Photophysical and Biological Properties of Diprotic Ruthenium(II) Complexes with Extended π-Systems

Abstract

Ruthenium(II) polypyridyl complexes can be harnessed for photodynamic therapy. In this work, diprotic ligands which can be deprotonated to change the metal complexes from dicationic to neutral are combined with extended π systems for light absorption at longer wavelengths. Herein, the diprotic ligands are 4,4′-dhbp = 4,4′-dihydroxybipyridine or 4,7-dhphen = 4,7-dihydroxy-1,10-phenanthroline and the π-expanded ligands are bphen = bathophenanthroline, dppz = dipyrido[3,2-a:2′,3′-c]phenazine, and dppn = benzodipyrido[3,2-a:2′,3′-c]phenazine. This report describes the synthesis and characterization of [(bphen)₂Ru(4,4′-dhbp)]Cl₂ (1_A), [(dppz)₂Ru(4,4′-dhbp)]Cl₂ (2_A), [(dppn)₂Ru(4,4′-dhbp)]Cl₂ (3_A), [(bphen)₂Ru(4,7-dhphen)]Cl₂ (4_A), [(dppz)₂Ru(4,7-dhphen)]Cl₂ (5_A), and [(dppn)₂Ru(4,7-dhphen)]Cl₂ (6_A). Single crystal X-ray diffraction data are reported for 1_A, 2_A, and two known precursor compounds: [(ղ⁶-p-cymene)RuCl(diprotic ligand)]Cl (P1, P2 where diprotic ligand = 4,4′-dhbp and 4,7-dhphen, respectively). Compounds 1_A-6_A have been studied for their photocytotoxicity vs. breast and melanoma cancer cells, lipophilic vs. hydrophilic properties, thermodynamic acidities (pK_a values), singlet oxygen quantum yields, and luminescence which is heavily influenced by the nature of the π expanded ligand. Four compounds (bearing bphen and dppn ligands) have promising photocytotoxicity including activity with green light and one compound has a phototherapeutic index (PI) as high as 145.

Graphical Abstract

INTRODUCTION

We have been interested in determining the photophysical and biological properties of ruthenium(II) tris diimine complexes which can be used for photodynamic therapy (PDT) applications. PDT involves using visible light and a photosensitizer to generate cytotoxic singlet oxygen (¹O₂).¹ PDT can be inherently cancer selective by shining the light directly on the tumor for spatio and temporal control. Several Ru(II) complexes have shown promise for PDT applications vs. cancer cells,^{2, 3} with TLD-1433 (Chart 1) having very potent light driven toxicity and advancing to Phase II clinical trials for non-muscle-invasive bladder cancer.¹ Typically, a type II PDT process generates singlet oxygen by energy transfer from ³MLCT, but it can also involve energy transfer from other excited states (³ILCT, etc.).⁴ However, very little work has been done on using protic ligands for PDT.^{5, 6} Our group has been interested in using ligands bearing OH groups (e.g. dihydroxybipyridine) for investigation of how ligand charge and protonation state influences the cellular uptake⁷ and excited state properties.^8–10

Chart 1.

Structure of TLD-1433 which is in clinical trials for bladder cancer. One enantiomer is shown, but it is used as a racemic mixture of both enantiomers.

We have studied complexes of the type: [(N,N)₂Ru(diprotic ligand)]²⁺ in this work (Figure 1). The (N,N) ligand was chosen from bathophenanthroline (bphen), dipyrido[3,2-a:2′,3′-c]phenazine (dppz), and benzodipyrido[3,2-a:2′,3′-c]phenazine (dppn) for their lipophilic and π extended structures. Bphen ligands are lipophilic and can favor Ru(II) complex accumulation in the mitochondria of cancer cells, as demonstrated in past work.^{9, 11} Ru(II) complexes of dppz and dppn have extended π systems and are known to intercalate between DNA base pairs.^12–16 Dppn, having an extended π-system when compared to its dppz parent, introduces a low-lying π-orbital resulting in a long-lived non-emissive ³π-π* excited state that can generate ¹O₂ more efficiently.^{14, 15, 17–21} Ru(II) complexes of dppz, dppn, and related ligands have been explored by Turro, McFarland, and others for promising PDT effects vs. cancer cells in vitro.^{4, 5, 22}

Figure 1.

Ruthenium complexes used in this study are shown in both their acidic (X_A = OH bearing) and basic (X_B = O⁻ bearing) forms.

Very little has been done to understand the role of acid-base chemistry in the behavior of photosensitizers (PSs). The hydroxy groups in the 4,4′-dihydroxybipyridine (4,4′-dhbp) ligand typically have pK_a values of 5–6 in aqueous solution with simultaneous removal of both protons.^{8, 10, 23} For complex 1_A, we previously measured a pK_a value of 5.9(3).⁸ These complexes are typically isolated as the acidic form (X_A = 1_A, 2_A, 3_A) which is dicationic and more hydrophilic.⁷ The pK_a values lead to the neutral and more lipophilic form of the complex (X_B) being the major species in solution at the physiological pH of 7.4. This also influences the photochemistry, with the OH bearing form having longer lived excited states vs. O⁻.⁸ In the current work, we investigated the role of increased π conjugation on the diprotic ligand by using 4,7-dihydroxy-1,10-phenanthroline (dhphen) in 4_A, 5_A, and 6_A. This ligand has been used in the catalysis literature,^{24, 25} but it has not been used for PDT studies to the best of our knowledge.

RESULTS AND DISCUSSION

Synthesis and Characterization.

Complexes 1_A and 4_A were synthesized from the reaction of cis-(bphen)₂RuCl₂ with 4,4′-dhbp and 4,7-dhphen, respectively, using the previously published procedure for 1_A.⁹ A different procedure was effective for the synthesis of 2_A, 3_A, 5_A, and 6_A in which the half sandwich complexes [(ղ⁶-p-cymene)RuCl(diprotic ligand)]Cl (P1, P2 where diprotic ligand = 4,4′-dhbp and 4,7-dhphen, respectively) were treated with dppz or dppn (Figure 2). These half sandwich complexes are known in the literature,^{26, 27} but the crystal structures of these compounds are new. The ¹H-NMR spectra of the five newly synthesized compounds were fully assigned using COSY, HSQC, and HMBC. Single crystal X-ray diffraction (SC XRD), IR, and MS were used to confirm the structures. The purity of the six compounds was analyzed using RP-HPLC. All the spectra and chromatograms obtained for characterization are in the Supporting Information.

Figure 2.

Synthetic procedures used to obtain complexes 1_A-6_A.

Single Crystal X-Ray Diffraction.

Complexes P1 and P2 are known in the literature and were synthesized by a slight modifications of known procedures in 83% and 80% yields, respectively.^{26, 27} Single crystals of both P1 and P2 were grown by vapor diffusion of diethyl ether into methanolic solutions of each compound (Figure 3). The bond lengths and angles shown for P1 and P2 (Table 1) are similar to other Ru(II) half sandwich complexes in the literature^{25, 28} including [(ղ⁶-p-cymene)RuCl(phen′)]Cl where phen′ is phenanthroline bearing an epoxide substituent.²⁷

Figure 3.

Molecular diagrams of P1 and P2 from the crystallographic data (ellipsoids shown at 50% probability). The chloride counter anions and water solvent (for P2) are removed for clarity. Grey = carbon, blue = nitrogen, red = oxygen, green = chlorine, teal = ruthenium.

Table 1.

Selected bond lengths (Å) and angles (°) for P1 and P2.

	P1	P2
Bond lengths
Ru1-arene (centroid)	1.674	1.676
Ru1-Cl1	2.397(1)	2.385(1)
Ru1-N1	2.087(1)	2.093(1)
Ru1-N2	2.096(2)	2.098(1)
O1-C3	1.346(2)	1.336(2)
O2-C15 for P1 O2-C8 for P2	1.337(2)	1.340(2)
Bond angles
N1-Ru1-N2	76.8(1)	77.6(1)
Nl-Ru1-Cl1	84.3(1)	85.8(1)
N2-Ru1-Cl1	82.6(1)	84.2(1)

Red single crystals of 1_A and 4_A suitable for X-ray diffraction were grown by diethyl ether vapor diffusion into saturated methanol solutions of the respective compound. The molecular diagrams are shown in Figure 4 with selected bond lengths and angles in Table 2. Overall, the octahedral Ru(II) centers show minimal distortion from an ideal octahedral geometry. The view with the protic ligand planar (left of Figure 4) shows minimal twisting of the OH substituted rings for 4,4′-dhbp or 4,7-dhphen. In contrast, the published structure of [(bphen)₂Ru(6,6′-dhbp)]Cl₂ shows more twisting of the OH substituted rings.⁹ The Ru-N bond distances involving the protic ligands are slightly shorter for 1_A (2.074(6) Å) vs. in [(bphen)₂Ru(6,6′-dhbp)]Cl₂ (2.100(2) Å) because the 6,6′-dihydroxybipyridine ligand introduces steric strain near the metal center.⁹ For 1_A and 4_A, the average C-O bond distance is ~1.33 Å, indicating an average bond order between single and double due to the resonance effect of the pyridinol ring. These values are very similar to other dhbp complexes.^{9, 23}

Figure 4.

