UV light activates a Gα_q/11-coupled phototransduction pathway in human melanocytes

Sunlight is crucial for life and has many beneficial effects, but, at the same time, the UV radiation (UVR) contained by sunlight is the most common environmental carcinogen (Routaboul et al., 1999; Bennett, 2008). Unlike other mammals that have fur to protect their skin, human skin is constantly exposed to solar UVR (280–400 nm) and is susceptible to its damaging effects, primarily skin cancers and photoaging. Human skin also has a unique protection mechanism against UVR: the presence of melanocytes in the epidermis allows skin to respond to UVR by increasing its pigmentation. Because UVR is omnipresent and is able to interact with human skin, identifying the molecular pathways that allow human skin to detect and elicit an immediate response to UVR is critical for developing new photoprotective methods.

How does human skin detect UVR? UVR consists of photons; photons can activate G protein–coupled opsin receptors (GPCRs) in the eye that elicit cellular responses through the activation of different G proteins and downstream effectors. Gα_i/o is used by vertebrate photoreceptors (Fung et al., 1981), whereas Gα_q/11 mediates Drosophila melanogaster phototransduction (Hardie, 2001) and non-image forming vision in the mammalian retina (Berson et al., 2002; Panda et al., 2005; Yau and Hardie, 2009). Activation of Gα_q/11 pathways leads to stimulation of phospholipase C β (PLCβ), which induces hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP₂) into diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP₃). Changes in the levels of PIP₂, DAG, and IP₃ modulate the activity of many proteins, including transient receptor potential (TRP) ion channels.

We recently characterized a retinal-dependent UVR-sensitive phototransduction pathway in human epidermal melanocytes (HEMs) that is G protein and PLCβ dependent and results in the activation of TRP subfamily A1 (TRPA1) channels; activation of this pathway results in a rapid increase in intracellular Ca²⁺ ([Ca²⁺]_ic) and increased cellular melanin content (Wicks et al., 2011; Bellono et al., 2013). In this study we investigated the G protein that mediates this pathway and the downstream molecular events. We found that UVR phototransduction in HEMs is mediated by Gα_q/11 signaling, and provide evidence for a phosphoinositide cascade involving IP₃-mediated intracellular Ca²⁺ release via IP₃ receptors (IP₃R) and PIP₂ regulation of Ca²⁺-permeable TRPA1 ion channels. The two sources of Ca²⁺ have different dynamics and, combined, result in a Ca²⁺ response with a fast rising phase and a slow decay. Our results demonstrate that UVR phototransduction in HEMs activates a Gα_q/11-dependent signaling pathway similar to well-characterized phototransduction pathways in the eye.

Cholera toxin (CTX), pertussis toxin (PTX), HC-030031, 1-oleoyl-2-acetyl-sn-glycerol (OAG), phosphatidylcholine phospholipase C (PC-PLC; from Clostridia perfringens), polylysine (PolyK, 70–150 kD), heparin, and ionomycin were purchased from Sigma-Aldrich. Endothelin, GPAnt-2, GPAnt-2a, and Xestospongin C (XeC) were from Tocris Bioscience. mSIRK and L9A were from EMD Millipore. DiC8-PIP2 was from Echelon Biosciences. Stocks of all reagents in water, DMSO, or ethanol were stored at −4°C or −20°C until use and diluted to the final concentration to contain <1% solvent. For Ca²⁺ imaging experiments, HEMs were preincubated with pharmacological reagents for 3–15 min, with the exception of PTX and CTX, which used 24 h incubations.

Primary HEMs isolated from neonatal foreskin were cultured in Medium 254 containing Human Melanocyte Growth Supplement (HMGS2; Cascade Biologics/Invitrogen) and 1% penicillin-streptomycin (Invitrogen), and propagated for a limited number of cell divisions (≤15). Vitamin A or retinoid derivatives are not components of either Medium 254 or HMGS2. Human embryonic kidney (HEK293) cells were cultured in Dulbecco’s Modified Eagle Medium and F12 nutrient mixture (DMEM-F12; Gibco/Invitrogen) containing 10% fetal bovine serum (Atlanta Biologicals) and 1% penicillin-streptomycin (Invitrogen).

