Novel Role of Cold/Menthol-sensitive Transient Receptor Potential Melastatine Family Member 8 (TRPM8) in the Activation of Store-operated Channels in LNCaP Human Prostate Cancer Epithelial Cells*

Recent cloning of a cold/menthol-sensitive TRPM8 channel (transient receptor potential melastatine family member 8) from rodent sensory neurons has provided the molecular basis for the cold sensation. Surprisingly, the human orthologue of rodent TRPM8 also appears to be strongly expressed in the prostate and in the prostate cancer-derived epithelial cell line, LNCaP. In this study, we show that despite such expression, LNCaP cells respond to cold/menthol stimulus by membrane current (Icold/menthol) that shows inward rectification and high Ca2+ selectivity, which are dramatically different properties from “classical” TRPM8-mediated Icold/menthol. Yet, silencing of endogenous TRPM8 mRNA by either antisense or siRNA strategies suppresses both Icold/menthol and TRPM8 protein in LNCaP cells. We demonstrate that these puzzling results arise from TRPM8 localization not in the plasma, but in the endoplasmic reticulum (ER) membrane of LNCaP cells, where it supports cold/menthol/icilin-induced Ca2+ release from the ER with concomitant activation of plasma membrane (PM) store-operated channels (SOC). In contrast, GFP-tagged TRPM8 heterologously expressed in HEK-293 cells target the PM. We also demonstrate that TRPM8 expression and the magnitude of SOC current associated with it are androgen-dependent. Our results suggest that the TRPM8 may be an important new ER Ca2+ release channel, potentially involved in a number of Ca2+- and store-dependent processes in prostate cancer epithelial cells, including those that are important for prostate carcinogenesis, such as proliferation and apoptosis.

Mammalian homologues of the Drosophila transient receptor potential (TRP) 7 channel, which initially emerged as a channel specifically linked to phospholipase C-catalyzed inositol phospholipid breakdown signaling pathways, have now grown into a broad family of channelforming proteins displaying extraordinarily diverse activation mechanisms (for reviews, see Refs. [1][2][3][4][5]. At present, these channels are grouped into six subfamilies based on structural homology and have been given a standard nomenclature (5).
A number of mammalian TRPs show a unique mode of gating, in response to thermal stimuli as well as to the chemical imitators of burning and cooling sensations, capsaicin and menthol, respectively. As such, they represent a group of thermal receptors covering a wide range of physiological temperatures. Most thermal receptors belong to the vanilloid TRP subfamily (TRPV, Ref. 6) including warm-sensitive (Ͻ40°C) TRPV3 (7-9) and heat-and capsaicin-sensitive TRPV1 (Ͼ43°C) (10) and TRPV2 (Ͼ52°C) (11). In contrast, sensitivity to cooling temperatures (Ͻ22°C) and menthol is mediated by a structurally distant thermal receptor, TRPM8, belonging to the melastatine (TRPM) subfamily of TRP channels (12,13); the ankyrin transmembrane protein 1 (ANKTM1 or TRPA1) is involved in the detection of noxious cold (14).
Consistent with their role in the sensation of distinct physiological temperatures, thermal receptors are mostly expressed in subsets of sensory neurons, where they participate in the conversion of various thermal stimuli into an electrochemical form valid for further propagation to the integrative centers of the central nervous system as well as to skin cells. Interestingly, the 92% identical human orthologue of a rodent cold receptor, initially termed as trp-p8 (15), is almost exclusively expressed in the male reproductive system. In fact, testis and prostate tissues were shown to be the only normal human tissues outside the nervous system to express cold receptor TRPM8-related transcripts (15). Moreover, while remaining at moderate levels in a normal prostate, trp-p8 expression greatly increases in prostate cancer. Other non-prostatic primary human tumors of breast, colon, lung, and skin also become highly enriched in trp-p8, whereas in the corresponding normal tissues it is virtually undetectable (15). All this suggests that aside from its already established role in cold sensation, TRPM8 is likely to have other important functional roles, especially in the prostate and during carcinogenesis.
In the present study, we focused on the characterization of the endogenous TRPM8 in the model system of human LNCaP (lymph node carcinoma of the prostate, Ref. 16) androgen-dependent prostate cancer epithelial cells. We show that LNCaP cells, although expressing TRPM8 transcripts and protein, nevertheless generate a membrane current in response to cold or menthol with drastically different biophysical properties than those initially described for sensory neurons and heterologously expressed TRPM8 (e.g. Refs. 12 and 13). In contrast to "classical" TRPM8-mediated current (I cold/menthol ), which is characterized by outward rectification and poor cationic selectivity (12,13), menthol-activated current in LNCaP cells shows strong inward rectification and high calcium selectivity; which could nevertheless, still be suppressed by experimental maneuvers that decrease endogenous TRPM8 mRNA and protein expression. We demonstrate that these puzzling results are explained by the exclusive extraplasmalemmal localization of TRPM8 mostly in the endoplasmic reticulum (ER) membrane of LNCaP cells, where in response to cold or menthol, it is able to support Ca 2ϩ release. Concomitant ER Ca 2ϩ store depletion activates Ca 2ϩ entry via plasma membrane store-operated channels (SOC). This could be detected in electrophysiological experiments as I SOC/menthol with biophysical properties identical to those of the store-operated currents described in our previous work (17)(18)(19)(20)(21). In contrast, TRPM8 cloned from human prostate and heterologously expressed in HEK-293 cells shows dual localization in the ER as well as in the plasma membrane (PM), where it mediates classical I cold/menthol . We also show that TRPM8 expression in LNCaP cells and the associated I SOC/menthol magnitude are androgendependent. Our results suggest that TRPM8 may be an important ER Ca 2ϩ release channel potentially involved in a number of Ca 2ϩ -and store-dependent processes in prostate cancer epithelial cells, including those that are important for prostate carcinogenesis, such as proliferation and apoptosis.
