Bone-seeking TRAP conjugates: surprising observations and their implications on the development of gallium-68-labeled bisphosphonates

General procedures and instrumentation (nuclear magnetic resonance (NMR), electrospray mass spectroscopy (ESI-MS), ultrafiltration/diafiltration, PET) have been described before [13]. [¹⁸ F]fluoride formulation for injection was prepared by adding 100 MBq of ¹⁸ F (obtained from routine cyclotron production at Klinikum rechts der Isar, Technische Universität München, München, Germany) to phosphate buffered saline (PBS) (1 mL).

Synthesis of bisphosphonate conjugates (Figure 2): TRAP·2H₂O (0.3 mmol, 185 mg), diisopropylethylamide (3 mmol, 388 mg, 510 μL), and the amino-bisphosphonate (1.5 mmol); for TRAP(MDP)₃, tetraethyl(aminomethylene)bisphosphonate 455 mg; and for TRAP(PDP)₃, tetraethyl(1-aminopropylene)bisphosphonate 500 mg, were dissolved in DMSO (2 mL). Then, HATU (2.4 mmol, 921 mg) was added with stirring. After 25 min, the reaction mixture was diluted with water (50 mL) and subjected to diafiltration (membrane with 500 Da MWCO). After 250 mL of water had passed, the cell content was concentrated in vacuo and subjected to preparative HPLC (column: YMC C18 ec 250 × 30 mm; detection wavelength, 220 nm; eluent A, MeCN with 0.1% TFA; eluent B, water with 0.1% TFA; gradient 25% to 50% B in 20 min, t _R(dodecaethyl-TRAP(MDP)₃) = 12 min, t _R(dodecaethyl-TRAP(PDP)₃) = 16 min). After evaporation of the solvents, the remaining viscous oil was dissolved in HBr/glacial acetic acid (33%) and stirred for 3 days. Addition of methanol to the reaction mixture yielded the products as colorless, crystalline solids. Data for TRAP(MDP)₃: yield 201 mg (61%); MW (calculated for C₂₁H₅₁N₆O₂₇P₉) 1,098.43; ESI-MS negative m/z = 1,097 (M-H⁺) and 548 (M-2 H⁺); ¹ H NMR (600 MHz, D₂O) δ = 2.13 (m, 6 H), 2.67 (m, 6 H), 3.48 (d, ³ J _HH = 5.4 Hz, 6 H), 3.56 (s, broad, 12 H), and 4.71 (t, J _PH = 21.3 Hz, 3 H) ppm; ¹³ C NMR (151 MHz, D₂O) δ = 26.11 (d, ¹ J _PC = 93.18 Hz), 28.65, 52.13, 54.74 (d, ¹ J _PC = 89.07 Hz), 47.45 (t, ¹ J _PC = 139.28 Hz), and 174.54 (dt, ² J _PC = 12.28 and ³ J _PC = 4.34 Hz) ppm; and ³¹P NMR (121 MHz, D₂O) δ = 14.10 (d, ² J _PP = 15.7 Hz) and 39.90 ppm. Data for TRAP(PDP)₃: yield 195 mg (55%); MW (calculated for C₂₇H₆₃N₆O₂₇P₉) 1,182.59; ESI-MS negative m/z = 1,181 (M-H⁺), 590 (M-2 H⁺), and 393 (M-3 H⁺); ¹ H NMR (600 MHz, D₂O) δ = 2.07 (m, 6 H), 2.10 (m, 6 H), 2.36 (tt, J _PH = 23.52 Hz, ³ J _HH = 5.97 Hz, 3 H), 2.53 (m, 6 H), 3.44 (t, ³ J _HH = 6.3 Hz, 6 H), 3.45 (t, broad, ³ J _HH = 6.6 Hz), and 3.52 (s, broad, 12 H) ppm; ¹³ C NMR (151 MHz, D₂O) δ = 25.36 (t, ² J _PC = 4.2 Hz), 26.29 (d, ¹ J _PC = 93.5 Hz), 28.63 (d, ² J _PC = 3.9 Hz), 35.77 (t, ¹ J _PC = 128.5 Hz), 39.48 (d, ³ J _PC = 7.4 Hz), 52.14, 54.82 (d, ¹ J _PC = 88.6 Hz), and 175.41 (d, ³ J _PC = 13.1 Hz) ppm; and ³¹P NMR (121 MHz, D₂O) δ = 21.53 (d, ² J _PP = 15.5 Hz) and 39.69 ppm.

