism of
toxicity was assessed using the ROS-sensitive cell permeable dye 2′,7′-dichlorofluorescin
diacetate (DCFDA). Our results demonstrate that the studied pyrimidinylhydrazones form
redox-active iron and copper complexes that are capable of producing intracellular ROS,
which might lead to cellular damage and cell death in cancer cells regardless of their
resistance status.
2

ACCEPTED MANUSCRIPT
1
. Introduction
The cellular homeostasis of metals is tightly regulated. As a result, only a minor
fraction of the intracellular pools is freely available, and the majority of metal ions are bound
to storage proteins or enzymes that utilize them as cofactors or as structural support elements
[
1,2]. It is estimated that at least one third of the human proteome consists of metalloproteins,
many of which have been associated with various pathological conditions including
infectious, cardiovascular and neurodegenerative diseases [3,4]. As compounds that can
directly target the metal ion cofactors of metalloproteins are expected to have a significant
biological effect, the development of chelators is a promising strategy in medicinal chemistry.
Chelators have been traditionally used for the treatment of metal overload and diseases
related to imbalanced metal homeostasis including hemochromatosis, β-thalassemia, and
Alzheimer’s or Parkinson’s diseases [5,6]. Several studies have described the deregulation of
iron homeostasis in cancer, suggesting that the altered metal homeostasis of cancer cells
represents a vulnerability that can be targeted by chelation strategies [7,8]. Cancer cells
increasingly rely on iron due to the crucial role of metalloproteins in proliferation. Since
metal homeostasis is tightly linked to the regulation of the intracellular redox balance,
anticancer metal complexes can also target cancer cells by the formation of redox active
complexes giving rise to reactive oxygen species (ROS) [5,8].
Examples for chelators with antitumor potential include aroyl- and arylhydrazones,
and variously substituted thiosemicarbazones [9–15]. The proposed mechanism of action of
these antibacterial and anticancer compounds is based on the inhibition of the iron-requiring
enzyme ribonucleotide reductase (RR) [5,8], the rate determining enzyme supplying
deoxyribonucleotides for DNA synthesis [16]. The mammalian enzyme consists of two
subunits: R1 harbours the catalytically active center; R2 contains a tyrosyl radical and a
3

ACCEPTED MANUSCRIPT
diiron center [17]. Compounds targeting RR include antimetabolites such as gemcitabine,
scavengers of the tyrosyl radical (hydroxyurea), gallium(III) complexes and various
chelators. Triapine, a tridentate -N-pyridyl thiosemicarbazone is a potent RR inhibitor
currently undergoing phase I and II clinical trials [18–20]. Unfortunately, Triapine is subject
to multidrug resistance (MDR), as it is recognized by the ATP-binding cassette (ABC)
transporter P-glycoprotein (Pgp/ABCB1) [21]. MDR is a major obstacle in the treatment of
cancer. Hence, there is an urgent need for new strategies aiming to overcome MDR [22].
Numerous metal complexes have been developed with the aim of bypassing or even targeting
MDR, while minimizing the deplorable range of side effects of clinically used platinum
complexes [23].
Arylhydrazones represent a promising chelator class characterized by antibacterial,
antimycobacterial, analgesic and anticancer activities [12,13,24,25]. Several pyridyl- and
pyrimidinylhydrazones were reported to be effective against multidrug resistant strains of
leprosy and tuberculosis [13,26] and cancer cell lines [14]. Compared to closely related
derivatives, pyrimidinylhydrazones with 4’-methoxy- and 5-ethyl-substituents have been
suggested to show a higher and broader activity against Mycobacteria [15]. In the case of
structurally related aroylhydrazones the introduction of an additional pyridine ring at the
imine carbon of pyridylcarboxaldehyde isonicotinoyl hydrazone (PCIH) led to increased iron
chelation and enhanced antitumor activity [11]. Here we have investigated the antitumor
activity of arylhydrazones from the pyrimidinyl type. Since the α-position to the hydrazine
moiety was shown to influence toxicity, we synthesized a derivative with methylation at the
imine carbon. The ligands were tested for their in vitro antiproliferative activity in six human
cancer cell lines. We show that three donor nitrogen atoms enable the formation of stable and
redox-active complexes with iron(III) and copper(II). UV-visible (UV‒Vis)
spectrophotometric titrations were applied to investigate speciation in aqueous solution, and
4

ACCEPTED MANUSCRIPT
cyclic voltammetry was used to characterize the redox activity of the complexes. The impact
of metal ions on the antiproliferative activity and the formation of ROS was also investigated.
2
. Materials and methods
2
.1. Chemicals
Chemicals used for synthesis were procured from Acros Organics (Geel, Belgium), Alfa
Aesar (Karlsruhe, Germany), Sigma-Aldrich (Schnelldorf, Germany) or TCI (Eschborn,
Germany) and used without further purification. Column chromatography was performed
using Silica gel 60 (40-63 µm, Merck, Darmstadt, Germany) as stationary phase. 3-(4,5-
dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), N-acetylcysteine (NAC) and
2
′,7′-dichlorofluorescin diacetate (DCFDA) were purchased from ABCR Chemicals
(Karlsruhe, Germany), TCI (Eschborn, Germany) and Sigma Aldrich (Schnelldorf,
Germany), respectively. The Pgp inhibitor WK-X-24 (XR9577) was used from prior
synthesized stocks (Laboratory of Prof. M. Wiese, Bonn university, Germany) [27],
Tariquidar was a kind gift from Dr. Susan Bates (NCI). Solid KOH, and tetrabutylammonium
chloride (TBACl), ferrocene, uracil, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
(
HEPES), 2-(N-morpholino)ethanesulfonic acid (MES), ascorbic acid (ASC) and glutathione
GSH) were purchased from Sigma Aldrich and HCl, KCl, CuCl , FeCl were Reanal
(
2
3
products (Budapest, Hungary). Fe(III) and Cu(II) stock solutions were prepared by dissolving
the appropriate amount of the metal chlorides in known amounts of HCl. Their concentrations
were determined by complexometry via the EDTA complexes. Accurate strong acid content
5

ACCEPTED MANUSCRIPT
of the metal stock solutions were determined by pH-potentiometric titrations. Ligand 3 was
dissolved in pure DMSO to obtain the stock solution (0.01 M).
2
.2. Synthesis and physical measurements
The ligands were prepared following the synthetic route suggested by Seydel et al. with
1
13
minor modifications [15]. H NMR and C NMR spectra were obtained on either a Bruker
Advance 300 or 500 spectrometer, respectively. DMSO-d or CDCl were used as solvents.
6
3
Standard pulse programs were applied. Chemical shifts are expressed in ppm values using the
residual solvent peaks as internal standards (DMSO-d 2.50; 39.52 ppm or CDCl 7.26; 77.16
6
3
13
ppm) [28]. C NMR signals were assigned with the aid of attached proton test (APT) spectra.
Elemental analyses of the final products were performed on a Vario EL elemental analyser
(Hanau, Germany). The values for carbon, nitrogen and hydrogen are given in percentage.
Electrospray ionisation-mass spectrometry (ESI-MS) measurements for the characterization
of the final products were carried out with a Waters Q-TOF Premier instrument (Waters Kft.,
Hungary) operating in positive ion mode; the samples were dissolved in methanol.
2
.2.1. 4-Chloro-6-methoxypyrimidine
To a cooled solution (-10 °C) of 4,6-dichloropyrimidine (1193-21-1, 5 g, 33.56 mmol) in dry
methanol, one equivalent of sodium hydride (titrated solution in dry methanol) was added,
the solution was allowed to warm to room temperature and react for 48 h (depicted in step a
of Scheme 1). The solvent was removed; the crude product was dissolved in brine, extracted
1
with CHCl , and was used without further purification. H NMR (500 MHz, DMSO-d ): δ =
3
6
5
5
8
.68 (d, J(H,H) = 0.8 Hz, 1H, H-2), 7.18 (d, J(H,H) = 0.9 Hz, 1H, H-5), 3.96 (s, 3H, OCH ).
3
6

