Photophysical Studies of Bioconjugated Ruthenium Metal-Ligand Complexes Incorporated in Phospholipid Membrane Bilayers

Abstract

Luminescent, mono-diimine, ruthenium complexes, [(H)Ru(CO)(PPh₃)₂(dcbpy)][PF₆] (1, dcbpy = 4,4′-dicarboxy bipyridyl) and [(H)Ru(CO)(dppene)(5-amino-1,10-phen)][PF₆] (2, dppene = bis diphenylphosphino-ethylene, phen = 9,10-phenanthroline), have been conjugated with 1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (DPPE) and with cholesterol in the case of 2. Compound 1 gives the bis-lipid derivative [(H)Ru(CO)(PPh₃)₂(dcbpy-N-DPPE₂)][PF₆] (3), while 2 provides the mono-lipid conjugate [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(S)-N-DPPE)][ PF₆] (4), and the cholesterol derivative [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(O)OChol)][PF₆] (5, Chol = cholesteryl), using standard conjugation techniques. These compounds were characterized by spectroscopic methods, and their photophysical properties were measured in organic solvents. The luminescence of lipid conjugates 3 and is quenched in organic solvents while compound 4 a weak, short-lived, blue-shifted emission in solution. The cholesterol conjugate shows the long-lived, microsecond-timescale emission associated with triplet metal-to-ligand charge-transfer (³MLCT) excited states. Incorporation of conjugate 3 in lipid bilayer vesicles restores the luminescence, but with blue shifts (~80 nm) accompanied by nanosecond-timescale lifetimes. In the vesicles conjugate 4 shows a similar short-lived and blue-shifted emission to that observed in solution but with increased intensity. Conjugation of the complex [(H)Ru(CO)(PhP₂C₂H₄C(O)O-N-succinimidyl)₂(bpy)][PF₆] (6”) with DPPE gives the phosphine-conjugated complex [(H)Ru(CO)(PhP₂C₂H₄C(O)-N-DPPE)₂(bpy)][PF₆] (7). Complex 7 also exhibits a short-lived and blue-shifted emission in solution and in vesicles as observed for 3 and 4. We have also conjugated the complex [Ru(bpy)₂(5-amino-1,10-phenanthroline)][PF₆]₂ (8) with both cholesterol (9) and DPPE (10). Neither 9 nor the previously reported 10 exhibited the blue shifts observed for 3 and 4 when incorporated into LUVs. The anisotropies of the emissions of 3, 4 and 7 were also measured in LUVs and of 5 in both glycerol and LUVs. High fundamental anisotropies were observed for 3 and 4 and 7.

Introduction

The objective of this study is to synthesize probes suitable for incorporation into biological membranes for membrane dynamics measurements. To achieve this objective, we have synthesized a series of luminescent probes derived from ruthenium-based metal complexes with long excited-state lifetimes and low molecular symmetry tethered to lipids or cholesterol.

The diffusion dynamics of proteins and protein assemblies that associate with membrane bilayers are slow, on a timescale of microseconds and longer, compared to the rotational diffusion of proteins in solution, which occurs on a timescale of several to tens of nanoseconds.¹ For example, the correlation times of the rotational motions of membrane-bound proteins can be microseconds to milliseconds.^2-5 The difference in timescales for these dynamical processes (microseconds versus tens of nanoseconds) is the result of interactions between the proteins and the membrane lipids. Fluorescence probes most useful for studying protein dynamics in solution have excited-state lifetimes in the range of 5–30 ns. Longer excited-state lifetimes are needed to measure dynamics of biomacromolecules on or in membranes. Microsecond and millisecond timescale dynamics are often studied by using phosphorescent probes.^6-8 Other techniques, such as EPR, using site-directed spin labeling are also useful for these purposes.⁹ However, excited-state probes potentially offer greater sensitivity for signal detection compared with EPR.

Transition-metal complexes containing one or more diimine ligands, exhibit tunable, long luminescence lifetimes (100 ns to ~ 10 μs), polarized emission, high photostability, large Stokes shifts and sensitivity to the probe environment.^10-11 In addition, the lifetimes of these probes can be tuned by varying the ligands attached to the metal center.^11,12 Microsecond excited-state lifetimes and polarized emissions make them useful probes for studying microsecond-timescale dynamics of membranes and macromolecular assemblies.

[Ru^II(bpy)₃]²⁺ and other similar transition-metal complexes are now extensively used to understand the nature of the charge-transfer excited state.^2,3,13-17 Typically these complexes contain diimine ligands such as 2,2′-bipyridyl (bpy), 1,10-phenanthroline and their derivatives 4,4′-dicarboxy-2,2′-bipyridyl (dcbpy) and 5-amino-1,10-phenanthroline (5-amino-1,10-phen), which provide low-energy π* orbitals for accepting the excited electron from the metal. Other ligands, such as phosphines, carbonyl and halides, can also be introduced along with the diimine ligands to tune the luminescence and solution properties. In these systems, the initial singlet excited state undergoes intersystem crossing with a quantum efficiency close to unity; the radiative lifetime of the triplet metal to ligand charge transfer (³MLCT) state reflects the degree of singlet-triplet mixing in the excited state due to strong spin-orbit coupling.^18,19 As a result, the luminescence lifetime and overall emission quantum yield of these complexes depends only on the radiative (k_r) and non-radiative (k_nr) decay rates of the triplet state. According to the energy gap law, k_nr increases exponentially as the emission energy decreases.^20-23 Other factors, such as Jahn-Teller distortion of the excited ¹MLCT state, also increase non-radiative decay (k_nr).^24-26 Therefore, in order to obtain luminescence from transition-metal complexes, a delicate balance of the metal’s and the ligand’s energy levels has to be established.

The highly polarized emission from some of these complexes stimulated our interest in using them as anisotropy probes for biophysical studies.^9,27 Luminophores covalently attached to macromolecules often undergo local (segmental) motions in addition to depolarization through global Brownian tumbling of the entire macromolecule. This results in complex anisotropy decays; time-resolved anisotropy measurements can be used to resolve information about motion, global motion, size and shape of the macromolecule, and flexibility of the system.⁵ From a practical point of view, the fundamental, zero-time anisotropy (r₀) should be at least 0.05 or greater.

The fundamental anisotropy is related to molecular symmetry. For example, [Ru^II(bpy)₂(dcbpy)]²⁺ and [Ru^II(bpy)₂(phen)]²⁺, which contain more than one type of diimine ligand, (i.e., less symmetric), show higher maximum fundamental anisotropies (excited near 490 nm, r₀ ~ 0.25 and ~ 0.175, respectively) than the more symmetric complex [Ru(bpy)₃]²⁺ (excited near 460 nm, r₀ ~ 0.13).⁵ Transition-metal complexes with a single chromophoric ligand have been reported for Re(I) and Ru(II) complexes (e.g., [Re(4,7-Me₂phen)(CO)₃(4-COOHPy)][PF₆]²⁸ and [(H)Ru(CO)(dcbpy)(PPh₃)₂][PF₆]¹¹) but their fundamental anisotropies have not been reported. Because low molecular symmetry would be expected to promote high anisotropy and high anisotropy is required for membrane dynamics measurements the complexes reported here were designed with one diimine ligand and in one case its anisotropy is compared with a tris-diimine complex.

