TRPM8 Activation via 3-Iodothyronamine Blunts VEGF-Induced Transactivation of TRPV1 in Human Uveal Melanoma Cells

In human uveal melanoma (UM), tumor enlargement is associated with increases in aqueous humor vascular endothelial growth factor-A (VEGF-A) content that induce neovascularization. 3-Iodothyronamine (3-T1AM), an endogenous thyroid hormone metabolite, activates TRP melastatin 8 (TRPM8), which blunts TRP vanilloid 1 (TRPV1) activation by capsaicin (CAP) in human corneal, conjunctival epithelial cells, and stromal cells. We compare here the effects of TRPM8 activation on VEGF-induced transactivation of TRPV1 in an UM cell line (92.1) with those in normal primary porcine melanocytes (PM) since TRPM8 is upregulated in melanoma. Fluorescence Ca2+-imaging and planar patch-clamping characterized functional channel activities. CAP (20 μM) induced Ca2+ transients and increased whole-cell currents in both the UM cell line and PM whereas TRPM8 agonists, 100 μM menthol and 20 μM icilin, blunted such responses in the UM cells. VEGF (10 ng/ml) elicited Ca2+ transients and augmented whole-cell currents, which were blocked by capsazepine (CPZ; 20 μM) but not by a highly selective TRPM8 blocker, AMTB (20 μM). The VEGF-induced current increases were not augmented by CAP. Both 3-T1AM (1 μM) and menthol (100 μM) increased the whole-cell currents, whereas 20 μM AMTB blocked them. 3-T1AM exposure suppressed both VEGF-induced Ca2+ transients and increases in underlying whole-cell currents. Taken together, functional TRPM8 upregulation in UM 92.1 cells suggests that TRPM8 is a potential drug target for suppressing VEGF induced increases in neovascularization and UM tumor growth since TRPM8 activation blocked VEGF transactivation of TRPV1.


INTRODUCTION
Among all cancers of the eye, uveal melanoma (UM) is the most frequent form in adults. Notably, UM is mostly found in the choroid (65% of all cases) and in ciliary body (15%), but it rarely occurs in the retina (1.4%; Singh et al., 2011). About 50% of the patients with primary UM will finally develop distant metastases predominantly in the liver (90%) (Spagnolo et al., 2012). To date, the etiology of UM is not fully understood and neither metastatic properties nor patient survival has significantly improved over the last decades (Tran et al., 2013). Accordingly, there is a pressing need for developing alternative approaches to treat this disease especially since there are no FDA approved drugs available for suppressing metastatic melanoma.
A preclinical approach targeting angiogenesis in combination with irradiation has been reported using bevacizumab, a monoclonal antibody binding and inhibiting vascular endothelial growth factor (VEGF; Sudaka et al., 2013). Nevertheless, the advantage of this combination therapy is unclear because this VEGF trap did not have a dramatic impact on any of the functional activities in UM cell lines (Logan et al., 2013). As a matter of fact, such treatment is reported to even promote expansion of melanoma cells in vitro (Dithmer et al., 2017). Furthermore, neoadjuvant intravitreous injection of this VEGF trap failed to shrink large size melanoma and is even counter indicated in these cases because it may instead even promote melanoma growth (Francis et al., 2017).
We show here that crosstalk between members of this receptor triad affects Ca 2+ signaling responses induced by VEGFR transactivation of TRPV1 in UM 92.1 melanoma cells. Therefore, selective targeting of TRPM8 control of TRPV1 responsiveness to transactivation by VEGF may ultimately provide an alternative approach to reduce tumor growth in a clinical setting.

Materials
BCTC, AMTB, and fura-2AM were purchased from TOCRIS Bioscience (Bristol, United Kingdom). CPZ and icilin were procured from Cayman Chemical Company (Ann Arbor, Michigan, U.S.A.). Medium and supplements for cell culture were ordered from Life Technologies Invitrogen (Karlsruhe, Germany) or Biochrom AG (Berlin, Germany). Melanocyte Growth Medium M2 was obtained from Promocell (Heidelberg, Germany). Dispase II was ordered from Boehringer (Ingelheim, Germany) and accutase was provided by PAA Laboratories (Pasching, Austria). Unless otherwise stated, all other reagents were procured from Sigma (Deisenhofen, Germany).

