Synthesis and in vivo evaluation of an ¹⁸F-labeled glycoconjugate of PD156707 for imaging ET_A receptor expression in thyroid carcinoma by positron emission tomography

Introduction

Papillary (PTC) and follicular (FTC) thyroid carcinomas together constitute differentiated thyroid carcinoma (DTC), which is the most common endocrine malignancy. The incidence of DCT has risen steeply in the past three decades [1], not totally attributable to increased detection [2]. Since these carcinomas grow slowly and are generally well-treated by surgical resection and postoperative radioiodide ablation with iodide-131, 80-90% of patients suffering from DTCs survive longer than 10 years [3]. Nevertheless, a significant number of DTCs eventually become unresponsive to radioiodide treatment [4]. Since thyroid tissue is highly vascularized [5] combination therapy with anti-angiogenic agents is potentially beneficial for these patients [6]. Indeed, DTCs are responsive to inhibitors of tyrosine kinase receptors of vascular endothelial growth factor (VEGF-R), and other tyrosine kinase receptors such as EGFR (Vandetanib; [7]), PDGFR (Axitinib; [8]) or KIT (Motesanib; [9]).

In addition to VEGF and its receptors, the so-called endothelin axis is involved in tumor growth and progression [10]. The endothelin axis consists of the vasoactive peptide endothelin (ET) [11,12], which occurs in three isoforms (ET-1, ET-2, and ET-3), along with two distinct receptor subtypes (ET_AR and ET_BR) [13-15]. Activation of ET_AR by ET-1 contributes to the progress of angiogenesis through stimulation of VEGF expression [10,16]. Since increased expression of the ET axis is reported in DTCs [17-19], the ET_AR is a potential target for molecular imaging by positron emission tomography (PET), with an aim to improved clinical diagnosis and the prediction of therapy response. To this end, we have developed ¹⁸F-labeled derivatives of the non-peptide ET_AR ligand PD 156707, such as a fluoroethoxy derivative [20,21]. However, these compounds suffered from an unfavourable biodistribution due to their high lipophilicity. Drawing upon our previous experience with click-chemistry of ¹⁸F-labeled carbohydrates [22,23], we now report an optimized radiosynthesis of a subtype-selective ET_AR radioligand with introduction of a glycosyl moiety as a radiolabeled hydrophilic building block into the lead structure (Figure 1). We tested the hypothesis that this approach should impart improved pharmacokinetics of the ¹⁸F-labeled ET_AR ligand by undertaking biodistribution studies and small animal PET imaging of ET_AR expression in living mice bearing ET_AR-positive papillary thyroid K1 tumors.

Materials and methods

Results and discussion

Starting from the alkyne precursor 2, which has been described previously [21], the synthesis of the glycoconjugate 1 was performed by click chemistry with applying the copper(I)-catalyzed azide alkyne 1,3-dipolar cycloaddition (CuAAC) in the presence of 2-deoxy-2-fluoro-β-D-glucopyranosyl azide [23], sodium ascorbate and CuSO₄ (Figure 2). After confirmation of identity and purity of title compound 1, competition binding studies were carried out for the endothelin receptor subtypes ET_AR and ET_BR using [¹²⁵I]ET-1 and membranes from mouse myocardial ventricles, as described previously [21]. We found that 1 inhibited [¹²⁵I]ET-1 binding at ET_AR with a K_i value of 4.5 nM, whereas inhibition at ET_BR required a 266-fold higher concentration (Table 1). This ET_AR/ET_BR subtype selectivity of 1 is three times lower than that of the ET_AR-selective reference ligand PD 156707. Therefore, glycosylation of 2 resulted in some loss of ET_AR subtype selectivity for glycosyl derivative 1, although this reduced selectivity is unlikely to be physiologically relevant. This result confirmed our previous study of a series of fluorinated derivatives of the ET_AR antagonist PD 156707 [21], indicating very similar receptor binding data for ET_AR affinity and subtype selectivity.