Covalently attaching a ruthenium polypyridyl probe with a long-lived excited state onto either cholesterol or a phospholipid requires complementary functional groups for conjugation. Metal polypyridyl complexes with carboxylate or amine functional groups are suitable for covalent conjugation to lipids, cholesterol and proteins.^5,9,29 Phosphatidylethanolamine, a glycerophospholipid found in biological membranes, contains an amine group that can be reacted with a carboxyl group on the metal ligand via formation of an activated ester. The chloroformate derivative of cholesterol, on the other hand, can be covalently bound to an amine-substituted ligand. In both cases, the resulting conjugates can be easily incorporated into lipid-bilayer vesicles or biological membranes for photophysical measurements.^2,17

Here, we report phospholipid and cholesterol conjugates for the complexes [(H)Ru(CO)(PPh₃)₂(dcbpy)][PF₆] (1) and [(H)Ru(CO)(dppene)(5-amino-1,10-phen)][PF₆] (2), (dppene = 1,2-diphenylphosphino-ethene), along with a detailed analysis of their polarized emissions when incorporated into different types of lipid large unilamellar vesicles (LUVs). To understand the effects of conjugation through the diimine luminophore on the photophysical properties of these complexes, we also present an investigation of the first example of an transition-metal complex conjugated through the phosphine ligand using trans-[(H)Ru(bpy)(Ph₂PCH₂CH₂COOH)₂][PF₆] (6’) as the precursor. To our knowledge, this is the first such report. For comparison with the phosphine containing complexes 1-6’, we also report the photophysical properties of the cholesterol and mono-lipid conjugates of the complex [Ru(bpy)₂(5-amino-1,10-phenanthroline)][PF₆]₂ (8). The lipid conjugate of 8 was previously reported.¹⁷

Experimental

General Methods and Materials

Reactions were carried out under nitrogen. Purification was carried out in air by using preparative thin-layer chromatography (10×20 cm plates coated with 1 mm silica gel PF 60254-EM Science). Activated neutral alumina (Aldrich, 150 mesh, 58 Å) was also used to purify compounds by column chromatography.

Reagent grade solvents were purchased from J.T. Baker. Methylene chloride (CH₂Cl₂) and acetonitrile (MeCN) were distilled from calcium hydride. Tetrahydrofuran was distilled from benzophenone ketyl. Ruthenium carbonyl was purchased from Strem Chemicals. 2,2′-bipyridine, 1,10-phenanthroline, 4,4′dicarboxy-2,2′-bipyridyl, 5-amino-1,10-phenanthroline, cholesteryl-chloroformate and thiophosgene were purchased from Sigma-Aldrich and used as received. 1, 2-Dihexadecanoyl-sn-glycero-3-phosphoethanolamine (DPPE), 1,2-dihexadecanoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), and L-α-phosphatidylcholine from chicken egg (egg-PC) were purchased from Avanti Polar Lipids Inc. and used as received. The complexes [(H)Ru(CO)(PPh₃)₂(dcbpy)][PF₆] (1) and [(H)Ru(CO)(dppene)(5-amino-1,10-phen)][PF₆] (2) were synthesized according to published procedures.¹¹ The compounds [Ru(bpy)₂(5-amino-1,10-phenanthroline)][PF₆]₂ (8) and [Ru(bpy)₂ (1,10-phen-5-NHC(S)-N-DPPE)][PF₆]₂ (10) were synthesized according to literature procedures.^17b

¹H NMR and ³¹P{¹H} NMR spectra were obtained on a Varian 400 MHz Unity Plus or a Varian NMR Systems 500 MHz spectrometer. Infrared spectra were obtained on a Thermo-Nicolet 633 FT-IR spectrometer. ESI-MS spectra were obtained on a Watts/Micromass LCT using 80% MeCN as the carrier solvent.

Luminescence spectroscopy

Steady-state UV-visible absorption spectra and emission spectra were recorded on a Molecular Device Spectra Max M2. The emission quantum yields (φ) for the ruthenium complexes in the presence of oxygen were calculated relative to a Rhodamine B standard (φ = 0.73, in ethanol).^{11, 30a}

$$\mathrm{\phi }=\frac{\mathrm{abs\; Rhoda\; mine\; B}}{\mathrm{area\; Rhoda\; mine\; B}}\times \frac{\mathrm{area\; Ru\; complex}}{\mathrm{abs\; Ru\; complex}}$$ (1)

Here “abs” refers to the absorbance of the luminophores at the excitation wavelength and “area” refers to the integrated area under the emission spectral curve. In the case of compound 7 the quantum yield was measure by a similar procedure but using fluorescein as the standard because of the blue-shifted emission of this complex.^30b Details of the methods used for the time-resolved spectroscopy are given in the supplementary materials.^31-34

Synthesis (see Schemes 1-3)

Scheme 1.

Scheme 3.

Synthesis of (H)Ru(CO)(PPh₃)₂(dcbpy-N-succinimidyl)[PF₆] (1’)

A mixture of [(H)Ru(CO)(PPh₃)₂(dcbpy)][PF₆] (1),¹¹ (155 mg, 0.16 mol) and N-hydroxysuccinimide (34 mg, 0.32 mmol) was stirred in 4 mL of dry MeCN at room temperature in a 10-mL round-bottom flask until all the reactants dissolved. N,N′-dicyclohexylcarbodiimide (DCC) (103 mg, 0.48 mmol) was added to the mixture and the reaction was stirred for three hours. The solid precipitate (dicyclohexylurea) was removed by filtration through a 0.2 μm syringe filter. The filtrate was added to 5 mL of iso-propanol, and the mixture was kept at −4 °C to complete the precipitation. The supernatant was evaporated, and the remaining orange residue was washed three times with 2-mL aliquots of dry ethyl ether. Compound [(H)Ru(CO)(PPh₃)₂(dcbpy-N-succinimidyl)][PF₆] (1’) was obtained in 32% yield (60 mg). IR in KBr: CO stretching frequency at 1956 (vs), 1775 (m), 1742 (s), 1650 (m) and CH aliphatic 2980 cm⁻¹. ¹H NMR (CDCl₃) δ = 9.6–7.2(m, 36H), −11.1(t, 1H), 2.8 (4H). ³¹P{¹H} NMR (CDCl₃) δ = 49.2 (s, 2P), −155(m, 1P).

Synthesis of [(H)Ru(CO)(PPh₃)₂(dcbpy-N-DPPE₂)][PF₆] (3)

30 mg of DPPE (0.043 mmol) was dissolved in CHCl₃ and 3.5 mL of triethylamine was added to the solution. The mixture was stirred for 15 min and then a solution of complex 1’ (60 mg, 0.021 mmol) in 2 mL dry MeCN was added drop-wise over 20 minutes. The reaction was stirred overnight and then the solvent was removed by rotary evaporation. The residue was purified by thin layer chromatography on silica gel. Two successive elutions with a mixture of hexane/methylene chloride/ethanol {6.5:3.5:0.5 (v/v)} yielded two bands. The baseline contained un-reacted complex 1’. The faster moving UV-absorbing band was identified as un-reacted DPPE, and the slower moving deep yellow band gave [(H)Ru(CO)(PPh₃)₂(dcbpy-N-DPPE₂)][PF₆] (3) in 15% yield (22 mg). IR in KBr: CO stretching frequency at 1956 (vs), 1734 (s), 1684 (vs) and CH aliphatic 2963 (s), 2924 (s), 2851 (m) cm⁻¹. ¹H NMR (CDCl₃) δ = 9.5–7.0 (m, 36H), 5.2(2H), 5.1–2.2 (35H), 1.9–0.78 (107H), −11.19 (broad, 1H); ³¹P{¹H} NMR (CDCl₃) δ = 49.6 (s, 2P), 25.04 (2P), −155 (m, 1P). ESI-MS: m/z 2034 [M⁺ − (C₁₅H₃₁ + PF₆)] (Calc. M⁺ − (C₁₅H₃₁ + PF₆) = 2034).

