[18F]Sulfonyl fluorides
The first account of the sulfur-[18F] fluorine bond was reported in the 1970s by de Kleijn and co-workers (De Kleijn et al. 1975; De Kleijn and van Zantan 1977). The preparation of [18F] tosyl fluoride (14) was reported by reacting tosyl chloride with the potassium [18F] fluoride in water, in an unreported yield (Fig. 3). The radiosynthesis of [18F] tosyl fluoride was also reported by Neal et al. (Neal et al. 2005); however, in this case, the compound was an undesired by-product from the reaction of bis (tosyloxy) methane (15) with [18F] fluoride to form [18F] fluoromethyltosylate (16, Fig. 4). The formation of the [18F] tosyl fluoride by-product was highly variable in the reaction. The ratio of [18F] fluoromethyltosylate : [18F] tosyl fluoride changed from 68:32 to 9:91 when the amount of Kryptofix 222 in the radiochemical reaction was reduced from 18 mg to 1.5 mg. The use of tetrabutylammonium bicarbonate as an activating agent also led to high yields of [18F] tosyl fluoride being formed as the by-product (85%). The authors were also able to increase the yield of the desired [18F] fluoromethyltosylate by adding water (up to 10% v/v) to the reaction, which was claimed to hydrolyze the [18F] tosyl fluoride. This justification is surprising, as sulfonyl fluorides are known to be synthesized from the chlorides even under aqueous conditions (Davies and Dick 1931). A more plausible explanation may rely on hardness-softness of the site of nucleophilic attack, but no data are currently available to support such hypothesis.
In recent years, attention has turned to using [18F]sulfonyl fluorides as prosthetic groups for ligands which can be used for PET imaging. In 2012, Inkster et al. reported four [18F]sulfonyl fluorides containing 4-formyl-, 3-formyl-, 4-maleimido- and 4-oxyalkynyl moieties (17–20, Fig. 5), synthesised in a 1:1 mixture of an organic solvent and an aqueous solution of caesium carbonate (Inkster et al. 2012). The 3-formyl analogue (18) was selected, due to its favourable stability in PBS buffer over 2 h, to be conjugated to the nonapeptide bombesin. The resulting radiolabelled peptide was stable in 10% DMSO in PBS buffer over 2 h at physiological temperature and pH, however, when analyzed in mouse serum, this product was found to only be 55% intact after 15 min, indicating defluorination was occurring.
Shortly after, Matesic et al. investigated the suitability of twelve aryl [18F] sulfonyl fluorides for applications in 18F-radiochemistry (Matesic et al. 2013). The [18F] sulfonyl fluorides were prepared by reacting the corresponding sulfonyl chlorides with [18F] fluoride under microfluidic conditions. These molecules bore neutral, electron-donating and electron-withdrawing functional groups. Additionally, sulfonyl chlorides containing varying degrees of steric bulk and a heterocyclic sulfonyl chloride were evaluated. Under microfluidic synthesis conditions, the [18F] sulfonyl fluorides could be prepared in less than 60 s at temperatures between 30–180 °C using a 0.5 mg/mL solution of the sulfonyl chloride precursor. The exception to this rule was the electron-withdrawing [18F] 4-nitrobenzenesulfonyl fluoride, which could not be produced at all using the parameters above. Interestingly, in the presence of 5% water in the reaction mixture, the compound could be formed in 91% radiochemical yield (RCY) at 30 °C (Pascali et al. 2014). This result is not unprecedented as several reports in recent years have described increased RCY when the molecules are prepared in solutions containing varying percentages of water (Sergeev et al. 2015). The reaction parameters of all analogues could be fine-tuned to produce >75% RCY at 30–180 °C by increasing precursor concentration, adding water, altering residence time, etc. The compounds were monitored for stability in buffers, before the stability of the leading candidates (21–23, Fig. 5) were evaluated in rat serum. After 2 h at physiological temperature in serum, the [18F] 2, 4, 6-triisopropylbenzenesulfonyl fluoride (23) was still 95% intact, suggesting that steric bulk around the sulfur-[18F] fluorine bond was more important in protecting the bond from hydrolysis, than the electron density of the molecule.
