Biological Investigations of Ru(II) Complexes With Diverse β-diketone Ligands

Abstract

The β-diketone scaffold is a commonly used synthetic intermediate, and is a functional group found in natural products such as curcuminoids. This core structure can also act as a chelating ligand for a variety of metals. In order to assess the potential of this scaffold for medicinal inorganic chemistry, seven different κ²-O,O’-chelating ligands were used to construct Ru(II) complexes with polypyridyl co-ligands, and their biological activity was evaluated. The complexes demonstrated promising structure-dependent cytotoxicity. Three complexes maintained high activity in a tumor spheroid model, and all complexes demonstrated low in vivo toxicity in a zebrafish model. From this series, the best compound exhibited a ~ 30-fold window between cytotoxicity in a 3-D tumor spheroid model and potential in vivo toxicity. These results suggest that κ²-O,O’-ligands can be incorporated into Ru(II)-polypyridyl complexes to create favorable candidates for future drug development.

Graphical Abstract

Dikontonate compounds can serve as ligands for metal complexes. Seven new complexes were generated to establish structure-activity relationships for Ru(II) polypyridyl complexes containing a single diketonate ligand. While there is limited correlation between physiochemical property and biological activity, the series resulted in the identification of a new lead compound with high potency in 2D and 3D cancer models and low toxicity in zebrafish.

Introduction

Acetylacetone and other β-diketones are important compounds for medicinal chemists. They are key intermediates for the synthesis of several heterocycles (pyrazoles, thiazoles, pyridazines, etc.),^[2] which are pharmacophores in many approved drugs and drug candidates under development.^{[2b–2e, 3]} Various β-diketones have also been investigated for biological and medicinal applications, and demonstrated potential anticancer,^[4] anti-inflammatory,^{[4g, 4h, 5]} antioxidant,^{[4g, 4h]} antibacterial,^{[4g, 6]} antifungal,^{[4g, 6]} and antiviral^{[4h, 6–7]} properties, while others are biologically innocuous and are used as UVA filters in sunscreens.^[8] β-Diketones are excellent κ²-O,O’-chelating ligands, similar to quinolone drugs,^[2b] with biological activities likely derived from the formation of metal complexes generated in situ.^{[6, 7b, 7c]}

Given their potential biological properties, synthetic accessibility, and the ability to chelate metals, many coordination and organometallic complexes with κ²-O,O’-ligands have been studied. Anticancer properties have been assessed using a range of different metals, including Pt(II), Ti(IV), V(IV),^[2b] Co(III),^[9] Ir(III),^{[2b, 10]} Os(II), Rh(III),^[10] Ru(II)^{[1, 10–11]} and La(III).^[12] Metal complexes made from similar κ²-O,O’-chelating ligands, such as 2-hydroxyacetophenones,^[13] flavonols,^{[10, 14]} naphthoquinones,^{[10, 15]} and acylpyrazolones^[7c] have also been investigated.

We have an interest in incorporating O-containing ligands into Ru(II) polypyridyl complexes,^{[11a, 16]} and recently we reported the anticancer properties of Ru(II) complexes of the UVA filter β-diketone, avobenzone (L8; 1-(4-tert-butylphenyl)-3-(4-methoxyphenyl)propane-1,3-dione).^[11a] We found that a range of promising spectroscopic and biological properties could be obtained by combining L8 in complexes by varying the ancillary polypyridyl ligands. For example, cytotoxic activity at low- to submicromolar concentrations were achieved, along with enhancement of potency upon light irradiation, and many of the complexes had no in vivo toxicity in a zebrafish model. However, our most potent compounds ([Ru(tpy)(L8)(pyrazine)]⁺ and [Ru(bpy)₂(L8)]⁺, (where tpy = 2,2′:6′,2′′-terpyridine and bpy = 2,2’-bipyridyl) suffered toxicity in vivo, indicating the need for further structural optimization.

Ruthenium(II) arene complexes containing κ²-O,O’ ligands have been evaluated extensively, and structure-activity relationships (SAR) developed for complexes containing different κ²-O,O’ ligands. ^{[4f, 11c–11e, 17]} In contrast, while a number of polypyridyl Ru(II) complexes containing κ²-O,O’ ligands have also been evaluated for anticancer properties,^{[1, 14a, 14b, 18]} SAR data is sparse. Additionally, many of these studies were performed with specialized κ²-O,O’ ligands (ferrocenyl β-diketonates and semiquinonates),^{[11b, 19]} which makes them hard to compare to simple β-diketonates.

Here we report the effect of various κ²-O,O’ ligands on the photophysical properties, potency in cancer cell lines, and in vivo toxicity for complexes with the structure [Ru(bpy)₂(κ²-O,O’)]⁺. Our studies provide evaluation of the complexes both as conventional chemotherapy and photodynamic therapy (PDT) agents.

Results and Discussion

Design and Synthesis

Seven compounds were synthesized by coordinating Ru(bpy)₂Cl₂ with β-diketone (1–4) or 2-hydroxyphenone ligands (derived from 2-hydroxyacetophenone and 2-hydroxybenzophenone; 5–7; Fig. 1A and 1C). The investigation of the 2-hydroxyphenonate ligands was motivated by the prevalence of the benzophenone scaffold in medicinal chemistry and natural products;^[20] Ru(II) complexes of related molecules may have interesting biological properties. All complexes were characterized by ¹H and ¹³C NMR, ESI mass spectrometry, and UV/Vis; purity was assessed by HPLC, and was ≥ 95% for all complexes. The structures of complexes 1, 2, 3, and 5 were confirmed by X-ray crystallography.

Figure 1.

(A) General synthetic scheme for complexes studied in this report. (B) Compound 8 was previously reported.^[11a] (C) Ligands L1–L7 used for complexes 1–7.

The anticancer potency of arene-Ru(II) complexes with curcuminoid ligands has been shown to correlate with increasing lipophilicity of the β-diketonate ligands.^[4f] Accordingly, we selected κ²-O,O’ ligands that provided a range of octanol/water partition coefficient (log Po/w) values to establish SAR. Ligand selection also took into consideration agents that have been identified as having anticancer properties or other medical or biological applications. For example, 2-benzoylacetophenone (L3),^{[4a, 4b, 4d, 21]} curcumin (L4)^{[1, 4b, 4e]} and oxybenzone (L6)^[22] were reported to exhibit anticancer properties. Ligand L4 was also chosen due to its biological properties (antioxidant, anti-inflammatory, etc.)^{[4e, 4g]} and metal-curcuminate complexes have been studied extensively for anticancer applications.^{[1, 2b, 4e, 4h, 10, 12, 23]} Ligand L6 and benzophenone-4 (L7) are used as UVA filters in commercial sunscreens.^{[8a, 20]} Acetylacetone (L1) and 2-hydroxyacetophenone (L5) were selected as simple β-diketone and 2-hydroxyphenone ligands, respectively. Hexafluoroacetylacetone (L2) was investigated as the CF₃ serves as a bioisosteric replacement for CH₃ groups but changes electronic features, which can modify biological properties of molecules.^[24]

X-ray Crystallography

Complexes 1, 2, 3, and 5 crystallized in monoclinic (P2₁/c (1), P2₁/n (3) and C2/c (5)) or triclinic ($P\overline{1}$, 2) space groups (Fig. 2, Tables 1, S2–S5). All complexes formed octahedral structures similar to other [Ru(bpy)₂(β-diketonate)]^{+[11b, 25]} and 2-hydroxyphenolate complexes.^[26] Despite the varying electronic characters of CH₃, CF₃ and C₆H₅ substituents, the bond lengths and angles in complexes 1, 2, and 3 are very similar. In contrast, [(η⁶-p-cymene)Ru(L2)Cl], has longer Ru–O bonds than the analogous complex with L1.^[11c] The lower trans influence of the κ²-O,O’ ligands compared to the bpy ligands is apparent from the longer Ru–N lengths trans to O (av. 2.046 Å) than those trans to N (av. 2.027 Å), consistent with similar reported complexes.^{[11b, 25]}

Figure 2.