Melanocyte Primary Cell Cultivation
PM were isolated from porcine eyes provided by a slaughterhouse. The preparation and primary cell cultivation were performed as described (Valtink and Engelmann, 2007). In brief, eyeballs were cut into two pieces. The choroid with the connected retinal pigment epithelium (RPE) layer was separated from the sclera and incubated in collagenase IV for several hours at 37 • C in order to release RPE cells from melanocytes. After a second treatment with dispase II, the choroids were put into a shaking device in order to better isolate the cells from the tissue. Finally, the cell suspension was passed through a cell strainer. After centrifugation, cells were washed in RPMI medium and seeded in tissue culture flasks. After 24 h, the medium was changed and cells were cultivated under the same conditions as those described for the UM 92.1 cells (Mergler et al., 2014). To avoid contamination with RPE cells or fibroblasts, the culture medium was supplemented with geneticin for about 5-7 days prior to subcultivation. Melanocyte cell cultivation was limited to no longer than 2 weeks to avoid cell dedifferentiation.

Intracellular Calcium Fluorescence Imaging
Semi confluent cells (≈80%) were loaded with fura-2/AM (2 µM) at 37 • C. After about 40 min, the cells were washed with a Ringerlike (control) solution containing (mM): 150 NaCl, 6 CsCl, 1 MgCl 2 , 10 glucose, 10 HEPES, and 1.5 CaCl 2 at pH 7.4 and 317 mOsM (Mergler et al., 2014). KCl was replaced with CsCl to characterize TRP channel activity (Voets et al., 2004). Following dye loading, the cells were exposed to this solution on the stage of an inverted microscope (Olympus BW50WI, Olympus Europa Holding GmbH, Hamburg, Germany), connected with a digital imaging system (TILL Photonics, Munich, Germany). Fura-2/AM fluorescence was consecutively excited at 340 and 380 nm for different times (Grynkiewicz et al., 1985). The 510 nm emission ratio (f 340nm /f 380nm ) is an index of relative intracellular Ca 2+ ([Ca 2+ ] i ) levels (Grynkiewicz et al., 1985). The 340 and 380 nm response signals were continuously detectable and did not distort the ratio. The changes in ratios were overall small because of the presetting of the single fluorescence signals at 340 and 380 nm, respectively. A control where TRPM8 was heterologously expressed and activated by their agonists is provided (Lucius et al., 2016). Before starting a measuring session, baseline stability was established for 8-20 min. All experiments were performed at a constant room temperature (≈23 • C). In addition, the fura-2-induced fluorescence signals were alternatively evaluated in a bath chamber using a Life Science fluorescence cell imaging software in conjunction with a high-resolution digital camera (Olympus XM-10) (Figures 9-11). Cutoff filters isolated alternative fluorescence excitation at 340 and 380 nm every 5 s wavelengths provided by a LED light source (LED-Hub by Omikron, Rodgau-Dudenhoven, Germany). Fura-2 fluorescence was alternately excited at 340 and 380 nm and emission was detected at 510 nm (250 ms−3.8 s exposure time). For image acquisition and data evaluation, the Life Science imaging software cellSens was used (Olympus, Hamburg, Germany). Results are shown as mean traces of the f 340nm /f 380nm ratio ± SEM (error bars in both directions) with n-values indicating the number of experiments per data point. The Ca 2+ data presented from many cells in several experiments were normalized (control set to 1.2 and 0.2, respectively) and averaged (with error bars). The time delay of 1-2 min in inducing a Ca 2+ transient stems from exposing the cells to a stationary bath rather than a flow through superfusion. When drugs were solubilized in dimethyl sulfoxide (DMSO) solution, their working concentration did not exceed 0.1%, which did not alter the Ca 2+ base line.

