Therapeutic potential of TRPM8 antagonists in prostate cancer

Transient receptor potential melastatin-8 (TRPM8) represents an emerging target in prostate cancer, although its mechanism of action remains unclear. Here, we have characterized and investigated the effects of TRPM8 modulators in prostate cancer aggressiveness disclosing the molecular mechanism underlying their biological activity. Patch-clamp and calcium fluorometric assays were used to characterize the synthesized compounds. Androgen-stimulated prostate cancer-derived cells were challenged with the compounds and the DNA synthesis was investigated in a preliminary screening. The most effective compounds were then employed to inhibit the pro-metastatic behavior of in various PC-derived cells, at different degree of malignancy. The effect of the compounds was then assayed in prostate cancer cell-derived 3D model and the molecular targets of selected compounds were lastly identified using transcriptional and non-transcriptional reporter assays. TRPM8 antagonists inhibit the androgen-dependent prostate cancer cell proliferation, migration and invasiveness. They are highly effective in reverting the androgen-induced increase in prostate cancer cell spheroid size. The compounds also revert the proliferation of castrate-resistant prostate cancer cells, provided they express the androgen receptor. In contrast, no effects were recorded in prostate cancer cells devoid of the receptor. Selected antagonists interfere in non-genomic androgen action and abolish the androgen-induced androgen receptor/TRPM8 complex assembly as well as the increase in intracellular calcium levels in prostate cancer cells. Our results shed light in the processes controlling prostate cancer progression and make the transient receptor potential melastatin-8 as a ‘druggable’ target in the androgen receptor-expressing prostate cancers.

www.nature.com/scientificreports/ PC is the second most frequently diagnosed cancer in men and several therapeutic approaches are currently available for patient's care. Nevertheless, PC might escape the treatments, giving rise to more aggressive forms of cancer. At this stage, PC often spreads, with a significant increase in the mortality rate 16 . Therefore, new approaches are needed for a better pharmacological management in PC patients with advanced disease.
TRPM8 gene was firstly identified as a reporter of the androgen receptor (AR) transcriptional activity in PC cells 17 and subsequently proposed as a PC biomarker. Its expression, indeed, increases during the initial stages of the disease, to decline after androgen deprivation therapy 18 . Although TRPM8 expression depends on AR transcriptional activity 19 , biochemical findings have reported a direct TRPM8 interaction with androgens 20 or the androgen receptor (AR) 21 . Nowadays, regulation of TRPM8 by the androgen/AR axis remains debated, although pharmacologic modulation of the channel is frequently proposed in PC therapy 20,22-24 . Our groups investigated the rapid androgen action in quite divergent cell types [25][26][27][28] and designed several TRPM8 modulators, characterized by selectivity towards other TRP channels, the highest 'in vitro' potency and metabolic stability so far described [29][30][31][32] .
Here, we report the effect of selected TRPM8 modulators and their analogues in various PC-derived cell lines, including LNCaP cells that represent a suitable model to investigate the role of sex steroids in PC-derived cells 33 . TRPM8 antagonists reverse the increase in calcium influx and the rapid signaling activation leading to the androgen-elicited invasion and proliferation in various PC-derived cells as well as to the increase in LNCaP spheroid size. These findings, together with the observation the TRPM8 modulators inhibit the proliferation of AR-expressing castrate resistant prostate cancer (CRPC) cells, while leaving unaffected the behavior of ARnegative PC cells, offer new insights into the knowledge of TRPM8 in PC pathogenesis and pave the way for novel promising strategies in clinical management of PC patients.

Results
In vitro pharmacological characterization. The molecular structure of tested compounds is shown in Figure S1 (Supplementary Information). They were synthesized following the procedures described in Figure S2 (Supplementary Information). Potency of these compounds over TRPM8 channels was evaluated by patchclamp assays. Compounds 3 and 6 have been previously tested for their agonist and antagonist effect, respectively, over TRPM8 (Table S1, Supplementary Information). Compound 3 showed a dose-dependent agonistic activity with higher potency than L-Menthol (EC 50 = 40 μM) 30 , while compound 6 is the most potent in vitro TRPM8 antagonist described so far with an IC 50 of 0.2 nM 29 . Both, agonist and antagonist were chosen as reference compounds to identify the type of TRPM8 modulator more suitable to produce an antiproliferative effect in PC cells. This matter is widely debated, since both TRPM8 agonist and antagonist have been proposed for PC treatment, albeit a different mechanism seems to be involved 9 .