Ellipsoid plot of complex 1 (A), 2 (B), 3 (C), and 5 (D) at 50% probability with hydrogen atoms, counterions, solvent molecules and second cation of 1 omitted for clarity. CCDC Deposition Number 2096762 for compound 1; CCDC Deposition Number 2096763 for 2; CCDC Deposition Number 2096765 for 3 CCDC Deposition Number 2096764 for 5.

Table 1.

Selected bond lengths and angles for complex 1, 2, 3, and 5

Bond Length (Å)	1 a	2	3	5
O1–Ru1	2.072(3)	2.0591(11)	2.0518(12)	2.047(3) b
O2–Ru1	2.064(3)	2.0703(11)	2.0536(12)	2.045(2) c
N1–Ru1d	2.046(3)	2.0537(13)	2.0425(15)	2.048(3)
N2–Ru1e	2.025(3)	2.0240(13)	2.0242(15)	2.030(3)
N3–Ru1 d	2.042(3)	2.0574(13)	2.0454(15)	2.037(3)
N4–Ru1 e	2.026(3)	2.0316(13)	2.0291(15)	2.029(3)
Bond angle (°)
O1–Ru1–O2	92.6(4)	92.24(4)	92.26(5)	90.35(10)
O1–Ru1–N1	85.1(2)	88.20(5)	86.28(5)	87.86(11)
O1–Ru1–N2	88.6(1)	85.37(5)	85.20(5)	87.49(11)
O1–Ru1–N3	95.8(2)	94.48(5)	95.69(6)	94.67(12)
O1–Ru1–N4	175.3(1)°	173.58(5)	174.92(6)	174.01(11)
O2–Ru1–N1	95.8(3)°	94.09(5)	94.07(5)	95.71(11)
O2–Ru1–N2	175.1(2)°	172.92(5)	173.23(5)	174.19(11)
O2–Ru1–N3	84.5(1)°	89.48(5)	89.49(5)	88.11(11)
O2–Ru1–N4	87.0(1)°	87.53(5)	86.86(5)	88.61(11)

^aAveraged over equivalent atoms in both cations.

^bKetone O.

^cPhenolate O.

^dTrans to N.

^eTrans to O.

Photophysical Properties

The use of analogous weak field ligands to induce bathochromic shifts in absorbance was pioneered by the Turro group to produce red-light absorbing polypyridyl Ru(II) complexes, which can act as photocages.^[27] Similar to these systems, complexes 1–7 exhibited shifts of the metal-to-ligand charge-transfer (MLCT) transition (λ_abs = 465–515 nm) compared to the archetypical polypyridyl complex [Ru(bpy)₃]²⁺ (λ_abs = 452 nm; Table 2 Fig. 3A and B).^[28] The absorption properties were tunable, as the electron-withdrawing CF₃ substituents in 2 resulted in a 50 nm hypsochromic shift compared to 1. As previously reported,^[1] 4 has an intense absorption at 395 nm (molar attenuation coefficient (ε) = 37,000 cm⁻¹ M⁻¹; π-π*) in addition to the MLCT peak (λ_abs = 510 nm, ε = 11,000 cm⁻¹ M⁻¹). With the exception of 2, the complexes display absorbance maxima similar to that of model compound 8. The complexes containing 2-hydroxyphenonate ligands (5–7) had very similar absorbance properties and peak shapes, despite their structural differences, and molar attenuation coefficients of 1–7 ranged from 6,300 to 11,000 cm⁻¹ M^-1.

Table 2.

Spectroscopic data for 1–8

	λabs (nm) (ε, cm−1 M−1)a	λem (nm) c	Singlet oxygen generation? e
1	515 (6,300)	719	Yes
2	465 (8,600)	N/A	No
3	495 (11,000)	N/A	No
4	395 (37,000) 510 (11,000)	N/A	No
5	505 (7,300)	N/A	Yes
6	500 (11,000)	N/A	No
7 b	500 (11,000) b	N/A	No
8 d	500 (13,200)	N/A	Yes

^aMeasured in acetonitrile.

^bMeasured in MeCN with 0.8% MeOH.

^cMeasured in water with excitation = 450 nm.

^dData for 1 previously reported ^[11a].

^eAs determined by direct measurement of singlet oxygen phosphorescence (λ_em = 1275 nm) in CD₃OD upon excitation of the complexes with λ_irr = 450 nm. N/A = not applicable.

Figure 3.

Absorbance of complexes 1–7 in MeCN. (A) Complexes with β-diketonates. (B) Complexes with 2-hydroxyphenonates. Calculated log Po/w (clog Po/w ) for ligands (C) and log Po/w for 1–8 (D). The log Po/w value for 4 was previously reported.^[1]

The photoluminescence of the complexes in water under air was also evaluated. Complex 1 was emissive, with λ_em = 719 nm. Despite 3–7 having π-extended κ²-O,O’ ligands, they were non-emissive under these conditions, similar to complex 8.^[11a] The replacement of CH₃ groups with CF₃ groups in 2 eliminated emission.

The potential for the complexes to generate singlet oxygen (¹O₂) and to act as agents for PDT was evaluated by direct detection of ¹O₂ phosphorescence at 1275 nm in CD₃OD (Table 2, Fig. S3). Similar to 8,^[11a] complexes 1 and 5, produced a detectable signal, with 1 producing a ¹O₂ signal ~2-fold more than 5 and 8.^[11a] As expected from their lack of emission, complexes 2, 3, 4, 6, and 7 did not produce detectable ¹O₂.

DNA damage studies were performed to assess the ability of 1 to generate ¹O₂ in aqueous environments (Fig. S2).^[29] Complex 1 was mixed with plasmid DNA and irradiated with light (λ_irr = 470 nm, 37 J/cm²) or kept in the dark as a control. Irradiation did not increase the nicked DNA band compared to the control samples, suggesting that 1 produced minimal or no ¹O₂. The lack of damaged DNA is likely due to increase quenching of the ³MLCT excited state by water; it is known that the use of CD₃OD in ¹O₂ studies increases the lifetime of the ³MLCT state and thus increases the signal for ¹O₂.^[30] This result suggests that the structure of the κ²-O,O’ ligand is vital for complexes to be emissive and generate ¹O₂. Based on this data we tentatively propose that alkyl κ²-O,O’ ligands are superior to conjugated κ²-O,O’ ligands for the creation of water-emissive [Ru(polypyridyl)₂(κ²-O,O’)]⁺ complexes.

