Synthesis and analysis of 4-(3-fluoropropyl)-glutamic acid stereoisomers to determine the stereochemical purity of (4S)-4-(3-[18F]fluoropropyl)-L-glutamic acid ([18F]FSPG) for clinical use

(4S)-4-(3-[18F]Fluoropropyl)-L-glutamic acid ([18F]FSPG) is a positron emission tomography (PET) imaging agent for measuring the system xC− transporter activity. It has been used for the detection of various cancers and metastasis in clinical trials. [18F]FSPG is also a promising diagnostic tool for evaluation of multiple sclerosis, drug resistance in chemotherapy, inflammatory brain diseases, and infectious lesions. Due to the very short half-life (110 min) of 18F nuclide, [18F]FSPG needs to be produced on a daily basis; therefore, fast and efficient synthesis and analytical methods for quality control must be established to assure the quality and safety of [18F]FSPG for clinical use. To manufacture cGMP-compliant [18F]FSPG, all four nonradioactive stereoisomers of FSPG were prepared as reference standards for analysis. (2S,4S)-1 and (2R,4R)-1 were synthesized starting from protected L- and D-glutamate derivatives in three steps, whereas (2S,4R)-1 and (2R,4S)-1 were prepared in three steps from protected (S)- and (R)-pyroglutamates. A chiral HPLC method for simultaneous determination of four FSPG stereoisomers was developed by using a 3-cm Chirex 3126 column and a MeCN/CuSO4(aq) mobile phase. In this method, (2R,4S)-1, (2S,4S)-1, (2R,4R)-1, and (2S,4R)-1 were eluted in sequence with sufficient resolution in less than 25 min without derivatization. Scale-up synthesis of intermediates for the production of [18F]FSPG in high optical purity was achieved via stereo-selective synthesis or resolution by recrystallization. The enantiomeric excess of intermediates was determined by HPLC using a Chiralcel OD column and monitored at 220 nm. The nonradioactive precursor with >98% ee can be readily distributed to other facilities for the production of [18F]FSPG. Based on the above accomplishments, cGMP-compliant [18F]FSPG met the acceptance criteria in specifications and was successfully manufactured for human use. It has been routinely prepared and used in several pancreatic ductal adenocarcinoma metastasis-related clinical trials.


