Synthesis and preclinical evaluation of DOTAGA-conjugated PSMA ligands for functional imaging and endoradiotherapy of prostate cancer

Background Due to its high expression in prostate cancer, PSMA (prostate-specific membrane antigen) represents an ideal target for both diagnostic imaging and endoradiotherapeutic approaches. Based on a previously published highly specific PSMA ligand ([68Ga]DOTA-FFK(Sub-KuE)), we developed a corresponding metabolically stable 1,4,7,10-tetraazacyclododececane,1-(glutaric acid)-4,7,10-triacetic acid (DOTAGA) construct for theranostic treatment of prostate cancer. Methods All ligands were synthesized by a combined solid phase and solution phase synthesis strategy. The affinity of the natgallium and lutetium complexes to PSMA and the internalization efficiency of the radiotracers were determined on PSMA-expressing LNCaP cells. The 68Ga- and 177Lu-labelled ligands were further investigated for lipophilicity, binding specificity, metabolic stability, as well as biodistribution and μPET in LNCaP-tumour-bearing mice. Results Radiochemical yields for 68Ga (3 nmol, 5.0 M NaCl/2.7 M HEPES (approximately 5/1), pH 3.5 to 4.5, 5 min, 95°C) and 177Lu labelling (0.7 nmol, 0.1 M NH4OAc, pH 5.5, 30 min, 95°C) were almost quantitative, resulting in specific activities of 250 to 300 GBq/μmol for the 68Ga analogues and 38 GBq/μmol for 177Lu complexes. Due to metabolic instability of l-amino acid spacers, d-amino acids were implemented resulting in a metabolically stable DOTAGA ligand. Compared to the DOTA ligand, the DOTAGA derivatives showed higher hydrophilicity (logP = −3.6 ± 0.1 and −3.9 ± 0.1 for 68Ga and 177Lu, respectively) and improved affinity to PSMA resulting in an about twofold increased specific internalization of the 68Ga- and 177Lu-labelled DOTAGA analogue. Especially, [68Ga]DOTAGA-ffk(Sub-KuE) exhibits favourable pharmacokinetics, low unspecific uptake and high tumour accumulation in LNCaP-tumour-bearing mice. Conclusions The pair of diagnostic/therapeutic PSMA-ligands [68Ga/177Lu]DOTAGA-ffk(Sub-KuE) possess remarkable potential for the management of prostate cancer.


Background
Prostate-specific membrane antigen (PSMA, EC 3.4.17.21, synonym: glutamate carboxypeptidase II) is an extracellular hydrolase whose catalytic centre comprises two zinc(II) ions with a bridging hydroxido ligand [1]. It is highly upregulated in metastatic and hormone-refractory prostate carcinomas, but its endogenous expression has also been reported in kidneys, salivary glands, small intestine, brain and, to a low extent, also in healthy prostate tissue [2,3]. In the intestine, PSMA facilitates absorption of folate by conversion of pteroylpoly-γ-glutamate to the pteroylglutamate (folate) [4]. In the brain, it hydrolyses N-acetyl-Laspartyl-L-glutamate (NAAG) to N-acetyl-L-aspartate and glutamate [5]. The enzymatic function of PSMA in normal and diseased prostate has not been clarified yet [6]. However, due to its overexpression on prostate cancer cells, PSMA represents an excellent target for molecular imaging and targeted radiotherapy of prostate cancer.
Amongst the currently available small-molecule PSMA inhibitors, only [ 131 I]MIP-1375 [23] and [ 131 I]MIP-1466 [24] (Figure 1) have been evaluated in therapy studies so far. In LNCaP-tumour-bearing nude mice, [ 131 I]MIP-1375 inhibited tumour growth in a time-and dosedependent manner. First, human applications of [ 131 I] MIP-1095/1466 revealed promising therapy response with moderate side effects [25]. The structurally corresponding diagnostics, 123 I-labelled analogues [ 123 I]MIP-1072 and [ 123 I]MIP-1095, were shown to possess suitable characteristics for SPECT imaging. However, these compounds are comparably lipophilic (clogP = 0.20 and 1.97, respectively [25]), and their compact structure complicates further optimization of PSMA affinity and pharmacokinetics. Furthermore, PET imaging is only possible using 124 I, a suboptimal PET-radionuclide with respect to resolution and dosimetry.
In contrast, radiometalated analogues based on the FFK(Sub-KuE)-scaffold represent a much more flexible and finely adjustable backbone for the development of KuE-based PSMA inhibitors, that allow labelling with both diagnostic and therapeutic radionuclides, e.g. 68 Ga for PET imaging and 177 Lu/ 90 Y for therapy after conjugation of DOTA. However, in order to further facilitate the labelling procedure for 177 Lu and 90 Y, improve ligand pharmacokinetics, and potentially obtain radiometalated derivatives with higher affinity, we substituted DOTA in DOTA-FFK(Sub-KuE) by 1,4,7,10-tetraazacyclodocecane,1-(glutaric acid)-4,7,10-triacetic acid (DOTAGA) [26,27]. Improved affinities, higher tumour uptake and faster kidney clearance have already been observed for the 68 Gacomplex (one free carboxylate) compared to the 90 Y-complex (all carboxylates coordinated) of DOTA-coupled somatostatin analogues before [28].
Consequently, we evaluated and compared the respective 68/nat Ga-and 177/nat Lu-DOTAGA analogues of FFK (Sub-KuE) and ffk(Sub-KuE) in terms of PSMA affinity, uptake in PSMA positive tumour cells, metabolic stability, in vivo biodistribution and μPET imaging. The previously described DOTA analogues [17] and [ 68 Ga]HBED-CC-Ahx-KuE [18] were also included to allow direct comparability of our results with the data in the literature.