Lipophilicity

The lipophilicity of a molecule, commonly quantified as the logarithm of the octanol-water partition coefficient (log Po/w), is an important feature that generally correlates with cellular uptake. Lipophilic molecules have positive log Po/w values, while hydrophilic molecules have negative values.^[31] The lipophilicity of 1–3 and 5–8 was assessed by measuring their log Po/w values,^[32] and the calculated partition coefficient (clog Po/w) for the protonated ligands was estimated. The ligands followed the expected trend of increasing clog Po/w values with increasing numbers of carbon and fluorine atoms. Complexes 1–8 exhibited a range of values (−1.1–1.4, Table 3, Fig. 3C and D) that generally followed the order of the clog Po/w of the ligands. However, log Po/w for 4 and 6 deviated from this trend, with 4 being more hydrophobic and 6 being more hydrophilic than would have been predicted. This gives a trend of lipophilicity that decreases as 8 > 4 > 3 > 6 > 2 > 7 > 5 > 1. Complexes 3, 4, 6, and 8 all fall within the clog Po/w range (−0.4–5.6) of drug-likeness;^[8b] complex 2 is borderline.

Table 3.

Cytotoxicity IC₅₀ values in HL60 cells and log Po/w values for L1–L8, 1–8, cisplatin, and phenanthriplatin

	IC50 (± Std Dev) (μM) of Ligands					IC50 (± Std Dev) (μM) of 1–8
Ligand	Dark	Light	PI	clog Po/w a	Complex	Dark	Light	PI	log Po/w	Relative potency
L1	> 100	> 100	n.d.	−0.88	1	31.9 (2)	2.95 (0.8)	10.8	−1.1 (0.03)	> 3
L2	> 100	> 100	n.d.	1.4	2	0.97 (0.04)	0.65 (0.2)	1.5	−0.44 (0.02)	> 103
L3	15.5 (0.3)	5.3 (0.1)	2.9	2.72	3	0.60 (0.2)	0.53 (0.06)	1.1	0.28 (0.02)	26
L4	4.1 (0.09)	< 0.1	40	2.56	4	0.38 (0.007)	0.34 (0.02)	1.1	0.75 d	11
L5	> 100	> 100	n.d.	0.96	5	16.25 (0.5)	5.47 (0.4)	3.0	−0.69 (0.02)	> 6
L6	~120	14.2	~8.5	2.73	6	0.37 (0.006)	0.21 (0.01)	1.8	−0.36 (0.04)	~333
L7	>100	>100	n.d.	1.03	7	>100	>100	n.d.	−0.61 (0.2) e	n.d.
L8 b	~30	~30	n.d.	4.3	8	0.30 (0.09)	0.17 (0.04)	1.8	1.4 ( 0.1)	~100
-	-	-	-	-	cisplatin c	3.1 (0.1)	3.5 (0.6)	n.d.	n.d.	n.d.
-	-	-	-	-	phenanthriplatin	1.8 (0.4)	n.d.	n.d.	n.d.	n.d

^aCalculated with ChemDraw^® 18.1

^bPreviously reported data.^[11a]

^cPreviously reported data. ^[34]

^dPreviously reported data.^[1]

^eA precipitate was observed in the water phase upon separation of the octanol and water layers. n.d. = not determined. PI = phototoxicity index ((Dark IC₅₀)/(Light IC₅₀)). Relative dark potency (RP) = (IC₅₀ ligand)/(IC₅₀ complex).

Complex stability

The stability of the complexes was assessed in water and phosphate-buffered saline (PBS) by monitoring their absorbance over 72 h at 37 °C (Fig. S4 and S5). All complexes exhibited a slight hyperchromic shift of their MLCT peak, which was previously attributed to slow precipitation over the course of the measurements.^[11a] Complexes 2, 6, and 7 maintained the same band shape throughout the experiment, suggesting that the main species in solution does not undergo chemical change. In contrast, an isosbestic point was observed at 325 nm for complex 5, which was accompanied by the disappearance of peaks at 350 and 490 nm and the appearance of a peak at 395 nm. Absorbance changes for 5 and 6 were suppressed when measured in PBS, while the difference between PBS and water for all other complexes was minimal.

High-performance liquid chromatography (HPLC) was used to elucidate the chemical changes of 1–7 over 72 h (Fig. S7–S11). Complexes 1 and 3 were stable, with no new peaks appearing, consistent with complex 8.^[11a] Minor degradation was observed for 4; 1.8% of the total area came from degradation products after 24 h, increasing to 5.6% at 72 h. The new species formed were more hydrophilic (retention time ~7.4–12.8 min vs 21.6 min for the intact complex), suggesting that a ligand is lost or degraded to produce a less hydrophobic ruthenium species, although no free ligand was observed.

Complex 2 was determined to partially degrade, with 6.0% lost at 24 h and 18.8% at 72 h. The HPLC showed two new peaks at 3.74 and 7.55 min. While complex 2 is stable for prolonged periods (> 1 year) at −20 °C in DMSO, it degraded into multiple peaks over the course of many months at room temperature (Fig. S6). The instability of this compound is likely due to the CF₃ groups making L2 a better leaving group than L1. The substitution of the L2 ligand has also been suggested to eliminate cytotoxicity for [(η⁶-p-cymene)Ru(L2)Cl].^[11c]

The 2-hydroxyacetophenonate (5) and 2-hydroxybenzophenonate (6 and 7) complexes were less stable than 1, 3, and 4, with degradation of 17.2%, 8.1%, and 11.1%^[33] for 5, 6, and 7, respectively, after 72 h. Similar degradation products were detected for 5–7 as those observed for 2, with peaks at ~3.7 and ~7.6 min. Minor peaks with the same retention time as the free ligands were also detected for 6 and 7 (Fig. S10 and S11), indicating that the κ²-O,O’ ligand was lost. Unfortunately, the crystal structure of 5 doesn’t give insight into the poorer stability of these complexes as they have structures very similar to the more stable β-diketonate complexes. This difference in stability between the β-diketonate and 2-hydroxyphenonate complexes demonstrates the need for researchers to carefully evaluate the stability of new [Ru(polypyridyl)₂(κ²-O,O’)]⁺ complexes in aqueous environments.

