Development of [18F]AmBF3 Tetrazine for Radiolabeling of Peptides: Preclinical Evaluation and PET Imaging of [18F]AmBF3-PEG7-Tyr3-Octreotide in an AR42J Pancreatic Carcinoma Model

Radiolabeled peptides have emerged as highly specific agents for targeting receptors expressed in tumors for therapeutic and diagnostic purposes. Peptides developed for positron emission tomography (PET) are typically radiolabeled using prosthetic groups or bifunctional chelators for fast “kit-like” incorporation of the radionuclide into the structure. A novel [18F]alkylammoniomethyltrifluoroborate ([18F]AmBF3) tetrazine (Tz), [18F]AmBF3-Tz, was developed for the [18F]fluorination of trans-cyclooctene (TCO)-modified biomolecules using Tyr3-octreotides (TOCs) as model peptides. [18F]AmBF3-Tz (Am = 15.4 ± 9.2 GBq/μmol, n = 14) was evaluated in healthy mice by ex vivo biodistribution and PET/computed tomography (CT), where the radiolabel in the prosthetic group was found stable in vivo, indicated by the low bone uptake in tibia (0.4 ± 0.1% ID/g, t = 270 min). TCO-TOCs tailored with polyethylene glycol (PEG) linkers were radiolabeled with [18F]AmBF3-Tz, forming two new tracers, [18F]AmBF3-PEG4-TOC (Am = 2.8 ± 1.8 GBq/μmol, n = 3) and [18F]AmBF3-PEG7-TOC (Am of 6.0 ± 3.4 GBq/μmol, n = 13), which were evaluated by cell uptake studies and ex vivo biodistribution in subcutaneous AR42J rat pancreatic carcinoma tumor-bearing nude mice. The tracer demonstrating superior behavior ex vivo, the [18F]AmBF3-PEG7-TOC, was further evaluated with PET/CT, where the tracer provided clear tumor visualization (SUVbaseline = 1.01 ± 0.07, vs SUVblocked = 0.76 ± 0.04) at 25 min post injection. The novel AmBF3-Tz demonstrated that it offers potential as a prosthetic group for rapid radiolabeling of biomolecules in mild conditions using bioorthogonal chemistry.


■ INTRODUCTION
Biomolecules are increasingly important in nuclear imaging due to their biocompatibility, precise targeting capability, and suitability to various diagnostic and therapeutic applications. 1,2 Chemical modification of naturally occurring peptides can serve as an avenue toward biologically more stable peptide derivatives, for example, by extending their biological half-life in vivo. 3 Additional functional groups can be included in the peptide structure, enabling chemoselective late-stage bioconjugation reactions. 4 Due to the ideal physical half-life and imaging properties of the radioisotope (t 1/2 = 109. 8 4,5 However, the direct incorporation of nucleophilic [ 18 F]fluoride into a molecule often requires leaving or protecting groups and generally harsher (e.g., alkaline) conditions, 6,7 limiting its use on structures sensitive to alkalinity or heat, such as proteins.
Mild incorporation of [ 18 F]fluoride into biomolecules chemoselectively by isotopic exchange (IE) can be applied instead of the canonical nucleophilic substitution. 8 However, some of the isotopic exchange reactions, such as the conventional silicon−fluoride (Si−F) exchange, require anhydrous conditions, adding a drying step crucial to the success of the radiolabeling. 9 When applying the Si−F isotopic exchange to an SSTR2-targeting TATE derivative, a hydrophilic silicon−fluoride acceptor (SiFA)-derivatized [ 18 F]F-SiFAlin-TATE is developed, and it has successfully entered clinical trials for neuroendocrine tumor (NET) imaging, 10−13 revealing the true potential of isotopic exchange reactions for clinical radiopharmaceutical development.