General
Fmoc-(9-fluorenylmethoxycarbonyl-) and all other protected amino acid analogues were purchased from Iris Biotech (Marktredwitz, Germany) or Bachem (Bubendorf, Switzerland). Tritylchloride polystyrene (TCP) resin was obtained from PepChem (Tübingen, Germany). The chelators DOTA-tris-tBu-ester and DOTAGA-anhydride were purchased from Chematech (Dijon, France). Solvents and all other organic reagents were purchased from Sigma-Aldrich (Munich, Germany) or CLN (Freising, Germany). Solid phase peptide synthesis was carried out manually using an Intelli-Mixer syringe shaker (Neolab, Heidelberg, Germany). Analytical reversed-phase high performance liquid chromatography (RP-HPLC) was performed on a Nucleosil 100 C18 (5 μm, 125 × 4.0 mm) column (CS GmbH, Langerwehe, Germany) using a Sykam gradient HPLC System (Sykam GmbH, Eresing, Germany). The peptides were eluted applying different gradients of 0.1% (v/v) trifluoroacetic acid (TFA) in H 2 O (solvent A) and 0.1% TFA (v/v) in acetonitrile (solvent B) at a constant flow of 1 mL/min (specific gradients are cited in the text). UV detection was performed at 220 nm using a 206 PHD UV-Vis detector (Linear™ Instruments Corporation, Reno, USA). Both retention times t R as well as the capacity factors K' are cited in the text. Preparative RP-HPLC was performed on the same HPLC system using a Multospher 100 RP 18-5 (250 × 20 mm) column (CS GmbH, Langerwehe, Germany) at a constant flow of 9 mL/min. Radio-HPLC of the radioiodinated reference ligand was carried out using a Nucleosil 100 C18 (5 μm, 125 × 4.0 mm) column. For radioactivity detection, the outlet of the UV-photometer was connected to a NaI(Tl) well-type scintillation counter from EG&G Ortec (Munich, Germany). Analysis of 68 Ga-labelled compounds was done as described previously [29]. ESI-mass spectra were acquired on a Varian 500-MS IT mass spectrometer (Agilent Technologies, Santa Clara, USA).
using an eluent mixture of ethyl acetate/hexane/TEA (500/500/0.8 (v/v/v)). Upon solvent evaporation, 4.80 g of 2 were obtained as a colourless, sticky oil (yield: 80% based on L-di-tert-butyl glutamate · HCl  (3): For Cbz deprotection, 6.037 g (9.71 mmol, 1.0 eq) of 2 was dissolved in 150 mL of ethanol (EtOH), and 0.6 g (1.0 mmol, 0.1 eq) of Palladium on activated charcoal (10%) was added. After purging the flask with H 2 , the solution was stirred overnight under light H 2 -pressure (balloon). The crude product was filtered through Celite, the solvent was evaporated in vacuo, and the desired product was obtained as a waxy solid (
Assembly of the peptide sequence H 2 N-Phe-Phe-on resin-bound Lys(Boc) was performed according to a standard Fmoc-protocol using 1.5 eq of 1-hydroxybenzotriazole (HOBt) and O-(1H-benzotriazol-1-yl)-N,N,N' ,N'-tetramethyluronium-tetrafluoroborate (TBTU) as coupling reagents and 4.5 eq DIPEA. After coupling of the last amino acid, the resin was washed, dried and stored in a desiccator until further functionalization.

