Ligand-centred redox activation of inert organoiridium anticancer catalysts

Article abstract of DOI:10.1039/d0sc00897d

Organometallic complexes with novel activation mechanisms are attractive anticancer drug candidates. Here, we show that half-sandwich iodido cyclopentadienyl iridium(iii) azopyridine complexes exhibit potent antiproliferative activity towards cancer cells, in most cases more potent than cisplatin. Despite their inertness towards aquation, these iodido complexes can undergo redox activation by attack of the abundant intracellular tripeptide glutathione (GSH) on the chelated azopyridine ligand to generate paramagnetic intermediates, and hydroxyl radicals, together with thiolate-bridged dinuclear iridium complexes, and liberate reduced hydrazopyridine ligand. DFT calculations provided insight into the mechanism of this activation. GS^?attack on the azo bond facilitates the substitution of iodide by GS^?, and leads to formation of GSSG and superoxide if O₂is present as an electron-acceptor, in a largely exergonic pathway. Reactions of these iodido complexes with GSH generateIr-SGcomplexes, which are catalysts for GSH oxidation. The complexes promoted elevated levels of reactive oxygen species (ROS) in human lung cancer cells. This remarkable ligand-centred activation mechanism coupled to redox reactions adds a new dimension to the design of organoiridium anticancer prodrugs.

Full text of DOI:10.1039/d0sc00897d

Chemical
Science
View Article Online
View Journal
EDGE ARTICLE
Ligand-centred redox activation of inert
organoiridium anticancer catalysts†
Cite this: DOI: 10.1039/d0sc00897d
a
a
a
All publication charges for this article
have been paid for by the Royal Society
of Chemistry
Wen-Ying Zhang,
Samya Banerjee,
George M. Hughes,
a
a
b
a
Hannah E. Bridgewater,
James P. C. Coverdale,
Ji-Inn Song,
Ben G. Breeze,
Guy J. Clarkson,
Fortuna Ponte,
a
a
c
Carlos Sanchez-Cano,
c
a
*
Emilia Sicilia
and Peter J. Sadler
Organometallic complexes with novel activation mechanisms are attractive anticancer drug candidates.
Here, we show that half-sandwich iodido cyclopentadienyl iridium(III) azopyridine complexes exhibit
potent antiproliferative activity towards cancer cells, in most cases more potent than cisplatin. Despite
their inertness towards aquation, these iodido complexes can undergo redox activation by attack of the
abundant intracellular tripeptide glutathione (GSH) on the chelated azopyridine ligand to generate
paramagnetic intermediates, and hydroxyl radicals, together with thiolate-bridged dinuclear iridium
complexes, and liberate reduced hydrazopyridine ligand. DFT calculations provided insight into the
mechanism of this activation. GSꢀ attack on the azo bond facilitates the substitution of iodide by GSꢀ,
and leads to formation of GSSG and superoxide if O2 is present as an electron-acceptor, in a largely
exergonic pathway. Reactions of these iodido complexes with GSH generate Ir-SG complexes, which are
catalysts for GSH oxidation. The complexes promoted elevated levels of reactive oxygen species (ROS) in
human lung cancer cells. This remarkable ligand-centred activation mechanism coupled to redox
reactions adds a new dimension to the design of organoiridium anticancer prodrugs.
Received 15th February 2020
Accepted 21st April 2020
DOI: 10.1039/d0sc00897d
rsc.li/chemical-science
a diselenobenzoquinone iridium complex which targets cyto-
chrome P450 reductase, exhibits comparable potency to
cisplatin.12
Introduction
Three key platinum drugs, cisplatin, carboplatin, and oxalipla-
tin are widely administered in worldwide cancer chemo-
therapy.1 However, platinum resistance and undesirable side
effects are now limiting their future use.2 Therefore, it is
important to discover other metal complexes with different
modes of action compared with platinum drugs.3,4 Third-row
5d6 iridium(III) complexes offer potential structural diversity of
octahedral coordination geometry, slow ligand exchange
kinetics, and facile synthesis.5–7 For example, kinetically-inert
octahedral iridium pyridocarbazole scaffolds can act as selec-
tive protein kinase inhibitors,8 and inert iridium polypyridine
anticancer complexes can have targets other than DNA.9 Also,
inert bis-cyclometalated iridium complexes are promising
photosensitizers for singlet oxygen production,10,11 and
Half-sandwich organometallic iridium cyclopentadienyl
complexes [(h5-CpX)Ir(L^L0)Z]n+/0 where CpX
Cp*
(pentamethyl-cyclopentadienyl), Cpxph (tetramethyl(phenyl)-
cyclopentadienyl), or (tetramethyl(biphenyl)-
Cpxbiph
,
¼
cyclopentadienyl), not only bind to DNA, but also target lyso-
somes and perturb the redox status of cells.13–15 The chelated
ligand L^L0, the ancillary leaving group Z and the p-bound CpX
ligand co-regulate the overall electronic structure and chemical
reactivity of such iridium complexes. Reported half-sandwich
iridium anticancer complexes with Z ¼ Cl are labile and most
likely activated in cancer cells through fast hydrolysis (minutes)
of the chlorido ligand.6 Extension of the CpX ring from Cp* to
Cpxph and Cpxbiph can slow down the hydrolysis rate, and
increase the extent of hydrolysis.16 Rapid hydrolysis sometimes
compromises anticancer activity due to rapid deactivation by
side reactions in advance of reaching targets. For example,
compared to the pyridine analogue [(h5-Cpxbiph)Ir(ppy)py]+ (ppy
¼ 2-phenylpyridine), the chlorido complex [(h5-Cpxbiph)Ir(ppy)
Cl] is more reactive, hydrolyzes more rapidly, and reacts readily
with the abundant (ca. 0.5–10 mM) cellular thiol tripeptide
glutathione (g-L-Glu-L-Cys-Gly, GSH), but has only one third the
potency of the pyridine analogue towards cancer cells.13 Hence,
a major aim of this work is to optimise the potency of organo-
aDepartment of Chemistry, University of Warwick, Coventry, CV4 7AL, UK. E-mail: p.j.
sadler@warwick.ac.uk
bSpectroscopy Research Technology Platform, University of Warwick, Coventry, CV4
7AL, UK
cDepartment of Chemistry and Chemical Technologies, University of Calabria, via
Pietro Bucci, 87036 Arcavacata di Rende, Cs, Italy
† Electronic supplementary information (ESI) available. CCDC 1860992–1860997,
1921724 and 1921725. For ESI and crystallographic data in CIF or other electronic
format see DOI: 10.1039/d0sc00897d
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
iridium(III) anticancer agents by rational control of their studied,21 their chemical and biological mechanisms of action
reactivity.
have been little explored.22–24 Also, the presence of the redox-
Here iodide is used as the monodentate ancillary leaving active azopyridine ligands provides low-lying p* orbitals. Such
group, a ‘so’ ligand expected to be strongly bound, and a poor ligands are known to participate in electron transfer processes
leaving group.17 Iodido ligands are known to confer inertness for organometallic catalysts.25–27 Previously we have found that
towards hydrolysis of Ru/Os arene complexes,18,19 and increase glutathione undergoes interesting reactions with phenyl-azo-
potency towards cancer cell lines compared to chlorido pyridine ligands in arene Ru(II) and Os(II) complexes.18,19,28 The
analogues.20 Although many half-sandwich iodido iridium inclusion of an azopyridine ligand azpyNMe2 (N,N-
complexes with the N-heterocyclic carbene ligands have been dimethylphenyl-azopyridine) in a chlorido Cpxph Ir(III) complex
Chart 1 Synthesis route for the organoiridium(III) complexes studied in this work, and formulae of the complexes (azpy ¼ 2-phenylazopyridine).
Crystals were obtained for complexes 10, 70, 80 as iodido salts, and 90 as a chlorido salt.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
produces a markedly different pattern of antiproliferative
Single crystals suitable forꢀX-ray diffraction of complexes 2,
activity in over 800 cancer cell lines compared to 253 standard 3$MeOH, 5$MeOH, 10 as PF6 salts, 10$MeOH, 70, 80 as iodido
drugs, suggesting a novel mechanism of action.29
salts, and 90$MeOH as a chlorido salt, were obtained at ambient
We have investigated the activation and reactivity of a small temperature by slow diffusion of Et2O into saturated methanol
family of novel inert half-sandwich iodido organoiridium(III) or dichloromethane (80) solutions. X-ray crystal structures of
complexes bearing variously substituted bidentate phenyl- complexes 10$MeOH and 90$MeOH are shown in Fig. 1, and the
azopyridine ligands, and two chlorido analogues (Chart 1). other six complexes in Fig. S3.† X-ray crystallographic data are
The X-ray crystal structures of eight complexes have been listed in Tables S1 and S2,† and selected bond lengths and bond
determined. The chemical reactivity of these iodido complexes, angles in Tables S3 and S4.† All the complexes adopt a typical
as well as their antiproliferative activity against human lung pseudo-octahedral geometry with a “piano-stool” shape. The Ir–
ꢀ
cancer cells and ability to induce cellular reactive oxygen species I bond lengths range from 2.6799(5)–2.6943(3) A. Except for
ꢀ
(ROS) were also studied. The antiproliferative activity of iodido complex 5$MeOH (1.870 A) recorded at 296 K, the distances
complex 1 against three other cancer cell lines and its toxicity in from the CpX ring centroid to iridium for all the other
ꢀ
vivo towards zebrash embryos compared to the chlorido complexes recorded at 150 K are within 1.820 ꢁ 0.009 A, similar
ꢀ
analogue was determined.