Synthesis of [(H)Ru(CO)(dppene)(1,10-phen-5-NCS][PF₆] (2’)

122 mg (0.13 mmol) of [(H)Ru(CO)(dppene)(5-amino-1,10-phen)][PF₆] (2),¹¹ was dissolved in 3 mL of dry acetone. Finely crushed CaCO₃ (45 mg, 0.45 mmol) was added to the solution of 2 followed by addition of thiophosgene (11 μL, 0.07 mmol). The reaction mixture was stirred at room temperature for 1 hour and then refluxed for 2.5 hour. After cooling to room temperature, CaCO₃ was removed by using a 0.45-μm filter, and acetone was removed by rotary evaporation. Compound [(H)Ru(CO)(dppene)(1,10-phen-5-NCS][PF₆] (2’) was obtained in 94% yield (50 mg). IR in KBr: CO stretching frequency at 1990 (vs), N=C=S at 2119 (m) and 2046 (m) cm⁻¹. ESI-MS: m/z 860 [M⁺ − PF₆] (Calc. M⁺ − PF₆ = 860).

Synthesis of [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(S)-N-DPPE][PF₆] (4)

A solution of [(H)Ru(CO)(dppene)(1,10-phen-5-NCS][PF₆] (2’) (50 mg, 0.049 mmol in 3-mL dry CH₂Cl₂) was added drop-wise into a stirring solution of DPPE (35 mg, 0.048 mmol in 5 mL dry CH₂Cl₂) over 1 hour at room temperature and the reaction was stirred overnight. Solvent was removed by rotary evaporation and the residue purified by thin-layer chromatography on silica plates. Three bands were resolved by elution with hexane/methylene chloride/methanol {3:6:2 (v/v)}. The fastest moving UV-absorbing band was identified as unreacted DPPE and the second moving yellow band was too small for further characterization. The slowest moving, deep-yellow band yielded [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(S)-N-DPPE][PF₆] (4) in 10% yield (15 mg). IR in KBr: CO stretching frequency at 1993 (vs), 1735 (vs), cm⁻¹; NH stretching at 3422 and aliphatic C–H stretching at 2920 (vs), 2849 (vs) cm⁻¹. ¹HNMR (CDCl₃) δ = 7.5–6.6 (m, 29H), 5.32 (s,br 1H), 4.0–3.4 (m, 9H), 2.9–0.2 (63H), -7.80 (1H). ³¹P{¹H} NMR (CDCl₃) δ = 68.30 (s, 2P), 58.19 (br, 1P), −145 (m, 1P).

Synthesis of [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(O)OChol)][PF₆] (5)

100 mg (0.10 mmol) of [(H)Ru(CO)(dppene)(5-amino-1,10 phen)][PF₆] (2) was dissolved in 15 mL dry CH₂Cl₂ and 1 mL dry MeCN, and then 1 mL triethylamine was added to the deoxygenated solution. A 10 mL CH₂Cl₂ solution of cholesteryl-chloroformate (45 mg, 0.10 mmol) was added to the probe solution drop-wise over 20 minutes, and the mixture was refluxed for 5 hours. Progress of the reaction was monitored by the disappearance of the peak at 1776 cm⁻¹ in the IR spectrum, corresponding to the chloroformate, and by the appearance of a new peak at 1730 cm⁻¹, corresponding to the amide. The solvent was removed by rotary evaporation and the residue purified by thin layer chromatography on silica gel. Elution with hexane/methylene chloride/methanol {1:1:1 (v/v)} yielded two bands. The [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(O)OChol)][PF₆] (5) was recovered in 20% yield (30 mg) from the orange, slower moving band while the faster UV-absorbing band contained unreacted cholesteryl-chloroformate. IR in KBr: CO stretching frequency at 1997 (vs), 1976 (vs), 1735 (s), and CH aliphatic 3054(w), 2926 (vs), 2850 (s) cm⁻¹. ¹H NMR (CDCl₃) δ = 8.5–6.5(m, 29H), 6.05 (2m, 1H), 4.3 (s,1H), 2.0–0.5 (44 H), -7.90 (m,1H). ³¹P{¹H} NMR (CDCl₃) δ = 75.71 (s, 2P), −145 (m, 1P).

Synthesis of [(TFA)₂Ru(CO)₂(PPh₂C₂H₄C(O)OH] (6)

A THF solution of K[Ru(CF₃CO₂)₃(CO)₃]¹¹ (500 mg, 0.90 mmol) and diphenylphosphino-propionic acid (425 mg, 1.8 mmol) was heated overnight at 45 °C. The solvent was removed by rotary evaporation and the residue vacuum dried yielding 625 mg (81%) as a pale yellow solid. IR in KBr: 2023 (vs), 2010 (vs), 1960 (m), 1790 (m, br), 1685 (vs, br) cm⁻¹. ¹H NMR in d₆-acetone : δ 7.9–7.3 (m, 20H), 3.90 (m, 2.8H, isomer a), 3.14 (m, 1.2H, isomer b), 2.57(m, 1.2H, isomer b), 2.10 (m, 2.8H, isomer); ³¹P {H} NMR: δ 26.59 (dt br).

Synthesis of [(H)Ru(CO)(PPh₂C₂H₄C(O)OH)₂(bpy)][PF₆] (6’)

The reaction of [(TFA)₂Ru(CO)₂(PPh₂C₂H₄C(O)OH)₂] (6) (300 mg, 0.35 mmol) with 2,2′-bipyridyl (55 mg, 0.35 mmol) in ethylene glycol (15 mL) was heated at 140°C for 72 h producing an orange solution. A deep-orange precipitate was obtained by the addition of NH₄PF₆ in DI water (1.0 g/10 mL) dropwise until precipitation. The precipitate was filtered and washed three times with cold DI water, three times with diethyl ether and dried under vacuum. Complex [(H)Ru(CO)(PPh₂C₂H₄C(O)OH)₂(bpy)][PF₆] (6’) was obtained in 41% yield (135 mg). IR in KBr: 1971 (vs), 1730 (s), 1740 (vs), 1605 (s) cm⁻¹. ¹H NMR in d₆-acetone: δ 8.38–6.95 (m, 28H), 3.99 (t, 4H), 3.61 (t, 4H), -11.1(t, 1H); ³¹P {H} NMR: 43.06 (s, 2P), −145 (m, 1P).

