Synthesis and Characterisation of a Boron-Rich Symmetric Triazine Bearing a Hypoxia-Targeting Nitroimidazole Moiety

Boron Neutron Capture Therapy (BNCT) is a binary therapy that promises to be suitable in treating many non-curable cancers. To that, the discovery of new boron compounds able to accumulate selectively in the tumour tissue is still required. Hypoxia, a deficiency of oxygen in tumor tissue, is a great challenge in the conventional treatment of cancer, because hypoxic areas are resistant to conventional anticancer treatments. 2-Nitroimidazole derivatives are known to be hypoxia markers due to their enrichment by bioreduction in hypoxic cells. In the present work, 2-nitroimidazole was chosen as the starting point for the synthesis of a new boron-containing compound based on a 1,3,5-triazine skeleton. Two o-carborane moieties were inserted to achieve a high ratio of boron on the molecular weight, exploiting a short PEG spacer to enhance the polarity of the compound and outdistance the active part from the core. The compound showed no toxicity on normal human primary fibroblasts, while it showed noteworthy toxicity in multiple myeloma cells together with a consistent intracellular boron accumulation.


Introduction
Boron Neutron Capture Therapy is a binary treatment in which boron compounds are introduced into cancer cells and then irradiated by thermal neutrons. Neutron capture yields excited 11 B which gives nuclear fission generating high-energy alpha particles ( 4 He nuclei) and lithium-7 ( 7 Li) nuclei causing damage to the cell. Due to their high linearenergy transfer (LET), α ( 4 He) and lithium particles deposit their energy within a range of 5-9 µm (approximately the diameter of a cell), limiting their effects mainly to the boroncontaining cells [1,2].
Although many compounds have been tested for their suitability for BNCT until today, BSH and BPA are still the only agents clinically used [3]. In the past, most clinical studies focused on the treatment of melanoma and glioblastoma, but BNCT has the potential to treat many types of cancer and even some other diseases like arthritis [4]. The tumor-targeting ability and the selectivity are among the most important requisites for new compounds. In particular, they should be able to accumulate the commonly accepted optimal boron amount for neutron capture therapy, namely ≥10 9 10 B atoms per cell, corresponding to 20 µg boron per g of tumour tissue [3]. Future agents are expected to be designed for specialized tasks and able to offer the use of BNCT to other types of cancer, treating tumour types with very low life expectancy and increase the life quality of patients by eliminating the side effects of conventional cancer therapies.
Hypoxia represents a great obstacle in treating malignant cells. The rapid and uncontrollable tumour growth prevents the build-up of an adequate blood vascular structure causing deprivation of nutrients and oxygen in deeper areas. The resistance of hypoxic areas to both radio-and chemotherapy is the main factor in poor prognosis and cancer recurrence [5,6].
2-Nitroimidazole is known to undergo bioreductive trapping in hypoxic cells, making it a promising starting point for the design of hypoxia-targeting agents [5]. After entering the cell through diffusion, it undergoes single-electron reduction. In non-hypoxic cells, it is oxidized by intracellular oxygen and can leave the cell again. In hypoxic tissue, however, it undergoes further reduction by intracellular nitroreductase enzymes like xanthine oxidase, lipoxygenase, and NADPH oxidase [7]. Therefore, its ability to accumulate in hypoxic cells by bioreductive trapping makes 2-nitroimidazole a promising starting point for the design of hypoxia-targeting agents.
In the last 25 years, various boron-containing 2-nitroimidazole derivatives have been synthesized [8][9][10][11][12]. Recently, we have reported an agent we named B381, which has shown specific accumulation in cancer tissues but, containing only one boron atom per molecule, it has furnished low boron accumulation [12]. To enhance boron molecular mass fractions, clusters like carboranes (C2B10H12) [13] can be used for new 2-nitroimidazolecontaining BNCT agents. Modular systems based on triazines are yet described and are combined with m-carboranes to boron-rich compounds [14][15][16]. In this paper, a symmetrical hybrid compound containing two carborane residues and 2-nitroimidazole moiety on a triazine scaffold is described as a potential agent for BNCT.

