An Automated and Putatively Versatile Approach for Labeling Peptides with Fluorine-18 Using the Fastlab TM Synthesizer; Exemplied with Glu-the CO-Lys Pharmacophore

Background: Noninvasive molecular imaging using peptides and biomolecules labelled with positron emitters has become important for detection of cancer and other diseases with PET (positron emission tomography). The positron emitting radionuclide uorine-18 is widely available in high yield from cyclotrons and has favorable decay (t 1/2 109.7 min) and imaging properties. 18 F-Labelling of biomolecules and peptides for use as radiotracers is customarily achieved in a two-step approach, which can be challenging to automate. 6-[ 18 F]Fluoronicotinic acid 2,3,5,6-tetrauorophenyl ester ([ 18 F]F-Py-TFP) is a versatile 18 F-prosthetic group for this purpose, which can be rapidly be produced in an one-step approach on solid support. This work details an automated procedure on the cassette-based GE FASTlab TM platform for the labeling of a peptidomimetic, exemplied by the case of using the Glu-CO-Lys motif to produce [ 18 F]DCFPyL, a ligand targeting the prostate specic membrane antigen (PSMA). Results: From uorine-18 delivery a fully automated two-step radiosynthesis of [ 18 F]DCFPyL was completed in 56 min with an overall end of synthesis yield as high as 37% using SPE purication on the GE FASTlab TM platform. Conclusions: Putatively, this radiolabeling methodology is inherently amenable to automation with a diverse set of synthesis modules, and it should generalize for production of a broad spectrum of biomolecule-based radiotracers for use in PET imaging.


Background
Cancer is the leading cause of premature deaths in the western world, a circumstance that has spurred the development of new radiopharmaceuticals for cancer diagnosis by positron emission tomography (PET) and for radiotherapy using α-and β-emitting radionuclides (Holland et al., 2012;Zhang et al., 2017).
For use in clinical PET imaging, the radiohalogen uorine-18 ( 18 F) is the preferred radioisotope due to its physical half-life (109.7 min), favorable decay properties, and its ready availability in multi-GBq quantities from medical cyclotrons. A considerable proportion of emerging radiopharmaceuticals are based on small to medium sized peptides, fusion proteins, and endogenous proteins. However, these products are not always obtainable through a direct radiolabeling approach due to their instability at high temperature or the otherwise harsh reaction conditions required to introduce uorine-18 into their structure. Although labeling of such molecules with uorine-18 using a chelator conjugated to the targeting moiety with A robust and high yield radiosynthesis amenable to automation is a prerequisite for adopting a PET radiopharmaceutical for widespread clinical or research use (Li et al., 2013). Production of PET radiopharmaceuticals with a multi-step synthetic process can be a challenging task using commercially produced cassette-based modules, mainly due to their lack of modi ability, and the frequent requirement  -TFP and the PSMA-avid substrate Glu-CO-Lys protected with t-butyl has subsequently been optimized (Chen et al., 2011). Currently, a direct precursor labeling with the trimethylammonium nicotinic acid moiety conjugated to the ε-nitrogen of the lysine residue of Glu-CO-Lys is the favored method for automated multi-dose production of [ 18 F]DCFPyL (Dornan et al., 2018), which has greatly facilitated the use of PSMA PET in clinical routine. However, this introduction of uoride-18 in organic solvents at a temperature approaching 100°C, is unlikely to provide high labeling e ciencies, or to generalize for larger molecular weight radiopharmaceuticals such as polypeptides and proteins with their bountiful functional groups.
Fairly recently, a simplistic and rapid preparation of [ 18 F]F-Py-TFP has been reported, wherein the prosthetic group is obtained directly by passing its precursor in a solution of acetonitrile/tert-butanol through a polymeric anion-exchange cartridge preloaded with [ 18 F] uoride (Basuli et al., 2016;Olberg, 2017). This procedure yields the bifunctional labeling agent in good yield, with an overall synthesis time of less than ten minutes (EOB), simultaneously avoiding the use of a reaction vessel and a phase transfer catalysts. The on-column radiosynthesis of [ 18 F]F-Py-TFP has a very small laboratory footprint and lends itself well for automation and offer radiolabeling of a wide variety of sensitive biomolecules (proteins eg.)
