Modular Synthetic Approach to Carboranyl‒Biomolecules Conjugates

The development of novel, tumor-selective and boron-rich compounds as potential agents for use in boron neutron capture therapy (BNCT) represents a very important field in cancer treatment by radiation therapy. Here, we report the design and synthesis of two promising compounds that combine meta-carborane, a water-soluble monosaccharide and a linking unit, namely glycine or ethylenediamine, for facile coupling with various tumor-selective biomolecules bearing a free amino or carboxylic acid group. In this work, coupling experiments with two selected biomolecules, a coumarin derivative and folic acid, were included. The task of every component in this approach was carefully chosen: the carborane moiety supplies ten boron atoms, which is a tenfold increase in boron content compared to the l-boronophenylalanine (l-BPA) presently used in BNCT; the sugar moiety compensates for the hydrophobic character of the carborane; the linking unit, depending on the chosen biomolecule, acts as the connection between the tumor-selective component and the boron-rich moiety; and the respective tumor-selective biomolecule provides the necessary selectivity. This approach makes it possible to develop a modular and feasible strategy for the synthesis of readily obtainable boron-rich agents with optimized properties for potential applications in BNCT.


Introduction
Since boron neutron capture therapy (BNCT) was ascertained to be a very promising binary cancer treatment [1][2][3][4], research has focused on the development of potent and selective boron-containing drugs [5,6]. The main advantage of this therapy is the generation of highly cytotoxic particles comprising a high linear energy transfer (LET) (α particle and Li particle). Their free mean path lengths of about 5 to 9 µm [7,8] roughly represent the diameter of a human cell [5]; therefore, these particles can only harm the surrounding tissue within this radius. However, only the lighter isotope of boron, 10 B (20% natural abundance) [9], produces high LET particles after irradiation with thermal neutrons [10]. Therefore, BNCT agents have to be enriched with 10 B [11,12]. Activation of the BNCT agents is caused by irradiation with thermal neutrons [13,14] for which 10 B exhibits a large capture cross section (3835 barn, 1 barn = 1 × 10 −24 cm 2 ) [9]. This renders BNCT a promising strategy to treat malignant tissue with tumor-selective boron-containing drugs [6,[15][16][17][18][19][20], as the thermal neutron beam can be focused on the affected area [21][22][23][24], thus generating therapeutic particles only upon neutron irradiation. In this manner, normal tissue can be spared and severe side effects, as known from pure radiotherapy or systemic effective chemotherapy, can be reduced.
The first boron-containing compounds used in clinical trials were L-boronophenylalanine (L-BPA) and sodium borocaptate (BSH) [5,6,10], but both compounds exhibit several drawbacks. For example, BSH and BPA do not exhibit optimized selectivity towards cancer cells (especially BSH), show a limited solubility in water (especially BPA [25]) and, in the case of BSH, are not able to penetrate cells due to their anionic character. Therefore, their application follows a tailored strategy where BSH is mostly applied for glioma treatment, as the dianionic compound is able to cross the damaged blood-brain barrier adjoining the malignant tissue in the human brain, and BPA is used as its fructose complex to overcome the low water solubility [5,6,10,25]. Since May 2020, the company Stella Pharma [26] has been allowed to market Steboronine ® [27] (generic name: Borofalan), which contains 10 Benriched (99%) L-BPA as its D-sorbitol complex. This BNCT agent, in comparison to the respective fructose complex, exhibits the advantage of being storable for about three years and does not have to be freshly prepared for each use with retention of its GMP grade.
