Synthesis and in Vitro Studies of a Series of Carborane-Containing Boron Dipyrromethenes (BODIPYs)

Abstract

A series of seven BODIPYs functionalized with ortho-carborane groups at the 8(meso) or 3/5(α) position were synthesized and characterized by NMR, HRMS, HPLC, and in the cases of 2b and 5b, by X-ray analysis. The BODIPYs exhibited low dark toxicity and phototoxicity toward human glioma T98G cells, and their cellular uptake varied significantly, with 5b accumulating the most and 7 the least. All BODIPYs localized mainly within the cell ER. The BODIPYs showed higher permeabilities than lucifer yellow across human hCMEC/D3 brain endothelial cell monolayers as the BBB model. Among this series, 1b showed the highest BBB permeability (P_e = 16.4 × 10⁻⁵ cm/s), probably as a result of its lower MW (366 Da) and favorable hydrophobicity (log P = 1.5). The combination of low cytotoxicity, amphiphilicity, high boron content, high cellular uptake, and moderate BBB permeability renders these compounds promising boron delivery agents for the BNCT of brain tumors.

Graphical Abstract

INTRODUCTION

Boron neutron capture therapy (BNCT) is a very promising binary anticancer methodology, especially for the treatment of brain tumors because it can selectively target and destroy malignant cells in the presence of healthy normal cells.^1–3 BNCT involves the irradiation of nonradioactive ¹⁰B-containing tumors with low-energy thermal neutrons, causing the excitation of ¹⁰B to ¹¹B, which rapidly produces cytotoxic high linear energy transfer (high-LET) α and ⁷Li particles, γ radiation, and about 2.4 MeV of kinetic energy through a nuclear fission reaction. The generated high-LET particles have short path lengths of less than 10 μm in tissue, therefore restricting the damage to ¹⁰B-containing tumor cells. Furthermore, biologically abundant nuclei including ¹²C, ¹H, and ¹⁴N possess very small nuclear cross sections, thus posing little interference with the ¹⁰B(n,α)⁷Li capture reaction. Another advantage of BNCT is that thermal and epithermal neutrons are able to penetrate deep into tissues to reach deep-seated tumors. However, a relatively high boron tumor concentration of at least 20 μg ¹⁰B/g tumor is required for effective BNCT treatment, and this has driven recent research in the areas of boron drug development and delivery methodologies for BNCT.

BNCT has been applied for the treatment of high-grade brain tumors, such as glioblastoma multiforme (GBM), and other difficult-to-treat malignancies, including melanomas and recurrent head and neck cancers.^3–5 One major challenge in the BNCT modality for brain tumors arises from the existence of the blood–brain barrier (BBB) that prevents most drugs from penetrating into the brain and from reaching the targeted tumor cells.^6–11 The BBB permeability of drugs has been shown to be related to the physicochemical properties of the drugs, including their lipophilic character, molecular weight (MW), size, polar surface area, charge, and extent of ionization.^12–14 Among these, the two main characteristics in a small molecule that favor its crossing of the BBB by passive diffusion are (1) molecular weight under 500 Da and (2) high lipophilic character, usually measured by the octanol–water partition coefficient, log P < 5.^6,12–14 Although significant hydrophobicity is important for enhanced permeability across the BBB, BNCT drugs also need to be soluble in aqueous media to enable their systemic administration.² Therefore, there is continued need to develop amphiphilic compounds as boron delivery drugs for BNCT of brain tumors, with appropriate balance between hydrophilicity and lipophilicity, as well as molecular weight of <500 Da for efficient BBB permeability and tumor cell uptake.

Two boron-containing drugs of low molecular weight have been extensively used in BNCT clinical trials: the sodium salt of the sulfhydryl boron hydride Na₂B₁₂H₁₁SH (BSH) and L-4-dihydroxyborylphenylalanine (BPA).^3–5 Although BSH and BPA have demonstrated efficacy in BNCT clinical trials, they have very limited BBB permeability.^10,15 Other boronated compounds have been investigated as potential BNCT drugs, including amino acids, peptides, carbohydrates, nucleosides, liposomes, porphyrins, and monoclonal antibodies (mAbs).^2,7 Among these, particularly promising are mAbs due to their very high specificity for a tumor-associated epitope, and porphyrin derivatives due to their ability to transport large amounts of boron within cells. However, these BNCT agents have limited BBB crossing ability, mainly due to their large size, high molecular weight, and hydrophobicity.^2,15

We have recently investigated the BBB permeability of a series of carboranylporphyrins conjugated to polyamines, glucose, arginine, and an opioid peptide,¹⁶ but they all showed low permeabilities (P_e < 3.3 × 10⁻⁶ cm/s) across hCMEC/D3 cell monolayers^17–19 as the BBB model. On the other hand, amphiphilic boron dipyrromethene (BODIPY) compounds of low molecular weight could show enhanced permeability and promise as boron delivery drugs for BNCT.²⁰ BODIPY dyes have attracted special interest in recent years due to their various applications in biological labeling, drug delivery, imaging, sensing, and theranostics.^21–24 BODIPYs are strongly UV–vis absorbing and generally emit sharp fluorescence with high quantum yields, which can facilitate detection and quantification of tissue-localized boron in BNCT. BODIPYs have also shown negligible sensitivity toward solvent polarity and solution pH, high permeability across cellular membranes, and relatively high stability under physiological conditions.