Coupling of chelating moiety
Fmoc-Phe-Phe-Lys(Boc)-TCP resin was allowed to preswell in N-methyl-pyrrolidon (NMP) for 30 min. After cleavage of the N-terminal Fmoc-protecting group using 20% piperidine in DMF (v/v), the resin was washed eight times with NMP. The coupling of the respective chelators is described below. Cleavage from the resin (2 × 30 min) and concomitant tBu-deprotection was performed using a mixture (v/v/v) of 95% TFA, 2.5% triisobutylsilane (TIBS) and 2.5% water. The combined product solutions were then concentrated, the crude peptide was precipitated using diethyl ether and was dried in vacuo. Due to sufficient purity of the crude products, they were used for the following reaction step without further purification.

Synthesis of unlabelled reference compounds
(I-BA)KuE: To a solution of 15 mg (0.031 mmol, 1.0 eq) 3 in 2 mL DMF was added 11.4 mg (0.046 mmol, 1.5 eq) of 4-iodo-benzoic acid, 6.3 mg (0.046 mmol, 1.5 eq) of 1-Hydroxy-7-azabenzotriazole (HOAt), 7.2 μL (0.046 mmol, 1.5 eq) diisopropyl carbodiimide (DIC) and 23.7 μL (0.138 mmol, 4.5 eq) DIPEA. The yellow solution was stirred for 20 h, diluted with ethyl acetate and extracted with water. The organic layer was dried over Na 2 SO 4 , filtered and evaporated to dryness. For tBu-deprotection, the crude product was dissolved in 200 μL TFA. After 30 min, the solvent was evaporated and the crude product was purified using preparative RP-HPLC (isocratic eluent:   (1/1) and incubated at RT for 10 min. The product was diluted with 10 mL of water and loaded onto a C18 Sep Pak Plus cartridge, which had been preconditioned with 10 mL of MeOH and rinsed with 10 mL of water. The cartridge was then washed with 10 mL water, and the product was eluted in 300 to 500 μL fractions with a 1:1 mix (v/v) of EtOH/MeCN (2 mL). The radioactive fractions were evaporated to dryness, and the residue was dissolved in 200 μL TFA. After 30 min, the solvent was evaporated in vacuo. The crude product was isolated from unlabelled tributyltin precursor by RP-HPLC (20% to 40% B in 20 min) to afford the desired product (10.9 MBq). HPLC (20% to 40% B, 220 nm): t R = 13.0 min, K′ = 6.22.
68 Ga-labelling: A 1.25 mL fraction of 68 Ge/ 68 Ga generator (iTHEMBA Labs, South Africa) eluate (1 M HCl) was loaded onto a self-filled cartridge containing 300 mg SCX material (Bond Elut-SCX, Varian). The cartridge was then washed with water (1.0 mL) and purged with air. The 68 Ga 3+ was eluted with aq. NaCl (5 M, 0.5 mL) and HEPES (2.7 M aq., 90 or 140 μL) was added (pH 3.5 or 4.5, respectively). That solution was mixed with the precursor (3 nmol in 10 μL) and heated at 95°C for 5 min. After cooling, labelling efficiency and radiochemical purity were determined using Radio-TLC and Radio-HPLC. Radiochemical purity of all 68 Ga-labelled conjugates was ≥95%. Therefore, the tracers were diluted and used in vitro experiments without further purification. Tracers for in vivo use were prepared in GMPcompliant procedure using a fully automated synthesis module (Scintomics GmbH, Germany) similar to the procedure described previously [33]. For animal studies, the EtOH for eluting the labelled tracer from a SPE cartridge was evaporated in vacuo.

