PSMA-D4 Radioligand for Targeted Therapy of Prostate Cancer: Synthesis, Characteristics and Preliminary Assessment of Biological Properties

A new PSMA ligand (PSMA-D4) containing the Glu-CO-Lys pharmacophore connected with a new linker system (L-Trp-4-Amc) and chelator DOTA was developed for radiolabeling with therapeutic radionuclides. Herein we describe the synthesis, radiolabeling, and preliminary biological evaluation of the novel PSMA-D4 ligand. Synthesized PSMA-D4 was characterized using TOF-ESI-MS, NMR, and HPLC methods. The novel compound was subject to molecular modeling with GCP-II to compare its binding mode to analogous reference compounds. The radiolabeling efficiency of PSMA-D4 with 177Lu, 90Y, 47Sc, and 225Ac was chromatographically tested. In vitro studies were carried out in PSMA-positive LNCaP tumor cells membranes. The ex vivo tissue distribution profile of the radioligands and Cerenkov luminescence imaging (CLI) was studied in LNCaP tumor-bearing mice. PSMA-D4 was synthesized in 24% yield and purity >97%. The radio complexes were obtained with high yields (>97%) and molar activity ranging from 0.11 to 17.2 GBq mcmol−1, depending on the radionuclide. In vitro assays confirmed high specific binding and affinity for all radiocomplexes. Biodistribution and imaging studies revealed high accumulation in LNCaP tumor xenografts and rapid clearance of radiocomplexes from blood and non-target tissues. These render PSMA-D4 a promising ligand for targeted therapy of prostate cancer (PCa) metastases.


Introduction
Once metastasized, prostate cancer (PCa) becomes one of the most aggressive cancer types. It is the second most common cancer in men and the fifth most common cause of cancer death among men [1,2]. The standard treatment of PCa is based on radical prostatectomy, external beam radiation therapy, or brachytherapy, chemotherapy, and hormonotherapy. Unfortunately, these therapies are often followed by the formation of metastatic castration-resistant prostate cancer (mCRPC) [3].
The PSMA (prostate-specific membrane antigen) known as glutamate carboxypeptidase II (GCP-II) is a transmembrane, 750 amino acid, type II glycoprotein, which is expressed virtually by almost all primary prostate cancer (PCa) and metastatic disease as well [4]. PSMA is highly homologous to N-acetylated R-linked acidic dipeptidase, a neuropeptidase that produces the neurotransmitter glutamate and N-acetylaspartate Trp) in PSMA-T4, as one of the linkers, instead of naphthylalanine (L-2NaI), has led to a significant improvement in the biodistribution of the labeled molecule (reducing kidney accumulation) and its increased affinity to PSMA in vivo [28].
Bearing in mind the high affinity of the developed [ 99m Tc]Tc-PSMA-T4 for PSMA in vivo and its confirmed high diagnostic efficacy in patients with PCa metastases, we decided to use this new system, containing two linkers (L-Trp-4-Amc) connected with the Glu-CO-Lys pharmacophore and HYNIC for 99m Tc chelation, to develop a new ligand intended for radiolabeling with radionuclides for therapy (i.e., 177 Lu, 90 Y, 47 Sc, and 225 Ac). For this purpose, we replaced HYNIC with the DOTA as BFC.
This work aims to present synthesis, physicochemical characteristics, and preliminary biological evaluation of a new carrier, PSMA-D4, as a potential candidate for targeted radiotherapy of PCa metastases.

Synthesis and Physico-chemical Characteristics of PSMA-D4
PSMA-D4 (Glu-CO-Lys-L-Trp-4-Amc-DOTA, Figure 1) in the form of TFA salt was synthesized following the protocol described herein (Appendix A) and obtained in 24% yield and purity greater than 97%. The identity of PSMA-D4 was confirmed by TOF-ESI-MS and 1 H, 13

Molecular Modeling
To understand the structural basis of the observed affinities, compounds PSMA-617, iPSMA-HYNIC (HYNIC-Lys(Nal)-CO-Glu), PSMA-D4, and PSMA-T4 were modeled in complex with glutamate carboxypeptidase II. Initial guesses of the complexes' structures were built by manual replacement of appropriate elements in the experimental structure of PSMA-1007 with GCPII (PDB accession code: 5O5T [31]). The complexes were then subjected to optimization by local search procedure in AutoDock 4.2.6 [32].
According to the results of this procedure (general overview in Figure 2a), modifications in either the linker or the chelating element do not influence the contacts of the Glu-ureido pharmacophoric fragment common to all studied compounds and the crystallographic ligand. Thus, Glu-ureido fragment is predicted to be located deep in the GCPII active site (Figure 2b), forming contacts with Arg210, Tyr552, Tyr700, (αcarboxylate of P1' Glu), Asn257, Lys699 (γ-carboxylate of P1'), Glu424, Gly518 (ureido The identity of PSMA-D4 was confirmed by TOF-ESI-MS and 1 H, 13