Synthesis of [(H)(CO)Ru(PPh₂C₂H₄C(O)-N-succinimidyl)₂(bpy)][PF₆] (6”)

The succinimidyl derivative was obtained by dissolving [(H)Ru(CO)(PPh₂C₂H₄C(O)OH)₂(bpy)][PF₆] (6’) (100 mg, 0.106mmol) in 5mL of MeCN in a round bottom flask at 0°C along with N-Hydroxysuccinimide (25mg, 0.212mmol) and N,N’-dicyclohexylcarbodiimide (DCC) (65mg, 0.32mmol) overnight. After stirring, the reaction was passed through a 0.2 μm syringe-filter to remove urea that had precipitated. The filtrate was added to an excess of cold isopropanol and recrystallized. The resulting precipitate was filtered and washed three times with diethyl ether. [(H)(CO)Ru(PPh₂C₂H₄C(O)-N-succinimidyl)₂(bpy)][PF₆] (6”) was obtained in 58% yield (70mg, 0.061mmol). IR in KBr: CO stretching frequency at 1939 (s), 1780 (s), 1736 (vs) and CH aliphatic 2930 (vs), 2853 (s) cm⁻¹. H NMR (CDCl₃) δ = 8.6–6.7 (28H), 4.3–3.2 (8H), 2.95-2.8 (t, 8H), -11.3(t, 1H). ³¹P{H} NMR 35.4 ppm(2P) and -145(1P).

Synthesis of [(H)(CO)Ru(PPh₂C₂H₄C(O)-N-DPPE)₂(bpy)][PF₆] (7)

An acetonitrile solution of [(H)Ru(CO)(PPh₂C₂H₄C(O)O-N-succinimidyl)₂(bpy)][PF₆] (6”) (60 mg, 0.048 mmol) was added drop-wise into a stirring methylene chloride solution of DPPE (68 mg, 0.096 mmol) in the presence of a catalytic amount of triethylamine. The reaction mixture was stirred overnight at ambient temperature. The solvent was removed on a rotary evaporator and the residue was purified by thin-layer chromatography on silica. Elution with hexane/methylene chloride/methanol (6:3:1) on silica gave a slower moving yellow band and a faster moving UV band with a heavy yellow baseline. The yellow compound on the baseline and the UV absorbing band were identified as un-reacted complex 6” and DPPE, respectively. The yellow band on the TLC plate gave [(H)Ru(PPh₂C₂H₄C(O)-N-DPPE)₂(bpy)(CO)][PF₆] (7) in 20% (~ 20 mg) yield. IR in KBr: CO stretching frequency at 1941 (s), 1735 (vs), 1653 (m) and CH aliphatic 2960 (s), 2918 (vs), 2850 (s) cm⁻¹. ¹H NMR (CDCl₃) δ = 8.6–6.7 (28H), 5.2–3.2 (44H), 3.0–0.4 (108H), -11.1(t, 1H). Peaks in both aliphatic and aromatic regions were broad. ³¹P: the phosphine peak and phosphate peak of the lipid were also broad and appeared at 37.94 (2P) and 22.5(2P) ppm respectively the PF₆ peak was at −145(1P) ppm.

Synthesis of [Ru(bpy)₂(1,10-phen-5-NHC(O)OChol)][PF₆]₂ (9)

[Ru(bpy)₂(5-amino-1,10-phen)][PF₆]₂ (8) was prepared according to a published method.³⁹ 100 mg (0.11 mmol) of 8 was dissolved in 10 mL dry CH₂Cl₂ and then 1 mL triethylamine was added to the deoxygenated solution. A 5-mL CH₂Cl₂ solution of cholesteryl-chloroformate (50 mg, 0.11 mmol) was added to the probe-containing solution drop-wise over 20 minutes, and the reaction mixture was refluxed for 4 hours. Progress of the reaction was monitored by the disappearance of the peak at 1776 cm⁻¹ in the IR spectrum, corresponding to the chloroformate, and by the appearance of a new peak at 1731 cm⁻¹, corresponding to the amide. The solvent was removed by rotary evaporation, and the residue was purified by thin layer chromatography on silica gel. Elution with the solvent mixture hexane/methylene chloride/methanol {1:2:1 (v/v)} yielded two bands. The [Ru(bpy)₂(1,10-phen-5-NHC(O)OChol)][PF₆]₂ (9) was recovered in 14 % yield (23 mg) from the orange, slower moving while the faster, UV-absorbing band contained un-reacted cholesteryl-chloroformate. IR (KBr) (υ cm⁻¹): CO stretching frequency at 1731 (s) and CH aliphatic 3139 (w), 2950 (vs), 2868 (s). ¹H NMR (CDCl₃) δ = 8.7–7.0 (24H), 5.37 (1H), 3.99 (1H), 2.0–0.5 (43 H).

Large Unilamellar Vesicle (LUV) preparation

A chloroform mixture of conjugated probe (3, 4, 5 or 7) and either DPPC, DMPC or a mixture of phospholipids containing a choline head group (egg-PC) was prepared in a molar ratio of 1:99. The organic solvent was removed by evaporation with argon gas, and the lipid/chromophore mixture was further dried under vacuum overnight. Then 0.52 mL HBS buffer (20 mM HEPES, 100 mM NaCl, pH 7.5) was added to the dried lipid and the solution was maintained above the phase-transition temperature of the corresponding PC (41 °C for DPPC, 23 °C for DMPC and less than 0 °C for egg-PC) ³⁴ to obtain a final lipid concentration of 1 mM. Addition of buffer to the lipid mixture produced cloudy suspensions. The suspensions were incubated above the phase-transition temperature for 1 hr with occasional stirring. Then a freeze/thaw cycle was carried out five times. Finally, clear suspensions of ~100-nm-diameter LUVs were obtained by extrusion through a 100-nm sizing membrane as previously described.³⁵

Results

Synthesis

Schemes 1 and 2 describe the ligand modification and conjugation of the ruthenium probes with lipids and cholesterol. For the phospholipid conjugations, we used diimine ligands containing either activated ester or highly reactive isothiocyanate functional groups. Complex 1 contains a 2,2′-bipyridyl ligand with two carboxylic acid groups, which were converted to the activated ester groups, and the activated ester groups were then used to form a peptide bond with the primary amine of DPPE. Complex 3, conjugated to two DPPE molecules, was synthesized and purified by standard chromatographic methods. Complex 4 was obtained by first converting the amine group on the 5-amino-1,10-phenanthroline ligand of complex 2 into 5-isothiocyanato-1,10-phenanthroline (SCN-phen), and then one molecule of DPPE was conjugated with the ruthenium probe through formation of a thiourea bond between SCN-phen-Ru and the primary amine group of DPPE. Because cholesterol is an important component of biological membranes, we synthesized the cholesterol conjugate of the ruthenium complex, 2. The amino group of the 5-amino-1,10-phenanthroline was used to form the amide bond in complex 5 by reacting complex 2 with the highly reactive cholesteryl-chloroformate.

Scheme 2.

All conjugated transition metal complexes, reported here, were characterized by IR, and ¹H and ³¹P NMR. In the IR, the terminal M–CO shows CO stretching modes around 2150–1850 cm⁻¹. Complexes 1–5 have only one M–CO ligand. The strong M–CO stretch appears at 1949 and 1956 cm⁻¹ for the complexes 1 and 3 respectively, and complexes 2, 4, and 5 showed strong M–CO stretches at 1990–1997 cm⁻¹. Strong absorptions in the organic carbonyl region were also observed for the carboxy-amide functional group in complex 5 and for the glycero-ester groups lipids in complexes 3 and 4. Medium intensity absorptions at 2102–2050 cm⁻¹ are observed for 4, which are assignable to the iso-thiocyanate (N=C=S) stretch.

