A non-anhydrous, minimally basic protocol for the simplification of nucleophilic 18F-fluorination chemistry

Fluorine-18 radiolabeling typically includes several conserved steps including elution of the [18F]fluoride from an anion exchange cartridge with a basic solution of K2CO3 or KHCO3 and Kryptofix 2.2.2. in mixture of acetonitrile and water followed by rigorous azeotropic drying to remove the water. In this work we describe an alternative “non-anhydrous, minimally basic” (“NAMB”) technique that simplifies the process and avoids the basic conditions that can sometimes limit the scope and efficiency of [18F]fluoride incorporation chemistry. In this approach, [18F]F− is eluted from small (10–12 mg) anion-exchange cartridges with solutions of tetraethylammonium bicarbonate, perchlorate or tosylate in polar aprotic solvents containing 10–50% water. After dilution with additional aprotic solvent, these solutions are used directly in nucleophilic aromatic and aliphatic 18F-fluorination reactions, obviating the need for azeotropic drying. Perchlorate and tosylate are minimally basic anions that are nevertheless suitable for removal of [18F]F- from the anion-exchange cartridge. As proof-of-principle, “NAMB” chemistry was utilized for the synthesis of the dopamine D2/D3 antagonist [18F]fallypride.

F radiopharmaceuticals for clinical use; and 3. some [ 18 F]F − is always lost due to volatilization and adsorption of the dried K[ 18 F]F onto the walls of the reaction vessel during the drydown process. Eliminating these limitations could improve the reliability of 18 F radiopharmaceutical production.
There have, therefore, been multiple efforts to improve this process. These have included: (1) The use of alternative AEX sorbents, including an N-vinyl lactam/divinylbenzene copolymer preloaded with long-chain quaternary ammonium salts 4 and C 18 cartridges impregnated with a phosphonium borane 5 ; (2) The use of elution matrices containing strong organic bases (e.g., phosphazenes 6 11 . Other examples include the work of Richarz, et al. who described a "minimalist" approach in which [ 18 F]F − is eluted from QMA cartridges using MeOH or EtOH solutions where the cation is the trimethylanilinium, diaryliodonium, or triarylsulfonium derivative of the aromatic target compound and the anion is TfO − or Br − 12 . The cation serves as the PTC and the TfO − or Br − counterion displaces the [ 18 F]F − from the AEX resin. However, the eluting solvent must still be removed by distillation and replaced with a polar aprotic reaction solvent. Neumann, et al. described a related approach in which [ 18 F]F − is eluted from the AEX cartridge using uronium salt-based precursors in 10:1 2-butanone:EtOH, and the 18 F-labeling reaction is carried out after addition of tributylamine to the eluate 13 . While not a general approach to [ 18 F]F − incorporation, this approach permits the incorporation of [ 18 F]F − into electron-rich (i.e. unactivated) aryl rings, and thus stands to make a significant impact in this research area. Another example of an innovative way to avoid the drydown step is the work of Basuli, et al. who discovered that certain 18 F-fluorination reactions will take place on the surface of a QMA resin at room temperature. Using this method, they prepared the prosthetic group [ 18 F]fluoronicotinic acid-2,3,5,6-tetrafluorophenyl ester by passing the trimethylammonium triflate precursor through an AEX cartridge impregnated with [ 18 F]F − 14 . The 18 F-labeled product was then immediately added to solutions containing amine-bearing targeting vectors [e.g. c(RGDfK), albumin, and a prostate-specific membrane antigen (PSMA) inhibitor] to produce the [ 18 F]fluoronicotinic amides 15 .
