Fully Automated Radiosynthesis of [¹⁸F]mG4P027 for mGluR4 Imaging

Abstract

Fluorine-18 labeled N-(4-chloro-3-(((fluoro-¹⁸F)methyl-d₂)thio)phenyl)picolinamide, [¹⁸F]mG4P027, is a potent positron emission tomography (PET) radiotracer for metabotropic glutamate receptor 4 (mGluR4). Our previous in vitro and in vivo evaluations have demonstrated that this tracer is promising for further translational studies. To automate the radiosynthesis of [¹⁸F]mG4P027, significant modifications were made to the manual process by carefully examining this process and addressing the root causes of the challenges associated with its automation. We successfully implemented its automated radiosynthesis using the TRACERlab FX2N module and consequently, obtained a high-purity radiolabeled [¹⁸F]mG4P027 in high yield, meeting the requirements for future human studies.

Graphical Abstract

The radiosynthesis of some radiotracers, such as [¹⁸F]mG4P027, requires a challenging two-step labeling process. In this study, we successfully automated this process using an FX2N module, achieving a high yield of radiolabeled [¹⁸F]mG4P027 with high purity suitable for human studies.

1. Introduction

Glutamate is the primary excitatory neurotransmitter in the central nervous system (CNS) of vertebrates, responsible for transmitting signals between neurons at over 50% of all synapses.(1) It acts as a neurotransmitter by binding to specific receptors in the CNS, which can be broadly categorized into two groups: ionotropic glutamate receptors (iGluRs) and metabotropic glutamate receptors (mGluR1–8). Dysregulation of mGluR signaling has been implicated in a range of neurological and psychiatric disorders, including neurodegeneration, addiction, anxiety, and depression, and thus drugs that have been developed to target mGluRs as potential therapeutics for these conditions. Of particular interest is presynaptic mGluR4, which is a promising target for treatment of various neurological diseases, especially Parkinson’s disease (PD).(2–4)

PET is a powerful and minimally invasive nuclear imaging technique used for clinical diagnosis and treatment monitoring, as well as for drug discovery and development, owing to its picomolar sensitivity (pM) and fully translational capability.(5–8) Regardless of the application, all PET imaging studies necessitate the use of a suitable tracer with a positron-emitting radioisotope. The availability of efficient PET radioligands for mGluR4 could facilitate investigations into the role of mGluR4 in both healthy and disease conditions as well as the development of novel drugs targeting mGluR4.(9) Therefore, we have designed and developed several mGluR4 tracers based on mGluR4 positive allosteric modulators (PAMs).(10–12) One such tracer, [¹¹C]mG4P012, was later renamed as [¹¹C]PXT012253 by Prexton Therapeutics, underwent evaluation in a clinical trial (NCT03826134) as a PET tracer for mGluR4 and served as a biomarker to support the phase II trial of a potential therapeutic drug, PXT002331 (Foliglurax), for Parkinson’s disease (PD) and levodopa-induced dyskinesia (LID).(2–4, 13)

Fluorine-18 (¹⁸F) is the most preferred radionuclide for PET imaging, owing to its ideal physical properties, such as long half-life (t_1/2 = 109.7 min). Specifically, its half-life is longer than that of carbon-11 (¹¹C), which enables the application of more extensive imaging protocols, leading to detailed kinetic studies and high-quality metabolic and plasma analyses. Additionally, the low positron energy associated with ¹⁸F decay results in high imaging resolution, making ¹⁸F a preferable choice for PET imaging studies. To develop a tracer labeled with ¹⁸F, we replaced the 3-methyl group of [¹¹C]mG4P012 with a 3-dideuterium-fluoromethyl moiety to obtain [¹⁸F]mG4P027, as shown in Scheme 1. Replacing the ¹¹C-methyl group with ¹⁸F-fluoromethyl group induces the least structural modification in the parent compound. Moreover, incorporating dideuterium in [¹⁸F]mG4P027 reduces the rate of in vivo defluorination initiated by metabolic cleavage of the C–H bond without altering the tracer’s binding affinity for mGluR4. The results indicated that [¹⁸F]mG4P027 is an improved PET tracer for imaging mGluR4, exhibiting good affinity, enhanced metabolic stability, higher brain uptake, and improved imaging capabilities (Table 1).

Scheme 1.

Structure modification from [¹¹C]mG4P012 to [¹⁸F]mG4P027.

