Novel Protein Kinase C-Mediated Control of Orai1 Function in Invasive Melanoma

Abstract

The incidence of malignant melanoma, a cancer of the melanocyte cell lineage, has nearly doubled in the past 20 years. Wnt5A, a key driver of melanoma invasiveness, induces Ca²⁺ signals. To understand how store-operated calcium entry (SOCE) contributes to Wnt5A-induced malignancy in melanoma models, we examined the expression and function of STIM1 and Orai1 in patient-derived malignant melanoma cells, previously characterized as either highly invasive (metastatic) or noninvasive. Using both fluorescence microscopy and electrophysiological approaches, we show that SOCE is greatly diminished in invasive melanoma compared to its level in noninvasive cell types. However, no loss of expression of any members of the STIM and Orai families was observed in invasive melanoma cells. Moreover, overexpressed wild-type STIM1 and Orai1 failed to restore SOCE in invasive melanoma cells, and we observed no defects in their localization before or after store depletion in any of the invasive cell lines. Importantly, however, we determined that SOCE was restored by inhibition of protein kinase C, a known downstream target of Wnt5A. Furthermore, coexpression of STIM1 with an Orai1 mutant insensitive to protein kinase C-mediated phosphorylation fully restored SOCE in invasive melanoma. These findings reveal a level of control for STIM/Orai function in invasive melanoma not previously reported.

INTRODUCTION

Melanoma is a cancer of the melanocyte cell lineage, pigmented cells predominantly found in the skin and eyes, and is responsible for the production of the melanin pigment that defines skin and eye tone (1). While cutaneous cancer is easily treatable by surgical excision when detected early, the 5-year survival rates for invasive melanomas is only 15% (2). Addressing the challenge of this aggressive cancer cell type requires new insight into the control mechanisms that distinguish noninvasive and invasive melanoma cells.

The transition from a highly proliferative to an invasive state is often referred to as phenotype switching and, in melanoma, is characterized by a corresponding shift in Wnt signaling. Hence, canonical β-catenin-mediated Wnt signaling driven by Wnt1 and Wnt3A has been shown to drive tumor development by promotion of melanocyte transformation (3). However, as melanoma cells transition into a metastatic phenotype, noncanonical Wnt signaling dominates with the activation of Wnt5A (3). Wnt5A activation downregulates β-catenin via the activation of SIAH2 (4, 5) and increases invasiveness. Wnt5A binds to the tyrosine kinase receptor ROR2 and members of the GPCR frizzled family (FZD2 and FZD5), leading to activation of phospholipase C-γ (PLC-γ) and downstream production of diacylglycerol (DAG) and the Ca²⁺-mobilizing second messenger inositol 1,4,5-trisphosphate (InsP₃). Ca²⁺ signals regulate numerous aspects of cell function which are altered during the transition to invasiveness, including adhesion, migration, autophagy, and apoptosis (6, 7). With such a wide array of potential Ca²⁺-sensitive physiological responses, current efforts are focused on linking specific Ca²⁺ signaling molecules with specific physiological and pathological situations.

Over the last 10 years, members of the STIM and Orai families have been well characterized as the molecular mediators of store-operated Ca²⁺ entry (SOCE) (8). In brief, STIM1 and STIM2 are endoplasmic reticulum (ER) Ca²⁺ sensors that respond to decreases in ER Ca²⁺ content by activating Orai1, Orai2, or Orai3, which are Ca²⁺ channels located on the plasma membrane (PM) (8). Although STIM1 and Orai1 are the primary mediators of this process in most cell types, in estrogen-responsive breast cancer, Orai3 replaces Orai1 as the mediator of this process (9), while in dendritic cells, STIM2 replaces STIM1 as the primary ER Ca²⁺ sensor (10). As such, there is a need to examine the function and regulation of SOCE in each physiological and pathophysiological scenario. Over the last few years, there have been several studies addressing the function and role of SOCE in invasive melanoma. Hence, in murine B16 melanoma cells, SOCE was shown to drive growth and survival via increased AKT activity (11, 12). Further, there were 3 studies published this year showing increased expression and/or function of STIM and Orai in invasive melanoma (13–15). Considered collectively, these studies show that Ca²⁺ entry supports the migration and invasion by melanoma cells. However, in 2004, an unbiased screen for metastasis-related genes in an invasive melanoma model revealed STIM1 as a repressor of metastasis (16). This apparent inconsistency with current thinking is supported by the surprising findings of the current study, in which we examined the expression and function of STIM1 and Orai1 in a series of melanoma cell lines exhibiting marked differences in Wnt5A expression with corresponding differences in invasive character (17, 18). Remarkably, we found that all of the cell lines exhibiting elevated Wnt5A expression exhibited a profound loss of STIM1/Orai1 function. However, this loss of function did not reflect any change in the level of STIM and Orai expression or coupling. Instead, loss of SOCE in Wnt5A-expressing invasive melanoma was attributed to Wnt5A- and protein kinase C (PKC)-dependent Orai1 phosphorylation.