Following these encouraging preliminary results, it was envisaged that a sterically hindered sulfonyl chloride could be produced as a synthon to be subsequently radiolabelled with [18F] fluoride and conjugated to a peptide or protein. Ideally, the radiosynthon would contain an alkynyl pendant, allowing it to undergo bioconjugation with an azido modified macromolecule using click chemistry (Roberts et al. 2015). The radiosynthon would contain di-tert butyl groups to provide additional protection against hydrolysis of the sulfur-[18F] fluorine bond and could be formed using a variety of routes (Fig. 6). In our initial testing (unpublished data) of the synthesis of appropriate precursors (King AT 2016), we subjected the di-tert butyl analogue 24 (Route A, Fig. 6) to acidic conditions using chlorosulfonic acid in an attempt to produce the desired sulfonyl chloride 25, however the reaction was unsuccessful. Alternatively, 24 could be used to synthesize an azo intermediate (26, Route B, Fig. 6), that could be reduced to the amino derivative 27 and converted to a sulfonyl chloride using a diazonium ion intermediate. Intriguingly, the azo formation did not proceed in the 4-position of the molecule, but rather in the 2-position, as confirmed by 2D NMR experiments. It is postulated that the azo group preferentially attached to the 2-position of the aromatic ring due to the decreased amount of steric hindrance compared to the 4-position. The third route (Route C, Fig. 6) involved a Friedel-Crafts alkylation on an amine (28), followed by chlorosulfonylation through a diazonium intermediate. A recent literature example reported the Friedel-Crafts alkylation of 4-methoxyaniline and diphenylmethanol to introduce two diphenyl groups around the amine bond (Meiries et al. 2013). Applying this strategy to introduce two tert-butyl groups in 27 was unsuccessful. The most likely explanation was the absence of the carbocation generation required for electrophilic substitution. Compounds containing amino substituents are also sometimes known to be meta-directing due to the acidic reaction conditions converting the amine into an ammonium ion (McMurry J 2012). However, in this case no addition of the tert-butyl groups was observed, as determined by NMR and mass spectrometry. These three synthetic routes highlight the complexities in synthesizing a sterically hindered sulfonyl chloride – the addition of the sulfonyl chloride moiety in between two sterically bulky functional groups (Routes A and B, Fig. 6) can be just as difficult as introducing sterically bulky groups onto a precursor molecule (Route C, Fig. 6). These results indicate that the design and synthesis of a sterically bulky sulfonyl chloride synthon require further investigation in the future.
Potentially inspired by the feasibility of using microfluidic devices to radiolabel small molecules with 18F, Fiel et al., revisited the radiosynthesis of the [18F] 3-formylbenzenesulfonyl fluoride 18 in Fig. 5, using a magnetic droplet microfluidic (MDM) platform (Fiel et al. 2015). The platform consists of a Teflon sheet mounted on a stage, with a robotic arm beneath the Teflon sheet which controls the movement of the magnetic particles on the stage. [18F] Fluoride preconcentration and subsequent radiolabelling were performed on the same platform, in contrast to most traditional radiolabelling procedures whereby the [18F] fluoride is preconcentrated in a vial external to the microfluidic device. The authors were able to radiolabel the 3-formyl analogue 18 in tert-butanol in 5 min at room temperature using a total volume of 100 μL. The RCY (72 ± 1%, n =3) was comparable with that reported by Inkster et al. (73 ± 7%) (Inkster et al. 2012), however, the reaction time had decreased from 15 min to 5 min.
Another sulfonyl fluoride bearing radiosynthon has been reported by Al-Momani et al. (Al-Momani et al. 2015). The radiosynthon, [18F] FS-PTAD (31) could be synthesised in 91% RCY by converting the sulfonyl chloride moiety on urazole 29 to the corresponding sulfonyl [18F] fluoride (30), followed by oxidation using 1, 3-dibromo-5,5-dimethylhydantoin (DBDMH) (Fig. 7). [18F] FS-PTAD was conjugated to model substrates in moderate to good RCY under mild aqueous conditions (phosphate buffer, pH 7), however, basic conditions (sodium hydroxide, pH 9–10) were required to conjugate [18F] FS-PTAD to tyrosine. The [18F] FS-tyrosine (32) was >95% intact after 2 h in a 8% ethanol/PBS buffer solution, without the requirement of steric bulk around the sulfur-[18F] fluorine bond to prevent it from hydrolysis, as suggested previously (Matesic et al. 2013). The authors also demonstrated moderate uptake of [18F] FS-tyrosine into two human glioblastoma and one rat glioma cell lines, which warrants further exploration.
