Cu-Mediated Aminoquinoline-Directed Radiofluorination of Aromatic C–H Bonds with K¹⁸F

Abstract

This communication describes a Cu-mediated ortho-C–H radiofluorination of aromatic carboxylic acids that are protected as 8-aminoquinoline benzamides. The method uses K¹⁸F and is compatible with a wide range of functional groups. The reaction is showcased in the high specific activity automated synthesis of the RARβ2 agonist [¹⁸F]AC261066.

Graphical Abstract

This communication reports a method for the Cu-mediated ortho-C(sp²)-H radiofluorination of aromatic carboxylic acids that are protected as 8-aminoquinoline benzamides with K¹⁸F. Fluorination of 18 examples in up to 62% RCC and high specific activity is reported, including the automated synthesis of [¹⁸F]AC261066 (after removal of 8-aminoquinoline auxiliary).

Aryl fluorides are widely prevalent in pharmaceuticals,^1,2 and their ¹⁸F isotopologs are important for positron emission tomography (PET) imaging.^3,4 As such, there is significant interest in methods for the late-stage ¹⁸F-fluorination of aromatic scaffolds.^5,6 The majority of existing methods for arene radiofluorination require a prefunctionalized starting material. For instance, hypervalent iodine reagents,^6c organoborons,^6d-f,j,k organostannanes,^6g Ni/Pd complexes,^6a,b and phenols^6h,i have recently been introduced as precursors for nucleophilic radiofluorination reactions. However, this requirement for pre-installed functionality at the target site can be a roadblock for the application of these methods to complex radiotracer targets.

A complementary approach would involve the direct radiofluorination of a C–H bond of an arene substrate. Several strategies have been developed for the radiofluorination of aliphatic⁷ and benzylic⁸ C–H bonds. However, analogous transformations of C(sp²)–H substrates have proven considerably more challenging. While C(sp²)–H radiofluorination can be accomplished via electrophilic aromatic substitution (S_EAr) with [¹⁸F]F₂ or [¹⁸F]Selectfluor,⁹ the generation and handling of these reagents requires specialized equipment that is not widely accessible. Additionally, the site- and chemoselectivities of S_EAr reactions are typically modest, and the final products generally have low specific activity.¹⁰ In principle, these limitations could be addressed through the development of nucleophilic (¹⁸F^–) C(sp²)–H radiofluorination methods. However, in practice, realization of this approach has remained elusive due to the inertness of C(sp²)–H bonds and the electronic mismatch between nucleophilic ¹⁸F^– and most arene substrates.¹¹

An attractive strategy to address these challenges would be to leverage modern advances in transition-metal catalyzed C(sp²)–H functionalization. For example, recent work by Daugulis demonstrated that 8-aminoquinoline directing groups enable Cu-catalyzed ortho-C(sp²)–H activation/nucleophilic fluorination reactions with AgF.¹² The directing group is easily cleaved, thus providing access to ortho-fluorinated carboxylic acids. This communication describes translation of this method to a radiofluorination process. While AgF was required in Daugulis’ original transformation, our studies reveal that K¹⁸F is optimal for radiofluorination. This nucleophilic radiofluorination of aromatic C–H bonds is applied to a variety of carboxylic acid derivatives and automated to access high specific activity radiotracers.

We initially examined the Cu-catalyzed radiofluorination of aminoquinoline substrate 1H with Ag¹⁸F¹³ under conditions closely analogous to those reported by Daugulis¹² (1H (20 μmol), CuI (5 μmol), N-methylmorpholine N-oxide (NMO, 90 μmol), K_2.2.2 (1.33 μmol), Ag¹⁸F (2500–3500 μCi) in DMF). However, these conditions did not afford detectable quantities of 1¹⁸F as determined by radio-TLC and radio-HPLC analysis (Table 1, entry 1). Notably, the Ag¹⁹F likely serves two roles in the original Daugulis reaction. First, it acts as the nucleophile to install the C(sp²)–F bond. Second, it serves as a base to sequester the proton that is generated during C–H activation. Since Ag¹⁹F is present in 3–4-fold excess relative to 1H, there is sufficient fluoride available for both of these functions. In contrast, under the radiofluorination conditions, the Ag¹⁸F is the limiting reagent. We hypothesized that an exogeneous base might be needed to sequester protons while preserving a reservoir of nucleophilic fluoride for the desired C(sp²)–F coupling reaction. Consistent with this hypothesis, the addition of 1,8-diazabicyclo(5.4.0)undec-7-ene (DBU) (20 μmol, 1 equiv relative to 1H) led to the formation of the desired product 1¹⁸F in 26 ± 1% RCC as determined by radio-TLC and confirmed by radio-HPLC (Table 1, entry 2).¹⁴ Further optimization revealed that switching from CuI to more soluble (MeCN)₄CuOTf resulted in a slightly improved RCC (29 ± 0%; Table 1, entry 3). Under these conditions, the [¹⁸F]fluoride source could be changed to readily accessible K¹⁸F³ to afford 33 ± 0% RCC of 1¹⁸F (Table 1, entry 4).

Table 1.

Optimization of C-H Radiofluorination^a

Entry	[Cu]	M18F	NMO	DBU	RCC (%)b
1	CuI	Ag18F	✓	--	nd
2	CuI	Ag18F	✓	✓	26 ± 1
3	(MeCN)4CuOTf	Ag18F	✓	✓	29 ± 0
4	(MeCN)4CuOTf	K18F	✓	✓	33 ± 0
5	(MeCN)4CuOTf	K18F	--	✓	31 ± 13
6c	(MeCN)4CuOTf	K18F	--	✓	50 ± 2

^[a]Conditions: 1H (20 μmol), Cu source (5 μmol), additives [NMO (90 μmol), K_2.2.2 (1.33 μmol), DBU (20 μmol)], M¹⁸F (2500-3500 μCi), DMF (1000 μL).