Introduction
Positron emission tomography (PET) is a noninvasive molecular imaging technique that uses special radiotracers and scanners to obtain cross-sectional images of the distribution of radiotracers in live animals and humans. From PET images, in vivo localization and quantification of brain and somatic proteins (e.g., transporters, receptors, and enzymes) and investigation of biochemical or physiologic processes (e.g., metabolic rate and blood flow) are feasible in both patients and healthy subjects. In addition, PET could be used to directly assess the effectiveness of novel agents at a very early stage, which enables earlier decision-making for selection of potential drug candidates and helps to elucidate the mechanisms of new drugs. Currently, PET and PET tracers have been widely used in oncology, cardiology, and neurology for diagnosis, in which [ 18 F]fluorodeoxyglucose ([ 18 F]FDG) is the most popular of the PET drugs used in clinics. Although [ 18 F]FDG is useful in cancer diagnosis, staging, and treatment monitoring, new PET tracers aimed at other molecular targets for early tumor detection with higher sensitivity and specificity are still pursued by numerous researchers. Novel PET imaging agents could provide more specific pathological information other than the rate of glucose metabolism. This means they could furnish complementary data for the diagnosis and evaluation of therapy outcomes [1].
The system x C − transporter (xCT) mediates sodium-independent cellular uptake of cystine in exchange for intracellular glutamic acid at the cell membrane and is highly expressed in the brain, pancreas, and immune cells. xCT has been reported to be expressed in hepatocellular carcinoma, breast cancer, and lung cancer [2] and plays a critical role in the regulation of the redox status in proliferating tumors [3][4][5]. Therefore, xCT is a promising target for PET molecular imaging of tumor metabolism and oxidative stress [6]. (4S)-4-(3-Fluoropropyl)-Lglutamic acid (FSPG, (2S,4S)-1) is a substrate of xCT and the 18 F-labeled FSPG ([ 18 F]FSPG) has been studied as a PET tracer for in vivo determination of xCT expression and activity ( Fig  1) [6,7]. FSPG PET has been used for the detection of various cancers, such as brain [8], breast [7], lung, and liver cancers in clinical trials [8][9][10][11][12]. In addition, [ 18 F]FSPG was found to be a more sensitive tool for the diagnosis and treatment evaluation of multiple sclerosis and detection of lung cancer with myocardial and pericardial metastases [13] in comparison to [ 18 F] FDG [14,15]. FSPG PET could be used to measure the antioxidant capacity of tumor and predict drug resistance in chemotherapy [16][17][18]. Furthermore, [ 18 F]FSPG is a potential tool for the diagnosis and therapy evaluation of inflammatory brain diseases (e.g., stroke) [19] and infectious lesions, as in sarcoidosis [20]. Early detection of liver metastases from pancreatic ductal adenocarcinoma (PDAC) is still an unmet medical need, which is a major reason for the low survival rate of PDAC patients. Contrast-enhanced computerized tomography (ceCT) and FDG PET are the most used methods to monitor metastases before surgery. However, due to the high uptake of [ 18 F]FDG in the liver, the sensitivity, specificity and diagnostic accuracy of FDG PET for liver metastases from PDAC were not satisfactory. In an effort to develop a more sensitive diagnostic method for metastasized PDAC, we noticed that the xCT is highly expressed in the pancreas and PDAC, whereas the expression of xCT in the liver is significantly lower. Therefore, [ 18 F]FSPG could be a more sensitive diagnostic drug for early detection of liver metastases from PDAC, which would provide valuable information for clinicians to choose more appropriate treatment strategies. It would also help to develop new drugs and therapies for metastasized PDAC and aforementioned diseases [21].
Recently, we found FSPG PET consistently found more PDAC metastases earlier than FDG PET in a rodent model [21]. Based on the promising preclinical data, we decided to conduct clinical trials to evaluate whether the FSPG PET is a more sensitive tool than FDG PET and ceCT in detecting liver metastases from PDAC. Due to the very short half-life of the 18 Flabeled PET tracers (i.e., t 1/2 = 110 min), the preparation, purification, compounding, and quality control need to be completed in two hours to ready the cGMP-qualified [ 18 F]FSPG injection for clinical use. However, [ 18 F]FSPG has not been approved by any countries for routine clinical use yet. Therefore, the chemistry, manufacturing, and controls (CMC) for the production of [ 18 F]FSPG need to be thoroughly studied to establish appropriate specifications of a qualified [ 18 F]FSPG injection for clinical use. These are the information needed to submit an investigational new drug (IND) application to the local authority.

Materials and methods
General procedures and chemicals 1 H-and 13 C-NMR spectra were recorded on DPX-200, AV-400, and AV-600 FT-NMR spectrometers (Bruker, Germany) at 298 K. 1 H-Decoupled 19 F-NMR spectra were recorded at 377 MHz in CD 3 OD using an AVIII 400MHz NMR spectrometer (Bruker, Germany). Chemical shifts are expressed in parts per million (ppm) on the δ scale relative to a tetramethylsilane (TMS) internal standard. A solution of 5-20 mg of sample in 0.6-0.9 mL D-solvent was used in NMR experiments and no solvent signal suppression was used. Electron spray ionization (ESI) mass spectra and high-resolution mass measurements (HRMS) were obtained using an Esquire 2000 and Bruker Daltonik micro TOF mass spectrometer, respectively. Optical rotations were obtained using a Jasco P-2000 Digital Polarimeter (Tokyo, Japan) and are reported at the sodium D-line (589 nm). Thin-layer chromatography (TLC) was performed on Merck (art. 5715) silica gel plates and visualized under UV light (254 nm), upon treatment with iodine vapor, or upon heating after treatment with 5% phosphomolybdic acid in ethanol. Flash column chromatography was performed with Merck (art. 9385) 40-63 μm silical gel 60. Anhydrous tetrahydrofuran (THF) was distilled from sodium-benzophenone prior to use. No attempt was made to optimize yields.