Chemical Synthesis
Reagents and solvents were purchased from TCI Europe (Italian distributor: Zentek SRL, Milan, Italy) or Carlo Erba (Milan, Italy) and used without further purifications. o-Carborane was purchased from Katchem (Praha, Czech Republic) Celite ® 577 from Sigma Aldrich (Milan, Italy). Thin-layer chromatography plates (TLC) equipped with fluorescent indicators were purchased from Merck (Milan, Italy). The visualization was accomplished by UV light (254 and/or 365 nm) and by staining in ceric ammonium molybdate. Palladium chloride was used for carborane derivatives. Solvents were dried with freshly activated molecular sieves (0.3 or 0.4 nm; Sigma Aldrich, Milan, Italy). Reactions were performed under nitrogen atmosphere in oven-dried glassware (when the use of water as reagent or solvent was not specified). Silica gel 230-400 mesh (Merck, Milan, Italy) was used for flash column chromatography.
Infrared spectroscopy was recorded with an FT-IR Thermo-Nicolet Avatar 370. Thermo Finnigan LCQ-Deca XP-plus mass spectrometer equipped with an ESI source and an ion trap detector was used for mass spectroscopy. Orbitrap QExactive PLUS (Thermo Fischer Scientific, Bremen, Germany) was used for the determination of the exact mass of final compound 1.
A JEOL ECP 300 spectrometer (JEOL Ltd., Tokyo, Japan) was used for NMR analyses. Stuart Scientific SMP3 apparatus were used for the determination of the melting points, which are uncorrected.

1-2-[2-(2-azidoethoxy)ethoxy]ethyl-2-nitro-1H-imidazole (5)
Compound 2 (290 mg, 1.14 mmol), dissolved in 6.5 mL anhydr. dimethylformamide (DMF) was added under nitrogen atmosphere to a three-necked flask together with Cs2CO3 (423 mg, 1.31 mmol) and 2-nitroimidazole 4 (163 mg, 1.43 mmol). Under stirring, N2 was passed through the solution for 0.25 h, then the reaction was heated to 110 °C for 18 h. After checking the complete conversion of 2 by TLC (EtOAc/cyclohexane 7:3), the solvent was removed under vacuum. The crude was dissolved in EtOAc, filtered through a Celite pad, . Therefore, an additional 0.1 eq. of PPh3 was added and the solution was stirred for another 24 h. After evaporation of the organic phase, the acidic aqueous phase was extracted 5 times with 5 mL dichloromethane (DCM) to remove Ph3PO from the solution and the pH was set to 12 by addition of NaOH pellets. The basic aqueous phase was extracted seven times with 5 mL DCM each, the combined organic phases were dried over MgSO4, filtered, and concentrated under vacuum.

Cell Viability Assay
Cell viability was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, as previously reported [17]. 5 × 10 3 fibroblast were seeded in 96-well plates in complete medium for 24 h. Attached fibroblasts were treated with increasing concentrations (0.1-100 µg/mL) of test compounds for 24-72 h. MM.1S cells at a concentration of 50,000 cells/well were treated with increasing concentrations (0.1-100 µg/mL) of test compounds for 24-72 h. The percentage of cell viability was calculated as [100 (x − y)/(z − y)], where x, y, and z were the absorbance read in compoundtreated, resting, and compound-untreated cells, respectively. Results are expressed as mean ± SD of at least three experiments run in triplicate.

