Photophysical and Photobiological Properties of Dinuclear Iridium(III) Bis-tridentate Complexes

Abstract

A series of cationic dinuclear iridium(III) complexes (Ir1 - Ir5) bearing terpyridine-capped fluorenyl bridging ligand and different polypyridyl or cyclometalating terminal tridentate ligands were synthesized, characterized, and evaluated for their photophysical and photobiological activities. The influence of the bridging and terminal ligands on the photophysical properties of the complexes was investigated by UV-vis absorption, emission, and transient absorption spectroscopy, and simulated by TDDFT calculations. All of the complexes displayed strong bridging-ligand localized visible ¹π,π* absorption and red- or near-infrared (NIR) phosphorescence as well as broad triplet excited-state absorption across both visible and NIR wavelengths. These triplet states were assigned as predominantly ³π,π* for Ir1 (τ = 3.1 μs) and Ir4 (τ = 48 μs), and predominantly ³CT (charge transfer) for Ir2, Ir3 and Ir5 (τ = 1.7–2.7 μs). Complexes Ir1 – Ir5 acted as in vitro photodynamic therapy (PDT) agents toward human SK-MEL-28 melanoma cells when activated with visible light, with submicromolar photocytotoxicity and phototherapeutic indices (PIs) ranging from 20 to almost 300. The in vitro PDT effects with visible light did not correlate with singlet oxygen (¹O₂) quantum yields or DNA photocleaving capacity probed under cell-free conditions. All of the Ir(III) complexes phosphoresced brightly when associated with compromised cells (with or without a light treatment) and exhibited photoactivated cellular uptake, highlighting the theranostic potential of this new class of Ir(III) complex photosensitizers.

Graphical Abstract

Dinuclear iridium(III) complexes bearing terpyridine-capped fluorenyl bridging ligand and different polypyridyl or cyclometalating terminal tridentate ligands exhibited in vitro photodynamic therapeutic effects toward SK-MEL-28 melanoma cells upon visible light activation, with submicromolar photocytotoxicity and phototherapeutic indices ranging from 20 to almost 300. They phosphoresced brightly when associated with compromised cells and exhibited photoactivated cellular uptake, highlighting the theranostic potential of these complexes. Both the bridging and terminal ligands impacted the photophysical and photobiological activities significantly.

INTRODUCTION

In recent decades transition-metal complexes have come to the forefront in the search for new chemical entities in drug discovery and biological chemistry.¹ Platinum (Pt)-based metal complexes have been investigated extensively for cancer therapy,² with cisplatin being arguably the most successful anticancer drug to date. Nevertheless, there is continued focus on developing alternatives to cisplatin^3,4 and other nonselective cytotoxic agents in an effort to reduce the systemic side effects associated with conventional chemotherapy approaches. Ruthenium (Ru)-containing compounds have been widely studied as alternatives to the Pt-derived drugs, with a few (e.g., NAMI-A, KP1019, and IT-139) entering clinical trials^5–7 but none in the clinic to date. With selectivity being a key consideration for any new drug, Ru(II) coordination complexes and other transition-metal complexes are also being considered for targeted modalities such as photodynamic therapy (PDT).^8–11

PDT has been known for over a hundred years¹² yet remains underexploited in mainstream cancer therapy despite its precise spatiotemporal selectivity. In its narrowest definition, PDT involves activation of an otherwise nontoxic photosensitizer with photons of appropriate energy to produce a triplet excited state that can relax through singlet oxygen (¹O₂) sensitization or the production of other reactive oxygen species (ROS).¹³ Collectively, cytotoxic ROS destroy tumors and tumor vasculature, and can even invoke an antitumor immune response under the right conditions. Porphyrin-based organic compounds (and the related chlorins, bacteriochlorins, and phthalocyanines) have traditionally served as ROS-generating photosensitizers for PDT.^14,15 However, transition-metal complex photosensitizers have the potential to both (i) expand the scope of PDT to include oxygen-independent mechanisms of action, and thus improve PDT efficacy in hypoxic tissue, and (ii) broaden the wavelength range that can be used to include deeper tissue-penetrating near-infrared (NIR) light. One recent example is our Ru(II) complex TLD1433,¹⁶ currently in clinical trials for treating bladder cancer with PDT (ClinicalTrials.gov Identifier: NCT03053635), and its transferrin conjugate Rutherrin™.¹⁷

TLD1433 incorporates a special π-expansive ligand derived from imidazo[4,5-f][1,10]phenanthroline appended to an α-terthienyl unit, which imparts a long-lived triplet intraligand excited state (³IL) of significant π,π* character that is lower in energy than the triplet metal-to-ligand charge transfer (³MLCT) state that typically dominates the photophysical dynamics of Ru(II) polypyridyl complexes. The reduced intersystem crossing (ISC) rates characteristic of spin-forbidden ³π,π* transitions centered on the organic moiety are responsible for the much longer intrinsic triplet lifetimes in these constructs that are known as metal-organic dyads.^18,19 The implication is that slow ISC back to the ground state from ³IL states provides ample opportunity for requisite bimolecular processes that constitute the phototoxic effects of PDT. We have demonstrated very potent in vitro PDT effects from a variety of Ru(II) dyads with low-lying, long-lived ³IL states, including both contiguously-fused π-extended azaaromatic ligands²⁰ as well as ligands tethered to π-extended aromatic chromophores.^16,21–23 More recently, we have incorporated π-expansive ligands into Ir(III) metal complexes to extend intrinsic excited state lifetimes for both reverse saturable absorption (RSA) and PDT applications.^24,25

Compared to the large number of Ru(II) systems that have been explored, investigations on Ir(III) complexes for in vitro PDT have been emerged in recent years.^24–44 While Ir(III) complexes derived from diimine ligands (N^∧N) may fall short of the optimal absorption window for PDT, we have demonstrated that incorporation of two cyclometalating ligands (C^∧N) to form biscyclometalated Ir(III) complexes extends the ground-state absorption spectrum significantly, and that installation of a π-extended N^∧N ligand alongside the two C^∧N ligands extends the lifetime through population of ³IL states. The result is nanomolar PDT potency with phototherapeutic indices (PIs) greater than 400.²⁵ We also showed that it was possible to use π-extended C^∧N ligands without compromising the potent in vitro PDT effects in Ir(III)-based systems that act as near-infrared-emitting theranostic agents.²⁴

The large luminescence quantum yields for certain Ir(III) complexes combined with their high yields for triplet state formation and good photostability make these metal complexes very attractive for photobiological applications such as PDT. In addition, their excitation and emission energies, as well as other photophysical and biological/chemical properties, can be systematically tuned via minor structural changes to a highly versatile and modular architecture. Despite these attributes, Ir(III) complexes studied to date³⁰ still fall short of some of the best Ru(II)-based in vitro PDT agents. The purpose of the present study is to use rational design principles to improve the water-solubility and in vitro PDT effects within the Ir(III) class of photosensitizers. Specifically, installing two Ir(III) centers in a single complex might simultaneously amplify photocytotoxicity and increase water solubility.

The solubility of organometallic complexes in aqueous solution can be improved by increasing the number of charges in the complex, which should be applicable to multinuclear iridium(III) complexes with C^∧N and/or N^∧N ligands.⁴⁵ Since the photophysical and biological properties of metal complexes bearing tridentate ligand(s) can be readily tuned by modification of the 4′-position of the tridentate ligand(s), and because the bis-tridentate ligand coordination prevents the formation of stereoisomers upon complexation with transition metals, tridentate ligands are chosen for this Ir(III) study.^46,47 Our previous work with bis-terpyridyl dinuclear Pt(II) complexes showed that the fluorenyl bridging group imparted these systems with intense absorption in the visible region (400-500 nm) and reasonably long-lived triplet excited states.⁴⁸ These desirable properties led us to investigate fluorenyl-linked Ir(III) systems as in vitro PDT agents.

