Structurally Characterized BODIPY-Appended Oxidovanadium(IV) β-Diketonates for Mitochondria-Targeted Photocytotoxicity

Abstract

Mixed-ligand oxidovanadium(IV) β-diketonates having NNN-donor dipicolylamine-conjugated to boron-dipyrromethene (BODIPY in L₁) and diiodo-BODIPY (in L₂) moieties, namely, [VO(L₁)(acac)]Cl (1), [VO(L₂)(acac)]Cl (2), and [VO(L₁)(dbm)]Cl (3), where acac and dbm are monoanionic O,O-donor acetylacetone and 1,3-diphenyl-1,3-propanedione, were prepared, characterized, and tested for their photoinduced anticancer activity in visible light. Complexes 1 and 2 were structurally characterized as their PF₆^– salts (1a and 2a) by X-ray crystallography. They showed V^IVN₃O₃ six-coordinate geometry with dipicolylamine base as the facial ligand. The non-iodinated BODIPY complexes displayed absorption maxima at ∼501 nm, while it is ∼535 nm for the di-iodinated 2 in 10% DMSO–PBS buffer medium (pH = 7.2). Complexes 1 and 3 being green emissive (λ_em, ∼512 nm; λ_ex, 470 nm; Φ_F, ∼0.10) in 10% aqueous DMSO were used for cellular imaging studies. Complex 3 localized primarily in the mitochondria of the cervical HeLa cells with a co-localization coefficient value of 0.7. The non-emissive diiodo-BODIPY complex 2 showed generation of singlet oxygen (Φ_Δ ≈ 0.47) on light activation. Annexin-V assay showed singlet oxygen-mediated cellular apoptosis, making this complex a targeted PDT agent.

Introduction

Photodynamic therapy (PDT) and photoactivated chemotherapy (PACT) together have emerged as two new non-invasive treatment modalities of cancer to reduce the toxic side effects and to avoid drug resistance resulting from conventional chemotherapeutic drugs like cisplatin and its analogues.¹⁻¹⁰ In PDT, photosensitizers with minimal dark cellular toxicity, after getting accumulated in the cancerous tissue, get activated on light-irradiation to generate reactive oxygen species (ROS) that damage the cancer cells while leaving the light unexposed normal tissue intact. Although beneficial compared to the chemotherapeutic drugs, prolonged use of the FDA (Food and Drug Administration)-approved PDT drug photofrin is known to cause severe skin sensitivity and acute hepatotoxicity due to generation of bilirubin on oxidation of the porphyrin core.^11,12 In addition, the macrocyclic porphyrin- and phthalocyanine-based photosensitizers in general have poor aqueous solubility, thus reducing their bioavailability. To address the solubility and tumor selectivity issues, transition metal-based PDT agents with desirable photochemical and photophysical properties are reported as potential alternatives to photofrin and its analogues.¹³⁻²⁰ Structurally simple β-diketonates derived from different derivatives of acetylacetone (Hacac) are suitable toward designing a variety of metal-based anticancer agents.²¹⁻²⁵ The first non-platinum metal-based anticancer compound to enter clinical trials was budotitane, specifically, [cis-diethoxy(1-phenylbutane-1,3-dionato)titanium(IV)], in which the β-diketonate ligand was used to enhance the lipophilicity of the complex.^26,27 Appropriate hydrophilic–lipophilic balance (HLB) is necessary for an anticancer agent to be accumulated inside the tumor cells in preference to the normal cells.²⁸

The present work stems from our interests to design oxidovanadium(IV) β-diketonates as targeted PDT agents with ancillary dipicolylamine-based photosensitizers in a mixed-ligand structure. The BODIPY (boron-dipyrromethene)-based ligands are suitable as photosensitizers-cum-cellular imaging agents due to their tunable photophysical properties that allow the emissive BODIPY moiety for cellular imaging to probe the accumulation of the complex in any particular organelle, while its non-emissive derivative can be used to produce singlet oxygen as the reactive oxygen species in a high yield for PDT activity.^29,30 In addition, BODIPY dyes are well documented for their mitochondrial localization ability.³¹⁻³³ Even in hypoxic condition of tumor cells, BODIPY units with their very high molar extinction coefficient values are efficient photosensitizers inducing apoptotic cell death.^34,35 The BODIPY dyes are also well known to possess low dark toxicity as they are redox stable within the biological potential window.^29,35 Besides the high molar extinction coefficient values, these dyes have sharp emission bands with minimal Stokes shift and excellent photostability. They can be synthetically fine-tuned to be either an imaging agent or as a PDT agent. However, to enhance their aqueous solubility and tumor selectivity, transition metal complexes with desirable photochemical and photophysical properties are reported as the potential PDT agents. With these characteristics, the highly emissive BODIPY complexes are extensively used for cellular uptake and imaging studies, while other complexes having BODIPY units with heavy atom(s) as photosensitizers are used for PDT activity in visible light (400–700 nm).

Herein, we report the synthesis, structural characterization, and PDT activity in cancer cells of a series of mixed-ligand oxidovanadium(IV) complexes of β-diketones having a tridentate dipicolylamine (dpa) ligand with pendant non-iodinated (L₁) and di-iodinated (L₂) boron-dipyrromethene (BODIPY) moieties, specifically, [VO(L₁)(acac)]Cl (1), [VO(L₂)(acac)]Cl (2), and [VO(L₁)(dbm)]Cl (3), where acac and dbm are monoanionic forms of acetylacetone and 1,3-diphenyl-1,3-propanedione (dibenzoylmethane), respectively (Figure 1). The choice of 3d¹-oxidovanadium(IV) moiety as a linker of the photosensitizer and the diketonate group is based on its redox inactivity within the biological potential window when compared to its redox active 3d⁹-Cu²⁺ analogue, which is known to display undesirable dark cellular toxicity.³⁶ The salient features of this work are as follows: (i) structural characterization of two mixed-ligand BODIPY complexes by single-crystal X-ray crystallography, (ii) significant mitochondrial localization of the complexes, and (iii) remarkable PDT activity on photoinduced generation of singlet oxygen (¹O₂) in visible light of 400–700 nm, causing apoptotic cell death.