ACCEPTED MANUSCRIPT
2
.2.2. 4-Hydrazinyl-6-methoxypyrimidine
To a solution of the 4-chloro-6-methoxypyrimidine, obtained as described in 2.2.1., (2.5 g
crude product) in methanol an aqueous solution of hydrazine (80%, 6.1 mL) was added and
refluxed (see step b of Scheme 1). After evaporation of the solvent, the aqueous solution of
the crude product was alkalized with NaOH and extracted with methylene chloride to give the
1
product as an ochre powder in 60% yield. H NMR (500 MHz, CDCl ): δ = 8.26 (s, 1H, H-2),
3
1
3
6
.42 (bs, 1H, NH), 6.07 (s, H-5), 3.92 (s, 3H, OCH ), 3.42 (bs, 2H, NH ). C NMR
3 2
(
126 MHz, CDCl ): δ = 170.72 (C-6), 167.17 (C-4), 157.68 (C-2), 84.90 (C-5), 53.87
3
(OCH3).
2
.2.3. 5-Ethyl-2-methyl-pyridine-N-oxide
3
5 mL of hydrogen peroxide (35% solution, 0.41 mol) was added to a solution of 5-ethyl-2-
methyl-pyridine (104-90-5, 21.76 mL, 0.17 mol) in glacial acetic acid (200 mL) and refluxed
for 16 h (depicted in step c of Scheme 1). After evaporation of the majority of the solvent, the
reaction mixture was neutralized with NaOH and Na CO and the starting material was
2
3
retrieved by extraction with petroleum ether. 5-Ethyl-2-methyl-pyridine-N-oxide was
extracted from the aqueous phase with CHCl , dried over Na SO and after removal of the
3
2
4
1
solvent obtained as a bright yellow liquid in 99% yield. H NMR (500 MHz, CDCl ): δ =
3
3
4
3
8
7
7
.14 – 8.09 (m, H-6), 7.12 (d, J(H,H) = 7.9 Hz, H-3), 7.00 (dd, J(H,H) = 1.3 Hz, J(H,H) =
3
3
.9 Hz, H-4), 2.55 (q, J(H,H) = 7.7 Hz, 2H, CH CH ), 2,48 (s, 3H, CH ), 1.20 (t, J(H,H) =
2
3
3
.7 Hz, 3H, CH CH ).
2
3
2
.2.4. 5-Ethylpyridin-2-yl-methyl acetate
As depicted in step d of Scheme 1, a solution of the 5-ethyl-2-methyl-pyridine-N-oxide,
obtained in 2.2.5., (22.28 g, 0.16 mol) and acetic anhydride (28 mL, 0.23 mol) in glacial
7

ACCEPTED MANUSCRIPT
acetic acid (7 mL) was refluxed for 3 h, put on ice, neutralized, extracted with methylene
chloride. After removal of the solvent, the extract was dissolved in diethyl ether, washed with
a saturated solution of Na CO and dried over Na SO to give the product in 85% yield.
2
3
2
4
1
4
4
H NMR (500 MHz, CDCl ): δ = 8.42 (d, J(H,H) = 2.1 Hz, H-6), 7.50 (dd, J(H,H) = 2.2 Hz,
3
3
3
3
J(H,H) = 7.9 Hz, H-4), 7.24 (d, J(H,H) = 7.9 Hz, H-3), 5.16 (s, 2H, H-7), 2.63 (q, J(H,H) =
3
7
.6 Hz, 2H, CH CH ), 2.11 (s, 3H, COCH3), 1.22 (t, J(H,H) = 7.6 Hz, 3H, CH CH ).
2
3
2
3
1
3
C NMR (126 MHz, CDCl ): δ = 170.79 (C=O), 153.04 (C-2), 149.35 (C-6), 138.70 (C-5),
3
1
36.11 (C-4), 121.89 (C-3), 66.98 (CH -OAc), 25.89 (CH CH ), 21.01 (COCH ), 15.36
2
2
3
3
(
CH CH ).
2 3
2
.2.5. 5-Ethylpyridin-2-yl-methanol
As depicted in step e of Scheme 1, 5-Ethylpyridin-2-yl-methyl acetate (24.29 g, 0.14 mol)
was dissolved in tetrahydrofuran (THF) and refluxed with an aqueous solution of NaOH (8.1
g, 0.20 mol) for 5 h. The reaction mixture was reboiled on activated charcoal and was
neutralized with glacial acetic acid and filtered. After concentration of the crude product
under reduced pressure, it was dissolved in a saturated solution of NaHCO , extracted with
3
diethyl ether and dried over sodium sulphate to give the product as a brown oil in 80% yield.
1
4
4
H NMR (500 MHz, CDCl ): δ = 8.32 (d, J(H,H) = 1.9 Hz, H-6), 7.48 (dd, J(H,H) =
3
3
3
2
.2 Hz, J(H,H) = 7.9 Hz, H-4), 7.20 (d, J(H,H) = 7.9 Hz, H-3), 4.70 (s, 2H, CH OH), 2.60
2
3
3
13
(
q, J(H,H) = 7.6 Hz, 2H, CH CH ), 1.20 (t, J(H,H) = 7.6 Hz, 3H, CH CH ). C NMR
2 3 2 3
(
126 MHz, CDCl ): δ = 157.10 (C-5), 148.12 (C-6), 137.99 (C-2), 136.34 (C-4), 120.55 (C-
3
3
), 64.29 (CH OH), 25.81 (CH CH ), 15.40 (CH CH ).
2
2
3
2
3
2
.2.6. 5-Ethylpicolinaldehyde
8