Biological activity

Cytotoxicity was evaluated by screening complexes and their corresponding free ligands in the acute promyelocytic leukemia(HL60) cell line, with and without light activation (Table 3). The majority of the free ligands had no effect up to 100 μM, but ligands with clog Po/w > 2.5 were found to be cytotoxic in the dark. Ligands L3, and L4 were more potent than avobenzone (L8 ~30 μM) with IC₅₀ = 15.5 and 4.1 μM. Ligands L3, L4, and L6 also had enhanced potency (~14–0.05 μM) when activated with light (>450 nm) to give phototoxicity index (PI = (dark IC₅₀)/(light IC₅₀)) values of 3, ~ 40, and 8.5, respectively. This is surprising, as L3 and L6 do not absorb light at 450 nm.^{[8a, 35]} However, as these types of ligands can coordinate metals in situ^{[6, 7b, 7c]} we hypothesized that this phototoxicity was due to coordination with cellular metal ions, such as Cu(II), leading to better absorbance of visible light. This was supported by an in vitro binding study that shows L3 capable of binding Cu(II) at 37 °C in a water/DMSO mixture (Fig. S19). Additionally, photolysis of [Cu(L3)₂] has been suggested to produce radicals,^[36] which may explain the phototoxicity of L3. A similar metal coordination mechanism for the phototoxicity of free ligands has also be been suggested for polypyridyl ligands.^[37] In contrast to L3 and L6, L4 absorbed light at 450 nm and has been demonstrated to both degrade and generate ROS under irradiation, which is likely the cause of its phototoxicity.^[4e]

With the exception of inactive 7, coordination of the ligands to [Ru(bpy)₂]²⁺ led to a dramatic increase in potency. The IC₅₀ values ranged from 0.37–32 μM in the dark (Table 3). The relative potency of the complexes and ligands (RP = (IC₅₀ ligand)/(IC₅₀ complex); Table 3) ranged from 3–333. A rough trend became apparent when pIC₅₀ was plotted against log Po/w (Fig. 4B), with a correlation between potency and hydrophobicity, except for 6 and 7. Most compounds (2–4 and 6) exhibited submicromolar IC₅₀ values (Fig. 4A), but no complex was as potent as 8. The inactivity of 7 (IC₅₀ >100 μM) is attributed to the inclusion of the sulfonic acid, similar to the behavior of [Ru(bpy)₂(8,8’-hydroxyquinolinato-5-sulfonate)]⁰, which was also inactive in multiple cell lines.^[16b] Light activation improved only compounds 1 and 5, with PI = 10.8 and 3, respectively, consistent with the ¹O₂ detection data in CD₃OD. However, these relatively low PI values and their cytotoxicity in the dark suggest that complexes 1 and 5 are not good candidates for PDT. Notably, the most active complexes 2–4 and 6 were more potent than the clinically relevant control, cisplatin (IC₅₀ = 3.1 μM), and a later-generation platinum agent, phenanthriplatin (IC₅₀ = 1.8 μM).

Figure 4.

(A) Bar chart of pIC₅₀ values for 1–8, cisplatin (Cis), and phenanthriplatin (Phen) in HL₆₀ cells (red, front row), DU145 cells (green, middle row), and MIA PaCa-2cells (orange, back row). (B) A plot of log P_O/W vs pIC₅₀ in HL₆₀ cells.

The SAR analysis revealed the following trends: 1) the simple ligands, L1 and L5, possessed moderate cytotoxicity when coordinated to [Ru(bpy)₂]²⁺; 2) the incorporation of fluorine (2) or phenyl groups (3) in β-diketonates improved the potency by 33–53-fold compared to 1; 3) coordination of L6 to form 6 resulted in a 44-fold more cytotoxic agent compared to L5; 4) the incorporation of the sulfonate group in L7 led to a total loss of activity; 5) the cytotoxicity of these Ru(II) complexes correlated to some degree with log P_ow.

Given the promising results in HL60 cells, the versatility of these complexes was evaluated by screening them in prostate (DU145) and pancreatic (MIA PaCa-2) cancer cell lines (Table 4, Fig. 4A). The complexes generally exhibited similar potencies, irrespective of cell line. The range of IC₅₀ values for 1 (IC₅₀ ~25–32 μM) and 4 (IC₅₀ = 0.77–0.29 μM) were ~4–10-fold less (1) or more potent (4) than previously reported IC₅₀ values for these two complexes,^[1] which may be due to different cell lines being used.

Table 4.

Cytotoxicity IC₅₀ values in DU145 and MIA PaCa-2 cells and in vivo toxicity for complexes 1–8, cisplatin, and phenanthriplatin ^a

		IC50 (± Std Dev) (μM) b			% Normal Zebrafish
	DU145	MIA PaCa-2	DU145 Spheroid	MCR Index d	Dosage, 10 μM	Dosage, 25 μMf
1	~25	~37.6	n.d.	n.d.	100 e	n.d.
2	0.37 (0.02)	0.89 (0.03)	3.5 (2)	9.5	100 f	100
3	0.29 (0.03)	0.35 (0.03)	0.97 (0.3)	3.3	100 f	0
4	0.29 (0.05)	0.77 (0.05)	n.d.	n.d.	100 f	0
5	~16.2	~21	n.d.	n.d.	100 f	100
6	0.37 (0.02)	0.49 (0.05)	0.82 (0.3)	2.2	100 f	100
7	>100	>100	n.d.	n.d.	100 f	100
8 c	0.093 (0.002)	0.068 (0.003)	n.d.	n.d.	0 e	n.d.
cisplatin	0.62 (0.06)	5.91 (0.7)	n.d.	n.d.	100 f	100
phenanthriplatin	0.14 (0.01)	0.15 (0.01)	0.78 (0.2)	5.6	100 f	100

^aAll data measured under dark conditions.

^bAverage of three trials.

^cPreviously reported data.^[11a]

^dThe multicellular resistance (MCR) index is the ratio of IC₅₀ values of the spheroid and 2-D cell assays.

^eComplex incubated with larvae for 48 h

^fComplex incubated with larvae for 96 h. n.d. = not determined

While 2–4 and 6 were not as potent as phenanthriplatin in DU145 or Mia PaCa-2 cells, the novel complexes were more potent in HL60 cells and maintained their submicromolar potency across the cell lines better than either cisplatin or phenanthriplatin.

The correlation to log Po/w was generally maintained, with a few exceptions (Fig. S17), consistent with the complete lack of cytotoxicity of [Ru(polypyridyl)₂(κ²-O,O’)]⁺ complexes with highly hydrophilic κ²-O,O’ ligands containing pendant sugars.^{[18c, 18g]} Compared to [(η⁶-p-cymene)Ru(κ²-O,O’)Cl] complexes with similar κ²-O,O’ co-ligands, compounds 2, 3, 4, and 6 are more potent, with submicromolar IC₅₀ values. Complexes 1 and 5 (in the dark) are within the moderate IC₅₀ value range that many arene complexes exhibit.^{[11c–11e, 17]} The complexes are also more potent than most^[8b] of the more water-soluble [(η⁶-p-cymene)Ru(κ²-O,O’)(PTA)]⁺ (PTA = 1,3,5-triaza-7-phosphaadamantane) complexes, with the exception of the analogous curcuminate complex which has potency within the same nanomolar range.^[38] The cytotoxicities of 2–4 and 6 are some of the most potent reported when compared to a number of similar [Ru(polypyridyl)₂(κ²-O,O’)]⁺ complexes.^{[13, 14b, 18f]} Generally, complexes that have been characterized with more potent cytotoxicity than 2–4 and 6,^{[11a, 11b, 18d, 19]} contain much more lipophilic ancillary ligands^{[18d, 19]} or novel κ²-O,O’ ligands,^[11b] which makes them hard to directly compare to the complexes reported here.