Figure 1.

Mixed-ligand oxidovanadium(IV) complexes 1–3 and the ligands used.

Results and Discussion

Synthesis and Characterization

The oxidovanadium(IV) complexes 1–3 were synthesized in good yield (∼75%) from a general method in which vanadyl sulfate was reacted initially with barium chloride in aqueous ethanol (1:5 v/v), and the filtrate on removal of insoluble barium sulfate was subsequently treated with an ethanol solution of the deprotonated β-diketone ligand followed by treatment of a solution of the tridentate dpa derivative (L₁ or L₂) in dichloromethane to obtain a dark precipitate of the product (Figure 1). The complexes were characterized from spectral and analytical data. Selected physicochemical data are given in Table 1. The mass spectra of the complexes showed respective single ion peak assignable to [M – Cl]⁺ in MeOH (Figures S1–S3). The complexes are 1:1 electrolytic giving molar conductivity values of ∼68 S cm² mol^–1 in DMF at 25 °C.³⁷ They are one-electron paramagnetic with effective magnetic moment values of ∼1.65 μ_B at 298 K, indicating the 3d¹ electronic configuration of the metal. Characteristic C=O stretching (from β-diketonate) and V=O stretching were observed in the FT-IR spectra of the complexes at ∼1592 and ∼972 cm^–1, respectively (Figures S4–S6). Aromatic C=N stretching was at ∼1610 cm^–1. The B–F stretching was also identified for the complexes at ∼1523 cm^–1 assignable to the BODIPY core. Cyclic voltammetry of the complexes showed irreversible reductive response within −1.0 to −1.3 V versus SCE (saturated calomel electrode) in DMF with 0.1 M [Buⁿ₄N](ClO₄) (TBAP) as the supporting electrolyte assignable to the V(IV)–V(III) reduction couple (Figures S7–S9). Additionally, a BODIPY-based reduction peak was observed near −1.06 V for 1 and 3, while the diiodo-BODIPY based peak appeared near −0.80 V in complex 2. The dbm ligand in 3 showed a reduction peak at −1.63 V. The complexes did not show any oxidative responses, indicating the redox stability within the biological potential window. This is expected to reduce any dark toxicity arising from the chemical nuclease activity of the complexes in the presence of glutathione (GSH) inside the cells.

Table 1. Selected Physicochemical Data, Partition Coefficient, and DNA Binding Parameters of the Complexes 1–3.

complex	λmaxa (nm) (ε/M–1 cm–1)	λemb(λexc) (nm) [ΦF]c	IRd (cm–1) V=O, C=O	Epce (V)	ΛMf (S cm2 mol–1)	μeffg (μB)	log Ph	Kbi (M–1)
1	501 (37500), 733 (50)	512(470) [0.10]	972, 1592	–1.32, −1.06	72	1.61	0.80 ± 0.02	(1.5 ± 0.2) × 105
2	536 (31910), 724 (66)	557 (510) [0.01]j	975, 1591	–1.28, −0.80	67	1.65	1.01 ± 0.02	(7.8 ± 0.6) × 104
3	500 (40560), 732 (56)	514 (470) [0.09]	971, 1593	–1.07, −0.98	70	1.63	1.56 ± 0.06	(1.8 ± 0.3) × 105

^aIn 10% DMSO–PBS buffer (pH = 7.2).

^bIn 10% aqueous DMSO.

^cFor emission quantum yield, fluorescein was used as a standard (Φ_F = 0.79 in sodium hydroxide solution of 0.1 M).

^dIn the solid phase.

^eIn DMF–0.1 M TBAP, E_pc = cathodic peak potential. All data were measured against saturated calomel electrode (SCE), and ferrocene was taken as a standard. Scan rate = 100 mV s^–1.

^fIn DMF at 25 °C.

^gPurified samples of the complexes (solid) at 298 K.

^hLog P values of Hacac and Hdbm are reported as 0.09 ± 0.35 and 3.08 ± 0.33, respectively (vide ref.²⁵).

ⁱIntrinsic equilibrium ct-DNA binding constant.

^jComplex 2 gave a singlet oxygen quantum yield (Φ_Δ) of 0.47.

The oxidovanadium(IV) complexes displayed weak and broad d-d transition band near 735 nm in 10% DMSO–DPBS medium (pH = 7.2) (Figure S10).³⁸ The BODIPY-based intense π → π* transition band was observed for complexes 1 and 3 at ∼501 nm, and a similar transition for complex 2 from the diiodo-BODIPY core appeared at a longer wavelength of ∼535 nm (Figure 2a). The bands within 350–400 nm were originated from both the β-diketonate ligand and from the LLCT (ligand–ligand charge transfer) transitions.³⁹ The non-iodo-BODIPY complexes 1 and 3 were green emissive in 10% aqueous DMSO, giving emission maxima at ∼514 nm with a moderate fluorescence quantum yield (Φ_F) value of ∼0.1 on excitation at 470 nm (Figure 2b).⁴⁰ Complex 2 having the diiodo-BODIPY unit was essentially non-emissive (λ_F ≈ 557 nm) with a low Φ_F value of ∼0.01 (λ_ex = 510 nm) due to facilitated effective intersystem crossing (ISC) in the presence of heavy iodine atoms that has resulted in a shift of the weak emission band to the longer wavelength.⁴¹ The emissive complexes 1 and 3 were suitable for cellular uptake and co-localization studies using their emission spectral property in flow cytometry and confocal microscopy, respectively. Complex 2 in the presence of the diiodo-BODIPY unit as a photosensitizer was used for the PDT study in visible light (400–700 nm). The complexes having the oxophilic oxidovanadium(IV) core are known to bind strongly to the O,O-donor β-diketonate ligands.^29a Mixed-ligand complexes 1 and 3 were found to be fairly stable in 10% DMSO–DPBS medium (pH = 7.2) up to 48 h. The diiodo-BODIPY complex 2 was observed to be relatively less stable than its non-iodo BODIPY analogues. This instability is ascribed to the attachment of two iodine atoms in ligand L₂ which was found to be less photostable than the non-iodo ligand L₁ (Figures S11–S15). Overall, the complexes exhibited good photostability with >70% even for complex 2 for a photoirradiation duration of 1 h in broad band visible light of 400–700 nm (Luzchem Photoreactor, 10 J cm^–2 power). The diiodo-BODIPY complex 2 was found to be more photostable than the ligand (L₂) itself under same experimental conditions.