The ¹H and ³¹P{¹H} NMR spectra of complexes 1’ and 3 obtained in CDCl₃ NMR are consistent with the proposed structures. The M–H resonance appeared as a triplet at δ −11.07 (J = 20 Hz) for complexes 1’ and a broad multiplet at δ −11.19 upon conjugation with lipids in complex 3. The hydride resonances for complexes 2, 4, and 5 appear as triplets at δ -7.61, -7.5 and -7.6 respectively. The aromatic region of the ¹H spectra is complex due to the presence of phenyl protons of phosphines ligands and the aromatic protons of diimine ligands. The CH=CH protons of diphenylphosphinoethylene are observed δ 6.2–6.9 ppm for complexes 2, 4 and 5. The conjugates showed chemical shifts in the aliphatic regions characteristic of the corresponding lipid and cholesterol. The ¹H NMR resonances for the lipid and cholesterol conjugates are slightly broader than those of the un-conjugated complexes (See supplementary materials Figures S6-S10). This is likely due to the fact that rotational correlation times of the complexes are long, which means that the molecules are not orientationally averaged and therefore do not display sharp signals. This could also be the result of aggregate formation in the polar organic solvents used.

The chemical shift of the metal-bound phosphines ligands in the ³¹P NMR spectra are in good agreement with those of similar Ru(II)phosphine complexes.¹¹ Complexes 1–5 show singlet resonances at 49.2–75.7 relative to external H₃PO₄, which are due to the triphenyl and diphenylphosphinoethylene ligands. The singlet observed for these complexes indicated that they have a symmetry plane that makes the two phosphorus nuclei magnetically equivalent in the case of 1 and 3, consistent with the proposed structures. That singlets are observed for 3–5, as well, suggests that the asymmetry in phenanthroline ring is not sufficient to preclude overlap of the phosphine resonances. This is also the case for 2.¹¹ The ³¹P resonances for the lipid phosphorus atoms are observed at 25.0 (2P) and 58.19 (1P) ppm for complexes 3 and 4. The higher-frequency shift in 4 relative to 3 might result from the different modes of binding to the diimine ring or to conformational effects. The counter anion [PF₆] appeared as a septet at −155 ppm in all the ³¹P NMR with an integrated relative intensity of 1:2 when compared with the phosphine ligand resonances.

To evaluate the effect of the site of lipid conjugation on the photophysical properties of the complexes in LUVs, we synthesized the complex [(H)Ru(PPh₂C₂H₄C(O)O-N-DPPE₂)₂(bpy)(CO)][PF₆] (7) (Scheme 3). This was done by converting the common starting material [K][Ru(CO)₃(TFA)₃] to the bis-diphenylphosphino-propionyl carboxylate, by reaction with diphenylphosphino-propionic acid (DPPA) which gave trans-[Ru(PPh₂C₂H₄C(O)OH)₂(CO)₂(TFA)₂] (6, two isomers were observed by ¹H NMR), followed by reaction with bipyridine giving (H)Ru(bpy)(PPh₂C₂H₄C(O)H)₂(CO)][PF₆] (6’). The bis–lipid conjugate was obtained by conversion of 6’ to the activated ester derivative [(H)Ru(bpy)(PPh₂C₂H₄C(O)-succinimidyl)₂(CO)] (6”). Then conjugation with DPPE, using a procedure similar to that used for 3, gave trans-[(H)Ru(PPh₂C₂H₄C(O)-N-DPPE)₂(bpy)(CO)][PF₆] (7) (Scheme 3). The complexes were characterized spectroscopically at each stage of the synthesis, to confirm evidence of formation of the expected analogs of 1 and 3. Under the conditions used for reaction with 2, 2’-bipyridine, refluxing ethylene glycol, all complexes are converted to their corresponding hydrides.

It should be noted here that complexes 4 and 5 are chiral, while complexes 3 and 7 are not, by virtue of the symmetry plane perpendicular to the two trans-phosphines and containing the other ligands. Because we observe only one set of NMR resonances for both complexes, the chemical shift differences for the diastereomers of 4 and 5 are not large enough to be resolved or only one of the diastereomers is populated.

Photophysical characterization of complexes 1-5, 6’, 7-10

Table 1 lists the absorption and emission maxima and the luminescence lifetimes for complexes 1–7 in ethanol. All of the compounds show intense, higher-energy absorptions at 270–295 nm due to the spin-allowed intra-ligand (π-π*) transitions. These absorptions are not shown in Table 1 in order to focus on the more important MLCT and phosphine absorptions. In the case of complex 7 the absorption at 295 nm is due the phosphine. The absorptions of this complex are all blue shifted relative to the others including the MLCT (vide infra) and this born out by the excitation spectra (see supplementary materials Figure S1). The absorptions observed between 356-366 nm for complexes 2, 4 and 5 are due to the presence of the double bond in the chelating phosphine ligand of these complexes. The less intense absorption bands (ε₄₅₀ ~ 2 ×10³ M⁻¹cm⁻¹) of all probes and their conjugates in the visible region (410–490 nm) are attributed to spin-allowed ¹MLCT (d-π*) transitions. The ¹MLCT absorption bands of the complexes containing dcbpy are at slightly lower energy than the lipid derivative complex 3. In the case of complexes 4 and 7 the MLCT absorption is blue shifted to ~400 nm (see supplementary materials Figures S4 and S5).

Table 1.

Absorption, emission, and excited-state lifetimes of ruthenium MLC probes in ethanol.

Compound	λab (nm)	λem (nm)	τ (μs)	φ
1 [HRu(CO)(PPh3)2(4,4’-dcbpy)][PF6]	303, 468	647	0.72	0.30a
2 [(H)Ru(CO)(dppene)(5-amino-1,10-phen)][PF6]	364, 442	610	0.25	0.25a
3 [HRu(CO)(PPh3)2(dcbpy-N-DPPE2)][PF6]	316, 442	----	----	-----
4 [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(S)-N-DPPE)][PF6]	360,400	520	-----	-----
5 [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(O)OChol)][PF6]	356, 440	605	0.47	0.49
6’ [(H)Ru(CO)(dppa)2(bpy)][PF6]	460	608	0.27	0.50b
7 [(H)Ru(CO)(dppa-N-DPPE)2(bpy)][PF6]	295, 400	505	0.004	0.019
8 [Ru(bpy)2(5-amino-phen)][PF6]2	350,445	625	0.22
9 [Ru(bpy)2()(1,10-phen-5-NHC(O)OChol)][PF6]2	350,445	625	0.22	0.25
10 [Ru(bpy)2(1,10-phen-5-N-DPPE)][PF6]2c	330, 460	625	0.22

^aFrom reference 11

^bThis work

^cFrom reference 17,39

All the complexes containing the chelating phosphine and phenanthroline ligands displayed ¹MLCT absorption bands at similar wavelengths. In ethanol, acetonitrile or methylene chloride, complexes 1, 2, 5 and 6’ displayed long-lived, orange-red luminescence characteristic of a ³MLCT excited state (Figure 1; the 5 and 6’ spectra – not shown – are very similar to that of 2 and 1 respectively). The conjugation with cholesterol (complex 5) resulted in an approximate two-fold increase of excited-state lifetime. No emission was observed from complex 3 and a very weak emission at 520 nm was observed for complex 4. This emission had a short lifetime (4-5 ns) and had an ill-defined excitation spectrum (see Figure S4). Complex 6’ exhibits ³MLCT emission at 608 nm, interestingly with a much shorter lifetime but with a higher quantum yield than 1. Complex 7 showed a blue-shifted ¹MLCT absorption band with a peak near 400 nm; excitation at 450 nm gave an emission with a maximum at 505 nm with a lifetime of ~4.46 ns in chloroform at 5 °C (Table 1). The quantum yield of this emission was found to be 0.019, making this a very weak singlet emission. Thus, bis-lipid conjugation via the phosphine ligand does not cause quenching of the luminescence as seen for 3, but gives a short-lived, blue-shifted emission in ethanol as observed for 4. Complexes 8-10, on the other hand, show identical long-lived ³MLCT emissions with a peak near 625 nm. The absorption and emission spectra of complex 9 are shown in Figure 2.