Therefore, the development of novel boron-containing tumor-selective agents for application in BNCT is important to overcome these limitations [19,20]. For all compounds, the basic requirements that must be fulfilled are: sufficient water solubility, low inherent toxicity, high boron content and high tumor selectivity. Water solubility can be increased by using charged compounds [28] or introducing hydrophilic moieties [29,30]. Tumor-selectivity can be achieved by using essential biological nutrients, substrates like boronated saccharides or amino acids [19,20,31], or even tumor-selective complex compounds, like boron-containing antibodies [20,29,30,[32][33][34][35][36]. A variety of different boroncontaining bioconjugates are known, including nucleosides [16], carbohydrates [37,38], amino acids [39][40][41] and peptides [29,30,42,43]. One main prerequisite of BNCT especially plays an essential role in this treatment, namely the selective accumulation of sufficient amounts of 10 B-containing compounds in cancerous tissues, so that the therapeutically active particles destroy only the malignant cells without destroying healthy tissue. An effective treatment requires boron concentrations of 10-30 µg 10 B/g tumor, or 10 9 10 Batoms/cancer cell [7,10]. One approach focuses on compounds with a very large boron content [29,30,42,44,45], another on the use of very selective BNCT agents over a longer period, taking advantage of specific shuttle systems that facilitate accumulation of the compound in the cells by internalization processes [17,29,30,33,46]. We pursued a combination of both strategies by combining tumor-selective small peptides, such as highly selective G protein-coupled receptor agonists, as biomolecules with meta-carborane derivatives to increase the boron load [29,30,43,47,48]. However, very high carborane loading (more than two carboranes attached to a peptide including 36 amino acids) results in loss of solubility or aggregation in aqueous media and, therefore, decreased potency and higher EC 50 values [29,30]. Carbohydrate moieties, such as galactosyl groups, can be employed to compensate the hydrophobic character of carborane clusters (up to eight modified carboranes attached to the same peptide comprising 36 amino acids).
Here, we report the development of small molecules representing potential boronrich coupling partners for tumor-selective molecules based on a modular strategy [47][48][49] combining readily available starting materials, like meta-carborane, α-D-galactopyranose and glycine or ethylenediamine derivatives (compounds 5 and 6 in Scheme 1). Compounds bearing a primary amine or carboxylic acid group represent potentially universal coupling partners for biomolecules [48]; representative coupling experiments are also included here to demonstrate the generalizability of this approach. In this regard, the synthesis of bifunctional anticancer drugs is of special interest. Several examples are known where drugs are used as theranostic compounds [50,51] or exhibit dual effects [52,53]. For example, derivatives of 7-amino-4-methylcoumarin are known for their anticancer activity [54]. Thus, combination with a carborane derivative can lead to drugs that possess anti-cancer properties and the ability to capture thermal neutrons for applications in BNCT. Furthermore, folic acid and its derivatives, which are already used as tumor-selective synthons for applications in BNCT [15,32,55], can act as diagnostic probes for some solid cancer types when combined with imaging agents [56]. Thus, conjugates of folic acid with carborane derivatives and an imaging agent using both carboxylic groups of folic acid could be used to generate highly selective BNCT agents with excellent imaging properties.
Beside the desired product 2, the disubstituted compound tert-butyl-(2-{[bis(1,2:3,4di-O-isopropylidene-6-deoxy-α-D-galactopyranos-6-yl)]amino}ethyl)-carbamate (2 ) was isolated with an 8% yield (see the Supplementary Materials). This product was formed due to the increased nucleophilicity of the secondary amine in 2 in comparison to a primary amine in the starting material; the amount corresponded to the small excess of the ethylenediamine derivative employed here.
The new compounds 1-6 were fully characterized by NMR spectroscopy (numbering scheme given in Figure S1), mass spectrometry and infrared spectroscopy and showed a purity of at least 95%. Furthermore, we were able to demonstrate that the same procedure can also be applied for the synthesis of the corresponding ortho-carborane derivative With the glycine derivative 5 and ethylenediamine derivative 6 in hand, exemplary coupling reactions with two selected biomolecules that play roles in cancer treatment were conducted. 7-Amino-4-methylcoumarin ( Figure 1, left), already employed in the preparation of polyfunctional cancer therapeutics based on cisplatin derivatives [62] but not for the preparation of potential BNCT agents [63][64][65], was selected as a coupling partner for 5. As coupling partner for amine 6, folic acid ( Figure 1, right) was chosen as this biomolecule has already been employed in the synthesis of potential BNCT agents [15,32,46,55,66]. Some cancer types overexpress folate receptors on their cell membrane surfaces, which can be used for selective uptake of the final BNCT agent in the respective cancer cells, increasing the efficacy of the therapy [32,67]. The new compounds 1-6 were fully characterized by NMR spectroscopy (numbering scheme given in Figure S1), mass spectrometry and infrared spectroscopy and showed a purity of at least 95%. Furthermore, we were able to demonstrate that the same procedure can also be applied for the synthesis of the corresponding ortho-carborane derivative N 1 -[(1,2-dicarba-closo-dodecaboran-1-yl)methyl]-N 1 -(1,2:3,4-di-O-isopropylidene-6-deoxy-α-D-galactopyranos-6-yl)ethane-1,2-diamine (see the supplementary materials for further details).
