Because of the strong relevance of androgens in all stages of prostate cancer (PCa), androgen deprivation therapy (ADT) is a mainstay in the treatment of metastatic PCa. However, progression to castration-resistant prostate cancer (CRPC) frequently occurs after 18 to 36 months on average [1
]. The process of CRPC is connected to various mechanisms, as reactivation of androgen signaling under castrated conditions, gain-of-function mutations within the androgen receptor (AR) gene, increased AR expression [3
], and elevated intratumoral androgen biosynthesis through upregulation of androgen-synthesizing enzymes like aldo-keto reductase family 1 member C3 (AKR1C3) [6
]. The development of novel anti-androgens such as abiraterone acetate or enzalutamide significantly improved the management of CRPC. These anti-androgens intervene with the AR signaling pathway via different mechanisms. Abiraterone acetate inhibits androgen biosynthesis by targeting cytochrome P450 17A1 (CYP17A1, also known as 17α hydroxylase/17,20-lyase), which catalyzes the conversion of pregnenolone to dehydroepiandrosterone (DHEA) in the adrenal glands [1
]. Enzalutamide inhibits androgen signaling through competitive binding to the AR, resulting in inhibiting translocation of the receptor into the nucleus and its binding to the DNA, leading to a reduced expression of AR-regulated target genes [2
]. Both drugs achieved stunning results. Notably, enzalutamide, which has been previously reserved for patients with metastatic CRPC based on the AFFIRM (2012) [10
] and the PREVAIL (2014) [2
] trials, has recently been released by the FDA to the use in all patients with CRPC—not metastatic and metastatic—based on the results obtained in the PROSPER trial [11
]. Unfortunately, resistances also occur with these novel drugs [12
]. An intriguing potential resistance mechanism is the occurrence of AR splice variants, the most frequent being AR-V7. These constitutively active splice variants lack the ligand-binding domain (LBD) and therefore become insusceptible to enzalutamide, which—like androgens—bind to the LBD within the AR protein [13
]. Hence, there is an urgent need to develop novel agents that are able to suppress both full-length AR (AR-FL) and AR-V7. Besides AR variants, increased AR expression, mutations within the AR gene (F876L) [17
], activation of the interleukin-6-stat3-AR axis [14
], elevated expression of the glucocorticoid receptor (GR) [18
], and aberrant intratumoral androgen synthesis [19
] are likely causes of anti-androgen resistance.
A number of different enzymes are involved in the biosynthesis of intratumoral androgens like CYP17A1 and AKR1C3. Various studies including recently published findings by our group associated increased expression of AKR1C3 with CRPC [6
] as well as with resistances to enzalutamide [14
] and abiraterone acetate [24
]. This key enzyme in steroid biosynthesis, also known as 17β-hydroxysteroid dehydrogenase type 5 (17β-HSD5, HSD17B5), catalyzes the conversion of weak androgen precursors, Δ4-androstene-3,17-dione and 5α-androstane-3,17-dione into the more potent androgens testosterone and 5α-dihydrotestosterone [6
]. Since it acts downstream of CYP17A1, targeting AKR1C3 would not cause an accumulation of deoxycorticosterone in the adrenal glands, and therefore, the co-administration of prednisone would not be necessary [25
]. In fact, various AKR1C3 inhibitors have been developed in the past as reviewed by Penning et al. [1
]. Of notice, some of them were found to exhibit dual-inhibitory function on AKR1C3 and the AR [30
]. The AR is a ligand-inducible nuclear steroid receptor transcription factor for testosterone and dihydrotestosterone that consists of an N-terminal domain (NTD) (carrying the constitutive activation function region), a DNA binding domain (DBD), a hinge region, and a ligand binding domain (LBD), the latter being the best studied section of the protein [34
]. The AR is an especially plastic protein, with the ability to rapidly change between several conformations [8
]. In addition, despite the relatively large number of available crystal structures (82 entries are reported in RCSB PDB (Protein Data Bank) for human AR, accessed 03.04.2020), structural information is mainly available on the LBD. As a matter of fact, most AR targeting drugs were designed to directly bind to this domain [34
]. There have been several efforts in the past to develop novel AR antagonists that bind to the NTD or the DBD. However, these are tricky due to the largely unstructured and unstable nature of the NTD and the high sequence similarity (77–80%) of the DBD with other steroid receptors like GR [13
The aim of our study was to investigate the anti-neoplastic effects of three natural AKR1C3 inhibitors, which originated from the plant Melodorum fruticosum
, in PCa with an emphasis on their efficacy in overwhelming enzalutamide resistance. To this end, we used 22Rv1 cells, which are enzalutamide resistant and express AR-V7 [37
] and an enzalutamide-resistant PCa cell line (DuCaP EnzaR) that has been previously established in our lab [38
]. In addition, we used a co-culture spheroid model consisting of DuCaP cells grown together with cancer-associated fibroblasts (CAFs) as 3-dimensional spheroids. Previous findings showed that DuCaP cells acquire enzalutamide resistance when co-cultured with CAFs, which is associated with increased expression of AKR1C3 [23
]. In addition, we demonstrate a bi-functional effect of MF-15, the most potent inhibitor, on ARK1C3 and AR, as shown by considerable downregulation of androgen-regulated genes, reduced production of prostate specific antigen (PSA), and a decrease in the expression of AKR1C3, full-length AR (AR-FL) and AR-V7. Furthermore, we provide evidence that MF-15 acts on the AR through binding to the DBD as revealed by molecular docking simulations and reporter gene assays. All experiments were conducted in charcoal-stripped serum to simulate conditions with ongoing ADT in the state of castration-resistance, as performed in clinical routine. These data let us conclude that MF-15 represents a novel potential AKR1C3 inhibitor with bi-functional activity on AKR1C3 and AR and may, therefore, be used to overcome enzalutamide-resistance in PCa.