Herein, we report the synthesis, characterization, and photophysical/photobiological properties of stable dinuclear Ir(III) complexes (Chart 1) of +2, +4, or +6 charges, with the charge determined by the identities of the metal coordinating atoms of different terminal tridentate ligands. Fluorene was chosen as the central bridging group for the two Ir-tpy components because it is a rigid π-conjugated linker expected to enhance molar extinction coefficients in the visible spectral region. Complexes Ir1 and Ir3 – Ir5 incorporate 9,9-dioctyl-2,7-di(terpyridyl)-9H-fluorene (L1) as the bridging ligand, with variation at the terminal tridentate ligands: 4′-phenyl-2,2′:6′,2″-terpyridine (N^∧N^∧N), 1,3-dipyridyl-4,6-dimethylbenzene (N^∧C^∧N), 4,6-diphenyl-2,2′-bipyridine (C^∧N^∧N), or 2,4,6-triphenyl-pyridine (C^∧N^∧C). Complex Ir2 uses 9,9-dioctyl-2,7-bis(2-phenylethynyl)-9H-fluorene (L2) as the bridging ligand to extend the π-conjugation length, which is anticipated to facilitate intraligand charge transfer (ILCT) transitions that fall in the PDT window and to also increase visible wavelength absorption.

Chart 1.

The molecule structure of target dinuclear Ir(III) complexes

EXPERIMENTAL SECTION

Synthesis and Characterizations.

Chemicals and solvents were purchased from Sigma-Aldrich or Alfa Aesar and used as received unless otherwise noted. The bridging ligands (L1 and L2) were synthesized following our previously reported procedures.⁴⁸ 4′-Phenyl-2,2′:6′,2″-terpyridine (N^∧N^∧N),⁴⁹ 1,3-dipyridyl-4,6-dimethylbenzene (N^∧C^∧N),⁵⁰ 4,6-diphenyl-2,2′-bipyridine (C^∧N^∧N),⁵¹ 2,4,6-triphenylpyridine (C^∧N^∧C),⁵² 4′-phenyl-2,2′:6′,2″-terpyridine-IrCl₃ (N^∧N^∧N-IrCl₃),⁵³ and {[1,3-dipyridyl-4,6-dimethylbenzene]IrCl₂}₂ (N^∧C^∧N Ir-dimer)⁵⁴ were synthesized according to established methods. Silica gel (60 Å, 230–400 mesh) and Al₂O₃ (activated, neutral) were purchased from Sorbent Technology. Complexes Ir1 – Ir5 were characterized by ¹H NMR, high resolution electrospray ionization mass spectrometry (ESI–MS), and elemental analysis. ¹H NMR spectra were recorded on Bruker-400 or Varian Oxford–500 spectrometers. ESI–MS analyses were collected using a Waters Synapt G2-Si Mass Spectrometer. Elemental analyses were performed by NuMega Resonance Laboratories, Inc. (San Diego, California). The synthetic schemes for complexes Ir1 – Ir5 are outlined in Scheme 1.

Scheme 1.

Synthetic routes for complexes Ir1 - Ir5.

L1-(IrCl₃)₂.

L1 (170.5 mg, 0.2 mmol) and IrCl₃ 3H₂O (141 mg, 0.4 mmol) were combined in degassed ethylene glycol (15 mL) and heated to 160 °C with protection from light. After 30 min, the reaction mixture was cooled to room temperature and filtered. The resulting precipitate was washed with ethanol, water, and diethyl ether to give Ll-Ir(Cl₃)₂ as a red solid (254 mg, 88%). ¹H NMR (500 MHz, d⁶-DMSO) δ 9.27 (d, J = 5.0 Hz, 4H), 9.18 (t, J = 8.8 Hz, 4H), 9.06 (dd, J = 11.3, 3.0 Hz, 2H), 8.98 – 8.94 (m, 2H), 8.70 (t, J = 9.8 Hz, 2H), 8.52 (dd, J = 8.2, 4.1 Hz, 2H), 8.38 (t, J = 7.8 Hz, 4H), 8.32 (d, J = 8.5 Hz, 2H), 8.06 – 8.01 (m, 4H), 2.43 (m, 4H), 0.95 – 0.69 (m, 20H), 0.48 (m, 10H).

Ir1.

4′-Phenylterpyridine-IrCl₃ (60.7 mg, 0.1 mmol) and L1 (42.6 mg, 0.05 mmol) were combined in degassed ethylene glycol (10 mL) and heated to 196 °C with protection from light. After 2 h, the reaction mixture was cooled to room temperature, and saturated NH₄PF₆ aqueous solution (20 mL) was added to precipitate the product. The crude material was purified by column chromatography on alumina gel. Unreacted ligand was eluted first with CH₂Cl₂, followed by elution of the desired product with an acetone/water gradient (100:0 to 95:5 (v/v)). The product was a yellow solid (60 mg, 65%). ¹H NMR (400 MHz, d⁶-DMSO) δ 9.76 (s, 4H), 9.67 (s, 4H), 9.50 – 9.34 (m, 4H), 9.29 (d, J = 8.2 Hz, 4H), 9.05 (s, 2H), 8.88 – 8.78 (m, 2H), 8.53-8.39 (m, 14H), 8.17 – 7.93 (m, 8H), 7.87 (t, J = 7.6 Hz, 4H), 7.78 (t, J = 7.4 Hz, 2H), 7.62 (dd, J = 12.2, 5.8 Hz, 8H), 2.61 (m, 4H), 0.89 (m, 20H), 0.62 (m, 10H). ESI-HRMS (m/z, in acetone): calcd. for [C₁₀₁H₉₀Ir₂N₁₂]⁶⁺, 309.4445; found, 309.4446. Calcd. for [C₁₀₁H₉₀Ir₂N₁₂PF₆]⁵⁺, 400.3262; found, 400.3270. Calcd. for [C₁₀₁H₉₀Ir₂N₁₂P₂F₁₂]⁴⁺, 536.6489; found, 536.6503. Anal. Calcd. (%) for C₁₀₁H₉₀F₃₆Ir₂N₁₂P₆·6H₂O: C, 42.80; H, 3.63; N, 5.93. Found: C, 42.61; H, 3.74; N, 5.98.

Ir2.

4′-Phenylterpyridine-IrCl₃ (60.7 mg, 0.1 mmol), L2 (45 mg, 0.05 mmol), and AgOTf (77 mg, 0.3 mmol) were combined in degassed ethylene glycol (10 mL) and heated to 196 °C with protection from light. After 24 h, the mixture was cooled to room temperature, and saturated NH₄PF₆ aqueous solution (20 mL) was added to precipitate the product. The crude material was purified by column chromatography on alumina gel. Unreacted ligand was eluted with CH₂Cl₂ first, followed by elution of the desired product using an acetone/water gradient (100:0 to 95:5 (v/v)). The purified product was a dark red powder (37 mg, 24%). ¹H NMR (400 MHz, d⁶-DMSO) δ 9.62 (t, J = 7.0 Hz, 6H), 9.22 (m, 8H), 8.48 (m, 6H), 8.32 (m, 8H), 7.92 (m, 8H), 7.77 (m, 6H), 7.71 (m, 4H), 7.51 (m, 10H), 2.51 (m, 4H), 1.02 (m, 20H), 0.67 (m, 10H). ESI-HRMS (m/z, in acetone): calcd. for [C₁₀₅H₉₀Ir₂N₁₂]⁶⁺, 317.4445; found, 317.4442. Anal. Calcd. (%) for C₁₀₅H₉₀F₃₆Ir₂N₁₂P6·13H₂O·CH₂Cl₂: C, 41.16; H, 3.85; N, 5.43. Found: C, 40.83; H, 3.47; N, 5.44.

Ir3.