Electrophysiology and Solutions-The whole cell patch-clamp technique was used for membrane currents recording. This technique has been described in detail elsewhere (17)(18)(19)(20)(21). The average cell capacitance of LNCaP and HEK-293 (wild-type and TRPM8-expressing) cells were 23.6 Ϯ 1.9 pF and 19.4 Ϯ 1.6 pF, respectively. The composition of the normal extracellular solutions used for electrophysiological recordings was (in mM): 140 NaCl, 5 KCl, 2 CaCl 2 , 2 MgCl 2 , 0.3 Na 2 HPO 4 , 0.4 KH 2 PO 4 , 4 NaHCO 3 , 5 glucose, 10 HEPES (pH adjusted to 7.3 with NaOH). The extracellular solution used to record Ca 2ϩ -carried I SOC/menthol contained (in mM): 150 N-methyl-D-glucamine (NMDG), 10 CaCl 2 , 10 HEPES (pH adjusted to 7.3 with HCl). Changes in external solutions were carried out using a multibarrel puffing micropipette with a common outflow, positioned in close proximity to the cell under investigation. During the experiment, the cell was continuously superfused with the solution via a puffing pipette to reduce the possibility of artifacts occuring because of the switch from a static to a moving solution and vice versa. Complete external solution exchange was achieved in Ͻ1 s. The intracellular pipette solution used to record I SOC/menthol in LNCaP cells contained (in mM): 100 NMDG, 20 NaCl, 10 HEPES, 8 EGTA, 3.1 CaCl 2 , 1 MgCl 2 (pH adjusted to 7.3 with HCl). Otherwise we used K ϩ -or Cs ϩ -based intracellular solutions (in mM): 100 KOH (CsOH), 40 KCl (CsCl), 5 HEPES, 8 EGTA, 3.1 CaCl 2 , 1 MgCl 2 (pH adjusted to 7.3 with glutamic acid). The free concentrations of divalent cations in the solu-tion containing a chelating agent were estimated using WinMaxc 1.7 software (22). All chemicals were obtained from Sigma. For all Ca 2ϩ imaging and electrophysiological experiments, the temperature of the cell bath solution was adjusted to 36°C, unless otherwise specified.
The trpm8 gene from LNCaP cells was cloned from 4 g of total mRNA by RT-PCR using SuperScript TM III Reverse Transcriptase (Invitrogen), Random Hexamer Oligonucleotides (PerkinElmer Life Sciences), forward primer: 5Ј-CCTGCTTGACAAAAACCGTC-3Ј, backward primer: 5Ј-TCTCAAGGTCTCAGCACACTA-3Ј and BD Advantage TM 2 PCR enzyme system (Clontech). PCR products of about 3.5 kbp were cloned into the pcR2.1 vector with a TOPO TA cloning kit (Invitrogen). After sequencing, the plasmid was digested by EcoRI (New England Biolabs), and the two restriction fragments were ligated into a pcDNA4 vector (Invitrogen) predigested with EcoRI. Correct trpm8 gene insertion was confirmed by restriction analysis of positive clones.
Immunohistochemistry-60% confluent LNCaP cells were fixed with 4% formaldehyde-1ϫ PBS for 15 min. After three washes, cells were either incubated in FITC-conjugated cholera toxin subunit B (CTB, C-1655, Sigma-Aldrich, dilution: 1:2000) diluted in 1ϫ PBS at room temperature for 20 min and then washed three times before permeabilization or were directly permeabilized in PBS-gelatin (phosphate-buff-ered saline, gelatin 1.2%) complemented with 0.01% Tween 20 and 100 mM glycine for 30 min at 37°C. Afterward, the cells were co-incubated with a primary rabbit polyclonal anti-TRPM8 antibody (code: ab3243, Abcam, Cambridge, UK) diluted (1:200) in PBS-gelatin for 1 h at 37°C. After thorough washes in PBS-gelatin, the slides were treated with the corresponding Alexa Fluor 546-labeled anti-rabbit IgG (A-21206, Molecular Probes, using the dilution: 1:4000) diluted in PBS-gelatin for 1 h at room temperature. After two washes in PBS-gelatin and one in PBS, the slides were mounted with Mowiol.
Fluorescence analysis was carried out using a Zeiss LSM 510 confocal microscope (488-nm excitation for FITC and 546 nm for Alexa-546) connected to a Zeiss Axiovert 200 M with a 40 ϫ 1.3 numerical aperture oil immersion objective. Both channels were excited, collected separately, and then merged to examine colocalization. The image acquisition characteristics (pinhole aperture, laser intensity, etc.) were the same throughout the experiments to ensure the comparability of the results. Using confocal microscope software (AIM 3.2, Zeiss, Le Pecq, France), the colocalization coefficients were calculated in the manner defined by Manders et al. (24).
For the immunofluorescence quantification of TRPM8 expression in LNCaP cells transfected with either control or anti-TRPM8 siRNAs, all confocal images were acquired under standard conditions (pinhole aperture, laser intensity, etc.) and analyzed using confocal microscope software (AIM 3.2, Zeiss) to determine the specific mean cell intensity for each experiment and then to calculate the average and standard deviation for four independent areas.
