Synthesis of an Azide- and Tetrazine-Functionalized [60]Fullerene and Its Controlled Decoration with Biomolecules

Bingel cyclopropanation between Buckminster fullerene and a heteroarmed malonate was utilized to produce a hexakis-functionalized C60 core, with azide and tetrazine units. This orthogonally bifunctional C60 scaffold can be selectively one-pot functionalized by two pericyclic click reactions, that is, inverse electron-demand Diels–Alder and azide–alkyne cycloaddition, which with appropriate ligands (monosaccharides, a peptide and oligonucleotides tested) allows one to control the assembly of heteroantennary bioconjugates.


■ INTRODUCTION
Since the discovery of Buckminster fullerene ( [60]fullerene, by Kroto et al.), 1 scientists have invested significant effort by investigating potential applications of this fascinating carbon allotrope. 2 Its synthetic availability, controlled functionalization techniques, and efficient conjugation chemistries have enabled the preparation of sophisticated C 60 conjugates, which have had a significant impact in material and biomedical sciences. 3,4 [60]Fullerene as such shows interesting physical and biological properties, and the spherical structure that allows radially symmetric and dense functionalization makes it an excellent scaffold to probe multivalent biomolecular interactions. For example, C 60 -based glycoballs have received marked interest as potential lectin binders. 5 More recently, 12-armed fullerene has been used for the assembly of molecular spherical nucleic acids (SNAs), 6−8 which is an attractive delivery and formulation option for therapeutic oligonucleotides. In most of the applications above, one type of ligand (e.g., sugars and oligonucleotides) is multiplied on an appropriately functionalized C 60 hexakis adduct. 9−13 However, some biomolecular applications would need a combination of different biomolecules (i.e., heteroantennary C 60 bioconjugates). For example, a drug delivery vehicle may need a system that allows an orthogonal loading of tissue-specific and cellpenetrating ligands and the drug payloads. Appropriate heterovalency is also needed to introduce reporter groups selectively to C 60 conjugates or to integrate them with other functionalities. A controlled mono Bingel cyclopropanation 14−16 of [60]fullerene, followed by full-decoration with the same reaction and using two different malonates, has enabled the synthesis of 1:5-, 1:11-, and even 1:1:10heterosubstituted C 60 scaffolds, which have been used for an orthogonal ligation, for example, via alkyne−azide and thiol− ene click reactions. 17−25 Recently, stereodefined C 60 [3:3] hexa-adducts were prepared by two subsequent click reactions. 26 For the controlled assembly, a C 60 tris-adduct was first regioselectively prepared using a macrocyclic malonate, which was then exposed to a Bingel cyclopropanation with another malonate.

■ RESULTS AND DISCUSSION
A Bingel cyclopropanation between [60]fullerene and heteroarm malonate 5 was used for the synthesis of 1 (Scheme 1). For the synthesis of malonate 5, one would use Meldrum's acid and a stepwise reaction with 3 and 4. However, the described one step-approach was chosen, as it gave 5 in an acceptable yield (21%) with a mixture of readily availabale alcohols (Scheme S1). A Bingel cyclopropanation between C 60 and 5 under standard conditions, 14−16 using CBr 4 as a bromination agent and 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) as a strong organic base in o-dichlorobenezene for 3 d at rt under argon, gave the hexakis-substituted C 60 core 1. The crude product mixture was purified first by silica gel column chromatography (in 24% yield) and then further by reversed-phase (RP) high-performance liquid chromatography (HPLC) to give homogenized 1 in 5% overall yield. The authenticity of 1 was verified by NMR spectroscopy and mass spectrometry (MS) (electrospray ionization−time-of-flight (ESI-TOF)) ( Figures S5−S8). It may be worth of mentioning that, while the cyclopropanated C 60 moiety of 1 is a wellorganized structure, with pyritohedral symmetry, 1 is obtained as a stereoisomeric mixture (in fact 2 5 = 32 steroisomers) due to the heteroarm malonate. The corresponding stereodefined structure would increase the biological and supramolecular value of these C 60 -derivatives. However, prior to a marked synthetic effort, needed to obtain the corresponding stereodefined C 60 derivative, 26 we wanted to evaluate first in this article how the tetrazine-azide combination on 1 works for the catalyst free assembly of heteroarm bioconjugates.
A set of TCO-and BCN-modified biomolecule ligands (cf. the tool box in Scheme 2) was used for the decoration of 1. The ligands were prepared by a carbamate coupling from commercially available TCO and BCN precursors, an aminomodified peptide, amino-modified oligonucleotides, and pnitrophenyl carbonate-modified sugars 27 (cf. Supporting Information, Schemes S2 and S3).
