68Ga, 44Sc and 177Lu-labeled AAZTA5-PSMA-617: synthesis, radiolabeling, stability and cell binding compared to DOTA-PSMA-617 analogues

Background The AAZTA chelator and in particular its bifunctional derivative AAZTA5 was recently investigated to demonstrate unique capabilities to complex diagnostic and therapeutic trivalent radiometals under mild conditions. This study presents a comparison of 68Ga, 44Sc and 177Lu-labeled AAZTA5-PSMA-617 with DOTA-PSMA-617 analogues. We evaluated the radiolabeling characteristics, in vitro stability of the radiolabeled compounds and evaluated their binding affinity and internalization behavior on LNCaP tumor cells in direct comparison to the radiolabeled DOTA-conjugated PSMA-617 analogs. Results AAZTA5 was synthesized in a five-step synthesis and coupled to the PSMA-617 backbone on solid phase. Radiochemical evaluation of AAZTA5-PSMA-617 with 68Ga, 44Sc and 177Lu achieved quantitative radiolabeling of > 99% after less than 5 min at room temperature. Stabilities against human serum, PBS buffer and EDTA and DTPA solutions were analyzed. While there was a small degradation of the 68Ga complex over 2 h in human serum, PBS and EDTA/DTPA, the 44Sc and 177Lu complexes were stable at 2 h and remained stable over 8 h and 1 day. For all three compounds, i.e. [natGa]Ga-AAZTA5-PSMA-617, [natSc]Sc-AAZTA5-PSMA-617 and [natLu]Lu-AAZTA5-PSMA-617, in vitro studies on PSMA-positive LNCaP cells were performed in direct comparison to radiolabeled DOTA-PSMA-617 yielding the corresponding inhibition constants (Ki). Ki values were in the range of 8–31 nM values which correspond with those of [natGa]Ga-DOTA-PSMA-617, [natSc]Sc-DOTA-PSMA-617 and [natLu]Lu-DOTA-PSMA-617, i.e. 5–7 nM, respectively. Internalization studies demonstrated cellular membrane to internalization ratios for the radiolabeled 68Ga, 44Sc and 177Lu-AAZTA5-PSMA-617 tracers (13–20%IA/106 cells) in the same range as the ones of the three radiolabeled DOTA-PSMA-617 tracers (17–20%IA/106 cells) in the same assay. Conclusions The AAZTA5-PSMA-617 structure proved fast and quantitative radiolabeling with all three radiometal complexes at room temperature, excellent stability with 44Sc, very high stability with 177Lu and medium stability with 68Ga in human serum, PBS and EDTA/DTPA solutions. All three AAZTA5-PSMA-617 tracers showed binding affinities and internalization ratios in LNCaP cells comparable with that of radiolabeled DOTA-PSMA-617 analogues. Therefore, the exchange of the chelator DOTA with AAZTA5 within the PSMA-617 binding motif has no negative influence on in vitro LNCaP cell binding characteristics. In combination with the faster and milder radiolabeling features, AAZTA5-PSMA-617 thus demonstrates promising potential for in vivo application for theranostics of prostate cancer.

The two basic questions addressed in the present study are 1. what are the radiolabeling and in vitro stability characteristics for [ 68 Ga]Ga-AAZTA 5 -PSMA-617, [ 44 Sc]Sc-AAZTA 5 -PSMA-617 and [ 177 Lu]Lu-AAZTA 5 -PSMA-617, and 2. how may the exchange of AAZTA for DOTA in the same targeting vector PSMA-617 influence the in vitro binding of the new compounds to prostate cancer cells in vitro.
Concerning complex formation aspects, there is a strong difference in complex geometry and amount of donor sites the chelator has to offer for stable complexation especially between 68 Ga and 44 Sc, two metals from the fourth period, and 177 Lu, as part of the lanthanides in the fifth period. Whereas gallium complexes in general need 6 coordination sites, scandium requires at least 7 coordination and lutetium prefer a coordination with up to 8 donor atoms (Nagy et al. 2017;Parker et al. 2013;Aime et al. 1996), so one might expect different answers for question 1.
Concerning the possible impact of the chelator on the radiopharmacology of the same targeting vector, there have been a number of systematic studies for various targeting vectors indicating, that the "chelator makes a difference" (Fani et al. 2011(Fani et al. , 2012. In the case of AAZTA, we recently reported promising results for radiolabeled AAZTA 5 -TOC directly compared with DOTA-TOC (Sinnes et al. 2019). This encouraged us to investigate whether the exchange of DOTA to AAZTA 5 conjugated to the Glu-urea-Lys binding motif through the 2-naphthyl-L-Ala-AMCH linker of PSMA-617, cf. Fig. 1, may affect the cell binding of the targeting vector or not. For PSMA derivatives, a well evaluated in vitro affinity assay based on LNCaP cells has been recently developed (Benešová et al. 2015). Data have been published already for 68  Lu-labeled DOTA-PSMA-617 (Benešová et al. 2015(Benešová et al. , 2016Umbricht et al. 2017), which can be used for direct comparison to the new AAZTA 5 -PSMA-617 derivatives.