Table 1.

Properties of mGluR4 PET tracers [¹¹C]mG4P012 and [¹⁸F]mG4P027

	[11C]mG4P012	[18F]mG4P027
cLogP	3.62	3.34
Binding affinity (nM)	3.5	3.5
Microsome stability t1/2 (min)	25.3	57.3
Cerebellum uptake (ID/cc%)	1.2	2.0
Radioisotope half-life (min)	20.4	109.7

Successful implementation of the radiosynthesis of PET tracers on cassette-based automation platforms is critical for translational research and future applications.(14, 15) Automated radiochemistry not only enables batch-to-batch reproducibility in PET radiopharmaceutical production,(16, 17) which is crucial for compliance with Good Manufacturing Practice (GMP) requirements,(18) but also provides radiation protection to users handling a large amount of activity.(19) While automated radiosynthesis of a specific tracer relies largely on the automation system available within the institute, it can be very challenging to implement multistep radiosynthesis in most cases. Herein, we report the fully automated two-step radiosynthesis of [¹⁸F]mG4P027 using a commercially available radiosynthesis module, GE TRACERlab FX2N platform.

2. Results and Discussion

We have described a manual radiosynthesis protocol for [¹⁸F]mG4P027.(20) As shown in Scheme 2, the manual radiosynthesis of [¹⁸F]mG4P027 was carried out in two steps: (1) The generation of [¹⁸F]fluoromethyl-d₂-4-methylbenzenesulfonate [¹⁸F]2 by [¹⁸F]fluorination of methylene-d₂ bis(4-methylbenzenesulfonate) 1, followed by (2) the radiosynthesis of [¹⁸F]mG4P027 by the reaction of thiophenol precursor 4 with [¹⁸F]2 (Scheme 2). This manual protocol includes two semi-prep high-performance liquid chromatography (HPLC) purifications for [¹⁸F]2 and final product [¹⁸F]mG4P027. However, automation of multiple-step reactions for translational research is a considerable challenge because extensive modification of manual protocols is typically required to implement novel ¹⁸F-labeling methodologies within automated modules.

Scheme 2.

Two-step radiosynthesis of [¹⁸F]mG4P027.

To develop an efficient automated protocol, we studied the key factors of the two-step manual labeling process of [¹⁸F]mG4P027 and analyzed the challenges involved in automating this process.(21) The results showed that (1) t-BuOH used in the first step causes problems during the Sep-Pak trapping of [¹⁸F]2; (2) remaining starting martial 1 and product [¹⁸F]3 affected the second step of the reaction, in which bistosylate 1 was more reactive than [¹⁸F]2 and reacted with 4 to give 6 (if 1 was not removed); (3) both steps were sensitive to water and because a large amount of water was introduced for the first Sep-Pak purification step, drying [¹⁸F]2 was a significant challenge; and (4) under neutral conditions, thiophenol precursor 4 easily transformed into its dimer 7 (Scheme 3). Furthermore, the automation of this process in a hot-cell environment presented additional hurdles that must be overcome. A schematic diagram of the commercial TRACERlab FX2N radiosynthesis module (GE Healthcare, Waukesha, WI) used for the synthesis of [¹⁸F]mG4P027 is shown in Figure 1.

Scheme 3.

Side reactions with compound 4.

Figure 1.

Schematic diagram of the TRACERlab FX2N radiosynthesis module.

As shown in Figure 1, the TRACERlab FX2N module consists of two separate reactors which can accommodate the two-step reactions. However, it is equipped with only one semi-prep HPLC unit, which, according to the module’s configuration, is designated for the purification of the final product. Consequently, an alternative approach was required to purify intermediate [¹⁸F]2 and remove unreacted starting material 1 from the first-step reaction mixture without relying on HPLC purification. In the first step of our manual protocol, the synthesis of [¹⁸F]2 was carried out in 25% t-BuOH in acetonitrile, with subsequent t-BuOH removal achieved through semi-prep HPLC purification.(20) However, the presence of this significant amount of alcohol posed challenges for trapping intermediate [¹⁸F]2 using a Sep-Pak column due to the dead volume of the reactor and the column, as well as the polarity of t-BuOH. Hence, in the automated module, we explored an alternative approach. It has been reported that the addition of a small amount of water to the reaction significantly increased the yield of [¹⁸F]2.(22) Therefore, we studied the effect of water content in the reaction on the yield of [¹⁸F]2 by using the FX2N module. According to our observation, a water content of greater than 3% results in complete failure of the reaction, whereas according to a previous report, a water content of up to 5% water still yielded 76% of [¹⁸F]2.(22) Furthermore, in dry acetonitrile, the predominant product obtained is undesired [¹⁸F]3.(20, 22) In our automated system, key intermediate [¹⁸F]2 was consistently produced by using 1% water.