to the C^N chelated complex [(Cp*)Ir(ppy)Cl] (1.82 A) (ppy ¼ 2-
The focus of this work is on elucidating the mechanism of phenylpyridine), but slightly longer than t0he N^N chelated
16,34,35
ꢀ
activation of these inert iodido organoiridium anticancer complex [(Cp*)Ir(bpy)Cl]Cl (1.78 A) (bpy ¼ 2,2 -bipyridine).
complexes, especially their reactions with GSH, both experi- The N]N azo bond lengths are correspondingly lengthened to
mentally and through DFT calculations. Intriguing is their 1.27–1.29 A from the uncoordinated mean length of 1.25 A.36 In
ability to participate in both oxidative catalytic pathways and addition, Ir–N1 (pyridine nitrogen) bond is longer than that of
reductive azopyridine release pathways, features which have not Ir–N8 (azo nitrogen) in the iodido Cp* complexes, in contrast to
been observed previously for other metals. This appears to be chlorido Cp* complex 90$MeOH. However, in the iodido/
the rst report of such reaction pathways for organoiridium chlorido Cpxph complexes, Ir–N1 is of similar length to Ir–N8.
ꢀ
ꢀ
anticancer complexes.
The crystal structures of four complexes (10$MeOH, 3$MeOH,
5$MeOH, and 10) contain neutral phenol groups, while
complexes 2 and 90$MeOH have a bridging proton shared
between the phenoxide oxygens of neighboring complexes as
illustrated in Fig. S4.†
Results
Synthesis and characterization
Complexes 1–6 which contain a phenolic substituent on the
azopridine ligand (Chart 1, R1 ¼ OH) exhibit pH-dependent
changes in their UV-vis absorption spectra in aqueous media
(Fig. S5 and Table S5†), from which pKa values of 3.90–6.49 were
determined (Fig. S6 and S7†).
Ten novel iridium(III) complexes were synthesized in good yields
by stirring mixtures of iridium chlorido/iodido dimer precur-
sors with 2 mol equiv. of the appropriate chelating azopyridine
ligand in dichloromethane (Chart 1). All the complexes were
characterized by NMR spectroscopy (1H, 1H–1H COSY, 13C,
HSQC, HMBC), elemental analysis and ESI-MS (details in S3.2 of
the ESI†).
Electrochemistry
The complexes are chiral at the Ir(III) centre. 1H NMR spectra Electrochemical reduction of iodido complexes 1, 3 and 7 in
of complexes 1 and 2 in d4-MeOD aer addition of the anionic comparison with unbound azpy (2-phenylazopyridine), HO-azpy
chiral shi regent D-trisphat30,31 at 298 K (Fig. S1 and S2†) show (2-phenolazopyridine) ligands and iridium dimeric precursor
splitting of the aromatic proton signals in a ca. 1 : 1 ratio, [(Cp*)Ir(m-I)I]2, was studied by cyclic voltammetry (CV) under
indicating the presence of two enantiomers in equal abun- N2. Complexes 1, 3, and 7 exhibited more facile sequential
dance, similar to previously reported chiral osmium/ruthenium reduction than the corresponding free ligands or [(Cp*)Ir(m-I)
arene
picolinamide
and
iminopyridine
anticancer I]2. The rst reduction step is quasi-reversible, ranging from
ꢀ0.07 to ꢀ0.28 V. The second is irreversible, ranging from
complexes.32,33
Fig. 1 X-ray crystal structures of (a) [(h5-Cp*)Ir(HO-azpy)I]I$MeOH (10$MeOH), (b) [(h5-Cp*)Ir(HO-azpy)Cl]Cl$MeOH (90$MeOH), with thermal
ellipsoids drawn at 50% probability. The hydrogen atoms, counter anions and solvent molecules have been omitted for clarity.
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Table 1 Electrochemical reduction potentials of selected ligands and
complexes
integrals. These adducts were also characterized by ESI-MS,
shown in Table S6.†
Next, the stability of the iodido complex [(Cp*)Ir(HO-azpy-Br)
I]PF6 (3) (1.0 mM) in MeOD/D2O (1/1, v/v, pH* 7.0) was studied
in the presence of 4.0 mM, 23.0 mM, or 103.0 mM NaCl, phys-
iologically relevant extracellular and intracellular concentra-
tions, over 24 h at 310 K. Integration of the methyl 1H NMR
peaks for Cp* at d 1.73 ppm (3) and d 1.59 ppm (3-Cl, the
chlorido analogue of 3), showed that the amount of 3-Cl formed
by iodide/chloride ligand exchange at these three NaCl
concentrations was 2.7%, 7.8%, and 16.5%, respectively
(Fig. S17†). LC-MS spectra for complex 3 at micromolar
concentrations with the three different NaCl concentrations
showed a positive-ion peak for the chlorido analogue 3-Cl with
a shorter retention time of ca. 17.6 min (c.f. ca. 21.0 min for 3,
Fig. S18†). However, when 100 mM 3 reacted with 103.0 mM
NaCl, a higher amount of 3-Cl (ca. 56%) was detected by HPLC
Ligand/complexa
Ered (V)
Complexa
Ered (V)
Azpy
Azpy-OH
[Cp*Ir(m-I)I]2
ꢀ1.07, ꢀ1.93
ꢀ0.94, ꢀ1.76
ꢀ1.21, ꢀ1.69
1
3
7
ꢀ0.28, ꢀ0.79
ꢀ0.13, ꢀ0.58
ꢀ0.07, ꢀ0.63
a
Conditions: 1 mM free ligand or iridium complexes with 0.1 M
Bu4NPF6 as supporting electrolyte in acetonitrile under N2 at ambient
temperature, E1/2 (ferrocene/ferrocenium) ¼ 0.063 V, scan rate ¼ 0.1
V$sꢀ1
.
ꢀ0.58 to ꢀ0.79 V (Table 1, cyclic voltammograms shown in
Fig. S8†).
Aqueous reactivity
The hydrolysis behavior of iodido complexes 1–8 in d4-MeOD/ peak integration (Fig. S18†).
1
D2O (1/5 v/v) over 24 h was studied at 310 K by H NMR. No
In addition, 1H NMR spectra of the iodido and chlorido
changes were observed in the spectra over this time. ESI-MS complexes in d6-DMSO were monitored over a time course of 19
analysis of the NMR solutions showed only peaks assignable d to investigate their stability at 298 K, since DMSO was used as
to cation [M ꢀ PF6]+ of the intact iodido complexes. To further a solvent in cell growth inhibition assays. However, no changes
1
verify that the H NMR spectrum contained only peaks for the to any 1H NMR peaks were observed (Fig. S19 and S20†). ESI-MS
intact iodido complexes, and that therefore no hydrolysis had data showed peaks only for the original complexes, providing
occurred, complex [(Cp*)Ir(HO-azpy-Br)I]PF6 (3) in MeOD/D2O evidence for the inertness of these iridium complexes towards
(1/1, v/v) was reacted with 1.2 mol equiv. AgNO3 at 310 K to solvolysis in DMSO.
remove the coordinated iodide. The resulting NMR peaks for
the aqua species were assigned (Fig. S9a†) and contrasted with
the hydrolysis-inert complex 3 (Fig. S9b†).
Reactions with N-acetyl-L-cysteine (NAC)
Reactions of complex [(Cp*)Ir(HO-azpy-Br)I]PF6 (3) with Initially reactions of complexes 1, 3 and 7 with N-acetyl-L-
3.0 mol equiv. nucleotide guanosine 50-monophosphate (50- cysteine (NAC), as an example of a biologically important
GMP), nucleobase 9-ethylguanine (9-EtG), and amino acids L- thiol, were studied by 1H NMR spectroscopy. Time-
histidine, N-acetyl-L-methionine, L-tryptophan, and L-arginine dependent 1H NMR spectra of [(Cp*)Ir(HO-azpy)I]PF6 (1) (1
were also studied over 24 h at 310 K in 0.1 M phosphate buffer mM) with 3 mol equiv. NAC in d4-MeOD/0.1 M phosphate
D2O/d4-MeOD (1/1, v/v, pH* 7.8) by 1H NMR and ESI-MS. No buffer D2O (3/7 v/v, pH* 7.4) at 310 K were recorded up to 24 h
adduct was observed between complex 3 and 50-GMP or 9-EtG, (Fig. S21†). In the aromatic region, a new set of ligand-based
nor with L-Trp or L-Arg by NMR (Fig. S10†) or by ESI-MS. Only peaks appeared with the set of ligand peaks of iodido
small amounts of adducts with N-acetyl-L-methionine (29%) or complex 1 decaying during the rst hour (Fig. S21†). While in
L-histidine (7%) were observed (Fig. S11†) based on 1H NMR the aliphatic area, the peak for Cp* protons of complex 1
peak integrals, and also evidenced by ESI-MS with assignable shied completely from d 1.71 to 1.59 ppm within the rst
peaks for adducts (Table S6†).
