Fluorine-18 Labeled Urea-Based Ligands Targeting Prostate-Specific Membrane Antigen (PSMA) with Increased Tumor and Decreased Renal Uptake

High expression of prostate-specific membrane antigen (PSMA) in prostate cancers prompted the development of the PSMA-targeted PET-imaging agent [18F]DCFPyL, which was recently approved by the FDA. Fluorine-18-labeled Lys–Urea–Glu-based oxime derivatives of [18F]DCFPyL were prepared for the comparison of their in vitro and in vivo properties to potentially improve kidney clearance and tumor targeting. The oxime radiotracers were produced by condensation of an aminooxy functionalized PSMA-inhibitor Lys–Urea–Glu scaffold with fluorine-18-labeled aldehydes. The radiochemical yields were between 15–42% (decay uncorrected) in 50–60 min. In vitro saturation and competition binding assays with human prostate cancer cells transfected with PSMA, PC3(+), indicated similar high nM binding affinities to PSMA for all radiotracers. In vivo biodistribution studies with positive control PC3(+) tumor xenografts showed that the kidneys had the highest uptake followed by tumors at 60 min. The PC3(+) tumor uptake was blocked with non-radioactive DCFPyL, and PC3(−) tumor xenograft (negative control) tumor uptake was negligible indicating that PSMA targeting was preserved. The most lipophilic tracer, [18F]2a, displayed comparable tumor-targeting to [18F]DCFPyL and a desirable alteration in pharmacokinetics and metabolism, resulting in significantly lower kidney uptake with a shift towards hepatobiliary clearance and increased liver uptake.


Introduction
Prostate cancer (PC) is the most common malignancy in men in the United States and Europe [1][2][3]. In recent decades, prostate cancer survival rates have improved; however, it is still a significant cause of death. Local PC is usually diagnosed with screening for prostate serum antigen (PSA serum testing), clinical examination, and imaging such as magnetic resonance imaging (MRI) followed by a biopsy of the prostate. Advanced PC, however, is commonly staged with computed tomography (CT), bone scans and positron emission tomography (PET), frequently using prostate-specific membrane antigen (PSMA)-targeted radioligands. Due to the higher sensitivity of PET over the other techniques, it is becoming more widely accepted as a diagnostic approach to identify sites of extra-prostatic disease. The metabolic radiotracer, 2-deoxy-2-[ 18 F]fluoro-D-glucose, [ 18 F]FDG, although commonly used in other cancers, has proven less useful in PC [4,5]. Carbon-11 or fluorine-18-labeled choline PET/CT showed promising results for the detection of bone metastases. However, these agents have limitations in terms of sensitivity and specificity [6]. This unmet clinical need led to the development of another class of radiotracers targeting the transmembrane protein PSMA, which is expressed in approximately 95% of PC cases including both primary and metastatic disease [7][8][9]. PSMA is a cell surface glycoprotein with carboxypeptidase and folate hydrolase enzymatic activities that has emerged as an important biomarker for PC and prompted the development of small-molecule inhibitors [10][11][12]. These smallmolecule inhibitors have proven to be suitable platforms for PET imaging with faster clearance rates and lower backgrounds.
The gallium-68 labeled PET tracer, Glu-NH-CO-Lys-(Ahx)-[ 68 Ga]Ga-N,N -Bis(2-hydroxy-5-(ethylene-betacarboxy)benzyl)ethylenediamine N,N -diacetic acid ([ 68 Ga]Ga-PSMA-11 (also named [ 68 Ga]Ga-PSMA-HBED-CC)), is the most widely studied PSMA radiotracer [13][14][15][16][17]. It was first reported by Eder et al. in 2012 [18]. The initial clinical PET imaging study with this tracer demonstrated a significant advantage compared to conventional imaging used for the detection of recurrent PC [13]. [ 68 Ga]Ga-PSMA-11 was recently approved by the Food and Drug Administration (FDA) for PET imaging of PSMA-positive lesions in men with prostate cancer [19]. However, the longer half-life of fluorine-18 (110 min) compared to gallium-68 (68 min) enables sufficient time for central production and local distribution of the tracers which is more pragmatic for most medical facilities. The extended imaging time with fluorine-18-labeled PSMA radiotracers may further increase the overall detection rate in patients with PC [20,21]. Moreover, fluorine-18 offers comparatively lower positron energy (fluorine-18, 633 keV vs. gallium-68, 1899 keV) with a resultant shorter positron range in the tissue, which may also improve image resolution [22,23]. Thus, the growing demand for PSMA-targeted PET imaging is likely to be better met by fluorine-18labeled radiotracers. Recently, Gust et al. proposed a molecular absorption spectrometry (MAS) method that uses fluorination as tool to improve bioanalytical labeling and suggested it as a potential alternative to 18 F-PET [24].
A variety of fluorine-18labeled PSMA-targeted PET radiotracers have been developed for PC imaging [25][26][27][28][29]. The most extensively studied tracers of these classes are urea-based small molecule inhibitors, e.g., N-  [30][31][32][33]. The clinical studies with the first-generation PSMA ligand [ 18 F]DCFBC demonstrated slow clearance with high background activity [34]. The second-generation ligands, [ 18 F] DCFPyL and [ 18 F] PSMA-1007 showed high tumor: background ratios and favorable pharmacokinetics compared to other small molecules [31,[35][36][37][38]. [ 18 F]-DCFPyL was approved by the FDA in 2021 for the detection of possible early metastatic PC involvement [39]. A wide range of prosthetic groups and linkers have been introduced to improve pharmacokinetics and detection rates with PSMA PET [18,[40][41][42][43]. These studies demonstrated favorable binding properties for more lipophilic compounds and inspired us to develop oxime derivatives with increased lipophilicity (Scheme 1). Herein, we report the synthesis of the precursor, radiolabeling, and biological evaluation of these oxime derivatives in comparison with previously reported tracers [ 18 F]DCFPyL and [ 18 F]1a. The biological evaluations include in vitro binding studies to assess the affinity (K d ) of these compounds for PSMA and in vivo biodistribution studies with PSMA-positive tumor mouse models to determine tumor targeting and differences in pharmacokinetics and metabolism.

