Novel late-stage radiosynthesis of 5-[18F]-trifluoromethyl-1,2,4-oxadiazole (TFMO) containing molecules for PET imaging

Small molecules that contain the (TFMO) moiety were reported to specifically inhibit the class-IIa histone deacetylases (HDACs), an important target in cancer and the disorders of the central nervous system (CNS). However, radiolabeling methods to incorporate the [18F]fluoride into the TFMO moiety are lacking. Herein, we report a novel late-stage incorporation of [18F]fluoride into the TFMO moiety in a single radiochemical step. In this approach the bromodifluoromethyl-1,2,4-oxadiazole was converted into [18F]TFMO via no-carrier-added bromine-[18F]fluoride exchange in a single step, thus producing the PET tracers with acceptable radiochemical yield (3–5%), high radiochemical purity (> 98%) and moderate molar activity of 0.33–0.49 GBq/umol (8.9–13.4 mCi/umol). We validated the utility of the novel radiochemical design by the radiosynthesis of [18F]TMP195, which is a known TFMO containing potent inhibitor of class-IIa HDACs.

. However, the carrier added radiolabeling methods were hampered by isotopic dilution with 19F-fluoride. This is detrimental to utility of such tracers for PET imaging since the nonradioactive fraction is predominant in the final product leading to excessive self-blocking. To overcome this limitation, classical direct nucleophilic radiofluorination via [ 18 F]-for-Br nucleophilic substitution was utilized to obtain [ 18 F]trifluoromethyl arenes under no-carrieradded conditions 39 . However, the multistep radiosynthesis and the presence of inseparable labelling precursor confounded the specific activity. An efficient, Cu-mediated coupling of difluoroiodomethane with aryl iodides for the radiosynthesis of [ 18 F]trifluoromethyl arenes was recently reported 40 . However, this radiofluorination is limited to aryl iodides precursors and the use of low boiling difluoroiodomethane starting material likely will further hamper the wide utility of this method. An efficient no-carrier-added, however, multicomponent protocol for facile [ 18 F]trifluoromethylation of aromatic and heteroaromatic systems using (hetero)aryl iodide, and [ 18 F] CF 3 Cu generated in situ from methyl chlorodifluoroacetate, CuI and TMEDA was recently reported to generate high radiochemical yield and moderate molar activity of 0.1 GBq/μmol (2.7 mCi/uM) 36 .
In contrast to the above reports, the radiofluorination reported here is a straightforward late stage, performed in a single radiochemical step via [ 18 F]-for-Br exchange under no-carrier-added conditions which led to tracers with high radiochemical purity and relatively higher molar activity of 0.33-0.49 GBq/umol (8.9-13.4 mCi/umol) when compared to the previously reported molar activity values for the [18F]trifluoromethyl-containing PET tracers. Moreover, the unreacted labelling bromo-precursor (starting material in large excess) is separable and did not confound the specific activity ( Fig. 1, Sl). Also, we experimentally demonstrated that the [ 18 F]- 19 F isotopic exchange does not occur and as such does not contribute to the radioactive product, and consequently does not influence the molar activity. Moreover, we expect the molar activity to improve significantly once starring with higher radioactivity using an onsite cyclotron for [18F]fluoride production. Currently, we are purchasing the [18F]fluoride from an outside source and it undergoes significant decay (> 2 half-lives) prior to the start of our experiments. Importantly, our radiolabelling approach can be widely applicable to develop new generation of TFMO-containing molecules without changing method of preparation for bromodifluoromethyl-1,2,4-oxadiazole, radiolabelling procedure or reaction conditions.

