Chelator-Accelerated One-Pot ‘Click’ Labeling of Small Molecule Tracers with 2-[18F]Fluoroethyl Azide

2-[18F]Fluoroethyl azide ([18F]FEA) can readily be obtained by nucleophilic substitution of 2-azidoethyl-4-toluenesulfonate with [18F]fluoride (half-life 110 min), and has become widely used as a reagent for ‘click’ labeling of PET tracers. However, distillation of [18F]FEA is typically required, which is time-consuming and unpractical for routine applications. In addition, copper(I)-catalyzed cycloaddition of [18F]FEA with non-activated alkynes, and with substrates containing labile functional groups, can be challenging. Herein, we report a highly efficient and practical ligand-accelerated one-pot/two-step method for ‘click’ labeling of small molecule tracers with [18F]FEA. The method exploits the ability of the copper(I) ligand bathophenanthrolinedisulfonate to accelerate the rate of the cycloaddition reaction. As a result, alkynes can be added directly to the crude reaction mixture containing [18F]FEA, and as cyclisation occurs almost immediately at room temperature, the reaction is tolerant to labile functional groups. The method was demonstrated by reacting [18F]FEA with a series of alkyne-functionalized 6-halopurines to give the corresponding triazoles in 55–76% analytical radiochemical yield.


Introduction
Labeling with 18 F is challenging due to the time constraints imposed by the short half-life (110 min), the need to manipulate minute amounts of [ 18 F]fluoride, and the need for remote automated synthesis of tracers for clinical use. [ 18 F]Fluoride can readily be produced with high specific activity (SA), and is by far the most common source of 18 F for tracer synthesis. Labeling is typically achieved by nucleophilic aliphatic substitution of suitable leaving groups, such as sulfonates, with [ 18 F]fluoride. Yet, 18 F-labeling of aliphatic groups is sensitive to the presence of hydrogen bond donors, as well as neighbouring group effects, and as a result tracer synthesis often requires the use of prosthetic labeling reagents. The copper(I)-catalyzed cycloaddition reaction of azides and alkynes to give 1,2,3-triazoles has recently emerged as a highly versatile method for conjugation of 18 F-labeled "click" reagents to small molecules and peptides [1][2][3][4]. Despite the broad substrate scope, the efficiency of the reaction is dependent on the nature of the "click" partners, and conjugation to substrates with low reactivity or labile functional groups can be challenging [5]. In addition, purification of the intermediate labeling reagent is often required prior to the cycloaddition reaction in order to limit the formation of side products.
Among the array of "click" reagents reported, 2-[ 18 F]fluoroethyl azide ([ 18 F]FEA) is attractive as its small size makes it particularly suited for labeling of small molecule tracers [2,5,[6][7][8]. However, purification can only be achieved by distillation, which is delicate, poorly compatible with automated synthesis modules, and also potentially hazardous as it can result in release of gaseous [ 18 F]FEA. Herein, we reported a convenient and highly efficient one-pot method for "click" labeling with [ 18 F]FEA. The method exploits the ability of the copper chelator bathophenanthrolinedisulfonate (BPDS, Figure 1) to accelerate the cycloaddition reaction, which overcomes the need to purify the intermediate [ 18 F]FEA, and enables conjugation to precursors with labile functional groups. The method was demonstrated by reacting [ 18 F]FEA in the form of a crude reaction mixture with a series of alkyne-functionalized 6-halopurines in the presence of copper(II)/ascorbate and BPDS to give the corresponding triazoles in 40-57% isolated radiochemical yield and with >99% radiochemical purity.

