Albumin-Binding PSMA Radioligands: Impact of Minimal Structural Changes on the Tissue Distribution Profile

The concept of using ibuprofen as an albumin-binding entity was recently demonstrated by the development of [177Lu]Lu-Ibu-PSMA-01. In the present study, we designed a novel ibuprofen-containing radioligand (Ibu-PSMA-02) with subtle structural changes regarding the linker entity in order to investigate a potential impact on the in vitro and in vivo properties. Ibu-PSMA-02 was prepared using solid-phase synthesis techniques and labeled with lutetium-177. [177Lu]Lu-Ibu-PSMA-02 was evaluated in vitro with regard to its plasma protein-binding properties, PSMA affinity and uptake into PSMA-expressing PC-3 PIP tumor cells. The tissue distribution profile of [177Lu]Lu-Ibu-PSMA-02 was assessed in tumor-bearing mice and dose estimations were performed. The in vitro characteristics of [177Lu]Lu-Ibu-PSMA-02 were similar to those previously obtained for [177Lu]Lu-Ibu-PSMA-01 with respect to plasma protein-binding, PSMA affinity and tumor cell uptake. The in vivo studies revealed, however, an unprecedentedly high uptake of [177Lu]Lu-Ibu-PSMA-02 in PC-3 PIP tumors, resulting in an increased absorbed tumor dose of 7.7 Gy/MBq as compared to 5.1 Gy/MBq calculated for [177Lu]Lu-Ibu-PSMA-01. As a consequence of the high tumor accumulation, [177Lu]Lu-Ibu-PSMA-02 showed higher tumor-to-background ratios than [177Lu]Lu-Ibu-PSMA-01. This study exemplified that smallest structural changes in the linker entity of PSMA radioligands may have a significant impact on their pharmacokinetic profiles and, thus, may be applied as a means for ligand design optimization.


Introduction
The prostate-specific membrane antigen (PSMA) was found to be overexpressed in the majority of prostate cancer cases [1,2] and is, therefore, a promising target for radionuclide therapy of metastatic castration-resistant prostate cancer (mCRPC) [3][4][5]. A wide variety of PSMA-targeting radioligands were developed in recent years, among them also conjugates for labeling with a variety of therapeutic radionuclides [6,7]. [ 177 Lu]Lu-PSMA-617 is the currently best-known PSMA Figure 1. Chemical structure of prostate-specific membrane antigen (PSMA) ligands. (a) Novel ligand, Ibu-PSMA-02, with a diaminobutyric acid entity as connecting entity (red) to conjugate ibuprofen (blue); (b) Previously developed Ibu-PSMA-01 with a lysine entity as a connecting entity (yellow) to conjugate ibuprofen (blue) [17] and (c) PSMA-617 [11] as a reference compound without an albuminbinding entity.