MATERIALS AND METHODS

Cell culture.

FS13, FS14, G361, UACC1273, FS4, FS5, M93-047, and UACC903 cells were cultured in RPMI 1640 with l-glutamine supplemented with 10% fetal bovine serum (FBS) and antibiotics. All cells were maintained at 37°C in 5% CO₂.

Generating stable cell lines.

Wnt5A short hairpin RNA (shRNA) was obtained from the TRC shRNA library available through the Molecular Screening Facility at The Wistar Institute. Clones transfected into FS4 cells were TRCN0000062717 (CGTGGACCAGTTTGTGTGCAA) and TRCN0000062716 (CACATGCAGTACATCGGAGAA).

Wnt5A overexpression vectors transfected into FS13 cells were the following: Wnt5a⁺, pLU-EF1a-Wnt5a(4) mCherry; and Wnt5a control, pLU EF1a-mcherry.

Lentiviral production was carried out as described in the protocol developed by the TRC library (Broad Institute). Briefly, 293T cells were cotransfected with the indicated vectors and lentiviral packaging plasmids (pCMV-dR8.74psPAX2 and pMD2.G). The supernatant containing virus was harvested at 36 and 60 h, combined, and filtered through a 0.45-mm filter. For transduction, FS4 and FS13 cells were layered overnight with lentivirus along with 8 mg/ml Polybrene. Cells were allowed to recover for 24 h. shRNA clones were selected using 1 mg/ml puromycin, and Wnt5A-overexpressing cells were selected by sorting for mCherry positivity on MoFlo AstriosEQ (Beckman Coulter) prior to antibiotic selection.

Cytosolic Ca²⁺ measurements.

Cells were plated on coverslips and placed in cation-safe solution (107 mM NaCl, 7.2 mM KCl, 1.2 mM MgCl₂, 11.5 mM glucose, 20 mM HEPES-NaOH, 1 mM CaCl₂, pH 7.2) and loaded with fura-2–acetoxymethylester (2 μM) for 30 min at 24°C as previously described (19). Cells were washed, and dye was allowed to deesterify for a minimum of 30 min at 24°C. Approximately 85% of the dye was confined to the cytoplasm, as determined by the signal remaining after saponin permeabilization. Ca²⁺ measurements were made using a Leica DMI 6000B fluorescence microscope controlled by SlideBook software (Intelligent Imaging Innovations, Denver, CO). Fluorescence emission at 505 nm was monitored while alternating between 340- and 380-nm excitation wavelengths at a frequency of 0.67 Hz; intracellular Ca²⁺ measurements are shown as 340/380-nm ratios obtained from groups (35 to 45) of single cells.

Electrophysiological recordings.

Conventional whole-cell patch clamp recordings were carried out using an Axopatch 200B and Digidata 1440A (Axon Instruments, NY) as previously published (20), with a few important modifications. To reduce the noise to a minimum, we added in series a humbug noise eliminator that eliminates electrical interference, such as simple 50/60Hz sine waves, mixtures of 50/60Hz harmonics, noise spikes from dimmers, and complex noise from fluorescent lamps. All experiments were performed at room temperature (20 to 25°C). Pipettes were pulled from borosilicate glass capillaries (World Precision Instruments, Inc., Sarasota, FL) with a P-97 flaming/brown micropipette puller (Sutter Instrument Company, Novatao, CA) and polished with DMF1000 (World Precision Instruments). Resistances of filled glass pipettes were 2 to 3 MΩ. Series resistances were in the range of 5 to 10 MΩ. The liquid-junction potential offset was around 4.6 mV and was corrected. Only cells with tight seals (>16 GΩ) were selected for break-in. Immediately after establishing the whole-cell patch clamp configuration, we start the recording by running a 250-ms voltage ramp (from +100 mV to −140 mV) every 2 s and performing a first divalent-ion-free (DVF) pulse when current development is minimal. The first current/voltage (I/V) curves obtained immediately after break-in in Ca²⁺-containing bath solutions and DVF bath solutions represent background currents that are subtracted from 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA)-activated Ca²⁺ currents and Na⁺ currents. After the currents are fully activated by store depletion, leak-subtracted I/V curves are obtained for Na⁺ currents (in DVF bath solutions). Using Origin Lab 7.5 software (OriginLab, Northampton, MA), I/V curves corresponding to background currents obtained in Na⁺ are subtracted from the I/V curves obtained in Na⁺ after maximal current activation. The subtracted I/V curves are represented as independent I/V curves. Cells were maintained at a 0-mV holding potential during experiments. High MgCl₂ (8 mM) was included in the patch pipette to inhibit TRPM7 currents.

