t-Bu2SiF-Derivatized D2-Receptor Ligands: The First SiFA-Containing Small Molecule Radiotracers for Target-Specific PET-Imaging

The synthesis, radiolabeling and in vitro evaluation of new silicon-fluoride acceptor (SiFA) derivatized D2-receptor ligands is reported. The SiFA-technology simplifies the introduction of fluorine-18 into target specific biomolecules for Positron-Emission-Tomography (PET). However, one of the remaining challenges, especially for small molecules such as receptor-ligands, is the bulkiness of the SiFA-moiety. We therefore synthesized four Fallypride SiFA-conjugates derivatized either directly at the benzoic acid ring system (SiFA-DMFP, SiFA-FP, SiFA-DDMFP) or at the butyl-side chain (SiFA-M-FP) and tested their receptor affinities. We found D2-receptor affinities for all compounds in the nanomolar range (Ki(SiFA-DMFP) = 13.6 nM, Ki(SiFA-FP) = 33.0 nM, Ki(SiFA-DDMFP) = 62.7 nM and Ki(SiFA-M-FP) = 4.21 nM). The radiofluorination showed highest yields when 10 nmol of the precursors were reacted with [18F]fluoride/TBAHCO3 in acetonitrile. After a reversed phased cartridge purification the desired products could be isolated as an injectable solution after only 10 min synthesis time with radiochemical yields (RCY) of more than 40% in the case of SiFA-DMFP resulting in specific activities >41 GBq/µmol (>1,100 Ci/mmol). Furthermore, the radiolabeled products were shown to be stable in the injectable solutions, as well as in human plasma, for at least 90 min.