Molecular Modeling
To understand the structural basis of the observed affinities, compounds PSMA-617, iPSMA-HYNIC (HYNIC-Lys(Nal)-CO-Glu), PSMA-D4, and PSMA-T4 were modeled in complex with glutamate carboxypeptidase II. Initial guesses of the complexes' structures were built by manual replacement of appropriate elements in the experimental structure of PSMA-1007 with GCPII (PDB accession code: 5O5T [31]). The complexes were then subjected to optimization by local search procedure in AutoDock 4.2.6 [32].
superposes closely on the position of the napthyl ring in L-2-Nal-bearing analogues (Figure 3b). The 4-aminocyclohexanoic acid fragment leads up the entrance funnel (its position is slightly displaced compared to the crystallographic position of the phenyl ring of PSMA-1007 in 5O5T) to enable the location of DOTA or HYNIC fragments against the residues of helix α15, strand β16, as well as the loop connecting α11 and α12. The chelators are predicted to be involved in several interactions (In Appendix C Figures A22 and A23), but it is to be noted that these interactions might be different in the presence of a chelated cation.
Given the flexibility of both the linker-chelator part of the studied conjugates and the protein by the entrance lid [31], it would be beneficial to inspect the complexes of GCPII with the presented compounds by the use of molecular dynamics. scheme of the interactions between Glu-ureido fragment and the internal pocket of the enzyme. In part (a), the protein is represented as a surface in a simplified manner, and the compounds are shown as colored sticks. Green-PSMA-1007 as found in 5O5T structure [30], light-blue-PSMA-617, yellow-PSMA-T4, and magenta-PSMA-D4.  (b) scheme of the interactions between Glu-ureido fragment and the internal pocket of the enzyme. In part (a), the protein is represented as a surface in a simplified manner, and the compounds are shown as colored sticks. Green-PSMA-1007 as found in 5O5T structure [30], light-blue-PSMA-617, yellow-PSMA-T4, and magenta-PSMA-D4.
Regarding the linker part, the naphtyl ring of the alanylnaphtyl residue (in PSMA-617, iPSMA-HYNIC) is predicted to be wedged between the P1 lysine aliphatic chain, Gly548, and Tyr552 (Figure 3a). If the L-2-Nal is exchanged for L-Trp (PSMA-D4, PSMA-T4), docking suggests no significant change in the binding mode. The L-Trp indole ring superposes closely on the position of the napthyl ring in L-2-Nal-bearing analogues (Figure 3b). The 4-aminocyclohexanoic acid fragment leads up the entrance funnel (its position is slightly displaced compared to the crystallographic position of the phenyl ring of PSMA-1007 in 5O5T) to enable the location of DOTA or HYNIC fragments against the residues of helix α15, strand β16, as well as the loop connecting α11 and α12. The chelators are predicted to be involved in several interactions (In Appendix C Figures A22 and A23), but it is to be noted that these interactions might be different in the presence of a chelated cation.  [31] or those described for other Glu-ureido GCPII binders [33].
Regarding the linker part, the naphtyl ring of the alanylnaphtyl residue (in PSMA-617, iPSMA-HYNIC) is predicted to be wedged between the P1 lysine aliphatic chain, Gly548, and Tyr552 (Figure 3a). If the L-2-Nal is exchanged for L-Trp (PSMA-D4, PSMA-T4), docking suggests no significant change in the binding mode. The L-Trp indole ring superposes closely on the position of the napthyl ring in L-2-Nal-bearing analogues (Figure 3b). The 4-aminocyclohexanoic acid fragment leads up the entrance funnel (its position is slightly displaced compared to the crystallographic position of the phenyl ring of PSMA-1007 in 5O5T) to enable the location of DOTA or HYNIC fragments against the residues of helix α15, strand β16, as well as the loop connecting α11 and α12. The chelators are predicted to be involved in several interactions (In Appendix C Figures A22 and A23), but it is to be noted that these interactions might be different in the presence of a chelated cation.
Given the flexibility of both the linker-chelator part of the studied conjugates and the protein by the entrance lid [31], it would be beneficial to inspect the complexes of GCPII with the presented compounds by the use of molecular dynamics. scheme of the interactions between Glu-ureido fragment and the internal pocket of the enzyme. In part (a), the protein is represented as a surface in a simplified manner, and the compounds are shown as colored sticks. Green-PSMA-1007 as found in 5O5T structure [30], light-blue-PSMA-617, yellow-PSMA-T4, and magenta-PSMA-D4.  Given the flexibility of both the linker-chelator part of the studied conjugates and the protein by the entrance lid [31], it would be beneficial to inspect the complexes of GCPII with the presented compounds by the use of molecular dynamics.