Encouraged by the favorable in vitro binding results, we proceeded to set up and optimize the radiosynthesis of ¹⁸F-labeled 1, and characterize its binding properties in vivo binding properties using nude mice bearing ET_AR-positive thyroid K1 tumors. As shown in Figure 3, the radiosynthesis of [¹⁸F]1 accomplished through concomitant ¹⁸F-labeling and glycosylation [22]. In general, this strategy followed the previously described concept for the fluoroglycosylation of peptides, although obtain an optimized radiochemical yield (RCY) entailed some modifications. In particular, the amount of precursor was increased from 100 nmol (sufficient for peptide labeling [22]) to 300 nmol of the alkyne-bearing precursor 2, and the ethanol content of the solvent was increased from 10% to 60% to provide higher solubility of the non-peptidic precursor 2.

Starting from triflate precursor 3 [23], 2-deoxy-2-[¹⁸F]fluoroglucopyranosyl azide ([¹⁸F]4; Figure 3) was produced in about 30 min, including the adjacent deacetylation under basic conditions. The subsequent CuAAC reaction in the presence of alkyne 2 (300 nmol) proceeded smoothly in a mixture of ethanol/PBS at 60°C, providing [¹⁸F]1 in a RCY of 70-75% after 15 min (Figure 3). In comparison with the ¹⁸F-fluoroglucosylation of peptides [22], the amount of the alkyne precursor had thus to be increased threefold to achieve the optimum RCY. After semipreparative HPLC, [¹⁸F]1 was obtained in an overall radiochemical yield of 20-25% (not corrected for decay, referred to [¹⁸F]fluoride) in a total synthesis time of 70 min. The specific activities of [¹⁸F]1 at end of preparation was in the range of 41-138 GBq/μmol. The chemical identity of [¹⁸F]1 and the high radiochemical purity (> 99%) were confirmed by HPLC. After formulation of an injectable solution of the radioligand, the stability of [¹⁸F]1 after incubation in human serum at 37°C was at least 99% (Figure 6).

To characterize the binding properties of [¹⁸F]1 in vitro, we performed saturation binding experiments with [¹⁸F]1 in a concentration range between 0.2 nM and 100 nM using the human papillary thyroid carcinoma cell line K1, which demonstrated ET_AR expression by Western Blot (Figure 4A and 4B). Notably, the expression of ET_AR on K1 cells was rather low in vitro, compared to its expression in K1 tumor tissue (Figure 4B). Figure 4A demonstrated saturable specific binding of radioligand [¹⁸F]1 to K1 cells, indicating a dissociation constant (K_d) of 0.7 nM. We estimated the ET_AR density (B_max) to be 191 fmol/mg, which represents only about 57,200 ET_A receptors per K1 cell (191 fmol/mg = 9.5 × 10^-5 fmol/cell = 57,200 receptors/cell). When analyzing the K1 tumor tissue derived from tumor-bearing nude mice, the overall expression of ET_AR was determined, which we ascribe to the endothelial cells in the highly vascularized K1 thyroid tumors. This association was demonstrated by co-staining for ET_AR and the murine endothelial cell antigen MECA32 in slices from K1 tumor tissue (Figure 4C).

The xenotransplantation of human thyroid K1 cells was performed by subcutaneous injection of a cell suspension of 5 million cells into nude mice, followed by growth of the tumors for about 14-20 days. The tumors then attained a diameter of 9-15 mm and showed only minor necrotic areas upon dissection. Our biodistribution study of [¹⁸F]1 in K1 xenografted nude mice at 10, 30, and 60 min post-injection (p.i.; Figure 5) revealed [¹⁸F]1 to have fast blood clearance, and low uptake in the kidneys and liver. In contrast, we observed very high uptake of the intact tracer as verified by HPLC in the bile and intestines (Figures 5A, 6). This indicates predominant excretion of [¹⁸F]1 occurred via hepatobiliary clearance, a finding in accordance with previously studied ¹⁸F-fluoroethoxy derivatives [20] and the PEGylated ¹⁸F-derivatives of PD 156707 [21].