BODIPYs functionalized with carborane clusters have been prepared via Suzuki²⁰ and Sonogashira²⁵ cross-coupling reactions of the corresponding 2,6-diiodo-substituted compounds. Alternatively, carboranes can be introduced onto the BODIPY at the 3(5)- and/or 8-positions through substitution reactions on the corresponding chlorinated derivatives.^20,26,27 Herein we report the synthesis of a series of seven carboranyl-BODIPYs from the corresponding chloro-BODIPY derivatives, with molecular weights in the range 366–527 Da and log P in the range 1.5–2.7. The cytotoxicity and uptake of the BODIPYs in human glioma T98G cells, as well as their permeability across the BBB using hCMEC/D3 cells, were investigated and compared.

EXPERIMENTAL SECTION

1. Synthesis

All the chemicals and reagents were purchased from Sigma-Aldrich and Fisher Scientific and used as received. 1-Mercapto-1,2-carborane was purchased from Katchem. Reactions were monitored by analytical thin-layer chromatography (TLC) performed on precoated plates (polyester-back, 60 Å, 0.2 mm, Sorbent Technologies). Purifications were conducted by column chromatography on silica gel (230–400 mesh, 60 Å, Sorbent Technologies) or preparative TLC plates (f254, VWR). ¹H and ¹³C NMR spectra were obtained using a Bruker AV-400 NanoBay (400 MHz for ¹H NMR and 100 MHz for ¹³C NMR) and a Bruker AV-500 spectrometer (125 MHz for ¹³C NMR) at room temperature. ¹¹B NMR was obtained on a Bruker AV-400 III (128 MHz), using BF₃·OEt₂ as reference. Chemical shifts (δ) are given in parts per million (ppm) in CDCl₃ (7.27 ppm for ¹H NMR, 77.0 ppm for ¹³C NMR); coupling constants (J) are given in Hz. High resolution mass spectrometry (HRMS) spectra were obtained using a 6210 ESI-TOF mass spectrometer (Agilent Technologies). Normal-phase HPLC was performed on a Dionex system including a P680 pump and UVD 340 detector connected to a Dynamax axial compression column packed with Rainin 60 Å irregular silica gel. The flow rate of 1 mL/min was used. For compound 1b, a stepwise gradient of 50% B (ethyl acetate) and 50% A (hexane) in the first 3 min to 80% B and 20% A during the next 18 min to 50% B and 50% A for the next 6 min was used. For all other compounds, a stepwise gradient of 10% B and 90% A in the first 5 min to 70% B and 30% A during the next 10 min to 10% B and 90% A for the next 10 min was used. All compounds were of ≥95% purity, as determined by HPLC (see Supporting Information, Figures S17–S23). The commercially available LY standard is of 97% purity, as provided by Life Technologies. 4-(1-Methyl-1,2-carborane)-methylphenylboronic acid²⁸ and BODIPYs 1a,²⁹ 3a,²⁶ 3b,²⁰ 6a,²⁷ and 7²⁷ were synthesized as previously reported.

General Procedure for Synthesis of BODIPYs via Nucleophilic Substitution Reaction

The starting chloro-BODIPY (0.05 mmol) was dissolved in 2 mL of THF. 1-Mercapto-1,2-carborane (0.055 mmol) and K₂CO₃ (0.5 mmol) were added, and the final mixture was stirred at room temperature. TLC was used to monitor the reaction until completion (2–5 h). The crude solid product was filtered and purified by column chromatography or preparative TLC using CH₂Cl₂/hexanes or ethyl acetate/hexanes for elution.

8-(1,2-Carboranyl-1-thio)-BODIPY 1b

16.1 mg (88%), mp 213–215 °C. ¹H NMR (CDCl₃, 400 MHz): δ 8.01 (2H, s), 7.36 (2H, s), 6.64 (2H, s), 3.77 (1H, s), 1.8–3.5 (10H, br). ¹³C NMR (CDCl₃, 100 Hz): δ 148.6, 139.4, 133.3, 132.2, 120.4, 72.2, 65.0. ¹¹B NMR (CDCl₃, 128 MHz): δ −0.08 (1B, t, ¹J_(B,F) = 27.8 Hz), −12.97 to −1.57 (10B, m). HRMS (ESI-TOF) m/z calcd for C₁₁H₁₇B₁₁F₁N₂S [M – F]⁺ 347.2199; found 347.2200.