With the glycine derivative 5 and ethylenediamine derivative 6 in hand, exemplary coupling reactions with two selected biomolecules that play roles in cancer treatment were conducted. 7-Amino-4-methylcoumarin ( Figure 1, left), already employed in the preparation of polyfunctional cancer therapeutics based on cisplatin derivatives [62] but not for the preparation of potential BNCT agents [63][64][65], was selected as a coupling partner for 5. As coupling partner for amine 6, folic acid ( Figure 1, right) was chosen as this biomolecule has already been employed in the synthesis of potential BNCT agents [15,32,46,55,66]. Some cancer types overexpress folate receptors on their cell membrane surfaces, which can be used for selective uptake of the final BNCT agent in the respective cancer cells, increasing the efficacy of the therapy [32,67]. The strategy for coupling glycine derivative 5 with the weakly nucleophilic coumarin derivative followed a procedure described by Quéléver and co-workers using phosphoryl chloride and pyridine [68] and resulted in (7), albeit in low yield (27%) (Scheme 2). Attempts to use less harsh conditions (N,N'-dicyclohexylcarbodiimide (DCC) and N-hydroxysuccinimide (NHS)) for this coupling reaction were not successful [69]. Using different coupling reagents (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDCI) and hydroxybenzotriazol (HOBt) with N,N-diisopropylethylamine (DIPEA) as base [69]) yielded the desired product; however, it was in a very low yield of only 18%, indicating that this method is inferior to the phosphoryl chloride approach and the low reactivity of the coumarin derivative is the main issue. Compound 7 was fully characterized by NMR spectroscopy, mass spectrometry and infrared spectroscopy, proving the successful synthesis (with at least 95% purity) of this bioconjugate as a proof of principle in this approach. The strategy for coupling glycine derivative 5 with the weakly nucleophilic coumarin derivative followed a procedure described by Quéléver and co-workers using phosphoryl chloride and pyridine [68] and resulted in (7), albeit in low yield (27%) (Scheme 2). Attempts to use less harsh conditions (N,N'-dicyclohexylcarbodiimide (DCC) and N-hydroxysuccini mide (NHS)) for this coupling reaction were not successful [69]. Using different coupling reagents (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDCI) and hydroxybenzotriazol (HOBt) with N,N-diisopropylethylamine (DIPEA) as base [69]) yielded the desired product; however, it was in a very low yield of only 18%, indicating that this method is inferior to the phosphoryl chloride approach and the low reactivity of the coumarin derivative is the main issue. Compound 7 was fully characterized by NMR spectroscopy, mass spectrometry and infrared spectroscopy, proving the successful synthesis (with at least 95% purity) of this bioconjugate as a proof of principle in this approach. The coupling reaction between folic acid and the primary amine 6 turned out to be more complicated due to the presence of two unprotected carboxylic acid groups in the former. Similar reactions have been reported where no additional protecting group was used for the secondary carboxylic acid group [55,66,70], as the primary COOH group exhibits a higher reactivity and, therefore, is more prone to undergo coupling reactions. However, here, the reaction of 6 with folic acid, following the procedure described by Trindade and co-workers, using DCC and NHS as activation reagents [70], gave both the desired monosubstituted species, namely , also verified by high-resolution mass spectrometry (m/z 883.5151 for 8 and m/z 1323.8829 for 9). Obviously, in this case, the protocol from the literature [70] for conjugate formation with folic acid was not applicable, as we got a mixture of 8 and 9 in an unknown ratio. However, the obtained disubstituted folic acid derivative 9 has a high boron contents which could be useful for applications as a BNCT agent. The coupling reaction between folic acid and the primary amine 6 turned out to be more complicated due to the presence of two unprotected carboxylic acid groups in the former. Similar reactions have been reported where no additional protecting group was used for the secondary carboxylic acid group [55,66,70], as the primary COOH group exhibits a higher reactivity and, therefore, is more prone to undergo coupling reactions. However, here, the reaction of 6 with folic acid, following the procedure described by Trindade and coworkers, using DCC and NHS as activation reagents [70], gave both the desired monosubstituted species, namely , also verified by high-resolution mass spectrometry (m/z 883.5151 for 8 and m/z 1323.8829 for 9). Obviously, in this case, the protocol from the literature [70] for conjugate formation with folic acid was not applicable, as we got a mixture of 8 and 9 in an unknown ratio. However, the obtained disubstituted folic acid derivative 9 has a high boron contents which could be useful for applications as a BNCT agent. Scheme 3. (a) 1.00 eq. folic acid, 1.00 eq. DCC, 1.00 eq. NHS, 1.10 eq. DIPEA, dimethylformamide, no yield determined.

