Asymmetric 18F-fluorination for applications in positron emission tomography

To date, both transition metal complexes and organomediators have been employed for enantiocontrolled 18F-incorporation as an alternative to conventional resolution of SN2-based radiochemistry.


Introduction
It is universally recognized that molecular chirality has a direct impact on function. In nature, many essential biological molecules exist only in one of two possible mirror-image structures, either because they possess a chiral unit or through their overall structure. 1 In the context of drug development, chirality dominates and it has been accepted since the early 1980s that most of the biological activity observed for a racemate oen resides within a single enantiomer. 2 As a result, it was anticipated that the proportion of racemic new molecular entities (NMEs) would decrease over time and possibly vanish. A recent survey indicates that the number of enantiopure NMEs approved since the mid-1990s has indeed increased, but the development and approval of racemic compounds remains a viable approach. 3 The deeply rooted importance of chirality in the pharmaceutical industry 4 and other areas, such as material science, has encouraged much research in asymmetric synthesis and catalysis, two active elds of modern chemistry. 5 The scientic complexity of drug discovery and the commercial challenges currently facing the pharmaceutical industry have led medicinal chemists to consider Positron Emission Tomography (PET) 6 more frequently as a technology for the identication of the most promising lead compounds much earlier in the drug discovery pipeline. 7 PET is a noninvasive quantitative imaging modality that can be employed to study drug pharmacokinetics and pharmacodynamics, and the relationship of these pharmacological characteristics to the behavioral, therapeutic and toxic properties of drugs. With Faye Buckingham completed her MChem in 2011 at the University of Oxford before joining the group of Prof. V. Gouverneur (University of Oxford, U.K.) to work towards a DPhil in Organic Chemistry. Her research is focused on the development of metal-free approaches for 18 Fradiosynthesis.
Véronique Gouverneur received her PhD in chemistry at the UCL (Belgium) with Prof. L. Ghosez. She moved to a postdoctoral position at the Scripps Research Institute (CA, USA) with Prof. R. A. Lerner. She then returned to Europe with a position at the University Louis Pasteur (France). She started her research career at the University of Oxford (UK) in 1998, and became Professor in 2008. Her research on uorine (radio)chemistry disclosed in more than 160 peer-reviewed publications was recently recognised with the ACS Award for Creative Work in Fluorine Chemistry 2015.
the current trend in the pharmaceutical industry to develop optically pure products, PET can also assess the behavior of individual enantiomers in living systems. 8 For chiral radiopharmaceuticals used in the clinic, the administration of a single enantiomer is benecial in terms of minimizing amount of radioactivity for the patient and, in some cases, by reducing background uptake due to non-specic retention of the inactive enantiomer. The advantageous characteristics of the positron emitting isotope 18 F, 9 and the prominent position of uorine substitution in drug discovery 10 have fuelled an upsurge of interest in 18 F-radiochemistry, with the appearance of novel methods for 18 F-labeling inspired by modern 19 F chemistry. Despite these advances, the production of chiral non-racemic 18 F-labeled drugs remains challenging, especially when the 18 F-tag is located on a stereogenic carbon. Fig. 1 presents the various approaches one may consider for this latter scenario.
Radiochemists would typically consider the separation of 18 F-labeled stereoisomers using High Performance Liquid Chromatography (HPLC) in preference to overcoming the obstacles associated with stereoselective or asymmetric 18 F-uorination; this is despite the fact that such separation leads to a substantial loss of radioactivity (50% loss for the separation of two enantiomers). In this essay, we discuss the challenges associated with 18 F-incorporation onto a stereogenic carbon and the current state of play of this eld of research; in the conclusive remarks, we question how important such developments are for drug developers and PET radiochemists.