{[2,2′-(4,6-Dimethyl-1,3-phenylene)bis-pyridine]IrCl₂}₂ (72 mg, 0.069 mmol) and L1 (59 mg, 0.069 mmol) were heated in degassed ethylene glycol (10 mL) at 196 °C with protection from light. After 1 h, the reaction mixture was added to water (10 mL) and filtered. Saturated NH₄PF₆ aqueous solution (20 mL) was added to the filtrate, and the resulting yellow precipitate was collected by centrifugation, washed with water (3×10 mL), and dried under vacuum. The crude product was purified by column chromatography on alumina gel. Unreacted ligand was eluted first with CH₂Cl₂, followed by elution of the desired product with an acetone/water gradient (100:0 to 95:5 (v/v)). The pure product was a yellow solid (94 mg, 58%). ¹H NMR (400 MHz, d⁶-DMSO) δ 9.65 (td, J = 5.0, 2.5 Hz, 4H), 9.18 (t, J = 9.9 Hz, 4H), 8.90 (m, 2H), 8.71 (m, 2H), 8.47 (dd, J = 9.6, 2.6 Hz, 2H), 8.37 (d, J = 8.4 Hz, 4H), 8.30 (dd, J = 8.8, 4.6 Hz, 4H), 7.99 (ddd, J = 7.5, 5.9, 2.7 Hz, 4H), 7.77 (m, 2H), 7.68 (ddd, J = 9.7, 4.8, 2.7 Hz, 4H), 7.52 (ddd, J = 6.7, 4.1, 1.2 Hz, 4H), 7.45 (dd, J = 5.7, 1.8 Hz, 2H), 7.40 (s, 2H), 7.12 (ddd, J = 8.4, 6.5, 1.7 Hz, 4H), 2.93 (s, 12H), 2.52 (m, 4H), 0.90 (m, 20H), 0.62 (m, 10H). ESI-HRMS (m/z, in acetone): calcd. for [C₉₅H₉₀Ir₂N₁₀]⁴⁺, 439.1652; found, 439.1668. Calcd. for [C₉₅H₉₀Ir₂N₁₀PF₆]³⁺, 633.8750; found, 633.8768. Anal. Calcd. (%) for C₉₅H₉₀F₂₄Ir₂N₁₀P₄₅H₂O: C, 47.03; H, 4.15; N, 5.77. Found: C, 46.97; H, 4.32; N, 5.98.

Ir4.

L1-(IrCl₃)₂ (72.4 mg, 0.05 mmol), 4,6-diphenyl-2,2′-bipyridine (30.8 mg, 0.1 mmol), and AgOTf (77 mg, 0.3 mmol) were mixed in degassed ethylene glycol (10 mL) and heated to 196 °C with protection from light. After 24 h, the mixture was cooled to room temperature, and saturated NH₄PF₆ aqueous solution (20 mL) was added to precipitate the crude product, which was purified by column chromatography on alumina gel. Unreacted ligand was eluted first with CH₂Cl₂,, followed by elution of the desired product with an acetone/water gradient (100:0 to 95:5 (v/v)). The pure product was a yellow solid (38 mg, 41%). ¹H NMR (400 MHz, d⁶-acetone) δ 10.05 (d, J = 7.2 Hz, 2H), 9.49 (d, J = 8.4 Hz, 2H), 9.30 (s, 2H), 9.00 (d, J = 15.3 Hz, 4H), 8.95 – 8.77 (m, 6H), 8.71 (s, 2H), 8.48 (dd, J = 18.4, 9.5 Hz, 8H), 8.36 (s, 2H), 8.28 (d, J = 6.3 Hz, 2H), 8.13 – 7.97 (m, 4H), 7.86 (dd, J = 10.8, 6.6 Hz, 4H), 7.75 (s, 2H), 7.61 (s, 6H), 7.48 – 7.33 (m, 2H), 7.19 (dd, J = 13.3, 4.5 Hz, 4H), 6.91 – 6.69 (m, 4H) 2.41 (m, 4H), 0.86 (m, 20H), 0.55 (m, 10H). ESI-HRMS (m/z, in acetone): calcd. for [C₁₀₃H₉₀Ir₂N₁₀PF₆]³⁺, 655.8756; found, 655.8798. Anal. Calcd. (%) for C₁₀₃H₉₀F₂₄Ir₂N₁₀P4·9H₂O: C, 47.69; H, 4.20; N, 5.40. Found: C, 47.56; H, 4.46; N, 5.25.

Ir5.

L1-(IrCl₃)₂ (72.4 mg, 0.05 mmol), 2,4,6-triphenylpyridine (30.7 mg, 0.1 mmol), and AgOTf (77 mg, 0.3 mmol) were combined in degassed ethylene glycol (10 mL) and heated to 196 °C with protection from light. After 24 h, the mixture was cooled to room temperature, and saturated NH₄PF₆ aqueous solution (20 mL) was added to precipitate the crude product. The pure product was obtained by column chromatography on alumina gel. Unreacted ligand was eluted first with CH₂Cl₂,, followed by elution of the desired product with an acetone/water gradient elution (100:0 to 95:5 (v/v)). The pure product was a red solid (16 mg, 8%). ¹H NMR (400 MHz, d⁶-DMSO) δ 9.56 (s, 4H), 9.19 – 9.09 (m, 4H), 8.95 – 8.85 (m, 4H), 8.60 (s, 4H), 8.33 (d, J = 6.6 Hz, 8H), 8.22 (s, 6H), 7.89 (s, 4H), 7.73 (s, 4H), 7.64 (s, 2H), 7.45 (s, 4H), 6.99 (s, 4H), 6.74 (s, 4H), 6.21 (s, 4H), 2.52 (m, 4H), 0.85 (m, 20H), 0.55 (m, 10H). ESI-HRMS (m/z, in acetone): calcd. for [C₁₀₅H₉₀Ir₂N₈]²⁺, 924.3283; found, 924.3291. Anal. Calcd. (%) for C₁₀₅H₉₀F₁₂Ir₂N₈P₂·0.6CH₂Cl₂: C, 57.94; H, 4.20; N, 5.12. Found: C, 57.92; H, 3.96; N, 5.27.

Photophysical Studies.

Spectrophotometric grade solvents were purchased from Alfa Aesar. The UV–vis absorption spectra of complexes Ir1 – Ir5 were collected on a Varian Cary 50 spectrophotometer, and steady-state emission measurements were carried out using a HORIBA FluoroMax 4 fluorometer/phosphorometer. The emission quantum yields of complexes Ir1 – Ir5 in argon-sparged CH₃CN solution were determined by the relative actinometry method⁵⁵ using [Ru(bpy)₃]Cl₂ (Φ_em = 0.097, λ_ex = 436 nm)⁵⁶ as the standard. The nanosecond transient absorption (TA) spectra and decays, triplet excited-state quantum yields, and triplet lifetimes were collected on argon-sparged (40 min) acetonitrile solutions on an Edinburgh LP920 laser flash photolysis spectrometer using the third harmonic output (355 nm) of a Nd:YAG laser (Quantel Brilliant, pulsewidth ~4.1 ns, repetition rate was set to 1 Hz). Molar extinction coefficients (ε_T1-Tn) for triplet excited states were determined by the singlet depletion method⁵⁷ at the TA band maxima, and quantum yields for triplet excited state formation were measured according to the relative actinometry method⁵⁸ using the ε_T1-Tn values, with SiNc in benzene as the reference (ε₅₉₀ = 70,000 M⁻¹cm⁻¹, Φ_T = 0.20).⁵⁹

Singlet Oxygen Quantum Yields.

Singlet oxygen quantum yields (Φ_Δ) were determined for the PF₆⁻ salts of the complexes directly from sensitized singlet oxygen emission centered at 1268 nm using a PTI Quantamaster equipped with a Hamamatsu R5509-42 near-infrared PMT. The metal complexes were prepared in spectroscopic-grade CH₃CN at 5 μM, and the measurements were made under ambient oxygen concentration (21%). Φ_Δ was calculated relative to the standard [Ru(bpy)₃](PF₆)₂ (Φ_Δ = 0.56 in aerated CH₃CN)⁶⁰ according to Eq 1, where I, A, and η represent integrated emission intensity, absorbance at the excitation wavelength, and refractive index of the solvent, respectively. Values calculated for Φ_Δ were reproducible to within <5%.

$${\Phi }_{\Delta }={\Phi }_{\Delta s}\left(\frac{I}{A}\right)\left(\frac{A_s}{I_s}\right)\left(\frac{{\eta }^2}{{\eta }_s^2}\right)$$ Eq1

DFT Calculations.