Planar Patch-Clamp Recordings
Whole-cell currents were measured with a planar patch-clamp setup (Port-a-Patch © ; Nanion, Munich, Germany) in connection with an EPC 10 patch-clamp amplifier (HEKA, Lamprecht, Germany) and the PatchMaster software (Version 2.6; HEKA, Lamprecht, Germany) (Mergler et al., 2012a(Mergler et al., , 2014Garreis et al., 2016). A standard intracellular solution containing (mM): 50 CsCl, 10 NaCl, 60 CsF, 20 EGTA, and 10 HEPES-acid at pH ≈ 7.2 and ≈ 288 mOsM was applied to the microchip (both provided by Port-a-Patch © , Nanion, Munich, Germany). The external solution contained (mM): 140 NaCl, 4 KCl, 1 MgCl 2 , 2 CaCl 2 , 5 D-glucose monohydrate and 10 HEPES, pH ≈ 7.4, and osmolarity ≈ 298 mOsM. At first, 5-10 µl of a single cell suspension were placed onto a microchip containing the aforementioned external solution. A software-controlled pump (Nanion) provided a connection between a single cell and the electrical system (sealing). The mean membrane capacitance was 10 pF ± 1 pF (n = 88) and mean access resistance was 25 ± 3 M (n = 88). Series resistances as well as fast and slow capacitative transients were compensated by the PatchMaster software. The liquid junction potential was calculated (≈3.8 mV; Barry, 1994) and offset by the software. Notably, current recordings were all leak-subtracted and cells with leak currents above 100 pA were excluded from analysis. All experiments were performed at 21-23 • C room temperature. The holding potential (HP) was set to 0 mV in order to eliminate any possible contribution of voltage-dependent Ca 2+ channel activity. Cells were kept in the whole-cell configuration for ∼10 min for control recordings and the compensation proceedings before starting the experiments (Pusch and Neher, 1988). Whole-cell currents were recorded over a voltage range of −60 to +130 mV for 500 ms each and measured every 5 s. The current densities (pA/pF) were The same effect could be observed in normal porcine melanocytes but with a time lag (n = 7; controls n = 11). (C) Summary of the experiments with CAP (n = 4-7). The asterisks (*) show significant differences between control and CAP (n = 4; 350s; *p < 0.05; n = 7; 590 s; **p < 0.01; paired tested). The hashtag shows a significant difference of the CAP effect at 350 s between UM 92.1 cells and melanocytes (n = 4-7; 590 s; # p < 0.05; unpaired tested). (D-F) Same experiments as shown in (A-C), but with icilin (20 µM) (n = 4-8). The icilin effect was markedly reduced in porcine melanocytes. The asterisks (*) show significant differences between control and icilin (n = 4; 350 s; *p < 0.05; paired tested). The hashtags (#) shows a significant difference of the icilin effect at 350 s and 590 s between UM 92.1 cells and melanocytes (n = 4-8; 350 s; ### p < 0.005; 590 s; ## p < 0.01; unpaired tested).
calculated by dividing the current (pA) by the cell membrane capacitance (pF). For purposes of comparison, the currents were normalized to control currents (set to 100%).

Statistical Analysis
The paired two-tailed Student's t-test was applied in conjunction with several normality tests (KS normality test, D'Agostino & Pearson omnibus normality test, and Shapiro-Wilk normality test). If these tests failed, non-parametric Wilcoxon matched pairs were used. The Student's t-test was also used for unpaired data, if the data also passed the aforementioned normality tests. If these tests failed, the non-parametric Mann-Whitney-U test was performed. Welch's correction was applied if data variance of the two groups were too different. Probabilities of p < 0.05 [indicated by asterisks ( * ) and hash tags (#)] were considered to be significant. The number of repeats is shown in each case in brackets, near the traces or bars. All values are means ± SEM (error bars in both directions). All plots were generated with SigmaPlot software version 12.5 for Windows (Systat Software, San Jose, California, United States) and with GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego California USA), respectively.
intracellular Ca 2+ increase in PM, which was absent in UM 92.1 tumorous cells. In UM 92.1 cells, CAP instead increased the f 340nm /f 380nm ratio more promptly but to the same level; namely, from 1.199 ± 0.001 (80 s) to 1.215 ± 0.002 after 590 s (n = 4, p < 0.01; Figures 1A,C). Overall, there is a difference in the Ca 2+ response pattern. While in PM, there was a large data scatter and a delayed [Ca 2+ ] i transient ( Figure 1B), this response was both more reproducible and prompt in UM 92.1 cells ( Figure 1A).