Introduction
Positron-Emission-Tomography (PET) is a non-invasive imaging technique using contrast-agents (radiotracers) labeled with radionuclides such as fluorine-18 which undergo positron emission decay. The resulting positron annihilates with an electron, producing two gamma photons, emitted at a 180° angle, which can be detected in coincidence with high sensitivity, thus yielding a spatial resolution in the mm-range. Although a high number of radiotracers has been developed only a limited number are commonly used as a result of their sometimes cumbersome and difficult synthesis.
The short half-life of many PET-nuclides ( 18 F t ½ = 109.7 min) makes it necessary to produce the radiotracer on site, resulting in high investment costs. The effort for a radiotracer synthesis is nearly independent of the number of patient doses produced per synthesis run. Therefore, PET-centers usually focus on the application of well-established radiotracers such as the glucose derivative 18 F[FDG] ( 18 F-2-fluoro-2-desoxyglucose) and only a few large PET-centers are able to provide a large number of other tracers. To bridge this gap, the development of new labeling techniques for the easy introduction of fluorine-18 into radiotracers without costly equipment would be favorable. A promising approach to simplify the radionuclide introduction significantly is the exploitation of the strong silicon-fluorine (Si-F) bond [1,2]. The silicon fluoride acceptor (SiFA) method, based on the efficient isotopic non-radioactive 19 F for radioactive 18 F exchange at the silicon atom, was recently developed and applied for simple one-and two-step 18 F-fluorinations of peptides [3]. The radiosynthesis of different SiFA-derivatized peptides (RGD-, octreotate-, as well as a bombesin-analogue) resulted in specific activities of up to 680 GBq/µmol (18.4 Ci/µmol) for the final radiotracers, surprisingly high for a carrier added radiosynthesis [4]. This finding can be explained by DFT (density functional theory) model calculations. The most convenient feature of this SiFA labeling technique is that a final HPLC purification of the radiotracer from the precursor is not necessary, since labeling precursor and labeled product are identical. Another approach with Si-18 F bearing building blocks was used for the in vivo evaluation of a Si-18 F-derivatized bombesin derivative in tumor-bearing rodents [5,6]. However, a heating-as well as an HPLC-purification step-were necessary to obtain the final 18 F-labeled compound in high specific activities. Most recently, the implementation of new functionalized SiFAs for the kit-like 18 F-labeling of biomolecules was reported [7][8][9][10]. In in vivo studies, the use of SiFA-amounts as low as a few nanomoles resulted in 18 F-labeled proteins with specific activities of up to 10-50 GBq/µmol (270-1,350 Ci/mmol), which would be suitable for receptor-imaging with PET. In this particular study, we aimed at evaluating the applicability of the SiFA technique for the derivatization of small molecule radiotracers, such as the D 2 receptor ligands fallypride (FP, 1), desmethoxyfallypride (DMFP, 2) and raclopride (3). It was expected that the original SiFA building block, which cannot be extensively modified without losing its stability against hydrolysis, might have a detrimental influence on the binding affinity of the SiFA derivatized D 2 receptor ligands. Several new Si-F bearing derivatives derived from basic model compounds were analyzed recently and evaluated as to their stability in aqueous solution with regard to the substitution pattern at the silicon atom [11]. The results are consistent with our previous findings that at least two sterically hindered substituents at the silicon atom are necessary to preserve the stability of the silicon-fluorine bond in vitro [3].
With respect to these steric requirements, SiFA derived model compounds of commonly used PET imaging agents were synthesized to evaluate the potential of the SiFA-concept for the syntheses of SiFA-type small molecule radiotracers. The benzamide derivatives [ 18 F]fallypride (1), [ 18 F]-desmethoxyfallypride (2) and [ 11 C]raclopride (3) radiotracers used for the PET-imaging of the dopaminergic system, were chosen as model compounds for this study [12,13] (Figure 1). All compounds are D 2 -receptor antagonists, which differ mainly in the receptor affinity, in the nanomolar (desmethoxyfallypride, raclopride) and picomolar range (fallypride), respectively [14]. These imaging agents are used for the diagnosis of different neurological disorders related to the dopaminergic system such as parkinsonism and craving [15,16].  Due to the bulkiness of the SiFA-moiety that has to be introduced into the potential radiotracer, the ligand fallypride, having one of the highest affinities to the D 2 receptor, was tested first as a scaffold for SiFA derivatization since even a certain loss of target affinity would not necessarily result in an unusable PET radiotracer. The two different strategies applied were: (i) the integration of the SiFA building block into the fallypride/desmethoxyfallypride general structure and (ii) the coupling of an already existing SiFA compound, namely SiFA-maleimide (SiFA-M, [17]), to a fallypride derivatized with an SH moiety at the butyl side chain (Figure 2). Besides the prerequisite of a good binding affinity to the targeted D 2 -receptor, the radiolabeling has to yield a radiotracer with a sufficiently high specific activity for D 2 receptor imaging with PET. In order to simplify the radiosynthesis of the desired 18 F fluorinated radiotracer, we only studied one-step radiosyntheses using the non-radioactive standards directly as the labeling precursors (isotopic exchange reaction). Hence, the amount of the precursor used determines the specific activity (in relation to the amount of radioactivity used for labeling).   The use of 10 nanomoles precursor would result in specific activities comparable to those of conventionally synthesized 18 F radiotracers. If, e.g., 500 MBq (13.5 mCi) 18 F are incorporated into 10 nmol of the labeling precursor the resulting specific activity would be as high as 50 GBq/µmol (1,350 Ci/mmol). Consequently, the identification of the lowest limit of the ratio precursor amount/concentration necessary for a sufficient radiochemical yield (introduction of fluorine-18) was the second important focus of this work.

Precursor Syntheses
The preparation of the SiFA-modified carboxylic acids 12a and 12b (Scheme 1) followed a general method described in our previous work [17]. Single crystals of compounds 10b, 12a and 12b suitable for X-ray diffraction analysis were obtained as colorless needles by re-crystallization from diethyl ether/hexane. The molecular structures of these compounds are presented in , and selected bond distances and bond angles are collected in Table 1.
All compounds crystallized monoclinically with eight (10b) or four (12a, 12b) molecules in the unit cell. There are two crystallographically independent molecules in the unit cell of compound 10b ( Figure 3) with one of them being disordered. In Table 1, only the data for the non-disordered molecule are given. The silicon atoms in these compounds are four-coordinate and show each a distorted tetrahedral configuration with average angles of 109.59 (10b), 109.70 (12a) and 109.04 (12b). The largest deviations from the tetrahedral angle are found for C(11)-Si-C(15) (119.40°, 12b) and F-Si-C(11) (104.15°, 12a).