Radiochemistry
Oxime formation of the aminooxy functionalized lysine-urea-glutamate scaffold (1 and 2, Scheme 1) with fluorine-18labeled aldehydes produced radiotracers [ 18 Figure 1) on the GE Tracerlab module (GE FX-N Pro) to accomplish the Sep-Pak fluorination of 6-[ 18 F]fluoronicotinaldehyde. The overall radiochemical yield (2 steps) of the synthesis was 15-42% (n >10, decay uncorrected) in a 50-60 min procedure. The radiochemical purity was >98% with a molar activity of 300-360 GBq/µ mol. The identities of the products were confirmed by comparing their HPLC retention times with co-injected, authentic non-radioactive standards. A representative HPLC profile for compound [ 18 F]1b is shown in Figure 2

Radiochemistry
Oxime formation of the aminooxy functionalized lysine-urea-glutamate scaffold (1 and 2, Scheme 1) with fluorine-18labeled aldehydes produced radiotracers [ 18 Figure 1) on the GE Tracerlab module (GE FX-N Pro) to accomplish the Sep-Pak fluorination of 6-[ 18 F]fluoronicotinaldehyde. The overall radiochemical yield (2 steps) of the synthesis was 15-42% (n >10, decay uncorrected) in a 50-60 min procedure. The radiochemical purity was >98% with a molar activity of 300-360 GBq/µmol. The identities of the products were confirmed by comparing their HPLC retention times with co-injected, authentic nonradioactive standards. A representative HPLC profile for compound [ 18 F]1b is shown in Figure 2

Radiochemistry
Oxime formation of the aminooxy functionalized lysine-urea-glutamate scaffold (1 and 2, Scheme 1) with fluorine-18labeled aldehydes produced radiotracers [ 18 Figure 1) on the GE Tracerlab module (GE FX-N Pro) to accomplish the Sep-Pak fluorination of 6-[ 18 F]fluoronicotinaldehyde. The overall radiochemical yield (2 steps) of the synthesis was 15-42% (n >10, decay uncorrected) in a 50-60 min procedure. The radiochemical purity was >98% with a molar activity of 300-360 GBq/µ mol. The identities of the products were confirmed by comparing their HPLC retention times with co-injected, authentic non-radioactive standards. A representative HPLC profile for compound [ 18 F]1b is shown in Figure 2