Solutions employed for whole-cell patch clamp electrophysiology. (i) Pipette solution.

The pipette solution contained Cs-methanesulfonate (105 mM), 20 mM Cs-BAPTA, 8 mM MgCl₂, and 10 mM HEPES (pH was adjusted to 7.2 with CsOH).

(ii) Bath solution.

The bath solution contained Na-methanesulfonate (115 mM), 10 mM CsCl, 1.2 mM MgSO₄, 10 mM HEPES, 20 mM CaCl₂, and 5 mM glucose (pH was adjusted to 7.4 with NaOH).

(iii) Divalent-ion-free bath solution.

The divalent-ion-free bath solution contained Na-methanesulfonate (140 mM), 10 mM HEDTA, 1 mM EDTA, and 10 mM HEPES (pH was adjusted to 7.4 with NaOH).

Cell lysis and Western blotting.

Cells were lysed in 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS) lysis buffer (35 mM CHAPS, 150 mM NaCl, 100 mM Tris-HCl, 10 mM EDTA, pH 8.0, with protease inhibitors), cleared by centrifugation, and normalized for protein content (determined using the DC protein assay kit; Bio-Rad). Proteins were resolved on SDS-PAGE gels, transferred to nitrocellulose paper, and analyzed with the indicated antibodies as previously described (19).

Invasion assays.

Invasion assays were performed using transwell migration chambers (Corning Life Sciences, Lowell, MA) as previously described (21). In brief, 8-μm filters were coated with 150 μl of 80 μg/ml Matrigel (Becton Dickinson, Franklin Lakes, NJ). Melanoma cells were seeded onto filters in media supplemented with 10% FBS at 3 × 10⁴ (FS4) or 5 × 10⁴ (FS5) cells/well. Media containing 20% fetal calf serum (FCS) was placed in the lower well to act as a chemoattractant. Cells were allowed to invade and adhere to the lower chamber, stained using crystal violet, and counted using ImageJ.

Fluorescence imaging and FRET measurements.

Fluorescence imaging and FRET measurement experiments were performed in melanoma cells transiently transfected with Orai1-cyan fluorescent protein (CFP) and STIM1-yellow fluorescent protein (YFP). Fluorescence was examined with a Leica DMI 6000B fluorescence microscope equipped with CFP (excitation [Ex] wavelength of 436 nm and emission [Em] wavelength of 480 nm [436_Ex/480_Em]) and YFP (500_Ex/535_Em) filters controlled by SlideBook software (Intelligent Imaging Innovations; Denver, CO). All images were obtained at room temperature with a 63× oil objective (numeric aperture, 1.4; Leica), and image analyses were performed using SlideBook software. CFP (436_Ex/480_Em) and the raw Förster resonance electron transfer score (FRET_raw) (436_Ex/535_Em) were monitored at a rate of 0.2 Hz. Three-channel corrected FRET (FRET_c) was calculated with the following formula: FRET_c = (F_raw − F_a/A_a × F_YFP − F_d/D_d × F_CFP)/F_CFP, where FRET_c represents the corrected total amount of energy transfer, F_raw represents fluorescence measured through the CFP/YFP ET filter cube, F_a/A_a represents direct excitation of the acceptor, F_CFP represents measured CFP fluorescence, and F_d/D_d represents measured bleed-through of CFP through the YFP filter (0.592248). Fluorescence images and FRET analyses shown are typical of at least three separate experiments.

Quantitative PCR.

RNA was extracted using TRI Reagent (MRC). Briefly, 1 ml of TRI Reagent was added per 10-cm culture dish. Homogenate was stored at room temperature for 5 min, followed by the addition of 0.2 ml chloroform to allow for phase separation. The mixture was incubated at room temperature for 10 min and then centrifuged at 4°C at 12,000 × g for 15 min. The aqueous phase was removed and mixed with 0.5 ml ethanol, followed by centrifugation at 4°C at 12,000 × g for 15 min to allow for RNA precipitation. Reverse transcription-PCR (RT-PCR) was completed as a two-step process using high-capacity cDNA reverse transcription (Applied Biosystems) followed by Sybr green PCR (Invitrogen) on a 7300 real-time PCR machine (Applied Biosystems). The real-time reaction was carried out by a 10-min incubation at 95°C and 40 cycles at 95°C for 15 s and 60°C for 1 min for annealing. Data are presented as a level relative to that of TATA-box binding protein (TBP) calculated using the formula 2^{(a − b)} (where a is the cycle threshold for the cDNA of interest and b is the cycle threshold of TATA-binding protein) and primers listed in Table 1.