MM.1S cells (2 × 106 cells/1mL media) were treated with compounds for 2 h at 37 °C.
Then the cells were washed 4 times with PBS, pellet weight was registered, then resuspended, vortexed, and digested in nitric acid overnight. Samples were diluted with deionized water to a final acid concentration of 5% (v/v), filtered through a 0.22-micron polyethersulfone syringe filter (DiKMA Technologies, Lake Forest, CA, USA), and analyzed using Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES, Optima 7300 V series, Perkin Elmer, Waltham, MA, USA). The following day, the compounds were added and the cells were trypsinized, collected, and proceeded cells after 2, 6, or 24 h. Samples were analyzed for boron content (λ = 249.677 nm) against a calibration curve of boron standards of 0-250 parts per billion (ppb) prepared from 10 parts per million boron standard solution (Inorganic Ventures, Christiansburg, VA, USA).

Statistical Analysis
Experiments were performed in triplicates and replicated independently at least three times and the results were shown as mean ± standard deviation (S.D.) or mean ± standard error of the mean (S.E.M.). For the statistical significance, one-way ANOVA was used and the values were considered significantly different for * p < 0.05, ** p < 0.01 and *** p < 0.001.

Synthesis of Carborane Hybrid Triazine
Compound 1 (Figure 1), the synthesis of which is reported in the Scheme 1, is composed of three elements: a nitroimidazole subunit, to target hypoxic tumour cells, and two o-carboranes as boron moieties, linked to the triazine through triethylene glycolderived spacers chosen to enhance the solubility of the compound. The spacers were equipped with a primary amine on one side to allow nucleophilic substitution of the chlorine groups of the cyanuric chloride. The spacer molecules 2 [18] and 3 [19] were prepared according to literature procedures.
Propargyl carborane 10 was prepared from o-carborane following our previously reported procedure [20]. The 2-nitroimidazole containing moiety was synthesized starting from spacer 2 by nucleophilic substitution of the mesylate with 2-nitroimidazole in DMF with Cs2CO3 as a base, and a reaction time of 18 h at 110 °C [7]. After the workup and flash chromatographic purification, 5 was obtained as a yellow oil in 72% yield. The following aromatic nucleophilic substitution on cyanuric chloride requires the reduction of the azido group of the compound 5 to the corresponding amine 6, which was achieved by a Staudinger reaction with triphenylphosphine. A two-phase system of Et2O and HCl was used to allow the easy removal of triphenylphosphine oxide side-product by extraction. Subsequent basification of the aqueous phase allowed the extraction of the desired product 6 with DCM, which was obtained after concentration under vacuum as a paleyellow oil in a 73% yield and was used directly in the following step.
The assembly of the substituents on the triazine scaffold was performed in two steps under temperature control, to allow the sequential substitutions of the chlorine atoms. To prevent double or triple substitution [21,22] the first reaction of cyanuric chloride 7 with 6 was performed using an excess of the reagent 7. In a first attempt, a ratio of 2:1 was used but was reduced to 1.2:1 in the follow-up reaction due to problems in separating the product from the remaining reactant. Under a nitrogen atmosphere, the reactants 6 and 7 were dissolved in dry THF. To keep the reaction under control, the solution was cooled to −10 °C before the dropwise addition of DIPEA. The progression of the reaction was monitored by TLC and stopped immediately when the spot of 6 had disappeared. Flash chromatography of the crude reaction mixture gave 8 as a dense yellow oil in 82% yield. The substitution of the other two chlorine atoms by 3 was started at room temperature. After the slow addition of DIPEA, the reaction was stirred under dry conditions for 90 min and 21 h at 80 °C. By flash chromatography (DCM with solvent gradient MeOH from 0-5%) 10 was obtained as a thick yellow oil (71% yield).
As last step 10 was connected to 9 exploiting the copper(I) catalysed variant of the azide-alkyne Huisgen cycloaddition [23]. This reaction allows the regioselective synthesis of 1,4-disubstituted triazoles at room temperature in aq. solutions catalyzed by Cu(I). Finally, t-BuOH/H2O (9:1) was used as the solvent. After 16 h the reaction was stopped and after purification by flash chromatography (DCM with 0-5% MeOH) compound 1 was obtained in 65% yield.