Liu et al. developed the radiolabeling of an alkylammoniomethyltrifluoroborate (AmBF 3 )-based prosthetic group, [ 18 F]-AmBF 3 -alkyne, 14 utilizing IE radiofluorination that tolerates aqueous conditions, making it well compatible with watersoluble molecules. The method provided [ 18 F]AmBF 3 -TATE in one step using IE after click chemistry conjugation of the prosthetic group to the peptide, 15,16 and the radiosynthesis of [ 18 F]AmBF 3 -TATE was successfully modified into a cassette system, yielding up to 10 patient doses in a single run by Lau et al. 17 Both SSTR targeting tracers, [ 18 F]F-SiFAlin-TATE and [ 18 F]AmBF 3 -TATE, showed favorable pharmacokinetics, high in vivo stability, and high image contrast. The AmBF 3 -chemistry has since been utilized for direct IE radiolabeling of various other peptides. 18−22 As an alternative modular strategy, Iddon et al. reported the development of 2-[ 18 F]fluoroethyl azide fluorination reagents suitable for radiolabeling 18 F-octreotides with reaction times as short as only 5 minutes at room temperature, using copper as a catalyst, 23 a method specifically useful for sensitive biomolecules. However, compared to other click-based methodologies, the exquisite reaction rate, absence of catalyst, and the biocompatibility of the bioorthogonal inverse electron-demand Diels−Alder (IEDDA) reaction have made it the focal point of click chemistry-based development in biomolecule radiolabeling, especially for pretargeted PET imaging. 24 Here, leveraging the aqueous compatibility of the AmBF 3 IE reaction in combination with the unsurpassed kinetics and selectivity of the IEDDA reaction, we report the development of a novel prosthetic group [ 18 F]AmBF 3 tetrazine ([ 18 F]-AmBF 3 -Tz) suitable for the chemoselective radiolabeling of trans-cyclooctene (TCO)-modified biomolecules. As a model system, we radiolabeled two Tyr 3 -octreotides (TOCs), analogues of somatostatin, 25 in a proof-of-concept study ■ RESULTS AND DISCUSSION Synthesis of AmBF 3 Tetrazine Precursor (6) for Radiolabeling. The synthesis of the AmBF 3 tetrazine was designed in a stepwise manner to incorporate the boronic acid pinacol ester selectively into the tertiary amine, followed by acid-catalyzed fluorination of the pinacol ester to afford the trifluoroborate. During the synthesis, it was crucial to take into account the susceptibility of the tetrazine, a redox mediator, 26 to readily reduce into "unreactive" dihydrotetrazine in the presence of a reducing agent, especially when heated. The synthesis of the trifluoroborate required anhydrous conditions for the nucleophilic substitution of the haloalkane in the pinacol ester, and the subsequent fluorination step required a corrosive-resistant reaction vessel, careful handling, and good ventilation due to the formation of corrosive and toxic HF (g), even if in small quantities. AmBF 3 -Tz (6) was synthesized with an overall yield of ∼36% (Scheme 1). The nuclear magnetic resonance (NMR) spectroscopy analysis revealed in the 1 H NMR a characteristic signal at the para-position of the Tz ring at 10 ppm (Supporting Figure S1), and the presence of the Tz ring was verified by high-performance liquid chromatography coupled to a diode-array detector (HPLC−DAD) at 534 nm, by the characteristic absorbance wavelength for Tz (>500 nm) (Supporting Figure S2). 19 F NMR spectra of 6 displayed splitting of the signal due to coupling to the trifluoroborate boron, and the 11 B NMR spectra likewise revealed the boron-11 coupling to fluorine-19, detected as a split quartet signal (Supporting Figures S3−S5 for 11 B, 19 F and 13 C NMR).
Compound 6 eluted at t R = 4.59 min, when analyzed by ultrahigh-performance liquid chromatography high-resolution mass spectrometry (UHPLC-HRMS), with a detected molecular ion peak corresponding to the protonated [M + H] + ion (Supporting Figure S6).
[ 18 F]Fluorination of 6. Prosthetic group 6 was radiolabeled with a protocol partly based on a methodology developed by Liu et al. 14 The radiosynthesis of [ 18 F]6 is presented in Figure 2. Modifications to the [ 18 F]fluoride eluent and radiolabeling buffer were done to alter the conditions more suitable for our prosthetic group and setup, ensuring repeatable radiolabeling yields (20.8 ± 10.3%, n = 7) in microliter volumes. The optimal reaction volume in our conditions was a mere 10−20 μL. Decreasing the volume by 2.5 times increased the yield by 6 times at 85°C (0.9% NaCl elution, 200 nmol of 6, Supporting Figure S15), and the radiochemical yield (RCY) decreased dramatically if the reaction mixture was evaporated to dryness or when the final volume exceeded 20 μL. However, for elution of reasonable amounts of [ 18 F]fluoride out of the PS-HCO 3 (Macherey-Nagel, Duren, Germany) solid-phase extraction (SPE) ion exchange cartridge, a minimum 20−30 μL of 0.9% NaCl was required. Therefore, we chose to substitute the commonly used aqueous 0.9% NaCl as the [ 18 F]fluoride eluent altogether and opted for a pyridazine HCl eluent formulation, similarly as reported by Kwon et al. 27 The pyridazine HCl buffer recipe was modified to best serve our setup, as a combination of pyridazine (9 v  decreasing the evaporation time from 45 min (100 μL of 0.9% NaCl as the eluent) to 10 min (100 μL of the modified pyridazine HCl buffer, pH 2.0) in our setup . The evaporation time was further cut in half by adding more DMF to the buffer (water quantity from ∼38 to ∼12% v/v), which made the control of the final volume easier, and improved the RCY ([ 18 F]6; 8−37% DCY), which reached the range of previously published [ 18 F]AmBF 3 tracers (∼16−35%). 14 15 (Scheme 2), resulting in a loss of radioactivity in each step, circumvented in the one-step radiofluorination of [ 18 F]AmBF 3 -TATE. Furthermore, the molar ratio of [ 18 F]6 to the TCOpeptide 12 or 13 was kept at least at 2:1, resulting in anticipated loss of radioactivity during the IEDDA.