TABLE 1.

qPCR primers

Target	Sequence
	Sense	Antisense
Wnt5A	5′-TAAGCCCAGGAGTTGCTTTG-3′	5′-GCAGAGAGGCTGTGCTCCTA-3′
MART1	5′-AGAAGATGCCCACAAGAAGG-3′	5′-CGCTGGCTCTTAAGGTGAAT-3′
STIM1	5′-TGGAGAAGGTCCATCTGGAAA-3′	5′-TGGGCTTCCTGCTTAGCAA-3′
STIM2	5′-GGGGTTCTCCCGACTGT-3′	5′-TGTGGCGAGGTTTAGGC-3′
Orai1	5′-CGCTGACCACGACTACCCA-3′	5′-CTCCTTGACCGAGTTGAGATT-3′
Orai2	5′-CCTCATCTTCGTGGTCTT-3′	5′-CTCGATCTCGCGGTTG-3′
Orai3	5′-GGGGCTCGTGTTTGTG-3′	5′-GCAGGCGATTCAGTTC-3′
TBP	5′-CAGCCGTTCAGCAGTCAA-3′	5′-GGAGGGATACAGTGGAGT-3′

Materials.

Anti-STIM1 antibody was from BD Biosciences (San Diego, CA). Anti-Orai1, histamine, and ionomycin were from Sigma Biosciences (St. Louis, MO). Anti-STIM2 antibodies were from Cell Signaling Technology (Danvers, MA). Thapsigargin was from EMD Biosciences (San Diego, CA), and gö6983 was from Tocris Bioscience (Bristol, United Kingdom). Fura-2–acetoxymethylester and Power SYBR green PCR master mix were obtained from Invitrogen (Grand Island, NY). All primers were synthesized by Biosearch Technologies (Petaluma, CA).

Plasmids.

cDNA encoding enhanced YFP-tagged human STIM1 (STIM1-YFP) was obtained from Anjana Rao (La Jolla Institute for Allergy & Immunology) via Addgene (Cambridge, MA). Orai1-CFP was generated as previously described (22). CFP-Orai1_Δ47 was generously provided by Christoph Romanin (Johannes Kepler University, Linz, Austria). CFP-Orai1_S27AS30A was generated by site-directed mutagenesis (BioInnovatise Inc., Rockville, MD).

RESULTS AND DISCUSSION

SOCE is reduced in invasive melanoma.

To assess possible differences in SOCE during melanoma progression, we selected 4 noninvasive (G361, UACC1273, FS13, and FS14) and 4 invasive (FS4, FS5, M93-047, and UACC903) melanoma cell lines characterized in prior studies based on both their capacity to metastasize in vivo and their levels of Wnt5A expression (17, 18). As expected, Wnt5A was virtually undetectable by quantitative PCR (qPCR) in all four of the noninvasive cell lines but was strongly expressed in all 4 invasive cell lines (Fig. 1A). We also confirmed that melanoma antigen recognized by T-cells 1 (MART1) expression was lost in the invasive melanoma lines only (Fig. 1B), consistent with prior studies indicating dedifferentiation as typically observed during melanoma progression. We then proceeded to measure the amount of SOCE in these cells. Preliminary experiments to identify receptor-induced Ca²⁺ responses in the noninvasive melanoma cell lines were unsuccessful using a wide array of different agonists, including bradykinin (100 μM), UTP (100 μM), histamine (100 μM), carbachol (100 μM), epidermal growth factor (EGF; 25 ng/μl), and platelet-derived growth factor (PDGF; 10 ng/μl), although histamine stimulated ER Ca²⁺ release in 3 out of 4 invasive cell lines (see Fig. S1 in the supplemental material). Intriguingly, the subsequent addition of 1 mM extracellular Ca²⁺ led to minimal Ca²⁺ entry in these 3 invasive cell lines. To avoid this apparent variability in receptor expression patterns (as well as the potential involvement of store-independent Ca²⁺ channels) among these different cell lines, all subsequent SOCE experiments were performed using the highly selective sarcoendoplasmic reticulum (SERCA) inhibitor thapsigargin (Tg) to deplete ER Ca²⁺ stores. We observed similar amounts of ER Ca²⁺ release in 7 of 8 cell lines, with FS5 exhibiting unusually small responses (see Fig. S2). Release of all sequestered Ca²⁺ using ionomycin revealed a similar pattern (see Fig. S3), indicating that this is a reflection of the cell types as opposed to differences in Tg sensitivity. Consistent with the experiments performed using histamine, SOCE was virtually undetectable in any of the invasive melanoma cell lines; no similar defect in SOCE was observed in noninvasive cells (Fig. 1C and D; see Fig. S1). Therefore, we performed a series of studies to reveal the mechanism(s) responsible for this surprising phenomenon.

