A prostate-specific membrane antigen (PSMA)-targeted prodrug with a favorable in vivo toxicity profile

Prostate-specific membrane antigen (PSMA) is a promising target for the treatment of advanced prostate cancer (PC) and various solid tumors. Although PSMA-targeted radiopharmaceutical therapy (RPT) has enabled significant imaging and prostate-specific antigen (PSA) responses, accumulating clinical data are beginning to reveal certain limitations, including a subgroup of non-responders, relapse, radiation-induced toxicity, and the need for specialized facilities for its administration. To date non-radioactive attempts to leverage PSMA to treat PC with antibodies, nanomedicines or cell-based therapies have met with modest success. We developed a non-radioactive prodrug, SBPD-1, composed of a small-molecule PSMA-targeting moiety, a cancer-selective cleavable linker, and the microtubule inhibitor monomethyl auristatin E (MMAE). SBPD-1 demonstrated high binding affinity to PSMA (Ki = 8.84 nM) and selective cytotoxicity to PSMA-expressing PC cell lines (IC50 = 3.90 nM). SBPD-1 demonstrated a significant survival benefit in two murine models of human PC relative to controls. The highest dose tested did not induce toxicity in immunocompetent mice. The high specific targeting ability of SBPD-1 to PSMA-expressing tumors and its favorable toxicity profile warrant its further development.


Results
SBPD-1 binds with high affinity to PSMA and contains a cathepsin B cleavable linker.. To achieve a specific, high-affinity interaction with PSMA we used the low-molecular-weight (LMW) scaffold Lys-Glu-Urea-DSS originally developed in our laboratory 32 . The synthetic tubulin inhibitor MMAE was conjugated to the Lys-Glu-Urea-DSS via a cathepsin B cleavable valine-citrulline linker (SBPD-1) or non-cleavable linker (SBPD-2), as a control to determine the utility of the linker (Fig. 1a).

SBPD-1 selectively kills PSMA-expressing PC xenografts in vivo.
Prior to in vivo potency we evaluated the stability of SBPD-1 and SBPD-2 in human and murine serum. SBPD-1 remained intact in human serum out to 48 h of incubation (Fig. 3). While 90% of SBPD-2 remained intact for 48 h in murine serum (data not shown), SBPD-1 was metabolized more quickly (Fig. 3). While more than 80% of SBPD-1 was intact in serum at 8 h of incubation, less than half represented parent compound at 24 h, and the majority of the prodrug was fully degraded by 48 h of incubation. It has been reported that the valine-citrulline linker is stable in human and monkey serum but that it can be hydrolyzed in mouse plasma via extracellular carboxylesterase 1c 37,38 . Based on those stability results, we applied small, fractionated doses for the murine efficacy study to avoid systemic toxicity that could affect the overall survival of the test animals.
To evaluate efficacy in preclinical models of human PC, we initially employed xenograft tumor models derived from PSMA+ PC3 PIP and PSMA− PC3 flu cells in NOD/SCID/IL2Rγnull (NSG) mice. Three weeks after injection of the cells, the average tumor volume reached 62.4 (± 11.6) mm 3 , and mice were treated with 20, 40 and 80 μg/kg of SBPD-1 via daily intraperitoneal (IP) injection for 30 days, n = 5. We monitored tumor growth and overall animal welfare (Fig. 4a). Animals were scored 'dead' when the tumor reached 4-times its original volume (Fig. 4b). Tumors in non-treated, control mice for both tumor types, in PSMA+ PC3 PIP mice treated with 20 μg/ kg and in PSMA− PC3 flu mice with all three doses, grew rapidly and all animals so treated were euthanized on day 20 post-initiation of treatment (Fig. 4b)   www.nature.com/scientificreports/ that they were undetectable by the completion of treatment (Fig. 4a). Approximately 1 week was required to be able to re-measure previously undetectable tumors in the group treated at 40 μg/kg. Two weeks were required for re-appearance of tumors in animals treated with the 80 μg/kg dose. In animals harboring PSMA + PC3 PIP tumors, both the 40 and 80 μg/kg doses provided significant survival benefits as the median survival times were 54 days [P = 0.003, Log-rank (Mantel-Cox) test] and 69 days (P = 0.003), respectively (Fig. 4b). Urine protein level and specific gravity measured for all test animals on Days 9 and 20 were normal, indicating that no acute renal toxicity occurred at any dose tested (Supplementary Table S1). SBPD-1 is non-toxic to C57BL/6 mice. We evaluated potential toxicity of SBPD-1 in immunocompetent animals. We administered MMAE (80 μg/kg), SBPD-1 (160 μg/kg), and 5% DMSO to healthy C57BL/6 mice (n = 5). We monitored animals for 80 days after initiation of administration. As previously reported 39 , MMAE demonstrated severe toxicity as all treated mice required euthanasia during treatment due to weight loss (Fig. 6b). Mice injected with vehicle or SBPD-1 did not show any signs of toxicity and steadily gained weight  www.nature.com/scientificreports/ ( Fig. 6a). We removed lung, liver, kidneys, salivary and lacrimal glands from all tested animals at Day 80 after initiation of SBPD-1 treatment. Histopathological examination revealed no tissue damage (Fig. 6c). We also obtained peripheral blood from mice injected with vehicle, SBPD-1, and healthy untreated animals, and prepared serum for chemistry studies (n = 5). Blood urea nitrogen (BUN), creatinine, glucose, alkaline phosphatase (ALP), total protein (T-Pro), and alanine aminotransferase (ALT) analyses showed that animals injected with either vehicle or SBPD-1 did not show differences in these values compared with those from untreated mice (Supplementary Table S2). Complete blood counts from the mice also showed no abnormalities except for lower white blood cell count for mice injected with SBPD-1, which may have resulted from the relative instability of the cathepsin B linker in murine serum and subsequent bone marrow toxicity of MMAE 37,40 .