ACCEPTED MANUSCRIPT
As depicted in step f of Scheme 1, MnO was prepared freshly by simultaneously adding a
2
solution of MnSO hydrate (14.5 g, 0.09 mol in 25mL water) and NaOH (7.6 g, 2.21 mol in
4
2
0 mL water) to a warm solution of KMnO (15.5 g, 0.10 mol in 100 mL water), refluxing
4
the mixture for 6 h, filtered and dried under reduced pressure. 5-Ethylpyridin-2-yl-methanol
12.82 g, 93.4 mmol) was dissolved in chloroform and given to the dried MnO (16.24 g,
(
2
1
86.8 mmol). The mixture was refluxed in an ultrasonic bath for 5 days, filtered and purified
with column chromatography using a mixture petroleum ether and ethyl acetate (4:1) on
1
silica gel to give the product as a bright yellow liquid in 61% yield. H NMR (500 MHz,
4
3
CDCl ): δ = 10.03 (s, 1H, CHO), 8.60 (d, J(H,H) = 1.9 Hz, H-6), 7.87 (d, J(H,H) = 8.0 Hz,
3
3
3
H-3), 7.70 – 7.63 (m, H-4), 2.73 (q, J(H,H) = 7.6 Hz, 2H, CH CH ), 1.28 (t, J(H,H) =
2
3
13
7
.6 Hz, 3H, CH CH ). C NMR (126 MHz, CDCl ): δ = 193.30 (CHO), 151.06 (C-2), 150.18
2 3 3
(
C-6), 144.67 (C-5), 136.28 (C-4), 121.72 (C-3), 26.43 (CH CH ), 15.02 (CH CH ).
2 3 2 3
2
.2.7. Pyrimidinylhydrazones: 1, 2 and 3
The three Schiff bases 1, 2 and 3 were obtained by adding an ethanolic solution of the
hydrazinyl component to the solution of the keto component in equimolar amounts (step g of
Scheme 1), refluxing the mixture in the presence of catalytic amounts of acetic acid and
recrystallization from aqueous ethanol.
2
.2.7.1. 4’-Methoxy-6’-(2-(pyridin-2-ylmethylene)hydrazinyl)pyrimidine (1)
The reaction of 2-pyridinecarboxaldehyde (191.1 mg, 1.78 mmol) with 4-hydrazinyl-6-
methoxypyrimidine (250 mg, 1.78 mmol) resulted in bright rosy crystals of the product in
1
7
2% yield (293 mg, 1.28 mmol). H NMR (300 MHz, CDCl ): δ = 8.78 (bs, NH), 8.60 (d,
3
3
5
3
J(H,H) = 4.8 Hz, H-6), 8.38 (d, J(H,H) = 0.6 Hz, H-2’), 8.00 (d, J(H,H) = 8.0 Hz, H-3),
3
5
7
.93 (s, 1H, C=NH), 7.75 (t, J(H,H) = 7.3 Hz, H-4), 6.63 (d, J(H,H) = 0.7 Hz, H-5’), 3.99
9

ACCEPTED MANUSCRIPT
1
3
(
s, 3H, OCH ). C NMR (126 MHz, DMSO-d ): δ = 170.00 (C-4’), 162.78 (C-6’), 157.77
3 6
(
C-2), 153.47 (C-2), 149.28 (C-6), 142.77 (HC=N), 136.59 (C-4), 123.62 (C-5), 119.44 (C-3),
5.38 (C-5’), 53.54 (OCH ). ESI-MS: m/z Calcd. for [M]: 229.0964. Found: 230.0979 for
8
3
+
+
[
M+H] , 252.0798 for [M+Na] . Anal. Calcd. for C H N O×0.75 H O: C, 54.43; H, 5.19;
11 11 5 2
N, 28.85. Found: C, 54.36; H, 4.49; N, 28.94.
2
.2.7.2. 4’-[2-[(5-Ethylpyridin-2-yl)methylidene]hydrazin-1-yl]-6’-methoxypyrimidine (2)
The reaction of 5-ethylpicolinaldehyde (241.1 mg, 1.78 mmol) with 4-hydrazinyl-6-
methoxypyrimidine (250 mg, 1.78 mmol) resulted in bright rosy crystals of the product in
1
6
3% yield (288.7 mg, 1.12 mmol). H NMR (500 MHz, DMSO-d ): δ = 11.45 (s, 1H, NH),
6
4
5
8
.43 (d, J(H,H) = 1.6 Hz, H-6), 8.33 (d, J(H,H) = 0.9 Hz, H-2’), 8.12 (s, C=NH), 7.93 (d,
3
4
3
5
J(H,H) = 8.1 Hz, H-3), 7.68 (dd, J(H,H) = 2.1 Hz, J(H,H) = 8.2 Hz, H-4), 6.49 (d, J(H,H)
3
=
0.6 Hz, H-5’), 3.89 (s, 3H, OCH ), 2.64 (q, J(H,H) = 7.6 Hz, 2H, CH CH ), 1.20 (t,
3
2
3
3
13
J(H,H) = 7.6 Hz, 3H, CH CH ). C NMR (126 MHz, DMSO-d ): δ = 169.96 (C-6’), 162.78
2
3
6
(C-4’), 157.74 (C-2’), 151.22 (C-2), 148.73 (C-6), 142.91 (HC=N), 139.16 (C-5), 135.87 (C-
4
), 119.11(C-3), 85.26 (C-5’), 53.50 (OCH ), 25.15 (CH CH ), 15.08 (CH CH ).ESI-MS:
3
2
3
2
3
+
+
m/z Calcd. for [M]: 257.1277. Found: 258.1272 for [M+H] , 280.1074 for [M+Na] . Anal.
Calcd. for C H N O: C, 60.69; H, 5.88; N, 27.22. Found: C, 60.32; H, 5.76; N, 27.28.
1
3
15
5
2
.2.7.3. 4’-Methoxy-6’-(2-(1-(pyridin-2-yl)ethylidene)hydrazinyl)pyrimidine (3, VP035)
The reaction of 2-acetylpyridine (0.2 mL, 1.78 mmol) with 4-hydrazinyl-6-
methoxypyrimidine (250 mg, 1.78 mmol) resulted in bright rosy crystals of the product in
1
7
8
0
9% yield (342 mg, 1.41 mmol). H NMR (500 MHz, DMSO-d ): δ = 10.43 (s, 1H, NH),
6
5
4
3
5
.57 (ddd, J(H,H) = 0.9 Hz, J(H,H) = 1.7 Hz, J(H,H) = 4.8 Hz, H-6), 8.37 (d, J(H,H) =
5
3
.9 Hz, H-2’), 8.15 (dt, J(H,H) = 0.9 Hz, J(H,H) = 8.1 Hz, H-3), 7.85 – 7.75 (m, H-4), 7.35
1
0