MicroRNAs (miRNAs) were designed and expressed in HEMs using a lentiviral system, as described previously (Wicks et al., 2011). BLOCK-iT miRNA oligos (Invitrogen) were cloned into the pcDNA6.2-GW/EmGFP-miR expression vector modified to contain mCherry instead of EmGFP to allow for simultaneous fluorescence detection and Fluo-4–based Ca²⁺ imaging. miRNAs were recombined from pcDNA6.2-GW into pDONR221 and pLENTI6/V5-DEST vectors (Invitrogen) for lentiviral production. Lentiviral particles containing miRNA were obtained as described previously (Wicks et al., 2011). HA-tagged RGS2 (Missouri S&T cDNA Resource Center) was recombined into pDONR221 and pLENTI6/V5-DEST vectors (Invitrogen) for lentiviral production as described previously (Wicks et al., 2011).

The mRNA expression level of Gα_q, Gα₁₁, or Gα_q/11 in control or targeted miRNA-treated cells was determined ≥7 d after infection using comparative C_T quantitative PCR (qPCR). Total RNA was extracted from infected HEMs using the RNeasy Plus kit (QIAGEN) and converted to cDNA using RT-PCR (SuperScript III; Invitrogen). qPCR reactions were prepared according to manufacturer instructions using Power SYBR green. All reactions were done in triplicate and actin was used for normalization.

Expression of HA-tagged RGS2 in HEMs was confirmed via Western blotting. Cells were homogenized ≥10 d after infection in ice-cold RIPA buffer (Thermo Fisher Scientific) containing protease inhibitor cocktail (Roche). Samples were agitated at 4°C for 30 min and then centrifuged at 16,000 g for 30 min at 4°C. Protein content was determined using the Pierce BCA Protein Assay kit (Thermo Fisher Scientific). Equal amounts of protein were loaded onto each lane, separated by electrophoresis on NuPAGE Bis-Tris gels (Invitrogen), and transferred to PVDF membranes (Roche). Membranes were blocked at room temperature for 1 h and incubated overnight at 4°C with rat monoclonal anti-HA antibody clone 3F10 (1:500; Roche), followed by 1 h at room temperature with HRP-conjugated goat anti–rat IgG affinity-purified antibody (1:5,000; EMD Millipore). Antibodies were detected using the SuperSignal West Femto enhanced chemiluminescence system (Thermo Fisher Scientific) and imaged using autoradiography film (Thermo Fisher Scientific).

Ultraviolet light stimulation of cultured HEMs was conducted using a 200 W Hg-Xe arc lamp with converging optics and appropriate filters (Wicks et al., 2011). A dichroic mirror (260–400 nm) was used in combination with 280-nm long pass and a 400-nm short pass filters (Newport). The levels of light lost due to scattering by imaging buffer were negligible. Physiological doses of UVR were applied by varying the duration and/or power of the pulse. A hand-held silicon detector was used to measure power (Newport).

Ca²⁺ imaging was performed as described previously (Wicks et al., 2011; Bellono et al., 2013). Cultured HEMs plated on glass coverslips were incubated for 15 min at room temperature in Ringer’s solution with 2 µM Fluo-4 (Molecular Probes/Invitrogen) and 250 µM sulfinpyrazone (Sigma-Aldrich), followed by dark incubation for 15 min with 12 µM of 9-cis or all-trans retinal (Sigma-Aldrich). Imaging was performed in modified Ringer’s extracellular solution containing (in mM): 150 NaCl, 1.8 CaCl₂, 1.2 MgCl₂, 10 D glucose, 25 HEPES, pH 7.4, and 310 mOsm/liter. Fluorescence images were acquired every 2 s, and 2 µM ionomycin (Sigma-Aldrich) was added at the end of some experiments to elicit a maximal Ca²⁺ response used for normalization.