18 F-C bond formation and stereogenicity
The slow progress of 18 F-radiochemistry in comparison with 19 F-chemistry is commensurate with the various hurdles associated with 18 F-labeling. Most academic and clinical research laboratories are not equipped to handle the cyclotron produced radioisotope 18 F, a non-trivial limitation preventing fast development in 18 F-radiochemistry. The half-life of the positron emitter 18 F (109 min) imposes time constraints that are not compatible with the lengthy reaction time required for many late stage 19 F-uorinations, and the stoichiometry of 18 Fradiochemical processes may lead to signicant differences in terms of reaction kinetics, in addition to complications for purication. 9 The 18 F source is indeed employed in nano-or picomolar quantity and is therefore in large sub-stoichiometry with respect to the precursor. Furthermore, for radiopharmaceuticals, additional complications may arise during isolation and formulation due to radiolytic decomposition. 11 Another signicant difference between 18 18 Flabeled molecules in higher specic activity, an advantageous property that widens considerably the range of PET studies possible to support drug discovery programs as well as clinical studies. Despite the high strength of the C-F bond (for CH 3 F, BDE ¼ 109.9 kcal mol À1 ), 12 metabolic paths leading to radio-deuorination with release of [ 18 F]uoride are problematic for tracers formulated with high specic activity and concentration, because [ 18 F]uoride binds strongly to the skeletal system. This effect is minimized with 18 F-labeled aryl uorides, which are more stable towards deuorination than alkyl uoride. The intrinsic sp 3 hybridization of stereogenic carbons makes chiral 18 F-labeled tracers susceptible to rapid metabolic degradation through oxidation and/or elimination pathways. Some reports suggest that the use of 18 F-labeled cycloalkyl uoride, and more generally 18 Fincorporation onto secondary instead of primary carbon atoms, typically enhances metabolic resistance; 13 these structurally rened compounds may feature 18 F-C stereogenicity and pose additional radiosynthetic challenges.

The conventional S N 2 approach
The most commonly employed radiotracer in the clinic is 2- , a non-racemic chiral molecule which presents the 18 F-substituent itself on a stereogenic carbon. 14 Its preparation, and more generally 18 F-uorination at a stereogenic carbon, involves S N 2 displacement of a leaving group positioned on an enantiomerically pure precursor (Scheme 1A). The leaving groups typically selected for displacement are triate, tosylate or halide functionalities. Alternatively, one could consider the regio-and stereoselective ring opening of a cyclic sulfate, a reaction that was successfully applied for the radiosynthesis of 16-a-[ 18 This S N 2-based approach typically employs high temperatures and is limited to substrates that are not prone to decomposition under the reaction conditions. Also, since uoride is a potent base as well as a nucleophile, both the substrates and newly formed 18 F-labeled product should be resistant to elimination and racemization (or epimerization) under the 18 F-uorination conditions. The possibility of incomplete inversion and the time-consuming synthesis of enantiopure precursors are further complications associated with this conventional S N 2 strategy. Such challenges are possibly best illustrated with the synthesis of 18 F-labeled 4-uoro-L-glutamine (4F-GLN) and 4-uoro-L-glutamic acid (4F-GLU) (Scheme 2). 