Ca 2ϩ Imaging-Cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] c ) was measured using ratiometric dye Fura-2 and quantified according to Grynkiewicz et al. (25). For confocal [Ca 2ϩ ] c imaging, cells were loaded with non-ratiometric Fluo-3AM dye. During measurements, the cells were bathed in the same normal extracellular solution used for electrophysiological recording. To produce Ca 2ϩ -free conditions, CaCl 2 was removed from this solution, and EGTA (0.5 mM) was added. For Ca 2ϩ imaging within the ER, LNCaP cells were grown on glass coverslips and loaded with 5 M Mag-Fluo-4AM (Molecular Probes), for 45 min at 37°C. After incubation with the dye, the plasma membrane was selectively permeabilized: cells were rinsed briefly in a high K ϩ solution of the following composition (in mM): KCl, 125; NaCl, 25; HEPES, 10; EGTA, 1; CaCl 2 , 0.5; MgCl 2 , 0.1 (free Ca 2ϩ clamped to 170 nM, pH 7.2), and exposed for 1 min to the same solution at 37°C in the presence of digitonin (0.5 mg/ml). Permeabilized cells were then continuously perfused with the high K ϩ solution supplemented with 0.2 mM Mg-ATP. Confocal imaging was performed using a Zeiss LSM 510 confocal microscope.
Data Analysis-Data were analyzed using PulseFit (HEKA Electronics) and Origin 5.0 (Microcal Software Inc., Northampton, MA). Results are expressed as means Ϯ S.E. Statistical analysis was performed using the Student's t test (p Ͻ 0.05 is considered significant) and analysis of variance tests followed by Tukey-Kramer post-tests.

RESULTS
Expression of Full-length TRPM8 in LNCaP Cells-Although previous studies have shown the expression of TRPM8 transcripts (15,26) and even protein labeling (26) in LNCaP cells, to be a necessary prerequisite for functional studies, we first sought to confirm endogenous  TRPM8 mRNA expression in these cells by RT-PCR and to compare that to other cell lines of prostatic origin. Fig. 1A shows that out of four prostatic cell lines tested: namely the normal prostate epithelial cell line PNT1A, the androgen-dependent prostate cancer epithelial cell line LNCaP and androgen-independent prostate cancer epithelial cell lines DU-145 and PC-3, only in LNCaP cells could the TRPM8 mRNA-specific amplification fragment be detected. No such fragment was evident in the most popular heterologous expression system, namely wild-type HEK-293 cells.
Given that expression of endogenous full-length TRPM8 transcript has never been demonstrated in LNCaP cells, and the degree of its homology to human TRPM8 is not known, we isolated the 3312-bp open reading frame of the full-length TRPM8 cDNAs from both LNCaP cells and human prostate (see "Experimental Procedures" for details). Sequencing of both clones showed quite perfect homology between them (99.97%) and to other published human sequences. Analysis of all TRPM8 clones available from GenBank TM revealed two sites of possible single nucleotide polymorphism (SNP) at positions: 173 (T 7 C) and 2383 (A 7 G) (see TABLE ONE). Thus, from a molecular perspective, endogenous TRPM8 in LNCaP cells is basically identical to the TRPM8 expressed in native human prostate.
Using the pEGFP-TRPM8 plasmid for HEK-293 cell transfection directs the expression of the fusion construct, in which GFP is attached to the TRPM8 C terminus. Such a construct is advantageous when studying the subcellular distribution of heterologously expressed TRPM8 based on GFP fluorescence. However, the influence of GFP on TRPM8 functional properties cannot be excluded. Therefore, for functional studies of recombinant TRPM8, we also used HEK-293 cells transiently co-transfected with separate pcDNA4 plasmid containing the TRPM8 cDNA insert and pEGFP plasmid (HEK-293/pcDNA4-TRPM8ϩpEGFP cells), as well as a HEK-293 cell line newly created in our laboratory, stably expressing TRPM8 under the tetracycline-inducible promoter (HEK-293/ TRPM8 cells). The same cells, but without tetracycline induction, were used for control purposes (HEK-293/ctrl cells), because RT-PCR detection of TRPM8 expression in such cells was negative (data not shown). Our control experiments did not reveal any notable differences in the protein targeting, menthol sensitivity, and voltage dependence of TRPM8-carried currents in HEK-293/pEGFP-TRPM8, HEK-293/pcDNA4-TRPM8ϩpEGFP, or HEK-293/TRPM8 cells.
LNCaP Cells and TRPM8-expressing HEK-293 Cells Exhibit Distinct Functional Responses to Menthol-Functional responses to temperature decrease or menthol exposure, which are the interventions known to activate endogenous and heterologously expressed TRPM8 (12,13), were investigated using a patch-clamp technique for activation of membrane currents.
