1. Introduction
At present, prostate cancer is still a major health burden with 1.4 million new cases worldwide and nearly 400,000 deaths annually [
1]. At early disease stages, treatment efficacy is high, but it decreases when the disease becomes more advanced. The most advanced stage of metastatic castration-resistant prostate cancer (mCRPC) is often incurable, resulting in a five-year survival rate of merely 15% [
2]. Recently, the Food and Drug Administration (FDA) approved radionuclide therapy with [
177Lu]Lu-PSMA-617, for treatment of mCRPC in patients with a progressive disease after chemo- and androgen deprivation therapy [
3].
[
177Lu]Lu-PSMA-617, targets the prostate specific membrane antigen (PSMA), also known as glutamate carboxypeptidase II (GCP II), which is shown to be overexpressed in 90–100% of prostate cancer cases [
4]. That, together with the large extracellular domain, makes PSMA a very interesting target for targeted radionuclide therapy. However, the expression of PSMA is not strictly limited to prostate cancer tissue, as PSMA expression was also observed in the vasculature of other neoplasms and healthy tissues such as the proximal tubules of the kidney and the salivary glands [
5]. The latter ones are of main interest, since the salivary glands are currently considered to be the dose-limiting organs following PSMA-targeted radionuclide therapy with small molecules [
6]. Human salivary glands can be subdivided into three major pairs (parotid, submandibular and sublingual) and minor salivary glands. Despite PSMA expression being much lower in the salivary glands compared to the prostate cancer tissue, salivary glands show a high uptake of PSMA-targeted small molecules, suggesting a non-specific retention mechanism [
7]. The ionizing radiation delivered to the salivary glands following PSMA-targeted radionuclide therapy damages the salivary glands, resulting in xerostomia, a debilitating condition, which severely decreases the quality of life and can even result in treatment discontinuation [
6]. When treating patients with PSMA-targeted small molecules combined with beta-emitting radionuclides, such as
177Lu, xerostomia is often reversible. However, irreversible damage has been observed using the alpha particle emitter,
225Ac [
8].
The mechanisms underlying the retention of [
177Lu]Lu-PSMA-617 in the salivary glands remain incompletely understood. Urea-based PSMA-targeted small molecules, such as PSMA-617, target PSMA via interaction between the glutamate moiety of PSMA-targeted molecules and the enzymatic pocket of PSMA having high glutamate affinity [
8,
9]. With this in mind, monosodium glutamate (MSG) has been investigated as a compound to decrease retention of PSMA-targeting small molecules in the salivary glands [
10,
11,
12]. While being successful at decreasing the healthy organ uptake of PSMA-targeting compounds, it also reduced tumor uptake, making it difficult to implement in clinical practice. However, these studies suggest that the uptake in the salivary glands of PSMA-targeted small molecules with a glutamate moiety might be reduced when interfering with the high affinity glutamate binding pocket of PSMA. Previous autoradiography studies have already indicated that the salivary glands suffer largely from non-specific accumulation of PSMA-targeted small molecules, resulting in salivary gland toxicity [
13]. In this study, we describe the uptake pattern of [
177Lu]Lu-PSMA-617 in prostate cancer and salivary gland cells. Further, autoradiography studies were conducted on pig salivary gland tissue and compared to mouse kidney tissue as a positive control. Additionally, histology was performed to characterize different pig salivary gland sections. Lastly, different glutamate receptor antagonists were used to investigate binding specificity of [
177Lu]Lu-PSMA-617 to the salivary glands (
Figure 1).
3. Discussion
Targeted radionuclide therapy using PSMA-targeted small molecules has emerged as an efficient new treatment option for patients with metastatic castration-resistant prostate cancer. Recently, this resulted in the FDA approval of [
177Lu]Lu-PSMA-617 as an end-stage treatment of this patient population. However, clinical results already indicated a significant salivary gland uptake of PSMA-targeted small molecules [
3]. At present, the salivary glands are considered to be the dose-limiting organ of treatment with PSMA-targeted small molecules, but mechanisms underlying the uptake and retention in these glands remain largely unknown [
6]. Therefore, our aim was to elucidate mechanisms of the uptake and retention of the well-known compound [
177Lu]Lu-PSMA-617 In vitro on cells by competition studies and on tissues by In vitro autoradiography experiments.