Functional TRPM8 Channel Expression in UM 92.1 Cells
Even though there was TRPM8 gene and functional expression in different UM cell lines including UM 92.1 cells, it was absent in human uveas (Mergler et al., 2014). To confirm that lack of TRPM8 expression is indicative of normal tissue, we probed for its presence in healthy PM. Icilin (20 µM), a mixed TRPM8/ TRPA1-agonist (Andersson et al., 2004;Rawls et al., 2007) induced a Ca 2+ transient in UM 92.1 cells (n = 4; p < 0.05; Figures 1D,F) whereas such an effect did not occur in PM (n = 8; p > 0.05; Figures 1E,F). Therefore, detectable functional TRPM8 and/or TRPA1 expression is a marker of UM cell line malignancy.
Irrespective of 3-T 1 AM ranging from 200 nM to 10 µM, its effects were essentially the same as those obtained with menthol (Figures 6A,B). The largest increases were obtained over a range between 1 and 5 µM (Lucius et al., 2016). In , and after addition of CAP (labeled as C). (C) Summary of patch-clamp experiments with VEGF and CAP. The asterisks (**) indicate statistically significant increase with VEGF (n = 7; p < 0.01; paired tested) and unchanged magnitude of currents in the presence of CAP (n = 4; p > 0.05; paired tested). (D) Maximal negative current amplitudes induced by a voltage step from 0 to −60 mV are depicted in percent of control values before application of 10 ng/ml VEGF. VEGF-induced inward currents (n = 7; *p < 0.05) were not increased in the presence of 20 µM CAP (n = 4; p > 0.05). (E) Maximal positive current amplitudes induced by a voltage step from 0 to +130 mV are depicted in percent of control values before application of 10 ng/ml VEGF. VEGF-induced outwardly rectifying currents (n = 7; *p < 0.05) were not increased in the presence of 10 µM CAP (n = 4; p > 0.05).

Icilin and Menthol Mimic Suppression by 3-T 1 AM of VEGF Transactivation of TRPV1
As a positive control, the effects of CAP and VEGF were determined using an alternative fluorescence Ca 2+ imaging data acquisition method as described in the method section. As shown in Figure 9A, a reduced CAP concentration (10 µM) led to an increase of f 340nm /f 380nm from 0.2021 ± 0.0004 (100 s) to 0.2094 ± 0.0013 (300 s) (n = 19; p < 0.005) whereas a washout did not reduce the Ca 2+ level. With 10 ng/mg VEGF instead of CAP (Figure 9B), this ratio increased from to 0.2011 ± 0.0004 (100 s) to 0.2331 ± 0.0029 (300 s) (n = 85; p < 0.005) and a washout did not augment this response Ca 2+ transient (Figures 9B,C). However, preincubation of the cells with icilin suppressed the VEGF-induced increase to 0.2058 ± 0.0023 at 300 s (p < 0.005) and to 0.2187 ± 0.0034 at 600 s (both n = 65; p < 0.01) (Figures 9C,D). Menthol had the same inhibitory effect as icilin. Functional TRPM8 expression was validated based on a correspondence between the transients induced by cooling from 20 to 18 • C (Figures 10A-C) and exposure to 200 µM menthol (Figures 10B,C). Furthermore, as with icilin, preincubation of the cells with menthol suppressed the VEGF-induced Ca 2+ transient even at a higher VEGF concentration since 20 ng/ml VEGF increased the f 340nm /f 380nm ratio from 0.1991 ± 0.011 (100 s) to 0.2212 ± 0.0021 (250 s) (n = 25; p < 0.005) ( Figure 10D). In contrast, 200 µM menthol completely blocked this effect (e.g., f 340nm /f 380nm = 0.2009 ± 0.0007 at 250 s; n = 32; p < 0.005) (Figures 10E,F). In summary, the near equivalence between the transients induced by either icilin, menthol, or 3-T 1 AM and their inhibitory effects on VEGF-induced TRPV1 transactivation confirms that this thyroid hormone metabolite is a selective TRPM8 agonist.