In Vitro Cell Binding Studies
All tracers exhibited high specific binding (B sp ; 85-98%) with sub-nM affinity for PSMA using PC3(+) tumor membrane preparations ( Figure 3A; Table 1

In Vivo Biodistribution
The biodistribution of [ 18 F]DCFPyL was determined in nude mice bearing human prostate cancer tumors transfected with PSMA (PC3(+) xenografts) at 30, 60, 90 and 120 min post-injection ( Figure 4A,B). [ 18 F]DCFPyL distributed rapidly and cleared from the blood and non-target tissues except for the tumor over the 120 min time course ( Figure  4A). The kidney exhibited the highest uptake (133%ID/g to 50%ID/g) at all time points and decreased by 65% from 15 to 120 min. All other tissue uptakes except tumor were > 30 fold lower than kidneys at all times indicating that [ 18 F]DCFPyL is dominated by renal clearance, as expected from published results [31]. The next highest uptakes after the kidneys occurred in the PC3(+) tumor in which [ 18 F]DCFPyL was highly retained from 15 (18.6%ID/g) to 120 min (20.8%ID/g; Figure 4A). The tumor tissue to muscle ratio (T:M) steadily increased over the time course with an 11-fold increase from 15 (24 T:M) to 120 min (260 T:M). These tumor T:M increases are reflective of an increased rate of clearance from the muscle while in tumors the majority of the radioactivity was retained ( Figure  4B). The retention of [ 18 F]DCFPyL in the tumor with an accompanying increase in tumor T:M over time would indicate high-affinity binding to PSMA. However, this was not the case with the salivary glands in which 92% of the radioactivity had been cleared at 120 min and T:M decreased 34% from 15 (2.7 T:M) to 120 min (2.0 T:M). This lack of retention of [ 18 F]DCFPyL and the low T:M ratios suggest that salivary gland PSMA expression levels in mouse, known to be lower than in humans, are insufficient to render a meaningful biodistribution model. With this in mind, salivary glands were not included in further biodistributions with the 18 F-labeled analogues [35].

In Vivo Biodistribution
The biodistribution of [ 18 F]DCFPyL was determined in nude mice bearing human prostate cancer tumors transfected with PSMA (PC3(+) xenografts) at 30, 60, 90 and 120 min post-injection ( Figure 4A,B). [ 18 F]DCFPyL distributed rapidly and cleared from the blood and non-target tissues except for the tumor over the 120 min time course ( Figure 4A). The kidney exhibited the highest uptake (133%ID/g to 50%ID/g) at all time points and decreased by 65% from 15 to 120 min. All other tissue uptakes except tumor were > 30 fold lower than kidneys at all times indicating that [ 18 F]DCFPyL is dominated by renal clearance, as expected from published results [31]. The next highest uptakes after the kidneys occurred in the PC3(+) tumor in which [ 18 F]DCFPyL was highly retained from 15 (18.6%ID/g) to 120 min (20.8%ID/g; Figure 4A). The tumor tissue to muscle ratio (T:M) steadily increased over the time course with an 11-fold increase from 15 (24 T:M) to 120 min (260 T:M). These tumor T:M increases are reflective of an increased rate of clearance from the muscle while in tumors the majority of the radioactivity was retained ( Figure 4B). The retention of [ 18 F]DCFPyL in the tumor with an accompanying increase in tumor T:M over time would indicate high-affinity binding to PSMA. However, this was not the case with the salivary glands in which 92% of the radioactivity had been cleared at 120 min and T:M decreased 34% from 15 (2.7 T:M) to 120 min (2.0 T:M). This lack of retention of [ 18 F]DCFPyL and the low T:M ratios suggest that salivary gland PSMA expression levels in mouse, known to be lower than in humans, are insufficient to render a meaningful biodistribution model. With this in mind, salivary glands were not included in further biodistributions with the 18 F-labeled analogues [35]. indicate that tumor uptake represents specific PSMA binding. Other significant decreases in T:M ratios occurred in the kidney (96%) and spleen (81%) compared to the [ 18 F]DCFPyLonly group. These decreases most likely are not entirely attributable to PSMA specific binding but could be a result of the altered metabolism in the kidney or cross-reactivity with glutamate carboxypeptidase III (GPCIII) in the spleen, respectively [44]. The only increase in T:M occurred in the liver (27:1 T:M; 2.7 fold) compared to the [ 18 F]DCFPyL only group which most likely is due to a shift from renal towards hepatobiliary metabolism.  Figure 5C). These results indicate that PSMA targeting has been preserved for all the analogues compared to [ 18     Since differences were observed in metabolism between [ 18 F]DCFPyL and the other tracers, the fraction of radioactivity that represented intact tracer (%Parent) in blood was determined at 60 min by TLC (  Since differences were observed in metabolism between [ 18 F]DCFPyL and the other tracers, the fraction of radioactivity that represented intact tracer (%Parent) in blood was determined at 60 min by TLC ( F ] D C F P y L  In  (Table 3) [44]. Similarly [ 18 F]DCFPyL parent kidney uptake was retained from 15 to 60 min although a decrease was observed at 120 min.