The simplest approach to avoiding the drydown step is to use the [ 18 F]F − eluted from AEX cartridges directly, without azeotropic drying. In one example of this approach, Brichard and Aigbirhio 16  The decay corrected (DC) elution efficiencies using these "damp" solutions of MeCN or DMF were 95-99% and 88-99%, respectively. When using DMSO, 5% water was required in order to achieve an elution efficiency of 89%. Subsequently, 100 µL fractions of the eluate, each containing 7.  17 . This distinction is necessary because in many cases the investigators did not isolate the product from the reaction mixture.] Another example is the work of Blecha, et al., who reported the quantitative elution of [ 18 F]F − from a 12.6 mg MP-1 cartridge (ORTG, Inc.) using 1-2 mL solutions of K 2 CO 3 /K 2.2.2. in 97% DMF or 99% MeCN 18 . Several aromatic and aliphatic substrates were successfully radiolabeled (e.g. 4-[ 18 F]fluorobenzonitrile from 4-nitrobenzonitrile, 50% RCC; from 4-trimethylammonium triflate benzonitrile, 95% RCC) by mixing a fraction of the eluate (1/8 to 1/4 of the total radioactivity) with a solution of the precursor in dry solvent. It's important to note that in both of these cases only a small fraction (10-25%) of the total amount of the [ 18 F]F − eluted from the AEX cartridge was used in each labeling reaction, which significantly reduces both the amount of base and the amount of water in the reaction mixture. Kniess  www.nature.com/scientificreports www.nature.com/scientificreports/ The following report describes an approach that, informed by these previous studies, facilitates the synthesis of 18 F-labeled PET tracers under non-anhydrous reaction conditions (3 or 5% water in 1 mL total solvent; Fig. 1b). In contrast to the approach of Brichard and Aigbirhio 16 , this approach uses the entire volume of [ 18 F]F − eluted from the anion exchange. The reaction conditions thus reflect typical radiopharmaceutical production conditions. This work also introduces the use of minimally basic tetraethylammonium salts (vs. HCO 3 − or CO 3 2− ) as [ 18 F] F − eluents with the expectation that the resulting [ 18 F]F − solutions might prove more suitable for 18 F-labeling of base-sensitive precursors. Finally, we report a protocol for the synthesis of [ 18 F]fallypride as the proof-of-concept for this "non-anhydrous, minimally basic" (NAMB) approach.

Results and Discussion
Commercially available AEX columns containing 10-12 mg of standard QMA (capacity = ~0.2 meq/g) or MP-1 (capacity = ~0.7 meq/g) resin were used for [ 18 F]F − trapping. These small columns were connected to disposable syringes via a Luer-lock/hose barb adapter ( Supplementary Fig. S2). After the [ 18 F]F − was trapped on the column, the column was washed with acetonitrile (3 mL) and argon was passed through the column for 10 min (Fig. 1b). The [ 18 F]F − was then eluted with a tetraethylammonium salt in 100-500 µL of MeCN or DMSO containing 10-50% water. We attempted to use lower water concentrations; however, at the lower water concentrations described by Brichard and Aigbirhio (0-2% H 2 O, 15 mg/mL TEAB) 16 , we observed that TEAB precipitated from the solution upon standing. Increasing the water content to 10-50% ensured that the salt remained in solution and facilitated [ 18 F]F − extraction. After elution, a solution of the precursor in anhydrous solvent was added, reducing the final water concentration to 3-5% in a 1 mL reaction volume. . Considering that some 18 F PET tracers suffer from low isolated yields due to the sensitivity of the precursor or product to heating in the presence of HCO 3 − or CO 3 2− , we also evaluated tetraethylammonium perchlorate (TEAP) and tetraethylammonium p-toluenesulfonate (TEATos) as minimally basic alternatives to TEAB. In this case, the only base present in the final "NAMB" reaction mixture was the small amount of carbonate present on the column after pre-conditioning by the manufacturer that is co-eluted with the [ 18 F]F − .
Columns were reversed before elution of the [ 18 F]F − . Elution in the same direction as [ 18 F]F − capture resulted in lower elution efficiency (<60%) in all cases, except when eluting a QMA column using 7.8 μmol AEX reagent in 50% MeCN/water (100 μL). Under these conditions, the elution efficiencies in the forward direction using TEAB, TEAP, and TEATos were 82, 79, and 98% respectively.