Radiochemistry
As formation of the N7-alkylated triazoles 6-8 under preparative conditions gave moderate yields (32-40%) with apparent decomposition of the corresponding precursors, we initially used nonradioactive FEA and the alkyne 2a to optimize the conditions for the cycloaddition reaction. Attempts to increase the reaction rate by the use of excess copper(II) sulfate and sodium ascorbate relative to the alkyne 2a failed to give the triazole 6 (15 min, rt), and instead resulted in formation of an unknown side product with a similar HPLC retention time to that of the target compound 6 [ Figure 2(a)]. The use of ascorbic acid to allow formation of copper(I) under acidic conditions was also unsuccessful, and resulted in complete consumption of the precursor 2a with formation of another unidentified side product [ Figure 2(b)]. To exclude the possibility that side product formation was caused by the in situ reduction of copper(II) sulfate, we investigated copper(I) chloride and triethylamine as an alternative catalytic system. However, we were unable to obtain the target triazole 6, and instead obtained product mixtures with similar HPLC profiles to that observed for the reaction of 2a and FEA in the presence of copper(II)/ascorbate [ Figure 2(a)].
Purines are known to form complexes with copper ions [12,13], and it is plausible that the combined interactions of the alkyne group and nitrogens in 2a with copper(I) results in formation of a complex that impairs subsequent cyclisation with FEA through unfavorable steric or electronic interactions. Instead of increasing the rate of the cycloaddition reaction, addition of excess copper(I) relative to the alkyne 2a may therefore favor hydrolysis of the precursor 2a by activation of the chlorine in the 6-position. To explore this possibility, we investigated the use of BPDS as an auxiliary  copper(I) chelator. The BPDS-copper(I) catalytic system was first identified from a fluorescent assay [14], and has since been used to accelerate triazole formation for reactions with peptides and substrates with high molecular weight for which the copper(II)/ascorbate system alone often provides unsatisfactory results [15,16]. Remarkably, in the presence of BPDS, copper(II) sulfate and sodium ascorbate, the cycloaddition of 2a with FEA occurred almost immediately at room temperature to give the target triazole 6 [ Figure 2(c)]. Encouraged by the superb efficiency of the cycloaddition reaction in the presence of the BPDS-copper(I) catalytic system, we explored the possibility of preparing the triazoles [ 18 6 in 60% analytical radiochemical yield (RCY) within 1 min at room temperature. Provided that the alkyne 2a was used in excess relative to copper(II) sulfate and BPDS, no other radioactive product was observed, and the precursor 2a remained largely intact after the reaction [ Figure 2(d)]. The apparent absence of non-radioactive triazoles in the product mixture was unexpected and suggests that the cycloaddition rate of FEA to alkynes far exceeds that for the corresponding 2-azidoethyl-4-toluenesulfonate precursor under these conditions. Comparable results were obtained for formation of triazoles [ 18 F]7-9, which were obtained in 55-76% analytical and 40-57% isolated (decay-corrected) radiochemical yields (Table 1). (57) b a Average analytical decay-corrected radiochemical yields ± standard deviation (n = 3). b Isolated RCY in brackets (n = 1). c Isolated yield obtained from a fully automated radiosynthesis.