Molecular diagrams of 1_A and 4_A from the crystallographic data (ellipsoids shown at 50% probability). Two views are shown for each molecule. The chloride counter anions, hydrogen atoms, and water solvent (for 4_A) are removed for clarity. Grey = carbon, blue = nitrogen, red = oxygen, green = chlorine, teal = ruthenium.

Table 2.

Selected bond lengths (Å) and angles (°) for 1_A and 4_A.

	1A	4A
Bond lengths
Ru-N (protic ligand)avg	2.074(6)	2.089(2)
Ru-N (bphen)avg	2.068(5)	2.056(2)
C-Oavg	1.338(8)	1.333(3)
Bond angles
Bphen N-Ru-Ncis (within chelate)	79.4(2)	79.39(9)
Protic ligand N-Ru-Ncis (within chelate)	79.3(3)	79.3(1)
N-Ru-Ntrans (avg trans angles)	171.5(2)	172.5(1)

Optical Absorption Spectroscopy.

The electronic spectra of compounds 1_A-6_A in dry acetonitrile display singlet metal-to ligand charge transfer (¹MLCT) bands between 429–483 nm (Table 3 and Figure 5). These bands undergo a red shift upon deprotonation of the protic ligands in acetonitrile, for example, with a shift from 475 nm for 1_A to 525 nm for 1_B^{8, 9} and from 483 nm for 4_A to 515 nm for 4_B (Figure S38). Similar changes are also observed for 4_A (broad shoulder at ~485 nm) red shifting to form 4_B (537 nm) in aqueous solution (90% aqueous, 10% ethanol added for solubility, see Figure S65). The red shift and broadening observed on deprotonation in the absorption spectra have been attributed to the generation of the singlet ligand-to-ligand charge transfer excited state (¹LLCT) by the mixing of the 4,4′-dhbp or 4,7-dhphen orbitals with the bphen orbital.

Table 3.

Optical absorption spectral data of compounds 1_A-6_A, dppz, and dppn as measured in dry acetonitrile. The molar absorptivities have estimated errors of (+/−)1000 M⁻¹cm⁻¹.

Compound	λmax (nm)	Charactera	ε (M−1cm−1)
1 A	~300	π- π*	n.d.b
	445	MLCT	17,000
	475	MLCT	15,700
2 A	359	π- π*	22,000
	370	π- π*	22,000
	429	MLCT	13,000
	480	MLCT	12,000
3 A	389	π- π*	5,000
	410	π- π*	6,000
	461	MLCT	4,000
4 A	~300	π- π*	n.d.b
	440	MLCT	20,000
	483	MLCT	18,000
5 A	359	π- π*	50,000
	370	π- π*	52,000
	425	MLCT	28,000
	483	MLCT	22,000
6 A	389	π- π*	7,000
	410	π- π*	12,000
	468	MLCT	7,000
dppz c	359	π- π*	14,000
	379	π- π*	14,000
dppn c	389	π- π*	10,000
	411	π- π*	13,000

^aSome of these π - π * transitions may also have n - π * character.

^bExtinction coefficient not available due to the peak being off scale for 1_A and 4_A (below 300 nm).

^cSimilar to a literature report.²⁹

Figure 5.

Absorption spectra of compounds 1_A to 6_A in dry acetonitrile.

Complexes 2_A and 5_A with the dppz ligand show two π-π* transitions (at 359 and 370 nm in acetonitrile, Figure 5) in the same positions as those for the dppz free ligand. MLCT bands for 2_A and 5_A are located at ~430 and 480 nm in acetonitrile. A similar pattern is observed for 3_A, 6_A, and the dppn free ligand with π-π* transitions at 389 and 410 nm. Thus, these ligand-based transitions are not perturbed significantly by the presence of the Ru(II) metal center. Extending the π conjugation from dppz to dppn resulted in a 30–40 nm red-shift in the π-π* transitions due to the extra aryl ring. Upon deprotonation of 3_A or 6_A to form 3_B or 6_B in aqueous solution, we observe a red shift of the charge transfer bands at low energy, with no significant changes to the dppn ligand based π-π* transitions (Figures S64 and S66).

Thermodynamic Acidity Measurements.

The deprotonation events described above have been studied in aqueous solution to measure pK_a values (Table 4) for the most potent light activated anticancer compounds: 1, 3, 4, and 6 (see below). Compound 1 was studied in past work by pH dependent UV-Vis, and an apparent pK_a value of 5.9(3) was observed for removal of both protons; therein 1% ethanol in aqueous solution was used to afford sufficient solubility.⁸ In the potentiometric titrations used herein, 10% ethanol was required to dissolve the complexes in the solution phase. Both individual proton removals for 3_A, 4_A, and 6_A are observed with pK_a1 values ranging from 4.2 to 5.3 and pK_a2 values ranging from 6.0 to 6.6 (Figure S61–S63, Table 4). While we have not measured the pK_a values for 2 and 5, we expect them to be in a similar range based upon similar structures to 3 and 6. As such, all compounds are expected to be fully deprotonated (present as the “basic” B forms) in cellular studies at physiological pH of 7.4 These are apparent pK_a values due to the use of ethanol (10%) as a co-solvent and pH is defined as being in 100% aqueous solution. Therefore, these values should not be compared directly with pK_a values for 1 (1% ethanol) or for compounds measured in pure aqueous solution.⁸

Table 4.

Emission data, singlet oxygen quantum yields, and log(D_o/w) values for 1_A-6_A.

Compound	λem (nm)a	Stokes shift (eV)a	Φlum (%)a	τlum (μS)a	ΦΔ (%)a	log(Do/w) (UV-Vis)b	log(Do/w) (RP HPLC)b,c	pKa
1 A	642	0.65	18.7	3.45	46	>3	3.3(3)	5.9(3)d
2 A	671	0.78	4.2	0.11, 0.52	17	>3	n.d.	n.d.
3 A	566	0.86 (PLE) / 0.52 (abs)	0.46	0.04	65	>3	n.d.	4.8(3), 6.4(3)e
4 A	653	0.65	9.4	3.12	17	>3	3.7(2)	4.2(2), 6.0(2)e
5 A	680	0.82	0.44	0.01, 0.07, 0.62	5	>3	n.d.	n.d.
6 A	571	0.88 (PLE) / 0.52 (abs)	4.65	0.04	32	>3	n.d.	5.3(2), 6.6(2)e

^aMeasured in dry acetonitrile under argon. See the experimental section for further details.

^bMeasured with the aqueous phase as pH 7.4 buffer.

^cPoor solubility for 2_A, 3_A, 5_A, and 6_A hindered our studies by reverse phase (RP) HPLC (n.d. = not determined).

^dApparent pK_a for removal of both protons as measured in prior work, with 1% ethanol in aqueous solution.⁸

^eApparent pK_a1 and pK_a2 measured with 10% ethanol in aqueous solution (Figures S61–S63).

Light Stability.

Typically, octahedral Ru(II) complexes lacking steric bulk near the metal center (e.g. 1-6) do not undergo photodissociation upon light irradiation. This was confirmed for 1 in past work,⁹ and for compounds 3, 4, and 6 in this work. White light irradiation (90–120 min) of samples of 3, 4, or 6 dissolved in pH 7.4 aqueous buffer with ethanol (8–16%) added for solubility resulted in no significant changes in the UV-Vis spectra with some slight ground state bleaching (Figure S60). MS analysis confirmed that the samples are intact after irradiation, thus ruling out ligand loss. Based upon a similar structure, we expect the same properties for 2 and 5.

Photoluminescence spectroscopy.