Discussion
Recently, Bouvet et al. reported the influence of different prosthetic groups on PSMA-targeted radiotracers (DCFPyL analogues) with improved tumor uptake and clearance profile [41]. The highest tumor uptake in their study was achieved with the most lipophilic compound (1a, Scheme 1), prepared via the oxime formation of 4-[ 18 F]fluorobenzaldehyde with aminooxy precursor 1. Moreover, they suggested that the low tumor uptake of the [ 18 F]FDG linked oxime tracer could be due to the combination of high hydrophilicity and steric crowding [36]. This result inspired us to further investigate the effect of adding an alkyl chain between the PSMA-inhibitor lysine-urea-glutamate scaffold and labeled prosthetic groups to increase lipophilicity and decrease steric crowding of the labeled PSMA probe. Therefore, compounds [ 18 F]1b and [ 18 F]2a-b were designed with alkyl chains of various lengths, and an arene or heteroarene substituent.
The biological evaluation of all tracers found that PSMA targeting was preserved both in vitro and in vivo and for the most part was comparable to [ 18 F]DCFPyL. In vitro, the labeled tracers and non-radioactive standards had retained specific and high nM binding to PSMA, with [

Discussion
Recently, Bouvet et al. reported the influence of different prosthetic groups on PSMAtargeted radiotracers (DCFPyL analogues) with improved tumor uptake and clearance profile [41]. The highest tumor uptake in their study was achieved with the most lipophilic compound (1a, Scheme 1), prepared via the oxime formation of 4-[ 18 F]fluorobenzaldehyde with aminooxy precursor 1. Moreover, they suggested that the low tumor uptake of the [ 18 F]FDG linked oxime tracer could be due to the combination of high hydrophilicity and steric crowding [36]. This result inspired us to further investigate the effect of adding an alkyl chain between the PSMA-inhibitor lysine-urea-glutamate scaffold and labeled prosthetic groups to increase lipophilicity and decrease steric crowding of the labeled PSMA probe. Therefore, compounds [ 18 F]1b and [ 18 F]2a-b were designed with alkyl chains of various lengths, and an arene or heteroarene substituent.
The biological evaluation of all tracers found that PSMA targeting was preserved both in vitro and in vivo and for the most part was comparable to [ 18 F]DCFPyL. In vitro, the labeled tracers and non-radioactive standards had retained specific and high nM binding to PSMA, with [ 18 F]1a tending to have higher affinity than [ 18 F]DCFPyL. All four analogues exhibited in vivo PSMA tumor-targeting comparable to [ 18 F]DCFPyL with tumor uptakes and T:M ratios, ranging from 27 to 17%ID/g and 203 to 74 T:M, respectively, which were at least 8-fold greater than non-target tissues except for the kidney and liver.
Tumor uptakes of [ 18 F]1a tended to be higher than [ 18 F]DCFPyL comparing favorably with previous findings [41]. Therefore, the modification of the canonical amide bond of DCFPyL to include an alkyl chain and oxime-linked [ 18 F]fluorobenzyl or -pyridinyl substituent minimally affected in vivo tumor targeting to PSMA, indicating that in human patients, all four [ 18 F]DCFPyL analogues would be expected to identify PSMA expressing lesions as has been clinically observed with [ 18 F]DCFPyL [45].
In human patients [ 18 F]DCFPyL has demonstrated favorable dosimetry within acceptable limits for diagnostic PET tracers, however high accumulation and retention in the kidneys and salivary glands could limit use in other clinical applications such as radionuclide therapies [35,46]. In a retrospective clinical trial Barber et.al reported 25% renal injury in patients treated with [ 177 Lu]Lu-PSMA-617 and currently sufficient kidney function is an important criterion for patient eligibility for this recently FDA-approved therapy [47][48][49]. In addition, retrospectively, xerostomia was found in 24% of patients by Heck et.al. which most likely is an underreported adverse effect resulting from high PSMA expression levels in the salivary glands [50]. The uptake of PSMA targeted imaging agents in human salivary glands is specific and consistent with known high PSMA expression levels which is not the case for mice. The mouse PSMA, a homolog of human PSMA, has a 12-fold lower expression level in salivary glands with higher levels of non-specific binding and therefore, may not be as reliable to detect changes in specific PSMA uptake [51,52]. In contrast, the elevated kidney uptake observed in mice is comparable to humans representing both specific PSMA