8-(1,2-Carboranyl-1-thio)-1,3-dimethyl-BODIPY 2b

18.7 mg (95%), mp 157–160 °C. ¹H NMR (CDCl₃, 400 MHz): δ 7.71 (1H, s), 7.10 (1H, s), 6.49 (1H, s), 6.29 (1H, s) 3.79 (1H, s), 2.65 (1H, s), 2.55 (1H, s), 1.8–3.5 (10H, br). ¹³C NMR (CDCl₃, 100 Hz): δ 166.9, 148.6, 141.4, 138.3, 137.2, 130.0, 127.0, 126.0, 117.4, 73.3, 65.1, 16.7, 15.7. ¹¹B NMR (CDCl₃, 128 MHz): δ 0.14 (1B, t, ¹J_(B,F) = 29.8 Hz), −12.97 to −1.60 (10B, m). HRMS (ESI-TOF) m/z calcd for C₁₃H₂₂B₁₁F₂N₂S [M + H]⁺ 395.2576; found 395.2566.

8-(1,2-Carboranyl-1-thio)-2-ethyl-1,3,6,7-tetramethyl-BODIPY 5b

20.9 mg (93%), mp 198–200 °C. ¹H NMR (CDCl₃, 400 MHz): δ 7.44 (1H, s), 3.70 (2H, s), 2.58 (3H, s), 2.44–2.45 (8H, overlap, m), 2.02 (3H, s), 1.8–3.5 (10H, br), 1.05–1.09 (3H, t, ³J_(H,H) = 7.4 Hz). ¹³C NMR (CDCl₃, 100 Hz): δ 162.7, 141.5, 141.1, 137.8, 137.2, 134.1, 130.1, 128.4, 74.9, 64.1, 17.3, 14.5, 14.2, 14.0, 13.4, 10.2. ¹¹B NMR (CDCl₃, 128 MHz): δ −0.06 (1B, t, ¹J_(B,F) = 30.0 Hz), −11.89 to −1.64 (10B, m). HRMS (ESI-TOF) m/z calcd for C₁₇H₃₀B₁₁F₂N₂S [M + H]⁺ 451.3204; found 451.3195.

8-(1,2-Carboranyl-1-thio)-3-chloro-2-ethyl-1,3,6,7-tetramethyl-BODIPY 6b

17.9 mg (74%), mp 192–194 °C. ¹H NMR (CDCl₃, 400 MHz): δ 3.69 (1H, s), 2.60 (3H, s), 2.44(8H, overlap, m), 2.00 (3H, s), 1.8–3.5 (10H, br), 1.06–1.10 (3H, t, ³J_(H,H) = 7.6 Hz). ¹³C NMR (CDCl₃, 125 Hz): δ 163.1, 141.3, 141.2, 137.8, 137.7, 137.6, 132.7, 128.5, 126.4, 74.9, 64.1, 17.3, 14.8, 14.6, 14.1, 13.5, 9.1. ¹¹B NMR (CDCl₃, 128 MHz): δ 0.08 (1B, t, ¹J_(B,F) = 30.0 Hz), −12.94 to −1.45 (10B, m). HRMS (ESI-TOF) m/z calcd for C₁₇H₂₉B₁₁ClF₂N₂S [M + H]⁺ 485.2818; found 485.2824.

8-[4-(2-Methyl-1,2-carboranyl)methylphenyl]-BODIPY 4

BODIPY 1a (11.3 mg, 0.05 mmol) was dissolved in toluene (4 mL). 1 M Na₂CO₃ (aq) (1 mL), Pd(PPh₃)₄ (5% mol), and 4-(1-methyl-o-carborane)methylphenylboronic acid (29.2 mg, 0.1 mmol) were added, and the final mixture was refluxed overnight. The mixture was poured into water (10 mL) and extracted with dichloromethane (10 mL × 3). The organic layers were collected, washed with H₂O and brine, and dried over anhydrous Na₂SO₄. The solvents were removed by rotary evaporation to give the crude products. The further purification was performed by column chromatography (ethyl acetate/hexanes as the eluent) to provide the titled product (10.6 mg) in 45% yield; mp 195–197 °C. ¹H NMR (CDCl₃, 400 MHz): δ 8.0 (2H, s), 7.56–7.58 (2H, m), 7.37–7.39 (2H, m), 6.92 (2H, s), 6.58 (2H, s), 3.57 (2H, s), 2.22 (3H, s), 1.8–3.5 (10h, br). ¹³C NMR (CDCl₃, 100 Hz): δ 146.4, 144.5, 137.7, 134.9, 133.6, 131.5, 130.7, 130.4, 118.7, 75.0, 40.9, 23.8. ¹¹B NMR (CDCl₃, 128 MHz): δ 0.18 (1B, t, ¹J_(B,F) = 28.6 Hz), −10.74 to −3.16 (10B, m). HRMS (ESI-TOF) m/z calcd for C₁₉H₂₅B₁₁FN₂ [M – F]⁺ 419.3108; found 419.3111.