The fluorescence intensity of individual cells (measured using ≥25% of the cell area, F_cell) was quantified using MetaMorph (Molecular Devices) and MATLAB (MathWorks) and plotted in Prism 6 (GraphPad Software). F_cell values were normalized as F_norm = (F_cell − F_min)/(F_iono − F_min), where F_iono is maximal fluorescence with ionomycin, and F_min is baseline fluorescence averaged from ≥15 data points acquired before light stimulation. Final data values for each dish were obtained by averaging F_norm values from individual cells. In experiments where ionomycin was not used for normalization, Fluo-4 fluorescence intensities were quantified as ΔF/F_o(t) = [F_cell(t) − F_baseline]/F_baseline and averaged as described previously. ΔF/F_o values were plotted as a function of time and fitted with a single-exponential function in Prism 6 (GraphPad Software) to calculate decay time constants. Ca²⁺ response initial slopes were calculated over the first 30 s after UVR stimulation.

Paired Ca²⁺ imaging experiments were used when the dish-to-dish variability was significant. Cells for each of the paired experiments were plated on glass coverslips and treated identically by incubation in the same Fluo-4 and retinal solution, then imaged sequentially in alternating order: control followed by experimental condition or experimental condition followed by control. The averaged fluorescence intensities of cells from one coverslip measured in each condition were plotted as the two experimental values connected by the dotted line. Statistical significance of paired experiments was evaluated using a paired Student’s t test. When pairing is not mentioned, the cells were cultured and treated identically, but we did not strictly alternate control and experimental condition measurements. In such cases, we averaged the values from all the control and all the experimental conditions and represented them as bar graphs.

Electrophysiology experiments were performed as described previously (Bellono et al., 2013). All-trans retinal was stored, solubilized, and applied as described previously (Wicks et al., 2011). Experiments were performed under dim-red or infrared illumination. Whole-cell patch clamp recordings were carried out using micropipettes with 3–6 MΩ resistance at room temperature using an EPC 10 amplifier (HEKA) with PatchMaster software (HEKA), filtered at 2.9 kHz and digitized at 20 kHz. Experiments were performed using modified Ringer’s solution (see “Calcium imaging”). Unless stated otherwise, internal pipette solution contained (in mM): 140 CsCl, 1 MgCl₂, 4 MgATP, 10 EGTA, 10 HEPES, pH 7.2, and 290 mOsm/liter. The low EGTA solution used in Fig. S3 contained 20 µM EGTA. UVR-induced currents were measured using a step protocol consisting of a step from a holding potential of −60 mV to +80 mV immediately before UVR exposure. Current values were calculated by subtracting initial current at +80 mV (I_o) from maximal current after UVR exposure: I_UVR = I_max − I_o. All recordings were inspected for baseline drift before analysis. In most recordings, the baseline did not drift significantly. If the baseline drift was >20% of the UVR response, cells were excluded. Cell membrane capacitance values were used to calculate current densities. Current–voltage (I-V) relations were established using voltage step protocol from the −60-mV holding potential to voltages between −80 mV and +80 mV in 20-mV increments.

HEMs are very difficult to patch clamp. Our experiments are further complicated by the fact that UV exposure often causes us to lose patches. Because UV photocurrents are small and exhibit strong outward rectification (Bellono et al., 2013), most of our voltage-clamp experiments use positive membrane voltages that are not physiological (+80 mV) in order to increase current amplitude. Although current amplitude is higher at positive membrane voltages, current kinetics are significantly altered compared with recordings carried out at more physiological membrane voltages (Bellono and Oancea, 2013). In addition, positive voltages are not well tolerated by these cells, preventing us from recording and plotting the current over a long period of time for every single trace. We have previously recorded currents over longer periods using voltage pulses (Bellono et al., 2013) and at more negative voltages (Bellono and Oancea, 2013), and showed that the current returns to baseline after the UV pulse. Our success rate, even for the short periods of UV irradiation, is well below 10%. Consistent with the difficultly of these experiments, the large noise is likely caused by the relatively low seal resistance (∼1 GΩ) and patch stability.

