In eukaryotic cells, mitochondria are crucial organelles for oxidative metabolism and adenosine triphosphate (ATP) production (Kadenbach, 2012). Mitochondrial respiratory chain complexes (I-V) of the inner membrane of mitochondria mediate a cascade of electron transfer for oxidative phosphorylation (OXPHOS) and aerobic adenosine triphosphate (ATP) (Kadenbach, 2012). The largest complex of the electron transport chain (ETC) is mitochondrial respiratory chain complex 1 (MC-1; NADH-ubiquinone oxidoreductase, EC 1.6.5.3), which catalyzes electron transfer from NADH to ubiquinone, translocating four protons through complexes I-V (Parey et al., 2020). MC-1 plays a crucial role in mitochondrial function (Tsukada et al., 2014b) and its dysfunction is linked to physiological brain aging (Tsukada et al., 2014b), kidney malfunctions (Saeki et al., 2020), and ischemic stroke (Suzuki et al., 2021). Human post-mortem and animal model data link MC-1 dysfunction to Alzheimer’s and Parkinson’s diseases (Tsukada, 2016; Kanazawa et al., 2017; Barron et al., 2020; Fang et al., 2020) Even a small loss of MC-1 activity significantly affects ATP synthesis and mitochondrial respiration in brain mitochondria (Pollard et al., 2016; Wilson et al., 2020). For better understanding of the metabolic role of MC-1 in the pathology of various diseases, positron emission tomography (PET) imaging of MC-1 is of interest and several MC-1 imaging agents are reported (Figure 1).

The University of Michigan PET Center conducted work with ¹¹C-labeled derivatives of rotenone in the past (Charalambous et al., 1995a,b; Kilbourn et al., 1997; Snyder et al., 1999), but these agents were not widely used, likely because of toxicity considerations. More recently, a pyridazinone analog, (¹⁸F)BMS-747158-01 (Flurpiridaz), which advanced to clinical trials as a myocardial perfusion imaging agent (Sattler et al., 2014), showed inhibitory activity for MC-1 function (Rokugawa et al., 2017). Studies with the agent demonstrated high uptake and long retention not only in the heart (Werner et al., 2019), but also in the brain. (¹⁸F)BMS-747158-01, however, exhibited high non-specific binding both in vivo and in vitro in the brain compared with the specific MC-1 known inhibitor rotenone (Maddahi et al., 2011).

Further work aimed at tuning the CNS imaging properties led to three additional analogues: BCPP-EF, -BF, and -EM (Figure 2) (Harada et al., 2013) Our interest was directed towards BCPP-EF and -BF, given the use of fluorine-18 (t_1/2 = 109.8 min) versus carbon-11 (t_1/2 = 20.4 min) for radiolabelling. Labelling with fluorine-18 offers better signal to background, potential additionial time points for study with the longer half-life, and easier distribution to off-site PET imaging facilities. Another issue associated with BCPP-EM was radiolysis at high molar activity (332 ± 228.5 GBq/μmol), which resulted in low radiochemical purity (Harada et al., 2013). The highly lipophilic nature of BCPP-BF (logD_7.4 4.2) resulted in high non-specific binding (34%), and increased bone uptake was observed with the agent (Harada et al., 2013). Based on these previous studies, it was determined that BCPP-EF (1) demonstrated promising quantitative imaging of MC-I activity and its ischemic damage in the living brain with PET without confounding microglial activation (Tsukada et al., 2014a). BCPP-EF exhibited greater sensitivity towards rotenone and lower non-specific binding (4%) (Harada et al., 2013). We decided to synthesize BCPP-EF and fully automate the method for clinical use at our PET Center. We report the first automated synthesis of (¹⁸F)BCPP-EF using a green radiochemistry approach (Shao et al., 2014; Stewart et al., 2015) that provides radiotracer with good radiochemical yield, excellent radiochemical purity, and high molar activity. We also provide quality control testing results that confirm the product is suitable for clinical use.

3.1.1. Chemistry. Based on previous reports (Nishiyama et al., 2014; Tsukada et al., 2018), the synthetic plan for BCPP-EF (1) was laid out, but steps were modified as needed to improve the synthesis (Harada et al., 2013). The retrosynthetic analysis of 1 resulted in three major fragments 2, 3, and 4 (Figure 4). Following literature reports, the synthesis of fragment 2 was achieved by treating mucochloric acid with tert-butylamine.hydrochloride in water with sodium carbonate (Harada et al., 2013). The generated hydrazone intermediate was subjected to acetic acid reflux conditions and resulted in dichloropyridazine analogue 6 in 59% yield. The intermediate 6, under basic hydrolysis, resulted in partial hydrolyzed fragment 2 in 87% yield (Figure 5A).