Radiolabeling of PSMA-D4 and logD
The radiochemical yield (RCY) was always >97% for all radiolabeled compounds and resulted in different specific activity (SA), (Table 1). In case of radiolabeling with 225 Ac for both used buffers, comparable results of radiochemical purity were observed. The partition coefficients between n-octanol and PBS (logD) were determined for PSMA-D4 labeled with 177 Lu and 90 Y using a shake-flask method. They were compared with the logD values measured under the same experimental conditions for the analogous radiocomplexes of PSMA-617, PSMAI&T labeled with 177 Lu, and PSMA-11 labeled with 68 Ga, Table 2. All complexes are highly hydrophilic. The logD values were in the same range for PSMA-D4 and PSMA-617 radiocomplexes, although lower than [ 68 Ga]Ga-PSMA-11 (containing the lipophilic HBED-CC chelator) and [ 177 Lu]Lu-PSMAI&T.

In Vitro
The competitive binding assay was performed to determine the half-maximal inhibitory concentration (IC 50 ) of PSMA inhibitors using cell membranes isolated from lymph node carcinoma of the prostate (LNCaP) cells and [ 177 Lu]Lu-PSMA-617 as a radioligand. The IC 50 values for PSMA-D4 were found to be 28.7 ± 5.2, pointing at their high affinity to PSMA antigen, while PSMA-I&T and PSMA-11 showed significantly lower affinity, and their IC 50 values were 61.1 ± 7.8 and 84.5 ± 26.5, respectively (Table 3).  Specificity for PSMA antigen was assessed on LNCaP and human prostate cancer (PC3) cell-lines membranes. Investigated PSMA compounds bound only to the PSMA expressing LNCaP cells, and the specific binding of all tested PSMA-D4 radiocomplexes was within the range of 98.8-99.9% (Appendix D).

Ex Vivo
In that study, ex vivo experiments with [ 90 Y]Y-PSMA-D4 and [ 47 Sc]Sc-PSMA-D4 were performed in tumor-bearing mice with LNCaP cell-line. Since no binding was observed in in vitro studies with PC-3 cell line, the in vivo studies in mice bearing PC-3 tumors were not performed. Side-by-side comparison of ex vivo distribution is shown in Table 5. Significant differences were observed in the kidneys and tumors. The radiocomplexes were similarly distributed in non-target tissues. Values are expressed as %ID g −1 and presented as mean ± standard deviation. Ratio: T/B-tumor to blood; T/M-tumor to muscle; T/K-tumor to kidneys.

of 35
The accumulation of [ 90 Y]Y-PSMA-D4 in the tumor was two-fold higher than that of [ 47 Sc]Sc-PSMA-D4 at 24 and 48 h after injection, which could be explained by the difference in molar activity of injected complexes (17.2 and 4.57 GBq mcmol −1 , respectively). The increased tumor uptake led to a high tumor-to-organ ratio. Both radioligands were rapidly cleared from blood circulation by kidneys.
The blood pharmacokinetics of [ 47 Sc]Sc-PSMA-D4 as a one-phase decay model are shown in Figure 4. Since the complex is rapidly excreted by the kidney, the excretion rate is proportional to the blood plasma concentration. Values are expressed as %ID g −1 and presented as mean ± standard deviation. Ratio: T/B-tumor to blood; T/M-tumor to muscle; T/K-tumor to kidneys.
The blood pharmacokinetics of [ 47 Sc]Sc-PSMA-D4 as a one-phase decay model are shown in Figure 4. Since the complex is rapidly excreted by the kidney, the excretion rate is proportional to the blood plasma concentration.    Due to the rapid excretion of [ 90 Y]Y-PSMA-D4 with urine, radioactivity accumulation in the urinary bladder was not observed. Moreover, in vivo, CLI revealed excellent tumor visualization and clearance of radiolabeled compound from kidneys and the whole body. These observations suggest that the linker system containing L-Trp favorably modifies pharmacokinetics and affinity of the designed ligand for PSMA positive tumors. Rapid Due to the rapid excretion of [ 90 Y]Y-PSMA-D4 with urine, radioactivity accumulation in the urinary bladder was not observed. Moreover, in vivo, CLI revealed excellent tumor visualization and clearance of radiolabeled compound from kidneys and the whole body. These observations suggest that the linker system containing L-Trp favorably modifies pharmacokinetics and affinity of the designed ligand for PSMA positive tumors. Rapid clearance of radioactivity from the kidney may reduce toxicity towards this critical organ.

In Vivo
The tumor uptake, calculated as %ID (per cent of injected dose) from in vivo studies, was comparable to that from ex vivo biodistribution studies (Figure 5f). Uptake measurements were done considering regions of interest (ROI's) corresponding to the tumor and the 0.1 mL of [ 90 Y]Y-PSMA-D4 as a standard (injected dose), placed in a black 24-well plate filled with gelatin solution. To refine the calculation of %ID, measurements considering background were done. The background was defined as regions near the heart and well without tracer.