Figure 2.

(a) Absorption spectra of the complexes 1–3 in 10% DMSO–DPBS medium (pH = 7.2). (b) Emission spectra of the complexes 1–3 in 10% aqueous DMSO (λ_ex = 470 nm). Color code: 1, blue; 2, red; 3, green.

X-ray Crystallography

Two BODIPY complexes, namely, [VO(L₁)(acac)](PF₆) (1a) and [VO(L₂)(acac)](PF₆) (2a) as PF₆ salts, were structurally characterized by the single-crystal X-ray diffraction method.⁴²⁻⁴⁴ ORTEP views of the cationic complexes are shown in Figure 3. Selected bond distances and angles of the complexes are given in Table 2. Complex 1a crystalized in orthorhombic space group Pbca with eight molecules (Z = 8) in the unit cell, while complex 2a crystallized in triclinic space group of P1̅ with a Z value of 2. The mixed-ligand structures of the complexes have a distorted octahedral geometry with a VO₃N₃ core. The tridentate chelating ligands L₁ for 1a or L₂ for 2a, using their N,N,N-donor dipicolylamine (dpa)-based binding sites, are bonded to the oxidovanadium(IV) unit in a facial manner, and the enolic tautomeric form of the monoanionic acetylacetonate (acac) ligand is attached in a O,O-bidentate chelating binding mode. The complexes are mono-cationic with the hexafluorophosphate acting as a counteranion. The V–O(acac) bond distance is ∼1.98 Å. The V–N bond lengths are longer, ranging within 2.10 to 2.36 Å. The V=O bonds are of 1.599(2) and 1.625(13) Å in 1a and 2a, respectively. The central amine nitrogen of the dpa moiety of L₁ and L₂ with sp³ hybridization is positioned trans to the V=O moiety, giving a longer V–N distance of ∼2.36 Å due to the trans effect. The structural data are similar to other reported mixed-ligand oxidovanadium(IV) complexes with V=O distances ranging within 1.56–1.63 Å in a distorted octahedral geometry and a V^IVO₃N₃ core (vide Table S1 for details).^29a,45,46 The angular disposition of the BODIPY unit with the −C₆H₄– linker reduces the effective conjugation between two units.

Figure 3.

ORTEP views of the cationic complexes in (a) [VO(L₁)(acac)](PF₆) (1a) and (b) [VO(L₂)(acac)](PF₆) (2a) showing thermal ellipsoids at 50% probability level (color code: red, V; green, N; blue, O; black, C). The hydrogen atoms and the PF₆ anion are omitted for clarity.

Table 2. Selected Bond Distances (Å) and Angles (°) for [VO(L₁)(acac)](PF₆) (1a) and [VO(L₂)(acac)](PF₆) (2a) with e.s.d.s. in Parentheses.

bond parameters	[VO(L1)(acac)](PF6) (1a)	[VO(L2)(acac)](PF6) (2a)
V(1)–O(1)	1.599(2)	1.625(13)
V(1)–O(2)	1.975(2)	1.979(11)
V(1)–O(3)	1.968(2)	1.976(13)
V(1)–N(1)	2.358(3)	2.350(12)
V(1)–N(4)	2.103(3)	2.141(13)
V(1)–N(5)	2.113(3)	2.126(13)
O(1)–V(1)–O(2)	103.83(11)	101.10(5)
O(1)–V(1)–O(3)	104.79(11)	105.90(6)
O(2)–V(1)–O(3)	88.04(10)	88.40(5)
O(1)–V(1)–N(1)	163.00(11)	159.50(5)
O(1)–V(1)–N(4)	94.31(12)	93.30(6)
O(1)–V(1)–N(5)	93.64(12)	96.80(6)
O(2)–V(1)–N(1)	87.06(9)	95.50(4)
O(2)–V(1)–N(4)	161.65(10)	164.80(5)
O(2)–V(1)–N(5)	88.87(10)	82.70(5)
O(3)–V(1)–N(1)	88.37(10)	86.30(5)
O(3)–V(1)–N(4)	84.65(11)	82.90(5)
O(3)–V(1) −N(5)	161.52(10)	156.90(5)
N(1)–V(1)–N(4)	75.95(10)	71.50(5)
N(1)–V(1)–N(5)	73.28(10)	73.40(5)
N(4)–V(1)–N(5)	92.75(11)	100.50(5)

The boron atom was found to lie within the dipyrrin plane and adopts a tetrahedral geometry in both 1a and 2a giving the N–B–N angles of 107.02(3)° and 110.49(14)°, respectively, while the respective F–B–F angles are 108.57(3)° and 107.29(14)°, respectively. The dihedral angles between the BODIPY unit and the phenyl ring were 76.86° and 91.41° for 1a and 2a, respectively. The angles between two moieties linked by CH₂ involving the dipicolylamine (dpa) and 4,4-difluoro-1,3,5,7-tetramethyl-4-bora-3a,4a-diaza-s-indacene (for 1a) or 4,4-difluoro-2,6-diiodo-1,3,5,7-tetramethyl-4 bora-3a,4a-diaza-s-indacene (for 2a) units were observed to be 115.89° and 111.98°, respectively.