ACCEPTED MANUSCRIPT
4
3
3
5
(
ddd, J(H,H) = 1.1 Hz, J(H,H) = 4.8 Hz, J(H,H) = 7.4 Hz, 1H, H-5), 6.57 (d, J(H,H) =
13
0
.9 Hz, 1H, H-5’), 3.89 (s, 3H, OCH ), 2.40 (s, 3H, C=NCH ). C NMR (126 MHz, DMSO-
3 3
d6): δ = 170.04 (C-4’), 163.51(C-6’), 157.53 (C-2’), 155.24 (C-2), 148.42 (C-6), 147.91
(
C=N), 136.44 (C-4), 123.40 (C-5), 119.93 (C-3), 86.10 (C-5’), 53.50 (OCH ), 11.76
3
+
(
C=NCH ). ESI-MS: m/z Calcd. for [M]: 243.1120. Found: 244.1128 for [M+H] , 266.0953
3
+
for [M+Na] . Anal. Calcd. for C H N O×0.5 H O: C, 57.13; H, 5.59; N, 27.76. Found: C,
12
13
5
2
5
7.23; H, 5.46; N, 27.09.
2
.3. Determination of relative lipophilicity by HPLC
HPLC-MSMS measurements were run on an AB Sciex 3200 QTrap tandem mass
spectrometer equipped with TurboV ion source. The injected mixture (5 µL) contained 1, 2, 3
and uracil at a concentration of about 1 µg/mL. The mass spectrometer was scanned in MRM
mode; the Q1/Q3 transitions were as follows: (1) 230/105 and 230/93; (2) 258/133 and
2
58/121; (3) 244/227 and 244/186; (uracil) 113/96 and 113/70. The source parameters were
as follows: dwell time: 200 msec; collision energy: 25 eV; IS (spray voltage): 5000 V; source
temperature: 450°C; curtain gas: 25 instrument unit; GS1: 40 instrument unit; GS2: 40
instrument unit. A Perkin Elmer Series 200 microLC system (consisting of a binary pump, an
autosampler and a column oven) was used for separation. Gemini C18 110A 150 x 4.6 mm, 5
µm was used at 40°C. The mobile phases were: A: 20 mM ammonium formate at pH = 10
adjusted by ammonia (pH = 10 was selected based on the log D versus pH curve. At pH = 10,
the compounds are present in neutral form, see 3.3); B: methanol. Flow rate was 1 mL/min.
The retention times of the compounds were measured at 45, 50, 55, 60 and 65% methanol for
the accurate determination of log k (retention factor). Data acquisition was stopped after the
elution of the last sample. A 15 minute equilibration step was used prior to the start of the
runs at each solvent composition.
1
1

ACCEPTED MANUSCRIPT
The reduced retention time was calculated using the following equation: k = (t -t)/t . Uracil
0
0
was used for the exact determination of t The retention time of uracil was 2.2 min at each
0
.
solvent composition. The log k values of the different eluent composition runs were
expressed as the function of the methanol content of the eluent, and extrapolated to pure
water for the determination of log Kw.
2
.4. UV‒Vis spectrophotometry
The pH-metric measurements for determination of the exact concentrations of HCl and KOH
o
stock solutions used for the spectrophotometric titrations were carried out at 25.0 ± 0.1 C in
aqueous solutions containing 0.5% (v/v) DMSO and at an ionic strength of 0.20 M KCl in
order to keep the activity coefficients constant. All the titrations were performed with
carbonate-free KOH solutions of known concentration (0.10 M) in the presence of 0.1 M
KCl. An Orion 710A pH-meter equipped with a Metrohm combined electrode (type
6
.0234.100) and a Metrohm 665 Dosimat burette were used for the pH-metric titrations. The
+
electrode system was calibrated to the pH = -log[H ] scale by means of blank titrations
(strong acid vs. strong base; HCl vs. KOH) according to the method suggested by Irving et al.
o
[
29]. The average water ionization constant, pKwater, is 13.76 ± 0.01 at 25 C, which
corresponds well to the literature data [30]. Samples were deoxygenated by bubbling purified
argon for ca. 10 min prior to the measurements and argon was also passed over the solutions
during further titrations.
A Hewlett Packard 8452A diode array spectrophotometer was used to record the
UV‒Vis spectra in the 200 - 800 nm interval. The path length was 1 cm. The
spectrophotometric titrations were performed on samples of ligand 3 alone or with Fe(III) or
Cu(II). The concentration of the ligand was usually 0.05 mM and the metal-to-ligand ratios
1
2

ACCEPTED MANUSCRIPT
were 1:1 and 1:2 over the pH range between 2 and 11.5 at an ionic strength of 0.20 M (KCl)
o
at 25.0 ± 0.1 C. The initial volume of the samples was 10.00 mL. Measurements for 1:1
ligand‐to-Fe(III) system were also carried out by preparing individual samples in which KCl
was partially or completely replaced by HCl. pH values, varying in the range ca. 0.7-2.0,
were calculated from the HCl content.
Protonation constants of ligand 3, the stability constants of its metal complexes and the
individual spectra of the species were calculated by the computer program PSEQUAD [31]
and literature data were used for Fe(III) hydroxido species [32].  (M L H ) is defined for the
p
q
r
p
q
r
general equilibrium pM + qL + rH
M L H as (M L H ) = [M L H ]/[M] [L] [H]
p
q
r
p
q
r
p
q
r
where M denotes the metal ion and L the completely deprotonated ligand. The calculations
were always made from the experimental titration data measured in the absence of any
precipitate in the solution.
2
.5. Electrochemical Studies
Cyclic voltammograms of the Fe(III) and Cu(II) complexes were determined at 25.0 ± 0.1 °C
in 90% (v/v) DMSO-water solution containing 1.0 mM metal ion or 1 mM ligand 3 while the
metal-to-ligand ratio was varied between 1:1-1:3. Ionic strength was 10 mM (TBACl).
Measurements were performed on a conventional three-electrode system under argon
atmosphere and a PC controlled Electrochemical Measurement System (EF 451). Samples
were purged for 15 min with argon before recording the cyclic voltammograms. A glassy
carbon electrode was used the working electrode, a platinum electrode as the auxiliary
electrode and Ag/AgCl/KCl (1 M) as reference electrode.
Electrochemical potentials were converted into the normal hydrogen electrode (NHE)
scale by adding 0.236 V [33]. The electrochemical system was calibrated with a solution of
1
3

ACCEPTED MANUSCRIPT
ferrocene in 90 % (v/v) DMSO-water (E1/2 = +0.428 ± 0.012 V vs. Ag/AgCl/KCl (1 M) in our
setup). Redox potentials were obtained at 100 mV/s scan rate in the range of –1.2 to +1.0 V.
2
.6. Oxidation of ascorbic acid (ASC) and glutathione (GSH)
Oxidation of ASC and GSH by the copper(II) and iron(III) complexes of ligand 3 was studied
by UV‒Vis spectrophotometry using Hewlett Packard 8452A diode array
a
spectrophotometer. A special, tightly closed tandem cuvette (Hellma Tandem Cell, 238-QS,
path length: 1 cm) was used and the reactants were separated until the reaction was triggered.
Both isolated pockets of the cuvette were completely deoxygenated by bubbling a stream of
argon for 10 min before mixing the reactants. Spectra were recorded before and immediately
after mixing, and changes were followed for maximum 90 min. One of the isolated pockets
contained the reducing agent (ASC or GSH) at ten times higher concentrations than that of
the metal ions. The other pocket contained the metal ions (Cu(II): 25 M; Fe(III): 50 M)
and ligand 3 at 1:1 metal-to-ligand ratio. The pH of the solutions was adjusted to 7.40 (copper
complexes) or 5.45 (iron complexes) by 50 mM HEPES or MES buffers, respectively, and an
ionic strength of 0.1 M (KCl) was applied at 25.0 ± 0.1 °C. At the end of the measurements
cc. H O (50 L / 1.4 mL) was added to the samples.
2
2
2
.7. Establishment of transgenic ABCB1 expressing Madin-Darby canine kidney II (MDCKII)
cell line
MDCKII-ABCB1 cell line stably expressing the human wild-type MDR1/ABCB1 was
established by the Sleeping Beauty transposon-based gene delivery system, using the 100x
hyperactive SB transposase [34]. MDCKII canine kidney cells were cotransfected with the
1
4