Experimental data are presented as mean ± SEM, where n refers to the number of dishes for imaging data or the number of cells for electrophysiology. We calculated p-values by unpaired or paired Student’s t test and considered results significant when P ≤ 0.05.

Fig. S1 shows that RGS2 expression in HEMs reduces Ca²⁺ responses to endothelin. Fig. S2 shows that HEMs expressing Gα_q-, Gα₁₁-, or Gα_q/11-targeted miRNA have reduced Ca²⁺ responses to endothelin. Fig. S3 shows that neither IP₃-mediated Ca²⁺ release nor DAG activate the UVR photocurrent. Fig. S4 shows selective inhibition of UVR phototransduction signaling components versus ion channels by cellular dialysis.

We have recently shown that exposure to physiological doses of UVR activates a retinal-dependent phototransduction pathway that is mediated by G protein activation (Wicks et al., 2011; Bellono et al., 2013). However, we have yet to identify the type of G protein responsible for initiating the retinal-dependent signaling pathway activated by UVR. HEMs express a variety of GPCRs and their related G proteins, including the melanocortin-1 receptor coupled to Gα_s (Park et al., 2009) and the Gα_q/11-coupled endothelin-1B (ET-1) receptor (Yada et al., 1991; Imokawa et al., 1992).

Because we previously showed that UVR exposure leads to retinal-dependent Ca²⁺ responses (Wicks et al., 2011), we monitored intracellular Ca²⁺ levels using the fluorometric Ca²⁺ indicator Fluo-4 in response to treatments that modulate the function of different G protein subunits. We first tested if UVR-induced Ca²⁺ responses in HEMs were mediated by members of the Gα_s or Gα_i/o family. CTX treatment inhibits Gα_s-mediated signaling and PTX inhibits Gα_i/o signaling (Beckman et al., 1974; Malbon et al., 1984). HEMs incubated for 24 h with 500 ng/ml CTX, 500 ng/ml PTX, or vehicle, and stimulated with 150 mJ/cm² UVR showed no difference in the amplitude of retinal-dependent Ca²⁺ responses (Fig. 1, A and B; amplitude of the response: F_{norm, max} = 0.55 ± 0.04 for vehicle, 0.56 ± 0.05 for CTX, 0.52 ± 0.04 for PTX). This result suggests that Ca²⁺ responses elicited by physiological doses of UVR (150 mJ/cm²; Wicks et al., 2011) are not mediated by Gα_s or Gα_i/o signaling.

The Gβγ subunit of heterotrimeric G proteins can lead to increase in intracellular Ca²⁺ levels ([Ca²⁺]_ic) by directly activating PLCβ (Park et al., 1993; Wu et al., 1993; Barr et al., 2000). To test if UVR-mediated elevation in [Ca²⁺]_ic, requires activation of Gβγ, we incubated HEMs with the Gβγ-inhibiting peptide mSIRK (10 µM) or its inactive analogue L9A (10 µM; Goubaeva et al., 2003; Malik et al., 2005; Wang and Hatton, 2007) before stimulation with 150 mJ/cm² UVR. No significant difference was measured between UVR-induced Ca²⁺ responses in HEMs treated with mSIRK compared with the inactive analogue L9A (Fig. 1, C and D; F_norm,max = 0.43 ± 0.05 for mSIRK vs. 0.43 ± 0.04 for L9A), which suggests Gβγ does not mediate UVR-induced Ca²⁺ responses in HEMs.