PSMA, also known as GCP II or folate hydrolase I, is an interesting target for prostate cancer therapy due to the overexpression on prostate cancer cells. However, this PSMA expression is not strictly limited to prostate cancer tissue, as PSMA expression has also been reported in the kidneys, small intestines and salivary glands, contributing to healthy organ toxicity. For the salivary glands, PSMA expression was investigated by quantitative polymerase chain reaction (qPCR), Western blots and immunohistochemistry and was reported to be rather low and restricted to specific subsites [
7]. Previous dosimetric studies also showed that the uptake of PSMA-targeted small molecules differed between the different salivary glands. The highest uptake (defined as the standardized uptake value, SUV) was observed in the submandibular gland (SUV
max 1 h post injection = 14.5), followed by the parotid gland (SUV
max 1 h post injection = 13.8) and lastly the sublingual gland (SUV
max 1 h post injection = 6.1) [
14,
15]. The parotid gland was reported to have the highest variation in SUV
max, which could explain the large difference in BP between salivary glands 1 and 2 observed in our study, both having parotid gland morphology. In contrast, salivary gland 3 had the highest BP amongst the salivary gland sections and our immunohistochemistry data showed mixed serous and mucous acini, being indicative of submandibular gland morphology and uptake pattern [
16,
17]. The binding of [
177Lu]Lu-PSMA-617 to the pig salivary gland was heterogeneous and confined to glandular areas as observed in the In vitro autoradiography study and H&E stainings. This finding was in line with previous reports [
15]. The glandular areas contain the salivary gland acinar structures responsible for saliva secretion. More so, PSMA glands are expressed in the acinar glandular cells and not in ductal cells [
15]. The higher [
177Lu]Lu-PSMA-617 retention in the glandular areas and subsequent damage to acinar structures might thus relate to clinically observed xerostomia and salivary gland dysfunction.
The low PSMA expression on salivary glands indicates that the high uptake observed in PSMA-PET/CT scans in patients treated with PSMA-targeted small molecules is at least in part non-specific [
18]. This hypothesis was validated in our research where at the cellular level, there was a high, specific binding of [
177Lu]Lu-PSMA-617 on PSMA positive PC3-PIP cells, while the binding on A-253 cells was low and mainly non-specific. We also performed saturation binding studies on mouse kidney tissue and different sections of pig salivary glands. These results validated the binding profile found on cells, where the mouse kidney tissue shows a high specific binding while the pig salivary glands showed mainly non-specific binding. The Bmax and Kd values we found for the tested pig salivary gland tissues are in line with previous research [
13].
Looking into the structure of the PSMA receptor, it consists of a large extracellular domain which can be divided into three separate domains, including the protease, apical and C-terminal domain. The enzymatic function of PSMA includes the hydrolysis of terminal glutaminyl residues of
N-acetyl aspartyl glutamate or folates, but its exact function in the salivary glands remains elusive [
19]. The majority of PSMA-targeted small molecules under investigation today are urea-based inhibitors consisting of a glutamate-urea-lysine moiety designed for PSMA targeting [
20]. In this regard, MSG was investigated preclinically for its potential ability to reduce binding of PSMA-targeted small molecules to PSMA in the salivary glands. Previous research already showed that MSG was indeed able to reduce the uptake of [
68Ga]Ga-PSMA-11 in salivary glands, while maintaining the tumor uptake. Therefore, it was suggested that MSG reduced the binding to PSMA by competing with off-target binding sites in healthy tissues such as kidneys and salivary glands [
11]. In our research, we confirmed that the co-incubation of [
177Lu]Lu-PSMA-617 with MSG indeed decreased the binding on salivary gland tissues. However, we also observed a decreased binding of [
177Lu]Lu-PSMA-617 when treated with higher MSG concentrations, suggesting that MSG might be able to reduce tumor uptake when dosing is too high. This data is in accordance with published reports, where decreased uptake was observed in patients treated with [
68Ga]Ga-PSMA-11 after the oral administration of MSG [
21].