Viability Test on Human Primary Fibroblasts
To evaluate the effect of compound 1 and BSH on cell viability human primary fibroblasts were used as representative cells, being fibroblasts present in many tissues and organs. Human primary fibroblast viability was assessed after cell treatment (24-72 h) with increasing concentrations (0.1-100 µM) of each compound, by MTT assay. No evident reduction of cell viability was measured at any concentrations and times considered (Figure 2) on human primary fibroblasts.

Viability Test and Intracellular Boron Content on Human Multiple Myeloma Cell Line
Next, we tested the effect of the compound on cell viability of a cancer cell line derived from multiple myeloma (MM) patients. MM.1S cells were treated (24-72 h) with increasing concentrations of compound 1 (0.1-100 µg/mL), as well as known MM chemotherapy including Ixazomib (0.1-1 ng/mL) and Pomalidomide (0.1-100 µg/mL) for 72 h, and cell viability was measured by MTT assay. Figure 3A demonstrates that Ixazomib in ng range is very toxic and MM.1S cells are very sensitive, Pomalidomide is less toxic and the compound 1 is moderate to MM.1S cells in µg range, implying that the compound 1 is being moderately toxic to MM.1S cells during longer exposures. Therefore, we tested the accumulation of the compound at 2 h based on boron content using ICP-OES. We demonstrate that MM.1S cancer cells accumulate boron at the highest concentration already at 2 h post-exposure ( Figure 3B). µg/mL) for 24 h, 48 h, and 72 h, as well as Ixazomib (0.1-1 ng/mL) and Pomalidomide (0.1-100 µg/mL) for 72 h (A). Cell viability was assessed by MTT assay. The concentration-response curves show the percentage of cell viability in comparison with controls (untreated cells). The data represent the mean ± SD of at least three independent experiments run in quadruplets. One-way ANOVA was performed showing significance between 24 vs. 48 h (p < 0.001), 24 vs. 72 h (p < 0.001) and 48 vs. 72 h (p = 0.1222, non-significant) MM.1S were treated with increasing concentrations (1-100 µg/mL) of compound 1 for 2 h and the boron content was measured by ICP-OES. The data represent the mean ± SD of three independent experiments run in triplicates. Student t-test was performed showing significance between 0 vs. 10 µg/mL (p = 0.0047), 0 vs. 100 µg/mL (p < 0.001) and 10 vs. 100 µg/mL (p < 0.001) (B).

Conclusions
As stated in the introduction, we have previously reported a hypoxia-targeted BNCT agent B381, based on 2-nitroimidazole, which has shown specific accumulation in cancer tissues in vitro and in vivo due to the specific reduction of the 2-nitroimidazole and conjugation to intracellular proteins [12]. However, because B381 compound included only one boron atom in each molecule (per 2-nitroimidazole), it resulted in relatively low boron tissue accumulation. In this study, and to improve the low boron accumulation observed with B318, we proposed the use of highly boronated 2-nitroimidazole derivative. Herein we have reported the synthesis, stability evaluation, and preliminary biological evaluation of a triazine-based derivatives containing 2-nitroimidazole and two symmetrical o-carborane units, 1 (20 boron atoms per molecule). Compound 1 showed no toxicity on normal human primary fibroblasts (representing normal tissue), while it showed significant toxicity in multiple myeloma (cancer) cells. The results revealed that at 100 ug/mL no toxicity was observed in normal cells, while it abolished the cancer cells with over 90% killing, demonstrating an outstanding therapeutic index. The cellular toxicity was driven by intracellular boron accumulation that was observed even within 2 h of exposure. These results were demonstrated in tissue cultures with normal oxygen; in the future we will investigate the biology of the compound and its derivatives in a deeper manner, especially exploring the mechanism of its accumulation in hypoxic tumour cells in vitro and its accumulation in cancer models in vivo, where tumour hypoxia is expected to develop.

Conflicts of Interest:
The authors declare no conflict of interest.