Results and discussion
Given our specific interest in targeting the class-IIa HDACs for PET tracer development, the TFMO bearing molecules presented an attractive target for PET tracer development. The TFMO is a distinctive class-IIa HDAC pharmacophore motif that interacts with the zinc ion at the bottom of the class-IIa HDAC catalytic pocket rendering high specificity and selectivity to class-IIa HDACs 32,34 . Due to the short half-life of the [18F]fluoride PET radiotracers, it is highly desirable to design a late-stage labeling site that allows for radiolabeling molecules in a short time with high radiochemical purity and sufficient molar activity. Therefore, we designed a radiochemical route (Figs. 2 and 3) to radiolabel the TFMO moiety with [18F]fluoride and we extended this novel radiofluorination method to incorporate the [18F]fluoride into TMP195 in a single radiochemical step as shown in Fig. 4. This approach enabled the very desirable one-step radiolabeling reaction thus facilitates straightforward automated routine production of [18F]TFMO based PET tracers in the future. By design this is also a universal labeling site that allows for radiolabeling many TFMO containing molecules without changing the radiolabeling strategy as evidenced by our current work.
To extend the utility of our new radiochemical design, we also radiolabeled TFMO-containing esters with [ 18 F]fluoride using the same strategy. Moreover, we generated the radioactive acids [ 18 F]1 and [ 18 F]2 which are expected to be generated in vivo from cleavage of the amide bond of the class-IIa HDAC inhibitors (i.e. [ 18 F] TMP195) by the metabolizing enzymes. As such, the acid tracers could be useful in determining the metabolic profile of the new tracers in vivo.
It is critical to note that despite the relatively low radiochemical yield, the radioactivity obtained was sufficient to perform our preclinical imaging studies routinely even when starting with as low as 5.5 GBq (150 mCi). In fact, due to the shorter time of radiosynthesis and dose formulation, it is likely that performing the radiosynthesis starting with high [18F]fluoride dose and using a fully-automated module will significantly improve the amount of the final dose and also expected to further improve the molar activity as was reported for other tracers 40 . Notably, the yield was sufficient for performing our preclinical studies and will also be sufficient to produce a clinical dose for human studies-that is, if a TFMO-based tracer from our ongoing studies becomes available for clinical translation. Notably, the radiosynthesis of [ 18 F]TMP195, an inhibitor class-IIa HDACs which is an important target for cancer and brain imaging, described in this publication underscores the significance and novellty of the radiochemical labeling method. Addtionally, it is likely that the radiochemical methods reported herein can be extended to other classes of small trifluoromethyl-containing heterocyclic molecules that appear in inhibitors of highly pursued imaging targets, such as cyclooxygenases (Cox-1 and Cox-2) and estrogen receptors [44][45][46] .
The new tracer [ 18 F]TMP195 was purified using a semi-preparative HPLC system (C18 column with 4.0 ml/ min flow rate using 75% acetonitrile/water). The overall radiochemical yield of [ 18 F]TMP195 was within 2-5% (end of radiosynthesis, decay-corrected, n = 10) and the radiochemical purity was > 98%. The identity of [ 18 F] TMP195 was confirmed by co-injection together with the corresponding cold (unlabeled) molecule using analytical HPLC, as shown in Fig. 5 (chromatograms a and c).
The molar activity of [ 18 F]TMP195 was determined from the area under the curve of the tracer that is attributed to the ultraviolet peak in the HPLC chromatogram (Fig. 5, chromatograms B) against a calibration curve pre-prepared with the unlabeled reference standard. The molar activity ranged from 0.33 to 0.49 GBq/umol (8.9-13.4 mCi/umol). Althogh these values are significantly higher than those obtained for [ 18 F]-labeled aryl-CF 3 applying [ 18 F]CuCF 3 -based cross-coupling strategies 36,47 , we expect the molar activity to improve significantly starting with a high dose using a fully automated radiofluorination. Currently, the [ 18 F]fluoride is purchased from an outside source and as a result, it undergoes significant decay (> 2 half-lives) before reaching our laboratory which reduces the overall molar activity. Despite the relatively low radiochemical yield, we obtained the tracer in very high purity and the no carrier added radiofluorination led to high molar activity. Furthermore, the labeling bromo-precursor such as compound 23 was easily separable from the reaction mixture using semi-prep HPLC (Fig. 1, Sl). Therefore 23 was not detectable in the final dose and did not confound the molar activity which is a major advantage over the previous reports 39 .