Potency in 3-D Tumor Spheroid Model

Complexes 2, 3, 6, and phenanthriplatin were selected for further evaluation in DU145 3-D tumor spheroids (Table 4). These 3-D models are considered superior to 2-D monolayers, as they exhibit some of the complexities found with in vivo tumors, such as hypoxic/necrotic regions and characteristic multicellular resistance (MCR) to cytotoxins, which can reduce the efficacy of chemotherapeutics to ranges similar to in vivo potencies.^[39] Importantly, all the complexes maintained low micromolar or submicromolar IC₅₀ values. The MCR index (MCR index = 2-D IC₅₀/3-D IC₅₀) values for 2, 3, and 6 were 9.5, 3.3, and 2.2, compared to MCR of 6 for phenanthriplatin. Complexes 2, 3, and 6 also maintain better MCR values than related [Ru(dip)₂(κ²-O,O’)]⁺ (dip = 4,7-diphenyl-1,10-phenanthroline and κ²-O,O’ = semiquinonate or maltol) complexes, with MCR values of 28 and 38. However, a direct comparison between these compounds is limited by the use of different cell lines.^[18d]

In vivo toxicity

The encouraging results from in vitro cytotoxicity experiments motivated us to assesses the toxicity of the complexes in vivo (Table 4). Zebrafish larvae are a well-established model to evaluate toxicity of novel compounds, and have been used successfully with various ruthenium complexes.^[40] The higher biological complexity of this model, relative to cellular-based assays, and its high sensitivity to developmental perturbations has made this model an important safety screen for early drug discovery.^{[40b, 40c]}

Complexes 1–6 were incubated with zebrafish larvae at concentrations above or near the IC₅₀ value of each complex. Complex 7 was evaluated at the same concentrations as 1–6 (10 and 25 μM) as a non-cytotoxic control compound. Gratifyingly, complexes 1–7 were non-toxic at 10 μM concentrations. Complexes 2–7 were further tested at 25 μM, and most showed no mortality at this concentration. Complexes 3 and 4 exhibited 100% toxicity at 25 μM (n = 3), but this concentration is 86-fold greater than the IC₅₀ of these complexes, which leaves a large window for evaluation of efficacy. Overall, these complexes exhibit a significant improvement compared to 8, which had a 100% mortality rate at 10 μM.

The most promising compounds were 3 and 6, which exhibited sub-micromolar potency in the spheroid model and the lowest MCR values. Assuming a correlation between IC₅₀ in the 3-D model and in vivo, there is a ~30-fold window between the concentration needed for tumor cell cytotoxicity and potential in vivo toxicity for compound 6. This provides a rational expectation that the compounds have a favorable therapeutic window suitable for future in vivo studies.

Conclusion

Seven complexes were synthesized by coordinating κ²-O,O’ ligands to [Ru(bpy)₂]²⁺, and compared to complex 8 with the AVB ligand. An evaluation of the stability of 1–7 revealed that, despite the similarities between the β-diketonate and 2-hydroxyphenonate ligands, they possessed different aqueous stability. Complexes coordinated with β-diketonate ligands were the most stable, though moderate levels of degradation for the fluorinated complex 2, and 2-hydroxyphenonate complexes (5–7) highlight the need for more robust stability studies to better anticipate speciation in biological studies. The impact of this degradation is currently unclear, as the majority of these compounds have submicromolar potencies. However, as complex stability is not correlated to activity, it is presumed that ligand exchange is not a key component of the mechanism of action, in contrast to the platinum agents used as references.

The in vivo toxicity in zebrafish embryos appears to correlate with the lipophilicity of the complexes. Complexes that were toxic at either 10 or 25 μM (3, 4, and 8) all had positive log Po/w values. When this in vivo data is considered in the context of the in vitro cytotoxicity data, it appears likely that there is an optimum range of log Po/w values to simultaneously give submicromolar potencies while remaining non-toxic in vivo. The most promising compounds were 2–4 and 6, but each faces challenges. Complexes 3 and 4 possessed promising cytotoxicity and thermal stability, but were toxic in vivo at 25 μM. In comparison, complexes 2 and 6 also have submicromolar cytotoxicities, are non-toxic in vivo at 25 μM, but degrade slowly in solution. Further optimization of these, and similar, structures to mitigate these problems may be required. This seems to be an achievable goal, given the promising results with a variety of κ²-O,O’-chelating ligands, and the synthetic diversity available with this motif.

It is challenging to identify a feature that may explain the biological activity of the ligands and their metal complexes. While we believe it may be correlated to the ability of both to generate radicals or ROS in cells, experimental support for this is inconsistent. There is a >0.4 V difference in the Ru(II/III) oxidation potentials of 1 and 2 in acetonitrile, and significant differences in their rates of electron transfer.^[41] However, the Ru(II/III) oxidation for 3 is similar to 1 (0.68 vs. 0.64 V),^[42] and there are large differences in their biological effects. It may not be effective to correlate redox potentials in organic solvents and biological results that rely on cellular uptake and localization properties. The measurement of free radical damage in cells would clarify if SAR can be established based on the redox behavior of the compounds under aqueous conditions.

Experimental Section

Materials and instrumentation

Commercial reagents were purchased from VWR, Fisher Scientific, or other commercial sources, and used without further purification. Water used for synthesis, purification, and biological studies was purified through a Milli-Q® purification system. Complex 8 was synthesized as previously reported.^[11a] Ruthenium complexes were purified by column chromatography using SiliaFlash® Irregular Silica Gel, F60, 40–63 μm (230–400 mesh). Thin-layer chromatography was performed with aluminum-backed EMD silica gel 60 F254 and visualized under 254 and 365 nm light. ¹H NMR and ¹³C NMR spectra were recorded on a Varian Mercury spectrometer (400, 100 MHz), a Bruker Avance NEO spectrometer (400 MHz, 100 MHz) or a JEOL ECZr (equipped with a Royal Probe, 500 MHz, 125 MHz). NMR chemical shifts are reported relative to the residual deuterated solvent peak of CD₃CN (δ 1.94 (¹H) and 1.32 (¹³C), acetone-d₆ (δ 2.05 (¹H) and 29.84 (¹³C)), or DMSO-d₆ (δ 2.50 (¹H) and 39.52 (¹³C)). Electrospray ionization (ESI) mass spectra were obtained on a Varian 1200L mass spectrometer at the Environmental Research Training Laboratory (ERTL) at the University of Kentucky or at the University of Kentucky Mass Spectrometry Facility. Compound purity was determined with an Agilent 1100 Series HPLC using a previously reported method.^[11a] Absorption spectra were obtained on a Cary 60 UV/Vis spectrophotometer using 3 mL cuvettes. Aqueous stability was monitored by UV/vis absorption using a BMG Labtech FLUOstar Omega microplate reader with Greiner UV-STAR® 96-well plates. Singlet oxygen was detected with a Horiba DSS-IGA020L NIR indium gallium arsenide solid-state detector controlled with a Horiba Fluoromax Plus-C fluorometer. The Prism software package was used to analyze data and plot graphs. Photoluminescence of complexes was measured using a Horiba Fluorolog-3 spectrofluorometer. DNA damage experiments were performed using a 470 nm LED array from Elixa (10.2 J/cm²). Agarose gels were digitally imaged using a BioRad ChemiDoc System. Cytotoxicity assays used an Indigo LOCTITE LED Flood System (>450 nm, 29.1 J/cm²). Synthesis was carried out under purified nitrogen where indicated. Prior to biological testing and all experiments performed in aqueous media, metal complexes 1–8 were converted to salts containing the chloride counterion by loading a solution of ~20 mg of the solid in 1–2 mL methanol onto an Amberlite IRA-410 chloride ion exchange column, eluting with methanol, and evaporating to dryness in vacuo. DMSO stock solutions of the metal complexes (10 mM) were made from the chloride salts for biological and spectroscopic measurements. Ligands L1–L8 were dissolved in DMSO to make 10 mM stock solutions for cytotoxicity studies.