hour, and two new sets of peaks in a 1 : 1 ratio for acetyl
The hydrolysis of the chlorido analogues was also inves- methyl protons of NAC appeared at d 1.88 and 1.77 ppm
1
tigated at 310 K for comparison. H NMR spectra of 100 mM (Fig. S21†).
solutions of [(Cp*)Ir(HO-azpy)Cl]PF6 (9) and [(Cpxph)Ir(HO-
Aer reaction of 1 (100 mM) with 3 mol equiv. NAC for 1 h,
azpy)Cl]PF6 (10) in d6-DMSO/D2O (1/9, v/v) showed that these LC-MS analysis showed a new ESI-MS peak with a shorter
complexes hydrolyzed to the extent of 53% (Fig. S12†) and retention time of ca. 12.5 min with m/z 689.26, assignable as [(1-
66% (Fig. S13†), respectively, at equilibrium over 24 h. The NAC)-I]+ with a bound deprotonated NAC thiolate (calcd 689.18,
aqua adduct of complex 9 was also detected by ESI-MS, giving NAC is [CH3CONHCH(COOH)(CH2S)]ꢀ in Fig. S22†). This
a positive ion peak m/z at 543.99 corresponding to the conrmed that the new sets of NMR peaks for ligand, Cp* and
formula [(Cp*)Ir(O-azpy)(H2O)]+ (9-H2O) (calcd 544.15). The NAC were from 1-NAC adducts. The two sets of acetyl methyl
hydrolysis of complex 9 followed pseudo rst-order kinetics peaks for bound NAC can be assigned to the diastereomers of 1-
(Fig. S14†) with an hydrolysis rate constant of 0.00698 ꢁ NAC (Fig. S21†) due to the chirality of the iridium centre and
0.00096 minꢀ1 at 310 K and the half-life of 99.3 min deter- NAC. Diastereomers were also evident in the 1H NMR spectra of
mined by HPLC analysis. The chlorido complex 9 not only isolated adducts of complex 7 (Fig. 2) and 3 (Fig. S23†) with
reacted with 9-EtG to the extent of 15% (Fig. S15†), but also NAC, as two sets of peaks. Similarly, all the 13C NMR signals
almost completely formed adducts with N-acetyl-L-methio- showed equally intense pairs of peaks for 7-NAC (details of 13
nine and L-histidine (Fig. S16†) based on 1H NMR peak NMR assignments for 7-NAC in S3.2 of the ESI†). This appears
C
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
within 15 min of the addition of GSH (2 mM) in d4-MeOD/
phosphate buffer D2O (0.1 M, 3/7 v/v, pH* 7.4) at 310 K, a new
set of aromatic ligand peaks as well as a new set of GSH protons
appeared (Fig. 3), and the Cp* methyl peak shied from 1.70 to
1.53 ppm and split into two peaks indicative of diastereomers of
[(Cp*)Ir(azpy)(SG)]+ (7-SG).
Meanwhile, HPLC separation of the NMR solution revealed
a new peak with shorter retention time of ca. 12.0 min
compared to the parent complex 7 (ca. 21.5 min, Fig. S24†). The
new ESI-MS peak with m/z 817.4 (Fig. S25†) can be assigned as
the glutathione thiolate adduct [(7-SG) + H]+ (calcd m/z 817.2).
When 7 was reacted with 10 mol equiv. of GSH under similar
conditions, there was a dramatic loss in intensity of peaks in the
aromatic region of the 1H NMR spectrum aer 15 min, together
with broadening of several peaks in the aliphatic region, shown
in Fig. 4b, perhaps due to the presence of paramagnetic species.
Aer 3 h, a new set of peaks appeared in the aromatic region
combined with a 1 : 1 doublet at d 1.79/1.82 ppm as well as new
peaks at d 3.30 ppm assignable to the b-CH2 of GSSG, Fig. 4c.
The main species in the NMR reaction mixture aer 3 h were
subsequently separated by HPLC and analyzed by positive-ion
mass spectra. The MS peak at m/z 183.65 was assigned as
unbound azpy ligand [(azpy) + H]+ (calcd 184.08), and the peak
with m/z 185.64 as the two-electron-reduced product phenyl-
hydrazo-pyridine [(H2azpy) + H]+ (calcd 186.10), Fig. S27.† In
addition, an MS peak with m/z 786.96 was assigned as binuclear
[(Cp*Ir)2(m-SG)3 + 3H]2+ (calcd 787.19; Fig. S27†). These species
indicated that aer 3 h, the new set of 1H NMR aromatic peaks
Fig. 2 1H NMR spectrum (400 MHz, d4-MeOD/0.1 M phosphate buffer
D2O, 3/7, v/v, pH* 7.4, 298 K) of isolated 7-NAC adducts. The bound
NAC (assigned as red i, j, and k), the Cp* ring methyl (assigned as blue
h) and the protons assigned as a and d on the azo ligand show two sets
of peaks due to the presence of diastereomers.
to be the rst characterization of half-sandwich iridium NAC
adducts.
In contrast, no adduct was observed by LC-MS upon incu-
bation of complex 7 with 10 mol equiv. of the amino acid b-
alanine under the same reaction conditions. Hence, the thiol
group of NAC appears to be a crucial site for reaction with these
iodido iridium complexes.
Reactions with glutathione (GSH)
Next the reactivity of complex 7 with glutathione was investi-
gated. In the 1H NMR spectrum of [(Cp*)Ir(azpy)I]PF6 (7) (1 mM)
Fig. 3 1H NMR spectra (400 MHz, d4-MeOD/0.1 M phosphate buffer D2O, 3/7 v/v, pH* 7.4) of (a) complex 7 (1 mM), and (b) 15 min after addition
of 2 mol equiv. GSH (2 mM) at 310 K, showing complete formation of 7-SG.
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Fig. 4 1H NMR spectra (400 MHz, d4-MeOD/0.1 M phosphate buffer D2O, 3/7 v/v, pH* 7.4, 298 K) of (a) complex 7 (1 mM), (b) 15 min, and (c) 3 h
after reaction with 10 mol equiv. GSH at 310 K. The broadening of peaks in (b) can be ascribed to paramagnetic effects, and the new set of peaks
in (c) assigned to the released phenyl-hydrazo-pyridine (H2azpy) and tri-SG bridged iridium dimer [(Cp*Ir)2(m-SG)3]+. Red squares denote the 7-
SG adducts and * denotes residual CHD2OD.
(Fig. 4c) was due to phenyl-hydrazo-pyridine arising from OH, R2 ¼ Br), 7 (Cp*, R1 ¼ R2 ¼ H), and 8 (Cpxph, R1 ¼ R2 ¼ H)
reduction and release of the azpy ligand, and the peaks at 1.79/ (Chart 1, 100 mM) in phosphate buffer (30 mM, pH* 7.4) at 310 K
1.82 ppm to the Cp* methyls of [(Cp*Ir)2(m-SG)3]+.
for 24 h by 1H NMR spectroscopy (Fig. S29†). The formation of
GSSG was evident from the appearance of new peaks at e.g.
d 3.30 ppm corresponding to the b-CH2 of GSSG (Fig. S29†),39
and was further conrmed by the HPLC/LC-MS peak with m/z at
612.62 (calcd 613.15 for [GSSG + 3H]+), Fig. S30.† The turnover
numbers (TONs) for complexes 7 and 8, with unsubstituted azpy
phenyl substituents, of 100 ꢁ 4 and 100 ꢁ 2, respectively, are
much higher than those of the phenolate azpy complexes 1 and
3 (29 ꢁ 2 and 18 ꢁ 1, respectively, Fig. 6). By contrast, the free
ligand phenol-azopyridine (HO-azpy) showed negligible cata-
lytic activity (low TON of 9 ꢁ 1, Fig. 6), and the GSH alone
underwent negligible oxidation (Fig. S29†). No gas bubbles or
pH changes were observed over 24 h incubation for the
complexes, nor was H2O2 detected by peroxide test sticks (even
with 1 mM catalyst).
To identify the active iridium catalyst, the reaction of
complex 7 (100 mM) with GSH (10 mM) in phosphate buffer
(30 mM, pH 7.4) was monitored over 24 h by HPLC/LC-MS.
From the ESI-MS of the HPLC peaks, ca. 100% of 7 was con-
verted to the glutathione adduct [(Cp*)Ir(azpy)(SG)]+ (7-SG)
within the rst 3 min aer mixing, and this was the major
species in the solution over 24 h (Fig. S31†). Hence the Ir-SG
adduct appears to be the active catalyst in GSH oxidation. This
was further conrmed by studying reactions of isolated
complexes 7-SG or 3-SG (100 mM) with GSH (10 mM) under
similar conditions (Fig. S32†). These reactions gave TONs of
Radical trapping by EPR
We investigated whether the NMR peak broadening (Fig. 4b)
might be due to the presence of radicals and attempted to detect
them by electron paramagnetic resonance (EPR) using a spin
trap. EPR spectra of a solution containing 7 (1 mM) with GSH
(20 mol equiv.) in phosphate buffer (0.1 M, pH 7.4) and the spin
trap DEPMPO (Fig. 5b) or DMPO (Fig. S28†) showed a strong
doublet of 1 : 2 : 2 : 1 quartets within the rst 87 min, which
decreased in intensity thereaer. These EPR signals are
assignable by simulation to trapped hydroxyl radicals (Fig. 5c
and S28†). GSH alone in the phosphate buffer (0.1 M, pH 7.4)
(Fig. 5a) or complex 7 (1 mM) with GSH (20 mol equiv.) in
phosphate buffer (0.1 M, pH 7.4) pre-deaerated with argon, were
EPR-silent. In addition, when superoxide dismutase was added
to the starting reaction mixture of complex 7 with GSH in
aerated phosphate buffer, the formation of hydroxyl radicals
was also not observed.