Resistances to potent anti-androgens like enzalutamide and abiraterone acetate are becoming a growing problem in the treatment of advanced prostate cancer. Therefore, the development of new drugs, which would help to overcome these resistances, is largely demanded. AKR1C3 has gained increasing attraction as a novel target based on studies demonstrating its impact on prostate tumor progression, mCRPC and development of resistances to enzalutamide and abiraterone acetate [14
Several potent AKR1C3 inhibitors, including steroidal, non-steroidal and natural compounds, have been developed so far [1
]. In the present study, we tested the antineoplastic effects of three chalcones in prostate cancer. They showed significant inhibitory activity on AKR1C3 in enzymatic assays with dihydrochalcone MF-15 as the most effective compound. This may be due to structural differences among the three compounds. In contrast to MF-11, MF-14 and MF-15 are 2′-(2-hydroxy)-benzylated. In addition, MF-15 is lacking the α-β double bond within the chalcone skeleton, thereby adding more flexibility to the overall structure of the protein. It therefore seems likely that for a favored protein binding of these compounds, both a 2′-(2-hydroxy)-benzyl moiety together with a rotable α-β single bond are beneficial.
In vitro cell culture experiments revealed that 10 µM MF-15 significantly inhibited various prostate cancer cell lines, including enzalutamide resistant DuCaP EnzaR and 22Rv1 cells, though with varying response rates. Western blot analysis, however, revealed substantial levels of AKR1C3 in all three cell lines, thereby suggesting that a low response rate to MF-15 can hardly be explained by faint AKR1C3 expression. In addition, we showed here that MF-15 also effectively prohibited 3D spheroid growth. Most importantly, MF-15 was also highly effective in inhibiting spheroid growth of DuCaP/CAF and DuCaP EnzaR co-culture spheroids, which were previously shown to exhibit increased AKR1C3 expression associated with enzalutamide resistance and a strong pro-inflammatory phenotype [23
]. In comparison with MF-15, two purchased AKR1C3 inhibitors, indomethacin and AKRi, only moderately affected cell and spheroid growth even when applied at concentrations of 20 µM and 50 µM. Previous studies have shown that indomethacin alone did not inhibit 22Rv1 cells at concentrations up to 20 µM, but efficiently re-sensitized cells to enzalutamide in vitro and in vivo [14
]. Indomethacin is a known AKR1C3 inhibitor, which also inhibits cyclooxygenases (COX-1, COX-2) and which is widely used as a nonsteroidal anti-inflammatory drug in the treatment of fever, pain or inflammation [29
Another interesting aspect of some previously described AKR1C3 inhibitors is their bifunctional activity on AKR1C3 and the AR. The group of Verma et al. for instance reported on the antineoplastic activity of an AKR1C3 inhibitor called KV-37, which reduced prostate tumor cell growth in vitro and in vivo associated with a reduction of PSA expression and reduced AR transactivation [22
]. In our study, MF-15 also significantly inhibited the expression of two androgen-regulated genes, FKBP5 and PSA, in enzalutamide-resistant 22Rv1 cells. In addition, we found that MF-15 significantly reduced PSA production and AR reporter gene activity associated with downregulation of AR protein, suggesting a direct impact on AR signaling. Even more important, MF-15 also inhibited AR-V7 reporter gene activity and induced a downregulation of AR-V7 expression in 22Rv1 cells. In numerous previous studies, the occurrence of the constitutively active AR-V7 splice variant lacking the C-terminal AR-LBD has been strongly linked to prostate tumor progression as well as to failures in the response to enzalutamide [15
]. Furthermore, the presence of AR-V7 mRNA in circulating tumor cells of patients with metastatic CRPC has been shown to predict poor response to AR antagonists [15
]. Drugs having the ability to suppress both AR-FL and AR-V7, and would therefore be of much better therapeutic efficacy, in particular in CRPC. MF-15 was shown to significantly and dose-dependently inhibit AR-FL and AR-V7, whereas enzalutamide—as expected—failed to inhibit reporter gene activity in the presence of AR-V7. Since patients, which are positive for AR-V7, usually show substantially low AR-V7 compared to relatively high AR-FL transcript levels at the same time [54
