Abstract
Herein, we describe an operationally straightforward radiosynthesis of a chiral transition metal fluoride catalyst – [18F](salen)CoF – and its use for late-stage enantioselective aliphatic radiofluorination. We demonstrate the utility of the method by preparing single enantiomer experimental and clinically validated PET tracers that contain base-sensitive functional groups, epimerizable stereocenters, and nitrogen-rich motifs. Unlike the conventional radiosyntheses of these targets with [18F]KF, labeling with (salen)CoF is possible in the last step and under exceptionally mild conditions. These results constitute a rare example of a nucleophilic radiofluorination using a transition metal fluoride and highlight the potential of such reagents to enhance traditional methods for labeling aliphatic hydrocarbons.
Graphical Abstract

The use of [18F]-labeled small molecules for positron emission tomography (PET) represents one of the most promising approaches to detect disease progression and evaluate therapeutic effectiveness in vivo.1–3 However, the radiochemical methods available to introduce [18F]fluoride into bioactive probes severely limit the potential scope of the imaging modality.4,5 The short half life (110 min) and low available concentrations of 18F (ranging from nM to μM), compounded with the general difficulties posed by C–19F bond formation, make the identification of broadly applicable radiofluorinations of complex molecules incredibly challenging.6,7 Nevertheless, the past five years have witnessed the discovery of new methods that begin to address the limited scope of radiolabeling with [18F]fluoride. The majority of these solutions have focused on the challenge of [18F]aryl fluoride synthesis.8–10 In contrast, methods for improving the scope of aliphatic radiofluorination remain significantly underdeveloped.11,12 Although numerous modern synthetic methods have been reported that achieve mild and selective aliphatic carbon–fluorine bond formation, these methods utilize electrophilic 19F sources.13 Nucleophilic fluoride is currently the only practical and generally available source of 18F to prepare PET tracers in high specific activity.13 As such, these electrophilic methods have proven less useful for applications in PET.
PET tracers containing aliphatic C–18F labels are typically prepared using a substitution reaction with activated alcohol derivatives (i.e. tosylate, mesylate) and [18F]KF in the presence of cryptands such as Kryptofix® 2.2.2. (K222). Substrates possessing protic functional groups (e.g. alcohols) and functionality prone to elimination are generally not tolerated under these reaction conditions due to the high temperatures (>100 °C) necessary for labeling and the basicity of [18F]KF/K222.6,7 Furthermore, despite the importance of stereochemistry with regard to biological activity, the preparation of single stereoisomer PET probes is often challenging owing to the propensity of [18F]fluoride reagents to induce epimerization. As such, PET tracers are often evaluated as racemic mixtures or they are subjected to time-consuming chiral HPLC separation.14,15 To the best of our knowledge, methods capable of late-stage enantioselective labeling with [18F]fluoride are completely unknown. Herein, we report an asymmetric, nocarrier-added radiosynthesis of [18F]fluorohydrins by ring opening of epoxides with chiral cobalt catalysts. In addition to offering direct access to single enantiomer tracers in the last synthetic step, the method also addresses many of the noted deficiencies associated with aliphatic labeling using [18F]KF.