Discussion
Radionuclide therapy with specific ligands targeting the PSMA inhibitor is a promising therapeutic strategy for patients with mCRPC. Several PSMA ligands and radionuclides have been developed and selected for this purpose. Among them, PSMA-617 and PSMA-I&T labeled with 177 Lu are the most common in clinical use. The utility significance of 225 Ac [35] and 47 Sc is also growing. PSMA therapies have strong potential for a subset of patients and have reached late stage trials; however, the exact patient populations and degree of health benefits in a greater population remain to be determined. On the other hand, despite the high efficacy of the current radioligands used for this purpose, there is always a need to improve treatment outcomes and minimize side effects, including using newer agents with more favorable therapeutic properties.
Bearing in mind the importance of each element of a potential isotope carrier molecule, including the linker connecting the bioactive part of the drug to the radionuclide [27] in the molecular targeting of the molecule to the binding site, we have developed a new ligand for targeted radiotherapy of mCRPC. From our previous studies on the PSMA ligand for 99m Tc labeling [28], we found that the presence of the L-Trp fragment in the linker system significantly reduces the accumulation of the radioactive product in the kidney, which in the case of a therapeutic formulation is of great importance due to the reduction of radiotoxicity towards this organ. Replacement of HYNIC in the developed PSMA-T4 molecule with DOTA (PSMA-D4) did not result in the loss of this property but reduced the renal affinity while maintaining a high affinity for the PSMA inhibitor.
In silico molecular modeling and docking studies of the new PSMA-D4 molecule to GCPII showed the high similarity of its binding mode to that of PSMA-617. When comparing to published data on PSMA-617 [36], the in vitro properties of PSMA-D4 are largely the same. These observations included n-octanol/PBS distribution ratio and K D and internalization in PSMA-positive and PSMA-negative cancer cells. In vitro studies also confirmed promising properties of a PSMA-D4 when labeled with therapeutic radionuclides.
Ex vivo studies demonstrated that [ 90 Y]Y-PSMA-D4 and [ 47 Sc]Sc-PSMA-D4 accumulated favorably in LNCaP tumor-bearing mice immediately after injection and rapid elimination with the urine. Key findings from these observations include low accumulation in the kidney, relatively fast clearance of radioactivity from blood and non-target tissues, and a high tumor-to-background ratio. These circumstances enable Cerenkov luminescence imaging, which confirmed the tumor-to-background contrast is increasing over time.
Radioactivity uptake in selected organs of investigated PSMA-ligands of LNCaP and PC-3 PIP (PSMA positive) tumor bearing mice is summarized in Table 6. The summary of radioligand accumulation in PSMA-positive tumors and critical organs in Table 6 allows the conclusion that labeled PSMA-D4 shows favorable affinity properties in vivo. However, it is difficult to make a quantitative assessment in comparison to other radioligands due to the variability as to the used experimental models, as well as due to the administered masses of the preparations.
Lutetium-177 (LutaPol) as lutetium chloride of SA higher than 555 MBq mg −1 Lu in 0.04 N HCl and yttrium-90 (ItraPol) as a yttrium chloride in a 0.04-0.05 N HCl of 0.925-37 GBq in a volume 0.010-2 mL were produced at Radioisotope Centre POLATOM, (Otwock, Poland). Scandium-47 was produced via the nuclear reaction 46 Ca(n,γ) 47 Ca→ 47 Sc, by irradiation of enriched 46 Ca target at the Institute Laue-Langevin (Grenoble, France) or Maria research reactor (Otwock, Poland), and 47 Sc was separated from the irradiated target by extraction chromatography on DGA resin [41]. Actinium-225 as actinium nitrate was purchased from the Institute of Physics and Power Engineering (Obninsk, Russia). Gallium-68 as a gallium chloride solution in 0.1 M HCl was obtained from the GalliaPharm ® (Ge-68/Ga-68 generator manufactured by Eckert & Ziegler (Dresden, Germany).

Synthesis of PSMA-D4
The PSMA-D4 synthesis was performed by standard solid-phase synthesis from the Fmoc-L-Lys(Alloc) attachment to the Wang polystyrene resin. Preparation of Glu(tBu)urea-Lys-NH 2 was carried according to Scheme 1.
Poland). Scandium-47 was produced via the nuclear reaction 46 Ca(n,γ) 47 Ca→ 47 Sc, by irradiation of enriched 46 Ca target at the Institute Laue-Langevin (Grenoble, France) or Maria research reactor (Otwock, Poland), and 47 Sc was separated from the irradiated target by extraction chromatography on DGA resin [41]. Actinium-225 as actinium nitrate was purchased from the Institute of Physics and Power Engineering (Obninsk, Russia). Gallium-68 as a gallium chloride solution in 0.1M HCl was obtained from the GalliaPharm ® (Ge-68/Ga-68 generator manufactured by Eckert & Ziegler (Dresden, Germany).