Discussion
Prostate-specific membrane antigen (PSMA) was first identified as a marker for PC through cloning of a monoclonal antibody raised against the patient-derived PC cell line, LNCaP 41 . Since PSMA was discovered to be the same as the N-acetyl-l-aspartyl-l-glutamate peptidase I (NAALADase I) 42 , PSMA has been pursed as a target for diagnostic imaging of advanced PC with various low-molecular-weight agents 35,[43][44][45][46] . Anti-PSMA antibodies have also been tested as PSMA-targeting entities for both molecular imaging and therapy of PC [6][7][8]47,48 . Other therapeutic approaches such as PSMA targeted-nanoparticles loaded with an anti-cancer drug 49,50 or photodynamic therapy [51][52][53] have been tested in preclinical and clinical settings. PSMA-targeted RPT has provided a new alternative to managing patients with advanced PC refractory to other therapies 54,55 . Recent prospective trials of 177 Lu-based therapies have demonstrated substantial imaging and PSA responses 56,57 . Fewer side effects than other systemic therapies, such as hormonal or chemotherapy, have repeatedly been shown 58 . Nevertheless, approximately 50% of patients were non-responders, and the majority of responders relapsed, requiring further cycles or other options 55 . Questions about long-term toxicity of this method remain, particularly for α-particle emitting versions of RPT 17,18,59,60 .
Although PSMA-targeted RPT is promising and fraught with fewer adverse events compared to the conventional cytotoxic therapies, radiation exposure to normal organs can result in xerostomia or other off-target effects [16][17][18]59,60 . A PSMA-targeted prodrug equipped with additional specificity to malignant cells may provide an enhanced therapeutic index. Several PSMA-targeted prodrugs were tested in both preclinical and clinical settings. Kularatne et al. tested various cytotoxic drugs as a form of prodrug by conjugating them to the PSMAtargeted agent, 2-[3-(1,3-dicarboxy propyl)ureido] pentanedioic acid 61 . Those prodrugs utilized a disulfide linker to enable drug release in the reducing environment of the cytoplasm. Some of the tested drugs exhibited cytotoxicity to PSMA-expressing LNCaP cells at single-or double-digit nanomolar concentration levels. However, in vivo safety and efficacy of those drugs have not been tested. Mipsagargin (G-202) is a prodrug consisting of an analog of thapsigargin conjugated to a PSMA-cleavable peptide 62 . Thapsigargin is a potent inhibitor of the sarcoplasmic/endoplasmic reticulum calcium adenosine triphosphatase (SERCA) pump essential for cell viability. Mipsagargin was used to target the PSMA-expressing tumor neovasculature of various solid cancers. Despite promising preclinical and phase I results 62 , phase II trials showed no clinical benefit for advanced hepatocellular carcinoma 63 . A PSMA-targeted antibody-MMAE conjugate (ADC) has been tested and showed favorable preclinical efficacy 25,26 . However, in a phase I trial with that conjugate, the therapeutic window proved narrow, necessitating modification of dose selection if the compound were to advance further 30 . The authors of that trial hypothesized that the toxicity may have been due to the systemic concentration of free MMAE released from the antibody 30 . The results from the corresponding phase II trial were recently published 64 . Toxicity was noted shortly after the initiation of the trial-particularly neutropenia and neuropathy-such that a dose reduction www.nature.com/scientificreports/ was necessary for it to continue. A partial radiologic response was obtained in only 2 of 119 participants, with none reporting a complete response 64 . Please note that we used PSMA + PC3 PIP cells to generate subcutaneous tumors that may not precisely reflect the case as it may occur in patients. Although we did not measure the number of PSMA molecules per PSMA + PC3 PIP cell in the current study, we have previously shown there to be an order of magnitude higher PSMA expression in these cells than in LNCaP cells, which are patient-derived 18 . However, PSMA + PC3/ML/ PSMA cells used for the metastatic model have comparable PSMA expression to that of LNCaP cells. Nevertheless, we used the PSMA + PC3 PIP/PSMA − PC3 flu cells to generate subcutaneous tumors in order to minimize the number of variables between cells used, as these lines are otherwise isogenic, and to see if any signal could be obtained in this proof-of-principal study. Future studies will explore tumor models that have a variety of levels of PSMA expression, including those that are more in line with what is seen in human specimens.
SBPD-1 was designed for safe delivery of the potent toxin MMAE to maximize its therapeutic index. There are three layers of specificity of this agent for malignant cells. First, there is high-affinity, specific PSMA targeting followed by internalization of drug-bound PSMA. Notably PSMA tends to localize to the centrosome upon internalization 65 enabling it to deliver a drug that interrupts microtubule formation to the compartment in which it can be most effective. Second, MMAE is released only upon enzymatic cleavage by cathepsin B, which is upregulated in the lysosomes of cancer cells 29 . The same drug with non-cleavable linker (SBPD-2) showed about 7,100-fold less potency in PSMA + cancer cells (Fig. 2). Third, MMAE inhibits microtubule polymerization, an essential process for cell division of cancer cells. A further advantage of the small-molecule approach is that drug conjugates tend to have superior tumor penetration and more rapid clearance from non-target sites than do ADCs 66 .
Since prior reports 37,38 as well as our results (Fig. 3a) have suggested that the valine-citrulline linker is unstable in murine serum, we modified the dosing plan to consist of several fractionated doses. Our in vivo safety results with an immunocompetent murine model showed no toxicity with the highest doses tested in the efficacy study (Fig. 6, Supplementary Table S2). It is likely that a clinical dosing plan could consist of less frequent administration as the valine-citrulline linker has been reported to be stable in human plasma 38 .
In summary, we have generated and tested in vivo a low-molecular-weight, PSMA-targeted prodrug that demonstrated tumor penetration and specificity sufficient to provide survival differences between PSMA + tumorbearing animals and animals bearing isogenic tumors devoid of PSMA, including in a metastatic model. Furthermore, despite carrying the potent anti-tumor agent MMAE, the conjugate was non-toxic. We believe that lower toxicity was due to the controlled environment to which MMAE was delivered, by virtue of the presence of a cathepsin B cleavable linker in the molecule. Compounds of this class or those employing similar strategies may enable safe and effective targeting of PSMA-expressing lesions in patients.

