The potential for spacial selectivity, as offered by photoactivation, together with
novel excited-state chemistry and accompanying mechanisms of action make exploration
of photoactivated metal chemotherapeutic complexes attractive for cancer therapy.1We
have been studying PtIV-diazidodihydroxido anticancer complexes, [Pt(N3)2(OH)2(Am1)(Am2)]
(Am1/Am2=am(m)ines),2 which, in the absence of light, exhibit minimal cytotoxicity
towards cancer cells and do not react with glutathione (GSH), 5′-guanosine monophosphate
(5′-GMP), or DNA in either cell-free media or aqueous solutions. By contrast, upon
irradiation with UVA or visible (blue/green) light, these complexes display potent
cytotoxicity towards a range of cancer cell lines. The reported photodecomposition
products include azide anions (N3
−), azidyl radicals (N3
.), nitrogen gas (N2), and oxygen gas (O2).3 A particularly potent photocytotoxic
anticancer complex is trans,trans,trans-[Pt(N3)2(OH)2(MA)(Py)] (1, MA=methylamine,
Py=pyridine).4 Herein we report the unprecedented oxidation of 5′-GMP by 1 upon irradiation
with UVA and the identification of some unexpected reaction pathways involving singlet
oxygen (1O2) and nitrene (Pt-N) intermediates. The source of the 1O2 was also investigated.
The photoreaction of 1 with 5′-GMP upon irradiation with blue light with a wavelength
of 450 nm for 1 h gave (SP-4-2)-[Pt(N3)(MA)(Py)(5′-GMP)]+ (1 a
+) as the major product and trans-[Pt(MA)(Py)(5′-GMP)2]2+ (1 b
2+) as the minor product (Figure 1 A). A similar result has been reported for 1 under
slightly different conditions, and for related compounds.2a, 4, 4 However, when the
reaction mixture was irradiated at 420 nm for 30 min, two new photoproducts, 1 c and
1 e, were observed by HPLC (Figure 1 B). When the sample was irradiated with UVA (365 nm)
for 15 min, one more species, 1 d, was found (Figure 1 C). The isotopic distributions
observed in the ESI-MS analysis revealed that the singly charged cations of 1 c (m/z=718.1),
1 d (m/z=684.1), and 1 e (m/z=700.1) all contain a Pt atom. A control experiment with
5′-GMP in the absence of 1 and UVA irradiation for 15 min showed no reaction (Figure 1 D),
thus suggesting that 5′-GMP is stable under these conditions. The photoreaction of
1 and 5′-GMP under an atmosphere of argon gave similar results, thereby excluding
the possibility that dissolved oxygen is the oxidant. Hence, 1 c, 1 d, and 1 e are
all produced by the photoreaction of 1.
Figure 1
Chromatograms for the photoreactions of 1 (0.67 mm) with 5′-GMP (1.0 mm) in aqueous
solution upon irradiation with A) 450 nm light, 50 mW cm−2, 60 min; B) 420 nm, 4.3 mW cm−2,
30 min; C) UVA (365 nm), 3.5 mW cm−2, 15 min; D) 5′-GMP (1.0 mm) only, UVA, 15 min.
High-resolution MS of 1 a, 1 c, 1 d, and 1 e, as well as their tandem MS (CID), were
performed to examine further the structure of each species (see Figures S1 and S2
in the Supporting Information). They are all assigned as PtII complexes, and their
structures are listed in Table 1, while fragment ions are listed in Tables S1–S4 in
the Supporting Information, together with their assignments.
Table 1
Positive ions for complexes 1 a, 1 c, 1 d, and 1 e observed by HRMS and the corresponding
assignments.