The photoluminescence spectra of compounds 1_A-6_A in dry acetonitrile display a triplet metal-to-ligand charge transfer (³MLCT) band between 566 and 680 nm (see Table 4 and Figures S42–S46). Compounds 1_A and 4_A, display typical Stokes shifts of 0.65 eV (Table 4). Compounds 2_A and 5_A display typical Stokes shifts of 0.78 – 0.82 eV (Table 4), which indicates that emission occurs after relaxation of an excited state. In contrast, 3_A and 6_A display smaller Stokes shifts of around 0.52 eV. This suggests that emission occurs from an excited state that is closer in energy to the initial excitation, which led us to investigate the origin of these ³MLCT excited states, as described below.

Complex 4_B is used as a representative example of a deprotonated compound. Absorption and luminescence studies show broadening and a red shift in the absorption and emission bands of 4_B, indicating the presence of a ¹LLCT and a lower energy ³LLCT (Figures S38 and S44). Introducing a ³LLCT excited state to the native ³MLCT excited state results from the electron-donating O⁻ group in the 4,7-dhphen ligand of 4_B which can quench the emission by a photoinduced electron transfer mechanism (PET) as described previously.^{30, 31} Our previous work has shown that formation of the deprotonated O⁻ bearing complex (e.g. in 1_B) leads to more rapid quenching of the excited state resulting in shorter photoluminescence lifetimes and lower quantum yields.^{8, 9}

Photoluminescence excitation measurement (PLE).

It is noteworthy that 3_A and 6_A were excited at the wavelengths corresponding to their UV-Vis absorption bands for ¹MLCT and π-π* states, and the steady-state luminescence spectrum observed displays an emission spectrum that is similar to that of the free ligand (Figure 6). This begs the question: from what excited state(s) does the emission arise? Is the emission a result of 1) the population of a triplet excited state from a ligand-centered singlet state, i.e., ¹π-π* to the ³MLCT, or 2) an overlap of the ³MLCT and the ³π-π* excited state, or 3) just the relaxation of ³MLCT excited state? To investigate these possibilities, a PLE measurement was carried out on compounds 1_A-6_A to investigate the excited state(s) involved in the emission of the compounds. The photoluminescence excitation spectrum almost always retraces the electronic transitions present in the absorption spectrum of the investigated compound. During PLE experiments, the range of the wavelengths in the absorption spectrum of the investigated compound is scanned while the emission wavelength is held constant.

Figure 6.

Optical absorption, steady-state photoluminescence (PL) spectra, and photoluminescence excitation (PLE) spectra in dry acetonitrile of free ligands dppn, dppz, and compounds 1_A-6_A. For each corresponding metal complex (1_A-6_A): black = absorbance, red = PL, and green = PLE. The blue curves show the absorbance of the free ligands: bphen for 1_A and 4_A, dppz for 2_A and 5_A, and dppn for 3_A and 6_A.

The PLE measurements showed that the dppn compounds (3_A and 6_A) display PLE spectra that retrace the free ligand absorption spectra. In contrast, the dppz and bphen compounds (1_A, 2_A, 4_A, and 5_A) all display PLE spectra that retrace the absorption spectra of the parent Ru(II) compound (Figure 6). Since the dppn compounds have PLE spectra that are replicates of the free ligand electronic spectra (Figure 6), it appears that dppn affords compounds 3_A and 6_A a low-lying ³π-π* deactivation pathway, which is populated from the ³MLCT state. The triplet excited state in the dppz compounds (2_A and 5_A) is believed to arise from overlap of the ³π-π* and the ³MLCT states. In contrast, the ³MLCT state for compounds 1_A and 4_A is populated from the ¹MLCT state. However, the exact energies of these triplet states have not been experimentally observed.

Photoluminescence quantum yield (ϕ_lum).

The photoluminescence quantum yields range from 18.7% to 0.44% for compounds 1_A-6_A. These range from brightly emissive with 1_A (18.7%) > 4_A (9.4%) to moderately emissive 2_A and 6_A (4–5%) to non-emissive 3_A and 5_A (<1%). There is no clear pattern with respect to the protic ligand (4,4′-dhbp vs. 4,7-dhphen), and it seems that the π-expanded ligand plays a greater role. Bathophenanthroline-derived compounds show greater luminescence quantum yields vs. dppz and dppn compounds. No clear trend is apparent for comparing dppz and dppn complexes. The ³π-π* state is a dark and non-emissive relaxation pathway that may lead to a lower photoluminescence quantum yield for Ru(II) complexes of dppz and dppn (2_A, 3_A, 5_A, and 6_A).

Using 4_B as a representative example of the deprotonated species, its photoluminescence quantum yield is low, with a value of less than 0.24% at room temperature. This is due to quenching of the ³MLCT state by the low-emissive ³LLCT state.^{8, 9} The ³LLCT excited state of complex 4_B is populated via direct photoexcitation in the ¹LLCT absorption band or through internal conversion from the photoexcited ³MLCT state.³²

Photoluminescence dynamics studies.

The time-resolved photoluminescence measurements (Table 5) show a relatively long lifetime value of over 3 μs for the bphen compounds, 1_A and 4_A. Much shorter lifetimes are observed for all other compounds. Four compounds show monoexponential decay, but compound 2_A has a biexponential decay, and compound 5_A has a triexponential decay, indicating multiple emission channels. This suggests that an emissive ³MLCT state for 2_A and 5_A transforms into other triplet emissive states. Similarly, the representative deprotonated species 4_B also shows a biexponential decay (see Figure S50).

Table 5.

Cytotoxicity and photocytotoxicity of compounds 1_A-6_A vs. MCF7 and MDA-MB-231 cells under normoxia (~18.5% O₂) via tetrazolium cell viability assay.^a

Compound	MCF7			MDA-MB-231
	EC50_dark (μM)	EC50_vis (μM)	PI	EC50_dark (μM)	EC50_vis (μM)	PI
1 A	3.8 ± 0.9	0.1 ± 0.1	29	12.3 ± 4.7	0.7 ± 0.5	19.4
2 A	>100	>100	N/A	n.d.b	n.d.b	n.d.b
3 A	>100	0.7 ± 0.1	>145	>100	3.8 ± 2.3	>35
4 A	11.0 ± 5.0	3.8 ± 2.2	2.9	36.1 ± 4.9	1.6 ± 0.5	23.8
5 A	3.9 ±0.3	2.5 ±0.8	1.7	n.d.b	n.d.b	n.d.b
6 A	>100	n.d.c	N/A	n.d.b	n.d.b	n.d.b

^aPhotocytotoxicity in MCF7 and MDA-MB-231 breast cancer cells were measured by irradiation for 2 h with white light (irradiance: 40 mW cm⁻², total fluence: 288 J cm⁻²).

^bThese values were “not determined” (n.d.) because the screening data showed no toxicity at 5 μM under dark and light conditions vs. MDA-MB-231.

^cThese values were “not determined” (n.d.) because the cellular viability (normalized absorbance) vs. log[Ru] curve had an unusual shape in all three replicates.

Singlet Oxygen Quantum Yields and Tests for Other Reactive Oxygen Species (ROS).

Singlet oxygen (¹O₂) is a product of the triplet excited states interacting with molecular oxygen, ³O₂. Therefore, the relaxation pathways, whether radiative or non-radiative, influence ¹O₂ generation. Complexes 3_A, 1_A and 6_A are relatively efficient at ¹O₂ generation with Φ_Δ values of 65, 46, and 32%, respectively (Table 4). Lower quantum yields of 17 and 5% are seen for the remaining compounds. Generally, the 4,4′-dhbp compounds have better singlet oxygen quantum yields as compared with 4,7-dhphen compounds if the other ligand (bphen, dppz, or dppn) is held constant. No clear trend emerges from comparing the singlet oxygen quantum yields to luminescence quantum yields or lifetimes.

To test for other reactive oxygen species (ROS), a sample of compound 3 or 6 in aqueous phosphate buffer (pH 7.5) was irradiated for one hour with white light under head space of air (see Experimental Section). These compounds were chosen for study as two of our most toxic compounds upon light irradiation (see below). After one hour, the sample was treated with catalase. Catalase converts peroxide to O₂,³³ and the amount of dissolved O₂ was quantified with a Clark electrode. Comparison with a control sample lacking the Ru complex showed no increased oxygen production (Figure S67). This suggests that peroxide is not produced by these Ru complexes. Furthermore, because superoxide undergoes rapid disproportionation to form peroxide,³⁴ this also suggests that superoxide was not present. This suggests that the major ROS present is ¹O₂. However, we caution that these experiments lack all the components of biological media, and this does not rule out the formation of superoxide or peroxide (perhaps from ¹O₂ reduction) in cell studies.