Materials and methods
Synthesis of AAZTA 5 -PSMA (AAZTA 5 -Chx-2-Nal-Lys-Urea-Glu) AAZTA 5 ( t Bu) 4 was synthesized in five steps according to the protocol from Sinnes et al. yielding a ready-for-coupling derivative (Sinnes et al. 2019). The PSMA-617 backbone was synthesized on solid phase following the established procedure from Heidelberg (Benešová et al. 2016). A standard amide coupling using HBTU/HOBt and DIPEA was performed to couple the AAZTA 5 to PSMA on solid phase. AAZTA 5 ( t Bu) 4 (59.0 mg; 0.09 mmol) was mixed with HATU (33.4 mg; 0.09 mmol), HOBt (35.6 mg; 0.26  For radiochemical evaluation with 68 Ga, a 68 Ge/ 68 Ga generator (TiO 2 -based matrix, Cyclotron Co. Obninsk, Russia) was used with online acetone post-processing separating iron and zinc impurities as well as 68 Ge breakthrough (Zhernosekov et al. 2007;Seemann et al. 2015b). Gallium-68 radiolabeling for in vitro studies was performed with a 68 Ge/ 68 Ga generator based on a pyrogallol resin (Schuhmacher and Maier-Borst 1981). Radiolabeling with 44 Sc was performed with a 150 MBq 44 Sc/ 44 Ti generator Pruszyński et al. 2010). 177 Lu, produced via the 176 Yb-pathway (Lebedev et al. 2000), was provided by ITG Munich. Labeling with 68 Ga and 177 Lu were performed in 1 ml of 0.2 M ammonium acetate buffer pH 4.5 at room temperature (25°C). For 44 Sc, due to the general post processing, the radionuclide is provided in 0.25 M ammonium acetate buffer pH 4 (Pruszyński et al. 2010). Kinetic studies were done with 50 MBq for 44 Sc and 100 MBq for 68 Ga and 177 Lu and aliquot were taken at different time points 1, 3, 5 and 10 min. The pH was controlled at start of labeling and after labeling was finished.
For reaction control thin-layer chromatography (TLC) with citrate buffer, pH 7, as eluent and radio HPLC (Merck Chromolith® RP-18e-column, water: MeCN with 0.1% TFA, 5 to 95% MeCN in 10 min) was used. TLC's were measured in RITA TLC imager (Elysia Raytest). The citrate TLC showed free radiometal with a R f of 0.9 and all labeled compounds were located at R f of 0.1 to 0.3. Radio-HPLC was used to characterize the labeled conjugates and to exclude the presence of colloidal radiometals not visible on TLC.

Stability studies
Stability studies were performed in human serum (HS), phosphate buffered saline (PBS) and EDTA/DTPA solution (pH adjusted to 7 by PBS buffer) in triplicate. Only batches obtained with > 95% labeling yield were used and time points were adjusted to the radionuclides' physical half-life: 68 Ga -0.5, 1, 2 h; 44 Sc -0.5, 1, 4, 8, 24 h; 177 Lu − 1, 2, 4, 24 h. HS (human male AB plasma, USA origin) were bought from Sigma Aldrich, PBS was prepared with a BupH™ Phosphate Buffered Saline Pack (PIERCE), EDTA/ DTPA solution were prepared using the prepared PBS buffer and adding ETDA/DTPA to a 0.01 M concentration. Final procedure used 50-70 μl of the labeling solution added to 1 ml of stability solution. Final pH was controlled to ensure no influence of the labeling buffer on the stability solution.