We also investigated the ability of ionotropic and metabotropic glutamate receptor antagonists to reduce the binding of [177Lu]Lu-PSMA-617 on prostate and salivary gland cells as well as mouse kidney and pig salivary gland tissues. We showed that both on cells and tissues, the ionotropic glutamate receptor antagonist kynurenic acid and the metabotropic glutamate receptor antagonists (RS)-MCPG and (RS)-CPPG had a decreasing effect on the [177Lu]Lu-PSMA-617 binding. However, for (RS)-CPPG, the cytotoxicity assays showed a decreased cell survival for all (RS)-CPPG concentrations except 50 µM. Therefore, the observed reduced binding might be attributed to a lower number of cells present in these conditions.
Most PSMA-targeted small molecules today utilize a glutamyl residue to bind to the active site of PSMA. Therefore, an alternative suggested mechanism for off-target binding of PSMA-targeted small molecules in the salivary glands involves the anion/H
+ transporter SLC17A5 (sialin). Sialin is highly expressed in salivary glands and, when present in the plasma membrane of salivary gland cells, it can mediate the electrogenic co-transport of anions such as aspartate and glutamate [
22]. This anion co-transporter expressed in the salivary glands with an affinity for glutamate might be a reason for the off-target binding related to the glutamyl residues in PSMA-targeted small molecules, and for the effect of MSG and specific ionotropic and metabotropic glutamate receptor antagonists on the binding and retention of these small molecules in the salivary glands. The hypothesis of off-target accumulation of PSMA-targeted small molecules is further supported by the fact that antibodies designed to target PSMA, such as the monoclonal antibody J591, show no uptake in the salivary glands [
23].
One of the functions of PSMA encompasses the modulation of glutamate signaling via the metabotropic glutamate receptor pathway, resulting in the cleavage of glutamate from dietary folic acids and the neurotransmitter
N-acetyl-L-aspartyl-L-glutamate. This function could result in the affinity of glutamate-urea-lysine moieties for other PSMA-like proteins such as metabotropic glutamate receptors. A computational study suggested that PSMA-targeted molecules with a glutamate-urea-lysine moiety can indeed bind to metabotropic glutamate receptors and
N-acetyl-L-aspartyl-L-glutamate [
24]. In our study, we utilized several metabotropic glutamate receptor antagonists and found that only (RS)-MCPG (a group I and II metabotropic glutamate receptor antagonist) had an effect on [
177Lu]Lu-PSMA-617 binding. However, our study was limited to three or four concentrations of each of the antagonists tested, so it might be warranted to screen more metabotropic glutamate receptor antagonists with more variety in tested concentrations.
Another suggested mechanism for off-target accumulation of PSMA-targeted small molecules is cross-reactivity with glutamate carboxypeptidase III, which is highly expressed in both kidneys and salivary glands. Additionally, it shows a high degree of homology with the PSMA receptor [
25]. To further elucidate the exact mechanism of off-target accumulation of PSMA-targeted small molecules, it might be interesting to repeat similar experiments to those we performed with inhibitors of sialin receptors and glutamate carboxypeptidase III.
A major limitation of this study comprises the use of A-253 cells to investigate the binding of [
177Lu]Lu-PSMA-617 to salivary glands. First, these cells are derived from a carcinoma of the submandibular gland, although literature has described these cells as having maintained many healthy tissue traits, including aquaporin expression important for salivary gland function [
26]. However, it remains less than adequate to study healthy tissue characteristics on cancerous tissue models. But the fact is, there is a lack of proper healthy salivary gland cell models that are readily available. On top of this, studying salivary gland uptake in vivo also remains challenging because mice show a higher kidney PSMA expression profile compared to humans, acting as a sink for circulating [
177Lu]Lu-PSMA-617 [
27]. However, we used this characteristic and included mouse kidney tissue as a positive control in the autoradiography studies. Pig salivary glands are described to be close human homologues regarding salivary gland structure and function. Furthermore, pig PSMA shows a 91% sequence homology to the human PSMA protein [
28,
29]. Western blot and immunohistochemistry on several pig tissues showed similar expression profiles of PSMA compared to human tissue. More so, the ratio of PSMA expression between pig salivary glands and prostate cancer cells is also similar compared to humans as pig salivary glands are reported to have a 500-fold decrease in PSMA expression compared to prostate cancer cells [
13,
30]. However, a complete in vivo evaluation on more relevant animal models with higher homology in the PSMA expression pattern to human salivary glands, such as pigs, remains difficult from an ethical point of view [
23,
26].