Following literature reports, we started to build fragment 3 by treatment of diethylene glycol with dihydropyran using p-toluenesulfonic acid monohydrate as an acid catalyst at −10°C in 15% yield (Figure 5B). Next, considering the synthesis of fragment 4, reaction conditions were modified from the reported route to have an improved and safe procedure (Harada et al., 2013). 6-Chloropyridin-3-yl-methanol 9 was treated with NaH to generate the alkoxide anion at room temperature. After evolution of H₂ stopped, a solution of 1-chloro-3-methyl-2-butene (10) in DMF was added into the reaction mixture. The reaction was heated at 50°C for 24 h and purified to isolate the desired fragment 4 in 26% yield (Figure 5C).

With fragments 2, 3, and 4 in hand, we started the synthesis of labeling precursor 15 and unlabeled BCPP-EF reference standard 1. In the original report (Harada et al., 2013), the coupling of 3 and 4 was conducted under microwave conditions. However, to make this synthesis more accessible we investigated the coupling of 3 and 4 under conventional heating conditions. To begin with, fragments 3 and 4 were dissolved in DMF and NaH was added. The reaction mixture was originally heated at 60°C for 6 h which resulted in desired product 11, albeit at low (26%) yield (Figure 6). In an effort to improve the yield of the 11, we increased the reaction temperature from 60 to 100°C and extended the reaction time to 12 h. To our delight, this resulted in an increase in the yield of 11 from 26 to 64%. Next, we deprotected the 1-chloro-3-methyl-2-butyl group from 11 with potassium tert-butoxide in DMSO at 60°C for 40 min, generating deprotected alkyl alcohol 12 at 75% yield. In literature precedents, we found that coupling of 12 with fragment 2 was carried out using Mitsunobu conditions. Unfortunately, in our hands the Mitsunobu conditions provided a low yield of 13, and the product was difficult to purify. In previous work on a similar scaffold we had employed the Tsunoda reagent (cyanomethylenetributylphosphorane) (Kaur et al., 2021), and were gratified to observe that treatment of 12 with fragment 2 and the Tsunoda reagent generated intermediate 13 at 76% yield.

Next, the THP group of protected product 13 was removed under acidic conditions using p-toluenesulfonic acid hydrate in methanol. This generated the desired deprotected product 14 at 80% yield. To access the radiolabeling precursor, the hydroxyl group of intermediate 14 was tosylated with p-toluene sulfonyl chloride in DCM which generated 15 at 79% yield (Figure 6). Finally, to synthesize unlabeled BCPP-EF reference standard 1, the tosylate 15 was reacted with TBAF at room temperature to give 1 at 49% yield.

3.1.2. Radiochemistry. To date there are no reports regarding an automated synthesis of BCPP-EF using a commercial radiochemistry synthesis module. To facilitate clinical studies under good manufacturing practice (GMP) we needed to develop an automated synthesis of (¹⁸F)BCPP-EF. We began by dissolving precursor 15 (10 mg) in anhydrous acetonitrile (1,000 μL) and reacted with azeotropically dried (¹⁸F)fluoride at 80°C for 10 min (Table 1, entry 1) as had been previously described. The crude reaction mixture was purified by semi-preparative HPLC using a perfluorophenyl-capped matrix column (Luna-PFP(2), Phenomenex) with an acetonitrile-based mobile phase (40% acetonitrile, 20 mM NH₄OAc, 0.2% acetic acid, 4 ml/min). ( ¹⁸ F)1 (t_R = 12–15 min) was isolated at 46.3% RCY and at RCP of 100%.