Both endogenous I cold/menthol in sensory neurons and the current associated with heterologous expression of rodent TRPM8 were shown to be outwardly rectifying and mainly carried by monovalent cations (12,13). Consistent with these findings, TRPM8-expressing HEK-293 cells responded to menthol application (100 M) by generating an outwardly rectifying current (I menthol ), which was absent in the ctrl/HEK-293 cells ( Fig. 2A). In contrast, application of menthol to the whole cell patch-clamped LNCaP cells under physiological ionic conditions from both sides of the membrane (i.e. K ϩ -and Na ϩ -based intra-and extracellular solutions, respectively; see "Experimental Procedures") failed to evoke any notable outward current on top of the voltage-dependent K ϩ one inherent to LNCaP cells (Fig. 2B, n ϭ 6) described in our previous reports (27,28). It should be noted that the endogenous K ϩ current in LNCaP cells, with respect to its apparent outward rectification, is similar to TRPM8-mediated I menthol . Thus, in the event of plasmalemmal TRPM8 expression, one would expect a menthol-induced enhancement of the outward current. Instead, at higher resolutions, we were able to detect the appearance of some small inward currents in response to menthol in LNCaP cells (Fig. 2B, inset).
The lack of classical TRPM8-mediated I menthol in LNCaP cells seems to be in conflict with the robust TRPM8 mRNA expression detected by RT-PCR (see Fig. 1A). This controversy can be explained by either predominant non-plasmalemmal localization of the mature TRPM8 protein or by specific post-translational processing resulting in the loss of functional properties. To investigate subcellular TRPM8 distribution we used two strategies. First, we transfected LNCaP cells with GFPtagged recombinant TRPM8 and examined the targeting of heterologously expressed construct based on GFP fluorescence Second, we used fluorescent-labeled anti-TRPM8 antibody to visualize endogenously expressed protein. Fig. 2C compares representative confocal images of HEK-293 cells (left panel) and LNCaP cells (right panel) heterologously expressing GFP-tagged TRPM8. Inspection of both images shows substantial GFP fluorescence associated with intracellular compartments, including perinuclear membranes in both cell types. However, if in HEK-293 cells clear clusters of GFP fluorescence are also evident on the cellular perimeter, then in LNCaP cells such peripheral fluorescence was completely lacking. Such patterns of GFP fluorescence are consistent with combined intracellular and plasmalemmal localization of chimera TRPM8-GFP protein in HEK-293 cells, but only intracellular localization in LNCaP cells. According to the differences in TRPM8/GFP localization, we were unable to detect any classical I menthol either in the regular LNCaP cells (i.e. Fig. 2B) or following heterologous expression of TRPM8-GFP construct (data not shown). However, in HEK-293/ pEGFP-TRPM8 cells classical I menthol could be readily activated (i.e. Fig. 2A).
To verify the subcellular localization of endogenous TRPM8 protein in LNCaP cells, we utilized double staining with specific anti-TRPM8 antibody and FITC-conjugated CTB. The latter specifically interacts with GM1 lipids and therefore serves as a plasma membrane marker (29, membrane (mean Ϯ S.E.) before (ctrl, circles, n ϭ 6) and after exposure to menthol (100 M, triangles, n ϭ 6).   Fig. 2D, green fluorescence of CTB is strictly limited to the PM (middle panel), whereas red detection of TRPM8 appears to be mostly dotted in the cell body (left panel). Overlay of the two images (right panel) did not reveal any yellow color, which would be indicative of TRPM8 and CTB colocalization, and thus provide evidence of TRPM8 location in the plasma membrane. This observation was quantitatively confirmed by the calculation of the colocalization coefficients (24) in 10 different PM-including regions of interest (ROI) belonging to 8 different cells, as exemplified in the right panel of Fig. 2D. (ROI is encompassed by the thin light blue line.) The CTP/TRPM8 colocalization coefficient averaged only 3 Ϯ 2%, indicating that only an insignificant proportion of TRPM8 proteins (i.e. 3 Ϯ 2%) is located within the PM. It is important to note that because of the optical limitations of a confocal microscope, a protein should be considered to have essential plasma membrane localization when the colocalization coefficient with PM marker exceeds 10% and is at least significantly different from 0.

30). As shown in
Interestingly, in contrast to the observations with heterologously overexpressed TRPM8/GFP, the immunodetection of endogenous TRPM8 did not reveal notable perinuclear localization. Therefore, it is quite likely that overexpression of TRPM8-GFP either produced an artifact of perinuclear localization or drastically potentiated a normally weak translocation to this membrane. Although the possibility of some presentation of TRPM8 protein in the perinuclear membrane cannot be excluded, its functional significance for Ca 2ϩ signaling and/or generation of membrane current in LNCaP cells is unlikely. Thus when taken together, our data suggest that neither endogenous nor heterologously expressed TRPM8 channels are targeted at the plasma membrane of LNCaP cells, which we believe is the primary reason for the lack of classical I menthol in this particular type of prostate cancer epithelial cells.
TRPM8 Mediates Qualitatively Different Cold/Menthol Responses in LNCaP Cells-Recently, it has been suggested that TRPM8 is present in the ER membrane of LNCaP cells, where it may function as a cold/ menthol-sensitive Ca 2ϩ release channel (26). Our data showing exclusive TRPM8 targeting of intracellular compartment(s) of LNCaP cells support the idea of such a functional role. If this is in fact the case, then TRPM8-mediated Ca 2ϩ release from the ER in response to cold/menthol should activate SOC plasma membrane channels, which can be detected as store-operated Ca 2ϩ current (I SOC ) in patch-clamp experiments. Our observation that exposure of LNCaP cells to menthol elicits not classical outwardly rectifying I menthol , but rather a small inward current (see Fig. 2B, inset) is consistent with such expectations. Moreover, the fact that this current was sensitive to the extracellular Ca 2ϩ removal (data not shown), suggested that Ca 2ϩ is a major charge carrier, which is typical of highly Ca 2ϩ -selective SOCs in LNCaP cells (17)(18)(19)(20)(21).