A second limitation of using the A-253 cells is the very low uptake of [177Lu]Lu-PSMA-617, as indicated by our binding and internalization studies. The lack of proper cellular models hampers conducting mechanistic studies on why PSMA-targeted small molecules are binding to, and retained in, the salivary glands. This highlights the need for better models to investigate salivary gland uptake of PSMA-targeted compounds, which better represent the uptake pattern as seen in humans, facilitating the development of appropriate countermeasures.
4. Materials and Methods
4.1. Chemicals
PSMA-617 (vipivotide tetraxetan, HY-117410), 2-PMPA (HY-100788) were purchased from MedchemExpress (Monmouth Junction, NJ, USA). The sulforhodamine B (SRB) based In vitro toxicology assay kit (TOX6) and L-glutamic acid monosodium salt hydrate (monosodium glutamate, MSG, G5889) were purchased from Sigma-Aldrich (Overijse, Belgium). Kynurenic acid (S4719), memantine HCl (S2043), capric acid (S6906) and (-)-Dizocilpine (MK 801) maleate (S2857) were purchased from SelleckChem (Planegg, Germany). UBP 302 (2079) was purchased from Tocris Biosciences (Abingdon, UK). (RS)-MCPG (sc-202325), (RS)-CPPG (sc-203448), LY341495 (sc-361244A) and UBP1112 (sc-204368) were purchased from Santa Cruz Biotechnologies (Heidelberg, Germany). Compounds were dissolved either in H2O, 0.05 M NaOH or DMSO.
4.2. Cell Culture
PC3-Flu (PSMA-negative) and PC3-PIP (PSMA-positive) prostate cancer cells (kindly provided by Dr. Pomper, John Hopkins University, Baltimore, MD, USA) were grown in RPMI-1640 High glucose Low sodium bicarbonate medium (Gibco™, Thermo Fischer Scientific, A1049101, Geel, Belgium) supplemented with 10% FBS, 100 µmL penicillin-streptomycin and 2 µg/mL Puromycin [
12]. A-253/HTB-41 submandibular salivary gland cells were purchased from the American Type Culture collection (ATCC, Manassas, VA, USA) and cultured using McCoy’s 5A medium (ATCC, 30-2007) supplemented with 10% FBS and 100 u/mL penicillin/streptomycin. All cell lines were maintained in a humidified 37 °C incubator with 5% CO
2 and sub-cultivated when at 80–90% confluency.
4.3. In Vitro Tissue Models
Healthy mouse kidneys were dissected from BALB/c mice. The mice were sacrificed by an overdose pentobarbital (200 µL of 60 mg/mL) after which the kidneys were dissected and rinsed with saline to remove blood. Pig salivary glands were obtained in collaboration with the Medanex clinic (Diest, Belgium). After dissection, tissues of interest were either prepared for snapfreezing or paraffin embedding. For snapfreezing, tissues were embedded in Tissue Tek (Tissue-Tek O.C.T., Sakura Finetek Europe B.V, Alphen aan den Rijn, the Netherlands) and snapfrozen using 2-methylbutane at −40 °C. Tissues were stored at −20 °C until further processing. For paraffin embedding, tissues were fixed using 4% PFA and stored in 70% ethanol until further use. Next, tissues were dehydrated in an ethanol/xylol series and submerged in paraffin. After hardening, tissues were stored until further use. Cryosections of 20 µm thickness were sliced using a cryotome (Cryostar NX50, Thermo Fisher Scientific, Geel, Belgium) and tissue sections were mounted on Superfrost Plus microscope slides. Slides were then stored at −20 °C until further use. Paraffin sections were sliced using a microtome at 7 µm thickness and mounted on microscope slides. Sections were stored at room temperature until further use.
4.4. Radiolabeling of PSMA-617 with Lutetium-177
[
177Lu]LuCl
3 was purchased from ITG (Munich, Germany). The radiolabeling with DOTA-PSMA-617 was performed as described previously [
31]. In brief, DOTA-PSMA-617 was labeled with
177Lu at a molar activity of 50 MBq/nmol in 0.5 M sodium acetate, pH 5. The reaction mixture was placed in a Thermomixer C (Eppendorf, Hamburg, Germany) to shake (500 rpm) at 95 °C for 10 min.
