For imaging of insulinoma lesions, a variety of exendin-4-based ligands have been or are currently investigated in clinical trials for targeting the glucagon-like peptide 1 receptor (GLP-1R) (Jansen et al., 2019; https://clinicaltrials.gov/ct2/results?cond=Insulinoma&term=glp-1R&cntry=&state=&city=&dist=&Search=Search, 2020). Labeled with 18F, 125I, 68Ga, 99mTc, 89Zr or 111In, these compounds show high potential for visualization of primary tumors and metastases. Whereas radiohalogenated compounds exhibit low kidney uptake and/or fast renal excretion (Lappchen et al., 2017; Kiesewetter et al., 2012a), elevated accumulation and retention in the kidneys (> 140% ID/g, 1–4 h p.i.) is always observable when residualizing radiometals are used for exendin-4 derivatization (Jansen et al., 2019; Kiesewetter et al., 2012b; Xu et al., 2015). This non-target tissue uptake not only impairs the detection of insulinomas but could also affect the localization of other GLP-1R positive tumors like pheochromocytomas and gastrinomas, due to the close local proximity of the kidneys to the respective target organs (pancreas (insulinoma), adrenal medulla (pheochromocytoma), duodenum, pancreas and periduodenal lymph nodes (gastrinoma)) (Luo et al., 2015; Sbardella & Grossman, 2020; Donow et al., 1991). Thereby, the instrinsically lower resolution of single-photon emission computed tomography (SPECT) revealed an inferior rate of detection compared to positron emission tomography (PET) (Velikyan & Eriksson, 2020). Moreover, the radiation burden of the kidneys might rapidly become unacceptably high with long-lived radioisotopes, especially in the context of peptide receptor radionuclide therapy (PRRT) approaches (Jansen et al., 2019).
A conventional method used for lowering the kidney uptake of radiotracers makes use of pre-administration of amino acids (Rolleman et al., 2003). Gotthardt et al. applied this method for [111In]In-DTPA-exendin-4 in rats and revealed that both gelofusine (18.7% decrease) and poly-L-glutamic acid (29.4% decrease) as well as the combination of both (47.9% decrease) had a significant impact on kidney uptake, whereas the administration of L-lysine did not show any effect (Gotthardt et al., 2007). Besides, coinfusion of albumin-derived peptide fragments reduced the kidney uptake of [111In]In-DTPA-exendin-3 by 26% in rats (Vegt et al., 2010).
In order to circumvent any co−/pre-administration steps, modifications of the exendin-4 scaffold itself were pursued to improve the tumor-to-kidney ratio. As a possible strategy, radiotracers with kidney-cleavable linkers have been introduced (Yim et al., 2013; Jodal et al., 2015), and resulted in a 50% reduction of radiation dose in mice, although a general proof-of-concept in humans is still pending (Zhang et al., 2019). For diagnostic purposes, studies with 18F-labeled exendin-4 revealed high tracer uptake in INS-1 tumor cells and xenograft models and a rapid clearance from the kidneys compared to radiometallated/[18F]AlF analogs (Kiesewetter et al., 2012a; Mikkola et al., 2016; Gao et al., 2011; Wu et al., 2013). Although these results suggested high potential of 18F-labeled exendin-4 derivatives, none of the investigated compounds were transferred into clinical application (Jansen et al., 2019; https://clinicaltrials.gov/ct2/results?cond=Insulinoma&term=glp-1R&cntry=&state=&city=&dist=&Search=Search, 2020). Occasionally, difficulties in separating the precursor from the respective radiolabeled agent resulted in low molar activities (Am) (Wu et al., 2013; Yue et al., 2014). As a consequence, optimal tumor uptake could not be attained, since a low peptide dose (~ 1 μg) and hence, a high molar activity (~ 200 GBq/μmol (Boerman & Gotthardt, 2012)) is mandatory. Moreover, dependent on the labeling methodology, a relatively high nonspecific accumulation in liver and intestines occurred in preclinical studies (Kiesewetter et al., 2012a; Gao et al., 2011). Direct 18F-labeling via a SiFA-modified precursor provided only low 18F-incorporation (6%), low radiochemical yields (RCY, 1.0–1.5%) and a slow blood clearance (3.3 ± 0.8%ID/g, 2 h p.i.) (Dialer et al., 2018). Moreover, PRRT of GLP-1R-overexpressing tumors has not been envisaged yet. Potential nephrotoxicity disallows the use of radiometallated α- or β−-emitting agents and for 131I-labeled derivatives, an ingenious treatment protocol with Irenat® would be indispensable to avoid severe 131I-induced damage of the thyroid (Lappchen et al., 2017). In consequence, the use of the exendin-3/4 scaffold currently allows for a very limited number of peptide-radionuclide conjugates that can be only applied for insulinoma imaging in patients.
Since a low amount of GLP-1R expression was detected in the kidneys and high activity accumulation by this organ could not be specifically blocked by an excess of unlabeled analog, a non-saturable, GLP-1R independent mechanism was presumed for tubular reabsorption (Gotthardt et al., 2006; Brom et al., 2010; Brom et al., 2012). Accordingly, further studies revealed that the megalin transporter system of the renal proximal tubules is crucial for uptake and retention of 111In-labeled exendin-4 and potential metabolites (Gotthardt et al., 2007; Jodal et al., 2015; Vegt et al., 2011). Kidney extracts of mice that received [18F]AlF-NOTA-exendin-4 showed a single very polar radioactive metabolite at 1 h p.i. and no detectable parent peptide. The identity of this metabolite could not be determined, but the general observation that exendin-4 peptides with C-terminal radiometal chelates exhibit very high uptake suggests that some metabolite from the C-terminal end might be responsible for the slow egress from the kidneys (Kiesewetter et al., 2012b). By comparing the renal accumulation of the 111In-labeled G protein-coupled receptor (GPCR) ligands octreotide, minigastrin, bombesin and exendin-4, Gotthardt et al. observed that the number of charged amino acids in these peptides correlates with their kidney uptake. Therefore, beside radiometal chelate-induced renal retention, the high number of charged amino acids of exendin-4-based radioligands is supposed to play a critical role for tubular reabsorbtion, although the exact mechanism still remains unknown (Gotthardt et al., 2007).