FIG 1.

SOCE is diminished in invasive melanoma cells. (A and B) RNA was extracted from melanoma cell lines, and then cDNA was prepared by reverse transcription. Custom qRT-PCR primers were designed to amplify 50- to 60-bp products against human Wnt5A (A) or MART1 (B). qRT-PCR was carried out using Power SYBR green (Applied Biosystems) per the manufacturer's instructions. Target levels were normalized to those of TATA-binding protein (TBP) and are depicted as means ± standard errors of the means (SEM) (n = 3). (C) Representative traces of store-operated Ca²⁺ entry in 4 invasive and 4 noninvasive human melanoma lines. Cells were plated on glass coverslips and loaded with fura-2–acetoxymethylester (2 μM) prior to treating with thapsigargin (Tg; 2 μM), a SERCA inhibitor, to deplete ER Ca²⁺ stores (not shown). Once cytosolic Ca²⁺ had returned to baseline, 1 mM Ca²⁺ was added to the cells, invoking SOCE. Ca²⁺ measurements are shown as 340/380-nm ratios obtained from groups of 30 to 50 single cells. (D) The magnitude of Ca²⁺ elevation (as depicted in panel C) was quantified and is presented as a scatter plot. Each symbol represents the average change in cytosolic Ca²⁺ content occurring after addition of Ca²⁺ to 30 to 50 Tg-treated cells in a single experiment. One-way analysis of variance (ANOVA) indicates significantly decreased SOCE in invasive melanoma (P < 0.001). (E to H) Representative electrophysiological recordings from noninvasive FS14 (E) and invasive M93-047 (F) cells were obtained in the whole-cell configuration at −100 mV. (G) Subtracted Na⁺ I/V curve obtained at the points marked by asterisks in panels E and F revealing the developing CRAC in FS14 but not M93-047 cells. (H) Average current density in FS14 and M93-047 cells (−100 mV; n = 5). ***, P = 0.0004.

Although Tg-induced Ca²⁺ release is a widely used assay for measuring SOCE, it is possible that the amount of Ca²⁺ entry observed using this assay can be skewed by alternative channels, differences in membrane potential, and/or differences in pumping activity. To address this possibility, we performed electrophysiological SOCE measurements (commonly referred to as Ca²⁺ release-activated Ca²⁺, or CRAC, measurements). As this assay is labor-intensive, FS14 and M93-047 cells were selected as representative examples of the noninvasive and invasive melanoma phenotypes. Although CRAC is a highly Ca²⁺ selective channel, it becomes permeable to Na⁺ in divalent-ion-free (DVF) solutions. Due to the extremely small size of the Ca²⁺ currents, this technique was used to amplify currents in both cell types for more accurate measurements, as previously described (20). Upon break-in, diffusion of the Ca²⁺ chelator BAPTA into the cell leads to ER Ca²⁺ depletion within ∼5 min. In FS14 cells, replacement of Ca²⁺-containing bath solutions with DVF solutions immediately after break-in leads to the appearance of a relatively small Na⁺ current that is symmetrical in nature, likely the result of cell membrane or seal leak due to the absence of divalent cations (Fig. 1E). Subsequent additions of DVF were made after BAPTA perfusion had started to deplete the ER of Ca²⁺. This led to steadily increasing inward Na⁺ currents, while outward currents remained unchanged. Note that a small Ca²⁺ current also was observed when Ca²⁺-containing bath solutions were used. Analysis of the leak-subtracted store depletion-activated Na⁺ currents revealed inward rectification and a highly positive reversal potential typical of CRAC currents (Fig. 1G). Further, these Na⁺ currents exhibited the rapid depotentiation typical of CRAC currents developing in DVF (Fig. 1E). In contrast, a comparable experiment in M93-047 cells failed to stimulate either a Ca²⁺ current or any change in inward Na⁺ current in DVF bath solutions upon store depletion (Fig. 1F and H). As such, we are confident that loss of SOCE in invasive melanoma reflects a bona fide loss of CRAC channel activity and not the involvement of alternative pumps or channels.

Changing Wnt5A expression alters SOCE.