The ability of tetraethylammonium salts to facilitate nucleophilic 18 F-fluorinations in "damp" MeCN or DMSO was assessed using [ 18 F]1 as a model compound (Fig. 2). Cartridge eluates (100 µL) containing 30% water/70% organic solvent or 50% water/50% organic solvent were diluted with a solution of precursor 2 (1.5 mg) in anhydrous organic solvent (900 µL) such that the final reaction volume was 1 mL and the final water content was 3% or 5%. The reaction mixtures were then heated by microwave irradiation (150 °C, 10 min) and assayed by radio-TLC (Table 2). Little difference in RCC was observed when comparing the three AEX reagents or the two organic solvents. However, the RCC of reactions carried out in 97% organic solvent (Table 2, entries 7-12) was generally higher (72-89%) than those carried out in 95% organic solvent (51-72%; entries 1-6). The highest yield was observed using TEATos in 97% MeCN (89%, Fig. 2). See Supplementary Fig. S3 for an example radio-TLC trace of an [ 18 F]1 reaction mixture.
The crude [ 18 F]1 reaction mixtures were largely free of radiochemical byproducts as determined by analytical HPLC (Fig. 3a). Furthermore, as the degree of precursor decomposition has been utilized by others as a metric of overall 18 F-fluorination reaction "mildness" 9 , it is worth noting that in reaction mixtures using TEAP or TEATos in 97% MeCN, precursor 2 remained largely intact after heating.
As some previous reports describing non-anhydrous 18 F-fluorination reactions are limited in either the choice of leaving group (e.g. -OTs only) 11 or the mechanism of 18 F incorporation (e.g. S N Ar reactions only) 5 , we tested the utility of the "NAMB" approach for the radiosynthesis of the 18 Fig. 4 and Table 3). Interestingly, this reaction proceeded in non-anhydrous solvent mixtures containing DMSO but not MeCN. As observed for model compound [ 18 F]1, a decrease in water content from 5% to 3% correlated with an increase in RCC from 43-50% (entries 4-6) to 64-76% (entries 7-9). See Fig. 3b for a representative radio-HPLC trace of the [ 18 F]3 reaction mixture and Supplementary Fig. S4 for an example radio-TLC trace.
In light of the promising results with [ 18 F]1 and [ 18 F]3, we sought to utilize "NAMB" 18 F-fluorination chemistry for the preparation of an established 18 F-labeled radiopharmaceutical. This method was, therefore, applied to the synthesis of [ 18 24,25 (Fig. 5). The standard synthesis of [ 18 F]fallypride is known to be base-sensitive, because of the tendency of the tosyl-fallypride precursor 6 to undergo hydrolysis and elimination side reactions 26 , making this compound a good candidate with which to evaluate this minimally basic synthesis.
The importance of the interplay between the [ 18 F]F − elution conditions and the [ 18 F]F − labeling conditions is highlighted by the work of Lemaire et al. 6 , who showed that [ 18 F]F − could be efficiently stripped from QMA resin with the phosphazine base P 2 Et in "damp" MeCN (0.63% H 2 O). Subsequently, the entire eluate volume (700 µL) was reacted with 6 in dry MeCN (0.5-1 mL) containing 2-t-butyl-1,1,3,3-tetramethylguanidine without azeo-tropic drying of the [ 18 F]F − . However, under these conditions, 20 mg of 6 was required to achieve high RCC (87%), presumably due to the larger final reaction volume (>1 mL) and the presence of a relatively large quantity of a very strong base (P 2 Et; 45 µmol) in the reaction mixture. Investigators were able to achieve an equally high product yield (86%) using only 1 mg of precursor, but this was only possible by employing a small fraction of the total eluate volume (50 μL out of 900 µL, or 5.6% of the total radioactivity), which reduces the amount of P 2 Et present in the reaction mixture from 45 µmol to 2.5 µmol. It is worth noting that P 2 Et does not itself displace the [ 18 F]F − from the anion exchange resin. The [ 18 F]F − is eluted by an anion, presumably OH − , produced in situ by P 2 Et deprotonation of the water present in the eluent.