binding in the renal cortex and non-specific radioactivity in the urinary tract due to excretion [44,53]. In studies investigating other fluorine-18labeled PSMA inhibitors the physiochemical properties of the labeled prosthetic groups were found to affect the biological clearance profiles, therefore, modification of the 18 F-labeled prosthetic group of [ 18 F]DCFPyL may offer a strategy to lower kidney uptake [41]. In these pre-clinical studies only [ 18 F]2a displayed a desirable alteration in pharmacokinetics and metabolism resulting in greater in vivo stability and significantly lower kidney uptake with higher liver uptake compared to [ [41]. The rank order of the liver uptake (%ID/g) of all the tracers indicated a switch to hepatobiliary clearance that corresponded to the tracers logP rank order suggesting that the lipophilicity plays a role in determining the clearance profile of the tracer. These results suggest that [ 18 F]2a may offer an alternative PSMA-targeting agent with decreased renal clearance in clinical applications.

Radiochemical Syntheses
All radiochemical syntheses were performed according to the following two general procedures described below. Fluorine-18 in target water (3700-7400 MBq) was diluted with 2 mL water and passed through an anion-exchange cartridge (Chromafix 30-PS-HCO 3 ). The cartridge was washed with anhydrous acetonitrile (6 mL) and dried for 1 min. The [ 18 F]fluoride from the cartridge was slowly eluted (0.5 mL/min) with its 4-formyl-N,N,N-trimethylbenzenaminium triflate precursor (5-7 mg) in 0.5 mL 1:4 acetonitrile: t-butanol. The Sep-Pak was further eluted with 0.5 mL acetonitrile and the eluent was collected in the same vial. The reaction mixture was heated at 120 • C for 2 min to produce 4-[ 18 F]fluorobenzaldehyde. The radiolabeled intermediate was purified by passing the reaction mixture through a pre-conditioned Oasis MCX Plus cartridge and collected in a vial containing aminooxy precursor 1 or 2 (5 mg) in 0.2 mL water. The cartridge was flushed with 1 mL acetonitrile and the eluent was collected in the same vial. The solution was stirred for 10 min at 70 • C and the solvent was evaporated under N 2 and reduced pressure. The HPLC buffer (3 mL) was added via a syringe. The mixture was injected into the HPLC for purification. The collected product was buffered to pH~7 with 45 mM sodium phosphate. The identity and purity of the product were confirmed by analytical HPLC. Fluorine-18 in target water (3700-7400 MBq) was diluted with 2 mL water and passed through an anion-exchange cartridge (Chromafix 30-PS-HCO 3 ) followed by anhydrous acetonitrile (6 mL) and the cartridge was dried for 3 min under vacuum. The [ 18 F]fluoride from the Sep-Pak was eluted with 5-formyl-N,N,N-trimethylpyridin-2-aminium triflate precursor (5-7 mg) in 0.5 mL 1:4, acetonitrile: t-butanol (in a syringe) via an external three-way valve. The mixture was passed through a pre-conditioned Oasis MCX Plus cartridge (incorporated between V13 and Reactor 1). The cartridge was flushed with 1 mL acetonitrile through the external three-way valve and the eluent was collected in the same vial (Reactor 1). To this solution in Reactor 1 was added the aminooxy precursor, 1 or 2, (5 mg) in water (0.5 mL) from Vial 3. The solution was stirred for 10 min at 70 • C and the solvent was then evaporated under N 2 and vacuum. The HPLC buffer (3 mL) was added from Vial 4. The mixture was transferred to Tube 2 and injected into the HPLC for purification. The collected product was buffered to pH~7 with 45 mM sodium phosphate. The identity and purity of the product were confirmed by analytical HPLC.
[ 18 F]1a: The radiochemical yield was 21-32% (uncorrected, n > 5) in 50 min with a molar activity of 300-330 GBq/µmol. HPLC conditions for purification: 30% B in A, t R =~18 min. HPLC conditions for analysis: 20% B in A, t R =~5 min.  [55,57]. Cell lines were grown at 37 • C in 5% CO 2 in RPMI-1640 supplemented with 10% FBS, 2 mM Lglutamine and Pen/Strep/Amphotericin B. PC3(+) and PC3(−) cell suspensions from in vitro cell culture were subcutaneously implanted (right shoulder) into athymic mice (Athymic NCr-nu/nu, Charles River Laboratory, 4 weeks old) for use as positive and negative controls, respectively, for in vitro or in vivo studies. When tumors reached the appropriate size (>100 mg) the xenograft mice were used for in vivo biodistributions and imaging studies or the tumors were excised and further processed to obtain membrane preparations for in vitro assays as described previously [58].