Nevertheless, in our previous manual radiosynthesis, we used 12.0 mg of di-tosylate 1 to ensure successful labeling, and the excess amount of 1 was subsequently removed via semi-prep HPLC purification.(20) In our efforts to optimize the automated synthesis, we reduced the amount of 1. However, we discovered that using less than 1.0 mg of 1 in the automated system resulted in either trace amounts of [¹⁸F]2 or complete failure, while using 2.0 mg of 1 consistently delivered [¹⁸F]2 in yields ranging from 30% to 57%.

To obtain/extract [¹⁸F]2 from the reaction mixture, the reaction solution was diluted with 10 mL of water, then passed through a C₁₈ Sep-Pak, and finally washed with an additional 5 mL of water. [¹⁸F]2 trapped on the C₁₈ Sep-Pak was dried using our previous procedure.(20) The trapped mixture containing 1, [¹⁸F]2 and [¹⁸F]3 was then washed off with DMSO and used directly for the second reaction.(20) Unfortunately, product [¹⁸F]mG4P027 was not obtained, indicating the need to remove 1 from the mixture used in the second reaction. As another possible approach, we increased the amount of 4 in the second reaction to enhance the reaction between [¹⁸F]2 and 4. However, we found that the amount of compound 4 could not exceed 1.5 mg because 4 tended to form compounds 6 and 7, leading to clotting caused by the poor solubility of compound 7 (Scheme 3).(21)

Therefore, we sought an alternative approach to remove 1 prior to the second reaction without disturbing [¹⁸F]2. recent study showed that azide reagents, especially tetrabutylammonium azide (n-Bu₄NN₃), can scavenge >99.5% of di-tosylate 1 without significantly attacking (approximately 4%) key intermediate [¹⁸F]2 in acetonitrile (ACN) at 80 °C.(23) However, it was also reported that only 43% of 1 and 3% of [¹⁸F]2 in t-BuOH can be consumed, which was another reason we aimed to avoid using t-BuOH for the first step. By using 20 mg of n-Bu₄NN₃, we scavenged most of 1 and a large amount of [¹⁸F]3, while preserving [¹⁸F]2, providing new hope for this complex reaction (ESI, Figures S1–3).

Considering all these factors, we initially performed the labeling according to our previous manual method, except the first prep-HPLC step.(20) Briefly, we tested the first reaction in 1% water in acetonitrile, with the addition of 20 mg of n-Bu₄NN₃ in 0.5 mL of acetonitrile at 80 °C for 10 min. The reaction mixture was then diluted with 10 mL of water and passed through a C₁₈ Sep-Pak to trap [¹⁸F]2, which was washed with an additional 5 mL of water. After drying for 30 min, [¹⁸F]2 was washed off the cartridge with dry DMSO (ESI, Figure S4), and then used in the second step to give >99% conversion (based on [¹⁸F]2) at 135 °C.(21)

Of note is that 4 and Cs₂CO₃ are added right before the second reaction in the manual labeling. However, we noticed that if 4 was loaded in the delivery vials in the saturated Cs₂CO₃ DMSO solution (without extra solid base Cs₂CO₃ available), then 7 formed before the second reaction (1 h later), which was visibly detected as a color change.(21) Furthermore, if loaded in the delivery vials, 10 mg of solid base Cs₂CO₃ is not proper to deliver into reactor 2 within the FX2N module through the fine tubing. Therefore, 4 was used in its solid state as in the manual labeling, preloaded in reactor 2 with a base, Cs₂CO₃.

As discussed above, we have modified the previous manual synthesis protocol and developed a new method to automate the synthesis. With this new method in manual operation, we obtained relatively clean intermediate [¹⁸F]2 (Figure 2 and ESI, Figure S4) and successfully performed the radiosynthesis of [¹⁸F]mG4P027. With success in manual synthesis, we applied this new method to automated synthesis. However, we found that the second reaction did not deliver product [¹⁸F]mG4P027, with the trapped intermediate [¹⁸F]2 was dried for 30 min by either helium blowing (ESI, Figure S5) or vacuum drying methods (ESI, Figure S6).