The HPLC profiles of [ 18 F]fallypride reaction mixtures employing TEAP and TEATos were very similar. In contrast to compound [ 18 F]1, significant decomposition of the starting material was observed after heating, with the major decomposition product found at 7.02 min in HPLC assays of the crude reaction mixtures (Fig. 6a). Nevertheless, only one small radioactive impurity was consistently observed (Fig. 6a, [29][30][31] . The highest yield (68%) was reported by Moon, et al. who used 10 µL of a 40% solution of TBAB as the phase transfer catalyst and observed that the yield decreased to less than 50% in the presence of higher amounts of base 26 .
The radiochemical purity of all final products was >99%. Product identity was verified by co-injection of the 18 F-labeled compound with non-radioactive fallypride (Fig. 6b). Molar activities (MAs) were calculated based on a calibration curve prepared from [ 19   Overall, for the synthesis of [ 18 F]fallypride under NAMB conditions, we did not observe a significant advantage of TEATos over TEAP in terms of ease of use (e.g. solubility), precursor tolerance (i.e. reaction "mildness"), radiochemical yield, or molar activity.

conclusions
Solutions of tetraethylammonium tosylate and tetraethylammonium perchlorate in non-anhydrous solvent mixtures offer a straightforward means to efficiently extract [ 18 F]F − from small AEX columns and facilitate the synthesis of 18 F-labeled aromatic and aliphatic compounds without the need for the azeotropic drying step that is ubiquitous in 18 F-PET chemistry. Since these anions do not contribute to the basicity of the reaction mixture, we describe this approach as "non-anhydrous, minimally basic" ("NAMB") 18 F-fluorination chemistry. Tetraethylammonium tosylate offers a particularly attractive alternative to standard reagents because the tosylate anion is both non-basic and non-oxidizing and because it is already present in many nucleophilic 18 F-fluorination reactions as a leaving group.
As shown by the synthesis of [ 18 F]fallypride, "NAMB" labeling conditions can be used to prepare 18 F-PET tracers from commercially available, GMP-compliant precursor molecules and single portions of [ 18 F]F − without the need to dry the [ 18 F]F − prior to use. Furthermore, the "NAMB" method can accommodate volumes of aqueous [ 18 F]F − (1-2 mL) and concentrations of precursor (1-3 mg in 1 mL) that are commonly used in automated synthesis systems. Further improvements in this technique are anticipated as "NAMB" chemistry is evaluated for the synthesis of a wider variety of clinically relevant 18  was trapped on an MP-1 anion-exchange column (MedChem Imaging, carbonate form, 10-12 mg), which was previously activated with H 2 O (1 mL). After washing the column with anhydrous MeCN (3 mL), Ar was passed through the column for 10 min. Fluorine-18 was eluted from the column in the reverse direction into a microwavable test tube using a solution of tetraethylammonium tosylate (TEATos, 23.5 mg/mL, 100 µL) in 7:3 MeCN:H 2 O. Residual liquid was removed from the column using a syringe filled with air (10 mL). 18 F-labeling reaction. Tosyl-fallypride (6; 1 mg) in dry MeCN (900 µL) was added to the [ 18 F]F − solution and the tube was crimp-sealed, magnetically stirred for 20 sec, and heated (microwave) to 150 °C for 10 min. After removing small aliquots for silica gel radio-TLC (10% MeOH in CH 2 Cl 2 , 72 ± 2% RCC, n = 4) and analytical HPLC (HPLC 1, Program A in the Supplementary Information), the reaction mixture was diluted with 0.1% TFA in water (1 mL) and injected onto a semi-preparative HPLC column (HPLC 2, Program B). The product was collected, diluted with water (50 mL), and trapped on a Sep-Pak ® C 18 Light cartridge that was previously activated with EtOH (3 mL) and water (10 mL). After washing the column with water (5 mL), [ 18 F]fallypride was eluted with EtOH (1 mL) and diluted with 0.9% saline (9 mL). The final formulation was passed through a 0.2 micron filter to afford 22.1 MBq (596 µCi) of [ 18 F]fallypride (21% non-decay corrected, 37% decay corrected). Product identity and molar activity were assessed by HPLC (HPLC 1, Program C). The synthesis time was 88 min. from start-of-synthesis.