To evaluate the applicability of core 1 in the preparation of heteroarm bioconjugates, we first synthesized C 60 -glycoconjugate C4. 1 (0.33 μmol) was dissolved in DMSO (100 μL) and exposed to iEDDA with TCO-Gal (9 equiv in 100 μL of DMSO added) to yield intermediate glycoconjugate C1, and then BCN-Glu (9 equiv in 100 μL of DMSO) was added to the same reaction mixture. The completion of the successive click reactions (both overnight reactions) was verified by RP HPLC (Figure 1 and Scheme S4), and the authenticity of the intermediate conjugate C1 and of the end product C4 was verified by MS (ESI-TOF) (Figures S15 and S16). The isolated C4 was characterized also by NMR spectroscopy ( 1 H, COSY, HSQC, HMBC), in which the correct 1:1 ratio of the galactose and glucose units could be clearly seen (Figures S17−S20 and Table S1). Because of the small-scale synthesis, the molar quantity of the obtained C4 was extracted from the 1 H NMR spectra by comparing the intensity of 1 H signals to an internal standard (a known quantity of acetonitrile used). Accordingly, C4 was obtained in 66% isolated yield. The other heteroarm glycoconjugates C5 and C6 and the peptideglycoconjugate C7 were next assembled in a similar manner by using 0.1 μmol of 1 and adjusting the excesses of the reagents and solvent volumes accordingly. As seen in the RP HPLC profiles (Figure 1) of the crude product mixtures, the selective assembly on the C 60 core (1) worked well in each case to yield the desired heteroarm conjugates C5−C7 in 56, 69, and 61% isolated yields (based on the absorbance at λ = 260 nm of the products compared to that of a known concentration of C4, ε = 120 × 10 3 L mol −1 cm −1 ). It may be worth mentioning that the idea of these small-scale trials was to further evaluate the proof of concept, in which the applicability of the scaffold 1 for the successive one-pot iEDDA and SPAAC conjugations was validated. Because of the difficulties related to the steroisomeric mixture (and the small scale), we did not invest in an effort for a complete NMR characterization, but the authenticity of the intermediate conjugates (C2 and C3) and of the end products (C5−C7) was verified by MS (ESI-TOF) (Figures S15, S16, and S21) only.
We recently communicated that multiarm C 60 -branching units (e.g., 1) may be contaminated by structurally similar and hardly identified derivatives. 8 Therefore, prior to the assembly of C 60 -based macromolecules, in which plausible errors may remain hidden, the homogeneity and applicability of the cores should be evaluated with small molecular ligands. After a successful synthesis of the glycoconjugates (C4−C7), we validated the applicability of 1 for the assembly of a molecularly defined SNA, in which both strands of the double helices were covalently bound to the C 60 core (C9). Relatively short oligonucleotides (TCO-ON1 and BCN-ON2) were used for the assembly. 1 (5 nmol) was exposed to iEDDA with TCO-ON1 (45 nmol, 9 equiv) in DMSO (45 μL). The mixture was stirred overnight at room temperature and purified by RP HPLC (Scheme S4) to give the hexa-arm intermediate conjugate (C8) in 50% isolated yield (according to the UV absrobance at λ = 260 nm). The authenticity of C8 was verified by MS (ESI-TOF) spectroscopy ( Figure S22). Because of the complementarity of the ON1 and ON2 strands, we did not try a one-pot assembly here. An aliquot (1.0 nmol) of purified C8 was exposed to SPAAC with BCN-ON2 (9.0 nmol) in H 2 O (20 μL). The mixture was mixed overnight at room temperature and purified by an RP HPLC (Figure 1), which gave the homogenized SNA C9 in 57% isolated yield (according to the UV absorbance at λ = 260 nm). As usual with SNAs, the MS-spetroscopic characterization of C9 proved complex. An aliquot of C9 was then introduced to Size Exclusion Chromatography equipped with a Multiple Angle Light Scattering detector (SEC-MALS), which showed a wellbehaving and pure (95.7% mass fraction of the total sample mass) macromolecule with a molecular weight of 41.4 ± 1.4 kDa (expected: 41.4 kDa) ( Figure S23). We also evaluated the homogeneity of C8, C9, and C8 in the presence of complementary strand (6 equiv of ON2, that is, hybridization-mediated SNA C10) by native PolyAcrylamine Gel Electrophoresis (PAGE). As seen in the electrophoregram (Figure 1, cf. also Figure S25), each of the SNAs showed a distinct band. C9 eluted markedly slower compared to C10. This is a surprising behavior, as the only difference between C9 and C10 is the covalent link of ON2 to the C 60 core. The thermal stability of the double helices on the SNAs C9 and Scheme 1. Synthesis of the Bifunctional C 60 Core (1) ACS Omega http://pubs.acs.org/journal/acsodf Article C10 was next evaluated by a UV-melting profile (T m ) analysis. C10 resulted in an 8°C decrease in the T m value, when compared to the corresponding free duplex (ON1 + ON2) ( Figure 2). This decreased duplex stability is consistent with the previous findings of SNAs, 8,28−31 caused by an electrostatic and steric repulsion between the densely