Density functional theory (DFT) and time-dependent DFT (TDDFT) calculations on Ir(III) complexes were performed using the Gaussian09 quantum software package.⁶¹ The basis sets used in all calculations were LANL2DZ^62–64 for Ir(III) and 6-31G*^65–69 for other non-metal atoms. Unlike the previous Ir(III) complexes studied,^25,70,71 complexes Ir1 - Ir5 have two metal centers connected by fluorenylbisterpyridyl ligand, which makes long-range interactions critical for the description of both geometry and excited state properties. Therefore, here we used the long-range corrected hybrid function, ωB97XD, which was designed to capture long-range atom-atom dispersion.⁷² Implicit solvent effects were incorporated by the conductor-like polarizable continuous model (CPCM)^73,74 simulating the effects of acetonitrile. The long aliphatic side chains on fluorene were replaced with butyl groups to reduce the computational cost. This reduction does not affect the optical properties of the complexes in the visible spectral region, since octyl groups do not contribute any electronic transitions in this energy range.

The absorption spectra for the Ir(III) complexes were generated by broadening the lowest 100 singlet vertical excitations computed by TDDFT^75,76 using the functional and basis sets described above. To broaden the vertical excitation to generate spectra in terms of molar absorptivity units (L·mol⁻¹·cm⁻¹) the method described by Bjorgaard and co-workers was followed.⁷⁷ The shape of the spectra generated by broadening the vertical excitation computed by TDDFT quantitatively agreed well, except for about ~0.6 eV blue-shift, which is expected for the ωB97XD functional applied to conjugated systems⁷⁸ (see Table S1 of Supporting Information for the energy differences of the experimental and theoretical transitions of the first absorption band). To align the theoretical spectra with the experimental spectra, all transitions energy were red-shifted by −0.55 eV To characterize the type of excitation for the Ir(III) complexes, natural transition orbitals (NTOs)⁷⁹ were generated using Gaussian09 software. NTOs allow for representing an excitation as the electron and hole pair, while preserving the many-body nature of the excited states. Due to the high symmetry of these dinuclear complexes, multiple transition densities that only differ on which metal center electronic density is localized contribute to some excitations. Therefore, only uniquely representative NTOs are shown for those excitations and those states are indicated by “*” in Tables 2 and S2. The visualization of the NTOs were done utilizing Visual Molecular Dynamics (VMD)⁸⁰ with isosurface of 0.02.

Table 2.

Natural Transition Orbitals (NTOs) for Low Energy Transitions of Ir1 - Ir5. For Transitions with Quasi-Degenerate Transition Orbitals, Only One Pair of Transition Densities Are Shown and Are Indicated By *.

	Sn	Hole	Electron
Ir1	S1 340 nm f = 2.637
	S5 303 nm f = 0.109
Ir2	S1 381nm f = 3.730
	S2* 336 nm f = 0.079
Ir3	S1 333 nm f = 0.010
	S3 328 nm f = 2.189
	S7 311 nm f = 0.368
Ir4	S1 351 nm f = 0.003
	S2 350 nm f = 0.004
	S3* 340 nm f = 2.002
Ir5	S1* 395 nm f = 0.933
	S6* 372 nm f = 0.050

Photobiological Activity Studies.

The experimental details for cell culture, cytotoxicity and photocytotoxicity studies, confocal microscopy, and DNA mobility-shift assays are the same as those described in our previous published work^24,25 and are presented in the Supporting Information.

RESULTS AND DISCUSSION

Electronic Absorption.

The experimental UV–vis absorption spectra of complexes Ir1 - Ir5 were recorded in acetonitrile (Figure 1a), and the absorption band maxima and molar extinction coefficients are listed in Table 1. The absorption follows Beer’s law in the solutions used in our study (i.e. 5×10⁻⁶ to 1×10⁻⁴ mol·L⁻¹), indicating that no ground-state aggregation was formed in the tested concentration range. The strong absorption bands in the range of 250-350 nm and 350-500 nm are predominantly assigned to ¹π,π* transitions localized on the terminal tridentate ligands, and the bridging ligand, respectively. These assignments are supported by the NTOs corresponding to the major transitions contributing to these bands (Supporting Information Table S2 and Table 2). Attribution of the absorption bands of 350-500 nm to the bridging ligand localized ¹π,π* transition is in line with that revealed in the dinuclear Ir(III) complexes with trisbidentate ligands and diethynylaryl substituted diketopyrrolopyrrole bridging ligand.^81,82 However, NTOs of Ir1 – Ir4 in Table 2 show that some charge transfer transitions, i.e. ¹MLCT, ¹LLCT (ligand-to-ligand charge transfer) or ¹ILCT (intraligand charge transfer) contributed to the 350-500 nm bands as well. Contributions of the charge transfer configurations to the 350-500 nm absorption bands are partially reflected by the insignificant but noticeable negative solvatochromic effects (Figure S3 in Supporting Information), especially in Ir2 that has more ¹ILCT character (see NTOs in Table 2). In contrast, the charge transfer transitions in Ir5 became more distinguishable and energetically separated from the bridging ligand localized ¹π,π* transition. This is clearly evidenced by the appearance of the new absorption band at 520 nm in Ir5.

Figure 1.

Experimental (a) and theoretical (b) UV–vis absorption spectra of Ir1 – Ir5 at room temperature in acetonitrile. The inset in panel (a) is the expansion of the spectra in the region of 450–800 nm. The theoretical spectra were computed using ωB97XD with mixed basis set. A redshift of 0.55 eV for the theoretical spectra in panel (b) was applied for better comparison with the experimental spectra.

Table 1.

Electronic Absorption, Emission, and Triplet Excited-State Absorption Parameters, as well as Singlet Oxygen Quantum Yields for Complexes Ir1 - Ir5

	λabs/nm (log ε)a	λem/nm (τem/μs); Φemb	λT1-Tn/nm (τTA/μs; log εT1-Tn); ΦTc	ΦΔd (λex/nm)
Ir1	251 (5.01), 297 (5.06), 427 (4.85)	583 (3.57); 0.024	498 (3.09; -), 765 (3.05; 4.62); 0.44	0.28 (430)
Ir2	251 (5.03), 285 (5.03), 364 (4.65), 420 (4.68)	576 (1.99); 0.003	385 (1.72; -), 640 (1.75; 5.04); 0.03	0.26 (468)
Ir3	260 (4.92), 289 (4.97), 386 (4.80), 405 (4.80)	608 (1.47); 0.025	513 (0.03 (17%), 1.72 (83%); -), 770 (0.03 (15%), 1.75 (85%)); 4.70); 0.14	0.04 (411)
Ir4	294 (5.00), 321 (4.88), 413 (4.75)	578 (53.3); 0.22	498 (48.6; -), 800 (48.3; 4.64); 0.28	0.38 (418)
Ir5	283 (4.97), 312 (4.89), 368 (4.72), 416 (4.65), 520 (4.13)	619 (1.92); 0.045	681 (2.68; 4.80), 787 (2.68; -); 0.07	0.22 (418)

^aAbsorption band maxima (λ_abs) and molar extinction coefficients (ε / L·mol⁻¹·cm⁻¹) of the UV-vis absorption in acetonitrile at room temperature.

^bEmission band maxima (λ_em), lifetimes (τ_em), and quantum yields (Φ_em) measured in acetonitrile (c = 1 × 10⁻⁵ mol·L⁻¹) at room temperature with Ru(bpy)₃Cl₂ (in degassed acetonitrile; Φ_em = 0.097, λ_ex = 436 nm) as the reference.