Materials and Methods
All reactions were carried out under nitrogen atmosphere using Schlenk techniques, if not reported otherwise. Anhydrous diethyl ether and DCM were obtained with an MBRAUN solvent purification system MB SPS-800 (M. Braun Inertgas-Systeme GmbH, Garching, Germany). MeCN and 2,4,6-collidine were dried over calcium hydride and distilled prior to use. Anhydrous tetrahydrofuran was dried over potassium and distilled prior to use. All solvents were stored over a molecular sieve (3 Å) under nitrogen atmosphere. 1,2-Dicarba-closo-dodecaborane(12) and 1,7-dicarba-closo-dodecaborane(12) are commercially [60] and 7-amino-4-methylcoumarin [72] were synthesized according to respective protocols from the literature. All other chemicals were commercially available and were used as received.
Thin-layer chromatography (TLC) with silica gel 60 F 254 on glass, available from Merck KGaA (Darmstadt, Germany), or ALUGRAM ® XTRA SIL G/UV 254 from Macherey-Nagel GmbH & Co. KG (Düren, Germany) on aluminum foil were used for monitoring the reactions. Carborane-containing spots were visualized with a 5% solution of PdCl 2 in methanol. Non-carborane-containing spots were visualized with a basic potassium permanganate solution. For chromatography, silica gel (60 Å) with a particle diameter in the range of 0.035 to 0.070 mm was used. Prior to column chromatography, raw products were adsorbed on Celite ® S from Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany).
NMR measurements were carried out on a Bruker AVANCE III HD spectrometer (Bruker Corporation, Billerica, MA, United States of America) with an Ascend TM 400 magnet (Bruker Corporation, Billerica, MA, United States of America) at room temperature. Tetramethylsilane was used as internal standard for 1 H-and 13 C{ 1 H}-NMR spectra; 11 Band 11 B{ 1 H}-NMR spectra were referenced to the Ξ scale [73]. NMR spectra were recorded at the following frequencies: 1 H: 400. 16 MHz, 13 C: 100.63 MHz, 11 B: 128.38 MHz. All chemical shifts are reported in parts per million (ppm). Assignment of the 1 H and 13 C signals was based on 2D-NMR spectra (H,H-COSY, H,C-HSQC, H,C-HMQC and H,C-HMBC). NMR data were interpreted with MestReNova [74]. NMR signals that appeared as broad overlapping signals with the shape of a multiplet or singlet in either 1 H-, 11 B{ 1 H}-or 11 B-NMR spectra were described as "br" (broad). The numbering scheme of the compounds for assignment of NMR signals is given in the Supplementary Materials (see Figure S1).
IR data were obtained with a PerkinElmer FT-IR spectrometer Spectrum 2000 (Perkin Elmer, Inc., Waltham, MA, United States of America) as KBr pellets and with a Thermo Scientific Nicolet iS5 with an ATR unit (Thermo Fisher Scientific, Waltham, MA, United States of America) in the range from 4000 to 400 cm −1 .