Cannabinoid Receptor Type 1 Activity Modulates Inhibition of TRPV1 by 3-T 1 AM
Since the G protein-coupled cannabinoid receptor 1 (CB1) is expressed in uveal melanoma cells (Mergler et al., 2014), we , and after addition of AMTB (labeled as C). (C) Summary of patch-clamp experiments with 3-T 1 AM and AMTB. The asterisks (**) indicate statistically significant increase with 3-T 1 AM (n = 9; p < 0.05 at the minimum; paired tested) and decreases in the presence of AMTB (n = 7; p < 0.05; paired tested).). (D) Maximal negative current amplitudes induced by a voltage step from 0 to −60 mV are depicted in percent of control values before application of 1 µM 3-T 1 AM. 3-T 1 AM-induced inward currents were clearly suppressed in the presence of 20 µM AMTB (n = 4; *p < 0.05). (E) Maximal positive current amplitudes induced by a voltage step from 0 to +130 mV are depicted in percent of control values before application of 1 µM 3-T 1 AM. 3-T 1 AM-induced outwardly rectifying currents (n = 5; **p < 0.01) were clearly suppressed in the presence of 20 µM AMTB (n = 4; *p < 0.05).
determined if either this receptor or its coupled G-proteins affect interactions between TRPM8 and TRPV1 in UM cells. To deal with this question, the individual effects of the selective CB1 receptor antagonist, AM251, and a corresponding agonist, WIN 55,212-2 (both 10 µM) were characterized by measuring their effects on [Ca 2+ ] i in UM 92.1 cells. WIN 55,212-2 induced a Ca 2+ transient at a different cell passage compared to our previous studies (n = 27; p < 0.005). This validates CB1 involvement in Ca 2+ regulation in UM 92.1 cells ( Figure 11A). Interestingly, the WIN-induced Ca 2+ increase was at significant higher levels under Ca 2+ free conditions (n = 53; p < 0.005) (Figures 11B,C). Similarly, 1 µM 3-T 1 AM also induced such a response. On the other hand, preincubation of the cells with the CB1 blocker AM251 (10 µM) augmented this rise induced by 3-T 1 AM. The transient reached with 3-T 1 AM by itself was 0.2155 ± 0.0014 (n = 13) at 400 s only whereas with AM251 (10 µM) it rose to 0.2446 ± 0.0037 at the same time (n = 46; p < 0.005) (Figures 11D,E,G). In contrast, 3-T 1 AM failed to induce a transient under Ca 2+ free conditions (n = 39) (Figures 11F-G). In summary, changes in CB1 activity and/or coupled G-protein activity modulate interactions between TRPV1 and TRPM8.

DISCUSSION
We describe here TRPV1 and TRPM8 functional activity and their interactions in modulating VEGF-induced signaling in UM cells. Even though a short-term UM cell culture (P < 20) was used in this study, variations may occur in gene expression profiles between UM primary tumors and their derived cell lines. Nevertheless, there were only moderate modifications in the gene products (Mouriaux et al., 2016). Accordingly, Mouriaux et al. suggested that cell lines might represent useful tools in functional show significant Ca 2+ increases with 3-T 1 AM (n = 9; 350 s; *p < 0.05; 490 s; **p < 0.01; paired tested). The hashtag (#) indicates a statistically significant difference of fluorescence ratios between 3-T 1 AM with and without BCTC (n = 6; 490s; # p < 0.05; unpaired tested). (D) Mean trace showing VEGF-induced Ca 2+ increase (n = 6) whereas non-treated control cells showed a constant Ca 2+ baseline (n = 9). (E) Same experiment as shown in (D), but in the presence of 3-T 1 AM (1 µM). 3-T 1 AM clearly suppressed the VEGF-induced Ca 2+ increase (n = 6). (F) Summary of the experiments with VEGF and 3-T 1 AM. The asterisk (*) shows a significant Ca 2+ increase with VEGF (n = 6; 400 s; *p < 0.05; paired tested). The hashtags (##) indicate statistically significant differences of fluorescence ratios between VEGF with and without 3-T 1 AM (n = 6; 400 s; ## p < 0.01; unpaired tested). assays, as well as pharmacologic and genetic studies (Mouriaux et al., 2016). The TRPV1 and TRPM8 functional activity identified in these UM cells is consistent with the correspondence between the mRNA and protein expression patterns previously described in several other UM cell-line types (Mergler et al., 2014). Similarly, functional interaction between TRPM8 and TRPV1 is evident because TRPM8 activation inhibited increases in TRPV1 functional activity induced by CAP in both corneal epithelial and endothelial cells (Khajavi et al., 2015;Lucius et al., 2016). Furthermore, TRPM8 activation blunts transactivation of TRPV1 by VEGF in UM cells (Figures 7, 8). This modulation of VEGF-induced increases in Ca 2+ influx mediated by TRPV1 activation shows that this receptor triad contributes through crosstalk to the growth promoting effects of VEGF in UM cells derived from malignant tumors. Such crosstalk is consistent with other studies wherein TRPM8 activation dampens CAP-induced TRPV1 activation by VEGF (Millqvist, 2016;Takaishi et al., 2016). This consistency in interactions among this receptor triad in different cell types suggests that results obtained with one cell type may be applicable to various cell types.
activity (Figures 2, 3). Its functional expression was also clearly detectable in both primary cultivated PM and healthy human uveas (Mergler et al., 2014). However, TRPV1 receptor density was probably at lower levels because of more extensive data scatter and a delayed response to CAP in most PM measurements ( Figure 1B). Nevertheless, the maximal rises in Ca 2+ influx in normal PM cells closely correspond to those in UM 92.1 cells.
The dynamic range of our Ca 2+ imaging system was limited to 0.20 for detecting increases in the fluorescence ratio because the initial fluorescence responses of the two exciting wavelengths were at a relatively high level, which compressed the dynamic range of our measurements due to mathematical reasons. Therefore, even ratio changes of only 0.015 for CAP were significant and the measurements were clearly discernible and reproducible ( Figure 1A). Another indication of the adequate resolving power of our measurements is that the effects of CAP and icilin especially in UM 92.1 cells were irreversible and reached a steady state in most experiments. Similarly, this irreversibility was described in other eye tumor cells (Mergler et al., 2012a(Mergler et al., , 2014Garreis et al., 2016) as well as healthy eye surface cells (Khajavi et al., 2015;Lucius et al., 2016).