The present findings for biodistribution of glycosyl compound [¹⁸F]1 match closely with those described for ¹⁸F-alkyl-labeled ET_A radioligands [20,21]. We recalculated the [¹⁸F]1 uptake values as an “uptake index” for direct comparison with corresponding results for the fluoroethoxy and other derivatives [20,21]; this comparison revealed no major differences in the excretion pathway between the several tracers. The logD_7.4 value of the title compound (-0.4) indicates that glycosylation increased the hydrophilicity of the structure by a factor of 19, as compared to the most hydrophilic of the compounds reported in the paper by Michel et al., which had a logD_7.4 of 0.89 [21]. This increased hydrophilicity did not markedly influence the biokinetics in vivo, and did not impair the hepatobiliary excretion. However, glycosylation of the lead structure PD156707 had greatly enhanced the metabolic stability of the tracer in blood of living mice. Whereas the fluoroethoxy derivative showed about 50% polar radiometabolites within 20 min after injection into mice [20], the glycosyl derivative [¹⁸F]1 was entirely stable in serum in vitro and likewise in vivo, as no radiometabolite was detected in blood or gall bladder at 60 min post-injection (Figure 6).

The uptake of [¹⁸F]1 in healthy myocardium and in K1 tumors was significantly higher in the unblocked condition than in animals co-injected with PD156707, providing evidence for specific ET_AR-mediated binding in these tissues in vivo (Figure 5B). The physiological specific binding in healthy myocardium suggests that this class of ligand might be a useful adjunct to ¹³N-ammonia PET studies of endothelial function in heart [26]. The specific uptake of [¹⁸F]1 was 0.62 ± 0.18 %ID/g in tumor samples collected at 10 min, declining to about 0.25 ± 0.09 %ID/g at 60 min p.i. (Figure 5A). These findings ex vivo were entirely consistent with the time-activity curves derived from small-animal PET, which likewise revealed specific binding of [¹⁸F]1 in the tumor region, based on the more rapid clearance in animals coinjected with PD156707 (Figure 7). The mean uptake value (%ID/g) of [¹⁸F]1 by PET imaging at 55 min p.i. was 0.35 ± 0.07 (n = 4), declining to only 0.14 ± 0.07 (n = 4; (P < 0.05, t-test) in mice co-injected with PD156707), in close agreement with the dissection study (Figure 5B). Although the tumor uptake of [¹⁸F]1 and the tumor-to-tissue ratios were rather low, the ET_AR expression in K1 thyroid tumors was detectable by small-animal PET (Figure 7A); we are unaware of any previous molecular imaging studies of these thyroid tumors, so comparison of [¹⁸F]1 with other tracers is not possible. Given the excellent metabolic stability of [¹⁸F]1 in vivo, we find that [¹⁸F]1 PET has considerable potential for imaging ET_AR expression in thyroid tumors.

In conclusion, we established a reliable and efficient strategy of concomitant glycosylation and ¹⁸F-labeling by click-chemistry of the ET_AR ligand PD156707, providing the ¹⁸F-labeled glycoconjugate ligand [¹⁸F]1 for preliminary PET studies. As with the lead compound, the excretion of [¹⁸F]1 in mice occurred via hepatobiliary clearance and was not significantly influenced by glycosylation, which did however impart nearly complete metabolic stability in vivo. Biodistribution studies ex vivo and small-animal PET imaging proved the specificity of [¹⁸F]1 for ET_AR binding in K1 thyroid tumors and in healthy myocardium. However, [¹⁸F]1 revealed significant hydrophobicity with only low tumor uptake, rendering the clinical translation of this tracer highly unfavorable. Despite the apparently low expression of the marker in K1 thyroid xenografts, [¹⁸F]1 shows promise for studying ET_AR expression and its regulation in animal models of angiogenesis, notably in thyroid carcinoma.