Found m/z
710.1205
718.1337
684.1228
700.1235
Proposed chemical structure
([1 a]+)
([1 c−H]+)
([1 d−H]+)
([1 e−H]+)
Theoretical m/z
710.1164
718.1314
684.1260
700.1209
Error [ppm]
5.8
3.2
4.7
3.7
Curiously, species 1 c, 1 d, and 1 e all have an NH3 ligand, which was not present
in the reactants. A reasonable source of the NH3 is the {Pt-N3} fragment, which can
lose N2 upon irradiation with light to form a {Pt-N} nitrene intermediate.6 This postulation
was verified by using 1*, azide trans,trans,trans-[Pt(N3
*)2(OH)2(MA)(Py)], where N3
*=[15N=14N=14N], and hence each bound N atom from the azide is now 50 % 15N. The photoreaction
of 1* with 5′-GMP, carried out under identical conditions, gave an identical chromatogram
as that shown in Figure 1 C. The ESI-MS spectra for 1 a*, 1 c*, 1 d*, and 1 e*, which
have the same retention times as 1 a, 1 c, 1 d, and 1 e, are shown in Figure S5 in
the Supporting Information. The molecular weight of compound 1 a* was 1 Da larger
than 1 a, thus suggesting that it has an intact N3
*
− ligand. The isotope distributions of 1 c*, 1 d*, and 1 e* indicate that they are
all 50 %/50 % mixtures of [M] and [M+1], so they are all considered as {Pt-NH3
*} fragments derived from {Pt-N3*} (NH3
*=50 % 15NH3/50 % 14NH3). This result suggests that N2 gas is released directly from
{Pt-N3}, thereby generating a {Pt-N} intermediate.
The release of N2 on photolysis of 1 was verified by 14N NMR spectroscopy (see Figure S6
in the Supporting Information); signals for free azide (N3
−) were also detected (see the Supporting Information for details). N2 may be released
directly from {Pt-N3} and also may be formed from the recombination of the azidyl
radicals (N3
.) generated in this photoreaction.7 The release of N3
. was confirmed by EPR spectroscopy by using 5,5-dimethylpyrroline-N-oxide (DMPO)
as the spin trap. Signals for the DMPO-14N3 spin adduct were detected (see Figure S7
in the Supporting Information), but hydroxyl radicals OH. were not trapped. It is
evident that 5′-GMP is not oxidized by N3
., as its presence did not affect the trapping of N3
. radicals (see the Supporting Information for details).
Species 1 e contains an 8-hydroxyguanine (8-OH-G) fragment, equivalent to 8-oxoguanine
(8-oxo-G), which is one of the most common products of DNA oxidation.8 The possibility
that the oxidation of guanine by photoactivated 1 involved singlet oxygen or nitrene
intermediates was investigated. Product 1 c has a similar structure as 1 e, but the
8-OH-G is replaced by RedSp (N-formylamidoiminohydantoin),8a hydrolyzed 8-OH-G (shown
in Scheme S1 in the Supporting Information).
During the photoreaction of 1 and 5′-GMP with irradiation at 450 nm the yellow color
of the solution became darker, and gas bubbles formed (see Figure S8 in the Supporting
Information). Similar results were obtained when the reaction was triggered with UVA.
The evolution of the gases O2 and N2 was verified by GC-MS performed in 18O-labeled
water under an argon atmosphere, with the aim also of examining the source of the
O atoms in the generated O2. After the photolysis of 1, the gas phase was analyzed
by GC-MS (see Figure S9 and the Supporting Information for details), and 16O2 (m/z=32)
and N2 (m/z=28) were both found. No 18O-substituted O2 was detected. This result verified
the release of N2 and O2 and, moreover, provided evidence that the two oxygen atoms
in the generated O2 are both from 1 rather than from the solvent.