Since all Wnt5A-expressing cells exhibit profound loss of SOCE, it is possible that Wnt5A is a negative regulator of SOCE. Therefore, we generated FS4 invasive melanoma cells stably expressing Wnt5A shRNA. Robust Wnt5A expression in FS4 cells was virtually eliminated in both Wnt5A shRNA-expressing cell lines (Fig. 2A), leading to a profound decrease in invasiveness (Fig. 2B). Wnt5A knockdown had a similar impact on the invasiveness of FS5 cells (see Fig. S4 in the supplemental material). Remarkably, this manipulation led to a severalfold increase in SOCE (Fig. 2C). To determine how exogenous Wnt5A expression would affect SOCE in noninvasive melanoma cells, we generated FS13 cells stably expressing Wnt5A. Western analysis confirmed that Wnt5A expression was observed only in transfected FS13 cells, as expected (Fig. 2A). Further, FS13 cells stably expressing Wnt5A exhibited a severalfold decrease in SOCE (Fig. 2D). The simplest interpretation of these observations is that Wnt5A activity negatively regulates SOCE; however, since Wnt5A influences the invasive phenotype of melanoma cells, it is also possible that increased SOCE in FS4 cells expressing Wnt5A shRNA reflects the fact that they are no longer invasive, while the corollary could be true in FS13 cells. Irrespective of which of these possibilities is correct, these observations provide clear support for the hypothesis that Wnt5A-dependent SOCE inhibition is a feature of invasive melanoma.

FIG 2.

Reciprocal control of SOCE in invasive versus noninvasive melanoma by Wnt5A. Assessment of FS4 and FS13 melanoma cells stably expressing Wnt5A shRNA or Wnt5A (overexpressed [o/e]), respectively. (A) Western analysis confirming predicted changes in Wnt5A expression. Hsp90 was used as a loading control. (B) Transwell assays in FS4 cells reveal significant Wnt5A-dependent loss of invasiveness. Images (top) show a representative field of invaded cells from the corresponding bar. (C and D) FS4 (C) and FS13 (D) cells were loaded with fura-2–acetoxymethylester (2 μM) prior to treating with thapsigargin (Tg; 2 μM), a SERCA inhibitor, to deplete ER Ca²⁺ stores (not shown). Once cytosolic Ca²⁺ had returned to baseline, 1 mM Ca²⁺ was added to the cells, invoking SOCE. Ca²⁺ measurements are shown as 340/380-nm ratios obtained from groups of 30 to 50 single cells. **, P < 0.001.

Differences in the expression and interaction of STIM and Orai do not account for differences associated with invasiveness.

A potential explanation for loss of SOCE in invasive melanoma would be dysregulated expression of a member of the STIM or Orai family. However, examination of STIM1, STIM2, and Orai1 expression levels by either Western analysis (Fig. 3A) or qPCR (Fig. 3B to D) revealed that while there was some variation in their expression among the different cell lines, there was no discernible pattern to these differences that could be associated with invasiveness. While we were unable to probe for Orai2 or Orai3 by Western analysis due to our inability to obtain reliable antibodies, qPCR similarly did not reveal any defects in channel expression in invasive melanoma cells (Fig. 3E and F).

FIG 3.

STIM/Orai expression levels do not account for differences in SOCE between invasive and noninvasive melanoma cells. (A) Cells growing under optimal growth conditions were lysed, normalized for protein content, and resolved on 8% SDS-PAGE gels. Western analysis of STIM1, STIM2, and Orai1 expression was performed, using GAPDH as a loading control. Western blots shown are representative of 4 experiments. (B to F) RNA was extracted from melanoma cell lines under optimal growth conditions, followed by analysis of STIM/Orai RNA expression levels using qPCR with Power SYBR green (Applied Biosystems) to detect amplification. Target levels were normalized to those of TATA-binding protein (TBP) and are depicted as means ± SEM (n = 3).

Further support for the concept that loss of SOCE in invasive melanoma was independent of expression level came from a surprising lack of SOCE recovery after STIM1/Orai1 overexpression (Fig. 4A and B). Indeed, even in noninvasive melanoma, SOCE elevation was modest relative to the dramatic elevation in SOCE that has been widely observed in response to STIM1/Orai1 overexpression in most cell types (23–26). To determine if this phenomenon reflected a failure of STIM/Orai association, we examined their localization and interaction both before and after store depletion. Prior to the addition of thapsigargin, STIM1 was localized diffusely throughout the ER in all 6 of the cell types examined (Fig. 4C and D). Upon treatment with Tg, STIM1 relocalized toward the periphery of the cell, where it colocalized with Orai1. To assess the possibility that subtle defects in protein-protein interaction were occurring, we measured Förster resonance electron transfer (FRET) between STIM1 and Orai1. Thapsigargin-induced ER Ca²⁺ depletion is passive and occurs relatively slowly. To avoid the potentially complicating effects of photobleaching, we used ionomycin to rapidly induce store depletion, leading to strong FRET within 2 min in all 6 cell types. That STIM1 relocalizes to the periphery and associates closely with Orai1 after treatment with either Tg or ionomycin excludes the possibility of a defect in ER Ca²⁺ depletion and strongly supports the hypothesis that STIM-Orai interaction occurs normally. As such, we investigated the possibility that the ability of STIM1 to activate Orai1 in invasive melanoma is regulated by alternative pathways, as outlined below.