We next investigated whether members of the Gα_q/11 family, known to promote Ca²⁺ mobilization through PLCβ activation (Smrcka et al., 1991; Taylor et al., 1991), mediate the UVR-induced Ca²⁺ response. To alter Gα_q/11 signaling, we expressed RGS2 in HEMs (Heximer et al., 1997; Heximer, 2004; Roy et al., 2006), a member of the regulators of G protein signaling (RGS) family, which inactivates Gα_q/11 by promoting the hydrolysis of GTP to GDP (De Vries et al., 2000). Lentiviral transduction of HA-RGS2 in HEMs resulted in the expression of a protein of the expected molecular size (Fig. S1 A) that reduced Ca²⁺ responses to activation of the Gα_q/11-coupled endothelin receptors endogenously present in HEMs (Yada et al., 1991; Imokawa et al., 1992; Fig. S1, B and C). Because HEMs expressing HA-RGS2 exhibited reduced Gα_q/11-mediated signaling via endothelin receptors, we tested if UVR-induced Ca²⁺ responses were also reduced. When compared with control transduced HEMs, cells expressing HA-RGS2 stimulated with 150 mJ/cm² UVR had significantly reduced retinal-dependent Ca²⁺ responses (Fig. 1, E and F; F_norm,max = 0.62 ± 0.03 for control vs. 0.37 ± 0.06 for RGS2), which suggests that members of the Gα_q/11 family mediate UVR signaling in HEMs. To determine which members of the Gα_q/11 family are expressed in HEMs and might be activated by UVR, we examined the mRNA expression levels of Gα_q/11 family members Gα_q, Gα₁₁, Gα₁₄, and Gα₁₅ by qPCR. We found that in HEMs Gα_q and Gα₁₁ are expressed at similarly high levels, whereas Gα₁₄ and Gα₁₅ are expressed at lower levels (Fig. 1 G).

We investigated the contribution of Gα_q versus Gα₁₁ to UVR signaling using RNA interference. Because Gα_q and Gα₁₁ share a high degree of homology, we designed miRNA targeting either Gα_q or Gα₁₁ individually, or both Gα_q and Gα₁₁ (Gα_q/11). Expression of Gα_q-targeted miRNA in HEMs resulted in a significant reduction in the mRNA transcript levels of Gα_q, but not of Gα₁₁, relative to control (scrambled) miRNA-expressing cells (Fig. 2 A; Gα_q mRNA relative to control = 0.14 ± 0.01, Gα₁₁ mRNA relative to control = 0.83 ± 0.16). Cells expressing Gα_q-targeted miRNA had a reduced Ca²⁺ response to 6 nM endothelin when compared with control miRNA-expressing cells, which suggests that Gα_q-targeted miRNA significantly reduced Gα_q signaling (Fig. S2, A and B). We then measured retinal-dependent Ca²⁺ responses elicited by 150 mJ/cm² UVR and found that HEMs expressing Gα_q-targeted miRNA had significantly reduced Ca²⁺ responses compared with HEMs expressing control miRNA (Fig. 2, B and C; F_norm,max = 0.57 ± 0.04 for control miRNA, 0.35 ± 0.03 for Gα_q miRNA).

Expression of Gα₁₁-targeted miRNA resulted in a significant reduction in the mRNA levels of Gα₁₁ (Gα₁₁ mRNA relative to control = 0.12 ± 0.03) and a smaller decrease in the mRNA transcript levels of Gα_q (Fig. 2 D; Gα_q mRNA relative to control = 0.75 ± 0.07). HEMs expressing Gα₁₁-targeted miRNA had reduced Ca²⁺ responses to 6 nM endothelin (Fig. S2, C and D), which suggests that Gα₁₁ signaling is also reduced. In response to stimulation with 150 mJ/cm² UVR, HEMs expressing Gα₁₁-targeted miRNA had a reduced retinal-dependent Ca²⁺ response compared with HEMs expressing control miRNA (F_norm,max = 0.61 ± 0.04 for control miRNA, 0.40 ± 0.05 for Gα₁₁ miRNA; Fig. 2, E and F). Collectively, our data from HEMs expressing Gα_q or Gα₁₁-targeted miRNA suggest that both Gα_q and Gα₁₁ contribute to the UVR-induced retinal-dependent phototransduction pathway in HEMs.