High performance liquid chromatography analysis
Two high performance liquid chromatography (HPLC) systems: 1) SHIMADZU 1 LC-6AD equipped with a SPD-20A UV detector; and 2) SHIMADZU 1 LC-10AT equipped with a SPD-M10A DAD detector (Kyoto, Japan), were used in this study. The columns and mobile phase combinations were used as follows:

Results/Discussion
There are two chiral centers in FSPG, and therefore four FSPG stereoisomers (i.e.,

Synthesis of (2S,4S)-and (2R,4R)-isomers of FSPG
Preparation of (2S,4S)-1 and (2R,4R)-1 is shown in S1 Scheme. An initial attempt to synthesize di-tert-butyl protected glutamate (S)-3 by treatment of N-Boc-L-Glu-OH ((S)-2) with (Boc) 2 O and t-butanol in the presence of 4-(dimethylamino)pyridine (DMAP) failed to provide (S)-3. Esterification of (S)-2 with t-butanol and dicyclohexylcarbodiimide (DCC) successfully yielded (S)-3, but significant racemization was observed. In the beginning, the enantiomeric purity of (S)-3 was not measured until the target compound (2S,4S)-1 afforded later was found to be in poor enantiomeric purity. Due to the lipophilic character of compound 3, the enantiomer excess (ee) was determined by normal-phase chiral HPLC using a Chiralcel OD column and 1% 2-propanol (IPA) in n-hexane as mobile phase. The retention times (t R ) of (S)-3 and (R)-3 were 10.3 and 7.7 min, respectively (SI, Fig 41 in S1 File). N-Boc-L-glutamic acid 5-tert-butyl ester ((S)-4) was selected as an alternative starting material for the preparation of (2S,4S)-1. Treatment of (S)-4 with (Boc) 2 O, t-butanol, and DMAP produced (S)-3 in high yield and optical purity, whereas treatment of (S)-4 with DCC, t-butanol, and DMAP obtained (S)-3 with a low ee. The relationship between racemization and the reagents used in the t-butyl protection of (S)-4 was investigated to determine the critical factors for enantiomeric purity of (S)-3 ( Table 1). As shown in Entry 1, using 2.4 eq of DCC, 1.0 eq of DMAP, and 2.2 eq of t-BuOH provided (S)-3 in 44% ee. The same reaction could not proceed without the presence of DMAP (Entry 2), whereas 0.2 eq of DMAP was sufficient for esterification without significant influence over enantiomeric purity (37%, Entry 3). Using 0.94 eq of (Boc) 2 O and 0.5 eq of DMAP in an excess amount of t-BuOH (i.e., used as solvent) yielded (S)-3 in 93% ee. When using 0.2 and 0.05 eq of DMAP in the same reaction, the (S)-3 was obtained in high yields with 93 and 95% ee, respectively (Entry 5 and 6; SI, Fig 42 in S1 File). From the above data, DCC was the most critical factor to cause racemization and a catalytic amount of DMAP was sufficient for the t-butyl protection reaction using (Boc) 2 O with satisfactory enantiomeric purity.