16 Protected precursors for the two possible diastereomers for these radiotracers were prepared by nucleophilic displacement with [ 18 F]uoride of the tosyl leaving group installed onto the requisite protected substrates. The challenges imposed by this strategy include the multi-step synthesis and fragile stability of the substrates, the occurrence of stereochemical erosion at C-2, and competitive cyclization upon 18 F-uorination. In vitro and in vivo studies were conducted with 18 F-(2S,4R)4F-GLN and 18 F-(2S,4R)4F-GLU, which are easier to access and to purify than diastereomers 18 F-(2S,4S)4F-GLN and 18 F-(2S,4S)4F-GLU. The development of these demanding 18 F-labeling experiments was however worthwhile, as evaluation studies showed that tumor cell uptake of 18 F-(2S,4R)4F-GLN is higher than that of 18 F-(2S,4R)4F-GLU, likely due to increased amino acid transport activity, protein incorporation, and non-protein metabolic pathways such as glutaminolysis. In contrast, 18 F-(2S,4R)4F-GLU is not incorporated into protein, with the uptake believed to be controlled by the transporter. In vivo studies showed that although 18 F-(2S,4R)4F-GLN exhibited a higher uptake and longer retention in rats bearing 9L tumor xenographs, 18 F-(2S,4R)4F-GLU showed a slightly higher tumor-to-background ratio due to a faster background clearance. Both 18 F-(2S,4R)4F-GLN and 18 F-(2S,4R)4F-GLU are useful as tumor metabolic imaging agents. 17

Asymmetric 18 F-uorination with metals
More recent reports have sidestepped the challenges associated with stereoselective S N 2 substitution, by favouring an alternative approach in which the product stereochemistry is set by an enantioselective 18 F-uorination. In 2011, the demonstration that a transition metal allowed for regioselective allylic 18 F-incorporation opened numerous opportunities towards stereoselective or asymmetric 18 F-uorination directly inspired by research carried out with the non-radioactive isotope 19 21 Doyle reported the enantioselective ring opening of meso and terminal epoxides by uoride, catalysed by (R,R)-Co(salen) and the chiral amine (À)-tetramisole (Scheme 3A). 22 Whilst the previous work had identied that chiral Lewis acid mediated epoxide ring-opening reactions with HF$pyridine complex suffered from low enantioselectivity due to racemic background reaction and catalyst degradation, Doyle discovered that a combination of benzoyl uoride and hexauoroisopropanol (HFIP) provided mild release of uoride, thereby enabling the formation of the uorohydrin product with excellent enantiocontrol. The reaction likely progressed via amine catalyzed in situ slow formation of HF, which in turn could generate the active (salen)Co(III) uorine complex. The desymmetrization of a range of meso epoxides occurred with enantiomeric excesses (ee) reaching up to 95%. In a later report, more detailed mechanistic studies on this reaction led to an improved protocol with low catalyst loading. 23 In 2014, Doyle in collaboration with Kung described extension of this methodology to 18 F-radio-uorination. 24 The [ 18 F](salen)CoF complex was synthesized by reaction of (R,R)-(salen)CoOTs with [ 18 F]uoride eluted from an ion-exchange cartridge, without the requirement for azeotropic drying, and was employed for reaction with epoxide precursors in methyl-tert-butylether (MTBE) to achieve hydrouorination with high RCY within 20 minutes. An excellent level of enantiocontrol was demonstrated, despite an increase of reaction temperature to 50 C. The synthesis of a range of radiotracers was well tolerated. In this report, the majority of substrates were terminal epoxides and only one example in which desymmetrization of a meso epoxide led to a product with a 18 F-substituted carbon stereocenter was disclosed (Scheme 3B). In 2013, Revunov and Zhuravlev took inspiration from the cobalt mediated enantioselective epoxide opening reaction. 25 This work employed [ 18 F]uoride treated with H 2 SO 4 , proposed by the authors to form [ 18 F]HF, which could be trapped by addition of (À)-tetramisole. The reaction took place by addition of (R,R)-Co(salen), HFIP and an epoxide precursor in either MTBE or 2-methyl-2-butanol (tAmOH) as solvent (Scheme 3C). The use of HFIP was found to be crucial to product formation. The reaction was performed at 100 C for 1 hour and afforded 3 examples of [ 18 F]uorohydrin products from cyclic epoxides with good RCY but low ee values, which the authors hypothesized was due to the high reaction temperature.