packed oligonucleotides. Interestingly, C9 showed no melting at all in the measured temperature range of 10−90°C. This indicates very stable double helices on C9. One may wonder whether this is a result of a negligible hyperchromic effect, even if an unwinding of the strands occurs on C9. (Note: a relative hyperchromic effect is considered in the melting profiles, Figure 2). Circular dichroism (CD) measurements with C9 and C10 ( Figure 2) were then performed to verify changes of helicity upon the temperature ramp. CD profiles of C10 represent a typical Btype double helix. The characteristic minimum at 250 nm gradually disappears upon heating, which indicates a thermal denaturation of the double helices. Typical B-type CD profiles may be observed also on C9, but what makes these data fascinating is that there is no marked change between the profiles upon heating. In the profile even at 90°C (the bold red line) a deep minimum at 250 nm can be seen, demonstrating that B-type double helices exist there. Very stable cyclic double helices are known structures, 32 but C9 seems to be an interesting example of dendritic nucleic acids, in which neighboring hairpin-type double helices stabilize each other via steric and electrostatic repulsion, in this manner preventing an unwinding of the strands. Polydisperse gold nanoparticle SNAs with covalently bound RNA double helices have been reported, 28 but no similar stability phenomenom has been studied in detail.
We also analyzed an aliquot of C9 on a polyethylenimine (PEI)-coated mica using atomic force microscopy (AFM). Particles of ca. 10 nm height, representing the monomeric C9 Scheme 2. Synthesis of Hetero-Antennary Bioconjugates using the Bifunctional C 60 Core (1) ACS Omega http://pubs.acs.org/journal/acsodf Article (Figure 3), were observed. We mention that the sample contained also larger but rather control-sized particles (ca. 25 nm height) that indicated aggregates of C9 on the PEI-coated mica ( Figure S24). The SNAs like C9 may find interesting applications as well-organized constructs in medicinal and supramolecular chemistry. Further studies considering a deeper understanding of the stability requirements (the effect of the sequence and its length, hydrogen-bonding stability, Dicermediated cleavage 28 of therapeutically relevant siRNAs on these SNAs, etc.) are currently under way in our laboratory.

■ CONCLUSION
In conclusion, a bifunctional C 60 core (1) has been described, which can be selectively functionalized by two subsequent pericyclic click reactions (iEDDA and SPAAC) in catalyst-free conditions. The applicability of the core in a one-pot assembly has been demonstrated for the synthesis of heteroantennary glycoballs (and one glyco-peptide C 60 -conjugate 33,34 ) C4−C7, which were obtained in 56−69% isolated yields. Furthermore, a novel type SNA (C9) with extraordinary stable covalently bound double helices, verified by UV-and CD-melting profile    ■ EXPERIMENTAL SECTION General Remarks. All reactions involving air-or moisturesensitive conditions were routinely performed and under an inert atmosphere. All reagents from commercial suppliers were used without further purification. The NMR spectra were recorded using 500 and 600 MHz instruments. The chemical shifts in the 1 H and 13 C NMR spectra are given in parts per million (ppm) from the residual signal of the deuterated solvents.
SEC-MALS. For the SEC-MALS analysis of C9 an Agilent Technologies 1260 Infinity II HPLC system (sampler, pump, and UV/vis detector) equipped with a Wyatt Technologies miniDAWN light-scattering detector and a Wyatt Technologies Optilab refractive index detector was used. An Agilent AdvanceBio SEC 300 Å 2.7 μm 4.6 × 300 mm column, 150 mM sodium phosphate pH 7.0 as mobile phase eluting at rate of 0.2 mL min −1 and run time of 20 min were used. Four microliters of a sample of C9 (1 mg mL −1 in Milli-Q water) was loaded onto the pre-equilibrated column. Detector signals were aligned with a bovine serum albumin (BSA) standard, which was analyzed prior to the C9 sample. The RI and MALS signals were used for the molecular weight (MW) calculations using an average refractive index increment (dn/dc) of 0.1703 mL/g. PAGE Analysis of SNAs. A native 6% Tris base, boric acid, ethylenediaminetetraacetic acid (EDTA), and acrylamide (TBE) gel were used to check SNAs purity. A pre-cast gel cover 10 cm × 10 cm in size (Thermo Fisher Scientific) was fixed into a vertical electrophoresis chamber, and 10% Trisborate-EDTA running buffer (VWR Life Science) was filled into an electrophoresis chamber. SNA samples (C8−C10, 5 μL of 0.3 μM SNAs mixed with 5 μL of 6× TriTrack DNA Loading Dye) and 5 μL of Gene Ruler Ultra Low Range DNA ladder 10−300 bp (Thermo Scientific) were loaded and electrophoresed at 150 V constant (45 mA) for ∼35 min. After the completion of the electrophoresis, the gel was removed from the electrophoresis chamber and stained by SYBRTM Gold Nucleic Acid Stain (Thermo Fisher Scientific) for 1 h and imaged under G-Box camera (Syngene).