The radiolabeling yield was determined using thin-layer chromatography. After cooling down, 2 µL of the reaction mixture was spotted on a glass microfiber chromatography paper strip impregnated with silica gel (iTLC-SG, Agilent Technologies, Diegem, Belgium). The elution was performed using a 0.5 M (pH 5.5) citrate solution after which the TLC papers were cut in half. Next, the activity of the top and bottom parts was counted using a 2480 WIZARD2 automatic y-counter (Perkin-Elmer, Mechelen, Belgium). Radiolabeling efficiency was calculated as: Radiolabeling yield = CPMBottom/(CPMTop + CPMBottom). Reactions with a radiochemical yield >95% were diluted in an appropriate buffer for further experiments. To test the stability of the radiolabeled product, [177Lu]Lu-PSMA-617 was diluted in several relevant media and tested at various time points after radiolabeling until 72 h using iTLC as described above.
4.5. Cell Binding and Internalization Assays
PC3-Flu, PC3-PIP and A-253 cells were seeded at 50,000 cells/well in a 24-well plate and incubated at 37 °C and 5% CO
2 in order for the cells to adhere to the surface. Cells were then treated with 5 nM [
177Lu]Lu-PSMA-617 and 5 µM of the highly potent PSMA-inhibitor 2-PMPA or vehicle to distinguish between specific and non-specific binding. After 4 h of incubation at 37 °C and 5% CO
2, cells were washed twice with PBS (corresponding to the wash fraction). Next, cells were incubated for 10 min at room temperature and washed twice with 50 mM glycine and 100 mM NaCl (pH 2.8) to collect the membrane bound fraction [
26]. Lastly, cells were incubated with 1 M NaOH for 30 min and washed twice to collect the internalized fraction. Then, the activity of the fractions was measured using a 2480 WIZARD
2 automatic y-counter. Activity measurements were converted to % added activity (%AA) and normalized for cell counts, performed by an automated MOXI Z cell counter, per cell type.
4.6. Tissue Saturation Binding Assay
The dissociation constant (Kd), the maximum density of receptors (Bmax) and the binding potential (ratio of Bmax over Kd) of mouse kidneys and pig salivary glands were determined using In vitro autoradiography. The autoradiography experiments were performed according to a published protocol [
13]. In brief, tissue sections were air dried before incubation with 170 mM Tris-HCl and 5 mM MgCl
2 (pH 7.4) for 10 min at room temperature. The slides were then incubated with increasing concentrations (10.5–1200 nM) of [
177Lu]Lu-PSMA-617 in 170 mM Tris-HCl, 5 mM MgCl
2 (pH 7.4) and 1% bovine serum albumin (BSA). In order to distinguish specific binding from non-specific binding, tissue sections were co-incubated with 100 µM 2-PMPA. After 1 h of incubation, the treatment was removed from the slides and slides were washed twice for 5 min in 170 mM Tris-HCl, 5 mM MgCl
2 (pH 7.4) and 1% BSA and once for 5 min in 50 mM Tris-HCl with 5 mM MgCl
2 (pH 7.4). Next, the slides were dipped in water, after which slides were dried using a heat gun. The activity on the tissue slides was measured using a Beaquant
®—real time autoradiography machine (AI4R, Nantes France). To determine the amount of activity corresponding to the measured counts, 1 µL of treatment solution was pipetted onto a microscope slide, air dried and measured together with the samples. From the known molar activity of the treatment stock, the corresponding relative concentration (fmol/mm
2) of the PSMA receptor could be calculated. The Kd and Bmax values for each tissue were calculated using a non-linear regression model in GraphPad Prism 9. The BP was calculated by dividing Bmax over Kd.
4.7. Cytotoxicity Assay of Monosodium Glutamate, Ionotropic and Metabotropic Glutamate Receptor Antagonists
The cytotoxicity of the different compounds under investigation was determined using the sulforhodamine B (SRB) based In vitro toxicology assay kit (TOX6, Sigma-Aldrich) as described below for the different compounds.
PC3-Flu and PIP cells were seeded at 10,000 cells/well, while A-253 cells were seeded at 15,000 cells/well in a 96-well plate. After overnight incubation at 37 °C and 5% CO2, cells were treated as follows.