Synthesis of Ibu-PSMA-02
Ibu-PSMA-02 was obtained with a yield of 15% after HPLC purification. Characterization by analytical HPLC and MALDI-TOF-MS confirmed the identity of the product and a chemical purity of >99% (Table 1, Supplementary Materials).
Overall, the uptake of [ 177 Lu]Lu-Ibu-PSMA-02 in the tumor was significantly higher (p < 0.05) and longer retained as compared to [ 177 Lu]Lu-Ibu-PSMA-01. Accumulation in non-targeted organs and tissue was also higher in the case of [ 177 Lu]Lu-Ibu-PSMA-02 at early timepoints. The fast clearance of [ 177 Lu]Lu-Ibu-PSMA-02 resulted, however, in similar or even lower retention of activity in kidneys at timepoints later than 1 h p.i. and in the liver later than 4 h p.i. [17]. Due to the presence of an albumin-binding entity and the resulting increased blood circulation time, maximum tumor uptake of [ 177 Lu]Lu-Ibu-PSMA-02 was reached at 24 h p.i. and was ~2.5-fold higher than the maximum tumor uptake of [ 177 Lu]Lu-PSMA-617 which was reached at 4 h p.i. Off-target accumulation of [ 177 Lu]Lu-Ibu-PSMA-02 as well as of [ 177 Lu]Lu-Ibu-PSMA-01 was, however, also higher than for [ 177 Lu]Lu-PSMA-617 at all investigated timepoints ( Figure 4, Supplementary Materials, Tables S1-S3). High blood pool activity 1 h after injection of [ 177 Lu]Lu-Ibu-PSMA-02 resulted in a low tumorto-blood ratio (2.2 ± 0.1), which increased significantly over time due to the efficient blood clearance. Tumor-to-blood ratios were in the range of 170−220 between 24 and 96 h after injection of [ 177 Lu]Lu-Ibu-PSMA-02 and reached a value of 918 ± 80 after eight days. The tumor-to-kidney ratio of [ 177 Lu]Lu-Ibu-PSMA-02 was low at 1 h p.i. (0.56 ± 0.09), but increased significantly with time to a tumor-tokidney ratio >50 eight days after injection. A similar time course was observed for the tumor-to-liver ratios. Overall, the tumor-to-background ratios of [ 177 Lu]Lu-Ibu-PSMA-02 were relatively low at early timepoints, but increased significantly over time. As a result, the tumor-to-blood, tumor-to-kidney and tumor-to-liver ratios of [ 177 Lu]Lu-Ibu-PSMA-02 were higher than those of [ 177 Lu]Lu-Ibu-PSMA-01 at timepoints later than 4 h p.i. ( Figure 5) [17]. Due to the lack of a designated albumin binder and thus extremely fast clearance from background organs, tumor-to-background ratios of [ 177 Lu]Lu-PSMA-617 were higher than those of the albumin-binding radioligands at all investigated timepoints (Supplementary Materials, Figure S4) [11].

Dosimetric Calculations
High accumulation and retention of [ 177 Lu]Lu-Ibu-PSMA-02 in the tumor led to a mean absorbed tumor dose of 7.7 ± 0.8 Gy/MBq. The mean specific absorbed dose to the kidneys was 0.71 ± 0.05 Gy/MBq, resulting in a tumor-to-kidney absorbed dose ratio of 11 ± 2. In contrast to [ 177 Lu]Lu-Ibu-PSMA-02, the mean absorbed tumor dose of [ 177 Lu]Lu-Ibu-PSMA-01 was lower (5.1 ± 0.2 Gy/MBq) and the mean absorbed dose to the kidneys higher (1.0 ± 0.2 Gy/MBq). This led to a lower tumor-tokidney absorbed dose ratio of 5.1 ± 0.5. Mean absorbed doses for both the tumor and the kidneys were clearly lower in the case of [ 177 Lu]Lu-PSMA-617 (3.90 and 0.05 Gy/MBq, respectively) and, the resulting tumor-to-kidney dose ratio (~87), therefore, much higher as compared to the albuminbinding radioligands ( Figure 6) [19].

