Investigation of Fluorine-18 Labelled Peptides for Binding to Cholecystokinin-2 Receptors with High Affinity

Many cancers of neuroendocrine origin overexpress cholecystokinin-2 receptors (CCK-2R) including medullary thyroid cancer, small cell lung cancer and other lung carcinoids. Fluorine-18 labelled peptides targeting CCK-2R enable direct visualization and quantification of this receptor in vivo using positron emission tomography imaging. CP04 1 and MG11 2 are two previously described truncated peptides derived from the native CCK-2R hormone ligand, gastrin. The N-terminus of the MG11 2 octopeptide was chemically modified with various fluorine containing aromatic (4-fluorobenzoate), heterocyclic (6-fluoronicotinate) and aliphatic (2-fluoropropionate) moieties. To assess the impact these modifications had on CCK-2R binding, ligand-binding assays were conducted using A431 cells overexpressing human CCK-2R. MG11 2 modified by 4-fluorobenzoate (FB-MG11 3) demonstrated the highest binding affinity (0.20 nM) followed by MG11 2 modified by 6-fluoronicotinate (FNic-MG11 4; 0.74 nM) and 2-fluoropropionate (FP-MG11 5; 1.80 nM), respectively. Whilst indirect labelling of MG11 2 using fluorine-18 labelled activated esters of fluorobenzoate and 6-fluoronicotinate was unsuccessful, direct fluorine-18 labelling at the N-terminus modified with 6-nitronicotinate afforded a 47.6% radiochemical yield of [18F]FNic-MG11. Unfortunately, [18F]FNic-MG11 4 was chemically unstable, decomposing slowly through defluorination, thereby impeding any further work with this radiotracer.