Figure 1.

Peak-normalized emission spectra of complex [HRu(CO)(PPh₃)₂(dcbpy)][PF₆] (1) and [(H)Ru(CO)(dppene)(5-amino-1,10-phen)][PF₆] (2) in ethanol.

Figure 2.

Absorption and emission spectra of complex [(H)Ru(CO)(dppene)(1,10-phen-5-NHC(O)OChol)][PF₆] (5) and complex [Ru(bpy)₂(1,10-phen-5-NHC(O)OChol)][PF₆]₂ (9) in ethanol.

Analysis of the time-resolved anisotropy decay of complexes 1, 2 and 5 in neat glycerol at 0 °C and with excitation at 470 nm, yielded r₀ values of 0.124, 0.077 and 0.121, respectively.

Photophysical studies of complexes 3-5 and 7, 9, 10 incorporated in lipid membrane bilayers

Lipid conjugates 3, 4, 7 and 10 and cholesterol conjugates 5 and 9 were incorporated in LUVs to study the photophysical properties of these probes in a membrane-like environment. The maximum of the low-energy absorption band was near 440 nm except for 4 and 7, which had this absorption at ~400. The dynamics of these probes incorporated in the LUVs was determined from the kinetics of the time-resolved emission anisotropy.

Although the absorption spectrum for complex 3 from 400 to 550 nm was characteristic of the charge-transfer band and essentially identical in chloroform, ethanol, and lipid LUVs—emission was only observed when complex 3 was incorporated in LUVs. Furthermore, the emission spectrum of complex 3 in the LUVs was blue-shifted (λ_max = 534 nm) with respect to the precursor probe, 1 (λ_max = 647 nm in ethanol, Table 1 and Figure 1). Complex 3 also exhibited a very short, excited-state lifetime (11 ns at 5 °C, air equilibrated) in PC LUVs. Complex 4 in ethanol solution showed a weak short lived emission at 520 nm. Complex 4 in PC LUVs also had a blue-shifted emission (545 nm) with a short lifetime (8 ns) (Figure 3) similar to its emission in solution but with much higher intensity. Both complexes showed more intense emission in LUVs compared to that of the red-shifted emission of the unconjugated precursors 1 and 2 in ethanol. The emission yield of complex 3 was greater than that of 4, as is the case for bipyridyl complex 1 relative to phenanthroline complex 2. Complex 7 showed the same blue-shifted emission in the LUVs as in ethanol. Complex 10 did not show this blue shift when incorporated in LUVs but did show a factor of two increase in excited-state lifetime. (0.22 to 0.52 μs).

Figure 3.

Peak-normalized emission spectra of complex [(H)Ru(CO)(PPh₃)₂(dcbpy-N-DPPE₂)][PF₆] (3), and [(H)Ru(CO)(dppene)(1,10-phen-5-NC(S)-N-DPPE][PF₆] (4) in egg-PC LUVs.

To eliminate the possibility that the blue-shifted, short-lifetime emissions of 3 and 4 in lipid LUVs were due to decomposition in the lipid bilayer, we synthesized a bis-lipid derivative of 4,4′dicarboxy-bipyridyl (dcbpy-N-DPPE₂) (11, see supplementary materials), and compared the photophysical behavior of this compound in egg-PC LUVs. This conjugate, which lacks the metal center, showed a less intense absorption band at 327 nm and an intense absorption band at 295 nm, characteristic of the un-conjugated dcbpy ligand. Further, the emission maximum of 11 in PC-LUVs was at 405 nm (excitation at 327 nm), not near 534 nm, and complex intensity decay kinetics were observed with a 5-ns intensity-averaged lifetime, <τ>. In another experiment, we prepared PC-LUVs without any probe incorporated. As expected, there was no emission whether excited at 327 or 450 nm. These LUVs, which lacked probe, were then incubated at 35 °C with complex 3, previously dissolved in THF (THF was approximately 2% of final volume), to adsorb the probe onto the LUVs. In contrast to the conjugate incorporated in LUVs by the standard reconstitution procedure, described previously, the emission spectrum of the bis-lipid-conjugate adsorbed onto the preformed LUVs had its maximum at 620 nm, characteristic of ³MLCT luminescence. However, when this preparation was subsequently extruded through the sizing membrane, the blue-shifted emission with a maximum near 530 nm was once again observed. These results indicate that the blue-shifted emission and short, ns-timescale excited-state lifetime observed for complex 3 are not due to decomposition of the complex to a free bpy-DPPE moiety, but are features of the system when the probe is incorporated into the LUV bilayer. This conclusion is also supported by the observation that the ¹MLCT absorption band of complex 3 is the same both in alcohol solution and in PC-LUVs.

A progressive decrease in the blue-shifted luminescence intensity with increasing temperature was observed over the temperature range 5 to 50 °C (Table 2). The change in excited-state lifetime and anisotropy decay of the blue-shifted emission of complexes 3 and 4 incorporated in PC LUVs was also measured over a range of temperatures to determine the sensitivity of these probes toward changes in the micro-viscosity of the bilayer environment. The excited-state lifetime decreased progressively with increasing temperature, consistent with the decrease in luminescence expected for quenching by thermally activated motions. An increase in the local motions, as reflected by the decrease in the rotational correlation times, was also observed with increasing temperature. The blue-shifted emission of lipid-conjugated probes 3 and 4 showed high fundamental anisotropy values (excitation at 470 nm, r₀ = 0.24 and 0.35 respectively) in LUVs compared to those of the red-shifted emission of complexes 1 (r₀ = 0.12), 2 (r₀ = 0.08) and 5 (r₀ = 0.12) in glycerol. The results of analyses of the time-resolved anisotropy data in terms of a double exponential decay for complexes 3 and 4 in LUVs at variable temperature are summarized in Table 2. At lower temperatures, the anisotropy decay revealed a significant contribution from the limiting anisotropy at infinite time (r_∞); a non-zero r_∞ is indicative of restricted motion in the membrane.³² Compound 7 was examined in DMPC LUVs and showed a slightly longer lifetime of 4.56 ns and a very high fundamental anisotropy of 0.31 and a significant r_∞ of 0.103. These properties closely parallel those observed for 3 and 4 in LUVs. The variable temperature study of this emission showed very little variation in lifetime over the range of 0 to 30 °C, which is likely due to the low quantum yield observed for 7 in solution.

Table 2.

Average lifetime, limiting anisotropy and rotational correlation times for complexes 3 and 4 in egg-PC LUVs (100 nm) from 5–50°C.