Discussion
In this study, the impact of a small structural change on the in vitro and in vivo behavior of a PSMA radioligand was investigated. The newly developed radioligand [ 177 Lu]Lu-Ibu-PSMA-02 differed from [ 177 Lu]Lu-Ibu-PSMA-01 by two carbon atoms in the linker entity which connected ibuprofen with the radioligand's backbone.
The synthesis of Ibu-PSMA-02 was performed in analogy to the previously published synthesis of Ibu-PSMA-01 [17] by means of a resin-immobilized precursor containing a diaminobutyric acid entity. Ibu-PSMA-02 was obtained after a single purification step in high purity. This was indirectly confirmed by the feasibility of radiolabeling at a molar activity up to 100 MBq/nmol with a radiochemical purity of >98%. Interestingly, [ 177 Lu]Lu-Ibu-PSMA-02 was considerably more stable than [ 177 Lu]Lu-PSMA-617 and other previously synthesized ibuprofen-derivatized PSMA radioligands [17]. In this regard, [ 177 Lu]Lu-Ibu-PSMA-02 rather resembled [ 177 Lu]Lu-PSMA-ALB-56 containing a p-tolyl entity as stabilizing albumin binder [16]. This result showed that the radiolytic stability of a radioligand can be effectively influenced by small structural changes. Similarly, Kelly et al. demonstrated that the structure of the linker entity might have a significant impact on the stability of PSMA radioligands [18].
Evaluation of the concentration-dependent affinities of both albumin-binding radioligands to mouse plasma proteins showed that the small structural changes in the connecting entity between ibuprofen and the PSMA ligand's backbone did not have an impact on the binding affinity. In analogy, binding to human plasma proteins and to human serum albumin showed similar results between the radioligands (Supplementary Materials, Figure S3) Even though the structural differences between Ibu-PSMA-02 and Ibu-PSMA-01 were small, the tissue distribution profile of [ 177 Lu]Lu-Ibu-PSMA-02 in PC-3 PIP tumor-bearing mice was different to the profile previously obtained with [ 177 Lu]Lu-Ibu-PSMA-01 [17]. Stability studies confirmed that both radioligands were entirely stable under the conditions of radioligand preparation for biodistribution studies (data not shown). These results ensured that the radioligand solutions injected into the mice did not contain any degradation products, which could have influenced the tissue distribution.
Mice injected with [ 177 Lu]Lu-Ibu-PSMA-02 showed higher accumulation of activity in PC-3 PIP tumors as compared to [ 177 Lu]Lu-Ibu-PSMA-01 and other ibuprofen-containing PSMA radioligands [17]. This may have been a consequence of the increased activity level in the blood pool of [ 177 Lu]Lu-Ibu-PSMA-02 at early timepoints. Obviously, in vitro albumin-binding properties of a radioligand under equilibrium conditions cannot be considered as the only predictor of its blood retention time, and consequently, tissue distribution profile. The somewhat higher retention of [ 177 Lu]Lu-Ibu-PSMA-02 in the blood resulted, however, also in higher uptake in non-targeted tissues and organs, particularly in the kidneys. Despite increased off-target accumulation of [ 177 Lu]Lu-Ibu-PSMA-02 at early timepoints after injection, the tumor-to-kidney and tumor-to-liver ratios at later timepoints were higher than for other albumin-binding radioligands evaluated under the same experimental conditions (Supplementary Materials, Figure S4) [11,17]. The high PSMA expression level of PC-3 PIP tumors does, however, not allow a direct comparison with other radioligands that were evaluated in LNCaP tumor-bearing mice [18,20] since LNCaP tumors express PSMA at considerably lower levels [21]. Moreover, the injected molar amount of the radioligand may have a critical impact on its tissue distribution profile as well. In agreement with biodistribution data, the SPECT/CT images showed increased accumulation of [ 177 Lu]Lu-Ibu-PSMA-02 in the PC-3 PIP tumor as compared to the uptake of [ 177 Lu]Lu-Ibu-PSMA-01 and [ 177 Lu]Lu-PSMA-617, but also higher uptake in the kidneys at 4 h p.i. Intense activity signals were also found in the urinary bladder for both albumin-binding radioligands. This can be ascribed to residual activity after emptying the urinary bladder before acquisition of the SPECT/CT image, and from the ongoing urine formation during the scan. The increased blood circulation of albumin-binding PSMA radioligands certainly poses the concern of an increased bone marrow toxicity. The absorbed dose to the bone marrow is, however, not easily assessable. Based on AUC calculations, the tumor-to-blood AUC ratio of [ 177 Lu]Lu-Ibu-PSMA-02 was, however, higher than for [ 177 Lu]Lu-Ibu-PSMA-01, which would certainly be favorable in view of a therapeutic application (Supplementary Materials, Figure S5, Table S4).
Additional in-depth studies will be necessary in order to fully understand the observed discrepancies among the biodistribution profiles of [ 177 Lu]Lu-Ibu-PSMA-02 and [ 177 Lu]Lu-Ibu-PSMA-01.