In addition, three new analogs (4, 5, and 9) were synthesized and characterized by patch clamp electrophysiological assays. These chemotypes are particularly attractive due to the potent and selective cytotoxicity shown in a previous drug discovery screening 34 . Compounds 4 and 5 showed high potency as TRPM8 antagonist with a IC 50s in the high nanomolar range (204-340 nM, Table S1 and Figure S3 in Supplementary Information). The inhibitory effects of both compounds on menthol-evoked TRPM8 currents were concentration-dependent (Figure S3 in Supplementary Information). Indeed, compound 9 proved to be an effective concentration-dependent TRPM8 antagonist although less potent, with an IC 50 in the low micromolar range (5.0 mM, Table S1 and Figure S3 in Supplementary Information). All the compounds behaved as reversible inhibitor, since menthol-evoked currents were fully recovered during the drug's washout.
The typical drawback of TRPM8 modulators is represented by the lack of selectivity over other TRP channels, particularly TRPA1 and TRPV1. Indeed, the same prototypical TRPM8 agonists (menthol) and antagonists (AMTB), modulate the activity of different TRP channels subtypes 35,36 . To address this issue, the selectivity of synthesized compounds was challenged by calcium fluorometric assays. Compounds 3 and 6 selectivity has been previously assessed 29,30 . Accordingly, the newly synthesized compounds showed no agonistic or antagonistic effect when tested in cell lines expressing TRPA1 or TRPV1 channels ( Figure S4 in Supplementary Information). Furthermore, the sodium channel Nav1.7 is involved in PC cell invasiveness 37 . To rule out the effect of synthesized compounds over this channel, calcium fluorometric assays using HEK-293 stably expressing the human isoform of Nav1.7 channel were performed. As in the previous assays, tested compounds were unable to act as agonist or antagonist of this channel ( Figure S4 in Supplementary Information).

TRPM8 modulators inhibit the DNA synthesis and cell-cycle progression induced by androgens in various AR-expressing PC cells.
We then used PC-derived LNCaP cells, which express AR and the isoform β of ER ( Figure S5 in Supplementary Information) and respond to androgens and estrogens in terms of proliferation 28 . In a first attempt, we evaluated the expression of TRPM8 in these cells. Stimulation of quiescent LNCaP cells with 10 nM of the non-aromatizable androgen R1881 very weakly increased the amount of TRMP8, in the presence of similar amounts of tubulin, as assessed by Western blot analysis of lysate proteins (Fig. 1a). We next analyzed the effect of all the compounds on DNA synthesis induced by androgens in LNCaP cells. Quiescent cells were left untreated or treated with 10 nM R1881, in the absence or presence of increasing concentration (from 1 nM to 1 μM) of the indicated compounds. Almost all of them reversed the DNA synthesis stimulated by 10 nM R1881 in LNCaP cells (panels b-f in Fig. 1). The inhibitory effect observed by 1 μM of the TRPM8 antagonists 4 (Fig. 1d), or 6 ( Fig. 1e) or 9 (Fig. 1f) was stronger than that observed using the same concentration of the agonist 3 (Fig. 1b) or antagonist 5 (Fig. 1c). All the tested compounds did not affect the DNA synthesis when added alone to LNCaP cells (panels b-f in Fig. 1). Representative IF images captured from one experiment in Fig. 1d,e shown in Fig. 2a. Of note, our results are consistent with the observation that WS-12, a potent and selective TRPM8 agonist, slightly modify the viability of LNCaP cells 15  www.nature.com/scientificreports/ exert a remarkable effect in the androgen-sensitive cell lines 23 . The higher efficacy of compounds 4, 6 and 9, as compared to the analogue 5, might be due to their different potency and efficacy as TRPM8 antagonists (Table S1 in Supplementary Information). Given the findings here presented, we investigated the compounds 4 and 6 throughout the manuscript (Table S1 in Supplementary Information).
One nM androgen concentration is optimal in BrdU incorporation studies reported in LNCaP cells 38 and AR occupancy is almost completely saturated by that ligand concentration 39 . Therefore, we verified the effect of low ligand concentration in our assay. A slight (almost 2-folds), but significant increase of the DNA synthesis was recorded in LNCaP cells stimulated with 1 nM R1881, as compared with the more robust increase (almost 3,5folds) in BrdU incorporation observed in cells stimulated with 10 nM R1881 (panels b-f in Fig. 1). These findings support a dose-dependent ligand effect. Noticeably, AR synthesis and degradation, the receptor's amount, the affinity as well as on/off rates of AR ligand interactions might influence the AR ligand binding activity 39 . However, the compounds 4 and 6 also reverted the effect by 1 nM R1881 (Fig. 2b).