ACCEPTED MANUSCRIPT
SB transposon vector containing the wild-type human MDR1 cDNA [35] and the SB
transposase vector construct, using the Lipofectamine 2000 reagent (Life Technologies) in
5
accordance with the manufacturer’s instructions. Briefly, 3×10 cells were seeded in 6-well-
plates; 24 hours later the cells were transfected with 2 g vector DNA per well in a 10:1 ratio
for the SB transposon and transposase constructs. 48 hours after transfection transgene
positive cells were sorted by flow cytometry (FACS Aria High Speed Cell Sorter, Beckton-
Dickinson) based on the cell surface expression of wild-type and mutant MDR1/ABCB1.
Protein expression was measured by antibody labeling using the human MDR1/ABCB1
specific monoclonal antibody MRK16 (Abnova). To obtain homogenous transgene positive
cell population, sorting procedure was repeated 4 times in 2-week intervals.
2
.8. Vector constructs
Sleeping Beauty transposon vector carrying the wild-type human MDR1 cDNA were
constructed as follows. The Sleeping Beauty (SB) transposon vector containing the cDNAs of
EGFP and the puromycin resistance gene (PURO) in separate transcription cassettes, each
driven by CAG promoter, was kindly provided by Dr. Tamás István Orbán (Hungarian
Academy of Sciences, Budapest, Hungary) [36]. Full-length wild-type MDR1 cDNA was
amplified by PCR from pAcUW-LMDR1 vector [37] with the following primers: 5’-
TAGAATACCGGTAGGTCGGAATGGATCTTGAA-3’
and
5’-
AGTGATGGATCCAACATCTCATACAGTCAGAG-3’ containing AgeI and BamHI
restriction sites, respectively. The digested PCR product was cloned into the SB-EGFP-
PURO transposon vector between the AgeI and BclI restriction sites, replacing EGFP,
resulting in SB-MDR1-PURO vector.
1
5

ACCEPTED MANUSCRIPT
2
.9. Cell culture
The human ovarian carcinoma cell lines A2780 and the doxorubicin selected multidrug
resistant counterpart A2780adr were obtained from ECACC, UK, (A2780: No. 93112519,
A2780adr: No. 93112520), and cultivated in RPMI (Sigma Aldrich, Germany) supplemented
with 10% fetal bovine serum and 50 unit/mL penicillin and streptomycin (Sigma Aldrich,
Germany). The human uterine sarcoma cell lines MES-SA and the doxorubicin selected
MES-SA-Dx5 were obtained from ATCC (MES-SA: No. CRL-1976™, MES-SA-Dx5: No.
CRL-1977™) and cultivated in Dulbecco’s Modified Eagle Medium (DMEM, Sigma
Aldrich, Hungary). The human cervix carcinoma cell line KB-3-1 and the vinblastine selected
KB-v1 (kind gifts from Dr. Michael M. Gottesman, National Institutes of Health) were
cultivated in DMEM. The phenotype of the resistant cells was verified using cytotoxicity
assays (not shown). MDCK II and the Pgp transfected MDCK-B1 were cultured in DMEM.
DMEM media (Sigma Aldrich, Hungary) were supplemented with 10% fetal bovine serum,
5
mmol/L glutamine, and 50 unit/mL penicillin and streptomycin (Life Technologies). All
cell lines were cultivated at 37 °C, 5% CO2.
2
.10. MTT viability assay
MTT viability assays were performed as described earlier with minor modifications [38,39].
Briefly, cells were seeded into 96-well tissue culture plates (Sarstedt, Newton, USA / Orange,
Braine-l'Alleud, Belgium) in the appropriate density evaluated for each cell line (5000 cells
per well for MES-SA-Dx5 and KB-3-1/v1 cells, 10000 cells per well for A2780 cells) and
allowed to attach for 12 h. Test compounds were added to achieve the required final
concentration in a final volume of 200 µL per well. After an incubation period of 72 h, the
1
6

ACCEPTED MANUSCRIPT
supernatant was removed and fresh medium supplemented with the MTT reagent
(0.083 mg/mL) was added. Incubation with MTT at 37 °C was terminated after 1 h by
removing the supernatants and lysing the cells with 100 µL DMSO per well. Viability of the
cells was measured spectrophotometrically by absorbance at 540 nm using either a Perkin
Elmer Victor X3 or a BMG POLARstar microplate reader. Data was background corrected
by subtraction of the signal obtained from unstained cell lysates and normalized to untreated
cells. Curves were fitted by Prism software [40] using the sigmoidal dose-response model
(comparing variable and fixed slopes). Curve fit statistics were used to determine the
concentration of test compound that resulted in 50% toxicity (IC ).
5
0
2
.11. ROS determination using DCFDA
Measurements were performed as reported in the literature with modifications [41–43]. After
harvesting and washing with phosphate buffered saline (PBS), cells were incubated with 10
µM DCFDA in a water bath shaker at 37 °C for 30 min in a density of 3 Mio cells/mL. After
washing with PBS cells were seeded to 96 well plates in PBS in a density of 20000 cells/well.
Basal fluorescence was measured, after addition of the test compounds in different
concentrations; samples were measured in time intervals of 10 min. DCFDA solution in
buffer was used as a cell free control to test for interaction of the test compounds with
DCFDA. Data was analysed as fold change of fluorescence compared to the basal levels.
3
. Results and discussion
3
.1. Synthesis
1
7

ACCEPTED MANUSCRIPT
As depicted in Scheme 1, the pyrimidinylhydrazones 1 to 3 were prepared by reacting 4-
hydrazinyl-6-methoxypyrimidine (derived from 4,6-dichloropyrimidine) with equimolar
amounts of the respective keto component under mild acidic conditions. While compound 1
serves as the non-alkylated core structure, 2 possesses an ethyl group as substituent in the
aromatic ring and 3 has a methylgroup at the imino carbon. 5-Ethylpicolinaldehyde used for
the synthesis of 2 was prepared following the method patented by Seydel et al. in a slightly
modified manner [15]. After formation of the N-oxide from 5-ethyl-2-methyl-pyridine using
hydrogen peroxide in glacial acetic acid, followed by alkaline reconditioning, an acetoxy
group was introduced using acetic anhydride, glacial acetic acid to provide 5-ethylpyridin-2-
yl-methyl acetate. The cleavage of the ester to the corresponding alcohol was carried out in
THF, as the suggested conditions using ethanol as a solvent provided the ethyl ether instead.
The oxidation to the desired aldehyde required a long time; and reflux with freshly prepared
MnO under ultrasound was found to be an efficient oxidizing method. Several other methods
2
including commercially available MnO [15], the application of SeO and tert-butyl
2
2
hydroperoxide [44], or the attempt to directly oxidize the 5-ethyl-2-methyl-pyridine [45–47]
were not successful. With the help of the modifications in the first and in the last steps of the
original approach, the yield could be increased from the reported 18% to 40%.
Physico-chemical properties are known to influence pharmacokinetics and
pharmacodynamics; therefore we were interested in the lipophilicity of the three compounds.
Since the lipophilicity of the compounds was too high to allow the exact determination of
logD7.4 values by the traditional shake flask method in n-octanol/buffered aqueous solution
(
1:1), logD values were estimated using MarvinSketch [48] and the relative lipophilicity of
7.4
the compounds was determined using HPLC. This method is commonly used for the
1
8