Introduction
Theranostic radiopharmaceuticals are key to the realization of personalized medicine in oncology, leading to high therapeutic efficacy with low toxicity in patients with cancer (Hicks et al. 2017;Hofman et al. 2018;Turner 2018). Theranostic radiopharmaceuticals use the same targeting agent, often a peptide, for both diagnosis and therapy, contingent upon the nature of the incorporated radionuclide. Diagnostic radiopharmaceuticals labelled with positron emitting radionuclides such as fluorine-18 (F-18) can be used for positron emission tomography (PET) imaging that enables quantification of receptor density and drug interactions in vivo with high sensitivity (pico-to nano-molar range) (Vallabhajosula et al. 2011). The diagnostic scan stratifies patients according to their likelihood of benefiting from targeted radionuclide therapy. Only patients demonstrating high tumour uptake and suitable biodistribution of the diagnostic radiopharmaceutical are advanced to receive a therapeutic dose that uses the same targeting agent radiolabeled with a cytotoxic radionuclide, such as the beta-emitting lutetium-177 (Lu-177) (Rodrigues et al. 2021).
Cholecystokinin receptor subtype-2 (CCK-2R) is an important G protein-coupled receptor (GPCR) overexpressed in many types of neuroendocrine cancers including medullary thyroid carcinoma (92% incidence), small cell lung cancer (57% incidence), astrocytoma (65% incidence) and stromal ovarian cancer (100% incidence). Accordingly, 6 Page 2 of 11 significant efforts have been made to develop theranostic radiopharmaceuticals to image and treat cancers that overexpress CCK-2R (Rangger et al. 2017;Reubi et al. 1997). Under normal physiological conditions, CCK-2R preferentially binds the peptide hormone gastrin in the stomach, but can also bind with high affinity to the related hormone, cholecystokinin. Both hormones share the same C-terminally amidated pharmacophore (Trp-Met-Asp-Phe-NH 2 ) that binds to CCK-2R, but they differ in the N-terminal sequence, with the pentaglutamic acid sequence of gastrin absent in cholecystokinin. A variety of peptides derived from these two CCK-2R ligands have been labelled with radionuclides and evaluated as CCK-2R targeting radiopharmaceuticals (Tornesello et al. 2011). Radiolabeled peptides derived from gastrin demonstrated improved tumour targeting in comparison to those derived from cholecystokinin (Laverman et al. 2011).
Radiolabelled derivatives of a CCK-2R targeting peptide derived from gastrin, CP04 1 ( Fig. 1) have undergone successful clinical trials for imaging neuroendocrine tumours. These trials demonstrated that these radiolabeled peptides possessed high receptor affinity (IC 50 in the sub-nM to low nM range), good cellular internalization and relative in vivo stability leading to high tumour to background ratios (Erba et al. 2018). We have recently developed 'foldamer' peptides based on CP04 1 that show significant improvements in tumour targeting and metabolic stability in comparison to CP04 1 (Corlett et al. 2021). Unfortunately, both CP04 1 and its analogues contain the highly ionic N-terminal hexa-glutamic acid sequence that results in undesired renal accumulation that when radiolabeled with a therapeutic radionuclide could lead to significant nephrotoxicity. Therefore, truncated analogues of CP04 1 have also been investigated, including the peptide MG11 2 that lacks all N-terminal glutamic acid residues. This shortened peptide displayed significantly reduced renal uptake and retention, spurring additional refinements of the core MG11 2 structure to further reduce kidney uptake and enhance tumor targeting (Erba et al. 2018;Grob et al. 2020;Hörmann et al. 2020;Klingler et al. 2019;Maximilian et al. 2020;Uprimny et al. 2020).
MG11 2 has been radiolabeled with different gammaemitting radionuclides (Tc-99m, Lu-177 and In-111) for single-photon emission computed tomography (SPECT) imaging Ocak et al. 2011;von Guggenberg et al. 2009), as well as with the short-lived positronemitting radionuclide Ga-68 for PET imaging (Summer et al. 2018). Whilst the superior resolution of PET imaging makes this the modality of choice compared to SPECT, logistical issues with Ga-68 (half-life: 68 min) mean that a longer lived more readily available PET radionuclide is a more feasible option. Fluorine-18 (half-life: 110 min) is the workhorse radionuclide for PET imaging as it can be generated in large amounts using low energy cyclotrons and importantly, it exhibits ideal physical properties. To date, an F-18 labelled MG11 2 analogue has yet got to be evaluated for PET imaging. Herein, we aim to develop fluorine-18 labelled MG11-based radiopharmaceuticals for evaluation as PET imaging agents.
Radioactivity measurements were carried out with a CRC-15PET dose calibrator (Capintec) that was calibrated daily using Cs-137 and Co-57 sources (Isotope Products Laboratories). Radiation was detected using a solid state photodiode scintillator crystal detector (Knauer). Preparative high performance liquid radiochemical chromatography (HPLC) was performed using a Knauer 1050 pump, 2500 UV detector, and 5050 manager. Radiation was detected using a Knauer solid state photodiode scintillation crystal detector in a TO-5 case. Analytical HPLC was performed using a Shimadzu HPLC system consisting of a SCL-10AVP system controller, SIL-0ADVP auto-injector, LC-10 ATVP solvent delivery unit, CV-10AL control valve, DGU-14A degasser, and SPD-10AVPV detector. This was coupled to a radiation detector consisting of an Ortec model 276 photomultiplier base with a 925-SCINTACE-mate preamplifier, amplifier, bias supply, and SCA and a Bicron 1M11/2 photomultiplier tube. Radiochemical LC-MS analysis was performed on a Shimdzu LCMS-8030 model.