Androgens upregulates cyclin D1 and simultaneously downregulates p27 in LNCaP cells 28 . Ten nM R1881 up-regulated cyclin D1 in LNCaP cells (upper section in panels c and d of Fig. 2). A simultaneous decrease in p27 levels was observed (middle section in panels c and d of Fig. 2). Consistent with data on the DNA synthesis, 1 nM R1881 slightly increased cyclin D1 expression, simultaneously with a p27 decrease (Fig. 2e,f). Thus, androgen modulation of cell cycle markers seems to depend on the ligand concentration. Irrespective of ligand concentration, compounds 4 (Fig. 2, panels c and e) and 6 ( Fig. 2, panels d and f) inhibited the androgen-triggered effect on cell cycle markers. Expectedly 25 , expression of Cdk4 was similar under the different experimental conditions (Fig. 2, panels c-f). Quantitative analysis from different Western blots was done and data presented in the Supplementary Information.
Since LNCaP cells express the isoform β of ER, while do not express ERα ( Figure S5 in Supplementary Information) 33,40 , we also analyzed the effect of compounds on estradiol-induced DNA synthesis in LNCaP cells. Estrogens induced a significant increase in BrdU incorporation, which was slightly inhibited by the compounds 4 or 6 (Fig. 2g). Both the compounds did not affect the BrdU incorporation in cycling or androgen-stimulated www.nature.com/scientificreports/ PC3 and DU145 cells ( Figure S5 in Supplementary Information). These cells, indeed, do not express AR and are insensitive to androgens, albeit they express similar amounts of TRPM8 ( Figure S5 in Supplementary Information). Thus, TRPM8 antagonists impair the mitogenic response and cell cycle progression controlled by AR in LNCaP cells, leaving unaffected the DNA synthesis in AR-negative PC cells. We next used normal epithelial prostate PNT-2 cells, which express appreciable levels of AR and TRPM8 ( Figure S6 in Supplementary Information). Although these cells are sensitive to androgen stimulation in terms of DNA synthesis, they scantly responded to our compounds ( Figure S6 in Supplementary Information), likely because of the differential regulation of TRPM8 activity reported in normal as well as transformed prostate cells 41 . At last, we used our selected compounds in C4-2B, 22Rv1 and DU-CaP cells. C4-2B cells derive from the LNCaP subline, C4-2 that acquires the metastatic potential targeting the bone when subcutaneously or orthotopically inoculated into both castrated and intact athymic male nude mice. In contrast with C4-2 cells that grow in the absence of androgens, but yet respond to manipulation of androgen levels, C4-2B cells mimic the natural course of PC progression from the androgen-dependence to -independence 42 . 22Rv1 cells, instead, derive from a xenograft that was serially propagated in mice after castration-induced regression and relapse of the parental, androgen-dependent CWR22 xenograft. 22Rv1 cells express the AR, release prostate specific antigen (PSA) and weakly proliferate in response to androgens 43 . At last, DU-CaP cells derive from a metastatic lesion to the dura mater of a patient with hormone refractory PC 44 . Albeit at different extent, the employed cell lines express significant levels of AR as well as TRPM8 (Fig. 3a). The compounds 4 and 6 inhibited the BrdU incorporation in cycling C4-2B (left section in Fig. 3b), with a more robust and significant effect exerted by the compound 4. We did not observe any effect in C4-2B cells treated with enzalutamide (right section in Fig. 3b) or in cells challenged with 10 nM R1881 (not shown), as these cells are androgen-independent. Figure 3 also show www.nature.com/scientificreports/ that the compounds 4 and 6 significantly inhibited the androgen-stimulated BrdU incorporation of 22Rv1 (c) or DU-CaP (d) cells. The finding that TRPM8 antagonists inhibit the mitogenesis of C4-2B (still expressing AR) or 22Rv1 (expressing both AR and its AR-V7 variant) 45 or DU-CaP (over-expressing AR) cells would be useful in clinical management of CRPC patients, as they often exhibit AR overexpression or the AR-V7 variant, which confers resistance to abiraterone acetate and enzalutamide 46 .