In Vitro Stability and Lipophilicity. demonstrated an overall higher cell uptake in vitro, which was effectively blocked by an excess of native TOC (baseline = 6.1 ± 0.6% at 120 min, n = 3, vs blocking = 0.7 ± 0.1% at 120 min, n = 3, p < 0.005, Supporting Figure S22), corroborating that the uptake was specific and receptor-mediated.
PET/CT and Ex Vivo Biodistribution of [ 18 F]6. The prosthetic group [ 18 F]6 was studied as a standalone tracer for evaluating the stability of its radiolabel (B-18 F) in vivo. Moreover, [ 18 F]6 was hypothesized to have beneficial properties, if stable in vivo, as a pretargeting tool. 11 nmol, 150 μL, was administered intravenously to male SCID (11.0 ± 0.5 MBq) and female C57BL/6JRj (11.3 ± 0.3 MBq) mice (n = 4 per strain) ( Figure 2D). Five minutes post injection, [ 18 F]6 demonstrated low uptake in major organs and fast clearance from the blood, as illustrated by the time−activity curve (TAC) for the heart (left ventricle, Supporting Figure S23). An elevated liver uptake, possibly due to the tetrazine moiety, which decreased steadily throughout the 50 min dynamic image acquisition, was also visible. The elimination of radioactivity from the tissues during the PET/CT image acquisition, presented as TACs, indicated that the prosthetic group eliminates quickly, mainly through the kidneys (Supporting Figure S23). PET/CT was followed by ex vivo biodistribution 270 min post injection, which confirmed the  6 19.3 ± 11.6 (n = 3) 21.4 ± 13.5 Yields are decay-corrected to the start of synthesis.
Bioconjugate Chemistry pubs.acs.org/bc Article optimal pharmacokinetics and high in vivo stability of the radiolabel in [ 18 F]6, indicated by the fast clearance of radioactivity from the major organs and the low bone uptake in the tibia (0.4 ± 0.1% ID/g for C57BL/6JRj, 0.3 ± 0.1% ID/ g for SCID; Figure 2F). Pronounced elimination into the gallbladder (6.4 ± 2.5% ID/g for C57BL/6JRj, 3.5 ± 2.4% ID/ g for SCID) was also seen, but the major elimination pathway was renal clearance. The radiolabel stability and the beneficial pharmacokinetic characteristics of [ 18 F]6 prompted its use for peptide radiolabeling and revealed its potential as a pretargeting radiotracer, currently under investigation by our group.    17 indicative of at least comparable stability of the radiotracers in vivo. Interestingly, the radioactivity in bone increased from 60 to 120 min post injection only for [ 18 F]14 (tibia: 2.9 ± 1.4% ID/g; occipital: 1.7 ± 0.1% ID/g) but not for [ 18 F]15 (tibia: 0.6 ± 0.4% ID/g; occipital: occipital 0.6 ± 0.1% ID/g). The ex vivo radiometabolite analysis by radio-TLC indicated that [ 18 F]14 was metabolized and two radiometabolites were detected in blood at 5 and 30 min post injection (radio-TLC; Supporting Figure S26), in accordance with the in vitro enzymatic stability assay results ( Figure 3B). A sample taken at 60 min post-injection revealed the same polar metabolite in blood, while in urine a less-retained, less-polar metabolite in trace amounts was seen, leaving approximately 99% of the radiotracer intact in both urine and blood. The prolonged blood residence of the TOC derivatives persisting at 60 min warrants further evaluation. After administration of [ 18 F]14, blood samples were taken, and the radioactivity in separated blood components was analyzed. The free fraction of the tracer was 72.9 ± 5.1% at 5 min and remained high until 60 min post injection (68.5 ± 5.3%). This indicates that the tracer was readily available at a steady rate throughout the study. Radioactivities of 22 and 25%, respectively, at 5 and 60 min, were bound to red blood cells (RBCs) (Supporting Table S1).