The nature of the released oxygen was investigated using a fluorescence probe for
singlet oxygen: SOSG. SOSG is a highly selective sensor for 1O2 without any appreciable
response to hydroxyl radicals or superoxide.9 In the absence of 1O2, SOSG exhibits
low fluorescence, but in the presence of 1O2, strong green fluorescence can be observed
with λ
ex=504 nm and λ
em=525 nm. Solutions containing 1 and SOSG were stable in the dark or even upon irradiation
at λ≥504 nm. However, when exposed to weak irradiation at 365 nm (21 μW cm−2), the
intensity of the fluorescence at 525 nm increased rapidly, thus indicating that 1O2
was generated (Figure 2). The dose-dependent efficiency of generating 1O2 upon irradiation
was higher the shorter the wavelength (UVA>420 nm>450 nm, see Figure S10 in the Supporting
Information). Control experiments carried out in the dark or in the absence of 1 showed
no change in the fluorescence intensity (see Figure S10 in the Supporting Information).
A sample saturated with argon was irradiated at 365 nm and gave stronger fluorescence
(Figure 2). This result revealed that the 1O2 was not generated from the dissolved
O2 through energy transfer from a photosensitizer. N2 had a similar effect as argon.
Neither argon nor N2 itself could trigger the fluorescence of SOSG. The release of
singlet oxygen from a PtIV-diazidodihydroxido complex upon irradiation with light
in the absence of any exogenous source of oxygen gas appears to be unprecedented.
Figure 2
Time-dependent fluorescence (λ
ex/λ
em=504/525 nm) from 1 (50 μm) and SOSG (1 μm) in H2O (3 % MeOH) upon weak irradiation
at 365 nm (21 μW cm−2) (293 K). ▪: no additive; □: 50 % D2O; ▴: saturated with argon;
•: 0.1 mm l-ascorbic acid (AscA). All the data points were the average of 2–4 independent
experiments.
The lifetime of 1O2 in D2O is known to be much longer than that in H2O.10 Indeed,
the fluorescence intensity arising from the reaction of 1 with SOSG in 50 % D2O with
λ
irr=365 nm was three- to fourfold higher than that of the reaction carried out in
H2O alone (Figure 2). The reaction was repeated in the presence of 100 μm l-ascorbic
acid (AscA) as a 1O2 scavenger. Complex 1 does not react with AscA in the absence
of light, even though AscA is a strong reductant. However, upon irradiation at 365 nm,
the fluorescence was totally quenched (Figure 2). These results again confirmed the
generation of 1O2.
Guanine and the other nucleobases may be oxidatively damaged by reactive oxygen species
(ROS), radicals, and ionizing/UVA radiation.11 The oxidation of guanine usually leads
to DNA damage so as to cause lethality, aging, and mutagenicity.8 Although it has
been reported that several PtIV-tetrachlorido complexes can directly oxidize guanine,12
we report here for the first time that the photodecomposition of a PtIV-diazidodihydroxido
complex can oxidize guanine. Complexes containing PtII and oxidized guanine as 8-OH-G
and RedSp were detected (Table 1).
Two pathways can be proposed for the oxidation of guanine. The first involves a nitrene
intermediate (Scheme 1, Mechanism 1). This can arise from loss of N2 from the {Pt-N3}
fragment upon irradiation with UVA.6 Two electrons are transferred from guanine to
the nitrene, and the guanine itself is oxidized to give 8-OH-G on addition of H2O.
The nitrene is reduced and finally forms the ammine adduct {Pt-NH3}. Another possible
oxidant is singlet oxygen. The 1O2 generated by photolysis of 1 can oxidize guanine
to 8-OH-G by a direct [4+2] cycloaddition (Scheme 1, Mechanism 2).8a
Scheme 1
Two possible mechanisms for the oxidation of 5′-GMP. Charges are omitted for clarity.
Singlet oxygen (1O2) is a highly reactive and toxic species in biological systems.