[18F]Fluorohydrins represent a useful motif in probe design and are featured in several experimental and clinically validated PET tracers.14,16–18 They are typically prepared through selective displacement of differentially protected diols followed by deprotection of the remaining protecting group (vide infra). As such, preparation of a single enantiomer PET probe containing an [18F]fluorohydrin requires that stereochemistry be set within an organic molecule prior to labeling. Asymmetric ring opening of epoxides is an attractive alternative approach to [18F]fluorohydrins because it obviates the need for multistep preparation of differentially protected enantiopure 1,2-diols and it introduces the radioisotope in the last step. However, ring-opening reactions of epoxides with [18F]fluoride generally cannot be achieved due to poor rates and regioselectivity.19,20
We recently demonstrated that a Lewis acidic chiral (salen)Co and an isothiourea or amidine co-catalyst promote enantioselective, nucleophilic fluoride ring opening of epoxides (Figure 1A).21 This approach represents a rare example of a catalyst-controlled enantioselective nucleophilic fluorination and features remarkably mild and functional group tolerant conditions. Key to its success is the use of benzoyl fluoride (PhCOF) and an alcohol additive as a latent source of HF, which avoids an unselective background reaction and catalyst poisoning. Based on these promising attributes, we sought to translate the methodology to a radiofluorination protocol. However, a number of issues inhibited direct application, not the least of which was that use of [18F]PhCOF would require its radiosynthesis and purification prior to labeling. Instead, we relied on insight gathered from our mechanistic analysis of the catalytic reaction, which revealed that a (salen)CoF(HF) species generated in situ from PhCOF serves as the active fluorinating agent and that fluorination occurs through a homobimetallic pathway in which one (salen)Co center activates the epoxide while the second delivers fluoride (Figure 1B).22 As such, we envisioned that an alternative synthesis of (salen)CoF by counterion metathesis between [18F]fluoride and a suitable (salen)Co(III)X (X = anionic counterion) species could lead to the development of a successful [18F]fluoride ring opening (Figure 1C).23 In the event, we found that using (R,R)-(salen)CoOTs (1) as a precursor, a [18F](salen)CoF (2) species suitable for the radiofluorination of epoxides was generated by elution of [18F]fluoride from an ion-exchange cartridge containing a quaternary ammonium cation (QMA). Importantly, the preparation of 2 is directly analogous to the preparation of [18F]KF and can be carried out under air without the use of rigorously dried solvents or glassware.6,7
Figure 1.
(A) Asymmetric fluoride ring opening of epoxides catalyzed by (salen)Co; DBN: 1,5-diazabicyclo(4.3.0)non-5-ene. (B) Proposed homobimetallic mechanism. (C) Strategy for direct radiosynthesis of [18F]fluorohydrins.
We next investigated the scope and generality of the epoxide radiofluorination with 2 (Table 1). Representative epoxides were added to the MTBE solution of 2 at 50 °C for 20 minutes and the radiofluorination reactions were analyzed by radio-TLC for radio-determination. Using this procedure, a variety of racemic terminal epoxides could be radiolabeled in 23–68% RCY and >90% ee. Ether, aliphatic, and styrenyl epoxides were all well tolerated (3, 4, 5, 7). Additionally an epoxide adjacent to relatively acidic C–H bonds, which undergo undesired elimination under standard conditions with KF, provided (6) in 23% RCY and >95% ee (see supporting information for details). Furthermore, a meso epoxide underwent desymmetrization under the same conditions in 25% RCY and >95% ee, allowing the generation of 8, bearing a stereogenic C–18F center.24 Disappointingly, labeling of epoxides possessing Lewis basic nitrogens or α-branching with 2 provided poor radioisotope incorporation (<1% RCY). This presented a significant limitation due to the prevalence of such functionality among bioactive molecules.
Table 1.
Scope of radiofluorination with 2a
|
Radiochemical yields (RCY) are the average of n runs and are based on activity added to each reaction. RCY was determined by radio-TLC. The identity and ee of the product was determined by radio-HPLC. See SI for details; MTBE: methyl t-butyl ether.
In order to address this limitation we explored the use of dimeric cobalt catalyst 9 (Figure 2A), reasoning that previously observed rate enhancements induced by 9 for the fluorination of epoxides may enable increased scope.22 The radiosynthesis of dimeric catalyst 10 was achieved in a manner directly analogous to 2 (see supporting information for details). In an effort to evaluate the reactivity of dimeric catalyst 10, we selected an array of clinically validated and recently disclosed pre-clinical PET tracers as case studies.25
Figure 2.
(A) Linked-catalyst employed in asymmetric radiofluorination of complex epoxides. (B) Comparative radiosyntheses of [18F]THK-5105. (C) Comparative radiosyntheses of [18F]isatin sulfonamides. Radiochemical yields (RCY) are the average of n runs and are based on activity added to each reaction. RCY was determined by radioTLC. The identity and ee of the product was determined by radio-HPLC. The dr of 15 was determined used Mosher ester analysis on [19F]15. See SI for details.