Manganese mediated benzylic C-H activation. A report by Groves and co-workers in 2014 set out to avoid a pre-functionalized approach to 18 F-labeling with the development of a method which involved the direct replacement of a benzylic sp 3 hydrogen with uorine. 26 This elegant transformation installs uorine substitution onto a stereogenic carbon. In an extension of the 19 F-uorination reactions reported by the same group, 27 this procedure facilitated the 18 F-uorination of a wide range of precursors containing benzylic C-H bonds. Analogously to the cobalt mediated reaction of Doyle, the reaction was proposed to proceed via formation of a [ 18 F](salen)Mn uorine complex, generated by reaction of Mn(salen)OTs with [ 18 F]uoride from an ion-exchange cartridge. In this case, this complex was proposed to undergo oxidation in the presence of iodosobenzene. Reaction at 50 C in acetone for 10 minutes afforded 18 F-labeled products with RCY reaching up to 72%, with an array of functional groups tolerated under these conditions. Starting with 3.5 mCi of [ 18 F]uoride, 18 F-labeled celestolide was obtained in 10% non-decay corrected RCY with a specic activity of 2.68 Ci mmol À1 (end of bombardment). In a singular example of the potential for synthesis of enantioenriched products employing this protocol, the Mn(salen)OTs mediated 18 F-uorination of celestolide afforded the radiolabeled product with an ee of 25% (Scheme 4). Currently the optimization of this reaction in terms of ee has yet to be reported. Nevertheless, this preliminary example highlights the potential opportunity for asymmetric 18 F-uorination with this methodology, which advantageously utilizes nucleophilic [ 18 F]uoride and employs precursors which do not require pre-functionalization with a leaving group.

Organomediated asymmetric 18 F-uorination
Encouraged by the benets associated with organocatalysis, 28 our research group recently reported a metal-free approach 29 to asymmetric 18 F-uorination by taking inspiration from the wealth of literature surrounding organocatalyzed asymmetric uorination. 30 Today, this eld of research is largely limited to processes relying on electrophilic uorination. A notable exception is the asymmetric nucleophilic oxidative uorination of ketoesters and aminouorination of alkenes reported by Shibata and co-workers. 31 An additional remarkable case of metal free catalytic nucleophilic uorination involves the natural uorinase discovered by O'Hagan and co-workers, an enzyme capable of inducing S N 2 substitution with uoride in water, a property exploited for the 18 F-labeling of small molecules and peptides under mild conditions; this enzyme has not been used in the context of asymmetric uorination. 32 Electrophilic 18 F-uorination remains a challenging process for radiochemists due to the narrow range of 18 F + sources available to date and the difficulties associated with their preparation. 33 Nevertheless, 18 F + radiochemistry offers great opportunities in asymmetric 18 F-uorination. An early example of organocatalyzed asymmetric uorination is the chiral amine mediated electrophilic a-uorination of aldehydes, a reaction independently reported by four research groups in 2005. 34 Translation of this reaction to radiouorination posed multiple challenges, especially in terms of avoiding the low temperature conditions and long reaction times associated with organocatalysis. Since a-uoroaldehydes are prone to decomposition and racemization, they are generally further derivatized prior to analysis. For application to radiosynthesis this two-step procedure should ideally occur in one-pot, without time-consuming purication of the intermediate. Our laboratory recently reported that prochiral aldehyde substrates are amenable to asymmetric 18 37 was not a suitable 18 F + source for this transformation. Aer stirring for 20 minutes at room temperature, reagents for derivatization of the enantioenriched 18 F-labeled aldehydes intermediates were added directly. The use of benzhydrazide in methanol afforded the corresponding hydrazone products with good radiochemical conversion (RCC) from [ 18 F]NFSI and ee of up to 92% in a one-pot procedure (Scheme 5).
The utility of a-[ 18 F]uoroaldehyde synthons was further demonstrated with the preparation of an enantioenriched a-[ 18 F]uoro carboxylic acid, primary and secondary amides and a secondary amine product (Scheme 6).
The Pinnick-Lindgren oxidation performed with sodium hypochlorite in acetonitrile in the presence of sodium dihydrogen phosphate and 2-methyl-2-butene led to the desired 18 F-labeled carboxylic acid with no erosion of ee. Oxidative amidation was performed in one pot; this process involved formation of an imine in the rst instance, followed by a Pinnick-Lindgren oxidation affording the desired 18 F-labeled amide with a slightly eroded ee of 83%. A representative enantioenriched 18 F-labeled primary a-uoroamide was also within reach in 49% RCC and 88% ee. Finally, a chiral b-uoroamine was prepared from the enantioenriched aldehyde via imine formation in dichloroethane (DCE) followed by reduction with sodium triacetoxyborohydride.
