Direct Radiofluorination of Arene C–H Bonds via Photoredox Catalysis Using a Peroxide as the Terminal Oxidant

Abstract

Herein, we describe an organic photoredox system for direct arene C–H radiofluorination using a peroxide oxidizing agent and LEDs as the light source. In conjunction with an optimized photocatalyst and a microtubing reactor, this system is applicable to a range of electron-rich aromatics and heteroaromatics. We also demonstrate the feasibility of C–H radiofluorination without an azeotropic drying step, which greatly simplifies the workflow of labeling process.

Graphical Abstract

Positron emission tomography (PET) is an important imaging modality that plays key roles in the biomedical field including disease diagnosis, prognosis, treatment monitoring and drug development.¹ One commonly used approach to generate novel contrast agents for PET is to radiolabel pharmaceuticals with known activity towards biological processes or targets of interest. Because fluorine-18 (¹⁸F) is the most widely used PET isotope, a significant amount of effort has been devoted to developing robust methods for radiofluorinating small-molecule pharmaceuticals.²

Traditionally, electron-deficient aromatic arenes are fluorinated via nucleophilic aromatic substitution.³ Recently, deoxyfluorination,^4–7 demetalative fluorination,⁸ copper catalyzed cross-coupling,^9,10 and substitution of iodonium intermediates^11–14 have been developed to radiofluorinate a larger spectrum of arenes. Comparison of these different strategies to C–H fluorination methods and their future challenges in the field are well discussed in the literature.¹⁵ In our labs, we have established a program studying direct C–H radiofluorination catalyzed by organic photooxidation catalysts under mild reaction conditions.¹⁶ Using laser irradiation to excite an acridinium photocatalyst, unactivated arenes undergo photoinduced electron transfer and the resultant cation radicals react with nucleophilic [¹⁸F]F^–. Molecular oxygen bubbled through the system then regenerates the ground state photocatalyst and induces rearomatization of the [¹⁸F]fluoride-arene cation radical adducts, generating radiolabeled products (See Supporting Information, Figure S28). While this approach offers a simple and efficient late stage radiofluorination method, it required an expensive laser light source and the bubbling of oxygen through the reaction vessel is difficult to operate in various cases. Aiming to make the method widely available to the field, we developed a photoredox system and apparatus that allows for direct C–H radiofluorination using a readily available LED light source and an organic peroxide as the terminal oxidant (Scheme 1).

Scheme 1.

Direct C–H radiofluorination via photoredox catalysis.

In order to replace the laser irradiation source with LEDs, the overall light influx needed to be significantly increased. Inspired by flow chemistry and microfluidic design, we first made a microtubing reactor that greatly increased the surface area being exposed to the light source. Unfortunately, performing the reaction in an enclosed microtubing reactor renders oxygen bubbling impractical. A screen of commonly used oxidizing agents was then performed in an attempt to replace oxygen as the terminal oxidant, the results of which are summarized in Table 1. Using diphenyl ether as the model substrate, the reaction mixture was incubated with various oxidants and [¹⁸F]TBAF under LED irradiation and the reactions were analyzed by radio-TLC. No desired product was detected when benzoyl peroxide (BPO) or pyridinium chlorochromate (PCC) were applied as the oxidant, and only trace amounts of product were detected when PhI(OAc)₂ or H₂O₂ were employed. tert-Butyl peroxyacetate (TBPA) was the second best performing oxidant among those tested, with a radiochemical yield (RCY) of 15.7%. tert-Butyl peroxybenzoate (TBPB) and tert-Butyl hydroperoxide (TBHP) were less reactive compared with TBPA, with a yield of 9.7% and 7.4%, respectively. Potassium permanganate (KMnO₄) proved to have the highest yield (19.2%) in our initial oxidant screen, which may be the result of highly reactive singlet oxygen generated by KMnO₄ during irradiation.^17,18 However, KMnO₄ has limited solubility in the reaction system, leading to difficulties in sample loading and yield inconsistencies. Despite slightly lower yields than KMnO₄, liquid oxidants were preferred due to ease of handling. Therefore tert-butylperacetic acid (TBPA) was chosen for further optimization. Solvents such as t-BuOH, acetonitrile, DMSO, toluene, dichloromethane, tetrahydofuran, N,N-dimethylformamide, 1,4-dioxane and methanol were evaluated and the results are shown in Table S2. After this round of screening, t-BuOH was identified as the optimal solvent with 1 equivalent of TBPA as terminal oxidant.

Table 1.

Screening of Oxidants for Photoredox Radiofluorination

Entry	Wavelength	Catalyst	[O]	RCY
1[a]	450 nm	Cat-1	TBPB	9.7%[b]
2[a]	450 nm	Cat-1	TBPA	15.7%[b]
3[a]	450 nm	Cat-1	BPO	N.D. [b]
4[a]	450 nm	Cat-1	TBHP	7.4%[b]
5[a]	450 nm	Cat-1	H2O2	2.4%[b]
6[a]	450 nm	Cat-1	PhI(OAc)2	0.5%[b]
7[a]	450 nm	Cat-1	KMnO4	19.2%[b]
8[a]	450 nm	Cat-1	PCC	N.D. [b]
9	365 nm	Cat-32	TBPA	7.7%[c]
10	385 nm	Cat-32	TBPA	4.6%[c]
11	410 nm	Cat-32	TBPA	12.6%[c]
12	425 nm	Cat-32	TBPA	20.2%[c]
13	450 nm	Cat-32	TBPA	17.4%[c]

^[a]Diphenyl ether (0.005 mmol), catalyst (0.00025 mmol), [O] 0.005 mmol, TEMPO (0.0025 mmol), [¹⁸F]TBAF in MeCN (0.5~1.5 mCi), and ^tBuOH (40 ul). The reaction mixture was then loaded on the capillary, sealed, then irradiated under LED 450 nm for 40 min at 0 °C.