In Vitro Evaluation
Compounds 4a-c and 5 (SiFA-DMFP 4a, SiFA-FP 4b, SiFA-DDMFP 4c and SiFA-M-FP 5) were tested for their affinity towards the human D 2 receptor (Table 3). All tested compounds showed K i -values in the nanomolar range. As expected for the derivatization of a small molecule with the bulky SiFA-moiety, the affinities are reduced by factors ranging between 22 [SiFA-DMFP 4a compared to desmethoxyfallypride (2) (1)]. Interestingly, SiFA-DMFP (4a) displays a higher affinity to the receptor compared to SiFA-FP (4b) and SiFA-DDMFP (4c). Among the newly developed substances 4a-c and 5, SiFA-M-FP (5) displays the highest binding affinity which might be explained with the higher distance of the derivatization site (SiFA is bound to the butyl side chain) to the receptor binding parts of the molecule. The methoxy moiety, which distinguishes the high affinity tracer fallypride (1) from the medium-affinity tracer desmethoxyfallypride (2) leads to a loss of binding affinity in the pair SiFA-DMFP 4a/SiFA-FP 4b inversely to the situation with fallypride (1)/desmethoxyfallypride (2). We can therefore assume that the benzoic acid ring system still influences the binding affinity but binds most probably in a different conformation to the binding site at the D 2 -receptor. However, we chose fallypride (1) as a scaffold for this study, because even an unavoidable decrease in bindingaffinity towards the D 2 -receptor might still lead to a medium affinity ligand comparable to raclopride (3-the gold standard for D 2 -receptor imaging with PET). Therefore, the direct comparison of SiFA-M-FP (5) to [ 11 C]raclopride [ 11 C]-3, which is most frequently used in clinical studies of the dopaminergic system, looked more encouraging: When comparing the medium-affinity D 2 -receptor radiotracer raclopride (K i = 1.21 nM) to SiFA-M-FP (5; K i = 4.21) and SiFA-DMFP (4a; K i = 13.6 nM) the affinity is only reduced by the factor 3.5 and 11, respectively. Thus, a successful in vivo D 2 receptor imaging using radiolabeled SiFA-M-FP (5) or SiFA-DMFP (4a) might be possible.

Radiolabeling
Radiolabeling reactions based on isotopic exchange usually have the drawback of a limited specific activity (SA), as the precursor cannot be separated from the final product. It is therefore crucial for in vivo applications of the synthesized radiotracer to reduce the precursor amount to the absolute minimum. Our previous studies showed that for radiofluorinations of small molecules such as a SiFA-aldehyde, amounts of 1-5 nmol were sufficient to achieve high radiolabeling yields [4]. For the direct labeling of SiFA-derivatized peptides, precursor amounts of 10-25 nmol were necessary [18]. The highest radioactivity incorporation rates were observed when the SiFA radiolabeling was carried out in polar aprotic solvents such as acetonitrile or DMSO. As the SiFA-moiety gets hydrolyzed very quickly under basic conditions, we tested the following two mild labeling procedures in acetonitrile, DMF or DMSO: (a) Kryptofix2.2.2/potassium oxalate/ 18 F − and (b) tetrabutylammonium bicarbonate/ 18 F − . Table 4 summarizes the labeling of SiFA-DMFP (4a) under different labeling conditions.
To keep the labeling procedure as simple and convenient as possible, we performed the purification step by using a reversed phase cartridge (SepPak C-18 light) for all tested combinations. We determined the 18 F-incroporation of 18 F − into SiFA-DMFP (4a) after a 5 min reaction time at room temperature with analytical radio-HPLC using samples of the crude reaction mixture. After dilution with 10 mL 1M HEPES buffer at pH = 4.0 (used to prevent hydrolysis) the desired radiotracer was purified using a reversed-phased SepPak-cartridge. The SepPak purification was necessary to remove unreacted 18 F − , solvent and K2.2.2/potassium oxalate or TBAHCO 3 from the product. After washing the cartridge with water, the products could be eluted very efficiently with 1 mL ethanol and were subsequently diluted with isotonic saline. The radiochemical yields (RCY) were calculated to the start of the synthesis and the radiochemical purity (RCP) analyzed by analytical radio-HPLC. Using SiFA-DMFP (4a) as the precursor, we found radioactivity incorporations in acetonitrile using TBAHCO 3 of up to 61% in the crude reaction mixture (Table 4) and an overall RCY after purification of up to 42%. Under these conditions, the specific activity was calculated to be in the range of 88 GBq/µmol (2.4 Ci/µmol) using a precursor amount of 5 nmol and at least 41 GBq/µmol (1.1 Ci/µmol) using 10 nmol precursor when starting from ~1GBq (27 mCi) 18 F-fluoride. This is in the suitable range for D 2 -neuroreceptor imaging with PET. The radiochemical purity of all products was >96% after purification. Using DMF or DMSO, the 18 F-radioactivity incorporation was drastically reduced to less than 25%. More importantly, the desired product could not be purified via cartridge separation. Likewise, the labeling procedure using K2.2.2 resulted in a lower relative 18 F-incorporation as well as a lower RCY compared to the procedure using TBAHCO 3 .  Moreover, radiolabeled [ 18 F]-SiFA-M-FP [ 18 F]-5 could not be purified via the SepPak separation method described above. We can only speculate that the chemical design of the phenolic ring system of SiFA-maleimide leads to a significantly different stability of the SiFA-moiety compared to the benzamides 4a-c. This finding is in line with our previous observation that protein labeling using [ 18 F]-SiFA-thiol bound to a maleimide-derivatized protein resulted in higher radiolabeling yields than a vice versa labeling of a thiol-derivatized protein with [ 18 F]-SiFA-maleimide.
The chemical purity was determined for all products by analytical HPLC at 214 nm and showed no side products for all radiolabeled derivatives with a RCP > 94%. The chemical stability of