], we also used various ratios of AR-FL:AR-V7 (95:5, 80:20) to mimic a more patient-like situation in our experimental settings. Of notice, not only was the basal reporter gene activity of AR-V7 much higher than that of AR-FL but, even more importantly, the inhibitory effect of enzalutamide decreased with increasing expression levels of AR-V7. Because MF-15 is also able to inhibit AR-V7 lacking the AR-LBD, it is highly considered that MF-15 acts via a different mechanism than enzalutamide, which is known as a competitive AR antagonist that binds within the AR-LBD [9
]. In fact, further AR activity assays revealed that MF-15 non-competitively inhibits the AR. This may also be a possible reason why a combination of MF-15 and enzalutamide did not give a benefit over the effects of MF-15 alone. Additionally, it should be noted that we only tested MF-15 at a concentration of 10 µM, which was found to be highly effective when the drug was administered alone. Further experiments with much lower drug concentrations are warranted to clearly define a combination of MF-15 with AR-targeting agents.
Molecular docking simulations on known binding sites of the AR-DBD were conducted to propose a mechanism of action for MF-15 on the AR. The best results were found directly in the AR-DBD. These findings were further corroborated by reporter gene assays using the glucocorticoid receptor, which exhibits high similarity with the AR in its DBD [13
]. MF-15 significantly inhibited GR activity in the absence and presence of dexamethasone, indicating that MF-15 binds to the DBD. Although a binding of MF-15 in the AR-NTD of course cannot be completely excluded, it should be further considered that the NTD of the AR represents an intrinsically disordered region, rendering the protein highly flexible, thereby allowing to form protein–protein interactions with multiple partners. In fact, more than 160 proteins including some inhibitors are known to interact with the AR-NTD [57
]. Due to the high flexibility of this domain, however, it is unlikely that a non-covalent binder like MF-15 could induce a potent inhibitory effect on this part of the protein.
An AKR1C3 inhibitor like MF-15 with dual-inhibitory function on AKR1C3 and AR as well as on truncated AR variants holds the advantage of simultaneously inhibiting intratumoral androgen biosynthesis and the AR signaling cascade, thereby covering the action of antiandrogens such as enzalutamide and abiraterone acetate. Nevertheless, it should be noted that the growth-inhibitory effects of MF-15 were not restricted to AR-positive prostate cancer cells, since also AR-negative PC-3 cells were affected by MF-15 in our study, exerting a similar effect as the chemotherapeutic drug docetaxel. Anti-proliferative effects of ARK1C3 inhibitors in PC-3 cells have previously been shown by several other groups [58
]. A potential mechanism was postulated by Sekine and coworkers, who showed that meclofenamic acid inhibited cell growth through attenuation of IGF-1-induced Akt activation [60
Another advantage of an AKR1C3 inhibitor like MF-15 would be that a co-administration of prednisone, as it is common to prevent side effects of abiraterone acetate, would not be necessary [25
]. MF-14 and other chalcones from Melodorum fruticosum
have been shown to also possess significant anti-inflammatory properties, inhibiting microsomal prostaglandin E synthase-1 (mPGES-1), which represents a key enzyme in the synthesis of pro-inflammation prostaglandins [61
]. Based on the knowledge that inflammation is correlated with tumor development and progression in general, a dual inhibition of mPGES-1 and AKR1C3 might represent a promising point in action [43
]. Moreover, MF-15 had a strong effect in our co-culture spheroid models, which are characterized by a strong pro-inflammatory phenotype [23
]. In summary, MF-15 is considered a novel and potent AKR1C3 inhibitor with additional AR targeting activity in the AR-DBD, thereby being potentially useful in the treatment of antiandrogen resistant CRPC.