Compound	Temp. (°C)	<τ>a (ns)	r∞	θ1 (ns)b	θ2 (ns)b	χ2
3	5	11	0.096	9.8(-1.62, 1.69)	2.46(-0.21, 0.24)	1.18
	10	9.4	0.07	8.62(-1.68, 2.4)	1.95(-0.18, 0.21)	1.10
	20	7.9	0.05	5.5(-1.31, 2.05)	1.2(-0.17, 0.18)	1.13
	30	6.7	0.03	4.1(-0.55, 0.68)	1.0(-0.06, 0.07)	1.19
	40	5.5	0.02	3.2(-1.84, 1.89)	0.58(-0.51, 1.0)	1.15
	50	4.5	0.01	1.6(-0.15, 0.42)	0.26(-0.06, 0.18)	1.14
4	5	7.2	0.09	8.4(-0.624, 1.2)	0.79(-0.21, 0.25)	1.19
	10	6.6	0.07	6.3(-0.54, 0.62)	0.83(-0.11, 0.13)	1.1
	20	5.8	0.04	5.4(-1.35, 2.01)	0.5(-0.03, 0.04)	1.2
	30	4.8	0.02	3.3(-0.15, 0.16)	0.36(-0.03, 0.02)	1.08
	40	3.8	0.01	2.0(-0.10, 0.11)	0.29(-0.04, 0.05)	1.0
	50	3.1	0.008	1.3(-0.08, 0.085)	0.13(-0.04, 0.041)	1.1

^aIntensity-average lifetime, <τ> = Σα_i τ_i² / Σα_i τ_i.

^bUpper and lower 95% confidence limits, calculated by support-plane method, are indicated within parenthesis. Johnson, M. L.; Frasier, S. G. in Methods in Enzymology Vol. 117, Academic Press, New York, 1985, p301.

The absorption and emission spectra of the cholesterol-conjugates complexes 5 and 9 in ethanol are shown in Figure 2. In contrast to the lipid conjugates, complexes 3 and 4, the emission spectra of complexes 5 and 9 are red-shifted and identical to those observed when incorporated in the egg-PC LUVs. In addition, the excited-state lifetimes, which at 23 °C were 0.47 and 0.22 μs for 5 and 9 in ethanol, increased to 0.89 and 0.52 μs, respectively, for the complexes in egg-PC LUVs.

Complexes 4 and 7 both showed blue-shifted luminescence with a maximum near 505 nm and a lifetime of ~4-5 ns in egg-PC LUVS, similar to that observed in ethanol. This indicates that the large blue shifts and short lifetimes observed for the emissions of 3 and 4 in LUVs are likely due to large perturbations in the geometry and/or electronic energies of the excited states.

In Complex 10 the lipid is conjugated to the phen rather than bpy ligand, which is the likely luminophore, does not show a blue shift and has an excited state lifetime typical of a ³MLCT (0.41 μ). The perturbations that result in the blue shifts and short lifetimes for 3, 4 and 7 are likely the result of conjugation of the large lipid molecules directly to the luminophore or to an ancillary ligand (phosphines) that makes a significant contribution to the MLCT excited-state, but this is not the case for 10.^17,39

Egg-PC has a low phase-transition temperature (less than 0 °C) because it contains mixed saturated and unsaturated acyl chains of different lengths, leading to a highly disordered phase. To understand the effect of a more ordered membrane on the observed rotational correlation times, we measured the photophysical properties of 5 incorporated in 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC) LUVs, which has two 15–carbon saturated acyl chains. The phase-transition temperature for DPPC is 41 °C;³⁵ the bilayer is in an ordered phase below this temperature.

As in egg-PC, the emission of 5 was red-shifted in DPPC. An analysis of the time-dependent anisotropy decay of 5 incorporated in either egg-PC or DPPC LUVs resulted in a fundamental anisotropy value of ~0.1. A single exponential satisfactorily fit the time-resolved intensity decay of 5. In the DPPC LUVs, the luminescence lifetime of 5 ranges from 1.10 μs at 10 °C to 0.43 μs at 50 °C. This temperature range spans the phase-transition temperature of DPPC (41 °C). In egg-PC LUVs, the lifetime is comparable (0.96 μs at 10 °C and 0.45 μs at 50 °C (Figure 4)). The long decay times suggest that these probes can be used to measure rotational motions as long as 3 μs (three times the mean intensity decay time). ^4,5

Figure 4.

Average lifetime of complex [(H)Ru(CO)(dppene)(1,10-phen-5-NC(O)OChol)][PF₆] (5) in LUVS over a range of temperatures. Error bars are based on the errors in the non-linear least squares fit using the support plane method developed by Johnson, M. L.; Frasier, S. G. in Methods in Enzymology Vol. 117, Academic Press, New York, 1985, p301.

The rotational motions of 5 in egg-PC LUVs were also analyzed over a range of temperatures. The rotational correlation time decreased from 112 ns to 14 ns as the temperature increased from 10 °C to 50 °C (Table 3). The recovered rotational correlation times are not due to the overall rotation of the 100-nm diameter LUVs, which would be much slower (sub-millisecond range), but are due to local motions. There is considerable uncertainty in measuring longer correlation times of the LUVs because of the difficulty of measuring accurately a correlation time above 3 μs with a probe of 1 μs lifetime. Considering its luminescence lifetime, probe 5 would be more appropriate for studying the overall rotational motion of small unilamellar vesicles (SUVs) of sizes less than 20-nm diameter, which would be in the sub-μs range. The time-dependent anisotropy decays at variable temperature were analyzed using single-exponential correlation times and a nonzero baseline limiting anisotropy (r_∞), which reflects restricted motion of the probe during the lifetime of the excited-state.^32,36-38

Table 3.

Average lifetime, limiting anisotropy and rotational correlation times for complex 5 at a range of temperature in 100-nm egg-PC LUVS.

Temp. (°C)	<τ>a (μs)	r∞	ϕ (ns)b	χ2
5	1.46	0.058	71(-1.61, 2.0)	1.00
10	1.24	0.055	54(-1.47, 1.93)	1.02
20	0.94	0.046	49(-1.46, 2.0)	1.08
30	0.68	0.051	44(-1.51, 2.4)	1.08
40	0.57	0.049	24(-1.2, 2.27)	1.18
50	0.47	0.050	10(-4.21,7.2)	0.98

^aIntensity-averaged lifetime, <τ> = Σα_i τ_i² / Σα_i τ_i

^aUpper and lower 95% confidence limits, calculated by support-plane method,are indicated within parenthesis.

One of the key design features of the series of complexes 1–7 is the use of only one diimine ligand in order to decrease the molecular symmetry; we reasoned that decreased symmetry would increase the excitation anisotropy of the transition-metal complex luminescence. To determine whether having only one diimine ligand in the cholesterol conjugate 5 has any significant effect on, or advantage for, the photophysical properties of this complex in membrane-like environments, we also synthesized, for comparison, a cholesterol derivative of 8,^17,39 which contains three diimine ligands (see supplemental materials for synthesis). This tris-diimine cholesterol conjugate, 9, had a ¹MLCT absorption band and red-shifted emission maximum similar to those of 5 (see Figure 2). The tris-diimine complex 9 also had a similar luminescence lifetime (~ 0.41 μs at 20 °C when incorporated in egg-PC LUVs). However, the fundamental luminescence anisotropy was much smaller (with excitation at 470 nm, r₀ ~ 0.02 for 9 versus r₀ ~ 0.12 for 5), consistent with the hypothesis that the larger fundamental luminescence anisotropy of the cholesterol conjugate 5 is due to the decreased symmetry of the mono-diimine complex. The anisotropies of parent complexes 1 and 2 have similar anisotropies to 5.