Different TRPM8 Expression Levels in UM 92.1 Cells and PM
In healthy human uveas, TRPM8 mRNA expression was not evident whereas the TRPA1 PCR signal was present at very high levels (Mergler et al., 2014). On the other hand, icilin (Rawls et al., 2007) increased Ca 2+ transients in UM 92.1 cells (Figure 1D) whereas this effect was not evident in PM (Figures 1E,F). Even though menthol activates TRPM8 (Eccles, 1994;Galeotti et al.,  2002; Bautista et al., 2007;Pedretti et al., 2009), the magnitudes of these transients did not correlate with the TRPM8 expression levels in certain cancer cells indicating a TRPM8-independent signaling pathway (Naziroglu et al., 2018). Irrespective of that, menthol also activated TRPM8 in the absence of extracellular Ca 2+ , whereas responses to icilin are Ca 2+ dependent (McKemy et al., 2002;Chuang et al., 2004). Therefore, icilin appears to induce Ca 2+ transients by increasing Ca 2+ influx from the extracellular medium (McKemy et al., 2002;Andersson et al., 2004;Behrendt et al., 2004). Nevertheless, TRPA1 involvement cannot be excluded because icilin can also interact with TRPA1 even though icilin was relatively ineffective at inducing Ca 2+ transients and there is no detectable TRPA1 mRNA expression in this cell type (Mergler et al., 2014). Therefore, TRPM8 activity in PM and human uveas is relatively low compared to higher levels in malignant uveal cell types (Mergler et al., 2014). In contrast, the CAP-induced Ca 2+ transients were comparable suggesting no difference in TRPV1 expression levels in these two different cell types ( Figure 1B).