Affinity studies/ cell binding studies
The in vitro experiments were performed using the PSMA-positive LNCaP cell line (androgen-sensitive human lymph node metastatic lesion of prostatic adenocarcinoma, CRL-1740 [American Type Culture Collection]) (Benešová et al. 2015). For negative control, PSMA-PC3 cells (bone metastasis of a grade IV prostatic adenocarcinoma, ATCC CRL-1435) were used (Eder et al. 2012a). The cells were cultured in RPMI1640 medium supplemented with 10% fetal calf serum and L-glutamine and incubated at 37°C in an environment of humidified air containing 5% CO 2 . The cells were harvested using trypsin-ethylenediaminetetraacetic acid (trypsin-EDTA; 0.05% trypsin, 0.02% EDTA, all from PAN Biotech).
Cell binding affinity was determined by competitive cell binding assay (Eder et al. 2012a). 10 5 LNCaP cells per well were incubated with 0.75 nM [ 68 Ga]Ga-PSMA-10 in presence of 12 different concentrations (0-5000 nM) of cold complexes (natural Ga, Sc and Lu) of AAZTA 5 -PSMA-617 in a volume of 100 μL by shaking for 45 min at room temperature and then removed using a multiscreen vacuum manifold (Millipore, Billerica, MA). Afterwards the cells were washed twice with 100 μL and once with 200 μL binding buffer at 5°C. The cell containing filters were stamped out and measured in a gamma counter (Packard Cobra II, GMI, Minnesota, USA). Using a nonlinear regression algorithm (GraphPad Prism Software) the 50% inhibitory concentration (IC 50 ) values were calculated. Each sample was done in quadruple while the whole experiment was done three times.
Nonradioactive complexes of the AAZTA 5 -PSMA-617 with nat Ga, nat Sc and nat Lu were synthesized by adding 15 μL (150 nmol  All AAZTA 5 -PSMA-617 complexes were tested 3 times in triplicate leading to n = 9. Same was done with the PSMA-617 complexes on the same cell-plates.

Internalization studies
For internalization studies 10 5 LNCaP or PC3 cells were seeded in poly(L-lysine)-coated 24-well cell culture plates at 37°C in an environment of humidified air containing 5% CO 2 for 24 h. (Mier et al. 2007). The medium is removed and 250 μl of radiolabeled 30 nM AAZTA 5 -PSMA is replaced for 45 min. One plate is being incubated at 37°C and the second one at 4°C to inhibit the internalization. The specificity of the ligands is proofed by addition 500 μM of 2-(phosphonomethyl)-pentanedioic acid (2-PMPA, Axxora, Loerrach, Germany). After incubation, the cells were washed three times with 1 ml ice cold PBS. To determine the surface-bound activity cells were incubated twice with 0.5 ml of glycine-HCl in PBS (50 mM, pH 2.8) each for 5 min at room temperature.
Both washing steps were collected for measuring with a gamma counter (Packard Cobra II, GMI, Minnesota, USA). Before lysating the cells with 0.5 mL of 0.3 M NaOH to determine the internalized fraction, they were once washed with 1 ml ice cold PBS (Eder et al. 2012a).

Radiolabeling and stability studies
AAZTA 5 -PSMA-617 (AAZTA 5 -Chx-2-NaI-Lys-Urea-Glu) was successfully radiolabeled in quantitative radiochemical yields (> 99%, as determined both by radio-HPLC and radio-TLC) with 68 Ga, 44 Sc and 177 Lu, in less than 5 min at room temperature. Precursor amounts were optimized to 5 nmol (5.4 μg) for 68 Ga and 44 Sc and 0.6 nmol (0.65 μg) for 177 Lu (molar ratio 10: 1 for chelator to radiometal). Final concentrations for the internalization studies of 6 μM (6 nmol/ml) were easily reached and final radiochemical purities of 99.9% (as determined by radio-HPLC) allowing direct use of the obtained product solution for the in vitro assay without further purification.