General
All solvents used for the syntheses of 4a-c (SiFA-DMFP, SiFA-FP, SIFA-DDMFP) were purified by distillation from appropriate drying agents under argon atmosphere. Solvents and chemicals used in the synthesis of 5, SiFA-M-FP, were of analytical grade and used without further purification. Chemicals and solvents used for the labeling experiments were purchased in the highest available grade and were used without further purification. The NMR experiments were carried out with Jeol AS500, Bruker DRX 400, Bruker DRX 300, and Varian Mercury 200 spectrometers. Chemical shifts (δ) are given in ppm and are referenced to the solvent peaks, with the usual values calibrated against tetramethylsilane ( 1 H, 13 C, 29 Si) and CFCl 3 ( 19 F). High-resolution mass spectra were obtained by using a Finnigan MAT95Q mass spectrometer and a LTQ Orbitrap mass spectrometer (Thermo Electron) using acetonitrile as the mobile phase. FT infrared spectra were recorded using a Bruker IFS 28 spectrometer. Elemental analyses were performed on a LECO CHNS-932 analyzer. The analytical and semi-preparative HPLC system used was an Agilent 1200 system equipped with a raytest Gabi Star radioactivity detector together with a Chromolith Performance (RP-18e, 100-4.6 mm, Merck, Germany) and a Chromolith (RP-18e, 100-10 mm, Merck, Germany) column, respectively. SiFA-M was synthesized according to a published procedure [17]. The synthesis of the thiol-substituted fallypride will be described elsewhere.