Lipophilic vs. Hydrophilic Properties of the Compounds.

Lipophilic compounds typically have better cellular uptake by passive diffusion and are often preferred for cellular studies.^35–38 The partitioning of a compound between octanol and water can be used to estimate the ability of a compound to diffuse through the phospholipid bilayer. The distribution coefficient, log(D_o/w), is used for ionizable compounds and is defined here as the log([Ru]_octanol/[Ru]_aq) where aq = aqueous buffer at pH 7.4. These compounds had good solubility in octanol but poor solubility in aqueous buffer, therefore, initially the log(D_o/w) values were all estimated as >3 (undetectable in aqueous solution by UV-Vis method, Table 4). We were able to determine log(D_o/w) values by reverse phase (RP) HPLC also, using a calibration curve in octanol solution. Values for log(D_o/w) of 3.3 and 3.7 were obtained for 1_A and 4_A, respectively, indicating that these compounds are very lipophilic. Similar log(D_o/w) measurements by RP HPLC were not possible for 2_A, 3_A, 5_A, and 6_A due to poor solubility; these compounds exhibited no solubility in aqueous buffer and while they would dissolve in octanol they also precipitated from octanol over time.

Cytotoxicity and Photocytotoxicity Cancer Cell Studies.

Compounds 1_A-6_A were evaluated for cytotoxicity and photocytotoxicitiy vs. MCF7 and MDA-MB-231 breast cancer cells (Table 5). These compounds were isolated as the OH bearing “A forms”, but at pH 7.4 in cell media, they are all expected to be predominantly deprotonated based on the measured pK_a values (above, see Table 4).^{8–10, 23} MCF7 are differentiated mammary epithelium cells, whereas MDA-MB-231 is a triple-negative breast cancer cell line with stem-cell like characteristics. All compounds were tested in MCF7 cells, but for MDA-MB-231, we only measured EC₅₀ values on compounds that showed toxicity at 5 μM under light and dark conditions. The phototherapeutic index (PI) is the ratio of EC₅₀ dark to EC₅₀ light, and this is our key metric for assessing the potential of these compounds for PDT. Only compounds 3_A, 1_A, and 4_A are promising vs. MCF7 with a PI value as high as >145 for 3_A with no observed dark toxicity up to 100 μM and EC₅₀ = 0.7 μM with visible light. Exactly why these three compounds perform best is unclear, but it appears to be correlated to good quantum yields for singlet oxygen (Φ_Δ = 65, 46, and 17% for 3_A, 1_A, and 4_A, respectively) and favorable cellular uptake. Similarly, the PI values are promising for these three compounds vs. MDA-MB-231 with a PI as high as >35 for 3_A.

Similarly, compounds 1_A-6_A were tested in a metastatic melanoma cell line, SKMEL28 (Table 6). These studies were performed in dark conditions and under irradiation with visible, blue, green, or red light. Only three compounds (4_A, 1_A, and 6_A) showed significant PI values greater than 1 with visible light, with the highest PI being 43 for 4_A. When considering longer wavelengths of light which penetrate tissue more deeply, only compounds 1_A and 4_A have significant photocytotoxicity with green light (PI = 14 and 9, respectively). These values may relate to favorable singlet oxygen quantum yields, which were 46, 32, and 17% for 1_A, 6_A, and 4_A, respectively.

Table 6.

Cytotoxicity and photocytotoxicity of the compounds 1_A-6_A against SKMEL28 cells under normoxia (~18.5% O₂) via resazurin cell viability assay.

Compound	EC50 ± SEM (μM)					PIe
	Dark	Visiblea	Blueb	Greenc	Redd	Visiblea	Blueb	Greenc	Redd
1 A	31.3 ± 5.1	0.869 ± n.d.	1.74 ± 0.23	2.25 ± 0.10	14.8 ± 0.7	36	18	14	2
2 A	53.0 ± 2.1	56.9 ± 1.5	56.0 ± 1.0	49.9 ± 1.3	55.3 ± 2.1	1	1	1	1
3 A	43.0 ± 1.8	50.8 ± 1.8	50.0 ± 1.1	43.4 ± 1.8	47.8 ± 1.9	1	1	1	1
4 A	39.8 ± 1.9	0.933 ± 0.025	2.36 ± 0.09	4.52 ± 0.33	39.6 ± 1.6	43	17	9	1
5 A	39.7 ± 1.6	50.7 ± n.d.	50.9 ± 1.7	42.9 ± 1.9	42.2 ± 2.0	1	1	1	1
6 A	50.6 ± 1.9	3.02 ± 0.20	4.07 ± 0.33	51.7 ± 1.7	52.1 ± 2.4	17	12	1	1

Light treatments were approximately 100 J cm⁻² delivered at 19–24 mW cm⁻² with ^a cool white Visible (400–700 nm), ^b Blue (453 nm), ^c Green (523 nm), ^d Red (633 nm), and ^e PI = phototherapeutic index.

^*n.d. = SEM not determined due to steep slope.

Since the studies using breast cancer cells and melanoma cells were done in different laboratories using different lamps, we are cautious about making comparisons. Nonetheless, a few general trends are apparent. The bphen compounds (1_A and 4_A) had good photocytotoxicity across assays in both breast and melanoma cell lines, and this may partially relate to better solubility (cf. the dppz and dppn compounds, 2_A, 3_A, 5_A, and 6_A) and long-lived excited states which may promote energy transfer to ³O₂ and good ¹O₂ quantum yields. No clear advantage is seen for 4,4′-dhbp vs. 4,7-dhphen as both 1_A and 4_A behave similarly and the structures are quite similar. The dppz and dppn Ru(II) compounds tended to aggregate and form a precipitate upon letting a solution sit for minutes to hours, and this may have impacted cellular studies and subcellular localization. This was especially problematic for the dppn compounds (3_A and 6_A). In particular, we were unable to obtain an EC₅₀ under visible light for 6_A vs. MCF7 cells, and we suspect that aggregation may have caused these studies to produce an unusual curve shape for plots of cellular viability vs. log[Ru].

CONCLUSIONS

In conclusion, several compounds have been synthesized, fully characterized, examined for their photophysical properties, and tested in cell culture for their photocytotoxicity. Comparing compounds containing either the 4,4′-dhbp or 4,7-dhphen ligand (e.g. comparing 1_A to 4_A, 2_A to 5_A, or 3_A to 6_A) showed similar photophysical properties, including photoluminescence and photoluminescence excitation spectra. Slightly improved singlet oxygen quantum yields are seen for the 4,4′-dhbp vs. 4,7-dhphen complexes. These similarities make sense as the 4,4′-dhbp and 4,7-dhphen ligands are very similar in structure. However, spectral changes are seen upon modifying the aprotic ligand from bphen to dppz to dppn. An expanded π system in dppn leads to a PLE spectrum that retraces that of the free ligand. Four compounds are promising for photocytotoxicity with visible light and these include 1_A, 3_A, 4_A, and 6_A. In a melanoma cell line, the bphen compounds (1_A and 4_A) were most promising and showed some activation with green light, but in breast cancer cells, compound 3_A, a dppn compound, was most promising. The reasons for this are unclear, but we note that the dppn compounds are prone to aggregation and low solubility, which may influence their subcellular localization.

EXPERIMENTAL SECTION

General.

All of the organic ligands (except dppz and dppn) and metal sources were purchased from commercial sources (VWR, Sigma Aldrich, etc.) and used as received. Dppz,³⁹ and dppn⁴⁰ were synthesized according to published procedures. NMR spectra were recorded in a Bruker AVANCE 360 (360 MHz, ¹H frequency) or AVANCE 500 (500 MHz, ¹H frequency) and the Neo Cryoprobe 500 (500 MHz, ¹H frequency) at room temperature if not mentioned otherwise. UV-Vis spectra were recorded on a Jasco V-780 spectrophotometer using a quartz cuvette of 1 cm path length. FT-IR spectra were recorded using a Bruker Alpha ATR-IR spectrophotometer. Mass spectra were obtained using a Waters AutoSpec-Ultima NT mass spectrometer or a Waters Xero G2-XS QTOF. HPLC studies were performed using an Agilent 1260 Infinity II HPLC system with with a diode array detector, vial sampler, quaternary pump, and column oven using an Agilent analytical column (poroshell 120 EC-C18, 4.6 × 150 mm, 2.7 μm).