^[b]Radiochemical yields (RCY) were calculated based on radio-TLC analysis with an eluent of ethyl acetate/hexane (v/v=1/20) on silica gel 60 aluminium plate.

^[c]2-Methoxybenzaldehyde (0.1 mmol), catalyst (0.025 mmol), [O] 0.05 mmol, TEMPO (0.025 mmol), MeCN 100 μl and ^tBuOH (40 μl). The reaction mixture was loaded into the capillary, sealed, and irradiated with a 450 nm LED for 40 min at 0 °C. Isolated RCYs were calculated by radio-HPLC.

Although photocatalyst 1 (Cat-1) provided the best radiolabeling results in our previous report, the availability of recently developed organic photocatalysts prompted us to perform a thorough catalyst screen. A library of 48 organic photocatalysts¹⁹ was evaluated (Figure S1) and the results are shown in Table S1. Generally, acridinium salts were the most suitable catalysts for radiofluorination. No desired products were detected when xanthylium salts (Cat-21 to Cat-31) or 2,4,6-triphenylpyrylium salts (Cat-13 to Cat-18) were applied as catalysts. Cat-32 provided the best results with an RCY of 42.4% (Table S1). No significant difference was observed when the reaction was carried out at room temperature or 40 °C. Because the catalyst has a wavelength-dependent extinction coefficient, we also evaluated the effect of LED wavelength on reaction yield. The UV-Vis spectrum of Cat-32 (Figure S24) revealed strong absorption at 365, 385, and 420 nm. Consequently, we performed a side-by-side comparison of LED lights with 365, 385, 410, 425, and 450 nm wavelength output (Table 1, entry 9–13). The optimal labeling yield was obtained with the 425 nm LED light (20.2%, compared with 23.5% yield using 3.5 W laser).

Using the optimized conditions, the scope of this radiofluorination method was evaluated and the results are shown in Scheme 2. Twenty-two examples were examined in this work. Eleven of the substrates demonstrated improved labeling yields and ten of them showed comparable or decreased yield compared with our previous report (which are listed in parentheses). We would also like to point out that all the yields in Scheme 2 are fully purified isolated RCYs. For compound 16 (4-chromanone), which is a key motif in many anticancer agents,²⁰ our new method underwent radiofluorination with 20% isolated RCY. In the previous method, no desired product was detected and many byproducts were observed in the reaction mixture. This improvement suggests that our LED system may be more suitable for substrates that require milder reaction conditions. Slightly improved yields were also achieved for heterocyclic substrates such as N,N-dihexylquinazolinedione (17) benzoxazole (18), and benzimidazole (19). The purified major products were structurally characterized by co-injection analysis on HPLC with corresponding ¹⁹F-standards (details in SI, Figures S2–S23). The radiofluorination regioselectivity observed in this study generally follows the same trends observed in previous photoredox arene C–H functionalization chemistry reported by Nicewicz and co-workers, with functionalization taking place ortho or para to an electron donating group.^{21, 22}

Scheme 2.

Scope of arene C–H radiofluorination and their radiochemical yields. The RCYs of our previous system are listed in parentheses (for which O₂ was used as the oxidant and a laser was used as the light source).

Having evaluated the scope of this radiofluorination, we sought to further simplify the labeling procedure by eliminating the azeotropic drying step in the preparation of [¹⁸F]TBAF. In brief, the target water containing ¹⁸F was directly trapped on a pre-activated mini-QMA cartridge. Anhydrous acetonitrile (5 mL) was then passed through the mini-QMA cartridge to wash out the residual water. After evaluating a series of elution conditions, we found that a mixture of TBAB solution (25 μl, 1.5 mg in MeCN), substrate, catalyst and oxidant solution in t-BuOH can efficiently wash out [¹⁸F]TBAF. The reaction mixture was then loaded in a quartz microtube and irradiated with LEDs at room temperature. After the reaction, the activity was collected in a 1.5 mL microcentrifuge tube and analyzed by radio-HPLC (Scheme 3). With this method in hand, we evaluated the radiofluorination of protected FDOPA precursor 23, affording the desired product [¹⁸F]-22 in 22.8% isolated RCY.

Scheme 3.

Schematic workflow of preparing ¹⁸F labeled agent through direct C–H fluorination and its application in protected ¹⁸F-FDOPA synthesis.

In summary, we have developed an efficient and user-friendly photoredox system for rapid and direct arene C–H radiofluorination, obviating the need for pre-functionalized starting materials. The mild reaction conditions, use of a relatively low energy and affordable LED light source, and readily available components will enable the general use of this methodology for the synthesis of novel ¹⁸F-labeled radiotracers.

Abbreviations

TEMPO

2,2,6,6-tetramethyl-1-piperidine 1-oxyl

TBAF

Tetrabutyl ammonium fluoride

^tBuOH

tert-Butanol

MeCN

Acetonitrile