Figure 2.

HPLC radioactivity trace of [¹⁸F]2 and [¹⁸F]mG4P027.

A large amount of water was introduced in the first Sep-Pak purification step, and thus drying the Sep-Pak column containing [¹⁸F]2 was critical for the second-step reaction, which is very sensitive to water.(21) To understand the issues with the automation system, we compared the drying methods used in the manual operation and automation system by using two C₁₈ cartridges to trap [¹⁸F]2 from the first-step reaction. One C₁₈ cartridge was used in the FX2N module, while the other was removed from the box for manual drying and labeling. After conducting manual labeling, we were able to produce [¹⁸F]mG4P027 with >99% conversion (based on [¹⁸F]2) by reacting resulting [¹⁸F]2 with the C₁₈ Sep-Pak column directly connected to a nitrogen tank.(20, 21) In contrast, [¹⁸F]mG4P027 was not detected following the reaction that took place in the FX2N module. This clearly indicates that the drying process in the box was not as efficient as in the manual labeling. Upon further analysis of the drying procedures, we discovered that water presented in reactor 1 and the waste bottle might introduce additional moisture into the lines. Additionally, both helium blowing and vacuum drying were insufficient to completely remove the moisture. Therefore, we changed the drying loop by using different ducts, in which no additional water was present (ESI, Figure S7). After drying for 40 min using the new drying ducts, [¹⁸F]2 was washed off and transferred to reactor 2 (ESI, Figure S8). We were delighted to obtain > 97% [¹⁸F]mG4P027 by using this new drying method (Figure 2 and ESI, Figure S9). These encouraging labeling results were repeated 5 times for quality control. High-purity [¹⁸F]mG4P027 (>99%) was finally produced in approximately 22% radiochemical yield (decay corrected) within 120 min, including semi-prep HPLC purification, with the fully automated system(Figure 2). The identity of [¹⁸F]mG4P027 was confirmed by co-injecting the cold compound for HPLC analysis (ESI, Figure S10).

3. Conclusion

In conclusion, automated radiosynthesis of [¹⁸F]mG4P027 was achieved through careful analysis of the automation system and the inherent challenges of the reactions involved. The use of n-Bu₄NN₃ for scavenging excess compound 1 along with an efficient drying procedure played a key role in the success of the radiosynthesis. The water sensitivity of the reactions was addressed by taking advantage of water effects in the first reaction and drying resulting intermediate [¹⁸F]2 via a different duct to avoid water in the system for the second step. Additionally, keeping thiophenol precursor 4 in the solid state until the second reaction was ready to proceed prevented dimerization of 4 to 5. Final product [¹⁸F]mG4P027 was produced in a reasonable yield with high purity suitable for clinical research. An IND application for human studies is currently underway. The successful synthesis of [¹⁸F]mG4P027 underscores the significance of meticulous optimization and troubleshooting in automated radiosynthesis, and we are optimistic that this study will serve as an inspiration for others in their pursuit of effective radiolabeling techniques.

4. Experimental

4.1. General

All reagents and starting materials were obtained from commercial sources including Sigma Aldrich (St. Louis, MO, USA), Thermo Fisher Scientific (Cambridge, MA, USA), Oakwood Chemical (N, Estill, SC, USA) and used as received. The precursors, intermediates and final products have been reported. (20) A 16/8.5-MeV cyclotron (PETtrace, GE Healthcare, Waukesha, WI) was used for [¹⁸F]F⁻ radionuclide production. GE high-yield niobium targets containing >97% enriched oxygen-18 water (Taiyo Nippon Sanso, Tokyo, Japan or Rotem Medical, Topsfield, MA) were bombarded with protons at integrated of currents up to 65 μA to produce [¹⁸F]F⁻. QMA-light cartridges and C₁₈ light cartridges were purchased from Waters (Milford, MA, USA). The GE TRACERlab^™ FX2N radiosynthesis module was used for the synthesis of [¹⁸F]mG4P027 (Figure 1). The reactions were monitored by a HPLC system (1260 Infinity II, Agilent, Lexington, MA, USA) coupled with a multi-wavelength ultraviolet detector equipped with a C₁₈ column (XBridge, 150 × 4.6 mm, 3.5 μm, 130 Å). The final labeled compound was purified with a reverse-phase C₁₈ semi-prep column (XBridge, 10 × 250 mm, 5 μm, 130 Å). Preparative HPLC was conducted at a flow rate of 5 mL/min using 100% 0.1 M ammonium formate in water for 5 min, followed by 40% ammonium formate in water and 60% ACN to obtain the final product. Analytical HPLC was carried out at a flow rate of 1.0 mL/min using a 15-min gradient method: eluent A: 0.1 M ammonium formate/H₂O; eluent B: CH₃CN; gradient: 50% B for 3 min, then 50% B to 95% B from 3 to 11 min, 95% B from 11 to 12 min, 95% B to 50% B from 12 to 15 min.