All of the photophysics data presented in this study were performed in deaerated and dry acetonitrile. HPLC grade acetonitrile, HPLC grade methanol, and HPLC grade water, reagent grade formic acid (FA), sodium phosphate dibasic heptahydrate, sodium phosphate monobasic monohydrate, and octanol were purchased from VWR. Deuterated DMSO for NMR was purchased from Cambridge Isotope Laboratories. All synthesized Ru(II) complexes were in their racemic forms if chiral.

Synthesis and Characterization data.

Synthesis of [(ղ⁶-p-cymene)RuCl(4,4′-dhbp)]Cl (P1).

A modification of the published procedure was performed.²⁶ [ղ⁶-p-cymene)RuCl₂]₂ (0.282 g, 0.713 mmol, 1.0 equiv.) and 4,4ʹ-dhbp (0.282 g, 1.496 mmol, 2.1 equiv.) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The Schlenk flask was then evacuated and refilled thrice with N₂. MeOH (15 mL) was then added to the Schlenk flask, and the solution was stirred under N₂ at 50 °C for 8 h. The red solution was cooled, filtered, and evaporated to dryness. The solid was rinsed with 10 mL of methanol, 10 mL of dichloromethane, 15 mL of hexanes, and 20 mL of diethyl ether. The reddish-yellow residue was left to air dry. (Yield: 0.285 g, 0.810 mmol, 81%). ¹H-NMR: 12.32 ppm (2H, s), 9.12 ppm (2H, d, 6.50 Hz), 7.81 ppm (2H, d, 2.59 Hz), 7.15 ppm (2H, dd, 6.55 and 2.60 Hz), 6.07 ppm (2H, d, 6.33 Hz), 5.84 ppm (2H, d, 6.32 Hz), 2.57 ppm (1H, m), 2.17 ppm (3H, s), 0.97 ppm (6H, d). Single crystals were grown by vapor diffusion of diethyl ether into methanolic solution of the compound.

Synthesis of [(ղ⁶-p-cymene)RuCl(4,7-dhphen)]Cl (P2).

A modification of the published procedure was performed.²⁷ [(ղ⁶-p-cymene)RuCl₂]₂ (0.302 g, 0.493 mmol, 1.0 equiv.) and 4,7-dhphen (0.220 g, 1.035 mmol, 2.1 equiv.) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The Schlenk flask was then evacuated and refilled thrice with N₂. MeOH (15 mL) was then added to the Schlenk flask, and the solution was stirred under N₂ at 50 °C for 8 h. The red solution was cooled, filtered, and evaporated to dryness. The solid was rinsed with 10 mL of methanol, 10 mL of dichloromethane, 15 mL of hexanes, and 20 mL of diethyl ether. The reddish-yellow residue was then left to air dry. (Yield: 0.217 g, 0.419 mmol, 85%). ¹H NMR spectra for this compound match the literature values:: 13.40 ppm (2H, s), 9.44 ppm (2H, d, 6.18 Hz), 8.14 ppm (2H, s), 7.42 ppm (2H, d, 6.21 Hz), 6.15 ppm (2H, d, 6.17 Hz), 5.94 ppm (2H, d, 6.24 Hz), 2.59 ppm (1H, m), 2.16 ppm (3H, s), 0.92 ppm (6H, d, 6.89 Hz). Single crystals were grown by vapor diffusion of diethyl ether into methanolic solution of the compound.

Synthesis of 1_A ([(bphen)₂Ru(4,4′-dhbp)]Cl₂).

This compound has been previously synthesized, characterized, and reported.⁹ Herein, we used the published procedure and report 100% purity by HPLC (see the SI).

Synthesis of 2_A ([(dppz)₂Ru(4,4′-dhbp)]Cl₂).

[(ղ⁶-p-cymene)RuCl(4,4′-dhbp)]Cl (0.075 g, 0.152 mmol, 1.0 equiv.), and dppz (0.096 g, 0.289 mmol, 1.9 equiv.) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The Schlenk flask was then evacuated and refilled thrice with N₂. DMF (10 mL) dispensed from the solvent purification system (SPS) was then cannula transferred into the Schlenk flask. The reaction mixture was stirred at 140 °C under N₂ for 6 d. The reaction mixture was then cooled to room temperature and stored in the freezer for 10 h. The dark purple solution was filtered and washed with copious amounts of ice-cold water. The dark solid residue was washed with 2 mL of methanol and 10 mL of dichloromethane, followed by 20 mL of diethyl ether and 10 mL of hexanes. The solid was then allowed to dry under vacuum. (Yield: 0.091 g, 0.098 mmol, 53%). ¹H-NMR: 12.23 ppm (2H, s), 9.68 ppm (2H, d, 8.3 Hz), 9.55 ppm (2H, d, 8.2 Hz), 8.54 ppm (4H, m), 8.46 ppm (2H, d, 5.4 Hz), 8.26 ppm (2H,d, 5.4 Hz), 8.21 ppm (4H, t, 9.5 Hz, 4.8 Hz), 8.16 ppm (2H, d, 3.2 Hz), 8.14 ppm (2H,d, 5.6 Hz), 7.87 ppm (2H, q, 5.4 Hz, 2.8 Hz), 7.46 ppm (2H, d, 6.4 Hz), 6.95 ppm (2H, dd, 6.6 Hz, 2.5 Hz). HRMS: M = C₄₆H₂₈N₁₀O₂RuCl₂, m/z for [M-2Cl⁻]²⁺ calcd: 427.0721, found: 427.07, [M-H⁺−2Cl⁻]⁺ calcd: 853.1374, found: 853.1360. IR (cm⁻¹) 3363 (OH str), 3073, 2666, 2558 (CH str), 1617, 1564 (C=C & C=N), 1490, 1442, 1419, 1354, 1301, 1266, 1227, 1115, 1110, 1030 (C-O & =CH def), 975, 870, 845, 815, 766, 722, 575 (oop vib). HPLC was used to determine a purity of 97.6% by peak area.

Synthesis of 3_A ([(dppn)₂Ru(4,4′-dhbp)]Cl₂).

[(ղ⁶-p-cymene)RuCl(4,4′-dhbp)]Cl (0.075 g, 0.152 mmol, 1.0 equiv.), and dppn (0.096 g, 0.289 mmol, 1.9 equiv.) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The Schlenk flask was then evacuated and refilled thrice with N₂. DMF (10 mL) dispensed from the SPS was then cannula transferred into the Schlenk flask. The reaction mixture was stirred at 140 °C under N₂ for 6 d. Then, the reaction mixture was cooled to room temperature and stored in the freezer for 10 h. The dark purple solution was filtered and washed with copious amounts of ice-cold water. The dark solid residue was washed with 2 mL methanol and 10 mL dichloromethane, followed by 20 mL diethyl ether and 10 mL hexanes. The solid was allowed to dry in a vacuum. (Yield: 0.122 g, 0.119 mmol, 71%). ¹H-NMR: 12.25 ppm (2H, s), 9.65 ppm (2H, d, 8.2 Hz, 1.1 Hz), 9.54 ppm (2H, d, 8.3 Hz, 1.1 Hz), 9.26 ppm (4H, d, 15 Hz), 8.48 ppm (2H, overlaid or embedded doublet), 8.44 ppm (4H, m), 8.29 ppm (4H, m), 8.18 ppm (2H, m, 2.6 Hz), 8.14 ppm (2H, q, 8.3 Hz, 4.1 Hz), 7.88 ppm (2H, q, 8.2 Hz, 4.1 Hz), 7.80 ppm (4H, q, 9.6 Hz), 7.52 ppm (2H, d, 6.5 Hz), 6.98 ppm (2H, dd, 6.4 Hz, 2.5 Hz). HRMS: M = C₅₄H₃₂N₁₀O₂RuCl₂, m/z for [M-2Cl⁻]²⁺ calcd: 477.09, found: 477.09. IR (cm⁻¹) 3347 (OH str), 3045 (CH str), 1619 (C=N str), 1452, 1348, 1313 (C=C str & C=N), 1215, 1109, 1070 (C-O str & =CH def), 881 (oop vib). HPLC was used to determine a purity of 97.4% by peak area.

Synthesis of 4_A ([(bphen)₂Ru(4,7-dhphen)]Cl₂).

