Jens Cardinaleab,
Johannes Ermert*a,
Sven Humperta and
Heinz H. Coenena
aInstitut für Neurowissenschaften und Medizin, INM-5: Nuklearchemie, Forschungszentrum Jülich GmbH, 52425 Jülich, Germany. E-mail: j.ermert@fz-juelich.de
bDepartment of Medical Physics in Radiology, German Cancer Research Center, 69221 Heidelberg, Germany
First published on 28th March 2014
Iodonium ylide precursors of electron rich arenes, i.e. the NET and SERT ligands 4-((3- and 4-fluorophenoxy)phenylmethyl)piperidine, served as model compounds for the direct substitution with n.c.a. [18F]fluoride. Good radiochemical yields of about 20% were obtained in reaction times of ca. 130 minutes with a molar activity of the labelled ligands of more than 50 GBq μmol−1. Those failed as in vivo probes in first evaluation studies. Several important insights, however, were gained into the reaction of ylides, e.g. an unexpected formation of regioisomers. The results clearly demonstrate that aryliodonium ylides are a promising alternative to the well-known diaryliodonium salts for the direct preparation of complex, electron rich n.c.a. [18F]fluoroarenes.
It is well known that iodonium compounds offer an alternative route even for the synthesis of electron rich [18F]fluoroarenes starting from n.c.a. [18F]fluoride.6,7 Their application as precursor for 18F-radiofluorination has mainly been limited to mechanistic studies, and they served in most cases as precursors for the preparation of versatile, non-activated small 18F-labelled molecules used as building blocks for further radiochemical transformations.5 So far, very little attention has been paid to other hypervalent iodine species for radiofluorination.
Recently, Satyamurthy and Barrio reported the successful radiosynthesis of n.c.a. [18F]fluoroarenes by the use of different iodonium ylides as precursors.8 Amongst those, the iodonium ylides derived from Meldrum's acid proved to be the most promising ones due to their relative high stability and good reactivity. Their general structure is shown in Fig. 1. These compounds are commonly formed by the reaction of Meldrum's acid with a suitable aryliodine-III compound under basic conditions.9–11
The reported radiofluorination of iodonium ylides was performed under similar conditions to those generally used for the radiofluorination of iodonium salts. After activation of the [18F]fluoride by a phase transfer catalyst/base activation system (e.g. Kryptofix®2.2.2/potassium carbonate) the iodonium ylide is added in a dipolar aprotic solvent such as DMF and heated to 110–130 °C for 10–15 minutes (Scheme 1), as described by Satyamurthy and Barrio.8
In 2003 Orjales et al. published a series of potential antidepressants.12 Amongst those are the fluorophenoxyethers 4-((3- and 4-fluorophenoxy)phenylmethyl)piperidine (3-FPPMP and 4-FPPMP, Fig. 2) which are potential ligands for the serotonin and norepinephrine (reuptake) transporters (SERT and NET). Being electron-rich fluoroarenes those offered themselves as good candidates to test an authentic labelling with fluorine-18 via the new approach. Thus, our objective was to use the radiosynthesis of n.c.a. 3- and 4-[18F]FPPMP as model reaction for the direct 18F-fluorination of corresponding iodonium ylides of electron-rich arenes as precursors.
![]() | ||
Fig. 2 Molecular structures of 3-FPPMP (1) and 4-FPPMP (2), ligands for the NE- and SE-reuptake transporters, respectively. |
The syntheses of 4-benzyloxyphenyliodonium-(5-[2,2-dimethyl-1,3-dioxane-4,6-dione])ylide (13) and of 4-methoxyphenyliodonium-(5-[2,2-dimethyl-1,3-dioxane-4,6-dione])ylide (14) were reported elsewhere.13 All iodonium ylides were synthesized under exclusion of light and stored at 2–8 °C. The reference compound 2-fluorobenzyloxybenzene was obtained by the reaction of 2-fluorophenol with benzylbromide. NMR spectra were recorded either on a Varian-Inova 400 or a Bruker DPX Avance 200 spectrometer. The chemical shifts are given in parts per million relative to the solvent signal.
Elemental analyses (EA, microanalyses) were carried out on a Vario EL cube, elemental analyser (at ZEA-3, Forschungszentrum Jülich).