4.2. Preparation for the automation

Before delivering [¹⁸F]F⁻ to the TRACERlab^™ FX2N synthesis module, each vial was filled with the appropriate solvents and/or reagents. Vial 1 was filled with 5.0 mg of tetraethylammonium bicarbonate (TEABC) dissolved in 0.5 mL of water and 0.5 mL of ACN. Vial 2 was filled with 15 mL of water. Vial 3 was loaded with 2.5 mg of compound 1, which served as the precursor in the first step. Compound 1 was dissolved in 0.8 mL of ACN containing 1% water. Vial 4 was filled with 1.0 mL of ACN. Vial 5 was filled with 20 mg of tetrabutylammonium azide dissolved in 0.5 mL of ACN. Vial 6 was filled with 1.0 mL of anhydrous dimethyl sulfoxide (DMSO). Vial 9 was filled with 1.0 mL of HPLC eluent, which consisted of 0.1 M aqueous ammonium formate. Vial 42 was filled with 1.0 mL of ethanol and vial 43 was filled with 9.0 mL of saline. Reactor 2 was loaded with 1.5 mg of the hydrogen chloride salt of solid 4 as the second-step precursor, along with 10 mg of Cs₂CO₃. A QMA cartridge preconditioned with 6 mL of water was placed between vials V10 and V11. Two C₁₈ Sep-Paks preconditioned with 4 mL of ethanol and then 10 mL of water were placed between vials VX3 and VX4, and between V17 and V15, as shown in Figure 1. The round-bottomed flask used to collect the HPLC fractions was filled with 25 mL of water to facilitate the trapping of [¹⁸F]mG4P027 on a C₁₈ Sep-Pak. The synthesis module was operated using an automated program. Following bombardment, [¹⁸F]F⁻ was transferred to the TRACERlab^™ FX2N radiosynthesis module via a helium gas stream.

4.3. Automation procedures

At the end of the bombardment, aqueous [¹⁸F]F⁻ in [¹⁸O]H₂O was transferred from the target into a collection vial (A) containing [¹⁸O]H₂O via a helium purge. [¹⁸F]F⁻ was then trapped in a QMA cartridge (B), subsequently eluted from the QMA cartridge and finally transferred to reactor 1 (C) using the solution in vial 1. The mixture was dried under azeotropic conditions, and after the first drying, the residue was dissolved in ACN from vial 4 and then further dried. Tosylate precursor 1 in vial 3 was transferred to reactor 1 (C) and heated at 90 °C for 10 min. After cooling, the azide solution in vial 5 was added followed by heating at 90 °C for another 5 min. After cooling down the reactor, the azide solution from vial 5 was added, and the mixture was heated to 90°C for an additional 5 min. After the reactor was cooled to 40 °C, 10 mL of water in vial 2 was added, and the diluted reaction mixture was transferred to the C₁₈ Sep-Pak (D) to trap [¹⁸F]2. The remaining 5 mL of water in vial 2 was then used to wash the Sep-Pak. The C₁₈ Sep-Pak was first dried using drying method 1 (ESI, Figure S5) for 5 min, followed by an additional 25 min of drying using drying method 3 (ESI, Figure S7). Then, [¹⁸F]2 was washed off C₁₈ Sep-Pak with anhydrous DMSO in vial 6 and transferred to reactor 2 (E) using the loop shown in Figure S8. Reactor 2 was then heated at 135 °C for 10 min followed by semi-prep HPLC purification to afford the final product.

DATA AVAILABILITY STATEMENT

The data that support the findings in the study are available in the supporting information of this article.