Based on the aforementioned study, we hypothesized that downsized GLP-1R-binding structures might have the advantage to exhibit a reduced kidney uptake and might provide accelerated urinary excretion, due to their inherent reduced number of charged amino acids. However, at present, GLP-1R-targeting radioligands, that structurally clearly differ from exendin-4 are not available. Although low-molecular-weight organic molecules were developed for addressing the GLP-1R (orally available antidiabetics) (Donnelly, 2012; Graaf et al., 2016), none of them was able to show favorable properties comparable or even superior to exendin-4/GLP-1-based polypeptide ligands (Graaf et al., 2016; Willard et al., 2012; Knudsen et al., 2007). As a result, these structures were not considered for further use as GLP-1R-targeted radiopharmaceuticals.
By contrast, short-chained undecapeptides developed in 2009 by Mapelli et al. were able to induce 3′, 5′-cyclic adenosine monophosphate (cAMP) production in chinese hamster ovary (CHO) cells, stably overexpressing the human GLP-1 receptor with similar EC50 values (87 pM) like GLP-1 (34 pM; in this study GLP-1 refers to GLP-1(7-36)amide unless otherwise stated) (Mapelli et al., 2009). Indeed, these high potencies were not plausible at first sight, since downsizing of GLP-1R ligands was assumed to lead to non-binding and hence, non-signaling agonists (Parker et al., 1998; Hjorth et al., 1994). However, the bulky hydrophobic residues at the C-terminus (position 10 and 11) of these GLP-1 mimetics, obviously compensate for the 21-residue GLP-1(16-36)amide fragment to a certain extent. 2-Aminoisobutyric acid (Aib) instead of L-alanine was introduced at the second position to confer dipeptidyl peptidase IV (DPP IV) resistance. For stabilization of an α-helical structure presumed to be crucial, also for binding of the endogenous ligand GLP-1 (Underwood et al., 2010), L-phenylalanine at position 6 was substituted with L-α-methyl-(2-fluoro)-phenylalanine. Additional screening experiments for alternative residues at positions 10 and 11, revealed lead peptide 1 which was even more potent (EC50 = 31 pM) bearing a simple L-homophenylalanine residue at position 11 and a more bulky (2‘-Et, 4‘-OMe)4, 4’-L-biphenylalanine at the penultimate position (Fig. 1) (Haque et al., 2010a; Haque et al., 2010b). Unfortunately, affinity data were not presented in these studies.
Follow-up studies, based on these undecapeptides mainly focused on GLP-1 peptidomimetics with cyclic constraints or cyclic α-conotoxin-GLP-1 chimeras (Hoang et al., 2015; Swedberg et al., 2015; Swedberg et al., 2016). These modifications led to a clear activity decline in cAMP signaling (EC50) in comparison to the respective non-cyclized 11-mer peptides.
Although none of these undecapeptides were converted into an imaging agent for insulinomas so far, they might potentially serve as a basis for GLP-1R radioligand with an improved pharmacokinetic profile. More precisely, reabsorption by megalin on the tubular cells and secretion of radioactive degradation products into the blood might be decreased for small peptides (< 30 amino acids (Liu & Edwards, 1999)) derived from 1, as this active process is possibly triggered to a lesser extent by molecules structurally altered to exendin-3/4 and exhibiting fewer charged amino acid residues (Gotthardt et al., 2007; Jodal et al., 2015; Vegt et al., 2011; Tojo & Kinugasa, 2012). Accelerated clearance from the blood and kidneys would result in earlier adequate tumor-to-background and tumor-to-kidney ratios, particularly advantageous for diagnostic and therapeutic applications, which might be also combined in a theranostic approach (Weineisen et al., 2015), provided that agonist-induced side effects (hypoglycemia, nausea, vomiting) are still tolerable.
In addition, cumbersome labeling procedures as currently required for radiofluorination might become dispensable if the respective ligands would allow for derivatization with a silicon-based fluoride acceptor moiety. Improved labeling conditions in combination with the ‘Munich method’ (Wessmann et al., 2012) for 18F-drying might enable direct radiofluorination with higher RCYs, exploiting the favorable imaging characteristics of 18F (low β+ energy of max. 0.635 MeV, β+ decay ratio of 97%, half-life 109.8 min) and the possibility of generating GLP-1R-targeting radiohybrid ligands (Jacobson et al., 2015; Wurzer et al., 2020).
Thus, the aim of this study was to investigate whether undecapeptide 1 (Fig. 1) might serve as an alternative lead structure for GLP-1R targeting, based on its favorable GLP-1R-activating properties (EC50 (1) = 31 pM vs. EC50 (GLP-1) = 34 pM (Mapelli et al., 2009; Haque et al., 2010a)).
Derivatives of 1, suitable for direct radiolabeling by isotopic exchange on a SiFA-bearing prosthetic group, that could also be valuable precursors for future studies on the insertion of other modifications (e.g. attachment of a chelator), were planned to be preselected by affinity determinations. As a result, the main objective of this study was to generate precursor molecules with low nanomolar IC50 values that preferably lie in the range of endogenous GLP-1.