Next, we investigated as to whether the [ 18 F]-19 F isotopic exchange is contributing to the radioactive product, and consequently influencing the molar activity. To rule out this possibility, we performed the radiofluorination   www.nature.com/scientificreports/ under the same conditions above using TMP195 as a starting material as illustrated in Fig. 6. [18F]TMP195 was not obtained and the non-radioactive TMP195 remined unchanged (Figs. 2, Sl) which likely explains the relatively higher molar activity obtained in this work compared to previous reports. Finally, [18F]TMP195 exhibited poor aqueous solubility which rendered it difficult to formulate for in vivo studies since high concentration of DMSO (~ 50%) was needed to solubilise the tracer. We determined the partition coefficients (LogD) using the octanol/PBS shake flask method 48 . The observed LogD value of [ 18 F] TMP195 was > 6.0 which was much higher than the calculated value of cLogD = 5.84 (ChemDraw). In fact, only a background radioactivity was detected in aqueous phase and most of radioactivity retained in the octanol layer. Therefore, other TFMO containing class-IIa HDAC inhibitors are being pursued in our laboratory to identify new PET tracer candidates with improved physiochemical properties and in vitro and in vivo pharmacokinetic profile. The new data will be published in due course.

Conclusions
In summary, we reported a late-stage radiofluorination of TFMO containing molecules and successful radiosynthesis of class-IIa HDAC targeting PET tracer [ 18 F]TMP195. The novel late-stage radiolabeling strategy produced an identical radioactive class-IIa HDAC inhibitor, thus ensuring maintenance of the identical inhibition affinity of TMP195. This strategy is being successfully applied in our laboratory to produce a new generation of [ 18 F] TFMO containing class-IIa HDAC inhibitor-based PET tracers. The single radiofluorination step is suitable for straightforward automation and routine production and can produce and formulate the tracer in relatively short period of time. It is also likely that the reported radiochemistry can be extended to other target molecules that contain trifluoromethyl-heterocyclic moiety. Finally. Our novel radiofluorination method reported in this publication will pave the way for the development of TFMO-containing PET tracers for PET imaging of class-IIa HDAC expression in cancer and the disorders of the CNS.

Methods
General information. Solvents and starting material were obtained from commercial sources and were used as received. High-Performance Liquid Chromatography (HPLC) was performed with a 1260 series pump (Agilent Technologies, Stuttgart, Germany) with a built-in UV detector operated at 250 nm and a radioactivity detector with a single-channel analyzer (labLogic) using a semipreparative C18 reverse-phase column (10 × 250 mm, Phenomenex) and an analytical C18 column (4.6 × 250 mm, ASCENTIS RP-AMIDE, Sigma). An acetonitrile/ammonium acetate buffer (MeCN/NH4OAc: 20 mM) or acetonitrile/water (MeCN/water) solvents with varying composition (solvent systems were developed specific to each compound) was used for quality control analyses at a flow of 1 mL/min. High resolution mass spectroscopy (HRMS) was performed using Agilent 1260HPLC/G6224A-TOF MS. NMR spectroscopy was performed using 400 MHz Bruker instrument.
Chemical synthesis. TMP195 was prepared similar to the previously described procedures 34 .
General procedure for amine/carboxylic acid coupling reactions (3-4, 12-13 and 23). The acid (1.0 eq) and HATU (1.2 eq.) in DMF (1.0-3.0 mL) were stirred for 15 min followed by simultaneous addition of the amine (1.2 eq) and NMM (excess: ~ 1.0 mL). The reaction continued for 3 h. The DMF was removed under vacuum and the residue was purified by column chromatography followed by trituration in cold pentane to afford the final product in 60-80% yield.