Automated radiosynthesis of purine [ 18 F] 7
To evaluate the suitability of the one-pot method for routine tracer production we established an automated synthesis of the triazole-functionalized purine [ 18 F]7 using a HBIII module (Scintomics). The configuration of the synthesis module is illustrated in Figure 3.  In order to maximize the trapping efficiency of [ 18 F]7, we evaluated a number of solid phases, sizes of cartridges, and solvent systems. The highest RCY was obtained by dilution of the product mixture with water (30 mL) to give a total of 2% acetonitrile, followed by trapping of the crude product on a C-18 cartridge (Sep-Pak® Plus, Waters). The product was eluted with 25% acetonitrile in water (2 mL), the resulting solution was diluted with water (3 mL) and subsequently purified by HPLC (Chromolith® RP18-e using 9% acetonitrile with 0.1% formic acid as the eluent) to give [ 18 F]7 in 40% decay-corrected RCY. While the radiochemical purity was > 99% [ Figure 4(a)], the specific activity was still below 1 GBq/μmol. The use of a smaller cartridge (C-18 Sep-Pak® light, Waters) allowed release of the product in a lower solvent volume (20% ethanol in water, 1 mL), however the trapping was less efficient and led to loss of 15-20% of the product. The resulting solution was diluted with water (1 mL) and purified by HPLC as described above. The collected fraction was diluted with water to give a total concentration of 5% in acetonitrile, the product was trapped on C-18 Sep-Pak® light cartridge (Waters), the cartridge was washed with water (1 mL), and the product eluted with 20% ethanol in saline (0.8 mL). The resulting solution was diluted with saline for injection (2.4 mL) to give a total of 5% ethanol, and filtered on sterile filter (Millex-GV 0.22 µm, 13 mm, Millipore). When starting with 1-1.5 GBq [ 18 F]fluoride, the formulated product [ 18 F]7 was obtained in 9 ± 2% (n = 3) decay-corrected RCY after sterile filtration, with >99% radiochemical purity and a specific activity in the range of 5-7 GBq/μmol [ Figure 4  An inherent limitation of the one-pot method for cycloaddition of [ 18 F]FEA with alkynes is that the total contents of the product mixture exceed the typical capacity of semi-preparative HPLC columns, and hence cartridge-based purification is required prior to sample loading. The high polarity of [ 18 F]7 presented particular challenges in this respect as cartridge trapping and HPLC purification only was achievable when using aqueous solutions with very low concentrations of organic solvents. The poor retention of [ 18 F]7 on reverse-phase stationary phases necessitated extensive dilution of product solutions, which added to the overall synthesis time, and also made it difficult to obtain the target compound with high specific activity. However, such complications should not affect tracers with more favorable physiochemical properties, in which case a shorter synthesis time with improved RCY and higher specific activity is likely to be achieved. While the outcome of the cycloaddition reaction will depend on the nature of the alkyne precursor, and the conditions used, in our hands we observed no competing reactions that are likely to complicate tracer purification.
It should be noted that one-pot "click" labeling with [ 18 F]FEA have been reported previously [17], and that BPDS has been used for one-pot labeling with other "click" reagents [16]. However, in the absence of a chelator to accelerate the reaction rate, one-pot labeling with [ 18 F]FEA is limited to highly reactive alkyne precursors. For "click" reagents other than [ 18 F]FEA, which readily can be purified with solid-phase extraction, one-pot 18 F-labeling offer fewer benefits, which may explain the limited use of BPDS despite the astonishing improvement in reaction rates that can be achieved with this chelator.

General
All reagents were purchased from Sigma-Aldrich (Dorset, UK) and used without further purification. NMR spectra were recorded on Bruker Avance (500 MHz or 600 MHz) spectrometers operating at the frequency of 500 or 600 MHz for 1 H, and 125 or 150 MHz for 13 C. Chemical shifts (δ) are reported in ppm downfield from the internal standard tetramethylsilane. High resolution mass spectra were recorded on a thermo Finnigan MAT900xp (CI/EI) or a Waters LCT Premier XE (ES) mass spectrometers. No-carrier-added aqueous [ 18 F]fluoride was provided by St Thomas" Hospital, King"s College London, UK. HPLC analyses were performed with an Agilent 1200 series system equipped with a diode array UV detector (results described are for UV at 265 nm), and a Raytest GABI star NaI detector. Analytical runs and HPLC purifications were carried out on a Chromolith® performance column, RP18-e, 100 × 4.6 mm and 100 × 10 mm (Merck), respectively. The solvent systems used were water and methanol, both containing 0.1% of formic acid. The flow rate was 3 mL/min for analytical and 5 mL/min for semi-preparative columns, respectively. Automated synthesis was carried out on HBIII module (Scintomics, Lindach, Germany). Isolated radiochemical yields (RCY) were measured using a Curiementor ® 4 ion-chamber (PTW).

General Procedure for Preparation of the Alkyne-functionalized 6-Halopurines
To a mixture of the 6-halopurine (1.0 equiv.) and NaH (60% dispersed in oil, 1.1 equiv.) in anhydrous DMF was added the alkylating agent (1.3 equiv.) under inert atmosphere. After stirring for 24 h at room temperature, the reaction mixture was diluted with dichloromethane (DCM) and extracted three times with water. The organic layer was dried over MgSO 4 and the solvent was removed under reduced pressure. The residue was purified by flash chromatography on silica gel using a gradient of DCM and ethyl acetate (from 0% to 40% of ethyl acetate) to give the N7-and the N9-alkynyl analogues.