Reversed phased-analytical HPLC was performed on a Phenomenex kinetex EVO C18 column (50 × 2.1 mm, 2.6 µm particles), at a ow rate of 0.5 mL/min, with a composition gradient from 2-95% solvent B (method 1) over ten min (solvent A: water/0.05% TFA, solvent B: acetonitrile/0.05% TFA). UV detection was at 214, 254 and 264 nm. Hydrophilic interaction liquid chromatography (HILIC) was performed on a SeQuant ZIC-cHILIC (100 × 4.6 mm, 3 µm particles) at a ow rate of 2.0 mL/min, with composition gradient from 2-98% solvent B (method 2a) or isocratically with 68% solvent B (method 2b) over ten min (Solvent A: 97% acetonitrile/3% 5 mM NH 4 OAc, solvent B: 5 mM NH 4 OAc). UV detection was at 210, 220, 254 and 264 nm. Thin layer chromatography (TLC) was run on gel 60F 254 plates (Merck) using acetonitrile as eluent. A Raytest miniGita (Raytest, Germany) equipped with a β-detector was used to record the radio-TLC scan. Gas chromatography (GC) was performed with an Agilent 6890N GC (Matriks, Norway) equipped with a ame ionization detector (FID), a column oven, and an auto-sampler with a direct injection system. The GC column was a fused silica capillary column with USP stationary phase G43 (6% cyanopropylphenyl -94% dimethyl polysiloxane) measuring 0.32 mm i.d. x 30 m. Automated radiochemistry was performed on the GE FASTlab synthesizer with the custom-made single-use FASTlab development kit (GE Healthcare) placed inside a lead-shielded hot cell. Fluorine-18 was produced in a GE PETtrace 6 cyclotron (General Electric Healthcare) with a GE 18 F -Nb 27 self-shielded target either by irradiation (40 µAh × two min) of [ 18 O]H 2 O (Taiyo Nippon Sanso, Japan), or from a target rinse after a routine 100 GBq production. No more than 150 MBq of radioactivity were handled outside the hot cell. Radioactivity was measured with a Capintec dose calibrator (New Jersey, USA). The identities of the radiolabeled products were con rmed by co-injection to HPLC along with authentic nonradioactive standards.
General radiolabeling procedure (manual) Aqueous [ 18 F] uoride was trapped on a Chromabond PS-HCO 3 anion cartridge followed by rinsing with hot acetonitrile (2 mL) and passage of air (2 × 5 mL) to dryness. A solution of precursor 2 (10 mg/mL, 22 mM), in the presence of TEA in t-BuOH/acetonitrile (50%, v/v, 1 mL) was used to initiate radiolabeling and to elute the radioactivity into a receiving vial (10 sec); the eluent was passed through the column back and forth into the syringe barrel for 20 seconds. After the nal elution into the vial, the reaction was quenched by rinsing the column with water (0.5 mL). We measured the radioactivity concentration in the quenched reaction mixture before analysis by HPLC and TLC with on-line gamma detection.
General automated radiolabeling procedure Cassettes were assembled according to the layout depicted in Fig. 2. External vials were vented with a syringe lter and connected to the cassette manifold with a silicone tube of speci c length, terminating in a needle as follows: 30 mL vial of H 3 PO 4 0.85%: 42-cm long tube, 0.80 i.d. × 80 mm needle (4) Collection vial: 13-cm long tube, 0.80 i.d. × 80 mm needle. This vial was changed during the synthesis from a 7 mL vial to collect the HILIC y-through (3 mL) to a 20-mL vial to collect the Oasis HLB PRiME y-through (5 mL) and washout (5 mL) (13) 25% ethanol vial: 13-cm long tube, 2.10 mm i.d. × 80 mm needle (12) Vials were prepared as follows: In brief, uorine-18 was trapped on the Chromabond PS-HCO 3 and then the column was dried with anhydrous acetonitrile. Fluorination occurred on-cartridge when the mixture of precursor 2 and base in acetonitrile/t-BuOH was cycled through the cartridge. The radiochemical yield is dependent of precursor concentration and the presence of an non-nucleophilic organic base helps to increase yield and reduce variantion between synthesis runs.
RCYs improved when the Chromabond column was dried using acetonitrile heated to 80°C in the reactor vessel. The best RCY (78±9%) was obtained using 11 mM of precursor and 0.2 equivalents of triethylamin (TEA) as the base. Translating these conditions (including SPE puri cation) into the FASTlab, [ 18 F]F-Py-TFP was obtained with RCY of 63±4% with TEA as the base. In our hands, yields were 15% lower using the automated FASTlab synthesis compared to manual synthesis. For the automated FASTlab The automated radiosynthesis entailed three sequential processes, i.e a ten min pre-labelling step, a 25 min on-cartridge radiolabeling step for [ 18 F]F-Py-TFP (including puri cation), and a 31 min conjugation and puri cation step affording the nal product [ 18 F]DCFPyL. Due to the inherent dead volumes, when implementing the radiochemistry to the FASTlab synthesizer, the reagent vials were lled with slightly larger volumes to retain the same stoichiometric amounts of reagents.

Pre-labelling
The vials, cartridges and reactor were conditioned for ten minutes before delivery of radioactivity. Vials were pressurized with 1 bar nitrogen to ensure optimal syringe lling. Ethanol and water were successively used to condition the tC18 Sep-Pak light cartridge. Acetonitrile (1.6 mL) was transferred into the FASTlab reactor and heated; the system was deemed ready to receive radioactivity when the reactor temperature reached 80°C.