Entry DCC (eq) DMAP (eq) t-BuOH (eq) (Boc) 2 O (eq) (S)-3 (ee, %)
1D and 2D NMR experiments were used for the stereochemical assignments of diastereomeric pyroglutamates 10 and 13. Selected chemical shift data is listed in Table 2, and the steric relationships between the protons in compounds 10 and 13 were determined by NOESY experiments, as shown in Fig 2. In the trans isomers (i.e., (2S,4R)-10 and (2S,4R)-13), correlations between the H 2α /H 3α and H 3β /H 4β were observed. In addition, a correlation was found between the H 3α and H 1' on the propyl side chain of (2S,4R)-13. In the cis isomers (i.e., (2S,4S)-10 and (2S,4S)-13), the NOE correlations were observed between the H 2α /H 3α and H 3α /H 4α . Moreover, a correlation was observed between the H 3β and H 1' on the allyl sidechain of (2S,4S)-10. Based on this data, the trans/cis configuration assignments could be established. Furthermore, the assignments of (2S,4R)-10 and (2S,4S)-10 in this study are consistent with those in the previous study, in which 1D NOE experiments were used to differentiate the two isomers [23].
From the chemical shift data shown in Table 2, several characteristic patterns are recognized for the 4-substituted pyroglutamate stereoisomers. The chemical shifts of the H 2α , H 3β , and C-3 for the trans isomers are more downfield (i.e., 0.05-0.07 ppm, 0.46-0.58 ppm, and 0.7-1.1 ppm, respectively) than those of cis isomers, whereas the chemical shifts of the H 3α for the trans isomers are more upfield (i.e., 0.47-0.59 ppm) than those of cis isomers. The chemical shifts of the H 4β for the trans isomers are more downfield (i.e., 0.08-0.09 ppm) than those of the H 4α for the corresponding cis isomers. Furthermore, the chemical shift differences between the H 3α and H 3β in the cis isomers (i.e., 0.75-0.90 ppm) are more prominent than those in the trans isomers (i.e., 0.18-0.27 ppm).