^[b]RCC was determined by radio-TLC (n ≥ 3); nd = not detected. The identity of 1¹⁸F was confirmed by radio-HPLC.

^[c]NMM added (90 μmol).

We next examined whether NMO is necessary for this transformation. In the Ag¹⁹F reaction (which is conducted under inert atmosphere), NMO acts as the terminal oxidant for Cu. However, the radiochemical reactions are conducted under ambient air, which could directly oxidize the Cu. Indeed, excluding NMO from the Ag¹⁸F reaction under otherwise identical conditions resulted in a comparable RCC (31 ± 13%, entry 5),¹⁵ although it did negatively impact the run-to-run reproducibility. We evaluated a number of additives to address this latter issue and found that the use of 90 μmol of N-methylmorpholine (NMM), the base counterpart of NMO, resulted in enhanced reproducibility as well as an improved RCC of 50 ± 2% (Table 1, entry 6).

The scope of this reaction was examined using aminoquinolines derived from a variety of substituted benzoic acids.¹⁶ As shown in Figure 1, electron-neutral (1¹⁸F-4¹⁸F), -withdrawing (5¹⁸F-10¹⁸F),¹⁷ and -donating (11¹⁸F) substituents were tolerated on the arene ring. Many functional groups, including benzylic C–H bonds, trifluoromethyl, cyano, nitro, ester, amide, and sulfonamide substituents, were compatible. This C(sp²)–H radiofluorination was also effective on pyridine- and indole-derived substrates, providing 12¹⁸F and 13¹⁸F in moderate RCC. A substrate containing a fluorine substituent at the activated 4-position on the quinoline reacted to afford the ortho-¹⁸F-labelled product 14¹⁸F in 50% RCC.¹⁸ This method was applied to the late-stage radiofluorination of a series of biologically relevant molecules. Four carboxylic acid-containing drugs, probenecid, ataluren, tamibarotene, and AC261066, were converted to the corresponding 8-aminoquinoline benzamides and then subjected to the optimal conditions. The [¹⁸F]fluorinated analogues (15¹⁸F-18¹⁸F, respectively) were obtained in 13–37% RCC.¹⁹

Figure 1.

Substrate Scope. Reported values indicate radiochemical conversion (RCC) determined by radio-TLC for n ≥ 4 runs. The identity of all products was confirmed by radio-HPLC. General conditions: Substrate (20 μmol), (MeCN)₄Cu(OTf) (5 μmol), NMM (90 μmol), K_2.2.2 (1.33 μmol), DBU (20 μmol), K¹⁸F (2500-3500 μCi), DMF (1000 μL), 90-110 °C, 30 min. [a] in cases where other products were observed by radio-HPLC analysis, RCCs from radio-TLC analysis were corrected as described in the Supporting Information.¹⁷

A final set of experiments involved automation of this reaction on a TRACERLab FX_FN synthesis module and hydrolysis of the aminoquinoline protecting group (Scheme 1). Initial automated studies were conducted with 1H, and afforded 1¹⁸F in 28 ± 6% (n = 6) automated RCC or, by incorporating semi-preparative HPLC purification, 9 ± 4% (n = 6) isolated decay-corrected radiochemical yield (RCY) and >98% radiochemical purity (RCP). Starting with 1.7 Ci of [¹⁸F]fluoride 1¹⁸F was obtained in 42 ± 3 mCi (n = 3) with high specific activity (6 ± 1 Ci/μmol). Hydrolysis of the aminoquinoline protecting group was then achieved with 4 M NaOH to afford 19¹⁸F in 90 ± 2% RCC from 1¹⁸F (n = 3) and 21 ± 2% RCC based upon starting [¹⁸F]fluoride.

Scheme 1.

Automated C(sp²)-H Radiofluorination

An analogous method was applied to the synthesis of [¹⁸F]AC261066 (20¹⁸F), a RARβ2 agonist (Scheme 1).²⁰ Subjecting 18H to the C–H radiofluorination conditions afforded 18¹⁸F in 12 ± 2% automated RCC (n = 3). Starting with 1.7 Ci of [¹⁸F]fluoride, 18¹⁸F was obtained in 36 ± 8 mCi (n = 3) after sep-pak purification, corresponding to 3 ± 1% isolated decay-corrected RCY. Manual hydrolysis of the amide with 4 M NaOH formed [¹⁸F]AC261066 (20¹⁸F) in 98 ± 1% RCC from 18¹⁸F (n = 5, determined by radio-TLC). Overall, the isolated decay-corrected RCY of 20¹⁸F was 9 ± 7 mCi (2 ± 1% based upon starting [¹⁸F]fluoride, n = 3). The product was obtained in high chemical and radiochemical (>98%) purity and high specific activity (0.80 ± 0.25 Ci/μmol).²¹

In summary, we describe the Cu-catalyzed, aminoquinoline-directed C(sp²)–H radiofluorination of arene C(sp²)–H bonds with K¹⁸F.²² The method has been applied to a variety of substrates, including the active pharmaceutical ingredients of probenecid, ataluren, and tamibarotene. In addition, it has been translated to an automated synthesis of high specific activity doses of RARβ2 agonist [¹⁸F]AC261066. We note that the automated radiochemical yields and directing group cleavage procedures will require additional optimization before they can be applied in routine radiosyntheses. In addition, future work should target the use of more practical directing groups as well as non-directed approaches to C–H radiofluorination. However, overall this operationally simple procedure demonstrates proof-of-concept that metal-catalyzed nucleophilic C(sp²)–H radiofluorination is feasible, and that this approach shows promise for the late-stage radiofluorination of bioactive molecules.