Determination of lipophilicity
To a solution of 0.5 to 1 MBq of radiolabelled peptide in 500 μl of phosphate-buffered saline (PBS, pH 7.4), 500 μl of n-octanol was added (n = 6). Vials were vortexed vigorously for 3 min. To achieve quantitative phase separation, the vials were centrifuged at 6,000 g for 5 min in a Biofuge 15 (Heraeus Sepatech, Osterode, Germany). The activity concentrations in 100 μl-samples of both the aqueous and the organic phase were measured in a γ-counter. Both the partition coefficient P (o/ w) , which is defined as the molar concentration ratio of a single species A between n-octanol and water at equilibrium, and logP (o/w) , which is an important parameter used to characterize lipophilicity of a compound, were calculated.

Metabolite analysis
177 Lu-10 and 177 Lu-11 were incubated in 80 μL of human serum (37°C) and 80 μL 0.4 M DTPA solution (RT), respectively and after 1, 24 and 48 h, the stability was determined by TLC analysis. The 68 Ga-labelled tracers (40 to 45 MBq) were injected into the tail vein of CD-1 nu/nu mice. The animals were sacrificed after 30 min, and urine, blood and kidney samples were taken. The kidney was frozen in liquid nitrogen, homogenized with a ball mill and extracted with 0.2 to 1 mL PBS containing 200 nmol 2-(phosphonomethyl)pentane-1,5-dioic acid (PMPA). The suspension was first centrifuged (15,000 g), and after ultrafiltration, the extracts were analysed by RP-HPLC. The blood samples were centrifuged to separate the plasma from the blood cells. Plasma proteins were removed by precipitation with acetonitrile (10 min, 4°C) and subsequent centrifugation and ultrafiltration. The blood extract was also analysed by RP-HPLC.

In vitro assays
Cell culture: PSMA + LNCaP cells (CLS: 300265) were grown in DMEM/Nutrition Mix F-12 with Glutamax-I (1:1) (Invitrogen, Life Technologies, Darmstadt, Germany) supplemented with 10% FCS. Cultures were maintained at 37°C in a 5% CO 2 /humidified air atmosphere. One day prior to the experiment, cells were harvested using Trypsin/EDTA (0.05% and 0.02%) in PBS, centrifuged and resuspended with culture medium. For cell counting, a Countesse automated cell counter (Invitrogen, Carlsbad, USA) was used. All in vitro binding and internalization studies were performed using live cells seeded 1 day prior to the experiment. For IC 50 determination, 150,000 cells/ mL were transferred to 24-well plates (1 mL/well), and for internalization studies, 125,000 cells/mL were transferred into PLL-coated 24-well plates.
Determination of IC 50 : The culture medium was removed, and the cells were washed once with 500 μL of HBSS (Hank's balanced salt solution, Biochrom, Berlin, Germany, containing 1% bovine serum albumin (BSA)), before being left to equilibrate in 200 μL of HBSS (1% BSA) on ice for 15 min. Then, 25 μL/well of either HBSS (1% BSA; control) or of solutions containing the respective unlabelled ligand of interest in increasing concentrations (10 -10 -10 -4 M in HBSS (1% BSA)) was added, followed by the addition of 25 μL of ([ 125 I]I-BA)KuE in HBSS (1% BSA). Experiments were carried out in triplicate for each concentration. The final radioligand concentration was 0.2 nM in all binding assays. Cells were incubated on ice for 60 min. Incubation was terminated by the removal of the incubation medium. Cells were thoroughly rinsed with 250 μL of HBSS. The wash medium was combined with the supernatant of the previous step. This fraction represents the amount of free radioligand. Cells were then lysed using 250 μL of 1 N NaOH, the lysate was transferred to vials and combined with 250 μL of HBSS used for rinsing the wells. Quantification of the amount of free and bound activity was performed in a γ-counter.
Internalization studies: The culture medium was removed, and the cells were washed once with 500 μL DMEM-F12 (5% BSA) before being left to equilibrate in 200 μL DMEM-F12 (5% BSA) at 37°C for a minimum of 15 min. Then, 25 μL (per well) of either DMEM-F12 (5% BSA) or of a 100-μM PMPA solution (blocking) was added, followed by the addition of 25 μL of 68 Ga/ 177 Lulabelled PSMA ligand. The final radioligand concentration was 0.2/0.5 nM in all internalization assays. To determine internalization kinetics, cells were then incubated at 37°C for 5, 15, 30 and 60 min, respectively. Experiments were carried out in triplicate for each time point (control and blocking). Incubation was terminated by placing the plate on ice for approximately 1 min and by subsequent removal of the incubation medium. Cells were thoroughly rinsed with 250 μL of PBS. The wash medium was combined with the supernatant of the previous step. This fraction represents the amount of free radioligand. To remove receptor surface bound radioactivity, the cells were then incubated for 10 min with 250 μL of ice cold PMPA solution (10 μM in PBS). After removal of the supernatant, the cells were thoroughly rinsed with another 250 μL of ice cold PBS. Both fractions were combined. The internalized activity was released by incubation with 250 μL of 1 N NaOH, transferred to vials and combined with 250 μL of 1 N NaOH used for rinsing the wells. Quantification of the amount of free, PMPA-releasable and internalized activity was performed in a γ-counter.