Synthesis of Ibu-PSMA-02
In analogy to previously synthesized ibuprofen-derivatized PSMA ligands [17], Ibu-PSMA-02 was synthesized via a solid-phase platform. The synthesis of Ibu-PSMA-02 was based on precursor 1, which was prepared in analogy to the methods previously described (Supplementary Materials, Scheme S1) [6,11]. Instead of a L-lysine residue, L-diaminobutyric acid was introduced as a linker to conjugate (RS)-ibuprofen with the PSMA ligand's backbone. (RS)-Ibuprofen was coupled to precursor 1 to obtain compound 2 in the same manner as previously reported by Deberle and Benešová et al. (Supplementary Materials) [17]. Ibu-PSMA-02 was obtained after cleavage of the compound from the resin, removal of the protecting groups and HPLC purification of the crude product (Scheme 1). The isolated, pure PSMA ligand was characterized by analytical HPLC and MALDI-TOF-MS (Bruker UltraFlex II

Determination of LogD Values
The in vitro evaluation of [ 177 Lu]Lu-Ibu-PSMA-02 was carried out as previously reported for other ibuprofen-modified PSMA radioligands [11,17]. The n-octanol/PBS distribution coefficient (LogD) of [ 177 Lu]Lu-Ibu-PSMA-02 (50 MBq/nmol) was determined in a mixture of n-octanol and PBS, pH 7.4, by a shake-flask method using liquid-liquid extraction followed by phase separation and measurement of activity in both phases using a γ-counter (Perkin Elmer Wallac Wizard 1480). The distribution coefficient was calculated as the logarithm of the ratio of counts per minute (cpm) measured in the n-octanol phase relative to the cpm measured in the PBS phase. Three experiments were performed in quintuplicate.

Albumin-Binding Properties
The binding of [ 177 Lu]Lu-Ibu-PSMA-02 and [ 177 Lu]Lu-Ibu-PSMA-01 to mouse plasma proteins was evaluated by incubating a fixed amount of radioligand with different concentrations of mouse plasma and followed by determination of the percentage of bound radioligand using an ultrafiltration method. Mouse plasma was diluted to obtain a mouse serum albumin (MSA) concentration of 0.07 to 550 µM in PBS pH 7.4. A fixed amount of the radioligand (50 MBq/nmol; 15 µL, ~300 kBq, 0.006 nmol,) was added to 150 µL of the mouse plasma dilutions, briefly vortexed and incubated for 30 min at 37 °C. Afterwards, each sample of mouse plasma-radioligand mixture was cooled on ice followed by addition of 300 µL ice-cold PBS. The samples were transferred to Amicon centrifugal filters (cut-off of 10 kDa), which ensured the retention of MSA (~67 kDa), and centrifuged (rcf of 14,000) for 30 min at 4 °C. The inserts of the filter devices were inverted and centrifuged (rcf of 200) for 3 min to recover the protein fraction. The activity of the protein fraction (Aprotein) and the activity in the filtrate and in the filter unit were measured separately in a γ-counter. The total activity (Atotal) was determined as the sum of counted activity of the protein fraction (Aprotein), the filtrate and the filter unit. The percentage of radioligand bound to mouse plasma proteins was calculated as (bound radioligand = Aprotein/Atotal *100).
The results were presented as average ± standard deviation (SD) of four independent experiments. The percentage of radioligand bound to mouse plasma (Aprotein) was plotted against the MSA-to-radioligand molar ratios and fitted with a nonlinear regression curve (one site, specific binding) using GraphPad Prism software (version 7, San Diego, CA, U.S.A.) to obtain the halfmaximum binding (B50). Control experiments were performed by incubating the radioligands with PBS only followed by filtration which resulted in ≤10% of unspecifically retained radioligand in the filter. The binding of [ 177 Lu]Lu-PSMA-617 to mouse plasma proteins was previously determined by Umbricht et al. using a similar procedure but a different ultrafiltration device [16].
Additionally, the radioligands were incubated in human plasma and in 700 µM human serum albumin (HSA) diluted in PBS to confirm the binding also to human plasma proteins and albumin, respectively (Supplementary Materials). The tumor cells were cultured in RPMI-1640 cell culture medium supplemented with 10% fetal calf serum, L-glutamine and antibiotics. Puromycin (2 µg/mL) was added to the cell cultures to maintain PSMA expression as previously reported [21].