Catalysis of GSH oxidation
We investigated whether these [(CpX)Ir(R1-azpy-R2)I]+
complexes can act as catalysts for the oxidation of GSH to GSSG.
We monitored reactions with GSH (10 mM) for complexes with
representative structural variability in the azopyridine ligand or
the Cpx ring, complexes 1 (Cp*, R1 ¼ OH, R2 ¼ H), 3 (Cp*, R1 ¼
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
17 ꢁ 2 aer 24 h at 310 K. Furthermore, the nal reaction
mixture analyzed by LC-MS, gave peaks assignable to the free
azpy ligand, the free (reduced) H2azpy ligand, and the dinuclear
adduct [(Cp*Ir)2(m-SG)3]+ (Fig. S33†). These are different from
the major product peak which was for 7-SG (Fig. S31†) under
aerated conditions.
DFT simulation of aquation and GSH reactions
The X-ray crystal structure of the [(Cpxph)Ir(HO-azpy)I]PF6
complex (2) was used as a starting point for DFT calculations
(details in S2.22 of the ESI†). DFT calculations suggested that all
the investigated lowest energy pathways involve deprotonated
GSH (GSꢀ). Although the pKa of the Cys thiol in GSH is high (ca.
9.4),40 there will be a small but signicant amount (ca. 0.3%) of
deprotonated GSH present at pH 7.41
DFT calculations were conducted rst to establish whether
direct attack of H2O or GSꢀ on the Ir centre to displace the
iodido ligand can occur. That is, aer initial formation of
slightly less stable adducts between 2 and H2O or GSꢀ, the
reactions proceed by second-order nucleophilic substitution
(SN2). The activation barriers for formation of the transition
states for these associative attacks are similar (19.9 vs.
20.2 kcal molꢀ1, Fig. 7). In contrast, the thermodynamics for
displacement of the iodide by H2O or GSꢀ are different, being
endergonic by 10.7 kcal molꢀ1 for H2O, and exergonic by
30.3 kcal molꢀ1 to form Ir-SG. Therefore, the calculations
support the experimental nding that the complex is inert
towards hydrolysis, and that direct GSꢀ attack is at least ther-
modynamically accessible.
Fig. 5 EPR spectra of (a) GSH alone; (b) radicals trapped by DEPMPO
(100 mM) in the reaction mixture of complex 7 (1 mM) with GSH
(20 mol equiv.) in 0.1 M phosphate buffer, pH 7.4, in the initial 87 min at
298 K; (c) simulated spectrum for trapped hydroxyl radicals DEPMPO-
OH (aN ¼ 14.0 G, aH ¼ 13.2 G, and aP ¼ 47.3 G)37 using the EasySpin
program.38
Attack of GSH on the azo bond
DFT exploration of possible, less energy-demanding, alternative
pathways for the substitution of iodide by GSꢀ or H2O showed
that the reversible attack of GSꢀ on the N]N bond of the azo-
pyridine ligand can assist in the process (Scheme 1).
Along pathway I for the substitution of iodide by H2O in
Fig. 8a, the attack of GSꢀ on the non-coordinated N atom of the
azo double bond, involves a barrier of only 9.2 kcal molꢀ1 to
reach the transitional state TS1. Then water coordination to the
vacant site on Ir (TS2 in Fig. 8a) occurs by surmounting an
energy barrier of 8.7 kcal molꢀ1. To complete the substitution of
Iꢀ by water, detachment of GSꢀ from the N atom (the TS3 in
Fig. 8a) corresponds to a very low energy barrier of only
0.2 kcal molꢀ1. The whole reaction is almost thermoneutral.
Along pathway II for the substitution of iodide by GSꢀ in
Fig. 8b, the rst step involving attack by GSꢀ on the N atom of
the azo bond and the simultaneous release of the iodide ligand
Fig. 6 Turnover numbers for the catalytic oxidation of GSH by
complexes 1, 3, 7, 8, 7-SG, 3-SG and free HO-azpy ligand. Reactions
were carried out in d6-acetone/phosphate buffer D2O (5/95, v/v,
30 mM, pH* 7.4) at 310 K for 24 h and monitored by 1H NMR
spectroscopy.
occur by surmounting
a much low energy barrier of
5.6 kcal molꢀ1. The free energy prole for the next step (coor-
dination of the second GSꢀ to iridium) is very at with a barrier
of only 1.3 kcal molꢀ1 to the transitional state TS2 in Fig. 8b.
With respect to pathway I, attack of water on the iridium centre
(Fig. 8a), the GSꢀ bound to the N atom of the azo group in
pathway II is not spontaneously released. Instead, in the pres-
ence of a third GSꢀ, the detachment of GSꢀ from the N atom
occurs to form oxidized glutathione, GSSG, which involves the
92 ꢁ 1 and 18 ꢁ 3, respectively, exhibiting similar catalytic
activity as the iodido complexes 7 and 3.
Since the EPR studies indicated that oxygen is involved in the
generation of hydroxyl radicals during these reactions, the
reaction of complex 7 with GSH was carried out under oxygen-
depleted conditions, which decreased the TON by ca. 83% to
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Fig. 7 DFT calculated free energy profiles for direct substitution of iodide in complex 2 by water (aquation) or GSꢀ. Relative energies are
in kcal molꢀ1 and calculated with respect to separated reactants.
transfer of two electrons to other molecules. Computations 0.3–1.6 mM. Particularly potent is complex [(Cp*)Ir(HO-azpy-Br)
suggest that the ground state O2 (3S) molecule can act as an I]PF6 (3) with an IC50 value of 0.33 mM, 10ꢂ more active than
electron acceptor for this step. Indeed, the O–O bond elongates, cisplatin. By contrast, complexes 7 and 8 bearing an unsub-
consistent with formation of superoxide O2ꢀc, whereas the stituted azpy ligand are less active than other iodido complexes
second electron is localized on the N atom of N–N single bond 1–6. The substitution of one methyl group on Cp* by a phenyl
which was previously bound to GSꢀ. The whole system is in substituent to give Cpxph does not enhance the anticancer
a triplet state again, and the spin is conserved. The singlet activity for iodido complexes, while the Cpxph chlorido complex
multiplicity of the complex can be recovered due to the presence 10 is slightly more active than its Cp* analogue 9.
of a second O2 molecule to form a second superoxide. In
Based on their promising in vitro activity, the potent iodido
pathway II of Scheme 1, the nal products are the singlet state complex 1 and its chlorido analogue 9 were selected for further
Ir-SG complex, GSSG, and two O2ꢀc radicals, to conserve the screening against human CNE2 (nasopharyngeal), A2780
triplet multiplicity. Overall, this reaction is largely exergonic.
(ovarian), and A2780cisR (ovarian) cancer cell lines to study the
inuence of the halido ligand on the anticancer activity.
A2780cisR is cisplatin-resistant through reduced drug trans-
port, enhanced DNA repair/tolerance and elevated GSH levels.43
Iodido complex 1 exhibits similar potency to chlorido analogue
9 against A549, CNE2, and A2780 cell lines, both more potent
than cisplatin (see IC50 values in Table 2). Surprisingly, iodido
complex 1 is ca. 2ꢂ more potent than the chlorido analogue 9
towards the cisplatin-resistant A2780cisR cell line. Both
complexes are not cross-resistant with cisplatin, with resistance
Antiproliferative activity
The antiproliferative activity of the complexes towards human
lung A549 cancer cells was determined using the SRB assay42
and compared with cisplatin (CDDP), Fig. 9 and Table S7.†
Iodido complexes 1–6 and chlorido complexes 9 and 10, which
share the common feature of a phenolic substituent on the
azopyridine ligand, are highly potent with IC50 values (the
concentration that inhibits cell growth by 50%) in the range
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
Scheme 1 Pathways I and II for the formation of Ir-OH2 and Ir-SG, as investigated by DFT calculations. Intermediates (abbreviated as [Int]) and
products involve the reversible attack of GSꢀ on the azo bond.
Fig. 9 Inhibition of growth of A549 human lung cancer cells in vitro by
complexes 1–10 in comparison with cisplatin (CDDP). The values of
half maximal growth inhibitory concentration (IC50) are given as the
mean ꢁ standard deviations for a duplicate of triplicate experiments
with cancer cells exposed to the tested complex for 24 h followed by
recovery for 72 h in fresh complex-free medium. See Table S7† for full
numerical data.
factors (IC50(A2780cisR)/IC50(A2780)) for 1 and 9 of 0.2 and 0.95,
respectively. This suggests that these iodido complexes have
a different mechanism of action compared to cisplatin.