Synthesis of PSMA-D4
The PSMA-D4 synthesis was performed by standard solid-phase synthesis from the Fmoc-L-Lys(Alloc) attachment to the Wang polystyrene resin. Preparation of Glu(tBu)urea-Lys-NH2 was carried according to Scheme 1. Scheme 1. Solid-phase synthesis of the Glu(tBu)-urea-Lys-NH2 moiety.
Further steps of synthesis carried on solid support were performed in an automatic peptide synthesizer LibertyBlue equipped with Discover microwave oven (CEM) [42,43] according to Scheme 2. Further steps of synthesis carried on solid support were performed in an automatic peptide synthesizer LibertyBlue equipped with Discover microwave oven (CEM) [42,43] according to Scheme 2. The product was detached from the resin and deprotected, based on the publication of Wängler et al. [45].

HPLC
Analytical HPLC was performed using a Shimadzu system consisting of LC-20AD pump, SPD-M20A diode array detector (DAD), CBM-20A controller, and the LC Solution software (Shimadzu Europa GmbH, Duisburg, Germany).
The product was detached from the resin and deprotected, based on the publication of Wängler et al. [45].
The raw PSMA-D4 (Glu-CO-Lys-LTrp-4Amc-DOTA) was purified on preparative HPLC on a reversed-phase column, resulting in >98% purity peptide. The identity of PSMA-D4 was confirmed by the mass spectrometry method with the use of ESI-IT-TOF detector. The peptide sample was dissolved in a mixture of AcN/water and, after injection, was ionized by the electro spray method. The molecular mass of the sample was determined in a positive mode.
The structure of the PSMA-D4 was determined by interpretation of the one-dimensional 1 H, 13 C, and DEPT-135 spectra, two-dimensional homonuclear COSY, TOCSY, and ROESY and heteronuclear 1 H-13C HSQC, HSQC-TOCSY, and HMBC NMR spectra. Proton connectivities were derived from COSY, TOCSY, and ROESY spectra. The 13 C resonances corresponding to carbons with directly attached protons were assigned using HSQC and HSQC-TOCSY spectra. HMBC spectra were used to assign resonances of the quaternary carbons and to validate the connectivities established by the other spectra. Results of both 1 H-15 N correlations (HSQC and HMBC) were used to confirm nitrogen atoms' character in the compound studied.
The details of the NMR analysis are presented in Appendix B.

Elemental Analysis
The percentage content of C, H, and N was analyzed in an automatic UNIcube analyzer. The measurement's basis was catalytic combustion of the analyzed substance in a special pipe, with oxygenation, at 1150 • C.
Combustion gas was separated from each other on adsorption columns and determined successively using a thermal conductivity detector (TCD).

Lipophilicity Determination
The partition coefficient logD of the radioligands was determined by the shake-flask method [46,47]. The solution of 10 to 50 MBq of radiolabeled PSMA ligands (D-4, 617, 11, I&T) in 1.0 mL of phosphate-buffered saline (PBS, pH 7.4) was added to 1.0 mL of presaturated n-octanol solution (n = 3). Vials were shaken vigorously for 10 min. To achieve quantitative phase separation, the vials were centrifuged at 1600 rpm for 5 min. The radioactivity concentration in a defined volume of both the aqueous and the organic phase in six replicates each was measured in a γ-counter (Wizard 1470, Wallac, Turku, Finland). The partition coefficient was calculated as the logarithm of the ratio between counts per minute (cpm) measured in the n-octanol phase to the PBS phase counts.
The starting point was the crystallographic structure of GCP-II with PSMA-1007 (PDB accession code: 5O5T [31]). In the first stage, the vector-linker substructure of the considered compounds was built by removing and replacing appropriate elements from PSMA-1007 (in the linker part). The linker was capped with the acetyl group. The complexes were subject to local search docking with AutoDock 4.2.6 [32] with the following parameters: 150 individuals in a population, 300 iterations of the Solis-Wets local search, local search space set to 30.0, and 100 local search runs. The results were clustered, and the representative pose of the best scored cluster was taken for further modeling. In this stage, structures of the chelators were built into the complexes.
In the case of DOTA structure, the coordinates of the chelator were taken from the NOJYIU entry [48] of the Cambridge Structural Database (DOTA complex with diaqualutetium(III)-sodium trihydrate) in order to maintain the conformation found with a chelated metal present. Then, vector-linker-chelator position was optimized again using local search docking with AutoDock 4.2.6 [32] with the following parameters: 150 individuals in a population, 300 iterations of the Solis-Wets local search, local search space set to 45.0, and 500 local search runs. The procedure was repeated several times to check convergence of the results. Additionally, the DOTA fragment was modeled with the flexibility of the carboxylate arms enabled or not, and with protonated and unprotonated state of this group, but upon finding no major differences in the poses, only the results of dockings with unprotonated, flexible carboxylates are discussed.
The receptor structure was prepared in AutoDockTools [32]. The box was set around the experimental position of the crystallographic ligand and extended. The grids were calculated with AutoGrid 4. Molecular graphics were prepared in the Discovery Studio Visualizer [49] and PyMOL [50]. Radiolabeling with 68 Ga was carried out by adding 5 mL of 68 Ga eluate (600-800 MBq) to a vial containing 40 mcg of PSMA-11 and 60 mcg sodium acetate. The mixture was incubated at 95 • C for 15 min.