The products of the labelling reactions were analysed by radio-HPLC. The HPLC system consisted of a Smartline Pump 1000, a Luna C-18(2) column (5 μm, 250 × 4.6 mm) or a Luna PFP(2) column (5 μm, 250 × 4.6 mm) (both from Phenomenex, Aschaffenburg, Germany). Further, two Rheodyne manual 6-position selector valves (Idex Health and Science, Wertheim-Mondfeld, Germany) were positioned in front of and behind the column, equipped with 50 μl sample loops. The detection of compounds by UV-absorption was carried out with a K-2501 detector (Knauer, Berlin, Germany). When the system was used for the isolation of products by analytical HPLC the first valve was equipped with a 200 μl loop and the column directly connected to the detectors. Semi-preparative HPLC-runs were conducted on a HPLC system consisting of a Merck-Hitachi L-6000 pump, a Rheodyne manual 6-position selector valve in front of the column, equipped either with a 2 ml or a 1 ml sample loop and a Phenomenex Luna PFP(2) column (5 μm, 250 × 100 mm). The detection of compounds by UV-absorption was carried out with a S3300 UV-detector (Sykam, Fürstenfeldbruck, Germany).
The detection of the radioactive products was performed with a NaI(Tl) detector crystal connected with an ACE Mate signal amplifier and an EG&G Ortec Model 276 photomultiplier base (Ortec, Meerbusch, Germany).
Measurement of the radioactivity of bulk samples was conducted on a Curiementor 2 (PTW, Freiburg, Germany) ionisation chamber.
Phosphor imager plates used for in vitro autoradiography of brain tissue slices were scanned with a laser phosphor imager BAS 5000 (Fuji, Düsseldorf, Germany) utilizing software from the vendor (Version 3.14, Raytest, Straubenhardt, Germany).
1H NMR (200 MHz, CDCl3) δ: 7.39–7.20 (m, 8H), 6.93–6.74 (m, 2H), 4.82 (d, 1H, J = 6.7 Hz), 4.25–4.10 (m, 2H), 2.77–2.57 (m, 2H), 2.02–1.92 (m, 2H), 1.69 (m, 1H), 1.49 (s, 9H), 1.44–1.29 (m, 2H).
13C NMR (200 MHz, CDCl3) δ: 158.9, 154.8, 139.2, 130.6, 129.9, 128.6, 128.0, 126.7, 125.5, 115.0, 94.2, 84.1, 79.4, 43.7, 43.4, 28.5, 28.4, 28.1.
MS (+ c ESI): m/z = 494.19.
Elemental analysis: calc.: C 55.99%, H 5.72%, N 2.84%; found: C 56.33%, H 6.21%, N 2.98%.
1H NMR (400 MHz, CDCl3) δ: 7.50–7.26 (m, 8H), 6.66–6.58 (m, 2H), 8.80 (d, 1H, J = 6.6 Hz), 4.24–4.09 (m, 2H), 2.77–2.57 (m, 2H), 2.09–1.87 (m, 3H), 1.59–1.22 (m, 13H).
13C NMR (400 MHz, CDCl3) δ: 158.1, 154.7, 138.0, 128.5, 127.8, 126.7, 118.3, 84.0, 82.9, 79.3, 43.7 (bs), 43.3, 28.4, 28.0 (bs).
MS (+ c ESI): m/z = 493.99.
Elemental analysis: calc.: C 55.99%, H 5.72%, N 2.84%; found: C 56.25%, H 6.2%, N 3.04%.
Yield: 292 mg beige solid (33%).
1H NMR (400 MHz, CDCl3) δ: 7.18–7.11 (m, 8H), 6.94 (t, 1H, J = 4.0 Hz), 6.80 (d, 1H, J = 4.2 Hz), 4.69 (d, 1H, J = 3.2 Hz), 3.98 (m, 2H), 2.49 (m, 2H), 1.81 (m, 2H), 1.52 (s, 6H), 1.31 (s, 9H), 1.22–1.08 (m, 3H).
13C NMR (400 MHz, CDCl3) δ: 163.6, 159.7, 158.7, 139.0, 138.1, 131.7, 130.5, 129.8, 128.6, 128.5, 128.1, 127.8, 126.8, 126.6, 125.3, 124.8, 119.8, 119.4, 114.9, 114.4, 104.3, 84.4, 74.3, 55.9, 43.0, 28.4, 28.0, 25.8.
MS (+ c ESI): m/z = 674.12.
Elemental analysis: calc.: C 54.81%, H 5.39%; found: C 54.81%, H 5.75%.
Yield: 42% as yellow solid.
1H NMR (400 MHz, CDCl3) δ: 7.71 (d, 2H, J = 8.4 Hz, C14–H), 7.33–7.21 (m, 6H), 6.78 (d, 2H), 4.79 (d, 1H), 4.10 (m, 2H), 2.60 (m, 2H), 1.91 (m, 4H), 1.63 (s, 6H), 1.41 (s, 9H), 1.33–1.24 (m, 3H).
13C NMR (400 MHz, CDCl3) δ: 163.4, 161.5, 154.7, 138.1, 136.1, 128.8, 128.3, 126.6, 119.2, 104.4, 102.6, 84.6, 79.4, 56.8, 43.2, 28.4, 25.8.
MS (+ c ESI): m/z = 674.10.