In vivo toxicity studies
To compare the in vivo toxicity of an iodido complex and its
chlorido analogue, the LC50 lethal concentration (concentration
which is lethal to half of the population) towards zebrash
Fig. 8 Calculated free energy profiles describing iodide substitution
(Danio rerio) embryos was determined for potent iodido
by (a) water in pathway I, and (b) GSꢀ in pathway II. Calculated energies
complex 1 and its chlorido analogue 9. This high-throughput
are in kcal molꢀ1 with respect to reference energy of separated
reactants.
vertebrate model is oen used as a predictor for drug toxicity
in humans.44–47 Strikingly, iodido complex 1 (LC50 ¼ 0.26 ꢁ 0.08
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Table 2 Antiproliferative activity of iodido complex 1 and chlorido analogue 9 towards human A549 lung, CNE2 nasopharyngeal, A2780 ovarian,
and cisplatin-resistant A2780cisR ovarian cancer cells for 24 h exposure to the complexes and subsequent 72 h cell recovery in fresh complex-
free medium
Complex
A549
CNE2
A2780
A2780cisR (RF)a
[(Cp*)Ir(HO-azpy)I]PF6 (1)
[(Cp*)Ir(HO-azpy)Cl]PF6 (9)
Cisplatin
1.01 ꢁ 0.08
1.5 ꢁ 0.3
3.3 ꢁ 0.1
1.26 ꢁ 0.04
2.3 ꢁ 0.3
7.7 ꢁ 0.3
0.25 ꢁ 0.02
0.12 ꢁ 0.04
1.2 ꢁ 0.2
0.049 ꢁ 0.001(0.20)
0.114 ꢁ 0.003 (0.95)
11.5 ꢁ 0.3 (6.41)
a
Resistance factor RF ¼ IC50(A2780cisR)/IC50(A2780).
mM) was ca. 25ꢂ less toxic than its chlorido analogue 9 (LC50
¼
Discussion
0.010 ꢁ 0.003 mM), although both are more toxic than cisplatin
(LC50 ¼ 0.6 ꢁ 0.2 mM).48
Iridium(III) is a classically-inert low-spin 5d6 metal ion, exem-
plied by the half-life for coordinated aqua ligands exchange on
[Ir(H2O)6]3+ of >300 years.49 However, it was apparent during our
study of [Ir(Cp*)(N,N)Cl]+ anticancer complexes (N,N ¼ e.g.
phenanthroline),16 and a previous report50 that the introduction
of a Cp* ligand into the coordination sphere can increase ligand
exchange rates by many orders of magnitude. Here we show that
the inertness of Ir(III) can be restored even in Cp* complexes,
when the N,N-chelated ligand is a strong p-acceptor, and the
monodentate ligand is a ‘so’ iodido ligand.
Our aim was to explore pathways by which such inert Cp*
complexes might become activated in cancer cells, to shed light on
possible mechanisms of action. In so doing we have discovered
some unusual reaction pathways for half-sandwich organoiridium
complexes, involving not only attack on the azo bond of coordi-
nated azopyridine by glutathione, but also catalysis of GSH oxida-
tion and generation of reactive oxygen species, and reductive release
of the azopyridine ligand, suggesting that these iodido iridium
complexes are likely to have a unique mechanism of action.
ROS detection
ROS generation was compared for the most cytotoxic complex,
3, and one of the least cytotoxic complex 7. The aim of this study
was to assess whether there is a direct correlation between in
vitro cytotoxicity and ROS generation. The levels of reactive
oxygen species (ROS) in A549 human lung cancer cells treated
with complex 3 or 7 were determined at equipotent 2ꢂ IC50
concentrations (Fig. 10 and Table S8†) by ow cytometry uo-
rescence analysis using a total ROS/Superoxide (SO) Detection
Kit. Superoxide production was monitored by following the
orange channel FL1, and total ROS species, including H2O2,
peroxy and hydroxyl radicals, peroxynitrite and NO, were
monitored by the green channel FL2. Aer exposure to the
complexes for 24 h, an increase in ROS/SO levels in cells treated
with 3 or 7 was observed, when compared to untreated cells.
Furthermore, a burst of superoxide production was also
observed in cells treated with complex 7.
Inertness of iodido complexes towards aquation
Transition metal anticancer complexes containing chlorido
ligands, including cisplatin and Ru(II) arene complexes, oen
undergo activation by hydrolysis, giving more reactive aqua
adducts.51 For the iodido complexes studied here, due to the
soer character of iodide, the Ir–I bonds have stronger cova-
lency, strengthened by p-acceptor chelated azopyridine ligands,
and become inert towards hydrolysis. This is paralleled by the
more stable iodido Ru(II)/Os(II) azopyridine complexes
compared to their chlorido counterparts.20 Thus chlorido
complex 9 hydrolyses much more slowly than diamine and
diimine N^N chelated chlorido half-sandwich Cp* Ir(III) anti-
cancer complexes.16
The inertness of iodido iridium complexes towards hydro-
lysis also leads to their much lower chemical reactivity towards
the nucleobase 9-EtG and various amino acids compared to the
Fig. 10 ROS induction in A549 cancer cells exposed to complex 3 or 7 chlorido analogue (Table S6†). Aer 24 h, ca. 56% of micro-
at 2ꢂ IC50 concentrations for 24 h with untreated cells as the negative
control. FL1 channel detects superoxide production, and FL2 channel
detects total oxidative stress. Normalized population data are pre-
sented as the mean ꢁ SD of triplicate samples for one experiment. p-
Values were calculated after a t-test against the negative control data,
*p < 0.05, **p < 0.01. See Table S8† for full numerical data.
molar iodido complex 3 was converted into its chlorido
analogue at high (extracellular) NaCl concentration (103 mM,
Fig. S18d†), and therefore much of this and other iodido
complexes would be expected to persist for transport into cells,
where the concentration of NaCl is much lower. Thus activation
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
than for scaffold-inert iridium pyridocarbazole complexes57 and
Os(II) azopyridine complexes,48 but there is scope in further
work for improving this by changes in ligand substituents.
GSH-mediated ligand-centred activation
These iodido iridium complexes are resistant towards hydro-
lysis and nucleobase binding, but exhibit more potent anti-
proliferative activity compared to the clinical anticancer drug
cisplatin against A549 lung, CNE2 nasopharyngeal, and A2780
ovarian cancer cell lines (Fig. 9 and Table 2). Moreover, iodido
complex 1 is more active than the chlorido analogue towards
cisplatin-resistant A2780cisR ovarian cancer cells. The high
potency of these inert iodido complexes raises intriguing
questions about their activation mechanism.
Fig. 11 Acidity of complexes 1–6.
of the iodido iridium complexes is likely to involve novel
pathways, distinct from chlorido complexes, and have targets
other than coenzyme NADH6 or DNA.52
Enhanced antiproliferative activity of zwitterionic complexes
GSH is an important cellular antioxidant which detoxies
various xenobiotics as well as protecting cells from toxic reactive
oxygen species.58 Thus, reactions of metal complexes with GSH
can perturb the redox state of cells. Unlike the detoxication of
platinum drugs by conjugation with GSH, some chlorido RuII
arene ethylenediamine anticancer complexes form Ru thiolate
adducts which are not the dead-end products, but further
oxidized to sulfenato and sulnato adducts, facilitating the
interaction of Ru complexes with DNA.59 Other Ru(II),60,61
Os(II),18,28 and Pt(IV)62 complexes have been reported to be acti-
vated by GSH. The high affinity of these iodido Ir(III) azopyridine
complexes for NAC or GSH to form Ir-thiolate adducts under
physiologically relevant conditions is evident (Fig. 3 and 4).
DFT calculations show that displacement of iodide by GSꢀ to
yield Ir-SG is thermodynamically favorable, being exergonic
overall (Fig. 7). Regardless of the ratio between the iodido
complex 7 and GSH, only the Ir-SG adduct is observed in the
reaction (Fig. S22 and S24†). This is a different behavior from
the iodido OsII azopyridine complexes, which undergo hydro-
lysis to Os–OH species in the presence of equimolar GSH.18 In
the present study, neither Ir–OH nor Ir–sulfenato species were
On account of their low pKa values (Fig. S7†), complexes 1–6
exist mainly in neutral zwitterionic forms (Fig. 11) at physio-
logical pH (7.4) in cell culture media under cell-screening
conditions.
These complexes all show higher anticancer potency against
human A549 lung cancer cells than non-phenolic complexes 7
or 8 (Fig. 9). The anticancer activity of metal complexes oen
correlates with cellular uptake and lipophilicity.53 The accu-
mulation of iridium in the lung cancer cells treated with the
most potent complex [(Cp*)Ir(HO-azpy-Br)I]PF6 (3) was 2-fold
higher than that in the cells treated with [(Cp*)Ir(azpy)I]PF6 (7)
at equipotent IC50 concentrations (Table S9†). The n-octanol/
water (pH 7.4) partition coefficients (log P7.4) for complexes 3
(ꢀ0.08) and 7 (ꢀ0.59), reveal that complex 3, with pKa of 6.31
(Fig. S7†) and being predominantly in its zwitterionic form, is
more lipophilic relative to the +1-charged complex 7. The higher
lipophilicity contributes to the higher cellular iridium accu-
mulation and anticancer activity of complex 3.