To better resolve menthol-activated current in LNCaP cells, we formulated intra-and extracellular solutions, which allowed the elimination of voltage-dependent K ϩ and the minimization of other possible currents (because of K ϩ , Na ϩ replacement with impermeable NMDG; see "Experimental Procedures") and raised the extracellular Ca 2ϩ concentration ([Ca 2ϩ ] out ), as a potential charge carrier, to 10 mM. Under such conditions, both cold stimulus (a temperature drop from 36°C to 22°C, Fig. 3A) and menthol (100 M, Fig. 3B) activated membrane currents with prominent inward rectification. These currents were indeed quite small at Ϫ100 mV averaging at only 1.13 Ϯ 0.36 pA/pF (n ϭ 6) and 1.24 Ϯ 0.37 pA/pF (n ϭ 5) for cold and menthol, respectively (Fig. 3, A and B, compare with ϳ60 pA/pF for voltage-dependent K ϩ current, Fig.  2B), had very positive reversal potential (around ϩ50 mV, inset of Fig. 3,  A and B), and disappeared upon extracellular Ca 2ϩ removal (data not shown). This is consistent with the activation of plasma membrane Ca 2ϩ -permeable cationic channels in LNCaP cells.
Cationic selectivity of menthol-activated channels in LNCaP cells was characterized by measuring current amplitudes (at Ϫ100 mV) following equimolar substitutions of Ba 2ϩ , Sr 2ϩ , Mn 2ϩ , Na ϩ , or K ϩ for 10 mM Ca 2ϩ in extracellular saline and relating them to the amplitude of the Ca 2ϩ current. A summary of the selectivity measurements for 4 -6 of each ion substitution is presented in Fig. 3C. Maximal currents carried by tested ions can be placed in the following order: Ca 2ϩ (1) Ͼ Ba 2ϩ (0.40 Ϯ 0.10) ϳ Sr 2ϩ (0.39 Ϯ 0.03) Ͼ Mn 2ϩ (0.16 Ϯ 0.14) Ͼ Na ϩ (0.12 Ϯ 0.04) Ͼ K ϩ (0.06 Ϯ 0.07). This order appears to be very similar to the one found for endogenous SOCs in LNCaP cells (19,20).
In addition, menthol-activated Ca 2ϩ current in LNCaP cells was inhibited by such common inhibitors of native cationic channels and heterologously expressed TRP members such as ruthenium red (10 M, n ϭ 4), SK&F 96365 (50 M, n ϭ 6), and flufenamate (10 M, n ϭ 5). The results of the respective experiments are summarized in Fig. 3D. All agents produced fast and reversible current inhibition, which was comparable in size to their effects on I SOC , as known from our previous reports (19,20).
Altogether, our patch-clamp experiments unequivocally demonstrate remarkable similarity of principal biophysical properties of menthol-activated and store-operated currents in LNCaP cells. Such a similarity strongly suggests that both currents are likely to be carried by the same system of plasma membrane store-dependent cationic channels. To highlight two distinguishing features: menthol as an initial activating stimulus and SOCs as the transmembrane pathway involved, we propose to call the menthol-activated current in LNCaP cells I SOC/menthol .
TRPM8 Acts as an ER Ca 2ϩ Release Channel in the I SOC/menthol Activation Pathway-Our working hypothesis is that application of menthol activates TRPM8, localized in the ER membrane of LNCaP cells, which produces ER Ca 2ϩ store depletion, followed by the activation of plasma membrane SOCs detected in the form of I SOC/menthol in electrophysiological experiments. To specifically prove the role of TRPM8 in this pathway, we utilized the strategy of knocking down TRPM8 mRNA either by means of antisense oligodeoxynucleotides (ODNs) or by siRNA, with a subsequent evaluation of what impact this may have on functional responses.
In the series of TRPM8 hybrid depletion experiments, we used LNCaP cells treated for 60 h with TRPM8-specific antisense ODNs (see "Experimental Procedures"). The cells treated for the same period of time with respective sense ODNs, which are not supposed to alter endogenous TRPM8 expression, served as a control. Western blot analysis with TRPM8-specific antibody (see "Experimental Procedures") confirmed a significant reduction in TRPM8 protein in antisenseversus sense-treated cells (Fig. 4A). Surprisingly, the Western blot of LNCaP and HEK-293/TRPM8 cells produced a number of specific bands. We assume that, with the exception of the 125-kDa band, which corresponds to the normal TRPM8, the highest band observed in both cell types may represent the TRPM8 dimer (about 250 kDa), whereas smaller bands of about 50 and 75 kDa might represent cleavage products of the normal TRPM8. Furthermore, LNCaP cells exhibited an additional intermediate triplet of about 95 kDa, which was not observed in HEK-293/TRPM8, but which could be specifically silenced by antisense treatment. It may well be that this triplet represents a new TRPM8 isoform.