Methods
General methods and materials for syntheses of prodrugs. Experiments were carried out in compliance with ARRIVE guidelines. Detailed methods for the syntheses of prodrugs are described in the Supplementary Information. Commercially available reagents and solvents for syntheses were analytical grade and used without further purification. Diisopropylethylamine (DIPEA), triflouroacetic acid (TFA), 4-(Dimethyl amino) pyridine (DMAP), pyridine (Py) and N-(3-dimethylaminopropyl)-N-ethylcarbodiimide (EDC) were purchased from Sigma-Aldrich (Allentown, PA, USA). l-Glutamic acid 5-tert-butyl ester, bis(4-nitrophenyl) carbonate and 1-hydroxybenzotriazole hydrate (HOBt) were purchased from Chem-Impex International (Wood Dale, IL, USA), disuccinimidyl suberate was purchased from TCI America (Pittsburgh, PA, USA) and monomethyl auristatin E (MMAE) was purchased from BroadPharm (San Diego, CA, USA). High performance liquid chromatographic (HPLC) purification of final compounds (SBPD-1 and SBPD-2) was performed using a C 18 Luna 10 mm × 250 mm column (Phenomenex, Torrance, CA, USA) on an Agilent 1260 infinity LC system (Santa Clara, CA, USA) and eluted with water (0.1% TFA) (A) and CH 3 CN (0.1% TFA) (B). 1 H NMR spectra were recorded on a Bruker Ultrashield 500 MHz spectrometer. Chemical shifts (δ) are reported in parts per million (ppm) downfield by reference to proton resonances resulting from incomplete deuteration of the NMR solvent and the coupling constants (J) was reported in Hertz (Hz). High resolution mass spectra were obtained by the University of Notre Dame Mass Spectrometry and Proteomics Facility, Notre Dame, IN using ESI by direct infusion on a Bruker micrOTOF-II.