Ex Vivo
In blocking conditions at 60 min, the free fraction seemed to decrease (55.7 ± 11.4%), and the RBC-bound fraction grew (29.7 ± 2.9%). The binding to RBCs slightly grew from 5 to 60 min post injection. This could have contributed to the long circulation time and high background radioactivity levels in organs with a large blood reservoir, such as the liver, and a slight rise in bone uptake detected for both  (Figure 4D). The tumor was well visualized, as seen in the maximum intensity projection (MIP) PET/CT image ( Figure 4C Figures S37, S38, and S39) moderate tumor-tobackground contrast. The prolonged availability of the radiopeptide in the blood pool likely contributed to the observed plateau in tumor uptake seen in baseline conditions ( Figure 4D upper panel), with no significant difference observed at 90 min post injection in the baseline and blocked conditions (baseline = 0.82 ± 0.14 SUV, n = 2, vs blocking = 0.76 ± 0.03 SUV, n = 2). As a possible contributor, close to 25% radioactivity in blood 60 min after administration of the other peptide analogue [ 18 F]14 was shown to be bound in RBCs ex vivo, contributing to the uptake in both tumor and nontarget tissues, such as the pancreas. This phenomenon, even when not studied for the more stable peptide [ 18 F]15, possibly accounted for the low efficiency seen in the PET/CT study. Furthermore, due to the highly similar biological behaviors and relatively small differences of the TOC analogues 14 and 15, the investigation of a non-PEGylated version would be warranted to assess the true benefit of adding a PEG chain to the structure.

Mouse Plasma Stability of [ 18 F]AmBF 3 -PEG 7 -TOC ([ 18 F]15) during Ex Vivo Studies.
After tracer injection, CO 2 asphyxiation, and cervical dislocation, blood was collected from a cardiac puncture into a tube containing 2 μL of 1% heparin (diluted from 5100 IU/mL) in 0.9% NaCl (aq.). The sample was centrifuged (1000g, 10 min) to separate the plasma from the blood cells. Cold acetonitrile (2 × vol of plasma) was added and centrifuged (10,000g for 5 min) to precipitate the proteins. A sample (100 μL) of the supernatant was injected into HPLC for radio-HPLC analysis. For the tracer [ 18 F]14, the supernatant was sampled also on TLC for digital autoradiography analysis.
Distribution of Radioactivity in Blood Components after Intravenous Administration of [ 18 F]14. Whole blood from mice were extracted during ex vivo studies, using cardiac puncture. The sample was applied in a microtube containing 1% heparin solution in 0.9% NaCl (aq.) (2 μL); the sample was centrifuged (1000g, 10 min), the total radioactivity in the sample was measured with a γ counter, the supernatant was separated from the pellet (RBC containing fraction), and cold ACN (500 μL) was added. The sample was centrifuged (10,000g, 5 min) to remove the free fraction from the precipitated protein-containing pellet. The pellet (proteinbound fraction) and the supernatant (free fraction) were measured with a γ counter, and a sample (100 μL) was injected into HPLC and spotted (4 μL) on a TLC plate (radio-TLC, TLC silica gel 60 F 254 , ACN/water 80:20).
Dosimetry. Dosimetry of [ 18 F]15 was calculated from the acquired PET/CT imaging data with Molecubes PET (β-CUBE) coupled with a CT (X-CUBE) (Ghent, Belgium). Regions of interest were drawn on source organs, namely, the heart, kidneys, liver, and lungs. Time−activity curves (TAC) were converted from mouse to human time −activity curves with the following equation where m organ,h and WB h are the organ and whole-body weights for human, respectively. Mass m organ,m and WB m are the organ weight and the whole-body weight for mouse, respectively. Time−activity curves were normalized to 1 MBq injection, and the physical decay correction was removed. After this, the TAC's were extrapolated into 3000 min, which corresponds in practice to infinity. Numbers of disintegrations in source organs are defined by integrating TAC from time 0 to 3000 min, and this value is input for OLINDA/EXM (version 2.1, Vanderbilt University, 2012) dosimetry software, where ICRP 89 reference adult male (73 kg) and ICRP 103 radiation weighting factors were used. Absorbed doses to each target organ are given in units mGy/MBq, and the effective dose is in units mSv/MBq. Statistical Analysis. The data were plotted and statistically analyzed with GraphPad Prism (version 9.1.1), and the results are presented as mean ± standard deviation (s.d.) with data points of n ≥ 3. The statistical analysis was done with the unpaired t-test with Welch′s correction, where p < 0.05 was regarded as statistically significant. The significances (p-value) were *p < 0.05, **p < 0.01, and ***p < 0.001. ■ ASSOCIATED CONTENT * sı Supporting Information