^cNanosecond TA band maxima (λ_T1-Tn), triplet excited-state lifetimes (τ_TA), triplet molar extinction coefficients (ε_T1-Tn / L·mol⁻¹·cm⁻¹), and quantum yields for triplet state formation (Φ_T) measured in acetonitrile at room temperature with SiNc (in degassed benzene; ε₅₉₀ = 70,000 L·mol⁻¹·cm⁻¹, Φ_T = 0.20) as the reference. λ_ex = 355 nm.

^dSinglet oxygen quantum yields in acetonitrile. Values are correct to within ±5%.

Comparison of the absorption spectra of Ir1 and Ir2 revealed that incorporation of the C≡C bonds to the bridging ligand caused a broadening and a red-shift of the bridging ligand localized ¹π,π* absorption band due to the extended π-conjugation. Replacing the terminal terpyridyl ligands in Ir1 by N^∧C^∧N (1,3-dipyridyl-4,6-dimethylbenzene) ligands in Ir3 induced a blue-shift of the 350-500 nm absorption band and incorporated more terminal ligands based ¹π,π* transition and ¹LLCT/¹MLCT character to this band (see NTOs for Ir3 in Table 2); while changing the terminal ligands to C^∧N^∧N (4,6-diphenyl-2,2′-bipyridine) ligands in Ir4 only caused a slight blue-shift of this band with respect to that in Ir1. In contrast, when the terminal ligands were changed to C^∧N^∧C (2,4,6-triphenylpyridine) ligands in Ir5, the transition energies, intensities, and the shape of the low-energy absorption bands changed pronouncedly from those of Ir1. This can be attributed to the distinct nature of the lowest energy optical transition in these two complexes. As the NTOs in Table 2 indicated, the stronger σ-donating ability of the phenyl rings on the C^∧N^∧C ligand delocalized the hole of the S₁ transition mainly to the 2,6-diphenyl rings and to the metal d orbitals, while the electron was predominantly on the terpyridyl ligands. Thus, the lowest-energy optical transition in Ir5 is predominantly the ¹LLCT/¹MLCT transition, which is in contrary to the bridging ligand localized ¹π,π* transition in Ir1. The drastic change of the dominant optical transitions accounts for the different features of the low-energy absorption bands in Ir5 with respect to that in Ir1.

Photoluminescence.

The emission of Ir1 - Ir5 was studied in different solvents at room temperature. The observed emission all exhibited large Stokes shifts with respect to the corresponding excitation wavelength, they were all long-lived (several to tens of μs), and sensitive to the presence of oxygen. Thus the emission was attributed to phosphorescence. The normalized emission spectra of Ir1 - Ir5 in acetonitrile are displayed in Figure 2, and the spectra in other solvents are given in SI Figure S4. The emission parameters are listed in Table 1 and Table S3 of SI. The emission of Ir1 and Ir4 resembled each other, both showing some vibronic structures, with much longer lifetimes and higher emission quantum yields compared to the other three complexes, and exhibiting minor solvatochromic effects. The vibronic spacing between the 580 nm and 620 nm bands is approximately 1150 cm⁻¹ and 1090 cm⁻¹ in Ir1 and Ir4, respectively, which is in accordance with the aromatic vibrational mode of the terpyridyl ligands. Thus, the emission of these two complexes can be assigned predominantly to the ligand localized ³π,π* state. However, the lifetime of Ir1 is one order of magnitude shorter than that of Ir4. This could be attributed to the weaker ligand field of the terpyridyl ligand with respect to that of the C^∧N^∧N ligand that contains the stronger σ-donating 6-phenyl ring. The nonradiative metal-centered ³d,d state is thus situated more closely to the low-lying emissive ³π,π* state and becomes thermally accessible in Ir1 compared to that in Ir4. This adds an additional decay path for the emitting ³π,π* state in Ir1 and consequently reduces its lifetime. For Ir2, Ir3, and Ir5, the emission spectra are featureless and broader, the lifetimes are less than 2 μs and the emission quantum yields are quite low, and the solvatochromic effect is more pronounced. All these characters imply charge transfer nature of the emitting states in these three complexes. Referring to the NTOs corresponding to the low-energy singlet charge transfer transitions shown in Table 2, it is reasonable to speculate that the emitting state of Ir2 could be the ³ILCT state and they are the ³LLCT/³MLCT states in Ir3 and Ir5. It appeared that either extending the π-conjugation of the bridging ligand in Ir2, or varying the terminal tridentate ligands in Ir3 and Ir5 changed the nature of the emitting state from the ³π,π* state in Ir1 to ³CT states. In addition, variation of the terminal tridentate ligands impacted the emission energies in Ir3 – Ir5 compared to that in Ir1, with a slight blue-shift of the emission in Ir4 while a salient red-shift in Ir3 and Ir5. The red-shifted emission in Ir3 and Ir5 with respect to that in Ir1 could possibly be rationalized by the stronger σ-donating ability of the phenyl rings on the terminal tridentate N^∧C^∧N and C^∧N^∧C ligands, which raised the energies of the terminal ligand and the metal d orbital based holes and thus reduced the energy gaps between the holes and electrons (likely localized on the terpyridyl motifs). Consequently, the ³LLCT/³MLCT emission energies of Ir3 and Ir5 are reduced.

Figure 2.

Experimental emission spectra of Ir1 (λ_ex = 426 nm), Ir2 (λ_ex = 420 nm), Ir3 (λ_ex = 405 nm), Ir4 (λ_ex = 413 nm), and Ir5 (λ_ex = 415 nm) at room temperature in deoxygenated acetonitrile (c = 1×10⁻⁵ mol·L⁻¹).

Transient Absorption.

To further understand the triplet excited–state characteristics, the nanosecond transient absorption (TA) studies of complexes Ir1 – Ir5 were conducted in acetonitrile solutions. The TA spectra of Ir1 – Ir5 at zero delay after excitation are presented in Figure 3 and the TA parameters are provided in Table 1. The time-resolved TA spectra of Ir1 – Ir5 are provided in Supporting Information Figure S5. The triplet lifetimes deduced from the decay of TA for Ir1 – Ir4 are similar to their emission lifetimes in acetonitrile. Therefore, we consider that the observed TA of these complexes was from the excited states that emit. In contrast, Ir3 exhibited a biexponential decay in its TA signals, with the longer lifetime being consistent with the emission lifetime. This implies that the long-lived TA signal in Ir3 could from the emitting excited state as well.

Figure 3.

Nanosecond transient absorption (TA) spectra of complexes Ir1 – Ir5 in deoxygenated acetonitrile at zero delay after 355 nm excitation. The inset shows the TA spectra of Ir3 at different delay time after excitation. A_{355 nm} = 0.4 in a 1-cm cuvette.

The TA spectra of Ir1 – Ir5 all possessed very broad positive absorption band(s) from the visible to the near-IR region, i.e. 463-800 nm for Ir1, 495-800 nm for Ir2, 459-800 nm for Ir3, 455-800 nm for Ir4, and 538-800 nm for Ir5. Bleaching occurred in the region corresponding to the low-energy absorption bands. Considering the similar shape of the TA spectra of Ir1 and Ir4 to that of our previously reported dinuclear platinum(II) complex with the same bridging ligand,⁴⁸ and the similar lifetimes to those of emission, we tentatively attribute the contributing transient absorbing excited state predominantly to the bridging ligand localized ³π,π* state. While for Ir2 and Ir5, the transient absorbing states are likely to be the ³CT state(s), i.e. predominantly ³ILCT for Ir2 and ³LLCT/³MLCT states for Ir5. In contrast to Ir1, Ir2, Ir4 and Ir5 that exhibited monoexponential decays in their TA signals, the TA signal of Ir3 followed a biexponential decay. The short-lived transient species had a lifetime of ~30 ns and gave rise to a spectrum reminiscent to those of Ir1 and Ir4; while the long-lived species had a lifetime of ~1.7 μs, which is consistent with the lifetime obtained from the decay of emission, and the TA was much weaker and featureless. In view of the different spectral features at the shorter and longer decay time and the reminiscence of the spectra to those of Ir1/Ir4 and Ir2/Ir5, respectively, we tentatively assign the short-lived species to the high-lying bridging ligand localized ³π,π* state; while the long-lived species to the emitting ³LLCT/³MLCT state. The formation of a rapidly decaying higher excited state that subsequently leads to the lower-lying, long-lived emitting state has been reported for a mononuclear Ir(III) complex [(dpb)-Ir(tpy-ph(tBu)₂]²⁺ that bears the same N^∧C^∧N ligand.⁸³