Antiproliferative activity and in vivo toxicity
The Ir(III) analogue of the Ru(III) clinical trial drug NAMI-A, observed.
trans-[IrCl4(DMSO)(Im)][ImH], (DMSO ¼ dimethyl sulfoxide,
The rst-step in the electrochemical reduction of azo ligands
Im ¼ imidazole), exhibits low cytotoxic activity, attributable can be assigned to the one-electron addition into the p* orbital
partly to its inertness towards hydrolysis.54 The cytotoxicity of centred on the azo group to give the azo anion radical.63–65 The
RuII arene complexes is generally greater for complexes which second one-electron reduction gives rise to the dianionic
hydrolyse relatively rapidly, and low for complexes which do not species ([–N–N–]2ꢀ). In aqueous media, two-electron reduction
hydrolyse.55 However, for half-sandwich iridium complexes is also accompanied by proton transfer to give hydrazo groups
bearing C^N chelating ligands, replacement of Cl by pyridine [–NH–NH–].19 The azo bond undergoes more facile reduction
decreases the rate of hydrolysis, but enhances the anticancer aer metalation with Cp*-Ir (Table 1). The redox potential of
activity.56 In this work, replacement of Cl by I in the iridium GSH/GSSG (ꢀ240 mV at pH 7) in cells and tissues,66,67 is more
azopyridine complexes does not result in loss of anticancer negative than that of the azo bonds in the active iridium anti-
activity towards A549 lung cancer cells (complexes 1 and 9, cancer complexes (ꢀ130 mV and ꢀ70 mV for 3 and 7, respec-
Chart 1 and Table 2). Remarkably, iodido complex 1 has 2-fold tively, Table 1), suggesting that azo bonds in the ligands are
higher activity towards cisplatin-resistant A2780cisR human likely to be reduced by GSH.
ovarian cancer cells with a much lower resistance factor than
Furthermore, in the NMR study of reactions of complex 7
the chlorido analogue 9 (Table 2). This suggests that the potent with 10 mol equiv. GSH under physiologically relevant condi-
iodido complexes have potential to overcome the cisplatin tions (Fig. 4b), there was a rapid disappearance (severe broad-
resistance.
ening) of the aromatic signals and the broadening of Cp*
The toxicity of iodido complex 1 towards zebrash embryos methyl peaks of the 7-SG adducts aer 15 min, which persisted
is 25ꢂ lower than that of chlorido complex 9 which opens up for a few hours. This broadening may be due to ligand exchange
a wider therapeutic window. However, this selectivity is lower reactions occurring at an intermediate rate on the NMR
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
Fig. 12 Proposed activation and catalysis pathways for the reaction of complex 7 with GSH following the DFT-calculated pathway II (Scheme 1).
The paramagnetic species detected by NMR may be a ligand-centred radical species as indicated by [Int3].
timescale, or to the formation of large, slow-tumbling polymeric instead of [(Cp*)(Ir-OH2)(N-N-SG)]+ ([Int2] in pathway I, Scheme
species, or to paramagnetic species arising from redox reac- 1). Next, along pathway II the intermediate [(Cp*)(Ir-SG)(N-N-
tions, such as azo ligand-centred radicals. The latter case SG)] reacts with a third GSꢀ by the detachment of the GSꢀ
seemed the most likely since the azo bond can take part in one- from the nitrogen atom to form oxidized glutathione GSSG. One
and two-electron reduction processes (Table 1).
electron can be transferred to O2 as evidenced by the O–O bond
EPR studies using spin traps for reactions carried out under elongation, and a superoxide anion is formed. Whereas,
similar conditions to the NMR study, conrmed the formation another electron is localized on the N atom previously attacked
of hydroxyl radicals (Fig. 5). Since no radicals were detected in by GSꢀ, which can be accepted by another oxygen molecule to
the presence of superoxide dismutase, or in the absence of O2, it form a second superoxide anion with the simultaneous regen-
is reasonable to assume that the hydroxyl radicals arise from eration of N]N bond in the Ir-SG adduct. The overall reaction is
superoxide, O2cꢀ, formed by the initial attack of glutathione on largely exergonic by 59.4 kcal molꢀ1. This thermodynamically
the N]N azo bond generating GS–N–N– species. It is known favorable pathway II and involvement of oxygen in the activa-
that trapped superoxide readily decomposes to hydroxyl radi- tion pathway correlate well with the trapping of hydroxyl radi-
cals.68 Aer 3 h, the presence of 1.0 mol equiv. free phenyl- cals by EPR as decay products of superoxide. Such a favorable
hydrazo-pyridine ligand and 0.5 mol equiv. tri-SG bridged di- attack of GSH on the azo bond is also consistent with the
iridium [(Cp*Ir)2(m-SG)3]+ adduct in the solution, suggests that observation of iodide release in alkaline solution, where Ir-SG
the azo group is the redox-active centre during the reactions of adducts are completely formed within minutes (Fig. S34†).
the complexes with GSH (Fig. 4c).
Thus, the reaction pathways depicted in Fig. 12 account for
the main features observed in the reaction of complex 7 with
GSH. These include the presence of paramagnetic species
(Fig. 4b) during the early stages (as a ligand-based radical
[Int3]), the involvement of O2 as an electron acceptor in
formation of Ir-SG adducts, the catalysis of GSH oxidation, and
in the later stages, formation of reduced free hydrazopyridine
together with thiolate-bridged dinuclear complexes (Fig. 4c).
The formation of hydrazo products by consecutive reductions of
azo bonds by thiols has been reported in the case of azo-
Activation mechanisms by DFT calculations
DFT calculations revealed that the subsequent step aer release
of the iodido ligand, involves attack by a second GSꢀ on the Ir
centre to form an Ir-SG complex requiring only 1.3 kcal molꢀ1
along pathway II (Scheme 1 and Fig. 8b), in contrast to the
higher energy requirement of 8.7 kcal molꢀ1 for water binding
to the Ir centre to give Ir-OH2 along pathway I (Scheme 1 and
Fig. 8a). Thus, aer the attack of the rst GSꢀ on the N atom of
the azo bond, the energy-preferred sequence of steps is pathway
II (Scheme 1), that is attack by a second GSꢀ on the Ir centre to
form [(Cp*)(Ir-SG)(N-N-SG)] (the [Int2] of pathway II, Scheme 1),
bipyridine-bridged
dinuclear
ruthenium/iridium
complexes,69,70 although it is notable that this appears not to
be observed in the activation of iodido arene Ru(II)19 and
Os(II)18,28 azopyridine complexes by GSH. Furthermore, the
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
released phenyl-hydrazo-pyridine ligands and the dinuclear towards aquation or nucleobase binding. Some exhibit a 10-fold
complex [(Cp*Ir)2(m-SG)3]+ might not be dead-end products as higher potency than the anticancer drug cisplatin against
they themselves may play a role in the biological activity. For human lung cancer cells and are not cross-resistant. Also,
example, the organic hydrazo compound procarbazine was complex 1 is 25-fold less toxic in vivo towards zebrash embryos
approved as an anticancer drug in the late 1960s.71,72 Also than its chlorido analogue.
Therrien et al. have reported that thiolate-bridged dinuclear
Experiments and DFT calculations suggest that reactions
Ir(III) complexes exhibit high anticancer activity towards A2780 with the abundant intracellular tripeptide glutathione play
and A2780cisR ovarian cell lines.73
a major role in activation of these iodido complexes in cells and
in generation of cytotoxic radicals (including superoxide),
reduced azopyridine ligand and thiolato-bridged dinuclear
complexes. A key feature in the activation is the attack of
glutathione on the azo bond of the coordinated azopyridine.
Reactions with GSH also give S-bound thiolato adducts which
are catalysts for the oxidation of GSH to GSSG in oxygen-
dependent mechanisms. It will be interesting in future work
to investigate the dependence of cancer cell cytotoxicity on
cellular oxygen concentrations because many tumours are
hypoxic.
It will also be interesting to explore the distribution of the
complexes and the active Ir-SG catalyst in cell compartments.
High concentrations of glutathione are present not only in the
cytoplasm, but also in mitochondria where it has crucial roles.76
Also the concentration of oxygen in mitochondria is expected to
be high since this is the site of the electron transport chain
which leads to oxidative phosphorylation. It will be interesting
to investigate reactions between these complexes and proteins,
which might be specic to those containing accessible free Cys
residues.
Catalysis of GSH oxidation
Cp* iridium complexes are widely exploited as transfer hydro-
genation catalysis with hydride donors such as formate or
reduced nicotinamide adenine dinucleotide.74 Such reactions
can lead to imbalance in redox homeostasis in cancer cells.