FIG 4.

Impaired Orai function is independent of STIM/Orai interaction in invasive melanoma. Cells were cotransfected with YFP-STIM1 and CFP-Orai1 prior to plating on glass coverslips. (A and B) Transfected noninvasive (A) and invasive (B) melanoma cells were loaded with fura-2–acetoxymethylester (2 μM) prior to treating with thapsigargin (Tg; 2 μM), a SERCA inhibitor, to deplete ER Ca²⁺ stores (not shown). Once cytosolic Ca²⁺ had returned to baseline, 1 mM Ca²⁺ was added to the cells, invoking SOCE. Ca²⁺ measurements are depicted as means ± SEM of 340/380-nm ratios determined from at least 3 separate experiments, each consisting of 30 to 50 cells. (C and D) Images of noninvasive (C) and invasive (D) melanoma cells transfected with YFP-STIM1 (green) and CFP-Orai1 (red) before and after treatment with Tg (2 μM) to deplete ER Ca²⁺ stores. Ctl, control. (E and F) FRET measurements between STIM1 and Orai1 in noninvasive (E) and invasive (F) melanoma cells transfected with STIM1-YFP and CFP-Orai1. Rapid store depletion was stimulated via the addition of ionomycin. Corrected three-channel Förster resonance energy transfer (FRET_c) was determined postacquisition based on the following formula: FRET_c = (F_raw − F_a/A_a × F_YFP − F_d/D_d ×F_CFP)/F_CFP. Each trace depicts means ± SEM from at least 3 separate experiments.

PKC inhibition leads to SOCE recovery in invasive melanoma.

Invasive melanoma cells have been shown previously to exhibit elevated PKC activity (27), and PKC has been shown previously to inhibit SOCE (28, 29). Since the loss of SOCE in invasive melanoma was independent of the level of expression, we considered that differential phosphorylation could be a factor. Cells were incubated for 90 min with the potent and selective PKC inhibitor gö6983 (10 μM) prior to measuring SOCE. Remarkably, whereas this had no effect on SOCE in any of the noninvasive melanoma cell lines (Fig. 5A, B, and E), we observed significant SOCE recovery in 3 out of 4 invasive melanoma cell lines (Fig. 5C, D, and F). We then assessed how PKC inhibition affects invasiveness using transwell invasion assays. For these studies, we examined the effect of gö6983 on the invasiveness of both FS4 cells (Fig. 5G) (no effect of gö6983 on SOCE) and FS5 cells (Fig. 5H) (gö6983 enhanced SOCE), and it decreased the number of invading cells by 2- to 3-fold. Interestingly, although gö6983 pretreatment decreased the invasiveness of both FS4 and FS5 cells, analogous to its ability to enhance SOCE, gö6983 more effectively attenuated invasiveness in the FS5 model.

FIG 5.

PKC inhibition leads to SOCE recovery in invasive but not noninvasive melanoma cells. (A to F) Cells were plated on glass coverslips and loaded with fura-2–acetoxymethylester (2 μM) before treating with gö6983 (10 μM) for 90 min. All cells were treated with thapsigargin (Tg; 2 μM), a SERCA inhibitor, to deplete ER Ca²⁺ stores (not shown). Once cytosolic Ca²⁺ had returned to baseline, 1 mM Ca²⁺ was added to the cells, invoking SOCE. Representative traces of SOCE in FS14 (A) and G361 (B) cells depict no effect of gö6983 in noninvasive melanoma. Representative traces of SOCE in M93-047 (C) and UACC903 (D) cells reveal SOCE recovery in invasive melanoma due to incubation with gö6983. Analysis of the effect of gö6983 on SOCE magnitude in all noninvasive (E) and invasive (F) melanoma lines reveals significant enhancement of SOCE in FS5, M93-047, and UACC903 lines only as determined by two-way ANOVA. Transwell assays in FS4 (G) and FS5 (H) cells pretreated with gö6983 (16 h) reveal significant decreases in invasiveness. Images (bottom) show a representative field of invaded cells from the corresponding bar. *, P ≤ 0.01; **, P < 0.001; ***, P < 0.0001.