Peptide Synthesis
All peptides were synthesized using Rink amide resin (ChemPep-Florida, loading capacity 0.8 mmol/g, bead size 100-200 mesh) as solid support. Peptide synthesis was performed on a 0.1 mmol scale with a five fold molar excess of amino acids (0.2 M) on a CEM Liberty Blue module. DIC (5 eq) and Oxyma pure (5 eq) were used for amino acid coupling. Coupling to the Nα-deprotected peptide-resin was performed under microwave irradiation for 3 min (30 W, 90 °C). Fmoc deprotection was performed using a 0.1 M Oxyma pure solution containing 20% (v/v) piperidine in N,N-dimethylformamide (DMF) under microwave irradiation for 30 s (40 W, 40 °C). The N-terminus of MG11 2 was capped with 2-fluoropropionic acid, 4-fluorobenzoic acid, 6-fluoronicotinic acid or 6-nitronicotinic acid to afford the respective non-radioactive fluorine containing peptides. Fully assembled peptides on resin were transferred into a syringe fitted with frit and washed three times with DMF and DCM respectively. A cleavage cocktail was prepared (8 mL/peptide) consisting of 5% water, 5% thioanisole, 2.5% DODT, and 87.5% TFA in a falcon tube and drained into the syringe vessel containing the peptide on dried resin beads. This mixture was mixed for 2-3 h on an orbital shaker. The solution was then drained into a falcon tube and TFA was removed by purging with nitrogen for ~ 20 min or until 1/3 of the total volume remained. The peptide was triturated using ~ 30 mL of diethyl ether before being centrifuged at 3500 rpm for ~ 5 min. This process was repeated three times. The precipitate was purged with nitrogen to remove the remaining diethyl ether, leaving a dried pellet of peptide. The crude peptide was then suspended in 50% MeCN:H 2 O solution and immediately analyzed by analytical HPLC and mass spectrometry to assess synthesis efficacy and confirm product formation. The sample was then lyophilized and subsequently purified on a preparative HPLC system. The pure fractions were combined and lyophilized.

Peptide Preparative HPLC Conditions
The peptides were purified using the preparative column described above, eluted at 8 mL/min with a gradient: 0.1% (v/v) TFA, starting at 0% MeCN for 1 min, increased to 85% over 30 min and then increased and maintained at 100% MeCN for another 10 min. Peptides were detected at 220 nm wavelength.

Peptide Analytical HPLC Conditions
The peptides were analyzed using the analytical column described above, eluted at 1 mL/min with a gradient: 0.1% (v/v) TFA, starting at 0% MeCN, increased to 100% over 30 min and then increased and maintained at 100% MeCN for another 10 min. Peptides were detected at 220 nm wavelength.

In Vitro CCK-2 Receptor Ligand Binding Experiments
The human epidermoid carcinoma cell line A431, transfected with a plasmid encoding full length CCK-2R, was used for all competitive ligand binding assays, as previously described (Corlett et al. 2021). To prove specificity, negative control A431 cells transfected with an empty vector were also analysed. Briefly, 2 days before experimentation 650,000 cells were plated into each well of a 6-well plate and allowed to grow to confluence. On the day of experimentation, each novel peptide 3-5 was serially diluted in binding buffer (DMEM + 1% FCS) and mixed with [ 177 Lu]Lu-DOTA-CP04 12 (~ 50,000 cpm). Each dilution was added to cells pre-washed with binding buffer and allowed to incubate at 37 °C for 1 h. Following incubation, peptide mixtures were removed and kept for analysis (unbound fraction). Cells were then washed twice in binding buffer, which was pooled with the unbound fraction. Cell lysates representing the bound fraction were created by incubating cells for 5 min with 1 M NaOH. All fractions were analysed using a gamma counter (Perkin Elmer 2480 Wizard2™; Perkin Elmer, Massachusetts, USA) and normalised according to protein concentration, which was determined using a Pierce BCA protein assay kit (ThermoFisher, Australia). Ligand binding curves were generated using GraphPad Prism software (GraphPad Software, California, USA) post non-linear regression analysis.
Half maximal inhibitory concentration (IC 50 ) values were approximated and expressed in nM of fluorinated peptides.
All peptides synthesized gave high yields and excellent chemical purity ( Table 1). Samples of pure FB-MG11 3, FNic-MG11 4 and FP-MG11 5 were then used to prepare solutions for the subsequent competition ligand-binding assays.