As documented in Fig. 4B, I SOC/menthol density (measured at Ϫ100 mV) following antisense treatment was about 72% lower than the control (i.e. from Ϫ0.81 Ϯ 0.18 pA/pF, n ϭ 6 to Ϫ0.14 Ϯ 0.06 pA/pF, n ϭ 8), suggesting a direct correlation between endogenous TRPM8 expression and I SOC/menthol magnitude. Qualitatively and quantitatively similar results were obtained using siRNA technology to down-regulate endogenous TRPM8 expression in LNCaP cells. Fig. 4D (upper panel) presents the dynamics of TRPM8 mRNA changes assayed by semi-quantitative RT-PCR on the 2nd, 4th, and 7th day after transfection with 100 nM siRNA TRPM8 . This enabled an estimation of TRPM8 mRNA reduction at these periods of 99 Ϯ 2, 65 Ϯ 10, and 10 Ϯ 5%, respectively. Immunostaining with anti-TRPM8 anti-body on the 2nd day after transfection followed by semi-quantitative analysis of confocal images (Fig. 4D, lower panel) demonstrated that the level of TRPM8 protein lagged somewhat behind, decreasing by only 83 Ϯ 3% (n ϭ 4) compared with control, yet still indicating the effectiveness of the adopted strategy in down-regulating endogenous TRPM8 expression. Substantial down-regulation of TRPM8 protein in response to siRNA TRPM8 treatment was also confirmed by Western blotting (Fig. 4A). As documented in Fig. 4E, such a down-regulation was paralleled by about a 42% decrease in maximal I SOC/menthol density at Ϫ100 mV (from control Ϫ1.24 Ϯ 0.43 pA/pF, n ϭ 5 to 0.72 Ϯ 0.27 pA/pF, n ϭ 8), further supporting the notion of the involvement of TRPM8 in the pathway leading to I SOC/menthol activation.
To directly demonstrate that the role of TRPM8 in this pathway specifically related to the mediation of Ca 2ϩ release from the ER in response to menthol, we performed a series of experiments with fluorometric [Ca 2ϩ ] c measurements in LNCaP cells subjected to TRPM8 knockdown by either antisense ODNs or siRNA technology. In this series of experiments, LNCaP cells treated with sense ODNs or a vehicle, served as respective controls. Fig. 4 (C and F) shows that when control LNCaP cells were exposed to menthol (100 M) in the absence of extracellular Ca 2ϩ , this caused a transient [Ca 2ϩ ] c elevation, obviously because of Ca 2ϩ release from the ER, whereas reintroduction of Ca 2ϩ in the continuing presence of menthol produced a rapid [Ca 2ϩ ] c elevation because of the initiation of store-operated Ca 2ϩ entry (open symbols of Fig. 4, C and F, n ϭ 72 and  69, respectively). Moreover, as one would expect in the framework of our working hypothesis, TRPM8 knockdown, independently of the strategy employed, abolished both menthol-induced responses, that associated with Ca 2ϩ release and that Ca 2ϩ entry (filled symbols of Fig. 4,  C and F, n ϭ 64 and 83, respectively). These results can only be explained if TRPM8 operates as a cold/menthol-sensitive ER Ca 2ϩ release channel in LNCaP cells.
The ability of menthol to produce TRPM8-mediated ER store depletion was further validated by directly monitoring the Ca 2ϩ content of the ER lumen ([Ca 2ϩ ] ER ) using compartmentalized fluorescent Ca 2ϩ dye Mag-Fluo 4 AM on digitonin-permeabilized LNCaP cells. This technology has been successfully used in our previous studies (18). Fig.  5A documents that application of menthol (100 M) does indeed produce a 23 Ϯ 3% (n ϭ 23) decrease in [Ca 2ϩ ] ER , whereas subsequently applying a saturated concentration of ionomycin (IM, 1 M, to check for the functionality of stores), depleted the ER store by another 40% (to total 65%). The ability of menthol to lower [Ca 2ϩ ] ER was not compromised by the combined action of the IP 3 receptor (IP 3 -R) blocker heparin (500 g/ml) and the ryanodine receptor (RyR) blocker ryanodine (20 M) (Fig. 5B), thereby suggesting the non-involvement of these Ca 2ϩ release channels in menthol effects. On the contrary, LNCaP cell treatment with TRPM8 antisense ODNs reduced the ER-depleting action of menthol by almost 2-fold (from 20 Ϯ 2%, n ϭ 12 in control to 8 Ϯ 2%, n ϭ 11, Fig. 5C), highlighting the significant role of TRPM8.
The specificity of the ER depletion to the ER-localized TRPM8 received additional strong support from the experiments on confocal imaging of cytosolic Ca 2ϩ in response to another TRPM8 agonist, the supercooling agent icilin (12). Fig. 6 shows that exposure to icilin (20 M) caused a strong elevation of [Ca 2ϩ ] c in LNCaP cells. The fact that reversible SERCA pump inhibitor cyclopiazonic acid (CPA, 20 M) did not produce much of an additional [Ca 2ϩ ] c rise on top of the icilininduced one (Fig. 6A), together with the preservation of icilin effects in the absence of extracellular Ca 2ϩ (Fig. 6B), indicated icilin action by releasing Ca 2ϩ from the ER. Icilin-induced Ca 2ϩ release was not impaired by the IP 3 -R and RyR blockers xestospongin C (10 M) and Insets show representative images taken at points indicated by arrows. B, same as in A, but in response to two successive menthol (100 M) applications, the second of which was made in the presence of IP 3 -R and RyR blockers heparin (500 g/ml) and ryanodine (20 M), respectively. C, same as in A, but acquired in LNCaP cells treated for 72 h with TRPM8-specific sense (filled circles, n ϭ 12) and antisense (open circles, n ϭ 11) ODNs. ryanodine (50 M), respectively; thus confirming its mediation by the ER-localized TRPM8, without any IP 3 -R or RyR involvement. Therefore, taken collectively, the results on cytosolic and ER luminal Ca 2ϩ measurements unequivocally demonstrated that the TRPM8 agonists, menthol and icilin, indeed act as an ER Ca 2ϩ store-depleting agent through ER-localized TRPM8, and that this depletion is sufficient for SOCE activation.