In Vitro Binding Studies
In vitro saturation studies were performed to determine binding affinities (K d ) and PSMA expression levels (B max ) using tumor membrane preparations from PC3(−) and PC3(+) PSMA xenografts (human prostate cancer cell line transfected with human PSMA; PC3(+)). A constant aliquot of the tumor membrane preparation was added to increasing concentrations of the tracers (0.5-70 nM) in duplicate (total bound activity (B t )); non-specific binding (B nsp ) was determined by adding non-radioactive DCFPyL (10 −6 M) to another set of duplicates. For competition assays a constant concentration of [ 18 F]DCFPyL (0.5 to 1.0 nM) and increasing concentrations (0-1000 nM) of competitors (non-radioactive standards; DCFPyL, 1a-b, or 2a-b) were added to membrane aliquots. After incubation (2 h at RT) separation of bound [ 18 F]DCFPyL from free was accomplished by filtration using GF/C filter papers followed by 2 washes with saline. Filter papers were collected, and the radioactive content was quantified by gamma counting (PerkinElmer 2480 Wizard3). From the saturation studies, the K d and B max were determined from 6-8 concentrations of the radiolabeled tracers and analyzed using non-linear regression curve fitting (one-site specific binding hyperbola); from the competition studies, inhibitory constants (K i )'s were determined from 8-10 competitor concentrations of non-radioactive standard/DCFPyL [PRISM (version 7.0 Windows), GraphPad software, San Diego, CA]. Aliquots of each membrane preparation were taken for the determination of the protein concentration (Bradford method).
In some cases, additional blood samples and/or tissue samples after gamma counting were taken for determining the fraction of intact radiolabeled tracer (parent) using thinlayer chromatography (TLC). For these TLC determinations: tissues were placed in equal volumes of acetonitrile and homogenized, or serum was obtained from the blood samples and mixed with an equal volume of acetonitrile. Following centrifugation of the samples, supernatants were collected and the radioactive content of the supernatants and pellets were determined. The supernatants were then applied to thin-layer chromatography (TLC) plates. The TLC plates were developed [solvent system: ethyl acetate (80%), methanol (10%) and acetic acid (10%)], and exposed on a phosphorimaging plate which was scanned the next day.

Conclusions
Fluorine-18-labeled urea-based PSMA inhibitors were prepared either manually or automatically in high radiochemical yield using the prosthetic group 4-[ 18