⁶⁸ Ga for labeling was obtained from a SnO₂-based ⁶⁸Ge/⁶⁸ Ga generator (iThemba LABS, Somerset West, South Africa), eluted with 1.0 M HCl. A 1.25 mL fraction of the eluate containing ca. 80% of the total activity (ca. 1.3 GBq) was adjusted to pH 3.3 by adding a solution of 600 mg 2-[4-(2-hydroxyethyl)-1-piperazinyl]-ethanesulfonic acid (HEPES) in 0.5 mL water. To a 90 μL aliquot of this mixture, 10 μL of 10^-4 M stock solution of the ligand was added. After heating for 5 min to 95°C, the solution was passed over a cation exchanger SPE cartridge (Chromafix HR-XC M, Macherey-Nagel, Düren, Germany) and purged with water (1 mL). This procedure removed non-complexed ⁶⁸ Ga as well as HEPES, which was confirmed by processing of blank samples. Radiochemical yields, determined by measuring the activity on the cartridge and in the eluate, were > 85%. Formulation was done by adjusting the pH of the eluate to 7.4 by adding approximately 50 μL of a solution of NaOH (1 g) and NaH₂PO₄ (483 mg) in water (20 mL) while monitoring the pH with a pH meter. 'Free' ⁶⁸ Ga formulation was prepared by addition of the generator eluate (40 μL, ca. 50 MBq) to PBS (1 mL), resulting in pH 7.2.

All animal experiments were carried out in accordance with the current animal welfare regulations in Germany. Five male Lewis rats (age 7 weeks, ca. 200 g) were kept under standard laboratory conditions (12 h light/12 h dark) and given standard diet and water ad libitum. For PET, ca. 35 MBq of tracer was injected into the tail vein under isoflurane anesthesia. Two subsequent scans of 15 min were recorded 60 min post injection, using two different axial bed positions in order to image the entire animals. Images were reconstructed using a OSEM3D algorithm without scatter and attenuation correction. For each full-body maximum intensity projection (MIP), two part-body MIPs were stitched together manually using graphics software. PET images are from representative animals reflecting the group.

Figure 3 shows that free ⁶⁸ Ga(III) (we use this generalized term since ⁶⁸ Ga species in PBS solutions are not well defined) provides almost no contrast of the skeleton over other tissues, as intravenous injection of free ⁶⁸ Ga(III) predominantly results in transferrin-bound activity [14–17]. In contrast, both bisphosphonate tracers ⁶⁸ Ga-TRAP(MDP)₃ and ⁶⁸ Ga-TRAP(PDP)₃ bind to bone while showing low levels in blood and soft tissues. Apparently, PET image quality achieved therewith cannot compete with that of [¹⁸ F]fluoride. ¹⁸ F possesses a lower positron energy than ⁶⁸ Ga, resulting in lower tissue penetration (FW20H of 0.54 mm and 2.12 mm in soft tissue for ¹⁸ F and ⁶⁸ Ga, respectively [18]), and therefore in a lower degree of image blurring. However, as the difference in resolution for a clinical 3-mm PET camera is small (3.05 mm for ¹⁸ F and 3.57 mm for ⁶⁸ Ga [18]), a successful application of ⁶⁸ Ga bone-imaging agents in patients is not precluded.

Upon investigation of the mode of gallium binding, we found that an equimolar mixture of ^69,71 Ga³⁺ and either ⁶⁸ Ga-TRAP(MDP)₃ or ⁶⁸ Ga-TRAP(PDP)₃ does not yield any signal in ⁷¹ Ga NMR spectra, not even after heating to 95°C for hours. However, the octahedral N₃O₃ coordination usually found for 'in-cage' Ga(III) complexes of TRAP ligands generally yields sharp ⁷¹ Ga NMR resonances at δ = 130 to 142 ppm [11, 12]. Obviously, Ga(III) ion is not located in the TRAP cavity and, therefore, must be complexed in an 'out-of-cage' manner by the bisphosphonate groups. Although this result is quite unexpected, PET images nevertheless prove that the degree of kinetic stability of these complexes is sufficiently high to carry ⁶⁸ Ga to the bone and retain it there. However, Figure 3 also shows a slightly higher background uptake for ⁶⁸ Ga-TRAP(MDP)₃, most likely caused by partial decomplexation in vivo due to lower complex stability. Clearance of both ⁶⁸ Ga tracers occurred faster than ¹⁸ F^- and exclusively via the kidneys.

⁶⁸ Ga-labeled trimeric bisphosphonate conjugates of TRAP were successfully applied for bone imaging in rats. Surprisingly, ⁷¹ Ga NMR investigation revealed that Ga(III) ion is not located in the macrocyclic cavity of TRAP and, therefore, must be complexed by one or more side chain bisphosphonates. Although the primary chelation site of TRAP possesses excellent Ga(III) complexing properties [12], it apparently cannot compete with the bisphosphonates. In ⁶⁸ Ga-TRAP(MDP)₃ and ⁶⁸ Ga-TRAP(PDP)₃, TRAP thus merely serves as a scaffold, and its ability for ⁶⁸ Ga binding is not required. We therefore conclude that in designing bisphosphonate tracers for ⁶⁸ Ga-based PET bone imaging, the introduction of chelating moieties other than the bisphosphonates themselves might be unnecessary. Rather, it appears to be sufficient to equip suitable branched scaffolds with multiple bisphosphonate units which serve both ⁶⁸ Ga-binding and bone-targeting purposes.