Radiolabeling and QC
The RCY of the final formulation was determined by thin-layer chromatography on silica-gel plates (ITLC SG) with 0.1 M citric buffer pH 5 as a mobile phase to differentiate between the free and PSMA-D4 bound radionuclide. The radiolabeling yield was evaluated in a competitor's presence (10 mM DTPA) in excess, which reacts with the non-incorporated radionuclide. In the case of [ 225 Ac]Ac-PSMA-D4, after development, the chromatography strips were stored for at least 45 min for the decay of unbound 221 Fr and to reach radiochemical equilibrium between 225 Ac and its daughter nuclide 221 Fr. Subsequently, radiochemical purity was determined by measuring the activity of the γ emission of 221 Fr. Due to a high RCY of radio conjugates, no additional purification step was necessary.

In Vitro
The LNCaP cell line used for in vitro and in vivo experiments was purchased from American Type Culture Collection (ATTC) and was maintained as per ATCC guidelines. As a control in in vitro studies, the androgen-independent human prostate cancer cells, PC3 (derived from bone metastasis), obtained from National Institute of Medicines (NIL), were used. PC3 cells were cultured in RPMI 1640 medium (IITD PAN Wroclaw, Poland) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% of penicillin/streptomycin (Gibco). All cells were grown to 80-90% confluence before trypsinization.
To The half-maximal inhibitory concentration (IC 50 ) of PSMA-D4 was determined in a competitive binding assay carried out on 96-well MuliScreen filter plates using LNCaP cell membranes (50 mcL) incubated with [ 177 Lu]Lu-PSMA-617 as a radioligand and ten different concentrations (2.4-9708 nM) of PSMA-D4. After 2 h incubation, plates were treated in the same way as in the saturation binding assay. The amount of bound radioactivity was measured in γ-counter. IC 50 value was evaluated using GraphPad Prism software (Darmstadt, Germany).

Animal Study
This study was conducted following the guidelines approved on 4 July 2018, by the first Local Animal Ethics Committee in Warsaw, Poland (the authorization number 681/2018), and was carried out in accordance with the national legislation regarding laboratory animals' protection and the principles of good laboratory practice.
Fifty males, 5-6 weeks old BALB/c NUDE mice, were obtained from Janvier Lab., France. Mice were housed under pathogen-free conditions with food and water ad libitum, and a 12-12 h light-dark cycle. Veterinarian staff and investigators observed the mice daily to ensure animal welfare and determine if humane endpoints (e.g., hunched, ruffled appearance, apathy, ulceration, severe weight loss, and tumor burden) were reached.
An experimental tumor murine model was induced using LNCaP cell, which grew to 80-90% confluence before trypsinization and formulation in Matrixgel™ Basement Membrane Matrix (Bedford, MA, USA) for implantation into mice. The mice were subcutaneously injected in the shoulder with 200 mcL bolus containing a suspension of 10 6 freshly harvested cell line LNCaP in Matrixgel™. This procedure was performed under anaesthesia with 2% isoflurane. The animals were kept under pathogen-free condition, and experiments were performed 2-3 weeks later when a tumor reached a volume of approximately 150 ± 60 mm 3 . Then, the animals were randomized into two groups ( At 1, 2, 4, 6, 24, and 48 h after injection, the animals were euthanized by cervical dislocation and dissected. Selected organs and tissues were assayed for their radioactivity and weighted. The physiological distribution was calculated and expressed in terms of the percentage of administrated radioactivity found in each of the selected organs or tissues per gram (%ID g −1 ).
The pharmacokinetic analysis of the [ 47 Sc]Sc-PSMA-D4 was based on the curves of the %ID g −1 accumulated in the blood. The pharmacokinetic parameters were determined according to a one-phase decay model. The following parameters were used: K (rate constant) and half-life calculated as ln(2)/K.