In vitro binding affinity and internalization studies
In competitive binding studies against [ 68 Ga]Ga-PSMA-10, all three non-radioactive metal complexes of AAZTA 5 -PSMA-617 indicated nanomolar binding affinities: 8.7 ± 0.8 nM for the gallium complex, 30.6 ± 11.5 nM for the scandium complex and 26.6 ± 11.1 nM for the lutetium complex. By applying the Cheng-Prusov equation on the IC 50 values the following inhibition constants were obtained (Table 1) (Craig 1993).

Radiochemistry
Quantitative radiolabeling with 68 Ga, 44 Sc and 177 Lu was achieved at room temperature for AAZTA 5 -PSMA-617 within less than 5 min at room temperature. This corresponds with literature data for the free chelator (Nagy et al. 2017) as well as with recent evaluations of AAZTA 5 -TOC (Sinnes et al. 2019).

Binding affinity
The binding affinities for all three non-radioactive metal complexes of AAZTA 5 -PSMA-617 are in the low nM range. [ nat Ga]Ga-AAZTA 5 -PSMA-617 reached the same K i value range of 8-9 nM as the PSMA-617 analogue, while proving a higher binding affinity as [ nat Ga]Ga-PSMA-11.
Both [ nat Sc]Sc-AAZTA 5 -PSMA-617 and [ nat Lu]Lu-AAZTA 5 -PSMA-617 displayed a somewhat lower affinity of 31 and 27 nM, respectively. The most noticeable difference, beside AAZTA 5 being a hybrid chelator, i.e. shows both cyclic and acyclic features, what mostly effects the radiochemistry, is the charge of − 1 for the AAZTA 5 scandium and lutetium complexes. For DOTA, the necessary donor set consists of N 4 O 2 to N 4 O 4 , and at least one of the acid groups is part of the bifunctionalization to the targeting molecule. As a M 3+ radiometal DOTA is forming a neutral complex, whereas AAZTA 5 with a N 3 O 4 core forms an overall negative charged complex (net charge − 1). This negative charge is created by the 4 acid groups that need to be deprotected for complexation and binding to the radiometal. The four negative charges of the acid groups overcompensate for the three positive charges of the radiometal. The complex of the gallium can be neutral because gallium may not require all donor atoms from the AAZTA 5 since gallium prefers an octahedral coordination sphere leaving one acid group empty. The negative charge definitely may have an influence on the in vitro as well as the in vivo behavior of the targeting molecule. Besides however, both chelator systems provide a suitable scaffold for a more efficient locally arrangement of the donor atoms (nitrogen and oxygens atoms) to achieve metal coordination. A relevant influence on the host of the radiometal inside the chelator ring arises the entropic effect. The relatively high SD for both affinity and internalization in case of the scandium and the lutetium complexes are justified by 9 total measurements over 9 experimental days (SD on each day ±1-2), whereas the gallium complex was tested twice on the same day. With each measurement being quadruple, total repetition was n = 8 for the gallium complex and n = 36 for the scandium and the lutetium complex, respectively.

Internalization
All AAZTA 5 -PSMA complexes internalize well and this was demonstrated to be a specific and active process. The internalization value for [ 68 Ga]Ga-AAZTA 5 -PSMA-617 is in the same range as literature values of [ 68 Ga]Ga-PSMA-617, with both reaching internalization ratios around 45%. The [ 44 Sc]Sc-AAZTA 5 -PSMA reached ratios of 41% internalized activity while [ 44 Sc]Sc-PSMA-617 showed up to 50% on the same day, while literature also giving around 41% internalized activity. For [ 177 Lu]Lu-AAZTA 5 -PSMA-617 ratios of 36% internalized activity were measured, which are higher than the 27% measured for [ 177 Lu]Lu-PSMA-617. Literature values for [ 177 Lu]Lu-PSMA-617 of 10-15% are much lower than the measured values in this assay.
Surface binding obtained for the radiolabeled AAZTA 5 -PSMA-617 complexes could be blocked by 2-PMPA. Similarly, [ 44 Sc]Sc-AAZTA 5 and [ 177 Lu]Lu-AAZTA 5 showed no surface activity or internalization on PSMA positive LNCaP cells and on PC3 cells as a second negative experiment.

Conclusion
Radiolabeled AAZTA 5 -PSMA-617 conjugates guarantee for almost quantitative yields of > 99% of 44 Sc-, 68 Ga-and 177 Lu-labeling under mild conditions in short time after less than 5 min at room temperature. Subsequent purification is obsolete for in vitro