It is noted that the measured triplet quantum yields of these complexes are not quite high, especially for Ir2, Ir3 and Ir5. This could be due to the following reasons: (i) The increased π-conjugation of the ligand would decrease the contribution of the transition metal d orbital to the frontier molecular orbitals of the complexes, which would reduce the spin-orbital coupling in the complexes and decrease the triplet quantum yield. Such a phenomenon has been reported in many Pt(II) and Ir(III) complexes.^25,70,84 (ii) When a transition-metal complex is excited, especially when high-energy excitation is utilized, population of more than one triplet excited states is possible.^70,85–91 However, not all of the populated triplet excited states contribute to excited-state absorption. In such a case, the calculated triplet quantum yield based on the observed TA signal could be significantly lower than the actual intersystem crossing quantum yield.

Singlet Oxygen Generation.

Production of ¹O₂ is known to have cytotoxic effects on cells, and thus compounds that generate ¹O₂ under cell-free conditions might be expected to act as in vitro PDT agents. The Ir(III) complexes Ir1 – Ir5 were assessed for singlet oxygen (¹O₂) sensitization in cell-free conditions through direct measurement of ¹O₂ emission at 1270 nm. [Ru(bpy)₃](PF₆)₂ was used as the standard, with a reported ¹O₂ quantum yield (Φ_Δ) of 0.56 in air-saturated CH₃CN.⁶⁰ The calculated Φ_Δ values for all of the complexes were less than 40%. The efficiencies for ¹O₂ production ranged from 4% for Ir3 to 38% for Ir4, with Ir1, Ir2 and Ir5 yielding similar values (22-28%). Despite having some absorption at wavelengths longer than 500 nm, ¹O₂ yields were maximal with blue excitation. For N^∧N^∧N terminal tridentate ligands, the presence of the ethynyl groups for extending π-conjugation did not alter the singlet oxygen quantum yield as Ir1 and Ir2 gave very similar values for Φ_Δ. When each terminal tridentate ligand of Ir1 had two of its nitrogens replaced with cyclometalating carbons (C^∧N^∧C) as in Ir5, the ¹O₂ yield decreased only slightly. These limited comparisons appear to indicate that substantial structural changes have little to no effect on Φ_Δ. However, when only one nitrogen of each terminal tridentate ligand of Ir1 was replaced by carbon (C^∧N^∧N) as in Ir4, the ¹O₂ yield increased to almost 40%. Clearly, there are structural combinations in this family of complexes that do influence Φ_Δ. The most dramatic impact on Φ_Δ occurred for Ir3, where the terminal tridentate ligands were N^∧C^∧N with methyl substitution at R₁ but no phenyl group at R. In this case, the ¹O₂ yield decreased by almost tenfold.

It is also worthy of noting that the Φ_Δ values for Ir2, Ir4 and Ir5 are higher than the measured triplet quantum yields (Φ_T, Table 1). This is not very surprising because population of excited states is wavelength dependent, which could result in different decay pathways.⁹¹ In the Φ_T measurement, 355 nm excitation was used; while low-energy excitation (i.e. 411 – 468 nm) was used in the Φ_Δ measurement. A 355-nm excitation in the TA measurement could populate more than one triplet excited states,^70,85–91 which might not only impact the Φ_T value measurement as discussed in the TA section, but could also reduce the population of the excited state that generates singlet oxygen because of the competing population of the other non-¹O₂-generating triplet excited states. In our previous study on the monocationic tris-bidentate Ir(N^∧N)(C^∧N)₂ complexes, we have demonstrated that the singlet oxygen generation efficiency is wavelength dependent, with lower-energy excitation resulting in higher Φ_Δ values in those Ir(III) complexes.²⁴ We speculate the same case for the complexes studied in this work.

Cytotoxicity and Photocytotoxicity.

To understand whether the photophysical properties of these Ir(III) complexes could lead to photobiological effects, the cytotoxicity profiles of Ir1 – Ir5 were assessed in SK-MEL-28 malignant melanoma cells under three conditions: (i) dark, (ii) illumination with broadband visible light, and (iii) illumination with red LEDs emitting at 625 nm. Cytotoxic and photocytotoxic activities were quantified as the effective concentration required to reduce the cell viability to 50% (EC₅₀) under a given condition. Briefly, cells growing in log phase were dosed with nine concentrations of the complex between 1 nM and 300 μM, incubated for 16 h, and were then subjected to a sham (dark) or light treatment. The light treatments were delivered at a fluence of 100 J·cm⁻² with an irradiance of 35.7 mW·cm⁻² or 32.3 mW·cm⁻² for visible and red light, respectively. After a 48 h incubation period, cell viability was quantified based on the ability of viable cells to reduce resazurin to resorufin. EC₅₀ values were determined from sigmoidal fits of the dose-response curves (Figure 4, Table 3). The phototherapeutic index (PI), a measure of the therapeutic margin for in vitro PDT, was calculated as the ratio of dark to light EC₅₀ values and determined for each complex and irradiation condition. The dark toxicities of Ir1 – Ir5 toward normal human skin fibroblasts (CCD-1064Sk) were also measured to determine any selectivity for cancer cells over normal cells. The ratio of the dark CCD-1064Sk EC₅₀ value for a given complex and its dark SK-MEL-28 EC₅₀ value yielded the selectivity factor (SF), where SF > 1 indicates selectivity toward the cancerous cell line. Selective activity toward the cancer cell line is not a requirement for the PDT agent as long as the dark toxicity of the photosensitizer is low and the PI is relatively large. Rather, the spatiotemporal control of the light treatment provides the selectivity known for PDT. Nevertheless, for in vivo applications, selective activity toward cancer cells over normal, healthy cells is an added benefit.

Figure 4.

In vitro dose-response curves for complexes Ir1 (a), Ir2 (b), Ir3 (c), Ir4 (d), and Ir5 (e) in SK-MEL-28 cells treated in the dark (black) and with visible (blue) or red (red) light activation.

Table 3.

Comparison of EC₅₀ values (μM) for SK-MEL-28 cancer cells and CCD-1064Sk normal skin fibroblasts dosed with complexes Ir1 – Ir5.

		SK-MEL-28 cells				CCD-1064Sk cells
	Dark	Visa	PIb	Redc	PId	Dark	SFe
Ir1	82.6 ± 1.5	0.75 ± 0.01	111	59.7 ± 0.4	1.4	102 ± 2	1.2
Ir2	16.9 ± 0.8	0.83 ± 0.05	20	16.2 ± 0.5	1.0	32.0 ± 2.0	1.9
Ir3	49.9 ± 0.1	0.17 ± 0.01	288	49.5 ± 0.1	1.0	49.1 ± 0.1	1.0
Ir4	17.0 ± 0.7	0.17 ± 0.01	102	14.3 ± 0.3	1.2	31.8 ± 1.8	1.9
Ir5	69.5 ± 1.0	1.05 ± 0.01	66	85.1 ± 1.0	0.82	142 ± 3	2.0

^aVis-PDT: 16 hours drug-to-light interval followed by 100 J·cm⁻² broadband visible light irradiation

^bPI = phototherapeutic index (ratio of dark EC₅₀ to visible-light EC₅₀)

^cRed-PDT: 16 hours drug-to-light interval followed by 100 J·cm⁻² light irradiation with 625-nm LEDs

^dPI = phototherapeutic index (ratio of dark EC₅₀ to red-light EC₅₀)

^eSF SK-MEL-28: selectivity factor (ratio of dark CCD-1064Sk EC₅₀ to dark SK-MEL-28 EC₅₀).