It is considered the principal antiproliferative species in photodynamic therapy (PDT),
in which it reacts with many biomolecules, including nucleic acids, proteins, and
lipids, thereby causing cancer cell death.8, 13 A common method to generate 1O2 is
photosensitization, which produces 1O2 by energy transfer from a photoexcited sensitizer
to ground-state triplet oxygen (3O2). This strategy is the basis of current PDT, but
requires the presence of oxygen at the target site. However, tumor cells are often
hypoxic. In contrast, the generation of 1O2 from trans,trans,trans-[Pt(N3)2(OH)2(MA)(Py)]
upon irradiation with UVA/blue light does not require any exogenous source of oxygen
gas (Figure 2). This feature may be beneficial for the potential clinical application
of 1 and killing of hypoxic cancer cells. Moreover, the wavelength can be tuned to
control its photocytotoxicity, as shorter wavelengths generate more 1O2.
Since neither of the oxygen atoms in the released 1O2 originate from water, the most
plausible source is the OH groups of 1. Furthermore, solvent substitution at PtIV
is not likely to occur before its reduction to PtII. A possible photolysis mechanism
is given in Scheme 2. Upon irradiation with short-wavelength light, such as UVA, photodecomposition
of 1 in the presence of 5′-GMP may occur through two pathways. In the first, two azidyl
radicals (N3
.) are lost and PtIV is reduced to PtII. The OH groups are protonated and thus are
substituted by 5′-GMP to give product 1 b. The second pathway is more complicated,
but perhaps more likely. The integration of the HPLC peaks in Figure 1 C suggested
that about 92 % of the Pt species were produced through this pathway. Complex 1 loses
one azide ligand (N3
−) and the OH ligands donate one electron each to reduce the PtIV to PtII; they then
rapidly recombine to generate oxygen gas, as singlet oxygen, through H2O2 formation.14
Then the binding of 5′-GMP to Pt produces 1 a, which can be further transformed to
1 e by loss of N2 gas from the N3 ligand, as in Scheme 1. Alternately, if N2 is expelled
from the N3 ligand before 5′-GMP binds, 1 d is produced, which can also be transformed
to 1 e by reacting with 1O2 (Scheme 1).
Scheme 2
Possible mechanisms for the photoreaction of 1 with 5′-GMP upon irradiation with UVA.
Species in square brackets are unstable intermediates. Charges are omitted for clarity.
Species percentages are average HPLC integrations for four experiments with UVA irradiation
(Figure 1 C).
There are a number of reports of the chemical reduction of PtIV to PtII, and it is
widely accepted that a concerted two-electron transfer from, for example, ascorbate,
GSH, or guanine, to PtIV is involved.12, 15 However, the photoreduction of 1 may not
follow the above pathway. PtIV is more likely to gain one electron from each of the
two N3 or two OH ligands and give rise to N3
. or OH. radicals, respectively. We were not able to trap OH. radicals, perhaps because
their lifetime is too short.
The in situ formation of an NH3 ligand can give rise to potential hydrogen-bonding
interactions with DNA. DNA adducts of the type 1 d, for example, may strongly inhibit
RNA polymerase II and nucleotide excision repair.16 Nitrenes are highly reactive intermediates
and are reported to be responsible for a wide range of DNA lesions.17 N3
. is a relatively mild and selective oxidant that can oxidize amino acids such as
tryptophan.3b N3
− is a mitochondrial inhibitor, and a myeloperoxidase and catalase inhibitor. These
species, together with 1O2, could all contribute to the potent photo-antiproliferative
effect of 1 on cancer cells.
In summary, we observed the unexpected oxidation of guanine during the photoreaction
of complex 1 (trans,trans,trans-[Pt(N3)2(OH)2(MA)(Py)]) with 5′-GMP. The photodecomposition
of 1 involves Pt-nitrene intermediates and formation of singlet oxygen, free azide,
azidyl radicals, and nitrogen gas. The oxidation of guanine is likely to arise from
reactions of singlet oxygen and nitrene intermediates. The generation of singlet oxygen
in the absence of oxygen gas and the oxidative damage to guanine may contribute to
the potent photocytotoxic effects of this complex.