TRPM8 modulators impair the migration and invasion induced by androgens in LNCaP cells.
Androgens increase the motility and invasiveness of various cell types through rapid activation of signaling effectors involved in cell locomotion 27,40 . Therefore, we evaluated the effect of TRPM8 modulators on androgen-induced motility by wound scratch assay in LNCaP cells. A robust increase in motility was observed on hormone challenging of LNCaP cells, as the wound gap was significantly reduced (Fig. 4a). Albeit at different extent, TRPM8 modulators (at 1 μM) inhibited the wound-gap closure in androgen-treated cells (Fig. 4a), with a stronger effect exerted by the compounds 4 and 6. Quantification of wound scratch assay at different concentrations of TRPM8 modulators was also done and graphically shown (Fig. 4b). The resulting graph confirms that the compounds 4 and 6 exhibited a stronger effect as compared with that obtained by using the other modulators. Irrespective of the experimental conditions, the compounds 4 and 6 did not modify the wound closure in AR-negative PC3 ( Figure S7 in Supplementary Information) or DU145 cells ( Figure S8 in Supplementary  Information). We next used a Transmigration assay in Boyden chamber. Quiescent LNCaP cells were challenged with 10 nM R1881 and allowed to migrate, in the absence or presence of the indicated compounds (at 1 μM). Results in Fig. 4c (left panel) show that androgens increased by almost 3 folds the number of migrating cells. Compounds 4 and TRPM8 antagonists inhibit the androgen-induced increase of LNCaP cell spheroids size. We next generated miniaturized LNCaP cultures in extracellular matrix (ECM) and observed a 3D structure after 3 days culture of LNCaP cells (Fig. 5a). Spheroids were then left untreated or treated with 10 nM R1881, in the absence or presence of 1 μM of the TRPM8 antagonists, 4 or 6. Changes in dimension and structure of organoids were monitored for additional 15 days. Phase-contrast microscopy ( Fig. 5a), together with quantification of data ( Fig. 5b), shows that androgen stimulation resulted in a robust increase of the LNCaP cell spheroid size. The compounds 4 and 6 both abolished the androgen-induced effect (Fig. 5a,b), leaving unaffected this response when used in the absence of hormone (Fig. 5b). Consistent with our previous findings 47 , only a weak response on LNCaP spheroid size was detected at 1 nM R1881. The compounds 4 and 6 reverted this effect ( Figure S9 in Supplementary Information). In summary, findings here obtained further make the TRPM8 as a putative target in PC.

TRPM8 antagonists inhibit the rapid androgen action in LNCaP cells.
Androgens modulate the activity of TRPM8 through non-genomic mechanisms 21,48,49 . Therefore, we investigated the mode of action of our compounds. As readout of the androgen-induced gene transcription, we evaluated the effect of compounds 4 and 6 on the androgen-induced PSA secretion from LNCaP cells. By increasing the ligand concentration (from 0.1 to 10 nM R1881), a significant increase in ligand-stimulated gene transcription was detected. At 10 nM ligand concentration, a negligible inhibition of PSA secretion was detected upon addition of the compounds 4 or 6. By contrast, enzalutamide strongly inhibited the effect exerted by androgens. Neither the two TRPM8 antagonists, nor enzalutamide did modify such response when used alone (Fig. 6a). Thus TRPM8 antagonists do not act on gene transcription mediated by the ligand-bound AR. Because of these results, we next analyzed the effect of compounds 4 and 6 on the androgen-induced rapid actions in LNCaP cells. Albeit at different extent, 10 nM R1881 triggered within 10 min the focal adhesion kinase (FAK) activation as well as paxillin phosphorylation. Both these responses are involved in the hormonal signaling leading to cell motility 50 . Addition of compounds 4 (left section in b) or 6 (right section in b) impaired the androgen-elicited effect. The compounds did not affect FAK or paxillin phosphorylation when used alone. Using the same lysate proteins, we analyzed the effect of the two compounds on the androgen-triggered extracellular-regulated kinase (ERK) and 90 kDa ribosomal S6 kinase (RSK). Hormone stimulation of LNCaP cells activated within 10 min ERK and RSK, which are needed for the androgen-elicited mitogenesis 28,33 . Treatment of androgen-stimulated cells with the compound 4 (left section in b) or 6 (right section in b) inhibited both the ERK and RSK activation, leaving unaffected the phosphorylation/activation of the two effectors when used alone. We then compared the effect of TRPM8 antagonists or enzalutamide on the androgen-induced rapid