Cell Culture and Cell Experiments
Cell uptake and internalization studies of [ 177 Lu]Lu-Ibu-PSMA-02 were performed as previously reported [11]. The radioligand was investigated in three experiments performed in six replicates with PC-3 PIP tumor cells and in triplicate with PC-3 flu tumor cells. The PSMA-binding affinity of [ 177 Lu]Lu-Ibu-PSMA-02, presented as KD value, was determined as previously reported [17]. Statistical significance of KD values was assessed using a one-way ANOVA with a Tukeys's multiple comparison post-test in GraphPad Prism software (version 7, San Diego, CA, U.S.A.). A p-value of <0.05 was considered statistically significant.

Tumor Mouse Model
All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. In particular, all animal experiments were carried out according to the guidelines of Swiss Regulations for Animal Welfare. The preclinical studies were ethically approved by the Cantonal Committee of Animal Experimentation and permitted by the responsible cantonal authorities (license N° 75668).
Mice were obtained from Charles River Laboratories, Sulzfeld, Germany, at the age of 5-6 weeks.

Biodistribution Study
[ 177 Lu]Lu-Ibu-PSMA-02 (5 MBq, 1 nmol, 100 µL) was diluted in saline containing 0.05% bovine serum albumin (BSA) and injected in a lateral tail vein of the mice. Mice were euthanized at different timepoints after injection (p.i.) of the radioligand. Selected tissues and organs were collected, weighed and measured for activity using a γ-counter. The results were decay-corrected and listed as percentage of the injected activity per gram of tissue mass (% IA/g) (Figure 4, Supplementary Materials, Table S1). The values of the areas under the curve (AUCs) for the PC-3 PIP tumor and the blood were calculated based on non-decay-corrected biodistribution data (Supplementary Materials).

Dosimetric Calculations
Dosimetric calculations for [ 177 Lu]Lu-Ibu-PSMA-02 and [ 177 Lu]Lu-Ibu-PSMA-01 were performed based on non-decay-corrected biodistribution data. The cumulated activity was estimated by calculating the time-integrated activity concentration coefficients (TIACCs) and used for calculation of the mean specific absorbed dose (Gy/MBq) to the tumor xenografts and kidneys (Supplementary Materials). The absorbed fractions for the tumors and the kidneys were assessed by Monte Carlo simulations using PENELOPE [22]. Dosimetric calculations for [ 177 Lu]Lu-PSMA-617 were previously performed with a different method as published by Müller et al. [19].

In Vivo SPECT/CT Imaging
SPECT/CT images were obtained using a dedicated small-animal SPECT/CT scanner (NanoSPECT/CT TM , Mediso Medical Imaging Systems, Budapest, Hungary) as previously reported [21]. [ 177 Lu]Lu-Ibu-PSMA-02 (25 MBq, 1 nmol, 100 µL) diluted in saline containing 0.05% BSA was injected in a lateral tail vein of the mouse. SPECT scans of ~45 min duration were performed 4 and 24 h after injection of the radioligand followed by a CT of 7.5 min duration. During the in vivo scans, the mice were anesthetized with a mixture of isoflurane and oxygen. Reconstruction of the acquired data was performed using HiSPECT software (version 1.4.3049, Scivis GmbH, Göttingen, Germany). All images were prepared using VivoQuant post-processing software (version 3.5, inviCRO Imaging Services and Software, Boston, MA, U.S.A.). A Gauss post-reconstruction filter (FWHM = 1.0 mm) was applied to the images and the scale of radioactivity was set as indicated on the images (minimum value = 0.7 Bq/voxel to maximum value = 70 Bq/voxel).

Conclusions
This study demonstrated that even subtle changes in the chemical structure may have a major impact on the pharmacokinetic properties of a PSMA radioligand, as exemplified by the characteristics of [ 177 Lu]Lu-Ibu-PSMA-02 and [ 177 Lu]Lu-Ibu-PSMA-01. [ 177 Lu]Lu-Ibu-PSMA-02 comprising a shorter linker showed unprecedentedly high tumor uptake as a consequence of a relatively high blood pool activity at early timepoints after injection. Further investigations are warranted to better understand the relationship between the structure of ibuprofen-derivatized PSMA radioligands and their in vivo properties.