To determine the cytotoxicity of MSG, the cells were incubated with increasing concentrations of MSG (0.9, 9, 90, 900 and 9000 µM) for 4 h. Next, the cells were washed twice and then fixed by adding 50% trichloroacetic acid (¼ of well volume) to the cell cultures. After 1 h of incubation at 4 °C, the wells were washed three times with water and air dried. Next, fixed cells were stained using 0.4% sulforhodamine B solution. After 30 min incubation at room temperature, cells were washed twice with 1% acetic acid wash solution and the incorporated dye was solubilized in 10 mM Tris. The absorbance was then measured using a CLARIOstar plate reader (BMG Labtech, Ortenberg, Germany). Cell survival was calculated relative to the untreated control condition and data were represented as a cell survival percentage. The cytotoxicity of ionotropic and metabotropic glutamate receptors was investigated as described above and the tested concentration for each compound is summarized in
Table 3.
4.8. Blocking Studies Using Monosodium Glutamate, Ionotropic and Metabotropic Glutamate Receptor Antagonists on Cells
PC3-PIP cells were seeded at 10,000 cells/well, while A-253 cells were seeded at 15,000 cells/well. After overnight incubation, the cells were treated with 5 nM [
177Lu]Lu-PSMA-617 and different concentrations of MSG and ionotropic and metabotropic glutamate receptor antagonists as summarized in
Table 4. To compare the blocking capacity of the different tested compounds, the cells were also treated with 5 µM 2-PMPA. The treatment was incubated for 4 h at 37 °C with 5% CO
2. Cells were then washed twice with PBS and the total bound fraction (membrane bound + internalized fraction) was collected by incubating the cells with 1 M NaOH for 30 min and washing them twice. Radioactivity in the collected samples was measured using a 2480 WIZARD
2 automatic y-counter (Perkin-Elmer, Mechelen, Belgium). Data are represented as the relative binding percentage normalized to the untreated control condition.
4.9. Blocking Studies Using Monosodium Glutamate, Ionotropic and Metabotropic Glutamate Receptor Antagonists on Tissue
To validate the results found with the cellular blocking experiments, similar experiments were performed on mouse kidney and pig salivary gland tissue sections.
A similar protocol as described above for In vitro autoradiography was followed. Tissue sections were incubated with 5 nM [
177Lu]Lu-PSMA-617, a vehicle (water or 2% DMSO) and different concentrations of the blocking compounds as summarized in
Table 5. After a 1 h treatment of incubation and washing steps, sections were exposed using a Beaquant
®—real time autoradiography machine (AI4R, Nantes, France). From the obtained counts per area, the relative binding percentage was calculated and normalized against the untreated vehicle condition. Data are represented as the relative binding percentage.
4.10. Immunohistochemistry
To characterize the different salivary glands sections, multiple tissue stainings were performed on paraffin-fixed tissues.
For all stainings, tissue slices were deparaffinized and rehydrated using xylol/ethanol and washed with demineralized water. For the hematoxylin and eosin staining, slices were then stained for 5 min with hematoxylin solution (105175, Merck millipore, MA, USA), washed and counterstained with Eosin Y-solution 0.5% aqueous (109844, Merck millipore, MA, USA). For the periodic acid—Schiff (PAS) staining, slides were incubated for 5 min in periodic acid 0.5% solution (100482, Merck millipore, MA, USA), washed and incubated for 15 min in Schiff’s reagent (3952016, Sigma-Aldrich, Hoeilaart, Belgium). Slides were then counterstained with hematoxylin. After staining and dehydration using ethanol/xylol, slides were mounted with a cover glass, dried and stored until imaging. Imaging was performed using a BioTek Cytation 5 (Agilent Technologies, Mechelen, Belgium)
4.11. Statistical Analysis
Tissue experiments were performed once in triplicate. Cell experiments were performed twice in triplicate. Tissue saturation binding experiments were analyzed using non-linear regression. Cytotoxicity data were analyzed using a 2-way ANOVA compared to the untreated control condition. Blocking studies were analyzed using a 2-way ANOVA with multiple comparisons and a Tukey post hoc test. For all experiments, outliers were detected and removed according to Tukey’s fences by the following formula: [Q1 − k(Q3 − Q1), Q3 + k(Q3 − Q1)], where k = 1.5. All statistical tests were performed using GraphPad Prism 9. Data were visualized as mean ± SD.