[18F] Ethenesulfonyl fluoride (ESF)
Still in the field of sulfonyl fluorides, ethenesulfonyl fluoride (ESF, 34) has been reported as one of the strongest Michael acceptors (Chen et al. 2016) and it has been reported to react with various “soft” nucleophiles under mild reaction conditions (Krutak et al. 1979). Consequently, [18F] ESF has significant potential to be used as a prosthetic group to radiolabel targets that contain lysine, cysteine residues or other ‘soft’ nucleophilic groups. The small size and hydrophilicity of [18F] ESF makes it ideal for radiolabelling proteins and polypeptides without drastically altering its pharmacological and polarity features, especially in comparison with currently used larger hydrophobic prosthetic groups. Our group has recently reported (Zhang et al. 2016) the first synthesis of radioactive [18F] ESF using a carried added (c.a.) process. We discovered that the route reported for the synthesis of non-radioactive ESF, employing fluorination and dehydrochlorination of 2-chloroethanesulfonyl chloride (33), was unsuccessful with [18F] fluoride. Therefore, a simple 19F/18F substitution reaction was attempted, and >80% RCY of [18F] ESF was obtained in diluted saline at 130 °C, employing a simple alumina cartridge purification approach (Fig. 8). This prosthetic group was employed in reaction with DMF solutions of seven model carboxy-protected amino acids (35), and the corresponding radioconjugate (36) was obtained in typically >40% yield at room temperature. Current investigations are undergoing towards synthesis of no carrier added (n.c.a.) [18F] ESF, its utilization for radioconjugation of proteins and polypeptides and the in vivo stability of the obtained products.
[18F] Fluorosulfate ion
Recent work into the utilization of [18F] fluorosulfate ion (38) as a PET imaging agent for the Na/I symporter (NIS) has been published by the group of Gee (Khoshnevisan et al. 2017). The function of NIS, a small anion transporter system, is important in the accumulation of iodide in thyroid follicles for the synthesis of hormones. The imaging of this system paves the way for the understanding of hormonal-related disease, as well as for assessing the efficacy of thyroid-targeted radiotherapies. Several radioactive anions have been used so far, mainly radioactive iodide ([131/123/124I] I-) or [99mTc] TcO4
-, with [186/188Re] ReO4
- and [18F] BF4
- evaluated more recently (Ahn 2012; Dadachova et al. 2002; Weeks et al. 2011). The use of [18F] BF4
- has demonstrated superior imaging features and optimal half-life characteristics, but the synthesis route afforded the radiotracer in low molar activity (1-8.8 MBq/nmol (Jiang et al. 2016)). Subsequently, the same group investigated different anions and their potential labelling with 18F. They have identified that at least three other fluorinated anions could be used for NIS imaging, due to their inhibitory potency: SO3F-, PO2F2
- and PF6
- (this last example being the most potent). On the basis of availability of amenable and practical radiosynthesis routes, [18F] SO3F- (38) was selected as a suitable candidate for novel NIS imaging. Compound 38 was synthesized by reaction of K222/[18F] KF dried complex with SO3-pyridine adduct (37, Fig. 9) in CH3CN. The radiochemical yield was up to 65% in 10 min at 80 °C. The purification was simply performed by passing the water diluted reaction mixture through neutral Al and QMA cartridges. [18F] SO3F- was recovered by eluting the QMA cartridge with 0.9% NaCl isotonic solution, providing the radiotracer in 31% RCY, >95% RCP, >48 MBq/nmol molar radioactivity and overall radiosynthesis time of <60 min. Radiotracer identity was confirmed by ion chromatography and residual amounts of pyridine, K222 and SO4
2- were also determined.
Tracer stability studies confirmed a RCP of >95% (after 4 h) in multiple media including formulation media, acidic environment and serum. Specific uptake of the tracer was confirmed in vitro in NIS-expressing cell lines, as well as in vivo in healthy mice using PET-CT. During the imaging studies, a steady increase of radioactivity was noticed in bones, especially after 30 min; however, the authors believe that this metabolism will likely not hamper the utility of [18F] SO3F- for NIS imaging. This work represented the first ever PET-CT imaging of a compound bearing an S-18F bond.
[18F]deoxyfluorinating agents
Despite the abundance of non-radioactive deoxyfluorination reactions published to date only a very limited number involving 18F have been reported up to date. From the chemical literature, there are multiple examples of deoxyfluorinating reagents containing an S-F bond (Fig. 10), such as DAST (diethylaminosulfurtrifluoride), Deoxo-Fluor (2-methoxy-N-(2-methoxyethyl)-N-(trifluoro-λ4-sulfanyl)ethanamine), XtalFluor –E and –M ((difluoro-λ4-sulfanylidene)(diethyl)ammonium tetrafluoroborate and 4-(difluoro-λ4-sulfanylidene)morpholin-4-ium tetrafluoroborate) and Fluolead (4-tert-butyl-2,6-dimethylphenylsulfur trifluoride) (Ni et al. 2014). The structure of these molecules features more than one fluorine atom, therefore posing limitations and posing significant challenges in regard to their radiosynthesis and the obtainable molar radioactivity.