Elemental analysis: calc.: C 54.81%, H 5.39%, N 2.20%; found: C 55.3%, H 5.73%, N 1.91%.
For deprotection of [18F]8 and [18F]10, 500 μl of trifluoroacetic acid was added and the whole solvent–acid mixture removed at 40–50 °C under reduced pressure (starting from 800 mbar) and a gentle argon stream. The crude products 3-[18F]FPPMP and 4-[18F]FPPMP were dissolved in 150 μl HPLC-solvent (60:
30
:
10 THF–MeOH–buffer: 1% aqueous TEA–H3PO4, pH 6.0; Luna C-18; 0.7 ml min−1) and purified by analytical HPLC.
Compound | k-value | HPLC conditions |
---|---|---|
3-FPPMP | 1.95 | 60![]() ![]() ![]() ![]() |
4-FPPMP | 2.12 | |
2-Fluoroanisole | 2.99 | 55![]() ![]() |
3-Fluoroanisole | 3.43 | |
4-Fluoroanisole | 3.19 | |
2-Fluorobenzyloxy-benzene (2-FBOB) | 4.81 | 65![]() ![]() |
3-Fluorobenzyloxy-benzene (3-FBOB) | 5.83 | |
4-Fluorobenzyloxy-benzene (4-FBOB) | 5.33 | |
Boc-3-[18F]FPPMP ([18F]8) | 10.68 | ACN–Buffer 60![]() ![]() |
Boc-4-[18F]FPPMP ([18F]10) | 11.04 |
![]() | ||
Scheme 3 Synthesis of iodonium ylide 11 by a one-pot procedure. (i) (1) mCPBA, CH2Cl2, (2) KOH, Meldrum's acid. |
![]() | ||
Scheme 4 Radiosynthesis of 3-[18F]FPPMP ([18F]1). (i) n.c.a. [18F]fluoride, Kryptofix®2.2.2., K2CO3, acetonitrile, 120 °C; (ii) CF3CO2H, DCM. |
The side products overloaded the cartridge and caused a breakthrough of the desired product resulting in losses of more than 50%. Thus, isolation by semi-preparative HPLC became mandatory which necessitated a change of the reaction solvent from DMF to acetonitrile. Surprisingly, this also led to an enhancement of the RCY to 20 ± 5%.
After the reaction in acetonitrile the product mixture was simply diluted with water and submitted to semi-preparative HPLC. A too strong dilution below 60:
40 (v/v) acetonitrile–water caused a precipitation of the non-polar components as oil, containing also a considerable fraction of the desired product Boc-3-[18F]FPPMP ([18F]8), and thus had to be avoided. However, the addition of 150 μl water to the reaction mixture (ca. 75
:
25 v/v acetonitrile–water) proved to be sufficient. The chromatogram of the separation is shown in Fig. 3. Although there were only small radiochemical impurities, many non-radioactive side products were detected in the UV-chromatogram. Also a large number of less polar side products were found which eluted later from the HPLC column (not presented in Fig. 3) and caused the breakthrough of product in the initial attempts to separate the product by solid phase extraction using a C-18 cartridge (see above).
![]() | ||
Fig. 3 Radiochromatogram of the separation of [18F]8 by semi-preparative HPLC. (60![]() ![]() |
After the chromatographic separation, product [18F]8 was extracted from the eluent with a C-18 cartridge, what could now be performed without product losses, and eluted in DCM for deprotection with trifluoroacetic acid. As expected, the removal of the Boc-group proved to be quite easy and was accomplished with quantitative yield. However, another HPLC separation was necessary to obtain n.c.a. 3-[18F]FPPMP in highest purity. This was done on an analytical column. The molar radioactivity of the product was determined in this final purification and was found to be higher than 50 GBq μmol−1. The RCY of [18F]1, related to starting [18F]fluoride, amounted to 20 ± 5% after a total synthesis time of about 110 minutes including separation from the HPLC product fraction and solvent change to water.
Since our objective was to demonstrate the preparation and isolation of a complex radiotracer via an iodonium precursor with a sufficient purity for a first preclinical evaluation, the radiochemical yield was not further optimised. Higher RCYs of up to 45% were observed when higher amounts of precursor in more solvent (15 mg of 11 in 1 ml of acetonitrile) were used. However, under these conditions the final product [18F]1 was usually contaminated with non-radioactive side products because of adverse effects in the chromatographic separation due to more side products and a higher solvent volume.
