Organoiridium Photosensitizers Induce Specific Oxidative Attack on Proteins within Cancer Cells

Abstract Strongly luminescent iridium(III) complexes, [Ir(C,N)2(S,S)]+ (1) and [Ir(C,N)2(O,O)] (2), containing C,N (phenylquinoline), O,O (diketonate), or S,S (dithione) chelating ligands, have been characterized by X‐ray crystallography and DFT calculations. Their long phosphorescence lifetimes in living cancer cells give rise to high quantum yields for the generation of 1O2, with large 2‐photon absorption cross‐sections. 2 is nontoxic to cells, but potently cytotoxic to cancer cells upon brief irradiation with low doses of visible light, and potent at sub‐micromolar doses towards 3D multicellular tumor spheroids with 2‐photon red light. Photoactivation causes oxidative damage to specific histidine residues in the key proteins in aldose reductase and heat‐shock protein‐70 within living cancer cells. The oxidative stress induced by iridium photosensitizers during photoactivation can increase the levels of enzymes involved in the glycolytic pathway.

Tables S3-4. Selected geometrical parameters for 1 isomers in the singlet and triplet states. Table S5. TD-DFT one-photon excitation data for 1 (CC isomer). Table S6. TD-DFT one-photon excitation data for 2 (CC isomer). Table S7. Photophysical data for the complexes. Table S8. (Photo)cytotoxicity of the compounds towards 2D and 3D cells. Table S9. T-test results for lysozyme peptide.                  The TD calculations were run for the first 30 singlet and triplet states. Tables S5 and S6 show the excitation energies, oscillator strengths, and characterization for the lowest excited states for the CC isomers of 1 and 2. Figs. S3 and S4 show a simulated one-photon absorption spectrum for the CC isomers of 1 and 2, while Fig. S5 shows canonical particle-hole orbital characterization of the lowest singlet and triplet states of 1, as obtained from the response eigenvectors. Two-photon absorption was calculated using the 3-and 4-state sum-over-states models of Ågren et al., [4] using the linear response ground to excited transition moments, and the approximate excited to excited transition moments obtained from the a posteriori Tamm-Dancoff approximation (ATDA) [5,6] . Fig. S8 shows the dominant canonical particle-hole orbital representation of the strong TPA state (S7) of 1. All calculations were performed using the Gaussian09 program suite [7] .

Determination of 2-Photon Absorption Cross
Sections. The 2-photon absorption spectra of probes were determined over a broad spectral range by a typical 2-photon induced luminescence (TPL) method using Rhodamine B in methanol as a standard. The 2-photon luminescence data were acquired using an Opolette TM 355II (pulse width ≤ 100 fs, 80 MHz repetition rate, tuning range 720-800 nm, Spectra Physics, Inc., USA). 2-photon luminescence measurements were performed in fluorometric quartz cuvettes. The experimental luminescence excitation and detection conditions were conducted with negligible reabsorption processes, which can affect the TPA measurements. The quadratic dependence of 2-photon-induced luminescence intensity on the excitation power was verified at an excitation wavelength of 808 nm. The 2-photon absorption cross section of the probes was calculated at each wavelength according to Eq. (1): where I is the integrated luminescence intensity, C is the concentration, n is the refractive index, and  is the quantum yield. Subscript '1' indicates reference samples, and '2' indicates experimental samples.
Quantum Yields of Singlet Oxygen. The quantum yields of singlet oxygen were determined using two different methods [8] .

Indirect
Method. An air-saturated PBS buffer solution, containing the complex (OD = 0.1 at irradiation wavelength), p-nitrosodimethyl aniline (RNO, 20 M), histidine (10 mM) were irradiated with blue light in a quartz cuvette for different time intervals. The absorbance of the solution was then recorded. Plots of variations in absorbance at 440 nm in PBS (A0-A, where A0 is the absorbance before irradiation) versus the irradiation times for each sample were S5 prepared and the slope of the linear regression was calculated (Ssample). As a reference compound, [Ru(bpy)3] 2+ (ref ( 1 O2) = 0.22 in H2O) was used in both methods, to obtain Sref. Equation (2) was applied to calculate the singlet oxygen quantum yields (sample) for every sample: I (the absorbance correction factor) was obtained using Equation ( Field modulation was applied at 100 kHz and 0.05 mT, and the microwave attenuation was 18 dB (~3.2 mW). The spin trap, 2,2,6,6-tetramethyl-piperidine (TEMP for trapping 1 O2, 20 mM), was used to verify the formation of 1 O2 generated by the iridium complexes (100 μM).
Monolayer Cell Culture. The cells were grown in RPMI-1640 with or without phenol red (for photoactivable experiment). All media were supplemented with 10% v/v of fetal calf serum (FCS), 1% v/v of 2 mM glutamine and 1% v/v penicillin/streptomycin. All cells were grown as adherent monolayers at 310 K in a 5% CO2 humidified incubator and passaged regularly at approx. 80% confluence. Photocytotoxicity for Monolayer Cells. The photocytotoxicity was determined in A549 lung cancer cells and normal MRC-5 lung cells. Approximately 510 3 cells/well were seeded into two 96-well plates ('dark' and 'light' plates), followed by 24 h incubation for attachment. All cells were then exposed to the iridium complexes with different concentrations. After 2 h incubation, each well of both plates was washed with phosphate-buffered saline (PBS), and fresh medium was added into the wells. Cells of irradiated plate were then irradiated (465 nm, 4.8 mW J/cm 2 , 10 min). After irradiation, cell incubation was continued for another 46 h. The 'dark' plate was treated similarly but without irradiation. The photocytotoxicity was measured by the standard MTT method. The change in optical density (OD) at 540 nm was monitored using a microplate reader (Promega).