actions in LNCaP cells. Panel c in Fig. 6 confirms that hormone stimulation rapidly triggered the activation of FAK, paxillin, ERK, RSK and Src tyrosine kinase. Addition of compounds 4 or 6 or enzalutamide impaired at similar extent the androgen-elicited effect. Hormone stimulation did not affect Akt activation, because of the PI3-K pathway deregulation in LNCaP cells 26 . In sum, while enzalutamide reverses both transcriptional and non-transcriptional events mediated by AR, the antagonists of TRPM8 only interfere in the rapid signaling circuits activated by androgens in target cells, leaving unaffected the AR-mediated gene transcription. These results raise the question of how the androgen/AR axis intersects TRPM8 in LNCaP cells. To address this issue, we investigated by Co-IP approach the complexation between AR and TRPM8 in LNCaP cells. In a preliminary time-course experiment, we observed that the maximal AR/TRPM8 Co-IP occurred within 10 min androgen stimulation of quiescent LNCaP cells to return to the basal level upon 20 min hormone challenging (not shown). Therefore, we used that time-point to investigate the effect of compounds 4 or 6. Regardless of experimental conditions, similar amounts of AR or TRPM8 were detected in lysate proteins (upper section in www.nature.com/scientificreports/ Fig. 6d). Ten minutes androgen stimulation significantly increased the Co-IP of AR with TRPM8. The compounds 4 or 6 abolished this association, while they did not affect the AR/TRPM8 complex assembly when used in the absence of hormone (lower section in Fig. 6d). Similarly, enzalutamide perturbed the androgen-induced AR/ TRPM8 complex assembly (Fig. 6e). The absence of AR or TRPM8 in lysate proteins immunoprecipitated with control antibodies confirms the specificity of our approach (middle section in d and panel e).

TRPM8 antagonists inhibit the androgen-triggered intracellular calcium increase in LNCaP cells.
We finally evaluated the impact of androgen-triggered AR/TRPM8 complex assembly on intracellular calcium levels. Imaging experiments using the Ca 2+ indicator Fluo 4-AM were then done. Fluorescence images (Fig. 7a) show that 10 nM R1881 increased within 240 s the intracellular Ca 2+ levels, as compared with the untreated, quiescent LNCaP cells. Compounds 4 or 6 reduced the androgen effect, while any response was detected by addition of TRPM8 antagonists to quiescent LNCaP cells. In summary, we posit that the androgen-triggered AR/TRPM8 complex assembly modulates the calcium channels, thereby inducing an increase in intracellular calcium levels. Interference in the androgen-activated AR/TRPM8 complex by TRPM8 antagonists reverses this effect (Fig. 7b).

Discussion
The role of TRPM8 in PC is not completely understood 9,15 . In this report, we aim at analysing the intersection of TRPM8 with androgen/AR axis in PC cells. Therefore, we used several PC-derived cell lines, representative of both androgen-dependent (LNCaP, 22Rv1 and DU-CaP cells) as well as the castrate-resistant C4-2B cells that are androgen-independent, albeit they still express AR. At last, PC3 and DU145 cells were used, since they do not express AR and are insensitive to hormone stimulation. Normal prostate PNT-2 epithelial cells were also employed. All the cell lines express appreciable amounts of TRPM8, as assessed by WB analysis of lysate proteins. Notably, LNCaP, 22Rv1, DU-CaP and C4-2B cells express AR and are almost all sensitive to androgens, except for C4-2B cells. In these cells, we have analyzed the biological effects of new synthesized compounds, which modulate the Ca 2+ permeable, non-selective cation TRPM8 channels. The presented findings show that www.nature.com/scientificreports/ nearly all the synthesized modulators inhibit the DNA synthesis elicited by androgens in LNCaP cells. Because of the higher potency and efficacy of compounds 4 and 6, we have used these compounds as the most suitable modulators throughout our experiments. PC often exhibits cyclin D over-expression or lose cell-cycle negative regulators, such as p21 and p27 51 . The compounds 4 and 6 significantly impair the androgen-induced up-regulation of cyclin D1 as well as p27 downregulation in LNCaP cells. The lead compounds specifically interfere with the androgen-induced mitogenic effects, leaving unaffected the DNA synthesis elicited by estrogens in LNCaP cells, likely because other channels, such as TRPM6 52 and TRPM4 53 are involved in the estrogen signaling. Of note, the compounds 4 and 6 do not affect the S-phase entry of AR-negative PC3 and DU145 cells, while they significantly inhibit the mitogenesis of cycling C4-2B cells or androgen-stimulated 22Rv1 and DU-CaP