The method was subsequently applied to access (2S,4S)-4-[ 18 F]uoroglutamic acid (Scheme 7). The 18 F-labeling of an enantiopure aldehyde precursor derived from L-glutamic acid under the optimized conditions, followed by a Pinnick-Lindgren type oxidation and TFA deprotection, afforded 4-(2S,4S)[ 18 F] uoroglutamic acid with very good RCC and d.r.; unlike the S N 2 approach described in Scheme 2, this method did not lead to unwanted epimerization at C-2. 38 Further experiments demonstrated a match/mismatch effect between the chiral imidazolidinone of opposite absolute conguration and the aldehyde substrate; therefore with (R)-amine I, signicant erosion of d.r. was observed.

Conclusion
There is ample evidence in the literature that PET imaging can facilitate the process of drug discovery and development. However, from a pragmatic viewpoint, the routine use of this imaging technology imposes non-trivial radiosynthetic challenges for medicinal chemists, especially when the drug candidate under study is a chiral non-racemic entity. Classical laboratory scale synthesis must be entirely revisited to allow for the nanoscale nature and challenges characteristic of 18 F-radiochemistry necessary to evaluate potential drug candidates in a living system. Following these imaging studies and aer passing all the hurdles of preclinical and clinical evaluation, large-scale production must then be implemented for the chiral non-racemic drugs selected for manufacturing operations (Fig. 2).
Such difficulties would be particularly stringent for chiral molecules with a stereogenic uorinated carbon. These considerations pose a fundamental question on the real value of asymmetric 18 F-uorination for radiotracer production since one could argue that separation techniques of stereoisomers may be more rapid or cost effective; this is assuming that the identication of suitable separation conditions is fast and facile. A related debate arose with the realization that although asymmetric catalysis is an attractive method to control stereochemistry into pharmaceutically active molecules, in practice, other techniques are oen used for pilot plant or production operations. 39 This dichotomy results from industrial constraints for the application of asymmetric catalysis to the large-scale synthesis of drug candidates and commercial drugs. Such constraints include economic considerations, the speed to implement a particular process, the freedom to operate and process robustness. Today, most medicinal chemists would accept that the enormous progress made in asymmetric catalysis will be used by the pharmaceutical industry on a large scale when the catalysts employed are inexpensive, readily available and can be implemented rapidly. Similarly, it is likely that the development of a range of versatile and effective asymmetric 18 F-uorination reactions will be considered by radiochemists in the future as an alternative to separation techniques or conventional radiochemistry requiring multistep synthesis to access the chiral enantiopure precursors necessary for 18 F-uorination. To reach such a situation, it is important that our community continues to develop methods for effective asymmetric 18 F-uorination, employing readily available starting materials and mild reaction conditions, with a protocol that is easy to implement and ideally amenable to automation. The eld of stereoselective and asymmetric labeling has been amply developed for a range of isotopes such as 11 C, 2 H or 3 H. 40 These advances have furthered our understanding of important biochemical processes and have highlighted the differential behavior of enantiomers in living systems. The radioisotope 18 F is now receiving similar attention; this is an encouraging trend that, at a more fundamental level, should increase our understanding of the effect of chirality and F-C stereogenicity on living systems. New asymmetric catalytic uorinations using transition metals, organocatalysts or other catalytic manifolds are being continuously developed in laboratories around the world; many of these methods offer attractive scope and selectivities. These will be used by radiochemists when the eld of asymmetric 18 F-uorination has matured to a point where it is demonstrated that such strategy presents clear advantages over more conventional radiochemistry or separation techniques for rapid in vivo evaluation.