Theoretical Studies

Time-dependent density functional theory (TD-DFT) and DFT studies were made to understand the electronic spectral data of the complexes 1–3 using the B3LYP level of theory with 6-311G+* (for C, H, N, O, B, and F) and Lanl2DZ (for V and I atoms) basis sets by the Gaussian 09 program (Tables S2–S4).⁴⁷⁻⁴⁹ The coordinates obtained from the crystal structures of 1a and 2a were used for the geometrical optimization of the complexes. The optimized structures and the FMOs (frontier molecular orbitals) are shown in Figure 4. The HOMO (highest occupied molecular orbital) is exclusively situated on the BODIPY core for complexes 1 and 3 and on the diiodo-BODIPY core for 2. In contrast, the LUMO (lowest unoccupied molecular orbital) is located on the VO²⁺ moiety and the β-diketonate ligand (acac for 1 and 2 and dbm for 3). The difference in energy between the FMOs for the complexes is ∼2.22 eV. The V^IV–O distances of ∼1.98 Å and V^IV=O distance of ∼1.61 Å are in good agreement with the bond distances obtained from the crystal structures.

Figure 4.

Optimized structures of the complexes 1–3 and the FMOs.

TD-DFT (solvent model: IEFPCM; solvent: DMSO) calculations for the complexes indicate a BODIPY-based strong transition (HOMO → LUMO+1) at ∼475 nm for 1 and 3 with oscillator strength values of 0.83 and 0.55, respectively (Tables S5–S7). A similar transition for the diiodo-BODIPY complex 2 was obtained at ∼496 nm with an oscillator strength value of 0.52. These transitions correspond to the experimentally observed band at ∼501 nm for 1 and 3 and at ∼535 nm for 2. The transitions within 350–380 nm were found to be originated from the β-diketonate ligand and BODIPY → β-diketonate ILCT transitions. Amolev’s rule (intersystem crossing quantum yield, Φ_ISC = 1 – Φ_F) predicts that the complexes 1 and 3 should possess an Φ_ISC of ∼0.90, while complex 2 having heavy iodine atoms is expected to have the value of ∼0.99.⁵⁰ The presence of diiodo-BODIPY ligand L₂ in complex 2 favors spin-orbit coupling (SOC) in a significant manner and hence enhanced the triplet state population due to “heavy atom effect”.⁵¹

Moreover, the energy difference between the ground doublet (D₀) and the first excited triplet (T₁) state of the complexes 1–3 were found to be 1.59, 1.76, and 1.59 eV, respectively, which are considerably higher than the required excitation energy of molecular oxygen (³Σ_g^– → ¹Δ_g, ΔE = 1.06 eV) to generate singlet oxygen (¹O₂). The complexes thus have the potential to act as type-II PDT agents in visible light.

Singlet Oxygen as a ROS

Singlet oxygen (¹O₂) as a reactive oxygen species (ROS) generated from a type-II mechanistic pathway is of importance in PDT.^1,2,7 Other ROS, like hydroxyl radical (^·OH), superoxide (O₂^·–) or peroxide (O₂^2–), generated by a type-I and/or photo-redox pathway, could be scavenged under physiological conditions by different enzymatic systems, namely, by superoxide dismutase (SOD), catalase, and peroxidase. Since ¹O₂ could not be degraded in enzymatic pathways in the human body, it is considered as the most lethal cytotoxic agent. To ascertain the formation of singlet oxygen (¹O₂), titration experiments using 1,3-diphenylisobenzofuran (DPBF) as a scavenger of ¹O₂ were performed. The decay in the absorbance intensity value at the DPBF-based λ_max of ∼417 nm, due to endoperoxide formation, was examined with different photoirradiation times in the light of 400–700 nm (dose = 10 J cm^–2) in the presence of constant concentration of the complexes (molar ratio of DPBF:complex = 50:1). The self-degradation possibility of DPBF was overruled as the control experiments in the absence of the complexes did not show any significant change in the absorbance value of DPBF. It was observed that in the presence of the complexes, the intensity at the absorption maximum at ∼417 nm gets reduced in a linear fashion with respect to the photoirradiation time (Figure 5). No thermal recovery of DPBF was noticed within the time span of the experiment.

Figure 5.

(a) Absorption spectral traces of DPBF and complex 2 on exposure to light (400–700 nm, 10 J cm^–2) for each exposure time of 15 s. (b) Plot showing changes in absorbance of DPBF at 417 nm with time on light exposure with 1–3 (color code: 1, black; 2, red; 3, blue). A higher slope for complex 2 indicates a significant quantity of singlet oxygen (¹O₂) generation.

The first-order decay constants were evaluated in the presence of different complexes, and the values are 6.31 × 10^–4, 7.93 × 10^–3, and 1.11 × 10^–3 for 1–3, respectively, giving an order 2 ≫ 3 > 1. The half-life time for the decay was also estimated to be 1098, 87, and 624 s for 1–3, respectively (Figure S16). The singlet oxygen quantum yield (Φ_Δ) values for the complexes were measured using Rose Bengal (RB) as a standard (Φ_Δ = 0.76 in DMSO) and keeping the optical density (O.D.) for RB or the complexes very low (O.D. of ∼0.1) to avoid any quenching of ¹O₂ by RB or the complexes.^29,30 Singlet oxygen quantum yield (Φ_Δ) values for the complexes are 0.14, 0.47, and 0.18 for 1–3, respectively, while the values are 0.20 and 0.88 for L₁ and L₂, respectively.⁵² Complex 2 with a diiodo-BODIPY unit with a moderately high quantum yield (Φ_Δ) generates a significant quantity of singlet oxygen (Figure S17). Having medium (for 1 and 3) to high (for 2) singlet oxygen generation efficiency, complex 2 is a potential type-II PDT agent, while the non-iodo BODIPY complexes with relatively low Φ_Δ values are suitable for cellular imaging study.