However, the formation of catalytic Ir–hydride intermediates is
likely to be inhibited in the presence of GSH, which is present in
millimolar concentrations in cells. Catalysis of the oxidation of
GSH to GSSG is another potential strategy for modulating
cellular redox metabolism, although it has not been well
studied. Iridium complexes of the type [(CpX)Ir(N^N0)SH]+ are
capable of oxidizing the GSH to GSSG, however, the mechanism
has yet to be investigated.75 Thiolato-bridged iridium dimers are
highly cytotoxic, but are poor catalysts.73 Here we have discov-
ered that Cp* Ir(III) azopyridine iodido complexes can catalyse
oxidation of GSH to GSSG under physiologically relevant
conditions (Fig. 6).
Based on the experimental data and DFT calculations, the
mechanism for activation and catalysis shown in Fig. 12 for
complex 7, can be proposed. DFT calculated pathway II (Scheme
1) is largely exergonic, indicating that the activation of iodido
complexes leads mainly to Ir-SG adducts. The azo bond of these
Ir-SG complexes is attacked by excess GSH giving rise to GSSG
and superoxide with oxygen as the one-electron acceptor.
Molecular oxygen was shown to play a crucial role in the catalytic
cycle since the TON number was lower under the oxygen-
depletion condition and the generation of hydroxyl radicals
detected by EPR required the presence of oxygen. Under the
oxygen-depleted condition, the Ir-SG adducts readily decompose
to give the free H2azpy ligand and dinuclear thiolato-bridged
[(Cp*Ir)2(m-SG)3]+ (Fig. 12). In the presence of O2 as the one-
electron acceptor from the paramagnetic species [Int3], the Ir-
SG adducts readily form, and participate in the catalytic cycle,
thus leading to a higher catalytic efficiency (Fig. 12). ROS accu-
mulation in A549 lung cancer cells treated with complex 7
(Fig. 10), suggests that such reactions can occur inside cells
inducing the oxidative stress. Although the high catalytic activity
of complexes 7 and 8 probably contributes to their high cyto-
toxicity, these complexes are not the most cytotoxic, suggesting
that other pathways resulting in the release of reduced azopyr-
idine ligand and dinuclear thiolato-bridged complexes may play
important roles in the mechanism of anticancer activity.
This work suggests that iodido Ir(III) Cp* azopyridine
complexes have a unique multi-targeting anticancer mecha-
nism of action, which is potentially important for combatting
cisplatin resistance. The existence of both metal- and ligand-
centred reactions provides wide scope for the design of novel
organoiridium chemotherapeutic compounds.
Conflicts of interest
There are no conicts to declare.
Acknowledgements
This research was supported by the EPSRC (Grant No. EP/
P030572/1 and EP/F500378/1 MOAC studentship for G. M. H.),
CRUK/EPSRC (Grant No. C53561/A19933),
a Chancellor's
International PhD Scholarship from the University of Warwick
(for W.-Y. Z.), and a Royal Society Newton International
Fellowship (NF151429 for S. B.). H. E. B thanks Mike and Enfys
Bagguley, and the EPSRC (grant no. EP/F034210/1) for PhD
studentship support. We thank Dr Lijiang Song, Dr Ivan Prokes
and Ian Bagley (University of Warwick) for excellent technical
assistance with MS, NMR and zebrash embryo experiments,
respectively.
Conclusions
References
Iodido Ir(III) Cp* azopyridine complexes exhibit potent cytotoxic
activity towards cancer cells despite being relatively inert
1 D. Wang and S. J. Lippard, Nat. Rev. Drug Discovery, 2005, 4,
307–320.
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.

View Article Online
Chemical Science
Edge Article
´
2 L. Kelland, Nat. Rev. Cancer, 2007, 7, 573–584.
21 A. Groue, J.-P. Tranchier, M.-N. Rager, G. Gontard, M. Jean,
3 (a) A. Bijelic, M. Aureliano and A. Rompel, Angew. Chem., Int.
Ed., 2019, 58, 2980–2999; (b) M. R. Gill and K. A. Vallis, Chem.
Soc. Rev., 2019, 48, 540–557; (c) Z. Yu and J. A. Cowan, Chem.– 22 P. Starha, Z. Dvorak and Z. Travnıcek, J. Organomet. Chem.,
N. Vanthuyne, H. R. Pearce, A. L. Cooksy and H. Amouri,
Inorg. Chem., 2019, 58, 2930–2933.
ˇ
ˇ´
´ ´ˇ
Eur. J., 2017, 23, 14113–14127; (d) T. Zou, C. T. Lum,
2018, 872, 114–122.
C.-N. Lok, J.-J. Zhang and C.-M. Che, Chem. Soc. Rev., 2015, 23 F. Chen, J. Moat, D. McFeely, G. Clarkson, I. J. Hands-
44, 8786–8801; (e) L. Zeng, P. Gupta, Y. Chen, E. Wang,
L. Ji, H. Chao and Z.-S. Chen, Chem. Soc. Rev., 2017, 46,
Portman, J. P. Furner-Pardoe, F. Harrison, C. G. Dowson
and P. J. Sadler, J. Med. Chem., 2018, 61, 7330–7344.
5771–5804; (f) S. M. Meier-Menches, C. Gerner, W. Berger, 24 J. M. Cross, N. Gallagher, J. H. Gill, M. Jain, A. W. McNeillis,
C. G. Hartinger and B. K. Keppler, Chem. Soc. Rev., 2018,
47, 909–928; (g) J. Liu, H. Lai, Z. Xiong, B. Chen and
T. Chen, Chem. Commun., 2019, 55, 9904–9914.
K. L. Rockley, F. H. Tscherny, N. J. Wirszycz, D. S. Yut and
J. W. Walton, Dalton Trans., 2016, 45, 12807–12813.
˜
25 S. Sinha, S. Das, R. Sikari, S. Parua, P. Brandao,
4 (a) E. Boros, P. J. Dyson and G. Gasser, Chem, 2020, 6, 41–60;
(b) M. P. Sullivan, H. U. Holtkamp and C. G. Hartinger, Met.
S. Demeshko, F. Meyer and N. D. Paul, Inorg. Chem., 2017,
56, 14084–14100.
Ions Life Sci., 2018, 18, 351–386; (c) A. Erxleben, Curr. Med. 26 S. P. Rath, D. Sengupta, P. Ghosh, R. Bhattacharjee,
Chem., 2019, 26, 694–728; (d) Y. Wang, H. Li and H. Sun,
Inorg. Chem., 2019, 58, 13673–13685.
M. Chakraborty, S. Samanta, A. Datta and S. Goswami,
Inorg. Chem., 2018, 57, 11995–12009.
5 A. Cusanelli, U. Frey, D. T. Richens and A. E. Merbach, J. Am. 27 R. Pramanick, R. Bhattacharjee, D. Sengupta, A. Datta and
Chem. Soc., 1996, 118, 5265–5271. S. Goswami, Inorg. Chem., 2018, 57, 6816–6824.
6 Z. Liu and P. J. Sadler, Acc. Chem. Res., 2014, 47, 1174–1185. 28 X. Zhang, F. Ponte, E. Borfecchia, A. Martini, C. Sanchez-
7 Y. Geldmacher, M. Oleszak and W. S. Sheldrick, Inorg. Chim.
Acta, 2012, 393, 84–102.
Cano, E. Sicilia and P. J. Sadler, Chem. Commun., 2019, 55,
14602–14605.
´
8 E. Meggers, Angew. Chem., Int. Ed., 2017, 56, 5668–5675.
9 M. A. Scharwitz, I. Ott, R. Gust, A. Kromm and
W. S. Sheldrick, J. Inorg. Biochem., 2008, 102, 1623–1630.
29 J. M. Hearn, G. M. Hughes, I. Romero-Canelon, A. F. Munro,
B. Rubio-Ruiz, Z. Liu, N. O. Carragher and P. J. Sadler,
Metallomics, 2018, 10, 93–107.
10 R. Gao, D. G. Ho, B. Hernandez, M. Selke, D. Murphy, 30 L. Mimassi, C. Guyard-Duhayon, M. N. Rager and H. Amouri,
P. I. Djurovich and M. E. Thompson, J. Am. Chem. Soc.,
2002, 124, 14828–14829.
Inorg. Chem., 2004, 43, 6644–6649.
31 L. Mimassi, C. Cordier, C. Guyard-Duhayon, B. E. Mann and
H. Amouri, Organometallics, 2007, 26, 860–864.
32 Y. Fu, M. J. Romero, A. Habtemariam, M. E. Snowden,
L. Song, G. J. Clarkson, B. Qamar, A. M. Pizarro,
P. R. Unwin and P. J. Sadler, Chem. Sci., 2012, 3, 2485–2494.
33 S. H. van Rijt, A. J. Hebden, T. Amaresekera, R. J. Deeth,
G. J. Clarkson, S. Parsons, P. C. McGowan and P. J. Sadler,
J. Med. Chem., 2009, 52, 7753–7764.
11 L. He, Y. Li, C.-P. Tan, R.-R. Ye, M.-H. Chen, J.-J. Cao, L.-N. Ji
and Z.-W. Mao, Chem. Sci., 2015, 6, 5409–5418.
12 H. Amouri, J. Moussa, A. K. Renfrew, P. J. Dyson, M. N. Rager
and L.-M. Chamoreau, Angew. Chem., Int. Ed., 2010, 49,
7530–7533.