This compound was synthesized using the literature procedure but substituting 4,7-dhphen for 4,4′-dhbp.⁹ The compound (bphen)₂RuCl₂ (0.359g, 0.429 mmol) and 4,7-dhphen (0.100 g, 0.472 mmol, 1.1 equiv) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The flask was then evacuated and refilled with N₂ thrice, after which a mixture of degassed 17 mL ethanol and 17 mL distilled deionized H₂O was added. The purple solution was refluxed at 120 °C for 72 h under N₂. The red solution was allowed to cool to room temperature and filtered to remove excess 4,7-dhphen. The product was precipitated out of solution using three drops of concentrated HCl. The product was washed with a copious amount of water, 10 mL diethyl ether, and 10 ml hexanes. The product was then air-dried. (Yield: 0.356 g, 0.339 mmol, 71 %). ¹H NMR: 13.07 ppm (2H, s), 8.35 ppm (2H, d, 4 Hz), 8.34 ppm (2H, d 4 Hz), 8.30 ppm (2H, s), 8.25 ppm (4H, s), 7.80 ppm (2H, d, 4 Hz), 7.78 ppm (2H, d, 4 Hz), 7.75 ppm (2H, d, 4.4 Hz), 7.68 ppm (8H, m), 7.64 ppm (4H, m), 7.61 ppm (8H, m), 7.29 ppm (2H, d, 4.5 Hz). HRMS: M = C₆₀H₄₀N₆O₂RuCl₂, m/z for [M-H-2Cl⁻]⁺ calcd: 977.22, found: 977.22. IR (cm⁻¹) 3383 (OH str), 3057 (CH str), 1572 (C=C str & C=N str), 1356, 1221, 1074, 1001, 905 (C-O str & =CH def), 850, 762, 695, 573, 445 (oop vib). HPLC was used to determine a purity of 100% by peak area.

Typical isolation of the deprotonated complex (4_B).

Compound 4_A (50 mg) was treated with 50 mL of basic MeOH at pH = 8.2 (prepared by methanol mixed with aqueous NaOH) and stirred for 3 h. The dark purple solution was filtered and evaporated to dryness. The dark purple solid was then dissolved in 2 mL dry EtOH and 40 mL dry acetonitrile. The purple solution was dried with magnesium sulfate and filtered. The solution was evaporated to dryness. The dark purple solid was then washed with diethyl ether and hexanes. The solid was dried under vacuum. The yield of 4_B was 66 %. The purity and identity of 4_B was confirmed by ¹H-NMR.

Synthesis of 5_A ([(dppz)₂Ru(4,7-dhphen)]Cl₂).

[(ղ⁶-p-cymene)RuCl(4,7-dhphen)]Cl (0.075 g, 0.15 mmol, 1.0 equiv.) and dppz (0.082 g, 0.29 mmol, 2.0 equiv.) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The Schlenk flask was then evacuated and refilled thrice with N₂. Dimethyl formamide (10 mL) dispensed from the SPS was then cannula transferred into the Schlenk flask. The reaction mixture was stirred at 140 °C under N₂ for 6 d. The reaction mixture was then cooled to room temperature and stored in the freezer for 10 h. The dark purple solution was filtered and washed with copious amounts of ice-cold water. The dark solid residue was then washed with 2 mL methanol and 10 mL dichloromethane, followed by 20 mL diethyl ether and 10 mL hexanes. The solid was then dried under vacuum. (Yield: 0.090 g, 0.09 mmol, 59 %). ¹H NMR: 13.09 ppm (2H, s), 9.58 ppm (4H, d, 5.9 Hz), 8.52 ppm (4H, m), 8.39 ppm (2H, d, 3.8 Hz), 8.29 ppm (2H, s), 8.20 ppm (4H,q, 2.5 Hz, 4.8 Hz), 7.93 ppm (4H, m), 7.79 ppm (2H, d, 4.5 Hz), 7.25 ppm (2H,d, 4.5 Hz). HRMS: M = C₄₈H₂₈N₁₀O₂RuCl₂, m/z for [M-2Cl⁻]²⁺ calcd: 439.0721, found: 439.0720, m/z for [M-H⁺−2Cl⁻]⁺ calcd: 877.1375, found: 877.1352. IR (cm⁻¹) 3342 (OH str), 3047–2333 (CH str), 1603, 1572, 1527, 1505, 1411 (C=C & C=N), 1350, 1205, 1115 (C-O str & =CH def), 824, 807, 756, 720, 577 (oop vib). HPLC was used to determine a purity of 97.6% by peak area.

Synthesis of 6_A ([(dppn)₂Ru(4,7-dhphen)]Cl₂).

[(ղ⁶-p-cymene)RuCl(4,7-dhphen)]Cl (0.075 g, 0.145 mmol, 1.0 equiv.) and dppn (0.096 g, 0.275 mmol, 1.9 equiv.) were added to an oven-dried Schlenk flask equipped with a stir bar and a septum. The Schlenk flask was then evacuated and refilled thrice with N₂. DMF (10 mL) dispensed from the SPS was then cannula transferred into the Schlenk flask. The reaction mixture was stirred at 140 °C under N₂ for 6 d. The reaction mixture was then cooled to room temperature and stored in the freezer for 10 h. The dark purple solution was then filtered and washed with copious amounts of ice-cold water. The dark solid residue was then washed with 2 mL methanol and 10 mL dichloromethane followed by 20 mL diethyl ether and 10 mL hexanes. The solid was then allowed to dry in vacuum. (Yield: 0.152 g, 0.144 mmol, 82%). ¹H NMR: 13.05 ppm (2H, s), 9.59 ppm (4H, d, 8.10 Hz), 9.26 ppm (4H, d, 5.33 Hz), 8.47 ppm (4H), 8.42 ppm (2H, d, 5.44 Hz), 8.32 ppm (2H, s), 8.30 ppm (2H,d, 5.47 Hz), 7.95 ppm (4H, m), 7.86 ppm (2H, d, 6.1 127 Hz), 7.80 ppm (4H, q), 7.24 ppm (2H, d, 6.1 Hz). HRMS: M = C₅₆H₃₂N₁₀O₂RuCl₂ m/z for [M-2Cl⁻]²⁺ calcd: 489.0879, found: 489.1134, m/z for [M-H⁺−2Cl⁻]⁺ calcd: 977.1690, found: 977.2155. IR (cm⁻¹) 3371 (OH str), 3067 (CH str), 1550 (C=C), 1417, 1348, 1297, 1264, 1258, 1213, 1107, 1070 (C=C & C=N), 869, 817 (C-O str & -CH def), 724, 575, 490 (oop vib). HPLC was used to determine a purity of 100% by peak area.

Single Crystal X-ray Diffraction.

CCDC Deposition Numbers 2479506–2479509 contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service www.ccdc.cam.ac.uk/structures. Single clear red block-shaped crystals of 1_A and 4_A and yellowish orange irregular shaped crystals of P1 and P2 were obtained by vapor diffusion of diethyl ether into a saturated methanolic solution of respective compounds. Suitable crystals 0.19×0.14×0.09 mm³ (1_A), 0.17×0.15×0.07 mm³ (4_A), 0.17×0.13×0.09 mm³ (P1), and 0.23×0.20×0.09 mm³ (P2) were selected and mounted on a suitable support on an XtaLAB Synergy R, DW system, HyPix diffractometer. The crystals were kept at a steady T = 100.01(16) K during data collection.

The structures were solved with the ShelXT 2018/2⁴¹ (Sheldrick, 2018) (1_A) and ShelXT 2014/5⁴² (4_A, P1, P2) structure solution program using the Intrinsic Phasing solution method and by using Olex2⁴³ as the graphical interface. The models were refined with version 2016/6 of ShelXL 2016/6⁴⁴ using Least Squares minimization.

Data were measured using ω scans of 0.5° per frame for 2.8/11.1 s (1_A) and 5.2/15.0 s (4_A) using Cu K_a radiation and ω scans of 0.5° per frame for 4.3s (P1) and 3.0s (P2) using Mo K_a radiation. The diffraction patterns were indexed and the total number of runs and images were based on the strategy calculation from the program CrysAlisPro⁴⁵ for 1_A, CrysAlisPro⁴⁶ for 4_A, and CrysAlisPro⁴⁷ for P1 and P2. The maximum resolution for achieved for each data collection is Θ = 77.221° (0.79 Å) (1_A), 77.111° (0.79 Å) (4_A), 32.469° (0.66 Å) (P1), 33.398° (0.65 Å) (P2).