cells. These latter findings are of interest in therapeutic approach of PC, which often develops resistance to the androgen deprivation therapy (ADT), although still expressing AR. In other cases, instead, PC might exhibit ADT resistance, because of the AR-V7 variant expression. As before described (see the Result's section), C4-2B and 22Rv1 cells are representative of the two different clinical patterns. Figure 6. TRPM8 antagonists inhibit the rapid androgen action by disrupting the ligand-induced AR/TRPM8 complex assembly in LNCaP cells. In (a-e), quiescent LNCaP were used. When indicated the compounds 4 and 6 were used at 1 μM. Enzalutamide (enz) was added at 10 μM. In (a), cells were left untreated or treated with increasing R1881 concentration (0,1, 1 or 10 nM), in the absence or presence of the indicated compounds for 48 h. Conditioned media were collected and PSA was assayed as described in "Methods". Means and SEMs from three different experiments are shown. *p < 0.05 for the indicated experimental points versus the corresponding untreated control. In (b-e), cells were unstimulated or stimulated with 10 nM R1881, in the absence or presence of the indicated compounds for 10 min. Lysate proteins were analyzed by WB technique using the antibodies against the indicated proteins. P-FAK stands for Tyr 397-P-FAK; P-paxillin stands for Tyr 118-P-paxillin; P-RSK stands for P-Ser 380 RSK; P-ERK stands for phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204); P-Src stands for P-Tyr 416 Src and P-Akt stands for P-Ser 473 Akt. The filters were stripped and re-probed using anti tubulin antibody, as loading control. In the upper panel in (d) and left section in (e), lysate proteins were analyzed by WB using the antibodies against AR or TRPM8. Lysate proteins were immunoprecipitated using the anti-AR antibody  1 μM). Different fields were analyzed by fluorescence (upper images) or phase-contrast (lower images) microscopy, as described in "Methods". Representative images were acquired using a DFC 450C camera (Leica). Bar, 10 μM. In (b), the mechanism of TRPM8 antagonist action in PC cells is shown. Androgens rapidly induce association of AR with TRPM8 in androgen-sensitive PC cells. Once assembled, this complex leads to intracellular calcium increase and promotes proliferation as well as pro-metastatic phenotype of PC cells. By perturbing the androgen-challenged AR/TRPM8 complex assembly, TRPM8 antagonists (compounds 4 and 6) lower the intracellular calcium levels and impair the proliferation, invasion and growth in 3D of PC cells. www.nature.com/scientificreports/ Ten nM testosterone positively affects the TRPM8-mediated migration in PC cells 21 . Consistent with these findings, our selected compounds impair the motility and invasiveness induced by 10 nM R1881 in AR-positive PC cells, but not AR-negative PC3 or DU145 cells. At last, a strong inhibitory effect is observed in 3D model, as the compounds reverse the androgen-induced increase in LNCaP cell spheroid size, further supporting the applicative potential of 4 and 6 compounds.
Steroid hormones regulate the Transient Receptor Potential (TRP) channels through rapid, non-transcriptional mechanism. Nevertheless, few reports have so far investigated the non-transcriptional effects related to the TRPM subfamily of TRP proteins. As before discussed, TRPM4 and TRPM6 have been linked to nontranscriptional estrogen action in different cell types and different experimental settings. Pregnenolone sulfate, the mother of all steroid hormones, rapidly and reversibly activates TRPM3, a divalent-permeable cation channel, thus inducing a rapid calcium influx as well as insulin secretion from pancreatic islets 54,55 . These findings identified TRPM3 as the playmaker between steroidal and insulin endocrine systems. A role for pregnenolone sulfate in activation of TRPM3 has been recently confirmed in somatosensory afferent neurons 56 . Relevant to our data, 10 nM testosterone challenges the interaction between AR and TRPM8 within lipid rafts microdomains of PC3 cells engineered to overexpress AR 21 . We here dissected an endogenous non-genomic mechanism of the TRPM8 channel regulation by androgens in PC cells. Ten nanomolar R1881 induces within minutes a significant increase in AR/TRPM8 Co-Ip in LNCaP cells. Once assembled, the complex triggers activation of various downstream effectors involved in the androgen-signaling leading to proliferation and motility of target cells. These findings encourage the development and use of novel TRPM8 antagonists, which perturb the androgen-elicited rapid responses, without affecting the secretion of PSA and gene transcription in PC cells.