´
13 Z. Liu, I. Romero-Canelon, B. Qamar, J. M. Hearn,
A. Habtemariam, N. P. E. Barry, A. M. Pizarro,
G. J. Clarkson and P. J. Sadler, Angew. Chem., Int. Ed., 2014, 34 L. Li, W. W. Brennessel and W. D. Jones, J. Am. Chem. Soc.,
53, 3941–3946. 2008, 130, 12414–12419.
14 A. J. Millett, A. Habtemariam, I. Romero-Canelon, 35 Z. Liu, L. Salassa, A. Habtemariam, A. M. Pizarro,
´
G. J. Clarkson and P. J. Sadler, Organometallics, 2015, 34,
2683–2694.
G. J. Clarkson and P. J. Sadler, Inorg. Chem., 2011, 50,
5777–5783.
15 Y. Yang, L. Guo, Z. Tian, Y. Gong, H. Zheng, S. Zhang, Z. Xu, 36 M. Shivakumar, J. Gangopadhyay and A. Chakravorty,
X. Ge and Z. Liu, Inorg. Chem., 2018, 57, 11087–11098. Polyhedron, 2001, 20, 2089–2093.
16 Z. Liu, A. Habtemariam, A. M. Pizarro, S. A. Fletcher, 37 H. Karoui, N. Hogg, C. Frejaville, P. Tordo and
A. Kisova, O. Vrana, L. Salassa, P. C. A. Bruijnincx, B. Kalyanaraman, J. Biol. Chem., 1996, 271, 6000–6009.
G. J. Clarkson, V. Brabec and P. J. Sadler, J. Med. Chem., 38 S. Stoll and A. Schweiger, J. Magn. Reson., 2006, 178, 42–55.
´
2011, 54, 3011–3026.
17 K. Fagnou and M. Lautens, Angew. Chem., Int. Ed., 2002, 41,
26–47.
39 T. Nakayama, T. Isobe, K. Nakamiya, J. S. Edmonds,
Y. Shibata and M. Morita, Magn. Reson. Chem., 2005, 43,
543–550.
18 R. J. Needham, C. Sanchez-Cano, X. Zhang, I. Romero- 40 N. W. Pirie and K. G. Pinhey, J. Biol. Chem., 1929, 84, 321–333.
´
Canelon, A. Habtemariam, M. S. Cooper, L. Meszaros, 41 S. G. Tajc, B. S. Tolbert, R. Basavappa and B. L. Miller, J. Am.
G. J. Clarkson, P. J. Blower and P. J. Sadler, Angew. Chem.,
Int. Ed., 2017, 56, 1017–1020.
19 S. J. Dougan, A. Habtemariam, S. E. McHale, S. Parsons and
P. J. Sadler, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 11628–
11633.
Chem. Soc., 2004, 126, 10508–10509.
42 P. Skehan, R. Storeng, D. Scudiero, A. Monks, J. McMahon,
D. Vistica, J. T. Warren, H. Bokesch, S. Kenney and
M. R. Boyd, J. Natl. Cancer Inst., 1990, 82, 1107–1112.
43 J. Ruiz, C. Vicente, C. de Haro and D. Bautista, Inorg. Chem.,
2013, 52, 974–982.
´
20 I. Romero-Canelon, L. Salassa and P. J. Sadler, J. Med. Chem.,
2013, 56, 1291–1300.
Chem. Sci.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Edge Article
Chemical Science
44 T. V. Bowman and L. I. Zon, ACS Chem. Biol., 2010, 5, 159– 58 A. Meister, Trends Biochem. Sci., 1981, 6, 231–234.
161.
59 F. Wang, J. Xu, A. Habtemariam, J. Bella and P. J. Sadler, J.
Am. Chem. Soc., 2005, 127, 17734–17743.
60 L. Zeng, S. Kuang, G. Li, C. Jin, L. Ji and H. Chao, Chem.
Commun., 2017, 53, 1977–1980.
61 Q.-X. Zhou, Y. Zheng, T.-J. Wang, Y.-J. Chen, K. Li,
Y.-Y. Zhang, C. Li, Y.-J. Hou and X.-S. Wang, Chem.
Commun., 2015, 51, 10684–10686.
62 Y. Shi, S.-A. Liu, D. J. Kerwood, J. Goodisman and
J. C. Dabrowiak, J. Inorg. Biochem., 2012, 107, 6–14.
45 P. M. Eimon and A. L. Rubinstein, Expert Opin. Drug Metab.
Toxicol., 2009, 5, 393–401.
46 A. J. Rennekamp and R. T. Peterson, Curr. Opin. Chem. Biol.,
2015, 24, 58–70.
47 W. B. Barbazuk, I. Korf, C. Kadavi, J. Heyen, S. Tate, E. Wun,
J. A. Bedell, J. D. McPherson and S. L. Johnson, Genome Res.,
2000, 10, 1351–1358.
48 J. P. C. Coverdale, H. E. Bridgewater, J.-I. Song, N. A. Smith,
´
N. P. E. Barry, I. Bagley, P. J. Sadler and I. Romero-Canelon, 63 J. L. Sadler and A. J. Bard, J. Am. Chem. Soc., 1968, 90, 1979–
J. Med. Chem., 2018, 61, 9246–9255. 1989.
49 L. Helm and A. E. Merbach, Coord. Chem. Rev., 1999, 187, 64 W. Kaim, R. Reinhardt, S. Greulich and J. Fiedler,
151–181. Organometallics, 2003, 22, 2240–2244.
50 T. Poth, H. Paulus, H. Elias, C. Ducker-Benfer and R. van 65 L. Sun, G. Li, X. Chen, Y. Chen, C. Jin, L. Ji and H. Chao, Sci.
Eldik, Eur. J. Inorg. Chem., 2001, 1361–1369. Rep., 2015, 5, 14837.
51 C. X. Zhang and S. J. Lippard, Curr. Opin. Chem. Biol., 2003, 7, 66 W. G. Kirlin, J. Cai, S. A. Thompson, D. Diaz, T. J. Kavanagh
481–489. and D. P. Jones, Free Radical Biol. Med., 1999, 27, 1208–1218.
52 S. Parveen, M. Hanif, E. Leung, K. K. H. Tong, A. Yang, 67 F. Q. Schafer and G. R. Buettner, Free Radical Biol. Med.,
J. Astin, G. H. De Zoysa, T. R. Steel, D. Goodman, 2001, 30, 1191–1212.
S. Movassaghi, T. Sohnel, V. Sarojini, S. M. F. Jamieson 68 V. Khramtsov, L. J. Berliner and T. L. Clanton, Magn. Reson.
¨
¨
and C. G. Hartinger, Chem. Commun., 2019, 55, 12016–
12019.
53 M. J. McKeage, S. J. Berners-Price, P. Galettis, R. J. Bowen,
Med., 1999, 42, 228–234.
69 G.-Y. Li, J.-P. Liu, H.-Y. Huang, Y. Wen, H. Chao and L.-N. Ji,
J. Inorg. Biochem., 2013, 121, 108–113.
W. Brouwer, L. Ding, L. Zhuang and B. C. Baguley, Cancer 70 G. Li, Y. Chen, J. Wu, L. Ji and H. Chao, Chem. Commun.,
Chemother. Pharmacol., 2000, 46, 343–350. 2013, 49, 2040–2042.
54 L. Messori, G. Marcon, P. Orioli, M. Fontani, P. Zanello, 71 Y. Kenis, J. De Smedt and H. J. Tagnon, Eur. J. Cancer, 1966,
A. Bergamo, G. Sava and P. Mura, J. Inorg. Biochem., 2003,
95, 37–46.
55 F. Wang, A. Habtemariam, E. P. L. van der Geer,
2, 51–57.
72 S. H. Kuttab, S. Tanglertpaibul and P. Vouros, Biomed. Mass
Spectrom., 1982, 9, 78–84.
´
R. Fernandez, M. Melchart, R. J. Deeth, R. Aird, 73 G. Gupta, A. Garci, B. S. Murray, P. J. Dyson, G. Fabre,
¨
S. Guichard, F. P. A. Fabbiani, P. Lozano-Casal,
I. D. H. Oswald, D. I. Jodrell, S. Parsons and P. J. Sadler,
Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 18269–18274.
P. Trouillas, F. Giannini, J. Furrer, G. Suss-Fink and
B. Therrien, Dalton Trans., 2013, 42, 15457–15463.
74 A. H. Ngo and L. H. Do, Inorg. Chem. Front., 2020, 7, 583–591.
ˇ
´
´
56 Z. Liu, I. Romero-Canelon, A. Habtemariam, G. J. Clarkson 75 P. Starha, A. Habtemariam, I. Romero-Canelon,
and P. J. Sadler, Organometallics, 2014, 33, 5324–5333.
G. J. Clarkson and P. J. Sadler, Inorg. Chem., 2016, 55,
2324–2331.
¨
57 A. Wilbuer, D. H. Vlecken, D. J. Schmitz, K. Kraling,
K. Harms, C. P. Bagowski and E. Meggers, Angew. Chem., 76 G. Calabrese, B. Morgan and J. Riemer, Antioxid. Redox
Int. Ed., 2010, 49, 3839–3842.
Signaling, 2017, 27, 1162–1177.
This journal is © The Royal Society of Chemistry 2020
Chem. Sci.