The Mode of Store-to-SOC Coupling Involved in Menthol Effects in LNCaP Cells-In our previous study, we have shown the existence of two types of SOCs in LNCaP cells with different coupling mechanisms to the ER Ca 2ϩ stores. The first type, termed SOC CC (for conformational coupling), relies on direct conformational coupling with an IP 3 -R for activation, and the second type, termed SOC CIF , requires the release of a diffusible calcium influx factor (CIF) (21). SOC CC , which carries I SOC/CC , is preferentially activated during IP 3 -induced store depletion, whereas activation of SOC CIF that carries I SOC/SIF occurs in response to thapsigargin (TG)-mediated store depletion. We then asked with what type of SOCs and store-to-SOC coupling mechanism menthol/ TRPM8-mediated store depletion interferes most. To answer this question, we utilized combined IP 3 /menthol and TG/menthol interventions to check for the additivity of the effects. Fig. 7A shows that activation of I SOC/CC by intracellular infusion of IP 3 (100 M) plus BAPTA (10 mM) via patch pipette, which recruits IP 3 R/SOC conformational coupling mechanism, did not prevent the activation of the full-size I SOC/menthol in response to menthol application. In contrast, the presence of TG (0.1 M) in the bath solution, which resulted in permanent store depletion, totally abolished the ability of menthol to activate a membrane current (Fig. 7B, n ϭ 5). Thus, these results suggest on the one hand that menthol and TRPM8 do not interfere with direct IP 3 /SOC conformational coupling and, on the other hand, that menthol and TG use common mechanisms, which are presumably CIF-mediated and probably also use a common plasma membrane SOCs for current activation.
TRPM8 Expression and Function Is Androgen-dependent-Recently it has been shown by several groups including ours (26,31), that TRPM8 expression is androgen-dependent. Thus, in view of our present findings on TRPM8 involvement in store-operated Ca 2ϩ influx, we wondered whether or not the androgen status of prostatic cells has any impact on menthol-activated current. Fig. 8A shows that steroid deprivation using charcoal-stripped culture medium, caused pronounced down-regulation of TRPM8 mRNA, as assessed by semi-quantitative RT-PCR, whereas a subsequent addition 5␣-dihydrotestosterone (DHT, 10 Ϫ9 M) caused rapid restoration of TRPM8 mRNA and even its further increase above the control levels. The deviations in TRPM8 mRNA expression were paralleled by the changes in I SOC/menthol . As Fig. 8B illustrates, exposure to menthol activated about 4-fold smaller I SOC/menthol in LNCaP cells (cultured for 96 h in steroid-deprived medium) compared with control cells. Subsequent addition of DHT (10 Ϫ9 M) completely restored I SOC/menthol . These variations of I SOC/menthol in response to both steroid deprivation and DHT supplementation basically coincided with the degrees of reduction and restoration of TRPM8 mRNA expression during suggesting nearly complete store depletion by icilin, whereas IP 3 and ryanodine receptor blockers, xestospongin C and ryanodine, respectively, did not prevent icilin-induced [Ca 2ϩ ] in increase, suggesting TRPM8 but not IP 3 -R or RyR involvement. respective periods (see Fig. 8A). Besides, as demonstrated in Fig. 8C, steroid deprivation prevents both menthol-induced Ca 2ϩ release and subsequent store-operated Ca 2ϩ entry, again confirming our conclusion that I SOC/menthol develops because of the ER store depletion via TRPM8.

DISCUSSION
In the present study we report on four major findings: 1) LNCaP prostate cancer epithelial cells express cold/menthol-sensitive TRPM8, 2) the functional response to cold/menthol involves TRPM8, yet properties of the membrane current are dramatically different from the classical TRPM8-mediated current, 3) TRPM8 is preferentially expressed in the ER membrane of LNCaP cells, where it plays the role of a cold/ menthol-sensitive ER Ca 2ϩ release channel, 4) cold/menthol-activated, TRPM8-mediated ER Ca 2ϩ store depletion is capable of activating store-operated Ca 2ϩ current in LNCaP cells, and 5) TRPM8-mediated ER Ca 2ϩ store depletion is tightly regulated by androgens.
Molecular Properties of Human TRPM8-The analysis of all published TRPM8 sequences revealed two possible single nucleotide polymorphisms at positions 173 (T 7 C) and 2383 (A 7 G) (see TABLE   ONE). We believe, however, that these two mutations are unlikely to account for the lack of TRPM8 plasma membrane expression and/or notable change in its functional properties. Indeed, both the rat and mice TRPM8 sequences (173C, 2383A) were shown to have clear plasma membrane functions (12,13). Furthermore, we detected plasmalemmal localization and expected classical TRPM8-mediated cur-  rent in HEK-293 cells heterologously expressing human TRPM8 clone with 173C, 2383G polymorphisms. Therefore, the lack of plasma membrane targeting of endogenous TRPM8 in LNCaP cells, which in positions 173 and 2383 positions coincides with the rodent isoform (i.e. 173C, 2383A), is most likely to be related to the specifics of the posttranslational processing of this protein and/or its interactions with some other factor(s), which take place in this type of prostate cancer epithelial cells. Such a conclusion is supported by the fact that heterologously expressed human TRPM8 does not target the PM of LNCaP cells either, although it shows clear PM targeting and function in HEK-293 cells.