To examine the compatibility of 10 with nitrogen containing substrates, [18F]THK-5105 (12) was selected (Figure 2B). [18F]THK-5105 (12) is an experimental PET tracer for imaging of tau pathology, which is an important biomarker for Alzheimer’s disease.14 The reported preparation of [18F]THK-5105 (12) relies on the commonly employed strategy of differential protection of a 1,2-diol as a tosylate and THP-ether (Figure 2B). Radiosynthesis is achieved by nucleophilic displacement of the tosylate 11, followed by acidic cleavage of THP-protected alcohol to provide racemic [18F]THK-5105 (12). In contrast, we found that dimeric cobalt catalyst 10 labeled racemic epoxide 13 in 25% RCY and 85% ee to directly deliver enantioenriched [18F]THK-5105 (Figure 2B, (S)-12). Although the radiochemical yield is diminished compared to that in the reported preparation of [18F]THK-5105, this result demonstrates the utility of the method with respect to tracer candidates containing Lewis basic functionality.
[18F]Isatin sulfonamide 15 was selected in order to probe the regioselectivity of [18F]fluoride delivery with catalyst 10. [18F]Isatin sulfonamide 15 is an experimental tracer for imaging of caspases-3 and −7, which are implicated in the process of apoptosis.16 It was reported that the stereochemistry of the fluorohydrin is integral to the selectivity of caspase inhibition. The reported strategy to prepare 15 utilizes a two step protocol employing ring opening of diastereomerically pure cyclic sulfonate 14 followed by deprotection to gain access to [18F]fluorohydrin 15 (Figure 2C). Unfortunately, under the described labeling conditions, cyclic sulfonate 14 undergoes unselective ring opening and results in the formation of a separable mixture of both possible regioisomers, (R)-15 and 16. In contrast, we observed that dimeric catalyst (S,S,S,S)-10 was capable of labeling a 1:1 mixture of diastereomers (17), producing (R)-15 as a single regioisomer in 62% RCY with high levels of diastereocontrol (>20:1), demonstrating the unique selectivity of 10 relative to traditional [18F]fluoride sources.
Given the mildness of these radiofluorinations (50 °C) in comparison to the fluorohydrin syntheses with [18F]KF (>100 °C), we sought to demonstrate the utility of the new process for the preparation of nitroimidazole based tracer [18F]FMISO (20), which was developed to image hypoxia within tumors (Figure 3A).18 Reported attempts at direct radiochemical synthesis of [18F]FMISO (20) from epoxide precursor 18 with [18F]KF require 3 hours at 100 °C and lead primarily to SNAr chemistry between the nitroimidazole and the pendant alcohol formed as a result of ring opening with fluoride (Figure 3A, 19).26 In contrast, we found that 10 facilitates the direct radiochemical synthesis of enantioenriched [18F]FMISO (20) in 67% RCY and 90% ee at 50 °C. Given these results, we anticipated that the related tracer [18F]FETNIM (23), which bears an additional hydroxyl group, might be amenable to labeling without the need for protection of epoxy alcohol 22 (Figure 3B).17 Gratifyingly, formation of [18F]FETNIM (23) was accomplished in 74% RCY directly from racemic epoxy alcohol 22 (2:1 syn:anti mixture of diastereomers), and the desired syn diastereomer was generated in >95% ee. These examples demonstrate that the low temperature and attenuated basicity of this new radiofluorination protocol enable efficient product formation in the presence of reactive functionality.
Figure 3.
(A) Comparative radiosyntheses of [18F]FMISO. (B) Comparative radiosyntheses of [18F]FETNIM. (C) Direct radiosynthesis of [18F]threonine precursor. Radiochemical yields (RCY) are the average of n runs and are based on activity added to each reaction. RCY was determined by radioTLC. The identity and ee of the product was determined by radio-HPLC. See SI for details.