In vivo experiments
All animal experiments were conducted in accordance with general animal welfare regulations in Germany (Deutsches Tierschutzgesetz, approval #55.2-1-54-2532-71-13). The prostate cancer cell line LNCaP was suspended 1/1 in serum-free medium and Matrigel (BD Biosciences, Germany) and approximately 10 7 cells in 200 μL were inoculated subcutaneously on the right shoulder of 6 to 8 weeks old CD-1 nu/nu mice (Charles River Laboratories). Tumours were grown for 2 to 4 weeks (males) and 4 to 6 weeks (females) to reach 4 to 8 mm in diameter.
Biodistribution: The 68 Ga-labelled PSMA ligands (approximately 7 to 15 MBq, 0.1 to 0.2 nmol) were injected into the tail vein of isoflurane anesthetized animals. The organ distribution was examined 1 h p.i. and quantified in a γ-counter.
μPET imaging: Imaging studies were performed at a Siemens Inveon small animal PET, followed by data analysis using the Inveon Research Workplace software. The animals were anesthesized with isoflurane and injected via tail vein with 14 to 18 MBq (0.2 nmol) of tracer. Dynamic imaging was performed after on-bed injection for 1.5 h. Static images were recorded at 1 h p.i. with an acquisition time of 15 min. For the blockade image, animals were coinjected with 8 mg/kg of PMPA. Images were reconstructed using 3D ordered-subsets expectation maximum (OSEM3D) algorithm without scanner and attenuation correction.
The ligands DOTA-FFK(Sub-KuE) (10), DOTAGA-FFK(Sub-KuE) (11) and DOTAGA-ffk(Sub-KuE) (12) were prepared according to The tBu-protected PSMA binding motif 3 was first reacted with DSS in solution and then coupled with the chelatorfunctionalized peptides 7, 8 and 9. After acidic cleavage of the remaining tBu-protecting groups, the final products were purified by RP-HPLC. The products were obtained in approximately 15% yield (based on tripeptide). The identity of 10, 11 and 12 was confirmed by MS.
The nat Ga III and nat Lu III complexes of compounds 10, 11 and 12 were prepared by incubation of 10, 11 and 12 with an equimolar amount of aq. Ga(NO 3 ) 3 at 40°C for 30 min and an 2.5-molar excess of LuCl 3 at 95°C for 30 min, respectively. Quantitative metal complex formation was confirmed by RP-HPLC and MS.

Radiochemistry
([ 125 I]I-BA)KuE, the reference radioligand in all PSMA binding assays carried out in this study, was synthesized in solution starting from 4-iodo-benzoic acid according to methods described in the literature [12,31,32]. The respective Bu 3 Sn-precursor 13 was destannylated with [ 125 I]NaI using peracetic acid as the oxidizing agent within 10 min at RT. After cartridge purification subsequent tBu-deprotection and RP-HPLC purification, the final product was obtained in a radiochemical yield of 44 ± 5% and radiochemical purity of >99%. The reference material for chromatographic analyses (cold) (I-BA)KuE was synthesized by coupling 4-iodo-benzoic acid to 3, subsequent tBu-deprotection and RP-HPLC purification.
Preparation of 68 Ga for labelling of 10, 11 and 12 was performed by combination of previously described methods [34,35] with minor modifications. The 68 Ga 3+ eluted with 1 M HCl from a 68 Ge/ 68 Ga generator (iThemba Labs, South Africa) was retained on a strong cation exchange cartridge followed by elution with 0.5 mL 5 M NaCl, resulting in highly concentrated 68 Ga activity. Quantitative 68 Ga-complexation was achieved using 3 nmol of the respective DOTA-or DOTAGAconjugated PSMA inhibitors (95°C, 5 min, pH adjusted to 3.5 and 4.5 by addition of 2.7 M HEPES, respectively), allowing their use in in vitro studies without further purification. Specific activities of the 68 Ga-labelled PSMA inhibitors were 250 to 300 GBq/μmol. The 68 Ga-labelling for in vivo biodistribution and PET imaging studies were carried out using a fully automated GMP-compliant procedure using a GRP synthesizer (Scintomics GmbH, Germany) [33]. In these cases, the obtained specific activities were 80 to 120 GBq/μmol.