General procedure for synthesis of (N′-Hydroxycarbamimidoyl)benzoic acid (7-8 and 16-17). To the nitrile (1.0 g) in ethanol (30 mL) was added first hydroxylamine hydrochloric acid (1.0 g) dissolved in water (8.0 mL) followed by sodium carbonate (1.2 g) dissolved in water (12.0 mL). The mixture was heated under reflux for 4 h. Ethanol was removed under reduced pressure and the residue was diluted with water, acidified with 10% HCl to pH ∼ 3, and filtrated, then washed with water and dried under reduced pressure to afford compound (N′-hydroxycarbamimidoyl)benzoic acid in 50-80% yield. www.nature.com/scientificreports/ General synthesis of the bromodifluorooxadiazoles (bromo-Precursors: 10-11, 20-21 and 23). Bromodifluoroacetic anhydride (2.0 mL) neat was added to N′-hydroxycarbamimidoyl benzoic acid, or benzamide. The reaction mixture heated to 50 °C for 3 h. The volatiles were evaporated. Benzoic acids were filtered, washed with water and dried under vacuum. The crude esters and amides were purified by column chromatography using ethyl acetate/hexanes to afford the bromodifluoro-analogs.
General synthesis of the trifluorooxadiazoles (19)(20). The synthesis of the TFMO moiety was performed similar to the synthesis described above for bromodifluorooxadiazole except that neat trifluoroacetic anhydride was used.
A solution of the bromo-precursor (6-8 mg) in the appropriate solvent (i.e. DMSO) (0.4 mL) was added to the dried K[18F]/kryptofix or Cs[18F]/kryptofix and the mixture was heated at 150 °C for 25 min. The reaction mixture was cooled and passed through a silica gel cartridge (waters, 900 mg) and eluted with 30% methanol in dichloromethane (2.5 mL). After evaporating of the solvent under a stream of argon at 60-80 °C, the residue was redissolved in the appropriate HPLC solvent and purified by semipreparative HPLC.
Authentication of the radioactive tracers. The radioactive peak was detected with a radioactivity detector co-injected with the relevent authentic cold compound which was detected with ultraviolet detector (250 nm) using analytical HPLC. The retention times for the radiotracers are: 18 F-3 was eluted at 5.93 with 60% acetonitrile/water solution. 18 F-4 was eluted at 6.0 with 60% acetonitrile/water solution. 18 F-19 was eluted at 6.8 with 70% acetonitrile/water solution. 18 F-20 was eluted at 6.2 with 70% acetonitrile/water solution. 18 F-1 was eluted at 1.8 with 50% acetonitrile/water solution. 18 F-2 was eluted at 1.8 with 50% acetonitrile/water solution.
Partition coefficient (logD). Log D for [ 18 F]TMP195 was determined using the method similar to our previous work. We determine the log D by partitioning the tracer between octanol and phosphate buffer at pH 7.4 and measuring the concentration of the tracer in each layer. The radioactivity of each layer is counted using gamma-counter. The partition coefficient (P) is calculated as [radioactivity (cpm/mL) in 1-octanol)]/[(radioactivity (cpm/mL) in phosphate buffer pH 7.4]. Molar activity. The specific activity was determined from the area under the curve of the tracer that is attributed to the ultraviolet peak in the HPLC chromatogram against a calibration curve pre-prepared with the unlabeled reference standard. Molar activity of our tracers ranged from 0.33 to 0.49 GBq/umol (8.9-13.4 mCi/ umol) which is remarkably high for [18F]trifluoromethyl moiety. It is important to note that we are currently purchasing F-18 from a commercial source with significant decay prior to the start of our radiochemical experiment (> 2 half-lives). We expect to obtain significantly higher molar activities for our tracers once our cyclotron becomes operational. Furthermore, automated synthesis is expected to further improve the radiochemical yield and molar activity due to more efficient synthesis and shorter overall production time. Moreover, starting with a higher amount of radioactivity may also further improve the yield and molar activities.