ACCEPTED MANUSCRIPT
determination of log K (log P) [49]. We used an isocratic method to measure the retention
w
times of the neutral forms of the compounds as a function of the organic composition of the
eluent. Results of five different eluent compositions (summarized in Supplementary Table
A1) were used for the accurate determination of the log k values as described in the Materials
and Methods section. Since in this setup the retention coefficients of the three analytes show
linear correlation with the volume fractions of the organic solvent in the binary eluent, the
relative lipophilicity of the compounds was determined using the linear solvent strength
model ([49]). The resulting log K values are shown in Table 1.
w
3
.2. Characterizing the in vitro antitproliferative activity of compounds 1 to 3
The cytotoxic potency of pyrimidinylhydrazones 1 to 3 was investigated in a diverse panel of
cancer cell lines including A2780 (ovarian carcinoma cells), KB-3-1 (cervix carcinoma) and
MES-SA (uterine sarcoma). To test the ability of the newly synthesized compounds to
overcome drug resistance, resistant derivatives of the cell lines (A2780adr, KB-v1 and MES-
SA- Dx5) were also included in the study. Resistance of these cell lines is primarily mediated
by the overexpression of Pgp, which effluxes structurally and mechanistically unrelated
cytotoxic compounds from the cells [50–52]. To clearly delineate the effect of Pgp on the
toxicity of the compounds, we included control MDCK-II and its drug resistant derivative
MDCK-II-ABCB1 engineered to overexpress cDNA-derived Pgp.
As summarized in Table 2, the three tested compounds proved to be toxic in all investigated
cell lines (with IC values of 0.03 ‒ 6.30 M) regardless of their resistance status towards
5
0
common cytostatic agents. Toxicity of the compounds does not correlate with lipohilicity
Table 1). Alkylation of 1 resulted in increased toxicity in all eight investigated cell lines.
(
1
9

ACCEPTED MANUSCRIPT
Introduction of a methyl at the imino carbon, as in 3 increases toxicity to a higher extent as
compared to the introduction of an ethyl group at the aromatic ring as in compound 2.
3
.3. Proton dissociation processes of ligand 3
Since the physico-chemical properties of a potential chemotherapeutic agent have a basic
influence on its pharmacokinetics and metal binding abilities, a detailed study on the proton
dissociation processes of the most promising compound 3 was performed. For this purpose
UV‒Vis spectrophotometric titrations were applied at a concentration of 50 M (in the
presence of 0.5% (v/v) DMSO). The recorded pH-dependent spectra (Fig. 1a) revealed two
proton dissociation processes at pH < 9; the calculated protonation constants (log(H L)) and
i
pK values are reported in Table 3. The spectra for the individual ligand species in different
a
2
+
+
protonation forms (H L , HL and L) were calculated by the deconvolution of UV-Vis
2
spectra recorded at different pH values (Fig. 1b). Titrations showed characteristic spectral
changes in the wavelength range 250 – 450 nm. Ligand 3 displays an intense absorption band
2
+
+
in all the protonation forms at 344 nm (H L ), 356 nm (HL ) and 312 nm (L), and less
2
2
+
+
intense bands at 276 nm (H L ) and 282 nm (HL ), which are assigned to n → *
2
transitions. The determined pK values demonstrate that the proton dissociation processes
a
overlap and the neutral form L predominates in the physiological pH range as illustrated in
Fig. 1c. pK can most presumably be attributed to the deprotonation of the pyrimidinium unit,
1
while pK belongs to the deprotonation of the pyridinium-N as shown in Scheme 2.
2
Deprotonation of the pyridinium moiety in the case of the -N-pyridyl thiosemicarbazones
such as the drug Triapine was also shown to result in similar spectral changes, namely a
significant shift of max value from 402 nm to 368 nm [53]. The possible proton dissociation
2
0

ACCEPTED MANUSCRIPT
of the hydrazinic =N-NH group of ligand 3 is likely to take place at higher pH values (pH >
1
1) and therefore its pK could not be determined accurately with the applied method.
a
3
.4. Complex formation processes of ligand 3 with Fe(III) and Cu(II)
The biological activity of metal binding compounds is often modified upon complex
formation with metal ions [18,54]. Characterization of metal binding ability (solution
speciation), and the most plausible chemical species formed in aqueous solution in the
biologically relevant pH range is a mandatory prerequisite for understanding the mechanism
of action of complexes. To this end, we determined the stoichiometry and stability of ligand 3
complexes formed with Fe(III) and Cu(II) in aqueous solutions using UV-Vis
spectrophotometry. The stoichiometries of the metal complexes and the overall stability
constants furnishing the best fits to the experimental data are listed in Table 4. The
appearance of well-defined charge transfer (CT) bands in the visible wavelength range of pH-
dependent UV‒Vis spectra unequivocally showed complex formation with both metal ions
(Figs. 2 and 3). These CT bands overlap considerably with the ligand bands, while the d-d
bands are too weak to be detected under the concentrations conditions used (cM = 25 -
3
+
3+
5
0 M). In the case of Fe(III), formation of the expected complexes [FeL] and [FeL2]
could be verified. In these complexes the ligand coordinates most probably via
3
+
(
Npyridyl,Nhydrazone,Npyrimidyl) donor sets. It is noteworthy that [FeL2] is the predominant
species at 1:2 metal-to-ligand ratio in the physiologically relevant pH range (see inset of Fig.
), although the hydrolysis of the metal ion suppresses the complex formation at pH > 8.5.
2
Surprisingly, the development of CT bands that are characteristic for Fe(II) complexes
formed with ligands consisting of aromatic nitrogen donor atoms was also detected at pH >
2
1

ACCEPTED MANUSCRIPT
~
8.5 in the 510–610 nm wavelength range. Formation of minor amounts of iron(II)
complexes is possible at lower pH values as well. The redox process of the complex seems to
be pH-dependent and much faster in the basic pH range. Thus data collected at pH > 7.5 were
not evaluated.
In order to compare the iron binding ability of ligand 3 with that of other well-known
iron chelators, the pM value (= -log [Fe(III)]; c /c =10; c =1 M) was calculated at
L
M
M
physiological pH, using the experimentally determined stability constants. For Fe(III) a pM
value of 17.6 was obtained, which is lower than those reported for the naturally occurring
iron binders such as human serum transferrin (20.3) [55], desferrioxamine B (26.5) [56] or
the chelator Deferiprone (19.3) used to treat thalassaemia major [57], but is higher than the
pM value of the thiosemicarbazone drug Triapine (15.7) [58]. Since higher pM values imply
that metal complex formation is more favored, these results suggest that the studied
pyrimidinylhydrazone forms less stable complexes compared to the aforementioned chelators
containing mainly oxygen donor atoms. However, the Fe(III) binding ability of ligand 3 can
-
exceed that of the tridentate -N-pyridyl thiosemicarbazones with a (Npyridyl,N,S ) binding
mode at pH 7.4.
In the case of the Cu(II) ion, the formation of mono-ligand complexes was expected
due to the usual square planar geometry of the Cu(II) coordination sphere, although the
2+
speciation model was significantly improved by the involvement of bis-ligand [CuL2]
species (see Fig. A1). Deconvolution of the UV‒Vis spectra recorded in the Cu(II) – ligand 3
system yielded the formation constants summarized in Table 4; individual spectra of the
3
+
2+
2+
[
CuLH] , [CuL] and [CuL ] complexes are shown in Figs. 3b and A2. Formation of the
2
bis-ligand Cu(II) complex is not as pronounced as in the case of Fe(III), which is reflected in
the fact that its molar fraction is only 0.52 at pH 7.4 (c /c =2; c = 50 M), while the
L
M
L
3
+
corresponding value is 0.80 in the case of [FeL2] under the same conditions. In the
2
2