High-resolution electrospray ionization mass spectrometry (ESI-HRMS) was performed with an ESI ESQUIRE 3000 PLUS spectrometer (Bruker Corporation, Billerica, MA, United States of America) with an IonTrap-analyzer from Bruker Daltonics or on a MicroTOF spectrometer from Bruker Daltonics with a ToF analyzer in negative or positive mode. As solvents for the measurements, DCM, MeCN, methanol or mixtures of these solvents were used. Interpretation of the spectra was carried out using MestReNova [74].
High-resolution electrospray ionization mass spectrometry (ESI-HRMS formed with an ESI ESQUIRE 3000 PLUS spectrometer (Bruker Corporation MA, United States of America) with an IonTrap-analyzer from Bruker Dalton MicroTOF spectrometer from Bruker Daltonics with a ToF analyzer in negative mode. As solvents for the measurements, DCM, MeCN, methanol or mixtur solvents were used. Interpretation of the spectra was carried out using MestRe
Method B. A Schlenk flask was charged with tert-butyl-N-[(1,7-dicarba-closo-dodecabor an-1-yl)methyl]-N-(1,2:3,4-di-O-isopropylidene-6-deoxy-α-D-galacatopyranos-6-yl)-glycina te (3) (0.50 g, 0.94 mmol, 1.00 eq.). Then, 4.0 mL anhydrous TFA (52.2 mmol, 55.6 eq.) were added and the resulting solution was stirred at room temperature for 3 h. Afterwards, about 4 mL DCM were added and both, TFA and DCM, were removed under reduced pressure. This process was repeated three more times with DCM as an entrainer to remove remaining TFA. Then, 3 mL concentrated NaHCO 3 solution were added to the obtained crude product while stirring and the mixture was subsequently sonicated for 15 min, resulting in a thick, cloudy white suspension with a brownish oil-like layer on top. Upon addition of 2 mL DCM, the oil-like layer was dissolved and gas evolution was observed. Once the gas evolution had almost stopped, the solution was stirred for five more minutes. Subsequently, the aqueous layer was removed using a syringe and extracted two times with 2 mL ethyl acetate each. The combined organic layers were washed twice with 2 mL of water. Afterwards, the combined organic layers were dried over MgSO 4 and filtered. Finally, the solvent was removed under reduced pressure, affording compound 5 as a yellowish foamy solid in 65% yield (290 mg, 0.612 mmol). 1  rted in parts per million (ppm). Assignment of the H and C signals NMR spectra (H,H-COSY, H,C-HSQC, H,C-HMQC and H,C-HMBC). terpreted with MestReNova [74]. NMR signals that appeared as broad ls with the shape of a multiplet or singlet in either 1 H-, 11 B{ 1 H}-or 11 Bdescribed as "br" (broad). The numbering scheme of the compounds MR signals is given in the supplementary materials (see Figure S1). btained with a PerkinElmer FT-IR spectrometer Spectrum 2000 (Perkin am, MA, United States of America) as KBr pellets and with a Thermo S5 with an ATR unit (Thermo Fisher Scientific, Waltham, MA, United in the range from 4000 to 400 cm −1 . on electrospray ionization mass spectrometry (ESI-HRMS) was per-SI ESQUIRE 3000 PLUS spectrometer (Bruker Corporation, Billerica, of America) with an IonTrap-analyzer from Bruker Daltonics or on a eter from Bruker Daltonics with a ToF analyzer in negative or positive for the measurements, DCM, MeCN, methanol or mixtures of these . Interpretation of the spectra was carried out using MestReNova [74].   (4) and 2.00 mL (26.0 mmol, 48.2 eq.) TFA were added. The reaction mixture was stirred for 3 h at room temperature. The reaction was stopped by adding 4 mL DCM with subsequent evaporation of all volatile components under reduced pressure. This procedure was repeated three times. The resulting crude product was further purified by adding 4 mL saturated NaHCO 3 solution and sonication for about 15 min. Again, 3 mL DCM were added with stirring, under observation of gas evolution, and after 5 min the resulting layers were separated. The aqueous layer was extracted with 3 mL ethyl acetate. The combined organic layers were washed twice with 2 mL distilled water each. The organic layer was dried over MgSO 4 , the drying agents were filtered off and the solvent was removed under reduced pressure. Compound 6 (0.25 g, 0.54 mmol, quant., R f = 0.03, nhexane/ethyl acetate, 5:1, (v/v)) was isolated as a colorless foamy solid. 