Synthesis and Characterization

Synthesis of [Ru(bpy)₂(acetylacetonate)]PF₆ (1).

[Ru(bpy)₂Cl₂]•2H₂O (110 mg, 0.21 mmol), acetylacetone (58.8 mg, 0.59 mmol) and NEt₃ ( 72.6 mg, 0.72 mmol) were dissolved/suspended in 9 mL EtOH:H₂O (1:2) in a pressure tube. The mixture was stirred at 90 °C for 3.5 h. The resulting purple solution was cooled to room temperature and EtOH was removed under vacuum. To the aqueous solution was added 6 mL of saturated aqueous KPF₆. The resulting suspension was allowed to settle at 4 °C for ~18 h. The obtained solid was vacuum filtered and washed with 10 mL H₂O, 10 mL of EtOH:H₂O (1:1), and copious ethyl ether. The resulting dark purple solid was dried under vacuum. Yield: 130 mg (94%). Purity by HPLC = 99.5% Spectroscopic data agreed with the literature for this compound.^[43] X-ray quality crystals were grown from vapor diffusion of hexanes into a solution of 1 in CH₂Cl₂.

Synthesis of [Ru(bpy)₂(hexafluoroacetylacetonate)]PF₆ (2).

[Ru(bpy)₂Cl₂]•2H₂O (110 mg, 0.21 mmol), hexafluoroacetylacetone (54.4 mg, 0.33 mmol) and NEt₃ (72.6 mg, 0.72 mmol) were dissolved/suspended in 9 mL EtOH:H₂O (1:2) in a pressure tube. The mixture was stirred at 90 °C overnight, turning dark red. The reaction mixture was cooled to room temperature and diluted with 60 mL of H₂O, giving a solid. The solid was redissolved by adding minimal EtOH (~40 mL). To the solution was added 6 mL of saturated aqueous KPF₆. The solution was further diluted with 30 mL of H₂O and allowed to settle at 4 °C for ~18 h. The resulting solid was vacuum filtered, washed with water (50 mL × 2), 15% EtOH in H₂O (10 mL), and copious ethyl ether. The solid was eluted from the filter paper with CH₂Cl₂ and filtered through Celite. The solvent was removed under vacuum to give a dark red solid. Yield: 152 mg (95%). ¹H NMR (400 MHz; acetone-d₆): δ 8.86–8.84 (m, 2H), 8.79 (ddd, J = 5.5, 1.5, 0.8 Hz, 2H), 8.69 (dt, J = 8.0, 1.0 Hz, 2H), 8.39 (td, J = 7.9, 1.5 Hz, 2H), 8.08 (ddd, J = 5.7, 1.4, 0.8 Hz, 2H), 8.02 (ddd, J = 8.1, 7.6, 1.4 Hz, 2H), 7.94 (ddd, J = 7.6, 5.6, 1.3 Hz, 2H), 7.36 (ddd, J = 7.5, 5.8, 1.5 Hz, 2H), 6.25 (s, 1H). ¹³C NMR (100 MHz, acetone-d₆): δ 172.13 (q, ²J = 34.1 Hz, CF₃CO)., 159.76, 158.59, 155.12, 150.77, 139.40, 137.70, 128.27, 126.92, 124.67, 124.48, 117.97 (q, ¹J = 282.9 Hz, CF₃), 92.96. ¹⁹F NMR (376 MHz, CD₃CN): δ −72.86 (d, ¹J_P-F = 706 Hz), −75.64. ESI MS calcd for C₂₅H₁₇F₆N₄O₂Ru [M]⁺ 621.03; found 621.0 [M]⁺. Purity by HPLC = 99.7%. X-ray quality crystals were grown from vapor diffusion of diethyl ether into a solution of 2 in methanol.

Synthesis of [Ru(bpy)₂(2-benzoylacetophenonate)]PF₆ (3).

[Ru(bpy)₂Cl₂]•2H₂O (110 mg, 0.21 mmol), 2-benzoylacetophenone (57.4 mg, 0.26 mmol) and NEt₃ (72.6 mg, 0.72 mmol) were dissolved/suspended in 9 mL EtOH:H₂O (1:2) in a pressure tube. The mixture was stirred at 90 °C for 3 h, cooled to room temperature, and the EtOH removed under vacuum. The mixture was diluted with 6 mL H₂O and 6 mL of saturated aqueous KPF₆. The suspension was allowed to settle at 4 °C overnight. The resulting solid was vacuum filtered, and washed with water, 10% EtOH in water (10 mL x 3) and copious ethyl ether. The solid was eluted from the filter paper with acetonitrile and the solvent was reduced under vacuum. The obtained solid was dissolved in 14 mL of methanol and ca. 35 mL of 2-propanol was added. The mixture was allowed to settle at 4 °C for ~18 h. The resulting dark solid was vacuum filtered, washed with 2-propanol (10 mL), hexanes (10 mL) and copious ethyl ether. The solid was dissolved in CH₂Cl₂, filtered through Celite and the solvent was reduced under pressure to give a dark red solid. Yield: 139 mg (84%). ¹H NMR (400 MHz; DMSO-d₆): δ 8.80 (d, J = 8.2 Hz, 2H), 8.72 (d, J = 8.1 Hz, 2H), 8.68 (d, J = 4.9 Hz, 2H), 8.17 (td, J = 7.9, 1.3 Hz, 2H), 7.95 (td, J = 7.8, 1.1 Hz, 2H), 7.89 (d, J = 5.6 Hz, 2H), 7.75–7.68 (m, 6H), 7.45 (t, J = 7.3 Hz, 2H), 7.36–7.31 (m, 6H), 6.78 (s, 1H). ¹³C NMR (100 MHz, DMSO-d6): δ 179.86, 158.71, 157.35, 153.10, 149.52, 138.99, 136.88, 135.23, 130.63, 128.49, 126.63, 126.27, 125.74, 123.48, 123.43, 93.77. ESI MS calcd for C₃₅H₂₇N₄O₂Ru [M]⁺ 637.12; found 637.1 [M]⁺. Purity by HPLC = 97.5%. X-ray quality crystals were grown from vapor diffusion of diethyl ether into a solution of 3 in CH₂Cl₂.

Synthesis of [Ru(bpy)₂(curcuminate)]PF₆ (4).

[Ru(bpy)₂Cl₂]•2H₂O (150 mg, 0.29 mmol), curcumin (108 mg, 0.29 mmol) and NEt₃ (44 mg, 0.44 mmol) were added to 10 mL of degassed EtOH:H₂O (1:1) in a pressure tube. The mixture was stirred at 100 °C for 1 h, cooled to room temperature, and transferred into 50 mL of H₂O. Following this, 1–2 mL of saturated aqueous KPF₆ was added to obtain a red precipitate. The solvent was removed by filtration and the solid was washed with water and ethyl ether. The purification of the solid was carried out by flash chromatography (silica loaded in MeCN). A gradient was run, and the pure complex eluted at 0.2% KNO₃, 5% H₂O in MeCN. The product fractions were concentrated under reduced pressure, and a saturated aqueous KPF₆ was added, followed by extraction of the complex into CH₂Cl₂. The solvent was removed under reduced pressure to give the product as a solid. Yield: 140 mg (52%). Purity by HPLC = 95.5%. Spectroscopic data agreed with the literature for this compound.^[25a]

Synthesis of [Ru(bpy)₂(2-hydroxyacetophenonate)]PF₆ (5).