Crystallography
Crystals of compounds 10b, 12a and 12b suitable for single-crystal X-ray diffraction analyses were grown by re-crystallization from diethyl ether/n-hexane. Crystallographic data are summarized in Table 1. Intensity data were collected with a Nonius KappaCCD diffractometer with graphitemonochromated Mo-Kα radiation. The data collections covered almost the whole sphere of the reciprocal space with 3 (5), 4 (7 and 8b) sets at different κ angles and 227 (5), 339 (7), 494 (8b) frames by ω-rotation (Δ/ω = 1°) at 2 × 160 s (5), 80 s (7), 60 s (8b) per frame. Crystal decay was monitored by repeating the initial frames at the end of the data collection. After analysis of the duplicate reflections, there was no indication of any decay. The structures were solved by direct methods (SHELXS97 [19]). Refinement applied full-matrix least-squares methods (SHELXL97). All H atoms were located in the difference Fourier map and their positions were isotropically refined with Uiso constrained at 1.2 times Ueq of the carrier C atom for non-methyl and 1.5 times Ueq of the carrier C atom for methyl groups. Atomic scattering factors for neutral atoms and real and imaginary dispersion terms were taken from International Tables for X-ray Crystallography [20]. The figures were created by SHELXTL. Crystallographic data are given in Table 6. Crystallographic data for the structures reported in this paper have been deposited with the Cambridge Crystallographic Data Centre as supplementary material publication No. CCDC-704771, CCDC-704583 and CCDC-704772. Copies of the data can be obtained free of charge on application to CCDC, 12 Union Road, Cambridge, CB2 1EZ, UK (Fax: +44 1223 336033; e-mail: deposit@ccdc.cam.ac.uk).       (5-(di-tert-Butylfluorosilyl)-2,3-dimethoxyphenyl)methanol (10b). The procedure was analogous to the synthesis of 10a starting from the protected alcohol 9b (4.69, 10.59 mmol). The crude product was purified by column chromatography (hexane/diethylether = 3/1 → hexane/diethylether = 2/1 → hexane/diethylether = 1/1 → hexane/diethylether) to afford 10b (  Elemental analysis calculated (%) for C 16 (14). The synthesis followed the same procedure as described in the literature [12]. The reaction of (S)-pyrrolinamide 13 ( (15). The synthesis followed the same procedure as described in the literature [12]. The reaction of (S)-  To an ice-cooled solution in dry CHCl 3 containing the substituted benzoic acid 12a (0.97 g, 3.10 mmol), (S)-(1-Allylpyrrolidine-2-yl)methanamine 15 (0.43 g, 3.10 mmol, 1.0 equiv.) and pyridine (0.25 mL, 3.10 mmol, 1.0 equiv.) dicyclohexylcarbodiimide (0.64 g, 3.10 mmol, 1.0 equiv.) and N-hydroxysuccinimide (0.36 g, 3.10 mmol, 1.0 equiv.) were added under stirring. The mixture was stirred at 0 °C for 5 h and 17 h at ambient temperature. After the white precipitate had been filtered, the filtrate was washed with saturated NaHCO 3 -solution (20 mL) and subsequently with H 2 O (20 mL). After extracting the aqueous phase with diethyl ether (20 mL) the combined organic layers were dried with MgSO 4 , filtered and the solvent was evaporated to give an oily residue. The latter was purified by column chromatography (CHCl 3 /Ethanol = 20/1 → CHCl 3 /Ethanol = 10/1) to afford benzamide 4a (0.55 g, 1.27 mmol, 41%) as a yellowish oil. 1  The cartridge was washed with 10 mL isotonic saline and subsequently eluted with 1 mL ethanol. After dilution with 9 mL isotonic saline the radiotracers were measured, analyzed by radio-HPLC and used for plasma stability experiments.

(S)-1-Allylpyrrolidine-2-carboxamide
In vitro stability in human plasma. To human plasma (500 µL) at 37 °C were added 10-100 MBq of the injectable solutions. The mixture was incubated at 37 °C. After 90 min, aliquots (75 µL, in triplicate) were removed and treated with acetonitrile (75 µL). Samples were then stored on ice for 5 min for complete precipitation of the plasma proteins. The precipitate was removed by centrifugation, and the supernatants were analyzed by radio-HPLC. Radioactivity in precipitate and supernatant was measured.

Conclusions
Motivated by our recent development of a kit-like radiolabeling strategy for the introduction of 18 F-fluoride into proteins and target-specific peptides using Silicon-Fluoride-Acceptors (SiFAs) and their application in preclinical studies, we synthesized a series of SiFA-containing fallypride/ desmethoxyfallypride derivatives ( Figure 2). Compared to other medium-affinity D 2 -receptor ligands such as DMFP 2, the compounds showed a reduced affinity to the targeted receptor. However, the affinity is still in the nanomolar range and should therefore be high enough for an in vivo application. The radiochemical synthesis of the potential radiotracers, most notably the synthesis time of only 10 min and the easy cartridge purification, could be a breakthrough in radiofluorinations of small-molecule radiotracers. However, the potential of these SiFA-radiotracers remains to be shown in ongoing in vitro studies regarding receptor subtype-selectivity as well as Pgp-activity and finally in preclinical in vivo studies.