PSMA binding affinity
The PSMA binding affinities were determined in a competitive binding assay using LNCaP human prostate carcinoma cells and the known high affinity PSMA ligand ([ 125 I]I-BA)KuE [10] (c = 0.2 nM) as the radioligand. The IC 50 values for the metal-free PSMA-inhibiting compounds and their respective nat Ga and nat Lu complexes are summarized in Table 1. HBED-CC-Ahx-KuE and [ nat Ga]HBED-CC-Ahx-KuE [12] were included in this study as reference compounds. The PSMA affinities of 11 and 12, as well as their respective nat Ga and nat Lu analogues were consistently higher than the affinities of the respective DOTA constructs (10). Substitution of the L-amino acids in the linker region of the ligands by D-amino acids showed only a negligible effect on PSMA affinity (12 vs 11).

Internalization kinetics and specificity of cell binding
The cellular uptake and internalization kinetics of the different 68 Ga-and 177 Lu-labelled PSMA ligands was determined using PSMA-expressing LNCaP cells. In the case of the 68 Ga-labelled compounds, the final peptide concentration in the assays was 0.2 nM. Since all 177 Lu analogues were obtained in significantly lower specific activities than the 68 Ga analogues, the ligand concentration had to be increased to 0.5 nM to obtain reasonable count rates for assay evaluation. All internalization studies were accompanied by reference studies using ([ 125 I]I-BA)KuE (c = 0.2 nM). This experimental setup allows data normalization and eliminates the influence of cell count and cell viability on absolute cellular tracer uptake. To investigate the specificity of binding to PSMA, experiments were also carried out in the presence of 10 μM PMPA, a known high-affinity PSMA inhibitor (1.4 nM, [7]). Furthermore, to be able to discriminate between total cellular activity (sum of membrane associated and internalized activity) and internalized activity, all incubations were followed by a washing step with 10 μM PMPA at 4°C to remove specifically cell-surface bound radioligand by displacement.
Non-specific binding in the presence of 10 μM PMPA was lower than 0.5% for all compounds investigated. As expected, the cellular binding and internalization of all radiometalated PSMA ligands in this study increased over the time of observation (1 h). For all compounds, the fraction of internalized activity ranged between 4.5 ± 0.2% and 27.7 ± 1.6% of the total added activity after 1 h. The fraction of non-internalized, i.e. membrane-bound activity was 1% to 3% in all experiments (exemplarily shown for 177 Lu-10 and 177 Lu-11 in Figure 5).
Variations in cell count and/or cell viability between experiments can significantly affect the cellular tracer uptake but cannot be fully avoided. Therefore, internalization data ( Table 2) are normalized to the uptake of ([ 125 I]I-BA)KuE, that was always assayed in parallel as an external reference. All DOTA and DOTAGAcoupled ligands showed a lower internalization efficiency compared to the radioiodinated reference compound, whereas [ 68 Ga]HBED-CC-Ahx-KuE was internalized almost to the same degree as ([ 125 I]I-BA)KuE.

Determination of lipophilicity
The partition coefficient logP between n-octanol and PBS was determined for the compounds listed in Table 3 using the shake flask method. All compounds are highly hydrophilic, with the DOTAGA compounds being up to one order of magnitude more hydrophilic than the corresponding DOTA analogues. Amongst all compounds tested, [ 68 Ga]HBED-CC-Ahx-KuE was the most hydrophilic compound.