[Ru(bpy)₂Cl₂]•2H₂O (111 mg, 0.21 mmol), 2-hydroxyacetophenone (45.3 mg, 0.33 mmol) and NEt₃ (72.6 mg, 0.72 mmol) were dissolved/suspended in 7 mL EtOH:H₂O (1:2) in a pressure tube. The mixture was stirred at 90 °C overnight, turning dark red. The EtOH was removed under vacuum and ~ 50 mL H₂O and 3 mL of saturated aqueous KPF₆ was added. The resulting suspension was allowed to settle at 4 °C for ~18 h. The obtained solid was vacuum filtered, washed with EtOH:H₂O (1:1, 5 mL × 2), and copious ethyl ether. Yield: 130 mg (89%). ¹H NMR (400 MHz; CD₃CN): δ 8.78 (d, J = 5.6 Hz, 1H), 8.65 (d, J = 5.6 Hz, 1H), 8.48 (t, J = 8.7 Hz, 2H), 8.36 (t, J = 9.4 Hz, 2H), 8.07 (dtd, J = 22.2, 7.9, 1.4 Hz, 2H), 7.84–7.74 (m, 3H), 7.70 (d, J = 5.7 Hz, 1H), 7.65–7.53 (m, 3H), 7.16–7.09 (m, 3H), 6.49 (dd, J = 8.8, 0.9 Hz, 1H), 6.38 (ddd, J = 8.3, 6.8, 1.3 Hz, 1H), 2.36 (s, 3H). ¹³C NMR (100 MHz, CD3CN): δ 198.56, 171.59, 160.42, 159.89, 158.92, 158.59, 154.64, 154.18, 151.27, 151.12, 137.75, 137.23, 136.06, 135.82, 135.71, 134.33, 127.46, 127.08, 126.62, 126.43, 126.19, 124.23, 124.15, 124.10, 124.06, 121.66, 114.83, 28.66. ESI MS calcd for C₂₈H₂₃N₄O₂Ru [M]⁺ 549.09; found 549.2 [M]⁺. Purity by HPLC = 95.5%. X-ray quality crystals were grown from vapor diffusion of diethyl ether into a solution of 5 in CH₂Cl₂.

Synthesis of [Ru(bpy)₂(oxybenzonate)]PF₆ (6).

[Ru(bpy)₂Cl₂]•2H₂O (150 mg, 0.29 mmol), oxybenzone (66 mg, 0.29 mmol) and NEt₃ (58 mg, 0.58 mmol) were added to 10 mL of degassed EtOH:H₂O (1:1) in a pressure tube. The mixture was stirred at 100°C for 1 h, cooled to room temperature, and transferred into 50 mL of H₂O. Following this, 1–2 mL of saturated aqueous KPF₆ was added to obtain a red precipitate. The solvent was removed by filtration and the solid was washed with water, MeOH:H₂O (1:2) and ethyl ether. Yield: 180 mg (79%). ¹H NMR (500 MHz, CD₃CN): δ 8.89 (d, J = 5.4 Hz, 1H), 8.73 (d, J = 5.5 Hz, 1H), 8.47 (dd, J = 17.0, 8.1 Hz, 2H), 8.37 (d, J = 8.4 Hz, 1H), 8.32 (d, J = 8.1 Hz, 1H), 8.07 (dt, J = 11.2, 8.2 Hz, 2H), 7.82–7.73 (m, 4H), 7.64 (s, 2H), 7.43 (t, J = 7.4 Hz, 1H), 7.33 (t, J = 7.7 Hz, 2H), 7.18 (d, J = 7.3 Hz, 2H), 7.15–7.06 (m, 3H), 6.02 (d, J = 2.5 Hz, 1H), 5.94 (dd, J = 9.4, 2.5 Hz, 1H), 3.66 (s, 3H). ¹³C NMR (125 MHz, CD3CN): δ 194.10, 175.77, 166.62, 160.36, 159.98, 158.86, 158.70, 154.69, 154.36, 151.45, 151.18, 141.55, 138.35, 137.74, 137.36, 135.97, 135.83, 130.90, 129.03, 128.72, 127.40, 127.28, 126.37, 126.29, 124.23, 124.20, 124.14, 124.07, 118.33, 116.33, 107.41, 105.61, 56.00. ESI MS calcd for C₃₄H₂₇N₄O₃Ru [M]⁺ 641.11; found 641.11 [M]⁺. Purity by HPLC = 98.2%.

Synthesis of [Ru(bpy)₂(benzophenonate-4)]⁰ (7).

[Ru(bpy)₂Cl₂]•2H₂O (150 mg, 0.29 mmol), benzophenone-4 (89 mg, 0.29 mmol) and NEt₃ (58 mg, 0.58 mmol) were added to 10 mL of degassed EtOH:H₂O (1:1) in a pressure tube. The mixture was stirred at 100°C for 1 h, cooled to the room temperature, and transferred into 15 mL of H₂O. The complex was extracted into CH₂Cl_2, and solvent was removed under reduced pressure to give a purple solid. The solid was dissolved in 20 mL of acetone and precipitated with 50 mL of ethyl ether. The solvent was removed by filtration and the solid was washed with ethyl ether. Yield: 153 mg (73%). ¹H NMR (400 MHz; DMSO-d₆): δ 8.79 (dd, J = 14.3, 8.4 Hz, 3H), 8.70–8.68 (m, 2H), 8.63 (d, J = 7.9 Hz, 1H), 8.17 (q, J = 8.3 Hz, 2H), 7.90–7.73 (m, 6H), 7.62 (s, 1H), 7.43 (t, J = 7.4 Hz, 1H), 7.34 (t, J = 7.6 Hz, 2H), 7.24 (dt, J = 13.6, 6.7 Hz, 2H), 7.09 (d, J = 7.5 Hz, 2H), 6.01 (s, 1H), 3.58 (s, 3H). No ¹³C NMR was obtained due to low solubility of this complex in all common NMR solvents. ESI MS calcd for C₃₄H₂₇N₄O₆RuS [M]⁺ 721.07; found 721.07 [M]⁺. Purity by HPLC = 95%

X-ray Crystallography

X-ray diffraction data were collected at 90.0(2) K on a Bruker D8 Venture kappa-axis diffractometer using MoKa radiation. Raw data were integrated, scaled, merged, and corrected for Lorentz-polarization effects using the APEX3 package.^[44] Corrections for absorption were applied using SADABS. The structure was solved by iterative dual-space methods (SHELXT)^[45] and refinement was carried out against F² by weighted full-matrix least-squares (SHELXL).^[46] Hydrogen atoms were found in difference maps, but subsequently placed at calculated positions and refined using riding models. Non-hydrogen atoms were refined with anisotropic displacement parameters. Atomic scattering factors were taken from the International Tables for Crystallography.^[47] For structures 2 and 5, a region of ambiguous solvent electron density was accounted for using the SQUEEZE procedure.^[48] Crystal data and relevant details of the structure determinations are summarized in Table S2–S5 and selected geometrical parameters are given in Table 1. Compound 1 CCDC Deposition Number 2096762; Compound 2 CCDC Deposition Number 2096763; Compound 3 CCDC Deposition Number 2096765; Compound 5 CCDC Deposition Number 2096764. A nickel complex analogous to 1 has been previously reported.^[49]

Aqueous stability

Measured by UV/Vis: The aqueous stability of complexes 1–7 was studied at 37 °C as 40 μM solutions in DI water and 1X PBS buffer. Each solution was measured in triplicate in a 96-well plate and monitored by UV/vis absorbance over the course of 72 h. Solvent evaporation was slowed during incubation by covering the plate with a Breath-Easy® membrane, which was removed before UV/Vis absorbance measurements.