![]() | ||
Scheme 5 Formation of regioisomers in the synthesis of 4-[18F]10. (i) n.c.a. [18F]fluoride, Kryptofix®2.2.2., K2CO3, acetonitrile, 130 °C. |
The resolution of the semi-preparative HPLC was not high enough to separate these regioisomers from each other. Therefore, the n.c.a. products were first isolated together by semi-preparative HPLC, the eluate concentrated, and the regioisomers separated on an additional HPLC-system with an analytical column. Following this, the desired product was deprotected with a quantitative yield as described above. The additional steps for the separation of the isomers took about 30–40 minutes prolonging the total synthesis time to about 140–150 minutes. By the procedure applied here, principally both regioisomers can be prepared from the same precursor (compound 12), albeit in different total RCY of about 20% ([18F]2) and 10% ([18F]1), respectively.
In order to further confirm the formation of regioisomers, the simpler but structurally equal compounds 4-methoxyphenyl-(5-[2,2-dimethyl-1,3-dioxane-4,6-dione])ylide 13 and 4-benzyloxyphenyliodonium-(5-[2,2-dimethyl-1,3-dioxane-4,6-dione])ylide 14 were 18F-labelled under similar conditions as used for the radiosynthesis of Boc-4-[18F]FPPMP 8 (Scheme 6).
![]() | ||
Scheme 6 Formation of regioisomers in the reaction of ylides 13 and 14 with n.c.a. [18F]fluoride. (i) n.c.a. [18F]fluoride, Kryptofix®2.2.2., K2CO3, acetonitrile, 130 °C. |
In fact, in both cases the corresponding 3-isomers were also formed in about 4% and 11% RCY, respectively, while the expected 4-isomers were formed in about 12% and 20%, respectively, as identified by comparison of the retention times with the respective macroscopic standards. Thus, the formation of regioisomers seems to be a problem with the radiofluorination of iodonium ylides. This surprising observation that the nucleophilic substitution on ylides is not regiospecific was not reported by Satyamurthy and Barrio.8 A similar effect, however, has been found by Graskemper et al. in the thermal decomposition of (4-methoxyphenyl)(5-methoxy [2.2]paracyclophan-4-yl)-iodonium hexafluorophosphate.17 There it is suggested, that the strong electron donating effect of the methoxy group leads to a competing fluorination via an aryne pathway.
Another interesting point to note is that the amount of non-radioactive side-products in the labelling reactions of ylides 13 and 14 was significantly lower than in case of the more complex ylides 11 and 12. This can either mean, that the presence of more functional groups opens alternative reaction paths, or, that the bulkier ylides simply tend to decompose via rather unspecific pathways. Due to the high number of side products formed in case of ylides 11 and 12 (cf. Fig. 3) an unspecific decomposition is more likely. In the initial labelling experiments with ylide 11 several reactions with lower temperatures were performed, but this only led to a decrease of the RCY while no improvement of the amount of side-products was observed.
The radiosynthesis of [18F]1 and [18F]2 proved feasible by direct nucleophilic substitution of the corresponding iodonium ylides. The isolation of the products, however, was challenging due to the high amount of non-radioactive side-products and, in case of [18F]2, due to the formation of a regioisomer. Here, even a second separation by HPLC was inevitable resulting in a prolonged synthesis time which might be reduced by further optimisation of the purification process, e.g. by application of a solvent gradient.
The preparation of the ylide precursors succeeded by a comparatively easy one-pot procedure in satisfactory yields of 33% and 42%. For the radiosynthesis the main challenge proved to be the purification of the labelled products due to many non-radioactive side-products formed. Additionally, during the preparation of 4-[18F]FPPMP the surprising formation of two regioisomers was found. The occurrence of positional isomers was further confirmed by the radiolabelling of comparable 4-methoxy- and 4-benzyloxyphenyliodonium ylides, demonstrating that the reaction is not regiospecific. Like in this study, separation of isomers might demand additional efforts in given case. In spite of these drawbacks the potential of aryliodonium ylides as precursors for nucleophilic 18F-labelling of more complex molecules was demonstrated here.
Today the most important alternative route to such 18F-labelled radiotracers is a multi-step synthesis in which either electron withdrawing groups are attached to the benzene ring to be 18F-fluorinated, or the whole molecule has to be synthesised by a two or more step procedure. In both cases the radiosynthesis gets more complex (e.g. due to additional reagents) while the necessity of purification by HPLC is also very likely, although the total amount of side-products might be significantly lower. The better approach for a given radiotracer will strongly depend on the individual case.
Since the iodonium ylides represent the more direct approach, their advantage increases the more complex the multi-step approach gets. In this respect, the reaction of n.c.a. [18F]fluoride with iodonium ylides proves to be a powerful alternative to the multi-step synthesis for the preparation of electron rich and non-activated [18F]fluoroarenes. Additionally, iodonium ylides also proved to be a better to handle alternative to the well-known aryliodonium salts. Those findings warrant further development of the ylide-method for the direct synthesis of non-activated or electron rich n.c.a. [18F]fluoroarenes.
This journal is © The Royal Society of Chemistry 2014 |