General Procedure for Cycloaddition of 2-fluoroethyl azide (FEA) with Alkyne-functionalized Purines
A solution of the alkyne-functionalized 6-halopurine (1.0 equiv.) in DMF was subsequently reacted with FEA (1.3 equiv.) in the presence of sodium ascorbate (0.1 equiv.) and CuSO 4 ·5 H 2 O (0.05 equiv.) at room temperature for 4 h. The reaction mixture was diluted with DCM and the organic phase was washed with water. The organic layer was dried with MgSO 4 and the solvent was removed under reduced pressure. The residue was purified by flash chromatography on silica using a gradient of ethyl acetate and MeOH (from 0% to 10% of MeOH).  analytical RCY: 60 ± 1% (n = 3); isolated RCY 41%. HPLC gradient system: the methanol content was increased from 5% to 12% over 7 min, then to 20% over 1 min and finally to 70% over 4 min. The flow rate was 3 mL/min and 5 mL/min for analytical and semi-preparative HPLC, respectively. Retention times of [ 18 F]6 were 5.6 min and 8.6 min for analytical and semi-preparative runs, respectively.

6-Chloro
. Hands-on reaction: analytical RCY: 76 ± 1% (n = 3). Automated synthesis: analytical RCY: 75 ± 1% (n = 3); isolated RCY was 40% when using a C-18 Plus cartridge. When using a C-18 light cartridge the RCY was 9 ± 2% (n = 3) after sterile filtration, with a specific activity of 5-7 GBq/µmol. Analytical HPLC was carried out with a Luna C18(2) column (3µm, 50 × 4.6 mm, Phenomenex) with a flow rate of 1 mL/min using the following gradient: the methanol content was increased from 5% to 70% in 13 min. Semi-preparative HPLC with a Chromolith® performance column and a flow rate of 5 mL/min was carried out using an isocratic solvent system consisting of 9% acetonitrile in water with 0.1% formic acid. Retention time of [ 18 F]7: 7.1 min and 17 min for analytical and semi-preparative runs, respectively. The specific activity was measured using a Chromolith® performance RP18-e (100 × 4.6 mm) column with a flow rate of 5 mL/min [ Figure 4(b)]. The methanol content was increased from 5% to 20% over 5 min, and then to 50% over 3 min. A calibration curve was obtained by correlating the area under the peak (UV = 265 nm) with the injected mass of the non-radioactive reference compound 7 across a predetermined range (measurements in duplicate, linear fit with R 2 = 99.4%). Following formulation of [ 18 F]7, an aliquot was analyzed and the correlation curve was used to estimate the total mass in the sample (UV integration from 2-8 min). The specific activity was calculated as the activity in the aliquot divided on the estimated mass expressed as µmol of 7.

Conclusions
Conditions for labeling of alkyne-functionalized 6-halopurines with [ 18 F]fluoroethyl azide have been investigated. The use of the copper(II)/ascorbate catalytic system failed to yield the target triazoles in satisfactory yields, and addition of excess of copper(II) relative to the alkynes led to decomposition of the precursors. In contrast, when BPDS was used as an auxiliary copper(I) chelator the cycloaddition reaction proceeded almost immediately at room temperature to give the corresponding triazole-substituted purines. The high reaction efficiency in the presence of BPDS-copper(I) was exploited to develop a one-pot method for labeling that circumvents the need to distill [ 18 F]fluoroethyl azide. The method enabled labeling of four 6-halopurines in 55-76% analytical radiochemical yield with short reaction time and under mild conditions (1 min at room temperature). The ability to carry out the reaction in one pot without the need for distillation makes this a highly convenient approach to prepare [ 18 F]fluoroethyltriazoles.