Optical Imaging of [ 90 Y]Y-PSMA-D4
The PhotonIMAGER Optima system (Biospace Lab, Nesles-la-Vallée, France) was used for the non-invasive detection, localization, and quantification of yttrium-90 signal from live animals based on Cerenkov luminescence imaging. It was feasible due to high-energy β − particle emission produces continuous spectrum light photons or Cerenkov radiation. Optical imaging of [ 90 Y]Y-PSMA-D4 in LNCaP cell grafted mice was performed contemporaneously with ex vivo studies. The animals were placed individually in an induction chamber, where anaesthesia was induced with 5% isoflurane (Iso-Vet, Piramal Healthcare UK Limited, West Drayton, UK) in 100% oxygen with a delivery rate of 2 L min −1 until loss of righting reflex. After induction, the animals were moved to the optical chamber. Anaesthesia was then maintained with 1.5-2% isoflurane in 100% oxygen with a flow of 1.5 L min −1 administered through a facemask connected to a coaxial circuit. Body temperature was maintained at 37 • C by a heating table inside the chamber. At recovery, all animals received 100% oxygen until recovery of righting reflex. No mice were restrained during anaesthesia.

Statistics
Results are provided as mean ± SD. The results of physiological distribution are presented as a percentage of the dose administered per gram of tissue (%ID g −1 ) as average value with standard deviation (%ID g −1 ; mean ± standard deviation (SD)) with n representing the number of animals per group. Data were statistically analyzed using GraphPad Prism version 8.0.0 for Window.
For blood activity data, a one-phase exponential decay model was used to model the percentage of remaining activity (%ID g −1 ) as a time post-injection function.
Two-tailed, unpaired Student's t-tests assessed results of normal distribution. Otherwise, outcomes were evaluated by a Mann-Whitney U-test. A one-way ANOVA test analyzed the differences between IC 50 and internalization result. A p-value of < 0.05 with two-tailed testing was considered statistically significant.

Conclusions
In this work, we present a novel PSMA-D4 ligand (Glu-CO-Lys-L-Trp-4-Amc-DOTA), which showed excellent radiolabeling characteristics, high selectivity towards PSMA receptors in vitro, and favorable tumor accumulation in LNCaP tumor-bearing mice. These features render PSMA-D4 a promising ligand for targeted radionuclide therapy of prostate cancer, enriching the state-of-the-art and paving the way to its further development for clinical use. Therefore, extended preclinical studies are planned to determine the pharmacokinetics and toxicity and to confirm the specificity of radio-labeled PSMA-D4 in vivo.

Synthesis of PSMA-D4
The PSMA-D4 synthesis was started from the preparation of Glu(tBu)-urea-Lys-NH 2 carried out on solid phase Wang polystyrene resin, crosslinked with divinyl-benzene, which is widely used in peptide synthesis. Fmoc-L-Lys(Alloc) was attached to hydroxyl groups in the support in the manner described previously [52].
After the reaction had been completed, the support was washed with N,N-dime thylformamide, 50% N,N-dimethylformamide solution in dichloromethane, and dichloromethane and dried in vacuum.
The amino acid loading was measured using a method described by Eissler et al., [53]. The Fmoc protection group was removed in the manner usually used in peptide synthesis [52].
A double molar excess of triphosgene in dry dichloromethane was cooled in a dry-ice bath to ≤−50 • C. The solution of L-Glu(tBu)OtBu*HCl 0.75% N,N-diisopropylethylamine in dry dichloromethane was prepared separately using sixfold molar excess. The solution was slowly added dropwise into triphosgene solution with stirring not to exceed −50 • C. After the solution was added, the bath was removed, and the solution was stirred until it reached room temperature. To the isocyanate solution prepared, dried resin with L-Lys(Alloc) was added, and the mixture was stirred overnight. The resin was filtered, washed with dichloromethane, and dried in vacuum.
The reaction of Alloc group detachment was performed in darkness. In the dark, glass vessel the following solution was prepared: 0.1 × molar excess of (catalyst) Pd[P(Ph)3]4 (tetrakis (triphenylphosphine)palladium(0)) in 10% solution of morpholine in dry dichloromethane. The resin was swollen in dichloromethane and stirred 3 × 1 h each time with the new solution described above. The resin was filtered off and washed with N,N-dimethylformamide, 2% N,N-diisopropylethylamine in N,N-dimethylformamide, the solution of 20 mg mL −1 sodium diethylotiokarbamate in N,N-dimethylformamide, N,N-dimethylformamide, and dichloromethane. It was dried under vacuum. Further synthesis steps on solid support were performed in an automatic peptide synthesizer LibertyBlue equipped with Discover microwave oven (CEM). The microwave radiation speeds up the reaction, especially when coupling bulky chelators like DOTA [42,43].
The chelator DOTA(tBu)3 required an extended reaction time in milder conditions; it Scheme A1. Solid phase synthesis of the Glu(tBu)-urea-Lys-NH 2 moiety.
Further synthesis steps on solid support were performed in an automatic peptide synthesizer LibertyBlue equipped with Discover microwave oven (CEM). The microwave radiation speeds up the reaction, especially when coupling bulky chelators like DOTA [42,43].
The chelator DOTA(tBu) 3 required an extended reaction time in milder conditions; it was coupled at room temperature for 1 h following microwave heating to 50 • C for 30 min. The microwave-assisted synthesis is, in this case, fast. It needs to be performed immediately after preparing the reagents solutions due to the instability of COMU and Fmoc-4-Amc over a long time [54]. The product was detached from the resin and deprotected by stirring with the solution of 2% of triisopropylsilane, 2% of phenol, 2% of water, 2% of thioanisole, and 2% of 1,2-ethanedithiol in trifluoroacetic acid for 6 h, based on the publication of Wängler et al. [45]. The resulting product was precipitated and washed with diethyl ether. The precipitate was washed with diethyl ether through centrifugation. The precipitate was dissolved in 0.1% solution of trifluoroacetic acid in water and heated up on a rotary evaporator to 60 °C at 800 mbar for 1 h to detach any remaining Boc groups from tryptophane. The solution was frozen and lyophilized. The data were processed with linear prediction in t1 followed by zero-filling in both dimensions. Gaussian weighting functions were applied in both domains prior to Fourier transformation. In the cases where the signal to noise was sufficient, the use of sine weighting functions facilitated a better resolution of the spectra.
In Figures A1-A20, all basic spectra and some important expansions for recoded experiments are presented.