Running Buffer Preparation. 100 mL of (10X concentrated solution of 0.9 M Tris, 0.9 M borate, and 0.02 M EDTA, 8.3 pH in distilled, deionized water) was dissolved in 900 mL of distilled, deionized water. Sample preparations: 5 μL of (6 × TriTrack DNA Loading Dye) and 5 μL of SNAs in water (0.3 μM based on SNAs concentration). Staining solution preparation: 5 μL of SYBRTM Gold Nucleic Acid Stain was dissolved in 50 mL of running buffer.
UV Melting Profile (T m ) Experiments. The melting curves (absorbance vs temperature) were measured at 260 nm on a UV−vis spectrometer equipped with a multiple cell holder and a Peltier temperature controller. The temperature was changed at a rate of 0.5°C min −1 between 10 and 80°C (ON1 + ON2-duplex and C10) and between 10 and 90°C (C9). The measurements were performed in 10 mmol L −1 sodium cacodylate (pH 7.0) with 0.1 mol L −1 NaCl and using 1.0 μmol L −1 ON1 + ON2 and 2.0 μmol L −1 C9 and C10. The T m values were determined as the maximum of the first derivate of the melting curve. The relative increased absorbance (hyperchromicity) of the C9 and C10 samples was compared to that of ON1 + ON2.
CD Spectroscopic Analysis. CD spectra of C9 and C10 were measured on an Applied Photophysics Chirascan spectrometer. The same mixtures used for the UV melting profile analysis were used. A quartz cell of diameter 10 mm was used for the measurements. The sample temperatures were changed from 20 to 90°C at a rate of 1°C min −1 .
AFM Images. 2.5 mg/mL PEI solution was deposited on cleaved mica. After 10 min of adsorption, the mica was rinsed with water and dried with nitrogen. A mixture of 10 nM C9 in water (10 μL) was deposited on the PEI-coated mica. After 10 min of adsorption, 50 μL of water was added, and the particles were scanned in water. The AFM images were obtained using a MultiMode 8 atomic force microscope (Bruker) with a NanoScope V controller, working in tapping mode with a ScanAsyst-Fluid+ probe (Bruker).
Synthesis of Conjugates C5−C7. The same procedure described for C4 was used to obtain two other C 60 -glyco conjugates C5 and C6 and C 60 -glyco-peptide-conjugate C7. 0.1 μmol of 1 was used for the assembly. The reagent excesses and solvent volumes were adjusted accordingly. The RP HPLC analysis of the individual reaction steps is described in Scheme S4. The authenticity of the intermediate products (C2 and C3, aliquots purified prior MS-characterization) and of the end products (C5−C7) was verified by MS (ESI-TOF) spectroscopy (Figures S15, S16, and S21). The isolated yields were extracted from the UV absorbance of the isolated products at λ = 260 nm. (The absorbance was compared to that of a known concentration of C4, ε = 120 × 10 3 L mol −1 cm −1 ; in the case of C7, the absorbance of the peptide moiety was considered.) Accordingly, C5, C6, and C7 were isolated in 56, 69, and 61% yields, respectively.
Synthesis of C8. TCO-ON1 (40 nmol in 40 μL of DMSO) was added to a mixture of 1 (5.0 nmol) in DMSO (5 μL) in a microcentrifuge tube. The mixture was mixed overnight at room temperature. Reaction mixture was introduced to an RP-HPLC (Scheme S4), and the product fractions were lyophilized to dryness. C8 was obtained in 50% isolated yield (based on UV absorbance at λ = 260 nm). The authenticity and homogeneity of C8 was verified by MS (ESI-TOF) ( Figure S22) and by PAGE (Figure 1).
Synthesis C9. BCN-ON2 (9.0 nmol in 10 μL of H 2 O) was added to a mixture of C8 (1.0 nmol) in H 2 O (10 μL) in a microcentrifuge tube. The mixture was mixed overnight at room temperature and introduced to an RP HPLC (Scheme S4), and the product fractions were lyophilized to dryness. C9 was obtained in 57% isolated yield (based on the UV absorbance at λ = 260 nm). The authenticity and homogeneity of C9 were verified by SEC-MALS ( Figure S23) and by PAGE (Figure 1).