2. Spectroscopic Studies

UV–visible and fluorescence spectra were collected on a PerkinElmer Lambda 35 UV/vis spectrometer and PerkinElmer LS 55 luminescence spectrometer at room temperature. Quartz cuvettes (10 mm path length) and spectroscopic grade solvents were used for both measurements. Optical density (ε) was determined by using the solutions with absorbance at λ_max (0.5–1). Quantum yields were determined by using the dilute solutions with absorbance (0.04–0.06) at the particular excitation wavelength. Cresyl violet perchlorate (0.54 in methanol) and rhodamine 6G in ethanol (0.95) were used as external standards for the carboranyl BODIPYs 5b, 6b, and 1b–3b, 4, 7, respectively. The relative fluorescence quantum yields (Φ_f) were determined by calculations using the following equation,³⁰

$${\mathrm{\Phi }}_{\mathrm{X}}={\mathrm{\Phi }}_{\mathrm{R}}(F_{\mathrm{X}}/F_{\mathrm{R}})(A_{\mathrm{R}}/A_{\mathrm{X}}){(n_{\mathrm{X}}/n_{\mathrm{R}})}^2{\mathrm{\Phi }}_{\mathrm{R}}$$

where Φ stands for fluorescence quantum yields, n stands for refractive indexes, F stands for the areas under the emission peaks, A stands for absorbance at the particular excitation wavelength, and subscripts X and R refer to the tested and standard samples, respectively.

3. Crystallography

X-ray data for 2b and 5b were collected at 90 K with Mo Kα radiation (λ = 0.710 73 Å) on a Bruker Kappa Apex-II DUO diffractometer. For BODIPY 2b, C₁₃H₂₁B₁₁F₂N₂S, monoclinic space group P2₁/c, a = 10.7729(5) Å, b = 15.7487(7) Å, c = 12.5738(6) Å, β = 104.222(2)°, V = 2067.88(17) ų, Z = 4, 22855 measured data. Final R = 0.040, Rw = 0.106 for 264 refined parameters and 7883 independent reflections having θ_max = 33.2°. For BODIPY 5b, C₁₇H₂₉B₁₁F₂N₂S, monoclinic space group P2₁/c, a = 13.688(2) Å b = 12.1938(17) Å, c = 15.488(2) Å, β = 113.231(6)°, V = 2375.3(6) ų, Z = 4, 47866 measured data. Final R = 0.040, Rw = 0.110 for 303 refined parameters and 8658 independent reflections having θ_max = 32.6°. The CIFs have been deposited at the Cambridge Crystallographic Data Centre (CCDC 1426277-1426278).

4. Octanol–Water Partition Coefficients

The partition coefficients (log P) were measured by adapting a reported procedure.³¹ 1-Octanol and Milli-Q water were mutually saturated, and the two phases were separated. A 64 μM stock solution was prepared by dissolving the BODIPY into water-saturated 1-octanol. Then 2 mL of the stock solution was added to 6 mL of Milli-Q water in a 15 mL volumetric tube and the mixture was intensively vortexed for 10 min. After complete separation of the two phases, an aliquot from the 1-octanol layer was diluted with water-saturated 1-octanol and its absorbance was recorded with a Varian Cary 50 Bio UV–vis spectrophotometer with a 10 mm path length quartz cuvette. The log P values were calculated as follows:

$$logP=log\left(\frac{A_{\text{oct}}}{A_0-A_{\text{oct}}}\frac{V_{\mathrm{W}}}{V_{\mathrm{O}}}\right)$$

where A₀ and A_oct are the absorbance of the compound in the water-saturated octanol before and after partitioning; V_w and V_O are the water and 1-octanol volumes, respectively.

5. Cell Studies

The T98G cell line used in this study was purchased from ATCC and cultured in ATCC-formulated Eagle’s minimum essential medium containing 10% FBS and 1% antibiotic (penicillin–streptomycin). The HEp2 cells were also purchased from ATCC and maintained in a 50:50 mixture of DMEM/AMEM supplemented with 5% FBS and Primocin antibiotic. The hCMEC/D3 cells were obtained from Dr. Pierre-Olivier Couraud from Institut COCHIN in Paris (France). All other reagents were purchased from Life Technologies.

5.1. Dark Cytotoxicity

A 32 mM compound stock solution was prepared by dissolving the BODIPY in 100% DMSO. The stock solution was diluted into final working concentrations (0, 6.25, 12.5, 25, 50, and 100 μM). Human glioma T98G cells were plated at 15 000 cells per well in a Costar 96-well plate (BD biosciences) and allowed to grow for 24 h. The cells were exposed to the working solutions of compounds up to 100 μM and incubated overnight (37 °C, 95% humidity, 5% CO₂). The working solution was removed, and the cells were washed with 1× PBS. The medium containing 20% CellTiter Blue (Promega) was added and incubated for 4 h. The viability of cells is measured by reading the fluorescence of the medium at 570/615 nm using a BMG FLUOstar Optima microplate reader. This fast, sensitive, and popular assay uses the indicator dye resazurin which is reduced to fluorescent resorufin in viable cells, while nonviable cells are not able to reduce resazurin or to generate a fluorescent signal. The fluorescence signal of the untreated cells was normalized to 100%.

5.2. Phototoxicity

Human glioma T98G cells were prepared as described above. The cells were incubated with compound concentrations of 100, 50, 25, 12.5, 6.25, 3.125, and 0 μM for 24 h. The loading medium was removed, and the cells were washed with 1× PBS buffer and then refilled with fresh media. The cells were exposed to a 600 W halogen lamp light source filtered with a water filter (transmits radiation 250–950 nm) and a beam turning mirror with 200 nm to 30 μm spectral range (Newport), for 20 min. The total light dose was approximately 1.5 J/cm². After light exposure, the cells were returned to the incubator for 24 h and assayed for cell viability as described above.