Appendix B.4. Results
A careful analysis of the results from various NMR spectra (298 and 353 K) confirms the compound studied structure shown in Scheme 1. However, 1 H and 13 C signals coming from ring A at room temperature (298 K) are significantly broadened due to the exchange process. Application of higher temperature (353 K) made this process faster, and NMR signals are thinner/narrower, allowing their assignment to proper atoms/nuclei. Halfheight widths in the case of proton signals of ring A (at room temperature and 353 K) were still too big to make 1 H-15 N transfer possible, and that is why no 15 N signals for ring A were observed in 1 H-15 N HMBC experiment. Additionally, in the long-accumulated 13 C NMR spectrum of compound supplied, a quartet at 158.5 ppm ( 2 J C-F = 32 Hz) was observed, suggesting the presence of trifluoroacetic acid in the sample.
The data were processed with linear prediction in t1 followed by zero-filling in both dimensions. Gaussian weighting functions were applied in both domains prior to Fourier transformation. In the cases where the signal to noise was sufficient, the use of sine weighting functions facilitated a better resolution of the spectra.
In Figures A1-A20, all basic spectra and some important expansions for recoded experiments are presented.

Appendix B4. Results
A careful analysis of the results from various NMR spectra (298 and 353 K) confirms the compound studied structure shown in Scheme 1. However, 1 H and 13 C signals coming from ring A at room temperature (298 K) are significantly broadened due to the exchange process. Application of higher temperature (353 K) made this process faster, and NMR signals are thinner/narrower, allowing their assignment to proper atoms/nuclei. Halfheight widths in the case of proton signals of ring A (at room temperature and 353 K) were still too big to make 1 H-15 N transfer possible, and that is why no 15 N signals for ring A were observed in 1 H-15 N HMBC experiment. Additionally, in the long-accumulated 13 C NMR spectrum of compound supplied, a quartet at 158.5 ppm ( 2 JC-F = 32 Hz) was observed, suggesting the presence of trifluoroacetic acid in the sample.

Appendix D
In vitro results . Total binding ( ) and non-specific binding ( ) were determined by incubation PSMA ligands with LNCaP (PSMA positive) and PC3 (PSMA negative) membranes, respectively. The specific binding ( ) was evaluated as a difference between the total and non-specific binding. . Total binding ( ) and non-specific binding ( ) were determined by incubation PSMA ligands with LNCaP (PSMA positive) and PC3 (PSMA negative) membranes, respectively. The specific binding ( ) was evaluated as a difference between the total and non-specific binding.
) and non-specific binding (

Appendix D
In vitro results . Total binding ( ) and non-specific binding ( ) were determined by incubation PSMA ligands with LNCaP (PSMA positive) and PC3 (PSMA negative) membranes, respectively. The specific binding ( ) was evaluated as a difference between the total and non-specific binding. . Total binding ( ) and non-specific binding ( ) were determined by incubation PSMA ligands with LNCaP (PSMA positive) and PC3 (PSMA negative) membranes, respectively. The specific binding ( ) was evaluated as a difference between the total and non-specific binding. ) was evaluated as a difference between the total and non-specific binding.