Lipophilicity

Lipophilicity of a drug molecule is important for its absorption, distribution, metabolism, and excretion (ADME) profile for being absorbed by gastrointestinal tract (GIT) and hepatocytes.⁵³ In addition, it is known to support passive diffusion through blood–brain–barrier (BBB). To estimate lipophilicity of the complexes, octanol–water partition coefficients (P) were determined using the shake-flask method. Complex 3 having a 1,3-diphenyl-1,3-propanedionate ligand was found to be the most lipophilic with a log P value of 1.56(±0.06). The observed order of lipophilicity for the complexes is 3 > 2 > 1 (Table 1 and Figures S18 and S19). Since BODIPY-appended L₁ ligand is invariant for 1 and 3, the difference in their log P values is attributed to the lipophilicity difference in the β-diketonate ligands. The log P values for the diketonate ligands follow the order dbm > acac. As expected, complex 2 having the diiodo-BODIPY unit is found to have a log P value greater than complex 1 with a non-iodo-BODIPY moiety. The most lipophilic complex is 3 with a dibenzoylmethane ligand with two pendant phenyl groups, that is, a ligand with high lipophilicity is found to possess a high log P value.

Cell Viability

The efficacy and selectivity of the complexes 1–3 as anticancer agents were verified by in vitro cell viability assay, viz., MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay, using cervical adenocarcinoma HeLa cells in the absence of light and on photoexposure with a visible light source (400–700 nm; Luzchem photoreactor, total dose = 10 J cm^–2; irradiation time of 1 h). The complexes were incubated for 4 h and then exposed to visible light. Another set was used as dark controls to compare the effect of light activation on the complexes. The complexes were generally less toxic (half maximal inhibitory concentration, IC₅₀ > 100 μM for 1 and 3; ∼60 μM for 2) when kept in the dark, whereas they show significant photocytotoxicity on light activation (Figure S20). The IC₅₀ values of 1–3 in HeLa cells are 12.5, 1.1, and 6.3 μM, respectively, giving an order [VO(L₂)(acac)]Cl (2) > [VO(L₁)(dbm)]Cl (3) > [VO(L₁)(acac)]Cl (1) (Table 3).^29a,d,54,55 The diiodo-BODIPY complex 2 was found to give a good phototoxic index (PI) value of >50. A comparison of the activity between complexes having L₁ and L₂ ligands indicates that the diiodo-BODIPY ligand acts as a better photosensitizer-cum-lipophilic moiety than the non-BODIPY ligand. The difference between complexes 1 and 3 is due to the higher lipophilic nature of dpm than acac.

Table 3. IC₅₀ Values (μM) of the Complexes 1–3 in HeLa Cells with Related Compounds.

	HeLa
entry	lighta	darkb
1c	12.5 ± 1.0	>100
2c	1.10 ± 0.04	58.0 ± 0.5
3c	6.30 ± 0.07	>100
[VO(L2)(cur)]Cld	2.5 ± 0.2	55 ± 2
[VO(L1)Cl2] e	1.8 ± 0.6	>50
photofrinf	4.3 ± 0.2	>41
cisplating		24

^aThe sample on photoexposure for 1 h in visible light of 400–700 nm at 10 J cm^–2.

^bThe sample on 4 h incubation in the dark without any photoexposure. In MTT assay, the maximum concentration of the complexes was 100 μM.

^cThe solvent medium was 99:1 (v/v) DMEM medium/DMSO.

^dThe data from ref.^29a Hcur is curcumin.

^eThe data from ref.^29b

^fThe data from ref.⁵⁴

^gThe data from ref.⁵⁵

The pathway of cell death for HeLa with complex 2 (1 μM) was found to be apoptotic in nature from the Annexin V-FITC and PI assay (Figure S21). HeLa cells treated with complex 2 with subsequent 1 h photoirradiation showed ∼33% apoptotic death of total population (18% in early apoptotic stage and 15% in late apoptotic stage), while the necrotic death population was found to be ∼21%. Again, when one set of the HeLa cells was incubated with the non-emissive diiodo-BODIPY complex 2 (1 μM) for 4 h in the dark, confocal microscopic images with nuclear straining Hoechst 33342 dye did not show any abnormal morphological changes. However, when HeLa cells were incubated for 4 h with complex 2 (1 μM) and photoirradiated for 1 h (400–700 nm, Luzchem photoreactor, total dose = 10 J cm^–2), it showed significant nuclear membrane blebbing, indicating its fragmentation, which corresponds to initiation of apoptosis (Figure S22). To analyze the trend in the IC₅₀ values, the cellular uptake of the emissive complexes 1 and 3 in HeLa cells was estimated using flow cytometry. It was found that incorporation of complex 3 having the dbm ligand was higher as compared to that of the acac complex 1 (Figure 6).

Figure 6.

Flow cytometric overlay plot showing the uptake of the complexes 1–3 in HeLa cells after 4 h incubation.

The higher cellular uptake of dbm complex 3 is mainly due to its lipophilicity property. Hence, the IC₅₀ values of 1 and 3 are in good agreement with their rate of incorporation inside the cancer cells. However, the non-emissive complex 2 was found to supersede the rest of the complexes in its photocytotoxicity probably due to its distinctly higher singlet oxygen generation efficiency with a high singlet oxygen quantum yield (Φ_Δ) in addition to its lipophilicity as mentioned above.

Subcellular Localization

The HeLa cells were incubated with the emissive dbm complex 3 (2 μM, Φ_F = 0.09) for 4 h in the dark after which the cells were processed, and confocal microscopic images were captured. To ascertain any nuclear localization of complex 3, confocal microscopic images were taken using a nuclear staining Hoechst 33342 dye, which displayed blue fluorescence (λ_ex = 350 nm; λ_em = 460 nm) in contrast to the green emissive BODIPY complexes (λ_ex = 470 nm; λ_em = 514 nm). The results showed predominant cytoplasmic localization of the complexes from the overlay of the images (Figure 7). Further organelle specific localization study was performed using Mito Tracker Deep Red (MTR) as a mitochondrion staining dye. The merged images revealed significant mitochondrial localization of the complex with a co-localization coefficient value of ∼0.7.

Figure 7.