Under the conditions utilized for labeling with [18F]KF, substrates prone to base induced epimerization frequently undergo loss of stereochemical information.15 Given the reduced basicity of 10, we expected that the integrity of epimerizable stereocenters within substrates could be maintained during labeling. We selected 4-fluorothreonine, a rare example of a fluorine-containing natural product, as a substrate bearing a potentially epimerizable stereocenter. Prepared as a mixture of enantiomerically pure diastereomers (4:1 syn:anti), epoxide 24 was subjected to 10, affording the desired product in 60% RCY (Figure 3C, 25). It was also observed that the matched catalyst preferentially reacted with the desired syn-diastereomer producing 25 as a single diastereomer (>99:1).
Having shown the utility of this method for the manual preparation of known tracers, we next sought to demonstrate the feasibility of a remote, semi-automated radiosynthesis of [18F]FMISO. Utilizing a remote-controlled microwave cavity integrated into an automated liquid-handler, 12.3 mCi [18F]FMISO was isolated in 10.6% non-decay corrected RCY (from activity delivered, 37 minutes total synthesis time) following semi-preparative-HPLC. The isolated tracer possessed a Co-content of 5 ppb (ICP-MS) and specific activity of 3.7 Ci/umol (EOB), indicating that this method is capable of providing a PET tracer useful for in vivo studies.16
In conclusion, we have demonstrated that commercially available catalyst 2 and dimeric catalyst 10 are capable of radiolabeling a diverse array of complex epoxides with high levels of stereocontrol and functional group compatibility under mild conditions. The protocol makes use of an air-stable catalyst and is operationally simple to carry out. We anticipate that it will facilitate the synthesis of novel PET tracers, and also allow investigators to better understand the relationship between stereochemistry and radiotracer imaging properties. Furthermore, the utility of 10 and other transition metal fluorides for mild and selective radiofluorination is a topic of ongoing research in our laboratory.
Supplementary Material
Supporting Information
Experimental procedures, additional reaction optimization and spectroscopic data for all new compounds. This material is available free of charge via the Internet at http://pubs.acs.org.
ACKNOWLEDGMENT
Dr. Julia A. Kalow is acknowledged for helpful discussions. Drs. Wenping Li and Eric Hostetler (Merck & Co., Inc., West Point PA) are acknowledged for providing the facilities utilized for the semi-automated synthesis of [18F]FMISO and assistance with specific activity determinations. We thank Princeton University, the National Science Foundation (CAREER-1148750), the National Institute of Health (CA164490 and DK081342), and the PA Health department for financial support. T.J.A.G. thanks BristolMyers Squibb and Eli Lilly for graduate fellowships. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE1148900 (R.F.L.). A.G.D. is an Alfred P. Sloan Foundation Fellow, Eli Lilly Grantee, an Amgen Young Investigator, and a Roche Early Excellence in Chemistry Awardee.
Footnotes
ASSOCIATED CONTENT
The authors declare no competing financial interests.