5.3. Time-Dependent Cellular Uptake

Human T98G cells were prepared as described above. The cells were exposed to 10 μM of each compound solution for 0, 1, 2, 4, 8, and 24 h. The loading medium was removed at the end of each incubation period, and the cells were washed with 1× PBS and solubilized by adding 0.25% Triton X-100 in 1× PBS. Standard curves using 10, 5, 2.5, 1.25, 0.625, and 0.3125 μM concentrations were obtained by diluting 400 μM of each BODIPY solution with 0.25% Triton X-100 (Sigma-Aldrich) in 1× PBS. A cell standard curve was prepared using 10⁴, 2 × 10⁴, 4 × 10⁴, 6 × 10⁴, 8 × 10⁴, and 10⁵ cells per well. The cell number was quantified using a CyQuant cell proliferation assay (Life Technologies). The compound concentration in cells at each time period was determined using a BMG FLUOstar Optima microplate reader at 485/590 nm. Cellular uptake is expressed in terms of compound concentration (nM) per cell.

5.4. Microscopy

Human HEp2 cells were incubated in a six-well plate (MatTek) and allowed to grow overnight. The cells were exposed to 10 μM of each BODIPY and incubated for 6 h (37 °C, 95% humidity, 5% CO₂), followed by the addition of organelle tracers obtained from Invitrogen. The organelle tracers were used at the following concentrations: LysoSensor Green, 50 nM; MitoTracker Green, 250 nM; ER Tracker Blue/White, 100 nM; and BODIPY FL C5 Ceramide, 50 nM. The cells were incubated with the BODIPY and tracers for 30 min and washed with PBS three times before imaging. The images were acquired using a Leica DMRXA2 upright microscope with a water immersion objective and DAPI, GFP, and Texas Red filter cubes (Chroma Technologies).

5.5. hCMEC/D3 Cell Line (BBB Model)

The BBB permeabilities were determined following a published procedure.^32,33 Specifically, the hCMEC/D3 cells were incubated in a six-well, 0.4 μm porosity PET Transwell plate (Corning) for 48 h, allowing the formation of a model brain capillary endothelial monolayer (checked by microscopy). EBM-2 medium containing 5% FBS, 1% penicillin/streptomycin, hydrocortisone, ascorbic acid, chemically defined lipid concentrate (1/100), HEPES, and bFGF was used as the growth medium. The coated PET Transwell plates with and without endothelial cells were transferred into six-well plates. The measurements were performed in triplicate for each compound, using three PET Transwell plates with cells and three without cells. The time points of the treatment were 0, 10, 25, and 45 min. At time 0, a 0.5 mL sample of each BODIPY or standard lucifer yellow (LY) at 50 μM concentration in transport buffer was added to the upper chamber (mimicking the blood), and 1.5 mL of transport buffer was added to the lower chamber (mimicking the BBB); see Figure S24 in the Supporting Information. The transport buffer was prepared by adding 5 mL of HEPES (1 M) and 5 mL of sodium pyruvate (100 μM) to 400 mL of HBSS. The plates were incubated at 37 °C, 95% humidity, and 5% CO₂. At time 10 and 25 min, each upper PET Transwell was transferred to the corresponding prepared six-well plate containing 1.5 mL of transport buffer: the so-called “25 min” first and then the “45 min”. At each time point, solution in the lower chamber was added into a 96-well plate with 100 μL for each well (five wells). The fluorescence intensity of the solution was measured by using a BMG FLUOstar plate reader at 485/590 nm and 425/538 nm (excitation/emission) for BODIPYs and LY, respectively, and the concentrations were determined from the corresponding standard curves. The calculations of permeability coefficients (P_e, in cm/s) were performed following the clearance principle as described in the equations below (eqs 1 and 2), where X is the amount of sample in the lower chamber and Cd is the concentration of sample in the upper chamber at each time point. The total cleared volume at each time point is calculated by summing the incremental cleared volumes up to the given time point.

$$\text{clearance}(\text{mL})=X/\text{Cd}$$ (1)

The cleared volume is plotted vs time, and a linear fit is applied.

$$P_{\mathrm{e}}(\text{cm}/\mathrm{s})=\left(\frac{1}{\text{PSt}}-\frac{1}{\text{PSf}}\right)/(A\times 60)$$ (2)

where PSt is the slope of the clearance curve for the culture, PSf is the slope of the clearance curve with the control Transwell plate without cells, and A is the surface area of the PET Transwell plate.