At last, to assess the effect of androgens on cytosolic Ca 2+ mobilization caused by TRPM8 activation we have also conducted Ca 2+ imaging experiments in LNCaP cells. Hormone stimulation of PC cells a rapid elevation of cytosolic Ca 2+ concentration in a pattern that depends on AR/TRPM8 complexation, since the TRPM8 antagonists, which perturb the complex assembly, reverse such increase. These results make TRPM8 as the molecular link between androgen and calcium systems. Noteworthy, the Ca 2+ signaling intersects sex steroid receptormediated events [57][58][59] and controls the behavior of cancer cells 60 . Thus, by altering the expression and/or activity of Ca 2+ channels, our selected antagonists affect Ca 2+ homeostasis with important consequences in cell outcome. These findings, together with the observation that TRPM8 sensitizes therapy-refractory models of PC to radio-, chemo-or hormonal treatments 15 , make TRPM8 as a 'druggable' candidate in PC. Therefore, the discovery of new selective TRPM8 antagonists represents a promising approach in PC patient's management. Moreover, our compounds combine the selective modulation of non-genomic actions mediated by AR with the targeted release of intracellular calcium. This bifunctional approach may be more effective than the currently used ADT. At last, the present results together with the identification of TRPM8 mRNA as a bloodstream signature for high-risk PC patients 61 , pave the way for new options in patient's stratification and treatment guidance.

Methods
Chemicals and reagents. Unless otherwise stated, reagents, starting materials, and solvents were from Sigma-Aldrich (St. Louis, MO, USA). Reactions were carried out with magnetic stirring in round-bottomed flasks. Analytical thin layer chromatography (TLC) was performed on pre-coated glass silica gel plates 60 (F254, 0.25 mm, VWR International). Purifications were performed by flash column chromatography on silica gel (230−400 mesh, Merck Millipore). 1D and 2D NMR spectra were recorded with Bruker Avance (400 MHz) spectrometer, at room temperature. Chemical shifts are reported in δ values (ppm) relative to internal Me 4 Si, and J values are reported in hertz (Hz). The following abbreviations are used to describe peaks: s (singlet), d (doublet), dd (double doublet), t (triplet), q (quartet), and m (multiplet). HR-MS experiments were performed using an LTQ-Orbitrap-XL-ETD mass spectrometer (Thermo Scientific, Bremen, Germany), using electrospray ionization. Analytical RP-HPLC was performed on a Phenomenex Synergi Fusion RP-80A (75 mm × 4.6 mm, 4 μm), with a flow rate of 1 mL/min, using a tunable UV detector at 254 nm. Mixtures of CH 3 CN and 0.05% TFA in H2O were used as mobile phase. All compounds showed a purity ≥ 95%. The synthetic androgen R1881 was used at 1 or 10 nM. Enzalutamide (Selleckchem, Munich, Germany) was used at 10 μM.

Patch-clamp experiments.
Patch clamp experiments for compounds 4, 5 and 9 were done as described 31 .
HEK-293/TRPM8 exon 1 cells were seeded 3 or 4 days before experiment at a concentration of 4 and 2.5 million cells, respectively, onto a T225 flask. Before each experiment, cells were washed two times with D-PBS without Ca 2+ /Mg 2+ (Euroclone, Milan, Italy) and trypsinized with trypsin−EDTA (Sigma-Aldrich; diluted 1/10 www.nature.com/scientificreports/ with PBS. They were left un-stimulated or stimulated for the indicated times, in the absence or presence of the selected compounds. To avoid cell proliferation, cytosine arabinoside (Sigma-Aldrich) at 50 μM (final concentration) was included in the cell medium. Different fields were analysed by contrast-phase microscopy, as below described in this section. Migration and invasion assays were done as reported 27 . In migration assay, we used 3 × 10 4 cells in collagen pre-coated Boyden's chambers with 8 μm polycarbonate membrane (Falcon). The indicated stimuli were added to the upper and the lower chambers. Cytosine arabinoside (at 50 μM) was included in the cell medium. After 7 h, non-migrating cells from the membrane upper surface were removed using a sterile cotton swab. In invasion assay, we used 5 × 10 4 cells in Boyden's chambers with 8 μm polycarbonate membrane (Falcon) pre-coated with growth factor reduced-and phenol red-free Matrigel (Corning). The indicated stimuli were added to the upper and the lower chambers. Cytosine arabinoside (at 50 μM) was included in the cell medium. After 24 h, non-invading cells from the membrane upper surface were removed using a sterile cotton swab. In migration and invasion assays, the membranes were fixed for 20 min in 4% paraformaldehyde, stained with Hoechst (Sigma-Aldrich), removed with forceps from the companion plate and mounted. Migrating and invading cells were scored by fluorescence microscopy, as below described in this section.