ACCEPTED MANUSCRIPT
3
+
protonated complex [CuLH] the ligand coordinates most probably in a bidentate manner
2
+
(
i.e. (Npyridyl,Nhydrazone); in [CuL] via the tridentate mode (Npyridyl,Nhydrazone,Npyrimidyl) was
suggested for pyridylhydrazones, which comprise similar binding moieties [59]. In case of
2
+
the [CuL2] species a distorted octahedral binding mode or a bidentate coordination of the
ligands may be possible, which is difficult to ascertain without EPR studies.
In order to compare the Cu(II) and Fe(III) binding ability of ligand 3, the pM value
was also computed on the basis of the stability data (conditions: pH = 7.40; c /c = 10 µM;
L
M
cM = 1 M) and a value of 8.6 was obtained for Cu(II). Since Fe(III) tends to hydrolyze at
physiological pH, its hydroxido species were also taken into account and a modified pM
value (pM*) was determined for Fe(III), by calculating the negative logarithm of the summed
equilibrium concentrations of the free metal ion and its hydroxido species (pM* = -
log ([Fe] + [Fe (OH) ]). pM* for Fe(III) was computed to be 9.5, which is almost one order
p
r
of magnitude higher than the pM of Cu(II). At an acidic pH (5.45) pM* can be considered
equal to pM, due to the lower extent of hydrolysis, therefore values of 8.5 and 13.3 were
computed for the Cu(II) and Fe(III) complexes, respectively. These results indicate that
whereas ligand 3 forms stable complexes with both Fe(III) and Cu(II), the unbound metal
fraction is somewhat higher in the case of Cu(II) showing the stronger Fe(III) binding ability
of ligand 3 under physiological conditions and in the slightly acidic pH range as well.
3
.5. Effect of the metal ions on the toxicity of ligand 3
Having shown the strong metal binding ability of ligand 3 to Fe(III) and Cu(II), experiments
were designed to evaluate the impact of metal ions on its toxicity. Ligand 3 was added to the
parental MES-SA cells and its resistant derivative MES-SA-Dx5 with increasing
2
3

ACCEPTED MANUSCRIPT
concentrations of metal ions. The metal salts alone had no impact on cell viability over a
three day long incubation period; therefore additive effects could be excluded. In MES-SA
cells, coincubation of 3 with Fe(III) was found to reduce the ligand’s cytotoxicity, while the
presence of Cu(II) had an opposite effect (Fig.4). Interestingly, the effects of both metal ions,
namely the rescuing by iron and the sensitization by copper was found to be significantly
attenuated in MES-SA-Dx5 cell line, the drug resistant derivative of MES-SA.
Reduced activity of the free ligand in the presence of the iron indicates that the toxicity may –
at least in part – be explained by the depletion of cellular iron levels. The differential
sensitivity of the drug resistant subline is reminiscent of the collateral sensitivity of MDR
cells [60]. The mechanisms underlying the paradoxical hypersensitivity of resistant cells are
not known. It has been suggested, but never shown, that the metal homeostasis of MDR cells
is altered as a result of Pgp function. Lack of the protective effect of Fe(III) in the MDR
subline suggests that these cells cannot be rescued by excess iron – perhaps due to the
imbalanced iron homeostasis of MDR cells.
In follow up experiments the Cu(II) and Fe(III) complexes of ligand 3 were prepared in situ
by mixing the ligand with half- and one equimolar concentrations of the metal ions. Using
this setup a constant molar ratio was ensured throughout the investigated concentration range
(1 nM – 250 µM). Results obtained in the presence of increasing metal ion concentrations at a
fixed ligand-to-metal molar ratio confirmed the reduced toxicity of Fe(III) complexes (not
shown), while the effect of Cu(II) was less pronounced in that setup as compared to the
coincubation (Fig. 7).
3
.6. Cyclic voltammetric studies of Fe(III) and Cu(II) complexes of ligand 3
2
4

ACCEPTED MANUSCRIPT
Besides depleting metal ions, the mechanism of toxicity of chelating compounds also relies
on redox active complexes that may be formed in the Fenton reaction [1,7,41]. To examine
the redox properties of the iron and copper complexes of ligand 3, we performed cyclic
voltammetry (CV) studies and the measured formal potentials were compared to those of
Triapine and physiological reductants. Illustrative voltammograms are shown in Fig. 5.
Measurements were performed in 90% (v/v) DMSO/H O in the potential range of ‒0.9 to
2
+1.0 V against Ag/AgCl/KCl (1 M). Ligand 3 alone was found to be redox-innocent in this
range. It should be noted that the ligand alone shows an irreversible process resulting in a
cathodic peak seen at ~ ‒1.01 V vs. Ag/AgCl/KCl (1 M) (which corresponds to ‒0.78 vs.
NHE, Fig. A3). The complexes of both metal ions showed redox activity. The redox potential
of the formed complexes is largely controlled by the donor atoms of the ligand that are
present in the first coordination shell. The addition of the chelating agent significantly
changed the peak positions in comparison to those of the aqua complexes. The detected redox
processes for the Fe(III)/Fe(II) and Cu(II)/Cu(I) couples were found to be quasi reversible
under the applied conditions in all cases. (For a detailed analysis of CV data see Table A2.)
Addition of ligand 3 to Fe(III) results in the formation of a metal complex and the
appearance of a new set of peaks in the voltammograms at ~ +0.68 V vs. NHE as calculated
from the measured data represented in Fig. 5a. Upon complexation the formal potential
increased by 0.31 V; the intensity of the aqua ion peaks decreased, while that of the metal
complex increased with increasing ligand excess. The positive shift of the formal potentials
due to the coordination by the ligand via (Npyridyl,Nhydrazone,Npyrimidyl) donor set indicates a
preference for the lower oxidation state, as in the case of other well-known ligands consisting
of aromatic nitrogen donor atoms such as 2,2′-bipyridyl or 1,10-phenanthroline. This can be a
possible explanation for the reduction of Fe(III) complexes in the basic pH range (vide supra).
2
5