RESULTS AND DISCUSSION

1. Synthesis and Characterization

A series of seven carboranyl-BODIPYs 1b, 2b, 3b,²⁰ 4, 5b, 6b, and 7,²⁷ five of which are new, were synthesized as shown in Scheme 1. The key chloro-BODIPY starting materials were synthesized from the corresponding dipyrroketones, as previously reported. ^26,27,29 The ortho-carborane cluster was chosen as the boron source because of its high boron content, high hydrophobicity, high stability under physiologic conditions, and low toxicity.^7,11 8-Chloro-BODIPYs 1a–3a and 5a were converted into their corresponding 8-carboranylthio-BODIPYs by reaction with 1.1 equiv of 1-mercapto-o-carborane in THF at room temperature, in yields ranging from 88% to 95%. We have previously reported that 3,8-dichloro-BODIPY 6a undergoes highly regioselective substitutions at the 8-position in the presence of N- and O-centered nucleophiles.²⁷ Using more reactive S-nucleophiles, the 8- vs 3-substitution regioselectivity tends to decrease. However, due to the electron-withdrawing nature of the carborane cluster, BODIPY 6a reacted smoothly with 1.1 equiv of 1-mercapto-o-carborane, affording 6b with high regioselectvity in 74% yield. On the other hand, BODIPY 7 was prepared from 6a via two successive regioselective reactions, a Stille cross-coupling at the 8-position using 1 equiv of tributylphenylstannane and Pd(PPh₃)₄, followed by substitution using an excess of 1-mercapto-o-carborane, as we have previously reported.²⁷ The Suzuki cross-coupling reaction of BODIPY 1a with 1.5 equiv of 4-(1-methyl-o-carborane)-methylphenylboronic acid in toluene and in the presence of Pd(PPh₃)₄ and 1 M Na₂CO₃(aq) produced BODIPY 4 in 45% isolated yield.

Scheme 1.

The structures of the new BODIPYs 1b, 2b, 5b, 6b, and 4 were confirmed by ¹H, ¹³C, and ¹¹B NMR, HRMS and in the cases of 2b and 5b by X-ray crystallography (Figure 1). We have previously reported the X-ray structures of 3b²⁰ and 7.²⁷ Crystals of BODIPYs 2b and 5b suitable for X-ray analysis were obtained by slow diffusion of hexanes into chloroform (Figure 1). In 2b, the B atom of the central C₃N₂B ring lies slightly (0.214 Å) out of the plane of the other five atoms, which are fairly coplanar, having a mean deviation of 0.013 Å. The carborane lies on the bisector of this plane, with C–C–S–C torsion angle 93.9°. In 5b, the central ring is slightly more planar, with the 8-C atom lying 0.098 Å out of the C₂N₂B plane. The S atom lies 0.319 Å out of this plane, and as in 2b, the carborane lies on the bisector, with C–S–C–C torsion angle of 94.1°. The ethyl group is also approximately perpendicular to the ring system, with C–C–CH₂–CH₃ torsion angle of 83.7°.

Figure 1.

X-ray crystal structures of 2b (left) and 5b (right).

The spectroscopic properties of all BODIPYs in dichloromethane solution were investigated, and the results are summarized in Table 1 (see also Figures S32–S44 of the Supporting Information). The absorption spectra of all BODIPYs followed the Lambert–Beer law, indicative of no aggregation in this solvent at the concentrations tested. The introduction of an ortho-carboranylthio group at the meso(8)-position in BODIPYs 1b–3b, 5b, and 6b caused large red-shifts on the maximum absorption (up to 57 nm) and emission (up to 73 nm) bands. This is probably due to the stabilization of the LUMO by this group, which decreases the HOMO–LUMO gap.³⁴ On the other hand, introduction of the same group at the α(5)-position, as in BODIPY 7, induced a slight blue-shift (~5 nm) relative to the starting BODIPY 6a.²⁷ Arylation at the 8-position, as in BODIPY 4, also produced a slight blue-shift compared with the starting 8-chloro-BODIPY 1a probably due to a large dihedral angle between the aryl group and the BODIPY core. The fluorescence quantum yields were higher for BODIPY 1b, suggesting low rotational freedom for the 8-carboranylthio group, and it decreased with increasing alkyl substitution for 2b, 3b, and 5b. This is attributed to increased energy lost to nonradiative deactivation processes.^27,34 On the other hand, BODIPY 4 shows much lower quantum yield than 7, probably due to the higher rotation of the 8-aryl group in the absence of 1,7-methyl groups.

Table 1.

Spectroscopic Properties of BODIPYs in Dichloromethane at Room Temperature

BODIPY	absorption λmax (nm)	log ε (M−1 cm−1)	emission, λmax (nm)	Φfa	Stokes shift (nm)
1b	544	4.42	556	0.52	12
2b	537	4.54	553	0.29	16
3b	554	4.45	576	0.065	22
4	502	4.50	517	0.034	15
5b	577	4.29	608	0.060	31
6b	582	4.16	609	0.090	27
7	521	4.18	540	0.58	19

^aRhodamine 6G in ethanol (0.95) was used as standard for all compounds except for 5b and 6b, which used cresyl violet perchlorate in methanol (0.54) as the standard.

2. Cytotoxicity and Uptake in T98G Cells

The cytotoxicity (dark and light, using 1.5 J/cm² light dose) and time-dependent uptake of all carboranyl-BODIPYs were investigated in human glioma T98G cells, and the results are summarized in Table 2 and Figure 2 (see also Figures S25 and S26 in the Supporting Information). The IC₅₀ values were calculated from dose–response curves. This is the first report on the cytotoxicity and cellular uptake of carborane-containing BODIPYs.