Crosstalk Between VEGFR and TRPV1
VEGF transactivated TRPV1 through its interaction with VEGFR since the Ca 2+ transients and their underlying currents were both fully blocked during exposure to CPZ. Furthermore, the stimulation by VEGF of TRPV1 was maximal since supplementation with CAP failed to augment the increases in whole-cell currents induced by VEGF application (Figures 4, 7, 8).
Unlike with CPZ, the TRPM8 blocker AMTB (Lashinger et al., 2008) failed to suppress these transients induced by VEGF suggesting that VEGFR solely transactivates TRPV1 (Figure 2). This agrees with what was described in a benign tumor (Garreis et al., 2016). The smaller inhibitory effects of BCTC on VEGFinduced increases in Ca 2+ influx than those induced by CPZ are supportive of the notion that BCTC is a mixed TRPV1/TRPM8 antagonist (Behrendt et al., 2004;Weil et al., 2005;Vriens et al., 2009;Benko et al., 2012;Liu et al., 2016). In contrast, AMTB is a more selective TRPM8 antagonist (Lashinger et al., 2008) since it failed to block VEGF-induced transactivation of TRPV1 (Figures 2D, 3A,B). The limited specificity of BCTC as a TRPV1 blocker is supported by our finding that at a relatively high CAP concentration (20 µM), BCTC was ineffective as a blocker of TRPV1 activation in human corneal epithelial cells (Lucius et al., 2016). The marked difference between the large inhibitory effect of CPZ and the minimal effect of AMTB on increases in currents induced by VEGF confirms that VEGF solely transactivates TRPV1 (Figure 3). However, VEGFR is also known to regulate multiple channels including TRPs (Garnier-Raveaud et al., 2001;Hamdollah Zadeh et al., 2008;Thilo et al., 2012;Reichhart et al., 2015;Wu et al., 2015;Qin et al., 2016). Specifically, McNaughton et al. demonstrated that nerve growth factor (NGF) receptors in HEK293 cells transfected with plasmids containing cDNAs coding for TRPV1 and for the Tropomyosin receptor kinase A (TrkA) receptor for NGF increased the expression level of TRPV1 but did not sensitize or activate the receptor (Zhang et al., 2005;Vay et al., 2012). One explanation may be that NGF is different from VEGF or that our study used non-excitable (tumor) cells.
The 3-T 1 AM mediated Ca 2+ transients as well as increases in their underlying currents occurred at lower concentrations in UM 92.1 melanoma cells than those in healthy cells or thyroid cells (Khajavi et al., 2017;Schanze et al., 2017). Specifically, 3-T 1 AM was used over a concentration range from 0.2 to 10 µM with 1-5 µM having maximal stimulatory effects on whole-cell currents, which agrees with previous studies using corneal epithelial cells (Lucius et al., 2016). 3-T 1 AM had a concentration dependent inhibitory effect on Ca 2+ transients that may be attributable to different modes of action. With 0.5 µM 3-T 1 AM, only whole-cell currents increased without any inhibitory effect on VEGF-induced rises in Ca 2+ influx (data not shown). However, with 1 µM 3-T 1 AM VEGF-induced rises in Ca 2+ influx also declined with a time course that was similar to that obtained with CPZ (Figures 2B, 7E, 11D). It is conceivable that the effect of 0.5 µM 3-T 1 AM is membrane delimited rather than causing release of Ca 2+ from intracellular stores. Perhaps, 1 µM 3-T 1 AM is at a high enough concentration for it to gain access to cytosolic intracellular TRPV1 binding sites with which CPZ also binds? If this supposition is proven to be correct, FIGURE 12 | Suggested Ca 2+ signal transduction model accounting for how TRPM8 activation affects receptor-linked signaling pathways (Mergler et al., 2014;Khajavi et al., 2017). Ca 2+ channels such as TRPs of the TRPV1 subtype (capsaicin receptor) can be selectively activated by heat (>43 • C) or capsaicin and blocked by CPZ ( Figure 1A) (Mergler et al., 2014). VEGF-A activating VEGFR-1/VEGFR-2 can also activate TRPV1 (Figures 2, 3). The TRPM8 subtype (menthol receptor) can be selectively activated by cold (23-28 • C), menthol, icilin or 3-T 1 AM and blocked by BCTC/AMTB ( Figures 1D, Figures 5, 6) (Mergler et al., 2014;Khajavi et al., 2017). Notably, a G-protein coupled receptor (GPCR) coupled to Gi/o proteins could be activated by 3-T 1 AM (Dinter et al., 2015a;Khajavi et al., 2017;Schanze et al., 2017). 3-T 1 AM may also directly activate TRPM8 by a GPCR-independent mechanism (↑[Ca 2+ ] i ]) (Khajavi et al., 2017). If TRPM8 is activated by 3-T 1 AM, 3-T 1 AM suppresses VEGRF via TRPV1 (Figures 7, 8). Notably, 3-T 1 AM may also directly suppress TRPV1 via VEGFR by a GPCR-independent mechanism (↓[Ca 2+ ] i ]) ( Figure 8) Menthol and icilin mimic the 3-T 1 AM effect and suppresses increases in TRPV1 activity by VEGF (Figures 9, 10).
3-T 1 AM may have dual effects that include activating TRPM8 and at higher concentrations also suppressing TRPV1 activation induced by VEGF. As the effects of modulators of TRP channel activity on Ca 2+ influx mirrored those on whole-cell currents (Figures 7, 8), this agreement supports the notion that 3-T 1 AM has a relevant role in regulating VEGF-mediated Ca 2+ regulation in tumor cells. In addition, we could demonstrate that TRPM8 activation by menthol or icilin mimic the 3-T 1 AM effect and is able to suppress increases in TRPV1 activity by VEGF (Figures 9,  10). To the best of our knowledge, this report is the first one describing such control in both benign eye surface tumor cells (Garreis et al., 2016) and now in metastatic UM 92.1 cells. On the other hand, the possible Ca 2+ signal transduction pathways activated by 3-T 1 AM as a specific activator of TRPM8 may be more complex as suggested in Figure 12 (Khajavi et al., 2017).