Confocal microscopic images of complex 3 in HeLa cells after 4 h incubation: Panels (a), (b), and (c) are green fluorescence images of 3, red fluorescence of mitochondria staining dye Mitotracker Red, and blue fluorescence of nuclear straining dye Hoechst, respectively. Panel (d) is the merged image of (a) and (b). Panel (e) is the merged image of (a), (b) and (c). Panel (f) is the merged bright-field image of (a), (b), and (c). Scale bar = 10 μm.

DNA Binding and Cleavage

Since the complexes were found to significantly localize in the mitochondria of the cells, mito-DNA could be their probable target. Thus, the binding propensity of the complexes 1–3 with calf thymus (ct) DNA was studied from UV–visible absorption spectral experiments. The complexes showed partial intercalative mode of binding with ct-DNA with intrinsic binding constant (K_b) values in the order of ∼10⁵ M^–1 in 5% DMF–Tris buffer (pH = 7.2). The binding strength follows the order [VO(L₁)(dbm)]Cl (3) > [VO(L₁)(acac)]Cl (1) > [VO(L₂)(acac)]Cl (2) (Table 1 and Figures S23 and S24). DNA binding properties of the complexes were also studied from competitive ethidium bromide (EB) displacement assay using the emission quenching method. The K_app values of the complexes 1–3 gave a similar order as observed from the K_b values (Table 1 and Figures S25 and S26). Binding propensity of the complexes with ct-DNA was studied in DMF–Tris-HCl buffer (pH = 7.2) using the viscometric titration method at 37 °C, and (η/η₀)^1/3 versus [Complex]/[DNA] plots were compared with standard DNA intercalator ethidium bromide and standard groove binder Hoechst dye (Figure S27). The binding trend of the complexes was found to be similar to the one observed by other methods.^56,57

DNA photocleavage activity of the complexes 1–3 and the ligands (L₁ and L₂) was studied using 10 μM concentration of each sample with appropriate dark controls using supercoiled (SC) pUC19 DNA (30 μM, 0.2 μg) in Tris-HCl/NaCl (50 mM, pH = 7.2) buffer (Figure S28).^58,59 Gel electrophoresis in agarose gel was carried out to estimate the amount of nicked circular (NC) DNA generation from the SC DNA on treatment of the complexes and the BODIPY ligands. A monochromatic green light source of 532 nm was used from a CW (continuous-wave) diode laser (100 mW laser power) for light irradiation. The wavelength selection for this study was based on the presence of the electronic spectral bands at ∼501 nm for 1 and 3 and 535 nm for 2. Each sample was subjected to 1 h photoirradiation after treatment with the DNA solution in Tris-HCl buffer and incubation for 1 h in the dark. The complexes and ligands (10 μM) exhibited negligible DNA cleavage activity (≤10% NC) in the absence of light, while generation of ∼15 and ∼50% of NC DNA was detected in the photoirradiated complexes having L₁ and L₂, respectively. However, the β-diketone ligands alone did not show any apparent DNA photocleavage activity. Complexes 1 and 2 showed ∼45% and 3 as ∼30% of NC DNA, thus giving an order 2 ≈ 1 > 3 (Figure S28). The DNA photocleavage activity of the non-iodo BODIPY complexes is similar to the diiodo-analogue, indicating the importance of the bulk of the BODIPY ligands in DNA cleavage activity. The ROS generation by complex 2 in the photo-triggered condition was also verified in HeLa cells by DCFDA assay (Figure S29).

The nature of ROS was ascertained from the DNA photocleavage mechanistic data using the diiodo-BODIPY complex 2 (10 μM) and different singlet oxygen quenchers and hydroxyl and other radical scavengers (Figure S30). A marginal enhanced percentage of NC DNA was observed in D₂O in which singlet oxygen has a longer lifetime. In contrast, addition of TEMP and NaN₃ as singlet oxygen quenchers significantly reduced the percentage of NC DNA. Addition of DMSO, KI, or catalase as a radical scavenger did not show any apparent change in the NC DNA percentage. A similar result was obtained on addition of SOD. The percentage of NC DNA diminished to a negligible value when the experiment was carried out in an argon atmosphere, thus indicating involvement of oxygen (³O₂) for the DNA photocleavage activity. The mechanistic data indicate generation of singlet oxygen as the sole reactive oxygen species (ROS). This observation is of importance as photofrin is also known to generate only singlet oxygen as the ROS in the PDT process and not any other radical species.

Conclusions

Mixed-ligand oxidovanadium(IV) complexes of monoanionic β-diketone and dipicolylamine (dpa) ligands having two pendant photoactive BODIPY moieties are designed and prepared for targeted PDT activity with the complexes showing significant visible light-induced (400–700 nm,10 J cm^–2) cytotoxicity in HeLa cervical cancer cells while being less active in the dark. Two BODIPY complexes are structurally characterized by single-crystal X-ray crystallography. The complexes having a green-emissive BODIPY moiety exhibited predominant subcellular mitochondrial accumulation and considerable uptake inside the cancer cells. The non-emissive diiodo-BODIPY complex as a photosensitizer showed a high singlet oxygen quantum yield and remarkable PDT activity. Apoptotic cell death of the HeLa cancer cells in presence of complex 2 was due to the formation of singlet oxygen species as evidenced from the annexin-V assay, and the nature of ROS was evidenced from the mechanistic data from the pUC19 DNA photocleavage study and DPBF titrations. Complex 2 having a diiodo-BODIPY ligand is found to be an excellent PDT agent with a good photocytotoxic index (PI) value of >50 (IC₅₀ in the dark: 58.0 μM, IC₅₀ in the light: 1.1 μM). Vanadium being a biocompatible metal ion, complex 2 having a lipophilic diiodo-BODIPY unit as a photosensitizer is a potential metal-based photodynamic therapy agent.