The diffraction patterns were indexed, and the total number of runs and images were based on the strategy calculation from the program CrysAlisPro. The unit cell for 1_A was refined using CrysAlisPro⁴¹ on 25109 reflections, 44% of the observed reflections. The Unit cell for 4_A was refined using CrysAlisPro⁴² on 44674 reflections, 37% of the observed reflections. The unit cell for P1 and P2 were refined using CrysAlisPro⁴³ on 25841 (P1) and 25896 (P2) reflections with the used reflections comprising 78% and 74% of the total observed reflections, respectively.

Data reduction, scaling, and absorption corrections were performed using CrysAlisPro. The final completeness is 100.00 % out to 77.221° in Θ for 1_A and 4_A and 99.90 % out to 32.469° and 99.90 % out to 33.398° for P1 and P2. A gaussian absorption correction was performed using CrysAlisPro. Numerical absorption correction based on gaussian integration over a multifaceted crystal absorption correction using spherical harmonics, implemented in SCALE3 ABSPACK scaling algorithm. For 1_A, the absorption coefficient μ of this material is 3.635 mm⁻¹ at this wavelength (λ = 1.542Å) and the minimum and maximum transmissions are 0.449 and 0.728. For 4_A The absorption coefficient μ of this material is 4.111 mm⁻¹ at this wavelength (λ = 1.542Å) and the minimum and maximum transmissions are 0.621 and 0.903. For P1 The absorption coefficient μ of this material is 1.007 mm⁻¹ at this wavelength (λ = 0.771Å) and the minimum and maximum transmissions are 0.892 and 1.000. For P2 The absorption coefficient μ of this material is 0.985 mm⁻¹ at this wavelength (λ = 0.771Å) and the minimum and maximum transmissions are 0.820 and 1.000.

The structure for 1_A was solved and the space group C222 (# 21) determined by the ShelXT 2018/2⁴¹ structure solution program using Intrinsic Phasing and refined by Least Squares using version 2016/6 of ShelXL 2016/6.⁴⁴

The structure for 4_A was solved and the space group I4₁/acd (# 142) determined by the ShelXT 2014/5⁴² structure solution program using Intrinsic Phasing and refined by Least Squares using version 2016/6 of ShelXL 2016/6.⁴⁴ All non-hydrogen atoms were refined anisotropically. Hydrogen atom positions were calculated geometrically and refined using the riding model.

The structure for P1 was solved and the space group P2₁/n (# 14) determined by the ShelXT 2014/5⁴² structure solution program using Intrinsic Phasing and refined by Least Squares using version 2016/6 of ShelXL 2016/6.⁴⁴ All non-hydrogen atoms were refined anisotropically. Hydrogen atom positions were calculated geometrically and refined using the riding model.

The structure for P2 was solved and the space group P2₁/c (# 14) determined by the ShelXT 2014/5⁴² structure solution program using Intrinsic Phasing and refined by Least Squares using version 2016/6 of ShelXL 2016/6.⁴⁴ All non-hydrogen atoms were refined anisotropically. Hydrogen atom positions were calculated geometrically and refined using the riding model.

Purity determination.

The analyte solutions (1_A-6_A) were made in methanol and stirred at room temperature for 2 h. The solution was filtered using a polytetrafluoroethylene (PTFE) filter (13 mm, 0.2 μm). The method employed three mobile phases for the gradient elution. Water plus 0.1% formic acid (FA) (v/v) and methanol were used in solvent lines A and B, respectively. At time t = 0, solvent line A was 100 % at the pump, and this was held for elution at a flow rate of 1.0 mL/min. This method employed needle wash to avoid cross-contamination between sample injections. At the sampler, the method also employs an injection volume of 5.0 μL, a draw speed of 100 μL/min, eject speed of 400 μL/min, a wait time after draw of 1.2 s, and a sample flush-out factor of 5.0 times injection volume. In the column, the analysis was set to run at 25 °C. At the diode array detector (DAD), the wavelengths acquired are 250, 270, 300, 320, and 350 nm. The wavelength with the highest response area is used for the purity determination. The full spectrum of the chromatogram was collected at a bandwidth of 4 nm. This method is tabulated in Table S3.

HPLC determination of lipophilic vs. hydrophilic properties.

The lipophilicity analysis by HPLC (log(D_o/w) values) was performed on an Agilent 1260 HPLC with a diode array detector (DAD), vial sampler, quaternary pump, and column oven. An Agilent analytical column (poroshell 120 EC-C18, 4.6 × 150 mm, 2.7 μm) was used. The dwell volume was measured according to the method in Table S2 and Figure S29.

HPLC reagent methanol, and HPLC grade water were obtained from VWR, reagent grade formic acid (FA), sodium phosphate dibasic heptahydrate, sodium phosphate monobasic monohydrate, octanol. Phosphate buffer (pH 7.4) was prepared and filtered using the Agilent InfinityLab filtration assembly system. The pH was measured using a calibrated Mettler Toledo pH meter. Stock solutions (100 μM) of the compounds 1_A and 4_A were made in pH 7.4 phosphate buffer pre-saturated octanol. The solution was filtered using a polytetrafluoroethylene (PTFE) filter (13 mm, 0.2 μm). The octanol layer (1 mL) was then transferred into a 2 mL HPLC vial to determine the concentration of Ru in octanol. The solution was stirred at room temperature for 18 h. The stock solution was filtered using a polytetrafluoroethylene (PTFE) filter (13 mm, 0.2 μm). Six dilutions from the stock solution were made up with pre-saturated octanol for the lipophilicity experiments. For the log(D_o/w) experiments, 5 mL of concentration ([X_oct]_i), where X is 1_A or 4_A from the diluted series (10 μM to 80 μM), is added to 5 mL pre-saturated pH 7.4 phosphate buffer. The 10 mL solution was stirred at room temperature for 48 h. For efficient separation, the solution was then centrifuged for 2 h. A sample from the octanol layer (1 mL) was then transferred into a 2 mL HPLC vial ([X_oct]_f). The quaternary solvent pump was used for gradient elution with two mobile phases. Water with 0.1% TFA (v/v) an methanol were used in solvent lines A and B, respectively. At time t = 0, solvent line A was 100 % at the pump, and this was held for isocratic elution at a flow rate of 1.5 mL/min. From t = 1 to 15 min, a linear gradient from 90% A to 0% A and 10% B to 100% B was applied at the pump at a flow rate of 1.0 mL/min. This method employed needle wash to avoid cross-contamination between sample injections. At the sampler, the method employs an injection volume of 1.0 μL, a draw speed of 100 μL/min, eject speed of 400 μL/min, a wait time after draw of 1.2 s and a sample flush-out factor of 5.0 times injection volume. At the column the analysis was set to run at 25.0 °C. At the DAD, the absorptions are acquired at 250, 270, 300, 320, 350, and 370 nm wavelengths. The wavelength with the highest response factor (peak area) is used for the calibration curve. The full spectrum of the chromatogram was collected at a bandwidth of 4 nm. A single phase (octanol) was analyzed because no response was recorded on the chromatogram for the aqueous phosphate buffer phase. A calibration curve of concentration versus area of response for each of the six compounds was obtained using the six dilutions, and the concentration of the analyte in the octanol ([X_oct]_f) is calculated from the calibration curve. Due to the insolubility of the compounds 1_A and 4_A in aqueous buffer, the final concentration of the compounds in aqueous buffer [X_Aq]_f was calculated as [X_oct]_i-[X_oct]_f. This method was developed to determine the lipophilicity of compounds 1_A and 4_A and verified using compounds with reported lipophilicity values. The lipophilicity of the compounds is calculated as log(D_o/w) = Log ([X_oct]_f)/([X_Aq]_f).

Absorption and Photoluminescence Studies.

The compounds were prepared in the glovebox using dry acetonitrile dispensed from the solvent purification system. The solution was then filtered using a 5-μm syringe filter to remove undissolved compounds. The absorption spectra of these solutions were recorded on a Jasco V-780 spectrophotometer from 250 nm to 800 nm. The photoluminescence spectra of the compounds were recorded on a Horiba Fluoromax+ fluorimeter The same samples used for absorption studies were utilized for luminescence studies. The samples were excited at the absorption maxima of the complexes, as obtained from electronic spectroscopy. The emission spectra were collected from 500 nm to 900 nm using appropriate band filters to avoid reabsorption. The emission and excitation splits were both 10 nm.