Analysis of FSPG stereoisomers
Because of the short half-life of 18 F-labeled PET drugs (i.e., 110 min), [ 18 F]FSPG needs to be prepared on a daily basis; therefore, fast analytical methods for quality control (QC) are especially essential and must be established to assure the safety and quality of [ 18 F]FSPG for clinical use. The activation of [ 18 F] fluoride by K 2.2.2. /K 2 CO 3 has been a commonly used method to form the naked and highly nucleophilic fluoride, but the basicity of K 2.2.2. /K 2 CO 3 may promote the epimerization of the C-2 position of [ 18 F]FSPG [22]. In fact, racemization and epimerization of the intermediates were also observed in other synthetic steps. Previous HPLC methods for the analysis of FSPG include using a Hypercarb column with a Corona detector and a C18 column with a precolumn derivatization using OPA-reagent and monitored at 340 nm [6]. To the best of our knowledge, these methods were only used to analyze diastereomeric (2S,4S)-1 and (2S,4R)-1 with moderate resolution [22], and no HPLC method has yet been reported for analysis of four FSPG stereoisomers. CuSO 4 has been used as an additive in the mobile phase for HPLC analysis, in which amino acid derivatives could be directly monitored by UV detector at 254 nm with sufficient sensitivity [25]. In this study, RP-HPLC analysis of FSPG stereoisomers using a Luna C18 column with a mobile phase of CuSO 4(aq) was conducted and monitored at 254 nm. All four stereoisomers displayed the same single peak at 4.4 min, which shows this method is useful for the determination of chemical purity of compound 1 but is not suitable for the determination of enantiomeric or diastereomeric purities of compound 1. A concise and efficient analytic method for the determination of all four FSPG stereoisomers was pursued.
The Chirex 3126 (d)-penicillamine column has been widely used in the analysis of amino acids, including glutamic acid and aspartic acid. When a solution of 2 mM CuSO 4(aq) /MeOH (85:15) was used as mobile phase, enantiomers of aspartic acid were separated with high resolution by HPLC. Therefore, an aqueous CuSO 4 -based mobile phase and Chirex 3126 column were chosen as the starting points for the development of a more convenient and faster HPLC method for analysis of four FSPG stereoisomers without derivatization. Initial analysis using Chirex 3126 (d)-penicillamine column (150 mm × 4.6 mm), with CuSO 4(aq) as the mobile phase, and monitored at 254 nm was not satisfactory. Due to the significantly higher lipophilicity of FSPG than that of glutamic acid, a much higher content of organic modifier was needed to elute stereoisomers of 1. This response therefore significantly increased the back pressure of the chromatographic system. For Chirex 3126 column, the recommended limits for 2-propanol (IPA) and acetonitrile (MeCN) are <15%, and the upper limits of back pressure was 3000 psi. Using Chirex 3126 column (150 mm) with a MeOH/CuSO 4(aq) mobile phase (15:85; 27˚C), none of FSPG stereoisomers could be eluted out in 60 min (Fig 3a). When IPA/CuSO 4 (aq) was used as mobile phase (15:85), only (2R,4S)-1, (2S,4S)-1, and (2R,4R)-1 were eluted out in 30 min at 25˚C (Fig 3b); whereas (2R,4S)-1, (2S,4S)-1, and (2R,4R)-1 could be eluted out in 20 min at 35˚C (Fig 3c). Under these conditions, the absorption peaks of (2R,4S)-1 and (2S,4S)-1 were completely overlapped. By using Chirex 3126 column (150 mm) with a MeCN/ CuSO 4(aq) mobile phase, the resolution of FSPG stereoisomers was significantly increased. The a The precipitated crystals possessed lower ee than the starting material, whereas the mother liquors exhibited higher ee than the starting material.
b Depicted are the ee values and the recovery of enriched (2S,4S)-6 in the mother liquors that were calculated based on the weight of (2S,4S)-6 used in each recrystallization. https://doi.org/10.1371/journal.pone.0243831.t003 (2R,4S)-1, (2S,4S)-1, and (2R,4R)-1 were eluted out in sequence with t R of 35.6, 39.3, and 46.1 min at 25˚C (Fig 3d), and with t R of 28.9, 31.5, and 38.0 min at 35˚C (Fig 3e), respectively. The t R and resolution of FSPG stereoisomers were sensitive to the temperature and organic modifiers: lower temperature resulted in longer t R and higher resolution; MeCN provided higher resolution and longer t R than IPA. The (2S,4R)-1 could not be eluted out under the above analytical conditions using Chirex 3126 column (150 mm). Furthermore, the long t R seems to be not compatible for analysis of the short half-life 18 F-labeled PET tracers. Therefore, chiral HPLC analysis using shorter Chirex 3126 column was conducted and the results are shown in Fig 4. Using a combination of two Chirex 3126 pre column (30 mm × 4.6 mm × 2) with MeCN/CuSO 4(aq) mobile phase successfully separated four FSPG stereoisomers in less than 40 min (25˚C ; Fig 4a), 35 min (30 C ; Fig 4b), and 30 min (30˚C ; Fig 4c), respectively.

Radiosynthesis of [ 18 F]FSPG
The [ 18 F]FSPG injection was prepared in the Radiochemistry Laboratory, PET center, National Taiwan University Hospital (NTUH). All the radiosynthesis and compounding operations for production of the [ 18 F]FSPG injection followed the Current Good Manufacturing Practice (cGMP) regulations for human pharmaceuticals standards and is regularly inspected by the Taiwan Food and Drug Administration (TFDA). A detail flow chart of the general procedure for the synthesis of [ 18 Table 2 in S1 File. [ 18 F] FSPG injection is a clear, colorless, and sterile solution. The radiochemical and enantiomeric purity should be not less than 90%, and the diastereomer contents should be not more than 5%. The limits of residual solvent are � 0.5% for EtOH, � 0.04% for MeCN, and � 0.5% for acetone, respectively. [ 18 F]FSPG injection contains residual Kryptofix 2.2.2 less than 50 μg/mL with a pH range of 5-8. [ 18 F]FSPG injection should be stored at controlled room temperature (18~25˚C) with an expiration time of 4 hours.