It should be noted that in the recent study by Zhang and Barritt (26), TRPM8 was detected in both the ER membrane and plasma membrane of LNCaP cells. The reason for such a discrepancy with our results is not clear, but may be related to the clonal variations of cell line.
Molecular Basis of Cold/Menthol Response in LNCaP Cells-The apparent similarity of fundamental biophysical properties of I menthol and I SOC (inward rectification and divalent cation selectivity) prompted us to conclude that I menthol is carried through the system of store-operated channels in LNCaP cells and, therefore, could be termed I SOC/menthol . This can only be the case if TRPM8 in LNCaP cells acts as a cold/mentholsensitive ER Ca 2ϩ release channel, and activation of I SOC/menthol through the system of plasma membrane SOCs is secondary to menthol-mediated ER store depletion. The bulk of our data, which prove: 1) exclusive TRPM8 localization in the ER membrane of LNCaP cells, 2) the existence of menthol-evoked ER store depletion, 2) concomitant activation of SOCE, and 3) the inability of menthol to produce its effects following store depletion by TG, fully agrees with such a possibility. I SOC/menthol activation secondary to menthol-mediated store depletion is also consistent with the slower kinetics of current activation in LNCaP cells compared with neuronal PM-localized CMR/TRPM8, which are directly gated by menthol.
Localization of TRPM8 on presynaptic Ca 2ϩ stores was also suggested for DRG neurons (32). It was concluded that menthol-induced Ca 2ϩ release might underlie synaptic transmission facilitation at sensory synapses by menthol. Thus, even in single cell types, such as DRG neuron, TRPM8 may play a dual role, both in cold stimulus perceptions and in the facilitation of synaptic transmission, and such a role is determined by protein localization. The possibility of differential TRPM8 localization in both plasma membrane and ER membrane depending on cell type may be a distinctive feature of this TRP member. One cannot exclude that the lack of plasmalemmal TRPM8 protein in LNCaP cells is explained by the dominant expression of its putative ϳ95-kDa isoform rather than the expected 125-kDa isoform (see Fig. 4A), which may differ in preferred targeting.
Possible Physiological Roles of TRPM8 in the Prostate-The role of TRPM8 cold receptor in prostate epithelial cells is not yet clear, but as it is because of their physiology, it is most likely to be related to such processes as proliferation (33,34), apoptosis (17,18,35), differentiation (35,36), or secretion. The extent of TRPM8 involvement in each of these processes may be determined by preferential subcellular localization of the protein.
Because the prostate is not subjected to any essential temperature variations, the existence of an alternative to cold endogenous agonist(s) involved in TRPM8 activation is quite likely. Indeed, other temperaturesensitive members of the TRP family have previously been shown to be potentially activated by such stimuli as anandamide and PIP 2 for TRPV1 (37), or anandamide, arachidonic acid, and phorbol esters for TRPV4 (38,39). Moreover, the possibility of functional control of TRPM8 by PIP 2 has recently been demonstrated (40). Because an application of exogenous PIP 2 was able to activate TRPM8 as well as to prevent a rundown of the menthol-activated current, it was suggested that the PIP 2 level may be an important regulator of cold transduction in vivo. However, because of the lack of evidence proving the involvement of temperature in the prostate function, further investigations are still needed to establish alternative TRPM8 activation pathway(s) in the prostate.
Our results show that TRPM8 represents an important route for intracellular Ca 2ϩ store depletion. Store depletion has previously been shown to be critical in promoting growth arrest and apoptosis of prostate cancer epithelial cells (17,18,33,41). Reduced basal filling of intracellular Ca 2ϩ stores is also the hallmark of androgen-independent, apoptosis-resistant cell phenotypes characteristic of advanced prostate cancer (18,35,41). Thus, as a molecular entity capable of influencing ER Ca 2ϩ store filling and promoting Ca 2ϩ influx, TRPM8 may be considered as an important determinant in controlling the apoptotic status of prostate cancer epithelial cells and as such, as a potential target for therapeutic interventions. In fact, the involvement of TRPM8 in LNCaP cells apoptosis has recently been suggested (26). However, these authors suggested that TRPM8 acts as a Ca 2ϩ permeable channel in both ER and PM membrane, and that menthol-induced cell death is mediated at least in part by the sustained increase in [Ca 2ϩ ] c . Considering our previous data on the prevalent role of ER store depletion over Ca 2ϩ entry in promoting androgen-dependent LNCaP cell death (17), we believe that the primary reason for menthol-induced apoptosis might be store depletion via ER-localized TRPM8. More interestingly, these authors demonstrated that the siRNA targeted against TRPM8 (down-regulating TRPM8 protein expression by ϳ50%) drastically decreasing the percentage (by about 40%) of viable LNCaP cells. These results suggest that the endogenous ER TRPM8 activity promotes cell survival by regulating the ER Ca 2ϩ store content and/or I SOC activity. This hypothesis is heightened by the fact that: (i) the trpm8 expression and the ER TRPM8 activity are both tightly regulated by androgens, and (ii) the trpm8 gene is overexpressed in androgen-dependent prostate cancer characterized by an increased cell viability (26,31). Given the importance of androgens in the regulation of the proliferative and apoptotic activity of prostate cancer cells, it is imperative to establish how these processes involve TRPM8.