2-Photon Photocytotoxicity for Multicellular Tumor Spheroids.
MCTSs were cultured using the liquid overlay method. A549 and MRC cells in the exponential growth phase were dissociated by a trypsin/EDTA to provide single-cell suspensions. 10,000 diluted cells were transferred to U-shaped 96-well plates in 200 L of medium. The single cells generated spheroids ca. 400 μm in semidiameter at day 3 in 5% CO2 at 37°C. After formation, each spheroid in the 96-well plate was imaged with a phase contrast microscope (10× objective, Zeiss, Germany) to monitor its integrity and semidiameter.
The spheroid-containing medium was replaced carefully with drug-supplemented medium (iridium complexes, ALA or cisplatin) using an eight-channel pipette. Four spheroids were treated per condition and drug concentration, and the DMSO volume was less than 1% (v/v).
After incubation in the dark for 2 h, the medium was replaced by fresh medium. The MCTSs in 'light' plate were exposed to irradiation (465 nm, 2.88 J/cm 2 or 750 nm, 10 J/cm 2 , 100 fs), then incubation of the MCTSs continued for another 46 h. The 'dark' plate was under the same S7 treatment except irradiation. The cytotoxicity of iridium complexes toward MCTSs was measured by ATP concentration with CellTiter-Glo® 3D Cell Viability Assay (Promega).
Briefly, an equivalent volume of CellTiter-Glo® 3D reagent was added to each MCTS sample in 96-wells plate; after 5 min shaking and 25 min incubation, the intensity of chemiluminescence was recorded on the Promega microplate reader. Nano-LC separations were achieved using an EASY nano-LC II system (Proxeon, Hemel Hempstead, UK) with a home-made 18 cm, C18 reverse phase (RP) nano capillary column (75 µm, 5 µm particle size) and a 3 cm C18 RP pre-column (150 µm, 5 µm particle size). Separation of tryptic cell digest was achieved using an acidified water/acetonitrile gradient from 5% ACN to 30% ACN over 120 minutes, followed by a second gradient of 15 minutes from 30% ACN to 80% ACN. Finally, with a 35-min wash of 80% ACN at a constant 600 nL/min flow rate. Singlet-oxygen-induced oxidation of proteins is complicated as discussed in several reviews [9,10] , being the result of direct attack on amino acids and secondary attack by initially formed peroxides. On the basis of previous reports, the following modifications were included in the search:
In addition, the fixed carbamidomethyl modification of Cys (from alkylation) and variable deamidation modifications of Asn, Gln and phosphorylation of Ser, Thr and Tyr were included.
Quantification of peptides was made by spiking with a lysozyme peptide in 1:6 lysozyme peptide:cell digest volume ratio (lysozyme peptide 0.01 µg/µL, cell digest 0.2 µg/µL) into each sample during the LC-MS/MS runs, and by determining the ratio between the cell peptide of interest and the specific lysozyme peptide. The differences between the ratios from the control sample set and the drug treated sample set were compared. T tests and p values were determined as measures of the level of significance between the different samples.
Pathway Analysis. Peptides that were detected 3 times or more out of the 5 replicates of data were selected for quantification of pathway analysis. The protein entry lists were imported and cross-matched with DAVID Bioinformation Database, then a list of related pathways was generated and exported by KEGG database [11][12][13][14] . Glycolysis, which was found to have the most number of protein counts and lowest Fisher Exact p-values were chosen for quantification using the methodology described above. The weighted area ratio was calculated using the equation shown below for proteins with multiple identified peptides. This provided a more comprehensive measure of the change in abundance under different conditions. S9 F = Fold of change of a peptide; a = average area ratio among the 5 replicates of data sets S10 Tables   Table S1. Crystal data and structure refinement for 2. The stability of the triplets follows that of the lowest singlet states: the CC isomer is the most stable, NC 6.07 kJ mol -1 higher, and NN 37.92 kJ mol -1 higher.