Experimental Section

Materials and Methods

The chemicals and reagents were purchased from standard commercial sources. Solvents used were purified and distilled by standard methods.⁶⁰ Synthesis of the complexes was carried out in a N₂ atmosphere utilizing the Schlenk technique. The chemical reagents, trackers, dyes, DNA, and buffer for DNA binding, photocleavage, and cellular experiments were also procured from standard commercial sources (vide ref^29a for details). The BODIPY (4,4-difluoro-1,3,5,7-tetramethyl-4-bora-3a,4a-diaza-s-indacene)-appended dipicolylamine (L₁) and diiodo-BODIPY (4,4-difluoro-2,6-diiodo-1,3,5,7-tetramethyl-4-bora-3a,4a-diaza-s-indacene)-appended dipicolylamine (L₂) were prepared by literature procedures.^61,62 Acetylacetone (Hacac) and 1,3-diphenyl-1,3-propanedione (Hdbm) were procured from Sigma-Aldrich, U.S.A. Tetrabutylammonium perchlorate (TBAP) was synthesized using tetrabutylammonium bromide and perchloric acid. TBAP was used in a small quantity with care (caution!).

Electrospray ionization (ESI) mass spectral measurements and elemental analysis data of 1–3 were obtained using a model 6538 UHD Accurate-mass Q-TOF LC/MS ESI mass spectrometer from Agilent Technologies and Thermo Finnigan FLASH EA 1112 CHNS analyzer, respectively. The electronic spectra (UV–vis) and infrared spectra were recorded with Perkin-Elmer spectrum one 55 and Perkin-Elmer Lambda 35 spectrometer, respectively, at ambient temperature. Fluorescence measurements were performed using a Perkin-Elmer LS 55 fluorescence spectrometer. A relative method, as described in the literature, was used to measure the fluorescence quantum yield values of the compounds.⁶³ A Control Dynamics (India) conductivity meter was used for molar conductivity measurements. Magnetic susceptibility was obtained using a magnetic susceptibility balance (Sherwood Scientific). Diamagnetic corrections were done using a literature report.⁶⁴ Electrochemical measurements were performed at room temperature using an EG&G PAR model 253 VersaStat potentiostat/galvanostat. Electrochemical analysis software 270 was used. A three-electrode setup of a glassy carbon working, platinum wire auxiliary, and a saturated calomel reference electrode (SCE) in DMF–0.1 M TBAP was used. Flow cytometric analysis and confocal microscopic measurements were carried out using FACS (fluorescence-activated cell sorting) Calibur Becton Dickinson (BD) cell analyzer at FL2 channel (595 nm) and ApoTome.2 fluorescence microscope, respectively.

Synthesis of [VO(L₁/L₂)(acac/dbm)]Cl (1–3)

To prepare the complexes, vanadyl sulfate (0.16 g, 1.0 mmol) and barium chloride (0.24 g, 1.0 mmol) were initially dissolved in 18 mL of aqueous ethanol (EtOH:H₂O = 5:1 v/v). The mixture that was stirred at room temperature for 3 h in a nitrogen atmosphere was filtered to discard the white barium sulfate as the precipitate. The blue colored filtrate was deaerated and then saturated with N₂. A deaerated ethanol solution (10 mL) of Hacac (0.10 g, 1.0 mmol) was slowly added to this filtrate for complexes 1 and 2 and Hdbm (0.22 g, 1.0 mmol) for 3, which was previously deprotonated with Et₃N (0.10 g, 1.0 mmol) for the enolic forms. A deep green solution thus formed, after stirring the mixture for 30 min, was reacted with the CH₂Cl₂ solution (10 mL) of L₁ (0.54 g, 1.0 mmol) for 1 and 3 or L₂ (0.79 g, 1.0 mmol) for 2. The product got precipitated out of the solution spontaneously on stirring for 2 h. The reddish brown solid was filtered and then isolated followed by washing with cold EtOH, CH₂Cl₂, and Et₂O. It was finally dried in vacuum in a desiccator over P₄O₁₀.

[VO(L₁)(acac)]Cl (1)

Yield = 78%. Anal. Calcd for C₃₇H₃₉BClF₂N₅O₃V: C, 60.30; H, 5.33; N, 9.50. Found: C, 59.99; H, 5.15; N, 9.34. ESI-MS in CH₃OH: m/z 701.3565 [M – Cl]⁺. IR data/cm^–1: 3365 m (br), 1608 m, 1592 s, 1542 vs, 1523 m, 1472 w, 1364 w, 1158 m, 1054 w, 972 m, 756 m, 584 w, 472 m (vs, very strong; s, strong; m, medium; w, weak; br, broad). UV–visible in 1:1 (v/v) DMSO:Tris-HCl [λ_max /nm (ε/M^–1 cm^–1)]: 733 (50), 501 (37500), 474 (8330), 370 (12560), 267 (18410). Molar conductivity in DMF at 298 K [Λ_M/S cm² mol^–1]: 72. μ_eff, μ_B at 298 K: 1.61.

[VO(L₂)(acac)]Cl (2)

Yield = 75%. Anal. Calcd for C₃₇H₃₇BClF₂I₂N₅O₃V: C, 44.95; H, 3.77; N, 7.08. Found: C, 44.77; H, 3.57; N, 7.16. ESI-MS in CH₃OH: m/z 953.0490 [M – Cl]⁺. IR data/cm^–1: 3364 m (br), 1613 m, 1591 s, 1539 vs, 1524 m,1471 w, 1375 w, 1166 m, 1076 w, 975 m, 763 m, 585 w, 525 s, 473 m. UV–visible in 1:1 (v/v) DMSO:Tris-HCl [λ_max/nm (ε/M^–1 cm^–1)]: 724 (66), 536 (31910), 505 (11670), 394 (4630), 260 (6810). Molar conductivity in DMF at 298 K [Λ_M/S cm² mol^–1]: 67. μ_eff, μ_B at 298 K: 1.65.