From the clinical point of view, AKR1C3 inhibitors could be used in patients with enzalutamide resistance as a new additional treatment option, especially in cases where the patients do not meet the criteria for undergoing another mCRPC treatment like chemotherapy. However, clinical trials are warranted to proof the findings of this study.
4. Materials and Methods
4.1. Isolation of MF-11, MF-14 from Melodorum Fruticosum, and Preparation of MF-15
The leaves of Melodorum fruticosum were collected and extracted with CH2Cl2 as described in Engels et al. (61). The crude extracts (23 g) were subjected to silica gel column chromatography (424 g silica gel, Ø = 6.1 cm, L = 61 cm) and eluted with a gradient from petrol ether (100–0%), to ethyl acetate (0–100%), and methanol (0–100%), and collected time-dependently. The 170 obtained fractions were inspected with thin-layer chromatography using MF-14 as authentic standards and pooled to 10 comprehensive fractions (A-J). Fraction I (689 mg) was then separated over Sephadex® LH-20 (Ø = 2 cm, L = 92 cm) using CH2Cl2:acetone (85:15, v/v) as a mobile phase to yield nine comprehensive fractions (I-a—I-i). Finally, 132.8 mg of fraction I-h were separated over Sephadex® LH-20 (Ø = 1.5 cm, L = 70 cm) using methanol as a mobile phase to yield 35.1 mg MF-11 and 53.5 mg MF-14. MF-15 was obtained through catalytic hydration of MF-14. To this end, 23.8 mg MF-14 were dried and dissolved in methanol before the addition of Pd/C (Sigma Aldrich, Buchs, Switzerland) and H2 at room temperature under stirring. The reaction was monitored with thin-layer chromatography and stopped after 15 min by filtration. MF-15 was ultimately purified over Sephadex® LH-20 with methanol as mobile phase and its identity and purity confirmed by 1H NMR by comparison literature (60). NMR spectra were acquired on a Bruker Advance II 600 NMR spectrometer (Bruker Biospin Rheinstetten, Germany). Finally, 18.5 mg MF-15 were obtained with a purity of >95%.
4.2. Docking Parameters for the AR
Docking simulations were carried out on a crystal structure of the AR-DBD (PDB entry 1r4i, Shaffer 2004) [63
], using GOLD (version 5.2). Since there is no co-crystallized small molecule ligand in the crystal structure that could have been used for redocking, we aimed to recreate the docking pose of compound 25 from Li et al. [13
] to optimize the docking workflow. DNA was deleted from the binding site and in the final settings the binding site was defined in a 9 Å radius around the coordinates 25.10, 81.08, 38.43. Goldscore was used as a scoring function.
4.3. Cell Lines
22Rv1 cells and PC-3 cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA) and cultured in RPMI (PAN-Biotech, Aidenbach, Germany) with 10% fetal calf serum (FCS, Gibco), 1% GlutaMAXTM
(Gibco) and 1% penicillin and streptomycin (Lonza). DuCaP cells were a gift from Prof. J. Schalken (Center for Molecular Life Science, Nijmegen, The Netherlands) and routinely cultured in RPMI with 10% FCS [36
]. Enzalutamide-resistant DuCaP EnzaR cells have been previously established and characterized in our lab [38
]. Immortalized CAFs have been previously established and described elsewhere [64
] and were grown in RPMI with 10% FCS and 1% penicillin and streptomycin and 1x GlutaMAX™ (Gibco). CAFs were stably transfected with green fluorescent protein (GFP) by our group as described previously [41
]. All cell lines were cultivated at 37 °C in a humidified atmosphere with 5% CO2
. AR-V7 status of the used cell lines: 22Rv1: AR-V7 positive [37
], DuCaP: AR-V7 positive [38
], DuCaP EnzaR: AR-V7 positive [38
], PC-3: AR-V7 negative [65
Enzalutamide (MedChemExpress), R1881 (Steraloids Inc. Newport RI), the AKR1C3 inhibitor (AKRi) (3-(4-trifluoromethyl)phenylamino) benzoic acid, Calbiochem), indomethacin (Sigma-Aldrich), dexamethasone (Selleck Chemicals), docetaxel (MedChemExpress) and mifepristone (Selleck Chemicals) were dissolved in dimethyl sulfoxide (DMSO). All solvents used for the isolation were purchased from VWR International (Darmstadt, Germany).