The dark cytotoxicities of complexes Ir1 – Ir5 toward SK-MEL-28 melanoma cells ranged from 16.9 to 82.6 μM, with Ir1 being the least cytotoxic in the absence of a light trigger and Ir2 and Ir4 being the most cytotoxic (Table 3 and Figures 4–5). With the exception of Ir3 (SF = 1), the other dinuclear Ir(III) complexes exhibited some selective cytotoxicity toward the melanoma cancer cells relative to the normal human skin fibroblast cells. SF values followed the order Ir5 > Ir2 ≈ Ir4 > Ir1 > Ir3, with Ir5 exhibiting two-fold greater dark toxicity toward SK-MEL-28 cells and Ir3 showing no selectivity. The selective cytotoxicity observed for Ir2 and Ir4 was almost as great as that for Ir5 (SF = 1.9 versus SF = 2.0). Ir1 and Ir5 had dark EC₅₀ values greater than 100 μM in the CCD-1064Sk cell line, and were thus considered to be completely nontoxic to the normal skin fibroblasts. In both cell lines, the dark toxicity was greatest for Ir2 and Ir4 and least for Ir1 and Ir5.

Figure 5.

Activity plot for complexes Ir1 - Ir5 in SK-MEL-28 and CCD-1064Sk cells treated in the dark (black) and with visible (blue) or red (red) light activation.

All of the complexes in the series could be activated with visible light to become powerful phototoxins, with EC₅₀ values ranging from 170 nM to 1 μM and PIs ranging from 20 to 288. Ir3 and Ir4 were the most phototoxic at 170 nM, while Ir1, Ir2, and Ir5 were similar (visible EC₅₀ = 0.75–1.0 μM). Photoactivation of the complexes with red light (625 nm) did not enhance the cytotoxicity over what was observed in the dark treatment, yielding PIs close to 1.0 in all cases. The photobiological activities of Ir1 - Ir5 were tested with red light, despite their very low molar extinction coefficients in this region, because other metal complex systems with π-expansive ligands have been shown to yield potent in vitro red PDT effects even with molar extinction coefficients less than 100 M⁻¹·cm⁻¹.²⁰

The presence of visible PDT effects (presumably due to the shorter wavelengths) but lack of red PDT effects suggests that direct population of the highly photosensitizing triplet states is not efficient in this class of complexes and that access to these states must be gained through ¹MLCT states. The photocytotoxicity profiles in SK-MEL-28 under two irradiation conditions along with the dark cytotoxicity profiles in two cell lines are summarized in the activity plot in Figure 5.

The visible PDT effect followed the order Ir3 > Ir1 > Ir4 > Ir5 > Ir2, with Ir3 being the most promising photosensitizer based on its PI of 288 and nanomolar photocytotoxicity. Ir1 and Ir4 both had PIs greater than 100, but the dark toxicity associated with Ir4 in both cell lines limits its potential for in vivo applications. The source of the PDT effect for this series has not been established. The ¹O₂ quantum yields measured under cell-free conditions followed the order Ir4 > Ir1 > Ir2 > Ir5 > Ir3, with Ir1 and Ir2 being very similar. Ir3 was the poorest ¹O₂ generator, yet it was one of the most phototoxic complexes of the series. On the other hand, Ir4 was the best sensitizer of ¹O₂ and was as phototoxic as Ir3. Therefore, ¹O₂ may play a role in the PDT mechanism for some complexes but not others in this series, or the intracellular ¹O₂ quantum yields may differ from those measured under cell-free conditions. Regardless, certain members of this new series of dinuclear Ir(III) complexes have been identified as promising PDT agents for further investigation.

While the structural diversity in such a small library is somewhat limited, it was possible to identify some trends regarding structural features that affect cytotoxicity. For example, incorporation of ethynyl linkers as in Ir2 turned the relatively nontoxic complex Ir1 into one of the most potent dark cytotoxic complexes of the series (Table 3). Likewise, replacement of the terminal tridentate N^∧N^∧N ligands of Ir1 with C^∧N^∧N as in Ir4 increased the dark cytotoxicity substantially, while replacement with C^∧N^∧C as in Ir5 had only a very minor effect that differed between the two cell lines. In SK-MEL-28, the dark toxicity increased slightly, and in CCD-106Sk, the dark toxicity decreased slightly. The complex that departed the most structurally from the other complexes in the series and was identified as being the most promising PDT lead, Ir3, was intermediate in terms of dark cytotoxicity (EC₅₀ ≈ 50 μM) with almost no difference between the two cell lines.

In terms of structural features affecting photocytotoxicity, the nature of the terminal tridentate ligand played some role as Ir4 (C^∧N^∧N) was more than six fold more phototoxic than Ir5 (C^∧N^∧C). While the presence of an ethynyl linker increased the dark cytotoxicity substantially in both cell lines, its presence did not impact the photocytotoxicity toward SK-MEL-28 in any significant way. The differences in dark and light-triggered cytotoxicity toward SK-MEL-28 cells alongside differences in dark cytotoxicity between normal and cancerous cells for certain members of this series indicate that even minor structural modifications can have a major impact on biological activity.

Cellular Imaging.

The phosphorescence from complexes Ir1 – Ir5 was used to probe cellular uptake by SK-MEL-28 melanoma cells with or without a light treatment (Figure 6). The excitation from a 458/488 nm argon-krypton laser matched the excitation maxima of the complexes and was used in conjunction with a 475-nm long pass filter to collect the emission from the complexes. The images were collected after a brief 1-h incubation period to ensure that some viable cells remained. Light-treated cells were illuminated with a broadband visible light (50 J·cm⁻²) that was 50% of the fluence used in the cellular assays in order to capture a fraction of viable cells.

Figure 6.

Confocal luminescence images of SK-MEL-28 cells dosed with Ir1 - Ir5 (a-e, 50 μM) in the dark (left) and with visible light (50 J·cm⁻²) (right).

Untreated SK-MEL-28 cells have a dendritic morphology. Treatment with the dinuclear Ir(III) complexes with or without illumination caused a conversion from dendritic to spherical morphology. The complexes showed detectable phosphorescence when associated with or in dead/dying and compromised cells with or without a light treatment. Only Ir1 appeared to be readily taken up into SK-MEL-28 cells in the dark at the observation time point. However, phosphorescence from the Ir(III) complexes in all cells was apparent after a light treatment, suggesting photoactivated uptake. For light-treated cells incubated with Ir1, it was not possible to discern subcellular localization because only cellular debris was present at the observation time point. However, Ir2 and Ir4 localized to the cytoplasm and multiple nucleoli whereas Ir3 and Ir5 were distributed throughout the cell and phosphoresced with a very intense signal by comparison.

Qualitatively, SK-MEL-28 cells treated with Ir4 with or without visible illumination appeared the most viable with healthy morphology in the imaging experiments but were the most susceptible in the cellular assays, highlighting the need to exercise caution when reconciling the cellular assay results with confocal imaging performed at different time points post-complex-delivery and post-irradiation and a different light fluence. When the conditions were similar, the imaging experiments did reflect the trends observed in the cellular assays with SK-MEL-28 but did not provide any information regarding uptake and localization since all cells were dead/dying but at slightly different stages. A quantitative comparison of the cellular uptake and induced morphological changes for the five complexes and correlations to cellular cytotoxicity or photocytotoxicity were not attempted given the need to alter incubation and illumination times to preserve some viable cells. Rather, the purpose of the imaging was to highlight the potential of these new Ir(III) complexes as theranostic agents based on their abilities to yield visible PDT effects and to be simultaneously imaged by their inherent phosphorescence.

DNA Interactions.