REFERENCES
- (1).Ametamey SM; Honer M; Schubiger PA Chem. Rev. 2008, 108, 1501–1516. [DOI] [PubMed] [Google Scholar]
- (2).Matthews PM; Rabiner EA; Passchier J; Gunn RN Br. J. Clin. Pharmacol. 2012, 73, 175–186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (3).Phelps ME Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9226–9233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (4).Zhu L; Ploessl K; Kung HF Science 2013, 342, 429–430. [DOI] [PubMed] [Google Scholar]
- (5).Furuya T; Kamlet AS; Ritter T Nature 2011, 473, 470–477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (6).Cai L; Lu S; Pike VW Eur. J. Org. Chem. 2008, 2853–2873. [Google Scholar]
- (7).Miller PW; Long NJ; Vilar R; Gee AD Angew. Chem., Int. Ed. 2008, 47, 8998–9033. [DOI] [PubMed] [Google Scholar]
- (8).Lee E; Kamlet AS; Powers DC; Neumann CN; Boursalian GB; Furuya T; Choi DC; Hooker JM; Ritter T Science 2011, 334, 639–642. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (9).Lee E; Hooker JM; Ritter TJ Am. Chem. Soc. 2012, 134, 17456–17458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- (10).Gao Z; Lim YH; Tredwell M; Li L; Verhoog S; Hopkinson M; Kaluza W; Collier TL; Passchier J; Huiban M; Gouverneur V Angew. Chem., Int. Ed. 2012, 51, 6733–6737. [DOI] [PubMed] [Google Scholar]
- (11).Topczewski JJ; Tewson TJ; Nguyen HM J. Am. Chem. Soc. 2011, 133, 19318–19321. [DOI] [PubMed] [Google Scholar]
- (12).Benedetto E; Tredwell M; Hollingworth C; Khotavivattana T; Brown JM; Gouverneur V Chem. Sci. 2013, 4, 89–96. [Google Scholar]
- (13).Liang T; Neumann CN; Ritter T Angew. Chem., Int. Ed. 2013, 52, 8214–8264. [DOI] [PubMed] [Google Scholar]
- (14).Okamura N; Furumoto S; Harada R; Tago T; Yoshikawa T; Fodero-Tavoletti M; Mulligan RS; Villemagne VL; Akatsu H; Yamamoto T; Arai H; Iwata R; Yanai K; Kudo YJ Nucl. Med. 2013, 54, 1420–1427. [DOI] [PubMed] [Google Scholar]
- (15).Qu W; Zha Z; Ploessl K; Lieberman BP; Zhu L; Wise DR; Thompson CB; Kung HF J. Am. Chem. Soc. 2011, 133, 1122–1133. [DOI] [PubMed] [Google Scholar]
- (16).Limpachayaporn P; Wagner S; Kopta K; Hermann S; Schafers M; Haufe GJ Med. Chem. 2013, 56, 4509–4520. [DOI] [PubMed] [Google Scholar]
- (17).Grönroos T; Eskola O; Lehtiö K; Minn H; Marjamäki P; Bergman J; Haaparanta M; Forsback S; Solin OJ Nucl. Med. 2001, 42, 1397–1404. [PubMed] [Google Scholar]
- (18).Lee ST; Scott AM Semin. Nucl. Med. 2007, 37, 451–461. [DOI] [PubMed] [Google Scholar]
- (19).Podichetty AK; Wagner S; Schroer S; Faust A; Schäfers M; Schober O; Kopka K; Haufe GJ Med. Chem. 2009, 52, 3484–3495. [DOI] [PubMed] [Google Scholar]
- (20).Schirrmacher R; Lucas P; Schirrmacher E; Wängler B; Wängler C Tetrahedron Lett. 2011, 52, 1973–1976. [Google Scholar]
- (21).Kalow JA; Doyle AG J. Am. Chem. Soc. 2010, 132, 3268–3269. [DOI] [PubMed] [Google Scholar]
- (22).Kalow JA; Doyle AG J. Am. Chem. Soc. 2011, 133, 16001–16012. [DOI] [PubMed] [Google Scholar]
- (23).Under radiolabeling conditions with low concentrations of [18F]fluoride, it is unlikely that a [18F]Co(III)F(HF) species is the active nucleophile as was proposed in our cold catalytic system. We therefore describe this species as a [18F]Co(III)F; however, the exact structure is unknown at this time. Furthermore, it is unclear if the OTs counterions undergo additional exchange (e.g. with MeOH or H2O) during the preparation of 10. [Google Scholar]
- (24).To date, attempts at employing non-symmetrical 1,2-disubstituted epoxides have met with failure with both the monomeric and dimeric catalyst systems. See SI for further details.
- (25).Our radiochemical yields are decay-corrected and based on activity added to each substrate. In contrast, the published radiochemical yields for 12, 15, 20, and 23 are based on initial activity delivered from a cyclotron and total activity isolated following purification and dose formulation. [Google Scholar]
- (26).Bejot R; Kersemans V; Kelly C; Carroll L; King RC; Gouverneur V Nucl. Med. Biol. 2010, 37, 565–575. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.