As we considered clinical use of [¹⁸F]BCPP-EF further, we were curious if a green radiosynthesis could be developed in line with our previous work in this area (Shao et al., 2014; Stewart et al., 2015). The underlying premise of green (radio)chemistry is the design of products and processes that minimize or eliminate the use and generation of hazardous substances (U.S. EPA, 2021). In the context of fluorine-18 radiochemistry, this mainly involves replacing more toxic class 2 solvents (e.g., MeCN) with less toxic class 3 solvents (e.g., ethanol, DMSO). An additional benefit is that for radiosyntheses employing only class 3 solvents, residual solvent analysis does not need to be conducted on every batch but instead can be reduced to an annual quality control test (Shao et al., 2014; Stewart et al., 2015). In order to avoid the use of class 2 solvents in the synthesis of [¹⁸F]BCPP-EF, we investigated whether acetonitrile (injection limit in humans 4.1 mg/day) could be replaced with ethanol (50 mg/day) for azeotropic drying of [¹⁸F]fluoride and in the HPLC mobile phase, and with DMSO (50 mg/day) for the reaction solvent. We replaced K₂₂₂ in acetonitrile (15 mg/ml in MeCN) for azeotropic drying with K₂₂₂ in ethanol (15 mg/ml in EtOH) as we have done in prior work (Stewart et al., 2015), since both the MeCN/H₂O and EtOH/H₂O azeotropes have similar boiling points and are interchangeable in our experience. Secondly, the reaction solvent was changed from MeCN to DMSO, another common solvent for ¹⁸F-fluorination reactions (Table 1, entry 2). Lastly, the crude reaction mixture was once more purified using a perfluorophenyl-capped matrix semipreparative column (Luna-PFP(2), Phenomenex) but instead using an ethanolic mobile phase (50% ethanol, 50 mM NH₄OAc, 0.2% acetic acid, 3 ml/min). To our delight, the (¹⁸F)BCPP-EF was obtained in 24.3% yield (Table 1, Entry 2) and in 97% RCP. It was found that reducing the amount of precursor from 10 to 5 mg did not decrease the yield of the desired product to a significant extent (Table 1, entry 2 vs. Table 1, entry 3). Thus a fully automated synthesis using 5 mg of precursor provided 15.99 ± 3.09 GBq (432.2 ± 83.6 mCi) of [ ¹⁸ F]1 in 21.6% decay-corrected RCY [n = 4, based upon 2.0 Ci of starting [¹⁸F]fluoride], >98% radiochemical purity and high molar activity [242.4 ± 105.7 GBq/μmol (6,521 ± 2,843 Ci/mmol)].

For reformulation, the purified [ ¹⁸ F]1 was trapped on a C18 (Waters, 1 cc vac) cartridge, and the cartridge was rinsed with water to remove the residual HPLC solvent/buffers. Different conditions were explored to find the best conditions for elution. We compared elution of the product with ethanol (0.5 ml), vs. ethanol/tween 80 with or without ascorbic acid (Table 2). All of the formulation conditions worked well to elute the product, did not show any evidence of radiolytic decomposition and maintained radiochemical purity of greater than 95% after 5 h. In our hands, there was no appreciable benefit to inclusion of tween 80, used in the original synthesis (Harada et al., 2013). Moreover, the formation of white precipitate was observed in ethanol/tween 80/ascorbic acid formulated dose after 2 h (Table 2, Entry 3). As such, we omitted the use of Tween 80 to simplify the final formulation as it was not required.

Lastly, encouraged by these results, we validated the synthesis of [ ¹⁸ F]1 to confirm suitability for production of batches for clinical use. Three process verifications runs were completed and produced [¹⁸F]1 at 154 ± 31 mCi activity yield [7.7 ± 1.6% non-decay corrected radiochemical yield based upon 2.0 Ci of [¹⁸F]fluoride], >95% radiochemical purity and molar activity of 6,801 ± 2,154 Ci/μmol (251,637 ± 79,726 GBq/μmol) (Table 3). Synthesis time was 61–65 min from EOB. Quality control testing of [¹⁸F]BCPP-EF 1 was conducted according to the guidelines outlined in Chapter <823> of the US Pharmacopeia, and previously reported QC procedures (Supplementary Material). As shown in Table 3, batches 1–3 met all acceptance criteria confirming suitability for future clinical imaging studies. HPLC analysis confirmed batches were stable for at least 1 h which is sufficient for our needs. However, we found evidence of slow radiolysis such that by 3 h post-end-of-synthesis, RCP had dropped to 87%. To remedy this issue, we conducted a fourth validation run in which ascorbic acid was included in the final formulation. This inclusion allowed extension of the product shelf life to at least 8 h.

A fully automated radiosynthesis of [¹⁸F]BCPP-EF for PET imaging of MC-1 was developed. The highlights of the current method are its straightforward chemistry, simplicity, good radiochemical yields, and compatibility with commercially available radiochemistry synthesis modules for automated production of the radioligand at excellent radiochemical purity and high molar activity. Validation batches confirmed that batches of [¹⁸F]BCPP-EF produced using this method are suitable for use in clinical PET studies.