Table 2.

Dark Cytotoxicity, Photocytotoxicity (1.5 J/cm²), and Uptake Plateau Concentration of BODIPYs in Human Glioma T98G Cells, Permeabilty Coefficients (P_e) of BODIPYs and Lucifer Yellow (LY) in Human Endothelial hCMEC/D3 Cells, and Octanol–Water Partition Coefficients (log P) of BODIPYs

compd	MW (g/mol)	log P	dark cytotoxicity, IC50 (μM)	phototoxicity, IC50 (μM)	cellular uptake at 24 h (nM/cell)	Pe × 10−5 (cm/s)
LY	457.24					2.2 ± 0.8
1b	366.25	1.50	>100	>100	0.23 ± 0.05	16.4 ± 3.3
2b	394.31	1.69	>100	98	0.36 ± 0.03	6.0 ± 1.0
3b	422.36	1.73	>100	80	0.11 ± 0.02	4.4 ± 0.5
4a	438.33	2.11	>100	>100	0.41 ± 0.03
5b	450.41	1.95	>100	40	1.5 ± 0.05	2.6 ± 0.9
6b	484.85	1.93	>100	>100	0.46 ± 0.06	5.4 ± 0.6
7a	526.50	2.70	>100	>100	0.0034 ± 0.0004

^aThe P_e values were not determined for these compounds due to their limited solubility in buffer.

Figure 2.

Time-dependent uptake of BODIPYs 1b (green), 2b (black), 3b (red), 4 (blue), 5b (purple), 6b (dark red), and 7 (orange) at 10 μM in human glioma T98G cells.

None of the BODIPYs, with the exception of 5b, showed any toxicity in the dark at concentrations up to 100 μM, as shown in Supporting Information, Figure S25. Upon irradiation with light, some BODIPYs showed enhanced cytotoxicity (see Supporting Information Figure S26), namely, 5b with calculated IC₅₀ = 40 μM and 3b with calculated IC₅₀ = 80 μM; for all other BODIPYs the IC₅₀ values were above 98 μM. The higher phototoxicity observed for 5b might be a result of its higher uptake by T98G cells, as shown in Figure 2. Indeed, 5b accumulated the most within cells at all the time points investigated, while 7 accumulated the least, among this series of BODIPYs. The very low uptake observed for BODIPY 7 is probably due to its poor aqueous solubility. The lipophilic character of this series of BODIPYs was evaluated by determining their distribution between 1-octanol and water, and the values obtained for the partition coefficients (log P) are given in Table 2. The extent of cellular uptake did not correlate with the hydrophobic character of the BODIPYs, which increased in the order 1b < 2b < 3b < 5b ~ 6b < 4 ≪ 7. With the exception of 7, all compounds were taken up rapidly in the first 2 h after which slower uptake was observed for all compounds except for 1b where a plateau was reached. After 24 h, the amount of compound accumulated within cells varied considerably (Table 2); compound 5b showed the highest uptake, about 436-fold higher than 7, followed by 6b (135-fold higher than 7), 4 (118-fold higher than 7), 2b (106-fold higher than 7), 1b (67-fold higher than 7), and 3b (33-fold higher than 7). These results indicate that among this series of compounds, BODIPY 5b could deliver the largest and therapeutic amount of boron within glioma cells, with very low dark toxicity.

To investigate the sites of subcellular localization of the carboranyl-BODIPYs, HEp2 rather than T98G cells were used for fluorescence microscopy, as the HEp2 cells facilitate imaging by nicely spreading on the surface of the six-well plate. The organelle-specific probes BODIPY Ceramide (Golgi), LysoSensor Green (lysosomes), MitoTracker Green (mitochondria), and ER Tracker Blue/White (endoplasmic reticulum) were used in the overlay experiments. The results are shown in Figures 3 and 4 for BODIPYs 1b and 5b, respectively (see also Supporting Information, Figures S27–S31). All carboranyl-BODIPYs localized preferentially in the cell ER, as shown in Figures 3d and 4d. In addition, the BODIPYs were also observed, but to a smaller extent, in the lysosomes, mitochondria, and Golgi apparatus. These results are in agreement with previous studies showing preferential ER-localization for BODIPY molecules.

Figure 3.

Subcellular localization of 1b in HEp2 cells at 10 μM for 6 h: (a) phase contrast, (b) overlay of 1b and phase contrast, (c) ER Tracker Blue/White, (d) overlay of 1b and ER Tracker, (e) BODIPY ceramide, (f) overlay of 1b and BODIPY ceramide, (g) MitoTracker Green, (h) overlay of 1b and MitoTracker, (i) LysoSensor Green, (j) overlay of 1b and LysoSensor. Scale bar: 10 μm.

Figure 4.

Subcellular localization of 5b in HEp2 cells at 10 μM for 6 h: (a) phase contrast, (b) overlay of 5b and phase contrast, (c) ER Tracker Blue/White, (d) overlay of 5b and ER Tracker, (e) BODIPY ceramide, (f) overlay of 5b and BODIPY ceramide, (g) MitoTracker Green, (h) overlay of 5b and MitoTracker, (i) LysoSensor Green, (j) overlay of 5b and LysoSensor. Scale bar: 10 μm.