Radiolabeling and puri cation of [ 18 F]F-Py-TFP
The next step provided puri ed [ 18 F]F-Py-TFP bound to the tC18 Sep-Pak light cartridge in 25 minutes. First, uorine-18 was trapped in the Chromabond PS-HCO 3 cartridge, followed by drying with two passages of hot acetonitrile drawn from the heated reactor. The acetonitrile was successively sent to syringes 2 and 3 to remove any water remaining from the pre-labeling step. Using syringe 1, a 1 mL volume of the precursor solution was pushed through the Chromabond PS-HCO 3  prepared for the nal puri cation step. The HILIC and Oasis PRiME HLB plus light cartridges were both rst conditioned with water from syringe 3. Syringe 2 was rst rinsed with ethanol; (1 mL) and then acetone (1.2 mL), which was used to condition the HILIC. After conjugation, the reaction mixture was diluted with acetone (2.4 mL) and transferred via syringe 2 to the HILIC cartridge, immobilizing the [ 18 F]DCFPyL product while more apolar impurities were unretained. Residual radioactivity was transferred from the reactor to the HILIC cartridge using a second portion of acetone (1.0 mL). The remaining acetone in vial 7 (approx. 1 mL) was used to rinse the HILIC cartridge followed by a N 2 -gas purge. Syringe 2 was rinsed with 0.85% phosphoric acid, and a 4.2 mL volume of this same acid solution was used to elute the [ 18 F]DCFPyL fraction off the HILIC column to syringe 3, which was further diluted with 2.7 mL of 0.85% phosphoric acid.
Considering that both [ 18 F]DCFPyL and unreacted precursor Glu-CO-Lys were retained by the HILIC SPE and both was eluted with weak phosphoric acid, an additional puri cation step to remove unreacted Glu-CO-Lys, which is a pseudo-carrier possessing PSMA receptor avidity, was required. After investigating different reverse solid-phase materials, including C18 and other polymeric phases, we identi ed the Oasis PRiME HLB plus light cartridge as being suitable, by retaining [ 18 F]DCFPyL but not the more polar Glu-CO-Lys. Consequently, [ 18 F]DCFPyL could be trapped selectively on the Oasis PRiME HLB light cartridge using syringe 2 as a waste receiver. The cartridge was washed with an additional 4.8 mL of phosphoric acid to elute any residual Glu-CO-Lys. Finally, syringe 3 was washed three times with water and [ 18 F]DCFPyL was eluted off the Oasis PRiME cartridge into the product vial with 2.1 mL 25% ethanolic solution.
Quality control of the drug product HPLC method 1 (reversed phase) was used in conjunction with on-line radioactivity detection for routine assessment of radiochemical and chemical purity. Chemical and radiochemical purity of the product was also monitored using method 2a. The HILIC HPLC method 2b was used for quantifying in the puri ed product any residual Glu-CO-Lys, which was too hydrophilic to be resolved by reversed phased HPLC. Representative chromatograms from individual steps of the automated synthesis are depicted in gure 3. The results for four automated runs, using starting activities ranging from 1 to 95 GBq, are shown in table 4. The radiochemical yield was fairly close to 40% irrespective of the starting activity. Furthermore, the radiochemical purity was never lower than 93% (range 93-98%) at end of synthesis, and declined in the rst hours after high radioactivity productions, in agreement with already published work showing radiolysis of multi-GBq productions of [ 18 F]DCFPyL (Ravert et al., 2016). In the nal radiopharmaceutical product, carrier DCFPyL is the main constituent of the PSMA-avid impurities; its concentration increases with higher starting radioactivities. The total chemical impurities were well below 200 µg per batch, such that formulation to a nal 20 mL volume drug product gave a mass concentration less than 10 µg/mL. This is comfortably below the purity speci cations adopted for PSMA-1007 ( , 2019). The overall molar activity was moderately high, increasing 7-fold as starting radioactivity increased from 1 to 95 GBq. Nevertheless, the entire range of obtained molar activity is suitable for PET imaging of highly expressed receptor targets such as PSMA (Pillarsetty et al., 2016). Residual solvents in the nal drug product (acetone, TEA, acetonitrile, DMSO and t-BuOH) where quanti ed using GC-FID and were found to be well below the Class II (acetonitrile) and Class III residual solvent limits speci ed in the European Pharmacopeia. with radiochemical yields su cient to support multiple clinical PET investigations from a single production. Residual solvents were well below the concentration limits adopted in the current version of the European Pharmacopeia. Furthermore, the fully automated proof-of-concept two-step process bodes well for a more general application to other radiosynthesis modules and for conjugation with large biomolecules such as oligopeptides and proteins.  Water bag 9. Abs. ethanol 10. HILIC SPE cartridge 11. Oasis PRiME HLB plus light cartridge 12. 25 % ethanolic solution external vial 13. Product collection vial. S1, S2 and S3, syringes 1, 2 and 3.