Photoluminescence quantum yield measurements.

The photoluminescence quantum yield (PLQY) of compound 1_A-6_A was measured using the Horiba Fluoromax. The sample solutions were prepared in the glovebox using acetonitrile dispensed from the solvent purification system (SPS). These compounds were dissolved, and the solution was transferred into a fluorescence cuvette equipped with a septum. The cuvette was wrapped with parafilm to prevent air and moisture from entering, after which the absorbance and absorption wavelength were measured. The sample solution was prepared such that the maximum absorbance of the samples is below 0.4 to avoid the inner filter effect. Although this may not pose a problem to our compounds because of the large Stokes shift,^{48, 49} absorbance correction was factored into the calculations to remove errors due to lamp output.

The quantum yield of luminescence measurements was calculated using the relative method, which compares the absorbance and emission of the standard with that of the sample under the same irradiance condition, and was calculated using equation (1), which calculates the quantum yield of the unknown as a product of the luminescence quantum yield of the standard (ɸ_s) and the integrated emission area (I), the corrected absorbance fractions (F) and the solvents refractive index (n) of the unknown (X) and the standard [(bpy)₃Ru]Cl₂.^{48, 49}

$${\Phi }_X={\Phi }_S\times \left(\frac{I_X}{I_S}\right)\times \left(\frac{F_S}{F_X}\right)\times {\left(\frac{n_X}{n_S}\right)}^2$$ (1)

Photoluminescence excitation measurement (PLE).

Samples were prepared in the same manner as mentioned above for photoluminescence experiments. The monitored wavelengths for the PLE experiments are 642 nm (1_A), 671 nm (2_A), 566 nm (3_A), 653 nm (4_A), 680 nm (5_A), and 571 nm (6_A).

Luminescence lifetime measurements.

Luminescence lifetime measurements of compounds 1_A-6_A were carried out on the Horiba Delta Flex fluorescence lifetime system using the time-correlated single-photon counting method (TCSPC) with the 404 nm excitation and 0.1 ns resolution.

Quantum Yield of Singlet Oxygen.

Using [Ru(bpy)₃]²⁺ as the standard (Φ_Δ = 0.56 in aerated MeCN), quantum yields (Φ_Δ) were calculated per the actinometric method described by eq. 2, where I denote the integration of the emission band, A is the solution UV-Vis absorption at the excitation wavelength, and η is the solvent’s refractive index (η²/η_S²=1 here, as MeCN was used in both). The [Ru(bpy)₃]²⁺ standard is denoted by the subscript S.

$${\Phi }_{\Delta }={\Phi }_{\Delta ,S}\left(\frac{I}{I_S}\right)\left(\frac{A_S}{A}\right)\left(\frac{{\eta }^2}{{\eta }_S^2}\right)$$ (2)

Emission spectra were measured on a PTI Quantamaster emission spectrometer equipped with a Hamamatsu R550942 near-infrared photomultiplier tube behind a 1000 nm long-pass filter. The emission and excitation spectra were corrected for lamp output and detector response nonlinearities. The longest wavelength in the excitation spectrum with the most intense emission at 1276 nm was selected as the excitation wavelength. The emission spectra were collected over 1200–1350 nm and integrated with baseline correction. Values were generally reproducible within ± 5%.

Cytotoxicity and photocytotoxicity.

Compounds 1_A-6_A were assessed for their biological activity in the light and in the dark against human skin melanoma cells (SK-MEL-28), hormone receptor-positive breast cancer cells (MCF-7), and triple-negative breast cancer cells (MDA-MB-231). The cells were cultured as 2D monolayer cells.

MCF-7 and MDA-MB-231.

Experiments to determine photocytotoxicity in MCF-7 and MDA-MB-231 were performed in triplicates. Breast epithelial adenocarcinoma cell lines MDA-MB-231 and MCF-7 (ATCC, Manassas, VA) were seeded in 96-well plates at a density of 10,000 cells per well using a 100 μL Dulbecco’s Modified Eagle Media phenol red-free (Gibco, Waltham, MA) supplemented with 10% v/v fetal bovine serum (Gibco) media. The compounds were dissolved in 1% v/v DMSO and diluted with cell media. The final DMSO concentration in the solution was kept at < 1% to prevent the cytotoxicity of DMSO to cells.

For initial screening, the cells were treated with 5 μM of the compounds. For EC₅₀ determination, the cells were treated with varying concentrations of the compounds. The compounds were incubated with the cells for 48 h in the dark. The incubated cells were then gently washed with phosphate-buffered saline (pH 7.2, 3 × 200 μL; Gibco) and left in the dark for additional 2 h (for EC₅₀ dark) or irradiated for 2 h with white light (STASUN 200 W LED flood light, 100–256 V, 20,000 lm, 40,000 lx, irradiance: 40 mW cm⁻², total fluence: 288 J cm⁻², irradiation time of approximately 2 h) (for EC₅₀ light). All cells were incubated in fresh media (200 μL per well) for 20 h in the dark.

Cell viability was measured using Cell Counting Kit-8 at 450 nm according to the manufacturer’s protocol (Enzo Life Sciences, Farmingdale, NY). The EC₅₀ of the compounds was then determined using a nonlinear regression fit of the dose-response curve with the following formula (eq 3) using GraphPad Prism (v10.0.3, La Jolla, CA):

$$A_{\text{obs}}=A_{\text{min}}+\frac{A_{\text{max}}-A_{\text{min}}}{1+{10}^{\left(\text{log}{\text{EC}}_{50}-\text{log}\left[\text{Ru}\right]\right)}}$$ (3)

where A_obs is the observed absorbance, A_max and A_min are the maximum and minimum absorbance, respectively.

Normoxia in SKMEL28.

Compounds 1_A-6_A were screened for their dark and light cytotoxicities toward SK-MEL-28 human melanoma cells under normoxia (~18.5 O₂). The cells (passage 11) growing in the log phase were first seeded in 384-well plates and incubated for 2–3 h before dosing with compound (1 nM to 300 μM). Stock solutions were made in 10% DMSO in water at 5mM, and the dilutions (1 nM – 300 uM) were made in DPBS.

Light treatment was delivered after PS addition with a drug-to-light interval (DLI) of 19 h, using a fluence of 100 J/cm² and irradiance of 19–24 mW/cm² (irradiation time of approximately 1 h) delivered from broadband visible (450 nm maxima, 400–700 nm) light source or LEDs emitting blue (453 nm), green (523 nm) or red (633 nm) light. The light-treated cells were then incubated for 23 h (post-PDT incubation) before measuring cell viability using the resazurin assay. Spectramax m2e fluorescent microplate reader was used for reading the plates after 4 h incubation with resazurin. EC50 values, the effective concentration to reduce relative cell viability by 50%, were obtained from logistic fits of the dose-response curves for the dark and light conditions. Phototherapeutic indices (PIs) were calculated as the ratio of dark to light EC₅₀ values and reflect amplification of cytotoxic effects by light and is the appropriate parameter to use for comparing PS potencies.

Reactive Oxygen Species (ROS) Assay: Peroxide.

A 10 mM stock solution of 3 or 6 was prepared in pH 7.5 phosphate buffer and each compound was tested separately. Samples of each compound and corresponding buffer-only controls were irradiated for 1 h at a fixed distance of 6 inches from the light source. Following irradiation, catalase (5 μM) was added to all samples and controls to convert photogenerated hydrogen peroxide into molecular oxygen. Oxygen evolution was quantified using a Clark-type O₂ electrode, and the corrected O₂ signal (sample minus control) was used to evaluate peroxide formation by each compound under these conditions.

Thermodynamic acidity (pK_a values) measurements.

Ten milligrams of compound 3_A, 4_A, or 6_A was dissolved in 10 % EtOH (with the other solvent being DI water). Five equivalents of HCl was added via 0.1M HCl to obtain a pH of ~2.6 to ensure protonation of the complex at the start and the UV-Vis spectra were taken (Figures S64–S66). A sample of the solution was then titrated against 0.00576 M of NaOH (standardized with KHP) to provide pK_a values (Figures S61–S63). A sample from the titration was used to provide a UV-Vis spectrum at pH of 9.8 for 3_B, 4_B and 6_B (Figures S64–S66).

Supplementary Material

The Supporting Information is available free of charge on the ACS Publications website at DOI: xxxx.

Experimental details including spectra and HPLC analysis (PDF)