The ability of the Ir(III) complexes to act as DNA photocleaving agents was investigated to establish whether light-mediated DNA damage could contribute to the observed in vitro PDT effects for this class of photosensitizers. Supercoiled plasmid DNA (20 μM bases) was treated with increasing concentrations of Ir1 – Ir5 and then exposed to a visible light treatment of 14 J·cm⁻² (Figure 7, lanes 3–14). The fluence is less that what was used in the cellular assays because the DNA is more susceptible to damage by the light treatment alone when not protected by the cellular environment. The photolyzed samples were then electrophoresed and compared to DNA alone with or without a light treatment (Figure 7, lanes 1 and 2) and DNA exposed to the highest concentration of the complex without a light treatment (Figure 7, lane 15). The gels were cast either with the DNA stain ethidium bromide (EB) incorporated or without EB and stained after electrophoresis (non-EB). EB gels allow detection of photocleavage not compounded by DNA unwinding; the non-EB gels allow detection of DNA unwinding in addition to photocleavage. Under the conditions employed for this gel electrophoretic mobility shift assay, undamaged supercoiled DNA (Form I) migrates the farthest in the gel, while aggregated/condensed DNA (Form IV) migrates very little from the loading well. Plasmid DNA that has undergone single-strand breaks (Form II) will relax and migrate between Forms I and IV, and plasmid DNA with frank double-strand breaks or double-strand breaks that arise from the build-up of single-strand breaks on opposing strands within about 16 base pairs (Form III) will migrate slightly faster than Form II. Forms I, II, and IV were detectable in both EB and non-EB gels. None of the complexes acted as DNA unwinders on the non-EB gel, indicating that they most likely do not act as DNA intercalators.

Figure 7.

DNA photocleavage of pUC19 DNA (20 μM) dosed with metal complex (MC) Ir1 (a), Ir2 (b), Ir3 (c), Ir4 (d), Ir5 (e) and visible light (14 J·cm⁻²). Gel mobility shift assays employed 1% agarose gels (0.75 μg·mL⁻¹ ethidium bromide) electrophoresed in 1× TAE at 8 V·cm⁻¹ for 30 min. Lane 1, DNA only (−hv); lane 2, DNA only (+hv); lane 3, 0.5 μM MC (+hv); lane 4, 1 μM MC (+hv); lane 5, 2 μM MC (+hv) lane 6, 3 μM MC (+hv) lane 7, 5 μM MC (+hv) lane 8, 8 μM MC (+hv); lane 9, 10 μM MC (+hv) lane 10, 12 μM MC (+hv) lane 11, 15 μM MC (+hv) lane 12, 20 μM MC (+hv); lane 13, 50 μM MC (+hv); lane 14, 100 μM MC (+hv); lane 15, 100 μM MC (−hv). Forms I, II and IV DNA refer to supercoiled plasmid, nicked circular plasmid, and aggregated plasmid, respectively.

All of the complexes showed some ability to photocleave DNA (Figure 7) in a cell-free environment. Qualitatively, DNA photocleaving ability appeared to increase in the order Ir5 < Ir2 ≈ Ir3 < Ir4 < Ir1. The formation of Form IV DNA and the disappearance of gel bands precluded a more quantitative comparison, but some general trends could still be discerned.

Despite its ¹O₂ quantum yield of 22%, Ir5 appeared to show the weakest interactions with DNA (although some strand breaks to yield detectable Form II⁹² were observed toward the highest concentrations). On the other hand, Ir1, with a similar Φ_Δ, acted as a much more potent DNA photocleaving agent, converting a significant amount of supercoiled Form I DNA to Form II DNA at a metal complex (MC) concentration of only 1 μM and 20 μM DNA bases (Figure 7a, lane 4). At similarly low [MC]:[bases] ratios of 0.05, Ir2 – Ir5 caused no detectable strand breaks, which can be seen by comparing lane 4 for all of the complexes. Ir4, with the largest value for Φ_Δ, photocleaved DNA in a concentration-dependent manner to yield Form II DNA as expected.

All of the complexes caused DNA aggregation/condensation, although Ir5 produced trace amounts of Form IV DNA only at the highest complex concentrations investigated. Interestingly, Ir5 was the only complex that did not cause the DNA gel bands to disappear. The lack of DNA staining by EB for the other complexes could be due to fluorescence quenching of the EB dye by the complex, their competition for EB intercalation sites, or their distortion of the DNA helix that prevents EB binding.

Clearly, the structural differences between the Ir(III) complexes of this small library resulted in markedly different interactions with DNA, and possibly different photophysical interactions with the EB dye. The observation that Ir5 shows marginal DNA interactions in the gel electrophoretic analysis yet acts as an in vitro PDT agent suggests that DNA may not be the intracellular target, at least for this particular complex. In fact, DNA photodamage did not correlate clearly with ¹O₂ quantum yields across the series, which also supports the notion that another biological target is likely involved. However, the cell-free experiment does not accurately mimic the complexity of the cellular environment and dynamic processes (e.g., uptake, efflux, metabolism, and localization), and in vitro DNA damage and ¹O₂ damage cannot be ruled out completely. What can be gleaned from the DNA photocleavage study is that minor structural changes in this series have profound effects on the complex interactions with biological macromolecules such as DNA, which is in agreement with their different profiles in the cellular assays and imaging studies.

CONCLUSIONS

The synthesis, photophysical and photobiological properties of a family of water-soluble cationic dinuclear iridium(III) complexes (Ir1 - Ir5) were explored. The influence of the bridging and terminal ligands on the photophysical properties of the complexes was investigated. Compared to Ir1 that had the single bond connection between the fluorenyl motif and the terpyridyl ligands on the bridging ligand, the extended π-conjugation afforded by the ethynyl connectors of the bridging ligand in Ir2 red-shifted the UV-vis absorption markedly, but the low-lying ³CT state of Ir2 accelerated nonradiative decay and resulted in weak phosphorescence. A considerable bathochromic shift also occurred in the absorption and emission of Ir5, owing to the stronger σ-donating ability of the negatively charged coordinating carbon relative to nitrogen and thus more charge transfer from the C^∧N^∧C ligands to the terpyridyl ligands. Complexes Ir1 – Ir5 all featured with broad positive absorption bands spanning the visible region and NIR regions in their nanosecond TA spectra. However, the triplet state TA lifetimes of Ir1 and Ir4 were much longer (3.1 μs and 48 μs, respectively) than those of Ir2, Ir3 and Ir5, implying the dominant bridging ligand-localized ³π,π* nature for the lowest triplet states in Ir1 and Ir4 rather than the ³CT states for the other three complexes. Based on photophysical properties alone, Ir4 was predicted to be the best in vitro PDT agent.

All of the Ir(III) complexes of this study exhibited photobiological effects when activated with visible light, but were inactive with single-wavelength red light (625 nm). Thus, the in vitro PDT effects with broadband visible light were attributed to the shorter wavelengths. Some of the complexes showed selective cytotoxicity toward cancerous human melanoma cells over normal human skin fibroblasts. The photobiological trends could not be readily correlated to any differences in photophysical properties despite accessible long-lived ³π,π* states often resulting in red PDT activity. But the long-lived Ir4 did not yield a red PDT effect, nor did any of the other complexes due to the lack of ground-state absorption in the red. Rather, Ir3 emerged as a promising photosensitizer for further investigation owing to its nanomolar photocytotoxicity and visible PI > 280, with Ir1 and Ir5 also having suitable profiles. This small library of just five complexes proved to be a rich source of photophysical and photobiological diversity with only minor structural modifications. They gave ¹O₂ quantum yields that ranged from 4 to 38%, light EC₅₀ values from nanomolar to micromolar, dark toxicities that ranged from 32 to >140 μM, and DNA interactions that were characteristic for a particular cationic complex. For in vitro PDT applications in particular, there was a clear indication that the terminal tridentate N^∧C^∧N ligand performed best when combined with methyl substituents on the central cyclometalating ring and no ethynyl linkers between terminal ligands and the central fluorene unit. Thus, Ir3 will serve as the lead complex for future studies and as the parent complex of a second-generation library.