Discussion

The ruthenium probes reported in this paper, synthesized with only one diimine ligand, showed both the long, microsecond excited-state lifetimes and sufficiently high fundamental anisotropies required to study dynamics in the submicrosecond-microsecond time range. It is of interest that the lipid conjugates showed no emission in the case of 3 and blue shifted, short-lived emissions in the case of 4 and 7, in alcohol or chloroform. This lack of emission is likely due to the large number of vibrational modes available which causes an increase in nonradiative decay when the conjugates are in organic solvents. Consistent with this, conjugate 3, which has two lipids, was non-emissive whereas conjugate 4, which has only one lipid and the more rigid phenanthroline ring, showed a weak emission but blue shifted and short lived emisson. Interestingly, 7 where the lipid is conjugated to the phosphine showed a blue shifted, short lifetime, weak emission in solution that is likely the result of perturbation of the orbitals contributing to the MLCT excited state or energy transfer to intraligand transitions.²⁵ In the more constrained environment of the PC LUVs, intense blue-shifted emissions were observed for 3 and 4. Complex 7 showed a similar blue shift but with lower intensity in both the LUVs and in organic solvents. Further, the fact that the MLCT absorption spectra of 3, 4 and 7 are very similar in solution and in the LUVs indicates that the orbital perturbation resulting in the blue shift must occur only in the excited state after the electron is transferred from the metal center to the aromatic ring. This suggests that in the initial excited state, there is perturbation of the orbital energies such that emission takes place from a singlet π* state. Similar effects have been observed in other ruthenium complexes.²⁵ Consistent with this interpretation, they also have very short excited-state lifetimes relative to the parent complexes, as well as much higher fundamental anisotropies (Table 2); the photophysical properties—Stokes shift and lifetime—observed for complexes 3, 4 and 7 in lipid LUVs are characteristic of a singlet emission, although a short-lived triplet cannot be strictly ruled out. It should be pointed out here that the previously reported tris-diimine complex lipid conjugated complex 10 does not show the anomalous blue-shifted, short-lived emissions observed for 3, 4 and 7. This could be due to the fact that in this complex the unsubstituted diimine ring is the electron acceptor from the metal and the lipid conjugated phenanthroline ligand makes no contribution to the excited state whereas in 3, 4 and 7 the phosphine ligand does contribute to the excited state. This is born out by the excitation spectra for 3, 4 and 7, where a significant contribution from their phosphine absorptions at about 325-350 nm is observed (see supplementary materials).

We considered the possibility that the anomalous blue shifts could be due to a fluorescent impurity. However, excitation of the emission spectra at varying wavelengths within the MLCT band results in identical emission line shapes characteristic of that compound, and the intensity varies, as expected for the differences in absorption at the different excitation wavelengths. This confirms that the spectra are not due to an impurity.

Accompanying the short excited-state lifetimes (11 ns and 8 ns) in LUVs, the lipid conjugates have high fundamental anisotropy and temperature sensitive rotational correlation times, which are helpful for studying faster, local motions (up to 33 ns) in the LUVs. Complexes 3 and 4, which have two and one lipid conjugate respectively, have double exponential anisotropy decays when incorporated in LUVs. Interestingly, the longer rotational correlation decay times are similar (7–8 ns at 10 °C, Table 2). Both the timescale and the insensitivity to the number of lipid anchors suggest that this motion reflects restricted diffusion, classically referred to as ‘wobble-in-a-cone’,^36-38 and is not due to axial rotation.⁴⁰

In the wobbling-cone model, it is assumed that the major axis of the probe wobbles randomly within a cone of semi-angle θ_c, which can be estimated using the following relationship:

$${\mathrm{r}}_{\infty }/{\mathrm{r}}_0={[1/2\cos {\theta }_{\mathrm{c}}(1+{\theta }_{\mathrm{c}})]}^2$$ (2)

The temperature-dependent motions of the lipid probes in egg-PC LUVs were analyzed using this model. Over the temperature range from 10–50 °C, the cone angle θ_c varied from 44–72° for complex 3 and from 55–74° for complex 4. In contrast to the longer correlation times, the shorter correlation times are significantly different for complex 3 and 4, (2.0 and 0.7 ns, respectively at 10 °C, Table 2). This timescale and dependence on the numbers of anchoring lipids indicate that these shorter rotational correlation times mainly reflect diffusive dynamics of the probe-labeled head group.⁴⁰ Thus, these probes could be useful for studying lipid head-group motions.

Recently, there has been a report of reversible coordination and lipid incorporation of an Ru(II) diimine aqua complex to a thioether cholesteryl conjugate that was previously incorporated into lipid vesicles.⁴¹ Complex 5, however, to our knowledge, represents the first cholesterol conjugate covalently linked to the diimine ring. The long excited-state lifetime, relative to fluorescence (microseconds versus nanoseconds), and high anisotropy values observed for probe 5 in glycerol and PC-LUVs make this probe an excellent candidate for studying membrane dynamics on the microsecond timescale. The reason that the cholesterol probes do not show the blue shifts observed for the lipid probes is likely related to the fact that the cholesterol molecule is more rigid, and this structural feature leads to less perturbation of the excited-state orbitals. Preliminary data from our laboratory,⁴² show that this probe is useful for studying the global dynamics of lipid nanodiscs, which are 10-nm diameter recombinant lipoprotein A-lipid constructs.⁴³

Conclusions

Three ruthenium-based luminescent bioconjugates with only one diimine ligand have been designed and synthesized as membrane probes. Steady-state and time-dependent photophysical properties of these complexes were studied in solution and in model membrane environments, where the probes would be distributed between the inner and outer leaflets (Figure 5). An important part of the design of the conjugates was use of phosphine ligands, which have previously been shown to improve luminescence quantum yields.^18,19 Lipid conjugates 3, 4 and 7 showed unexpectedly blue-shifted, relatively intense emissions with short, nanosecond-timescale luminescence lifetimes in solution and in the LUV’s. In the LUV’s these emissions were sensitive to changes in membrane viscosity. These complexes would not be useful for studying microsecond-timescale dynamics on membranes, but could be useful for nanosecond-timescale processes. These results sharply contrast with the previously reported tris-diimine lipid conjugates, which exhibited the typical red-shifted, long-lived emissions.^2,17 Complexes 5, 9 and 10 could be used as probes for studying the slower dynamics. Our results point to the sensitivity of the transition metal complex-lipid interaction to the ancillary ligands of the complex. A similar blue shift and short decay time was recently observed for the related complex, [Ru(bpy)₂(dpp)]²⁺ (dpp=2,3-bis(2-pyridyl)pyrazine) upon protonation of the pyrazine nitrogen.⁴⁴ This suggests that these blue shifts are due to perturbations in the orbital energies of the diimine ligand, which is further supported by the absence of the blue shift in 9 and 10. The cholesterol conjugate 5 incorporated in phosphatidyl-choline LUVs had lifetime and anisotropy decays that were sensitive to temperature-dependent motions and conjugation to cholesterol did not significantly perturb the fundamental anisotropy. In addition, the comparison with the tris-cholesterol conjugate 9 revealed that having only one diimine results in a greater fundamental luminescence anisotropy.

Figure 5.

Representation of the MLC-LUV conjugate interactions showing the differences in probe incorporation into the lipid bilayer.

In summary, the unusual behavior of lipid conjugates 3, 4 and 7 relative to 10 points to the importance of the phosphine ligands in controlling photophysical properties via their contribution to the excited state electron distribution when present in combination with multiple vibrational modes of the attached lipids. These contributions are not apparent in the excitation spectra of 8-10 (see Figure 3 and supplementary materials).¹⁷