Metabolite analysis
The stability of 177 Lu-10 and 177 Lu-11 was investigated by incubation in human serum for 48 h at 37°C and by DTPA challenge (0.4 M, RT, 48 h) and subsequent TLC analysis. Release of 177 Lu III from the complexes was not observed under the respective experimental conditions in the examined time frame.
To investigate the influence of the amino acid composition of the spacer (KFF vs kff ) on the metabolic stability in vivo, tissue homogenates and body fluids of mice injected with 68 Ga-11 and 68 Ga-12 were analysed by radio-HPLC. Representative HPLC profiles of extracts and body fluids are shown in Figure 6. Extraction efficiencies from the blood and from the kidney were 56% and 43% for 68 Ga-11 and 61% and 62% for 68 Ga-12, respectively. Rapid in vivo degradation was observed for 68 Ga-11 consisting of the L-amino acid tripeptide FFK, resulting in only 21% intact tracer in blood at 0.5 h. 68 Ga-12 that consists of the D-amino acid tripeptide ffk, was found to be stable in blood (100% intact tracer) at 0.5 h p.i..  Figure 8 confirm higher tumour-to-background ratios for 68  compared to the respective DOTA analogue 68 Ga-10 (b). In agreement with the results obtained from the biodistribution (Figure 7), the metabolically stable 68 Ga-12 (d) exhibits increased tumour accumulation and a lower unspecific whole body uptake and tumour-to-background ratios even higher than those of the reference compound [ 68 Ga]HBED-CC-Ahx-KuE (a). That tracer uptake into tumour and kidneys is specific and PSMA mediated [18], as illustrated by the blocking experiment with PMPA (8 mg/kg) (Figure 8d′).

Biodistribution and small-animal PET studies
In addition, the biodistribution kinetics of 68 Ga-10, 68 Ga-11, 68 Ga-12, and the reference compound [ 68 Ga] HBED-CC-Ahx-KuE were investigated by carrying out dynamic μPET scans over a period of 1.5 h in CD-1 nu/ nu mice bearing LNCaP tumour xenografts (Figure 9). Despite the somewhat lower individual tumour uptake of 68 Ga-12 observed in this μPET-study, the markedly enhanced clearance of this compound from non-target  tissues resulted in higher tumour-to-tissue ratios for 68 Ga-12 compared to [ 68 Ga]HBED-CC-Ahx-KuE.