ACCEPTED MANUSCRIPT
The formal potential of the complexed Fe(III)/Fe(II) redox couple is significantly higher (~
+0.69 vs. NHE) than that of other iron chelators such as Triapine (+0.07 V vs. NHE) [58] or
tris-hydroxamato complexes (-0.43 V) [61] resulting in an easier reduction by physiological
reductants.
Ligand 3 also influences the redox potential of the Cu(II)/Cu(I) couple. Addition of
the ligand diminished the intensity of the aqua complex peaks, while a new reduction peak
appeared upon complexation. In case of the copper complexes, the formal potential was
shifted by 0.25 V to the negative direction, indicating the ligand preference for the metal ion
in the higher, +2 oxidation state. The formal potential of the Cu(II)/Cu(I) in the case of ligand
3
(+0.36 V vs. NHE) is higher than that of the biological reductants (glutathione, ascorbate
1,62]), thus the reduction of this Cu(II) complex may lead to redox cycling and the
generation of ROS.
[
3
.7. Reduction of complexes formed with ligand 3 by intracellular reducing agents
In order to further investigate the potential of the copper and iron complexes formed with
ligand 3 to undergo intracellular redox reactions, the ability of biologically relevant reducing
agents such as ascorbate (ASC) or glutathione (GSH) to reduce the complexes was tested in
in vitro assays. The redox processes were followed spectrophotometrically under strictly
anaerobic conditions via monitoring the overlapping ligand and CT bands in the wavelength
range of 300-600 nm. Kinetic runs were performed at pH 7.40 in the case of the copper
complexes, while pH 5.45 was chosen for the iron complexes in order to avoid metal ion
hydrolysis and the possible formation of the Fe(II) species in the Fe(III) - ligand 3 system in
the basic pH range (see Section 3.3). Oxidation of ASC yields dehydro-L-ascorbic acid; the
2
6

ACCEPTED MANUSCRIPT
oxidized form of GSH is glutathione disulfide (GSSG). Neither the reduced and oxidized
form of ASC nor GSH absorbs light at  > ~310 nm [63], thus spectral changes above this
wavelength are characteristic only to the absorption of the metal complexes. Oxidation of
ASC or GSH is often monitored at their max values (265 and 262 nm, respectively) [9,64],
but since the studied metal complexes, as well as the free ligand 3 absorb light strongly in the
UV range, the UV–Vis spectra were recorded between  ~ 300 nm and 600 nm upon addition
of ASC and are shown in Fig. 6. No time dependent spectral changes were observed (tmax
5 min) in the case of the Cu(II) complexes (Fig. 6a), which suggests that this metal complex
could not be reduced by ASC under the applied conditions.
We also monitored the spectral changes of the formed complexes upon reduction by GSH
Fig. 6c). Glutathione is a predominant intracellular reducing agent with a lower formal
=
7
(
potential compared to that of ASC. (The formal potential at pH 7.4 of the GSSG/GSH and the
dehydro-L-ascorbic acid/ASC redox pairs are -0.26 V and +0.05 V, respectively [62,65].) In
the case of copper a significant decrease of the absorbance was observed at the max value of
the Cu(II) complex (408 nm) during the redox reaction; while the absorbance value was
increased at the max of the free ligand (~310 nm), most probably as a result of the
decomposition of the generating instable Cu(I) complex. These results suggest that the
stronger reducing agent GSH is able to reduce the Cu(II) complex. Addition of H O
2
2
regenerated the Cu(II) complexes, suggesting that a reversible redox process can contribute to
the toxicity of the compound.
In line with the high redox potential of the Fe(III)-ligand 3 complex (see Section 3.6), this
complex was reduced by ASC (Fig. 6 b); yet the redox reaction was relatively slow. When
H O was added to the sample at the end of the kinetic run only negligible spectral changes
2
2
2
7

ACCEPTED MANUSCRIPT
were observed even after 0.5 h, suggesting that the iron complex is not likely to undergo
intracellular redox cycling.
Surprisingly, ten-fold excess of GSH resulted in almost no changes in the UV–Vis spectrum
of the Fe(III) - ligand 3 system (Fig. 6 b) within 90 min, suggesting that the kinetic of the
reduction is fairly slow. A similar trend was observed in the reduction kinetics of vanadium
complexes by the two reducing agents [63,64]. Furthermore, the formal potentials of both
reducing agents are pH-dependent, which might also explain the lack of GSH-induced
reduction of the iron complex at pH 5.45. Based on studies on the pH dependence of the
redox potentials, it can be estimated that at pH 5.45, the redox potential of GSH and ASC are
shifted by 116 mV and 52 mV, respectively [62,65,66]).
Taken together, our results indicate that the Cu(II) complex of ligand 3 has the potential to
undergo redox-cycling reactions under physiological conditions, whereas the iron complex is
probably not able to undergo intracellular redox cycling.
3
.8. Involvement of ROS in the toxicity of 3
To evaluate the role of ROS production in the toxicity of 3, the cytotoxicity experiments were
repeated in the presence of the antioxidant NAC. Coincubation with NAC resulted in a
significant increase of the IC values (Fig. 7), suggesting that ROS play a role in the toxicity
5
0
of ligand 3. Due to the relative redox potentials a direct reduction of compound 3 by NAC is
unlikely; therefore a more probable source of ROS induction is the intracellular formation of
redox active complexes undergoing Fenton-like chemistry. In the next set of experiments
Cu(II) complexes of ligand 3 were coincubated at 1:1 and 1:2 metal-to-ligand ratios with
NAC and cytotoxicity was measured in MES-SA and MES-SA-Dx5 cells. Since
2
8

ACCEPTED MANUSCRIPT
complexation with Fe(III) significantly attenuates the toxicity of ligand 3, the effect of NAC
on the toxicity of the iron complexes could not be reliably determined.
As expected, NAC diminished the toxicity of the Cu(II) complexes, suggesting that ROS
may, at least in part, be responsible for the toxicity of the ligand and its metal complexes
(Fig. 7c). Intriguingly, the (relative) protective effect of the antioxidant appeared to be more
pronounced in drug resistant cells as compared to their sensitive counterparts (compare
A2780adr, Fig. 7b to A2780, Fig. 7a, and dx5 to MES-SA (Fig 7c)).
To characterize the extent of ROS formation, intracellular ROS were measured using
the ROS-sensitive cell permeable dye DCFDA. MES-SA and MES-SA-Dx5 cells were
preincubated with DCFDA and the effect of ligand 3 and its complexes on fluorescence was
evaluated following an 80 minute-long incubation period. Fig. 8 summarizes the results
obtained with MES-SA cells (Dx5 cells yielded similar results; Fig. A4). At 50 µM, ligand 3
did not induce an increase in fluorescence. Complexation with iron resulted in ROS
3
+
3+
production in the case of both metal-to-ligand ratios used ([FeL] and [FeL ] ) (Fig. 8a).
2
2+
Addition of Cu(II) resulting in 1:1 metal-to-ligand ratio (mostly [CuL] , see Fig. A2)
2
+
induced the formation of ROS, while the [CuL ] complex was not redox active (Fig. 8a). In
2
all cases, ROS induced by the metal complexes could be scavenged by the addition of NAC.
A similar pattern was observed at 100 µM ligand concentration (Fig. 8b). Unexpectedly,
2
+
addition of the ROS scavenger NAC to 100 µM of the [CuL] complex increased, rather than
decreased ROS production. Since this effect could also be observed in the cell free condition,
we speculate that the paradoxical effect of NAC is explained by the ability of the complex to
participate in redox cycling as a result of the reduction by NAC (see also Figures 5b, 6c). The
Fenton-like reaction results in increased ROS production, but direct oxidation of DCFH may
also contribute to the observed signal [67,68]. Induction of ROS by NAC was only observed
at 100 µM ligand concentration, which is between 300 and 1000 times higher than the IC50
2
9