[VO(L₁)(dbm)]Cl (3)

Yield = 80%. Anal. Calcd for C₄₇H₄₃BClF₂N₅O₃V: C, 65.56; H, 5.03; N, 8.13. Found: C, 65.33; H, 4.84; N, 7.95. ESI-MS in CH₃OH: m/z 825.2279 [M – Cl]⁺. IR data/cm^–1: 3364 m (br), 1610 m, 1593 s, 1540 vs, 1523 m,1478 w, 1366 w, 1156 m, 1055 w, 971 m, 755 m, 576 w, 480 m. UV–visible in 1:1 (v/v) DMSO:Tris-HCl [λ_max/nm (ε/M^–1 cm^–1)]: 732 (56), 500 (40560), 472 (10550), 350 (16720), 257 (27470). Molar conductivity in DMF at 298 K [Λ_M/S cm² mol^–1]: 70. μ_eff, μ_B at 298 K: 1.63.

Solubility

The complexes showed good solubility in common organic solvents, namely, EtOH, MeOH, MeCN, DMSO, and DMF. They were moderately soluble in chlorinated solvents (chloroform and dichloromethane) and insoluble in hydrocarbon solvents and diethyl ether. The complexes as their chloride salts showed good aqueous solubility (∼150 μM in 99:1 v/v DPBS:DMSO) and hence were suitable for anticancer activity studies.

X-ray Crystallographic Method

The methanol solution of [VO(L₁)(acac)]Cl (1) and the acetonitrile solution of [VO(L₂)(acac)]Cl (2) were treated with NH₄PF₆ to isolate the complexes as their hexafluorophosphate salts (1a and 2a) in crystalline forms as attempts to isolate the single crystals of the chloride salts were unsuccessful. The crystal structures of [VO(L₁)(acac)]PF₆ (1a) and [VO(L₂)(acac)]PF₆ (2a) were obtained by single-crystal X-ray diffraction method.⁴⁶⁻⁴⁸ Single crystals of 1a were obtained upon slow evaporation of the 1:3 MeOH/CHCl₃ (v/v) solution of the complex, whereas single crystals of 2a were obtained from diethyl ether vapor diffusion into the acetonitrile solution of the complex. Among the crystals, a crystal of suitable quality and dimensions was mounted on a crystal mounting loop with paratone oil. The intensity data were collected using graphite monochromated Mo Kα radiation (0.7107 Å) at 100 K for 1a and 293 K for 2a. The structures of 1a and 2a were solved in Pbca and P1̅ space group of orthorhombic and triclinic crystal system with Z values of 8 and 2, respectively. The structures solved by direct methods using SHELXL-2014 incorporated in WinGX (Version 1.63.04a) exhibited no abnormality in the core molecular structure of the complexes. The high R-indices (all data) for 2a were due to poor quality of the diffraction data. Reflections collected/Unique are 8862/5563 for complex 1a and 8784/2608 for complex 2a (vide Table S8 in the Supporting Information). All atoms barring the hydrogen atoms were refined anisotropically for complex 1a. Due to a low data-to-parameter ratio, only few nonhydrogen atoms for 2a were refined anisotropically. The hydrogen atoms were refined using a riding model. An analysis of the data indicated that the reduced number of unique reflections for 2a was due to photodegradation of its crystals on exposure to X-ray radiation during data collection (vide Figures S12 and S14 in the Supporting Information). The data-to-parameter ratio for this structure was not suitable enough to refine all nonhydrogen atoms anisotropically. This constraint led to the A-alert in the checkCIF(PLATON) file for 2a. The molecular structure obtained from the available data did not show any abnormal structural features and compare well with that of 1a. The atomic coordinates were successfully used to obtain the energy minimized structure of this complex for the photophysical studies to interpret the spectral data. Selected crystallographic parameters are given in Table S8. The CCDC deposition numbers are 1815715 and 1815716.

Theoretical Method

The geometric optimization of the complexes were performed by hybrid density functional theory (DFT) using the B3LYP level of theory with 6-311G+* (for C, H, N, O, B, and F) and Lanl2DZ (for V and I) basis sets by the Gaussian 09 program.⁵¹⁻⁵³ The coordinates were initially acquired from the crystal structures of the complexes 1a and 2a. Time-dependent DFT (TD-DFT) was employed to obtain the electronic transitions. The optimized coordinates and the electronic transitions are given in Tables S2 and S4–S7 (see the Supporting Information).

Partition Coefficient Determination

The shake-flask method was used to perform the experiment. At first, the absorbance calibration curve was prepared for the complex under examination using different concentrations of the complexes in the aqueous medium. Then, known concentrations of the aqueous solutions of the complexes were mixed in vessels with the same volume of analytical grade n-octanol. After shaking the mixture well at room temperature for 4 h, it was allowed to settle down. Two phases got separated, and the remaining concentration of the complex in the aqueous phase was determined by using the previously prepared calibration curve.

DNA Binding and Cleavage Experiments

UV–visible spectral studies were performed to measure the DNA binding constants of the complexes using calf thymus (ct) DNA following reported methods.^56,57 DNA binding by the ethidium bromide (EB) displacement method and viscosity method were also carried out (vide the Supporting Information). A Schott Gerate AVS310 automated viscometer fitted with a thermostatic bath of 37 °C was used. The details on photocleavage experiments are given in the Supporting Information.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.9b04204.

• Singlet oxygen quantum yield (Φ_Δ) measurement, cellular experiments, and DNA binding and cleavage experiments; reaction schemes of ligands and complexes (Schemes S1 and S2); ESI-MS (Figures S1–S3); IR spectra (Figures S4–S6); cyclic voltammograms (Figures S7–S9); d-d bands of 1–3 (Figure S10); stability data from UV–visible plots (Figures S11–S15); DPBF absorbance decay plots (Figures S16 and S17); partition coefficient calibration curves (Figures S18 and S19); MTT assay plots (Figure S20); annexin-PI assay (Figure S21); confocal images (Figure S22); DNA binding data (Figures S23–S27); DNA photocleavage gel and bar diagrams of ligands and complexes (Figure S28); DCFDA assay (Figure S29); DNA photocleavage gel and bar diagrams from mechanistic study with complex 2 (Figure S30); (Table S1) comparison of bond parameters of mixed-ligand oxidovanadium(IV) complexes; (Tables S2–S7) DFT data; and (Table S8) selected crystallographic data of 1a and 2a (PDF)

The authors declare no competing financial interest.