G Protein-Coupled Cannabinoid Receptor-1 Modulates 3-T 1 AM Suppression of TRPV1 Channels
It has been suggested that 3-T 1 AM is a multitarget ligand (Zucchi et al., 2014) interacting with different TRP channel subtypes including TRPM8 (Khajavi et al., 2015(Khajavi et al., , 2017Lucius et al., 2016). Since functional CB1 expression has been described in ocular tumor cells (Mergler et al., 2012a(Mergler et al., , 2014 and in healthy ocular cells (Stumpff et al., 2005;Yang et al., 2010Yang et al., , 2013, we determined if 3-T 1 AM interacts also with the cannabinoid receptor 1 CB1. CB1 activation by WIN 55,212-2 induced Ca 2+ transients, which were larger in a Ca 2+ free conditions than with 2 mM external Ca 2+ (Figures 11A-C). On the other hand, 3-T 1 AM failed to elicit a Ca 2+ transient in a Ca 2+ free medium ( Figure 11F) whereas in the presence of AM251 and external Ca 2+ in the medium 3-T 1 AM-induced Ca 2+ transients that were larger in the presence of the CB1 blocker than in its absence (Figures 11D,E). Since CB1 activation also suppresses TRPV1 activation (Yang et al., 2013;Mergler et al., 2014), there may be an inverse relationship between increases in TRPM8 activity and declines in CB1 activity. Overall, the mechanisms involved in 3-T 1 AM modulation of TRPV1 channels may also involves contributions by other receptors such as CB1 and/or its coupled G-protein mediators.

Clinical Relevance
3-T 1 AM is a potential therapeutic agent for suppressing UM expansion as already indicated in other studies demonstrating that this thyroid hormone metabolite reduces cancer cell growth and viability (Rogowski et al., 2017;Shinderman-Maman et al., 2017). In UM cells, modulation of their metastatic activity appears to include changes in TRPV1 activity induced by TRPM8 and possibly CB1. The model for such control shown in Figure 12 proposes that VEGF secreted by UM cells stimulates intracellular Ca 2+ influx in endothelial cells, which is a requisite for driving angiogenesis and promotes UM proliferation and metastasis. Since TRPM8 activation has an opposing effect on TRPV1 activity, targeting TRPM8 may provide an effective alternative to suppress metastatic melanoma progression without side effects. Such an approach appears to be tenable based on the fact that functional TRPM8 expression is only detectable in the UM cells rather than PM cells. There is an urgent need for further assessment of the validity of this option since there are no measures currently available to reverse melanoma metastasis.

AUTHOR CONTRIBUTIONS
LW, CB, SM, and PR designed the study, analyzed the data, wrote and edited the manuscript. LW and CB contributed equally to the work. JK contributed with his expertise on thyroid hormone metabolites, discussed data and their interpretation and helped edit the manuscript. LW, CB, AB, PD, NL, MS, HvdW, IR, and SM performed calcium measurements and planar patchclamp recordings as well as plot analyses. LW, CB, AB, PD, MS, HvdW, and SM created diagrams. SM created the schematic drawing.