4.5. Cell Viability
Cell viability was determined via CellTiter 96® Aqueous one solution cell proliferation assay (Promega). Briefly, 10 µL of reagent were added to 100 µL of cell culture medium, and absorbance was measured at 490 nm on a Cytation™ 5 Cell Imaging Multi-Mode Reader (BioTek). In each individual experiment, changes in cell viability were expressed as percentage of mock control.
4.6. Spheroid Culture
3D spheroids were established as described previously [23
]. Briefly, 8000 cells were seeded into each well of a 96-well ULC ultralow attachment plate (Costar, 7007) and cultivated at 37 °C in a humidified atmosphere with 5% CO2
. To obtain co-culture spheroids, DuCaP tumor epithelial cells were seeded together with CAFs at a ratio of 1:1 as previously described [41
]. Spheroid size was determined with IncuCyte®
S3 LiveCell Analysis System.
4.7. Real Time Quantitative RT-PCR (qPCR)
Total RNA was isolated from cells with ExtractMe total RNA isolation Kit (Blirt, Gdansk, Poland) and quantified with the NanoDrop ND-2000c (Thermo Scientific). RNA was transcribed into cDNA by reverse transcription using SuperScript III reverse transcriptase (Invitrogen). qPCR was performed with TaqMan™ Assays (Thermo Fisher Scientific) for the quantification of FKBP5 (Hs01561006_m1), AKR1C3 (Hs00366267_m1), and hydroxymethylbilane synthase (HMBS, Hs00609297_m1), which was used as the endogenous control. Primer and probe sequences of AR-FL, AR-V7 and PSA were as follows: AR-FL (forward 5′-AGGATGCTCTACTTCGCCCC-3′, reverse 5′-ACTGGCTGTACATCCGGGAC-3′, probe 5′-FAM-TGGTTTTCAATGAGTACCGCATGCACA-TAMRA-3′), AR-V7 (forward 5′-CGGAAATGTTATGAAGCAGGGATGA-3′, reverse 5′-CTGGTCATTTTGAGATGCTTGCAAT-3′, probe 5′-FAM-GGAGAAAAATTCCGGGT-TAMRA-3′), PSA (forward 5′-GTCTGCGGC GGTGTTCTG-3′, reverse 5′-TGCCGACCCAGCAAGATC-3′, probe: 5′-FAM-CACAGCTGCCCACT GCAT CAGGA-TAMRA-3′). qPCR was carried out with ABI Prism 7500 Fast RT-PCR System (Applied Biosystems) cycler. Fold change in gene expression was determined using the mathematical model ratio 2−ΔΔCT
]. Values of genes of interest (GOI) were determined relative to HMBS. “Fold change” expression was calculated relative to the mock control (set 1) for each individual experiment.
4.8. Western Blotting
Cells were lysed using RIPA buffer supplemented with 0.5 mM PMSF, 2 µg/mL leupeptin, 2 µg/mL aprotinin, 2.5 mM NaF and 1% Triton X-100 by shaking at 4 °C for 1 h. Fifty micrograms of protein were separated using 4–12% NuPAGE Bis-Tris protein gels (Thermo Fisher ScientificTM, Vienna, Austria) and transferred onto a 0.2 µm nitrocellulose membrane (GE Healthcare, Vienna, Austria). Blocking of membranes and antibody incubation was performed in StartingBlockTM (PBS) blocking buffer (Thermo Fisher ScientificTM, Vienna, Austria) using the following antibodies: anti-AKR1C3 (clone NP6.G6.A6, 1:500, Sigma), anti-androgen receptor (1:500, Cell Signaling), anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH, 1:50000, Millipore). Visualization and quantification of protein bands were performed with Image Studio software Version 5.2 (LI-COR Biosciences).
4.9. Luciferase Reporter Gene Activity Assay
To assess reporter gene activity, AR negative PC-3 cells were transiently transfected with a CMV promoter-driven pFlag tagged full-length AR (AR-FL), the AR truncated splice variant 7 (AR-V7) or GR (pCMV6-XL5NR3C1, Origene), and a NanoLuc luciferase reporter gene under the control of a promoter element consisting of two consecutive canonical androgen response elements (AREs) and a TATA box (pNL1.1.-ARE2TATA) using ViaFect lipofection reagent (Promega). Cells were then seeded into 96-well plates in RPMI + 2% FCS at 37 °C overnight. Then, 0.2 nM R1881, 2.5 µM enzalutamide or 10 µM MF-15 were added. After 24 h, luciferase reporter gene activity was determined with Nano Glow Dual Assay (Promega) and measuring absorbance on a CYTATION multiplate reader instrument.