Receptor Binding Study
To determine the relative binding affinity of FB-MG11 3, FNic-MG11 4 and FP-MG11 5 for the CCK-2 receptor, radioligand − receptor binding assays were performed using A431 cells stably expressing the CCK-2 receptor, with [ 177 Lu]Lu-DOTA-CP04 12 employed as the competitive radioligand (Fig. 3). A total of seven concentrations Fig. 2 Synthesis of MG11-based peptides capped with different fluorine-and nitro-containing moieties of peptide were assayed, ranging from 0 to 40 nM. Unfortunately, due to solubility issues we were unable to assay at higher concentrations. This complicated the analysis of the CCK-2 receptor binding properties of these peptides with the IC 50 values calculated to be used as a guide only for comparison. Nevertheless, the apparent binding affinity of all peptides tested were in the low nanomolar to subnanomolar range (Fig. 3). FB-MG11 3 demonstrated the highest binding affinity followed by FNic-MG11 4 whilst FP-MG11 5 clearly exhibited the lowest affinity. Accordingly, we chose to prepare fluorine-18 labelled versions of FB-MG11 3 and FNic-MG11 4.
Given the difficulties in radiolabeling the N-terminus of the unprotected MG11 peptide in solution, the attachment of 6-nitronicotinate on solid phase was investigated. Within aromatic systems the nitro group is a good leaving group that often effectively facilitates radiofluorination substitutions (Preshlock et al. 2016). We hypothesized that 6-nitronicotinoyl-MG11 11 (Nitro-MG11) could be an effective precursor to allow the synthesis of [ 18 F]FNic-MG11 4 via direct labelling. Nitro-MG11 11 was successfully prepared using Fmoc-SPPS, with the N-terminus of the protected, resin bound MG11 peptide capped with Table 1 Analytical data of peptides 2-5 and 11 a ESI-MS base peak corresponds to [M+H] + b HPLC conditions: Hypersil BDS C18, 150 mm × 4.6 mm, 5 µm 130A, eluted at 1 mL/min with a gradient: 0.1% (v/v) TFA, starting at 0% MeCN, increased to 100% over 30 min and then increased and maintained at 100% MeCN for another 10 min. Peptide purity was determined at 220 nm wavelength c Reported yield of purified and lyophilized material  Fig. 3 Calculated IC 50 values from binding assays performed on human epidermoid carcinoma cells (A431), engineered to overexpress the full-length CCK-2 receptor. Competition binding curves were generated by incubating the peptides 3-5 with the known CCK-2 receptor binding agent, [ 177 Lu]Lu-DOTA-CP04 12. IC 50 experiments were performed in three technical replicates 6-nitronicotinic acid. The peptide was then cleaved off the resin and fully deprotected and purified to generate Nitro-MG11 11 with good yields and purity. Direct radiofluorination of Nitro-MG11 11 via nucleophilic aromatic substitution of the nitro group with fluoride-18 to generate [ 18 F]FNic-MG11 4 was then investigated (Fig. 6).
Dried nucleophilic 'naked' [ 18 F]fluoride was prepared using a mixture of Kryptofix 222 (K 222 ) and K 2 CO 3 . The reactive K 222 /K 2 CO 3 / 18 Fcomplex was treated with Nitro-MG11 11 in DMSO at 110 °C. To monitor the progress of the reaction, a sample of the reaction mixture was taken every 5 min for up to 20 min and analysed using HPLC. Pleasingly, within 5 min this reaction generated [ 18 F]FNic-MG11 4 with a 47.6% radiochemical yield (based on HPLC integration) (Fig. 7). Intriguingly, the yield decreased to 33.4%, 28.8% and 27.3% at 10, 15 and 20 min, respectively. Furthermore, the fraction of [ 18 F] FNic-MG11 4 present in the reaction mixture heated for only 5 min decreased to 20.7% after three hours at room temperature, indicating product instability under basic conditions. The remaining radioactivity present was free [ 18 F]fluoride ions, indicating that [ 18 F]FNic-MG11 4 decomposes through a substitution process that releases fluoride-18.