Measured by HPLC: Compounds 1–7 were diluted in water to 250 μM. Their HPLC chromatograms were recorded and the samples were incubated at 37 °C for 72 h. Chromatograms were obtained for each compound before incubation, at 24 h, and at 72 h.

Singlet Oxygen Detection

Singlet oxygen generation by the Ru(II) complexes with an excitation source of 450 nm was measured by monitoring the phosphorescence of ¹O₂ at 1275 nm in CD₃OD. Solutions were tested with absorbance of ~0.2 at 450 nm. Excitation and emission slits spectral widths were set to 29 nm. Integration was set to 5 s and emission was collected from 1220–1350 nm.

DNA Gel Electrophoresis

Compound 1 was serially diluted 1:2 to give final concentrations of 0, 7.8, 15.6, 31.3, 62.5, 125, 250, and 500 μM of with 40 μg/mL of pUC19 plasmid in 10 mM phosphate buffer pH 7.4 in a 96-well plate. The dark control samples were removed prior to exposure of the plasmid solution to light. The samples were then irradiated with 470 nm light for 1 h (37 J/cm²). Irradiated and control samples were incubated overnight at 37 °C in closed microcentrifuge tubes. Following this, 6X DNA loading dye was added to each sample and the plasmid samples were resolved on a 1% agarose gel in 1X Tris-Acetate (TA) buffer, with 0.3 μg of plasmid loaded per lane. The samples were run for 75 min at 100 mV followed by staining the gel with a solution of ethidium bromide in 1X TA buffer for 40 min. The gels were then destained in 1X TA buffer for 30 min and digitally imaged.

Octanol-water Partition Coefficient Determination

The lipophilicity of 1–3 and 5–8 was measured using the shake-flask method, following a previously reported procedure.^[32] Briefly, complexes were dissolved in 0.5 mL of n-octanol presaturated with water to give 100 μM solutions, and 0.5 mL water presaturated with n-octanol was added. The mixture was vigorously shaken 200× by hand. The layers were allowed to stand and separate over 24 h. The layers were separated, and each phase was analyzed for the presence of compound by UV/vis spectroscopy. Each experiment was carried out in triplicate.

Copper Binding of L3 and L6

The ligands were diluted in DMSO to give a final volume and concentration of 100 μL and 400 μM. Then 100 μL of 200 μM CuCl2 were added to give a final ratio of 2:1 (ligand:Cu²⁺). The absorbance of each solution was monitored by UV/vis over 24 h and compared to controls that consisted of the ligand with no CuCl₂ in DMSO:H₂O (1:1).

Cell Culture

All cell lines used in this study were obtained from the American Type Culture Collection (ATCC). HL60 human leukemic cells were maintained in Iscove’s media supplemented with 10% FBS, 100 U penicillin, and 100 mg/mL streptomycin. DU145 and MIA PaCa-2 cell lines were cultured in DMEM with 10% FBS, 100 U penicillin, and 100 mg/mL streptomycin. Cells were cultured and maintained at 37 °C with 5% CO₂.

Cytotoxicity Assay

For cytotoxicity assays with HL60 cells, 30,000 cells/well were plated in extracellular solution (10 mM HEPES pH 7.4, 10 mM glucose, 1.2 mM CaCl₂, 1.2 mM MgCl₂, 3.3 mM KH₂PO₄, 0.83 mM K₂HPO₄, and 145 mM NaCl). Compounds were serially diluted in extracellular solution, added to the cells, and incubated for 1 h at 37 °C with 5% CO₂. The cells were then irradiated with 29.1 J/cm² light (>450 nm using the Indigo LED) for 1 min or kept in the dark. Immediately after irradiation an equal volume of Opti-MEM supplemented with 4% FBS was added to each well and the cells were incubated with the compounds for 72 h. A 440 μM solution of resazurin was prepared in PBS and added to the cells to a concentration of 70 μM. The plates were incubated for an additional 3 h and the change in emission determined using a SpectraFluor Plus plate reader with 535 nm excitation and 595 nm emission filters. For DU145 and MIA PaCa-2 cytotoxicity assays, the cells were seeded into 96-well plates at 2,000 cell/well in DMEM media with 10% FBS,100 U penicillin, and 100 mg/mL streptomycin and incubated overnight. The media were then removed and replaced with extracellular solution, followed by the addition of compound. After a 1 h incubation the cells were irradiated, followed by the addition of opti-MEM as described above. After 72 h resazurin was added to the cells, incubated for 1–2 h and emission was quantified as described above.

Spheroid cytotoxicity measurements

The DU145 cell line was seeded into Nunclon Sphera 96-well U-bottom plates (Thermo Scientific) with 5,000 cells/well in a 50 mL volume of opti-MEM supplemented with 2% FBS, 100 U penicillin, and 100 mg/mL streptomycin (opti-MEM). After 48 h an additional 50 mL volume of media was added. The cells were incubated for an additional 4 d, whereupon the cells formed spheroids with an average diameter of 450 mm. Visual inspection was used to ensure that the spheroids were uniform in size and shape across all wells. Compounds were prepared in opti-MEM, added to the spheroids, followed by a further incubation of 72 h. Viability was measured with the Cell Titer Glo 3D Viability Assay (Promega), where an equal volume of Cell Titer Glo was added to each well, followed by titrating the solution 3–4 times to disrupt the spheroids. The solution was then transferred to a 96-well plate and luminescence measured with the SpectraFluor Plus plate reader (Tecan).

Toxicity assays in zebrafish.

Animal studies were approved under the University of Kentucky’s Institutional Animal Care and Use Committee, protocol 2019–3399. Healthy 2 day post fertilization (dpf) Casper strain zebrafish larvae were pipetted into 96-well plates, at 1 larva per well in 150 μL 1X E3 media (5 mM NaCl, 0.17 mM KCl, 0.33 mM MgSO₄ in dH₂O). Compounds were prepared at twice the desired concentration in E3 media and 150 μL added to each well. Plates were incubated in the dark for 96 h, with drug refreshed during media change at 48 h. Animals were imaged using a Vertebrate Automated Imaging System (Union Biometrica) as previously described.^[50] Care was taken to keep all compounds in the dark throughout their use, and each compound was tested in triplicate at two concentrations.