Cathepsin B cleavage.
Release of MMAE from prodrugs by a recombinant cathepsin B was analyzed using a modified method from previously published work 33  PSMA affinity and in vitro cytotoxicity. PSMA affinities of SBPD-1and SBPD-2 were measured using the modified Amplex Red glutamic acid/glutamate oxidase assay as previously described 34 . PSMA-expressing PC3-PIP, PSMA-negative PC3-flu, PSMA-positive PC3/ML/PSMA and PSMA-negative PC3/ML were maintained as previously described 18  In vivo toxicity. Male C57BL/6 mice were purchased from Jackson Laboratory. Ten-week-old mice were injected with the indicated doses of MMAE (formulated in 5% DMSO), SBPD-1 (formulated in saline) or 5% DMSO intraperitoneally (daily for 30 days, n = 5). Animals were monitored daily for weight changes and other abnormalities for 80 days. Animals were euthanized in a CO 2 chamber at day 80, and blood, lung, liver, kidney, salivary gland, and lacrimal gland were collected for complete blood counts, blood chemistry, and histopathological analyses. Complete blood counts including white blood cells (WBC), red blood cells (RBC), hemoglobin (HGB), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), and platelet (PLT) were measured using scil Vet ABC Hematology Analyzer (scil animal care company, Gurnee, IL). Blood chemistry parameters including blood urea nitrogen (BUN), glucose (GLU), Alkaline Phosphatase (ALP), total protein (T-Pro), Alanine aminotransferase (ALT) and Creatinine (Cre) were measured with Spotchem EZ chemistry analyzer (Arkray USA, Edina, MN). Hematoxylin and eosin slides were generated for five organs and examined by certified veterinary pathologist.

Data availability
All data used in this submission are included in the body of manuscript or in the Supplementary Information. Additional data related to the paper are available upon request. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.