Discussion
Over the last two decades, many CCK-2R targeting radiotracers with high affinity have been developed for SPECT and PET imaging. However, fluorine-18 labelled CCK-2R targeting radioligands have remained out of reach. Although the N-termini of MG11 2 and other CCK-2R targeting peptides based on the gastrin hormone do not interact directly with the receptor, they have been shown to induce interesting folding characteristics of the peptides that may augment binding affinity (Morley et al. 1968). Accordingly, we chose to investigate the CCK-2R binding affinity of MG11-based peptides modified with a range of fluorine-containing aliphatic, aromatic and heteroaromatic moieties that act as scaffolds to introduce fluorine-18 at the N-terminus.
Ligand binding assays with our novel MG11 peptides demonstrated that N-terminal modification with a 4-fluorobenzoyl moiety led to the highest increase in CCK-2R binding affinity, compared with 6-fluoronicotinate and 2-fluoropropanoate (Fig. 3). These results indicate that the introduction of lipophilic/aromatic groups at the N-terminus of MG11 2 improves binding affinity. It is proposed that the Ala-Tyr-Gly-Trp region of the gastrin hormone is involved in a β-turn, inducing a "U" shape conformation (Morley et al. 1968). We have also recently demonstrated that 13-residue peptide analogues of the gastrin hormone (based on CP04) adopt a β-hairpin conformation (Corlett et al. 2021). As such, we speculate that the aromatic fluorobenzoyl or fluoronicotinoyl moieties may facilitate intramolecular π-π interactions with the tyrosine and/or tryptophan residues stabilizing the biological active conformation. Peptides are commonly labelled with fluorine-18 via indirect methods using pre-labelled fluorine-18 synthons. Thus, we also investigated the use of known Surprisingly, acylation of MG11 2 did not proceed, suggesting the N-terminus of this peptide is inaccessible (Fig. 4). This lack of reactivity is consistent with the proposed folded conformation of MG11, which may result in occlusion of the N-terminal amine.
While indirect labeling is the most common method routinely employed for the preparation of fluorine-18 labelled peptides, recent advancements have demonstrated that direct labeling of peptides can also be performed with clear advantages using highly reactive precursors. Nicotinate scaffolds with leaving groups at the 2-position undergo facile radiofluorination through fluoride-18 ion attack at the ipso-carbon atom. Thus, peptides incorporating N,N,N-trimethylpyridin-2-aminium groups encompass an important class of precursors for direct radiofluorination of peptides (Cardinale et al. 2017;Dornan et al. 2018;Naka et al. 2020). Using this knowledge, we investigated the preparation of an MG11 analogue capped at the N-terminus with an N,N,Ntrimethylpyridin-2-aminium group to facilitate direct radiofluorination. Unfortunately, treatment of MG11 2 with activated ester TFP-NMe 4 6 did not generate the N,N,Ntrimethylpyridin-2-aminium tethered peptide 14 (Fig. 5), further confirming the inaccessibility of the α-amine of the N-terminal D-Glu residue.
The challenges of modifying the N-terminus of the MG11 peptide in solution led us to incorporate a reactive nicotinate moiety directly onto the fully protected peptide on solid phase. The nitro group is a well-established leaving group on aromatic systems that can facilitate fluorine-18 substitution (Preshlock et al. 2016). As such, we prepared 6-nitronicotinate capped MG11 (Nitro-MG11 11) by standard SPPS techniques. Treatment of Nitro-MG11 11 with fluoride-18 successfully generated the radiolabelled [ 18 F]FNic-MG11 4 with a 47.6% radiochemical yield within 5 min at 110 °C. However, [ 18 F] FNic-MG11 4 was not stable under basic conditions and decomposed with increased reaction time, and even at room temperature after 3 h, releasing free [ 18 F]fluoride. Further, the precursor Nitro-MG11 11 demonstrated similar instability, decomposing to a compound with m/z 1120.4 Da, indicating loss of HNO 2 from 11.
We hypothesise that the decomposition of Nitro-MG11 11 and [ 18 F]FNic-MG11 4 progresses through an intramolecular nucleophilic aromatic substitution with an internal nucleophile, resulting in displacement of NO 2 from 11 or [ 18 F] Ffrom 4. The most likely nucleophile is the phenol of the tyrosine residue (Fig. 8), though other groups including the tryptophan residue cannot be ruled out. The carboxylate of the neighboring glutamate residue is unlikely to be a suitable nucleophile as this would lead to a highly strained paracyclophane ring. This cyclisation process accounts for both the observance of the product with m/z 1120.4 Da and the decomposition of [ 18 F]FNic-MG11 4 with loss of [ 18 F]fluoride.

Conclusion
Three MG11 peptide analogues containing fluorinated acyl groups were prepared and shown to bind with nM affinity to the CCK-2R receptor. Fluorine-18 labelling of these peptides