3D culture in ECM. LNCaP cell spheroids were generated as reported 26 . Briefly, 2 × 10 4 LNCaP cells were mixed with 200 μL of phenol-red free growth factor-reduced Matrigel and 50 μL of organoid plating medium in each well. The mixture was pipetted in 24-well plate and allowed to solidify for 45 min at 37 °C, before the addition of 400 μL organoid plating-medium, which was made using phenol red-free DMEM/F12 medium, containing 7% CSS, penicillin (100 U/mL), streptomycin (100 U/mL), diluted GlutaMAX 100X (Gibco), 10 mM Hepes, B27 supplement (50 × stock solution; Thermofisher, Waltham, Massachusetts, USA), 1 M nicotinamide (Sigma-Aldrich), 500 mM N-acetylcysteine (Sigma-Aldrich) and 10 μM Y-27632 (Millipore, Burlington, MA, USA). After 3 days, the organoid-plating medium was replaced with a similar medium in the absence of N-acetylcysteine and Y-27632. At the 3rd day, spheroids/organoids were untreated or treated with the indicated compounds. The medium was changed every 3 days and fields were analyzed by phase-contrast microscopy.
Phase-contrast microscopy. In wound scratch assay, different fields were analysed using DMIRB inverted microscope (Leica) equipped with N-Plan 10 × objective (Leica), as reported 49 . LNCaP organoids were analysed using the same microscope equipped with C-Plan 40 × or HCX PL Fluotar 63 × objectives (Leica). Images were captured using a DFC 450C camera (Leica) and acquired using the Application Suite Software (Leica). They are representative of three different experiments. The relative organoid size was calculated using the same software and expressed as a fold increase over the basal organoid size (measured at 3th day). The wound gap was calculated using Image J Software and expressed as % of the decrease in the wound area. Migrating or invading cells from at least 30 fields/each membrane were scored, using a DMLB (Leica) fluorescent microscope, equipped with HCPL Fluotar 20 × objective.
Fluorescent Ca 2+ imaging. LNCaP cells in 6 multi-wells were made quiescent and incubated for 60 min at 37 °C with 1 µM 4-Fluo AM (Abcam, ab241082). Cells were washed twice with Hank's Balanced salt solution (HBSS) to remove any dye non-specifically associated with the cell surface. They were incubated with medium for 30 min to allow complete de-esterification of intracellular AM esters, and then left untreated or treated for 240 s with 10 nM R1881, in the absence or presence of the indicated compounds. Different fields were analyzed using DMIRB Leica (Leica) microscope equipped with C-Plan × 40 or HCX PL Fluotar × 63 objectives (Leica). IF microscopy images were generated using a DFC 450C camera (Leica).

DNA synthesis and immunofluorescence (IF).
Cells on coverslips were pulsed in vivo with BrdU (Sigma-Aldrich; at 100 μM, final concentration) and left unchallenged or challenged with 10 nM R1881 for 18 h, in the absence or presence of the indicated compounds. Cells were fixed, permeabilized, washed and the BrdU incorporation was analysed as reported 63 . Nuclei were stained with Hoechst and cells on coverslips were scored by IF microscopy, using a DMLB Leica (Leica) fluorescent microscope, equipped with HCX PL Apo 63 × oil objective. Different fields from each coverslip were analyzed and BrdU incorporation was calculated by the formula: percentage of BrdU-positive cells = (No. of BrdU-positive cells/No. of total cells) × 100. Data in the graphs derive from at least 1000 scored cells for each coverslip. Representative images were captured using DC480 camera (Leica) and acquired using Leica Suite software. They are representative of three different experiments.
Prostate specific antigen (PSA) assay. It was done in culture medium using the PSA ELISA kit www.nature.com/scientificreports/