Nonetheless, [18F] DAST was synthesized and used (Straatmann and Welch 1977) in the fluorination of short chain model alcohols. In this 1977 paper, the group of Welch tested three routes for the synthesis of this fluorinating reagent: (a) a de-novo synthesis of [18F] DAST from target produced [18F] SF4, (b) 18F/19F isotopic exchange on DAST with [18F] F2 and finally, (c) an exchange via treatment with [18F] HF. The best route was determined to be (c), that yielded [18F] DAST (39) in >80% RCY and was then used to deoxyfluorinate methanol, ethanol and ethylene glycol, with RCY of 20%, 25% and 12% respectively (Fig. 11). Given that the radiofluorinating reagent produced bears 3 fluorine atoms and therefore can provide a maximum RCY of 33%, the reported yields for the employed [18F] fluoroalkanes are relatively high. The authors however recognize that the achievable molar radioactivity is particularly low, due to the use of isotopic exchange reactions.
More recent literature has indicated that sulfonyl fluorides are viable as [18F] deoxyfluorinating reagents (Takamatsu et al. 2002; Guo and Ding 2015), and indeed some applicative value has been demonstrated by perfluorobutanesulfonyl fluoride (PFBS) (Vorbrüggen H. The conversion of primary or secondary alcohols with nonaflyl fluoride into their corresponding inverted fluorides. Synthesis (Stuttg) 2008; Bennua-Skalmowski and Vorbruggen 1995). Inspired by these contributions, Nielsen et al. have reported the synthesis and use of [18F] PyFluor (2-pyridinesulfonyl fluoride, 41) (Nielsen et al. 2015). In this work, the authors first studied a range of sulfonyl fluorides (comprising PFBS) to fluorinate a model alcohol, using 1.1 eq of fluorinating reagent and 2 eq of 1,8-diazabicyclo-[5.4.0] undec-7-ene (DBU) or 7-methyl-1,5,7-triazabicyclo [4.4.0] dec-5-ene (MTBD) as a base. This preliminary set of experiments revealed PyFluor as the most promising reagent, with the highest fluorination yield and selectivity (towards dexoyfluorination reaction). The postulated mechanism involves the formation of an intermediate sulfonate between Pyfluor and the alcohol, and its subsequent fluorination by nucleophilic substitution by the liberated fluoride ion. DBU or MTBD act as Brønsted bases and are proposed to help stabilize the developing fluoride ion, rendering it available to nucleophilic reaction. The wide scope demonstrated, allowed for Pyfluor to now be commercially available through Sigma-Aldrich as a deoxyfluorinating reagent.
In the same contribution, [18F] Pyfluor (41) was synthesized from the respective sulfonyl chloride (40) in CH3CN at 80 °C, giving the radio-synthon in 88% RCY in 5 min (Fig. 12). The addition of a benzyl protected tetrahydro-2H-pyran-pyran-2-ol to the same pot gave a 15% RCY of the desired deoxyradiofluorinated compound (42) to be obtained in 20 min. Currently, this has been the only reported radiofluorine application of the Pyfluor approach. Even if promising, a wider utilization of this approach might be hindered by the difficult access to the needed labelling precursor 40.
[18F] SF6 production and uses
[18F] Fluoride atoms produced by the irradiation of SF6 gas through the19F(n,2n)18F nuclear reaction in fast neutron generators were first reported in the 1970s (Colebourne and Wolfgang 1963). Even though these 18F atoms lost the majority of their kinetic energy in the presence of SF6 (Smail et al. 1972), they were still able to undergo addition onto olefins, ethylene and acetylene to produce the 18F-fluorinated analogues (Williams and Rowland 1971), and this observation was even proposed as a useful method for scavenging undesired 18F atoms. The production of [18F] SF6 remained largely unexplored in the radiochemistry literature for 40 years, until Gómez-Vallejo reported the cyclotron production of [18F] SF6 in 2016 (Gómez-Vallejo et al. 2016). In this method, the authors verified that new cyclotrons are able to exploit the (p, pn) route to obtain c.a. fluorinated gases (in this test CF4 was also produced), but the yield obtained are insufficient for imaging uses. Therefore, they optimized a double irradiation method, in which [18O] O2 was used as target material in the first irradiation, and CF4 or SF6 were used to fill the target in a second irradiation. In this way, almost 7 GBq of [18F] SF6 was produced using an integrated current of 4 μA and the authors hinted at the possibility of using such 18F-radiolabelled fluorinated gases for in vivo PET imaging assessment of lung ventilation. The availability of [18F] SF6 may also turn useful in multimodal imaging approaches, due the current use of SonoVue® (from Bracco), an SF6 microbubble formulation, as an ultrasound contrast imaging agent.