1 (9): A 100 mL Schlenk flask was charged with 0.24 g (0.55 mmol, 1.00 eq.) folic acid and 20 mL dimethylformamide were added. The mixture was sonicated for 15 min and, subsequently, warmed to 37 • C for 15 min until a clear solution was obtained. To this mixture, 0.11 g (0.55 mmol, 1.00 eq.) DCC and 0.06 g (0.55 mmol, 1.00 eq.) NHS were added. The reaction mixture was stirred for 16 h at room temperature. Afterwards, 0.25 g (0.55 mmol, 1.00 eq.) N 1 -[(1,7-dicarba-closo-dodecaborane-1-yl)methyl]-N 1 -(1,2:3,4-di-Oisopropylidene-6-deoxy-α-D-galactopyranos-6-yl)ethane-1,2-diamine (6), dissolved in 7 mL dimethylformamide, were added to this solution and the mixture was stirred overnight at room temperature. The resulting suspension was filtered under inert conditions. Subsequently, 0.10 mL (0.61 mmol, 1.10 eq.) N,N-diisopropylethylamine were added and the mixture was stirred overnight at room temperature. The reaction was stopped by adding 50 mL ice-cold diethyl ether. Simultaneously, the mixture was cooled in an ice bath. Completion of the precipitation was achieved by storage for one additional night at −20 • C. The resulting precipitate was filtered off and washed with 10 mL ice-cold diethyl ether. The precipitate was suspended in 5 mL diethyl ether and subsequently sonicated for 15 min. Afterwards, the raw product was filtered again. The orange-red solid was washed with 5 mL diethyl ether and the previously described procedure was repeated one more time. Afterwards, the precipitate was dried in vacuo. It was not possible to isolate the desired compound 8, but mass spectrometry revealed the presence of   [74]. NMR signals that appeared as overlapping signals with the shape of a multiplet or singlet in either 1 H-, 11 B{ 1 H}-o NMR spectra were described as "br" (broad). The numbering scheme of the compo for assignment of NMR signals is given in the supplementary materials (see Figure  IR data were obtained with a PerkinElmer FT-IR spectrometer Spectrum 2000 (P Elmer, Inc., Waltham, MA, United States of America) as KBr pellets and with a Th Scientific Nicolet iS5 with an ATR unit (Thermo Fisher Scientific, Waltham, MA, U States of America) in the range from 4000 to 400 cm −1 .
High-resolution electrospray ionization mass spectrometry (ESI-HRMS) was formed with an ESI ESQUIRE 3000 PLUS spectrometer (Bruker Corporation, Bil MA, United States of America) with an IonTrap-analyzer from Bruker Daltonics o MicroTOF spectrometer from Bruker Daltonics with a ToF analyzer in negative or po mode. As solvents for the measurements, DCM, MeCN, methanol or mixtures of solvents were used. Interpretation of the spectra was carried out using MestReNova

Conclusions
In this work we reported the successful design of a novel modular, small-moleculebased approach to synthesizing boron-rich compounds bearing a carboxylic acid group or a primary amine group as potential coupling partners for suitable tumor-selective biomolecules. As proof of concept, conjugates with 7-amino-4-methylcoumarin and folic acid were obtained. While the present work focused on the development of a synthetic protocol, the next steps will include the deprotection of the respective galactopyranosyl protecting groups under acidic aqueous conditions [61] followed by biological investigations.
Supplementary Materials: Supplementary information is available online, including the numbering scheme of the isolated compounds 1-7, NMR spectra of compounds 5, 6 and 7 and mass spectra of 8 and 9, additional synthetic procedures and analytical data for 2 , ESI-3, ESI-3 , ESI-4 and 1-(trifluoromethanesulfonylmethyl)-1,7-dicarba-closo-dodecaborane (12), crystallographic information for compound ESI-3 , information about the exploration of the optimization for the synthesis of 3 and the deprotection protocol for 4, and the extension of the synthetic protocol to ortho-carborane derivatives.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.