3. BBB Permeability

The hCMEC/D3 cell line is a useful model for studies of BBB permeability because it retains many of the morphological and functional characteristics of human brain endothelial cells.^17–19,33 We have recently used this model to investigate the BBB permeability of a series of carboranylporphyrins¹⁶ and BODIPYs.²⁰ All compounds of these types tested to date showed low permeability values (P_e < 3 × 10⁻⁶ cm/s) with exception of BODIPY 3b²⁰ which showed higher permeability (P_e = 4 × 10⁻⁵ cm/s) than LY. This was attributed to its smaller MW and lower hydrophobic character among the compounds tested. In this study we evaluated a new series of derivatives of 3b (MW = 422, log P = 1.7), with MW and hydrophobic character in the ranges 366–527 Da and 1.5–2.7, respectively, chosen to favor BBB and tumor cell permeability. For comparison purposes and for evaluation of the cell monolayer integrity, we also determined the BBB permeability for LY, a polar fluorescent molecule with MW in the same range (457.24 Da) as the BODIPYs. The results obtained are shown in Table 2. The P_e values could not be determined for 4 and 7, due to their poor solubility in buffer, which caused these compounds to precipitate. All carboranyl-BODIPYs tested showed higher BBB permeability compared with LY, a marker for low BBB permeability. Among this series of compounds, 1b showed the highest P_e value (7-fold higher than LY), maybe due to its lower MW (366 Da) and favorable hydrophobic character (log P = 1.50), conferring it the highest solubility in buffer as well as lipophilicity. Compound 2b, of similar MW (394 Da) and hydrophobicity (log P = 1.69) to 1b, showed the second highest P_e value. The other BODIPYs, including 3b, with increased alkyl and/or aryl substitution, MW > 400 Da, and log P > 1.7, showed lower P_e values although still higher than that determined for LY. Compound 5b showed the lowest P_e value among this series of BODIPYs, slightly higher than LY, and therefore is considered to have low BBB permeability, although it was the most efficiently taken up by the T98 glioma cells (Figure 2). All other BODIPYs display BBB permeabilities higher than that reported for phenytoin, which has been used as a marker for medium BBB permeability.^33,35 Among the new BODIPYs, 1b, 2b, and 6b showed higher permeability than 3b,²⁰ which displays a slightly higher P_e value than phenytoin. These results show that both the MW and amphiphilicity of the BODIPYs influence their permeability across the BBB and that small amphiphilic carboranyl-BODIPYs of MW < 400 Da and log P < 1.7, such as 1b and 2b, are the most efficient at diffusing across the BBB by passive diffusion.

CONCLUSIONS

A series of seven amphiphilic carboranyl-BODIPYs with MW within the range 366–527 Da, including a previously reported BODIPY known to permeate across a BBB model, were synthesized in good yields from the corresponding chloro-BODIPYs by nucleophilic substitution. The structures of the BODIPYs were confirmed by NMR, HRMS and, in the cases of 2b and 5b, by X-ray crystallography. The carboranyl-BODIPYs display intense absorption and emission bands in the visible region of the spectrum and quantum yields in dichloromethane in the range 0.6–0.03. All BODIPYs showed low dark toxicity (IC₅₀ > 100 μM) in human glioma T98G cells, an important property of potential boron delivery agents because of the high boron concentration requirement in BNCT (>20 μg/g tumor). The BODIPYs also showed low phototoxicity (IC₅₀ > 80 μM) with the exception of 5b (IC₅₀ = 40 μM), probably as a result of its remarkably high uptake into T98G glioma cells. On the other hand, BODIPY 1b showed the largest permeability across the BBB model consisting of hCMEC/D3 cells. Our results show that the BODIPYs with MW < 400 Da and log P < 1.7 are the most efficient at crossing the BBB model. The most hydrophobic compound 7 (log P = 2.7) was poorly soluble in aqueous solutions, showed very low uptake into T98G cells, and its precipitation in buffer precluded determination of its BBB permeability. All BODIPYs tested showed higher BBB permeability compared with LY, as well as low dark cytoxicity and therefore could potentially be efficient boron delivery agents for BNCT of brain tumors. Among this series, 1b and 2b showed the highest BBB permeability while 5b and 6b accumulated the most within tumor cells; therefore, these are the most promising BNCT agents.

ABBREVIATIONS USED

BNCT

boron neutron capture therapy

BBB

blood–brain barrier

LET

linear energy transfer

PDT

photodynamic therapy

PBS

phosphate buffered saline

EBM-2

endothelial cell basal medium 2

bFGF

human basic fibroblast growth factor

FBS

fetal bovine serum

HEPES

2-[4-(2-hydroxyethyl)-piperazin-1-yl]ethanesulfonic acid

HBSS

Hanks’ balanced salt solution

TLC

thin layer chromatography

HPLC

high performance liquid chromatography

ER

endoplasmic reticulum