Discussion
Due to the consistent overexpression of PSMA in especially hormone-refractory and metastatic prostate cancer, this cell surface enzyme represents an excellent target for high-contrast PET imaging and also for therapeutic applications. The currently available PSMA-targeted radiopharmaceuticals are highly optimized either for imaging [36,37] or endoradiotherapeutic applications [25], but so far none of the compounds was evaluated for a theranostic concept. Our approach was therefore focused on the development of a PSMA ligand allowing for both, complexation with, e.g. 68  In a first attempt to improve the pharmacokinetics of the DOTA-coupled PSMA ligand 10, we increased the hydrophilicity of the ligand by substitution of DOTA by DOTAGA, resulting in 11. Although [ 68 Ga]HBED-CC-Ahx-KuE contains the lipophilic HBED-CC chelator, which improves binding affinity [18], a logP of -4.1 ± 0.1 was determined, revealing a significantly higher hydrophilicity compared to logP = -3.1 ± 0.2 for 68 Ga-10. The hydrophilicity of the 68 Ga-labelled DOTAGA-coupled ligand 11 increased to log P = -3.6 ± 0.1. For 68 Ga, as well as 177 Lu labelling, the DOTA-conjugated ligand 10   Ga-11 (3% for 3 min, 3% to 95% in 6 min, 95% for 3 min) and 68 Ga-12 (3% to 95% in 6 min, 95% for 3 min). and the DOTAGA-conjugated ligands 11 and 12 exhibited no differences in labelling efficiency.
To examine the metabolic stability of these new PSMA tracers in vivo, CD-1 nu/nu mice were injected with the 68 Ga-labelled ligand 11. Due to its L-amino acid spacer, 68 Ga-11 was rapidly metabolized. Furthermore, we demonstrated that fast metabolization of 68 Ga-10 and 68 Ga-11 resulted in discontinuous clearance kinetics and low tumour accumulation in vivo (Figure 7, 8, 9). This most likely explains the finding in the literature, that despite moderate affinity towards PSMA (IC 50 = 29.5 ± 6.6 nM) the DOTA ligand 68 Ga-10 exhibited unfavourable tumour targeting compared to [ 68 Ga]HBED-CC-Ahx-KuE. Further, we observed significant inter-individual differences in metabolization kinetics for 68 Ga-11 (with its L-amino acid spacer) in mice, which might explain inconsistent in vivo results obtained with PSMA ligands with L-amino acid spacers [38,39]. To overcome the problem of rapid proteolytic cleavage of radiolabelled 11, we substituted the Lamino acid spacer (FFK) by its D-amino acids counterpart ffk (12) resulting in high metabolic stability in vivo.
In addition to the improvement in metabolic stability for 12, we found that the DOTAGA-conjugated ligands 11 and 12, as well as their gallium(III) and lutetium(III) complexes, showed a significantly increased affinity towards PSMA on LNCaP cells compared to the DOTA-  coupled ligand 10/ nat Ga-10/ nat Lu-10. Especially, nat Lu-10 (IC 50 = 54.7 ± 6.1 nM) was less affine compared to nat Lu-11 (15.1 ± 1.5 nM) or nat Lu-12 (13.1 ± 2.2 nM). This finding most likely is due to the increased charge of the DOTAGA-ligands. In our competitive binding assay, unlike reported for other assays in the literature [18], a significantly lower affinity for nat Ga-10 (IC 50 of 29.5 ± 6.6 nM) was determined compared to [ nat Ga]HBED-CC-Ahx-KuE (IC 50 = 6.1 ± 0.8 nM).
As expected from their enhanced PSMA affinities, the DOTAGA ligands 68 Ga/ 177 Lu-11 and 68 Ga/ 177 Lu-12 showed up to twofold increased internalization compared to the corresponding DOTA ligands 68 Ga/ 177 Lu-10. Interestingly, 68 Ga/ 177 Lu-12 with the ffk spacer showed a higher internalization rate than ligand 11 with the FFK spacer. Blocking experiments with PMPA highlight the high PSMA-specificity of these tracers.
After the initial distribution phase of a radiopharmaceutical, straightforward clearance kinetics is typically characterized by a linear decrease in a semilogarithmic plot (two-compartment model). Interestingly, this kinetic profile (Figure 9) was only observed for 68 Ga-12, whereas Figure 9 Time-activity-curves (logarithmic plot) in % ID/mL for blood pool (heart), muscle, kidney and tumour. Graphs for (a) [ 68 Ga]HBED-CC-Ahx-KuE, (b) 68 Ga-10, (c) 68 Ga-11, (d) 68 Ga-12 are derived from dynamic small animal PET data. all other tracers (or their radiolabelled metabolites) exhibit unspecific retention or redistribution effects, resulting in non-linear kinetics (e.g. between 10 and 60 min) in the semilogarithmic plot. Considering the time-activity curves, it is worth noting that based on the kinetics of the reference compound [ 68 Ga]HBED-CC-Ahx-KuE redistribution from tissues and compartments with unspecific uptake in the early phase seems likely and might explain the discontinuous clearance and increasing tumour uptake at later time points. The uptake of DOTA conjugate 68 Ga-10 in the tumour is continuously increasing over time, but surprisingly without a corresponding decline of activity in muscle or blood (as measured over the heart). In contrast, a slow washout from the target and non-target tissues was observed for 68 Ga-11. Most likely as a result of its metabolic instability, 68 Ga-11 exhibited discontinuous and only moderate clearance kinetics. However, 68 Ga-12 shows typical and nearly ideal in vivo pharmacokinetics, both for the target tissue (tumour and kidney) as well as the non-target tissue muscle and the blood. Equivalent to our DOTAGA ligand 12, the biodistribution of a naphthyl-containing [ 177 Lu]DOTA-PSMA ligand was described recently [40], therefore, we assume that 177 Lu-12 also reveals favourable properties in vivo.

Conclusions
Compared to the metabolically unstable DOTA conjugate 10 and DOTAGA-conjugate 11, 68 Ga-and 177 Lu-12 were found to be rapidly internalized into PSMA expressing cells, to have a favourable pharmacokinetic profile in vivo and only negligible unspecific uptake with no redistribution in vivo. Based on the DOTAGA chelator and the use of a D-amino acid spacer, compound 12 possesses remarkable potential for a PSMA based theranostics concept and is a suitable lead structure for the continuing development of PSMA ligands with further improved affinity.