The targets of PKC-mediated inhibition of Orai1 function were identified previously as serines 27 and 30 on the N-terminal tail of Orai1 (29). Since the first 72 amino acids of Orai1 are not required for Orai1 activation (30–32), we cotransfected FS4 or G361 cells with STIM1 and either wild-type Orai1 (Orai1_WT), an Orai1 truncation mutant lacking the first 47 amino acids (Orai1_Δ47; generously donated by Christoph Romanin, Johannes Kepler University, Linz, Austria), or an Orai1 moiety in which serines 27 and 30 both were mutated to alanine (Orai1_S27AS30A). In addition, cells transfected with STIM1/Orai1_WT also were treated with vehicle or gö6983. To measure SOCE, ER Ca²⁺ content was depleted by treatment with thapsigargin (not shown), followed by measuring the amount of Ca²⁺ entry. Whereas Orai1 phosphorylation status had no impact on SOCE levels in G361 cells (Fig. 6A and C), marked phosphorylation-dependent differences in SOCE levels were observed in FS4 cells (Fig. 6B and C). Hence, overexpressing STIM1 and Orai1_WT failed to recover SOCE in FS4 cells unless they were treated with gö6983 (Fig. 6B and C). Further, this elevated SOCE was indistinguishable from what was observed in STIM1/Orai1_Δ47- or STIM1/Orai1_S27AS30a-expressing FS4 cells.

FIG 6.

Overexpression of STIM/Orai does not recover SOCE in invasive melanoma unless PKC phosphorylation is blocked. G361 (A and C) or FS4 (B and C) cells were transfected with YFP-STIM1 and Orai1_WT, Orai1_Δ47, or Orai1_S27AS30A as indicated and then plated on glass coverslips and loaded with fura-2–acetoxymethylester (2 μM). A subset of YFP-STIM1/Orai1_WT-expressing cells was treated with gö6983 (10 μM) for 90 min prior to Ca²⁺ measurements. Ca²⁺ measurements are shown as 340/380-nm ratios obtained from groups of 30 to 50 single cells. All cells were treated with thapsigargin (Tg; 2 μM), a SERCA inhibitor, to deplete ER Ca²⁺ stores (not shown). Once cytosolic Ca²⁺ had returned to baseline, 1 mM Ca²⁺ was added to the cells, invoking SOCE. (C) The magnitude of Ca²⁺ elevation (as depicted in panels A and B) was quantified and presented as bar charts. *, P < 0.05; **, P < 0.001.

Concluding remarks.

Metastatic melanoma is a poorly controlled disease, and there remains a great need for new insight into features that differentiate it from nonmetastatic forms. The well-established role of Ca²⁺ signals in migration has led us and others to examine STIM and Orai function in this disease model (11–15). It is both intriguing and confusing that we observe negative regulation of Orai1 in Wnt5A-dependent invasive melanoma, while each of these previous studies report an invasiveness-associated increase in STIM/Orai expression and function. Differences in the melanoma models chosen and experimental conditions are undoubtedly a factor. Indeed, it is interesting that serum deprivation was reported in two prior studies reporting elevated SOCE in invasive melanoma relative to noninvasive cells (13, 15). Further, examination of M93-047 and FS4 cells after serum deprivation revealed significant SOCE recovery under these conditions (see Fig. S5 in the supplemental material). While neither condition is a true reflection of the in vivo scenario, melanoma cells must be exposed to serum factors while being transported through the vasculature. Based on the current study, this would be predicted to lead to significant SOCE inhibition. Less certain is how differential regulation of Orai1 activity by Wnt5A and serum factors would impact its function at primary and secondary tumor sites. It has been reported previously that STIM1 mediates local elevation of Ca²⁺ near the front of a migrating cell (33); however, these Ca²⁺ hotspots occur in an otherwise low Ca²⁺ environment (33, 34). PKC-mediated inhibition of Orai1 function may serve to support the formation of this Ca²⁺ gradient, counterintuitively supporting Ca²⁺-dependent migration.

It has been well established that the first 72 amino acids of Orai1 are dispensable for channel activation (30). As such, it is surprising and intriguing that phosphorylation at a site located within these amino acids could have such a profound impact on channel function. Since no effect on STIM1-Orai1 interaction was observed, it is tempting to speculate that the phosphorylated residues actually impede channel conductance, perhaps by interacting with positively charged residues within the helical extensions of TM1 or Orai1. Future investigations may shed additional light on this possibility, as well as the potential for additional regulatory sites in the seemingly dispensable N-terminal 72 amino acids of this channel.

By definition, SOCE is regulated primarily by ER Ca²⁺ content; however, in the current study, we show that kinase activity regulates this pathway under a novel pathological condition. This dependence on alternative signaling is consistent with several recent findings by us and others showing that STIM1 can be activated by ROS (35) and temperature (36), while Orai activation is regulated by pH and hypoxia (37). Future investigations may reveal additional physiological and pathological scenarios where these alternative regulatory pathways uncouple Orai1 activation from ER Ca²⁺ depletion, as currently reported.