We next examined if decreasing Gα_q and Gα₁₁ levels simultaneously had a larger effect on UVR-induced Ca²⁺ responses in HEMs. Expression of Gα_q/11 miRNA that targets both Gα_q and Gα₁₁ significantly reduced the mRNA levels of Gα_q and Gα₁₁ relative to control miRNA-expressing cells (Fig. 2 G; Gα_q mRNA relative to control = 0.24 ± 0.08, Gα₁₁ mRNA relative to control = 0.32 ± 0.08). Stimulation with 6 nM endothelin resulted in reduced Ca²⁺ responses in HEMs expressing Gα_q/11-targeted compared with control miRNA (Fig. S2 E and Fig. 2 F; F_norm,max = 0.72 ± 0.04 for control miRNA, 0.44 ± 0.07 for Gα_q/11 miRNA). UVR stimulation of HEMs expressing Gα_q/11-targeted miRNA led to smaller increases in intracellular Ca²⁺ levels compared with control miRNA-treated cells (Fig. 2 H). Quantification of these responses showed that the UVR-induced Ca²⁺ responses were significantly reduced in the presence of Gα_q/11-targeted miRNA compared with control miRNA (Fig. 2, I and J), which suggests that both subunits Gα_q and Gα₁₁ contribute to retinal-dependent UVR-induced Ca²⁺ responses in HEMs.

UVR stimulation of HEMs leads to activation of a whole-cell current mediated by TRPA1 ion channels in a retinal and G protein–dependent manner (Bellono et al., 2013). We thus investigated if Gα_q/11 is also required for UVR-induced whole-cell currents. We first tested the effect of Gα_q/11-targeted miRNA on whole-cell currents measured at +80 mV in response to 240 mJ/cm² UVR and found that expression of control miRNA had no effect on the photocurrents, whereas Gα_q/11-targeted miRNA nearly abolished them (Fig. 3, A and B; I_{UVR/ctrl miRNA} = 4.06 ± 0.29 pA/pF, I_{UVR/ Gαq/11 miRNA} = 0.28 ± 0.17 pA/pF) at all voltages (Fig. 3 B, inset). The whole-cell patch clamp technique used to measure the photocurrents allowed us to use an alternative method to block Gα_q/11 signaling by dialyzing GPAnt-2a, a Gα_q/11 inhibitory peptide, into HEMs (Mukai et al., 1992). The time-dependent effect of peptide inhibitor dialysis allowed us to measure the UVR-induced current at 2 min after break-in, when the peptide was not effective, and thus the measurement was used as control, and at 10 min of dialysis, when the peptide became effective. This experimental protocol allowed us to compare the effects of the peptide inhibitors in the same cell. When GPAnt-2a was included in the pipette solution, the UVR photocurrents were similar to control cells 2 min after break-in, but decreased significantly after 10 min of dialysis (Fig. 3, E and F; I_UVR/GPAnt-2a = 3.98 ± 0.50 pA/pF at 2 min, 0.69 ± 0.08 pA/pF after 10 min of dialysis with GPAnt-2a). As a control we performed a similar experiment using the GPAnt-2 peptide, which inhibits Gα_i/o signaling (Mukai et al., 1992), and detected no change in the amplitude of the UVR current after 2 or 10 min of dialysis (Fig. 3, C and D; I_UVR/GPAnt-2 = 3.68 ± 0.49 pA/pF at 2 min, 3.64 ± 0.26 pA/pF after 10 min). These results suggest that UVR-induced photocurrents are dependent on Gα_q/11 signaling.

Our results so far indicate that UVR phototransduction leads to activation of Gα_q/11, which in turn activates PLCβ, required both for intracellular Ca²⁺ release (Wicks et al., 2011) and TRPA1 activation (Bellono et al., 2013). Nonetheless, the mechanism by which UVR leads to TRPA1 activation downstream of PLCβ remains unknown. Because PLCβ hydrolyzes plasma membrane PIP₂, generating DAG and the soluble messenger IP₃, we reasoned that PLCβ-dependent signaling could modulate TRPA1 channel activity in HEMs via IP₃, DAG, or PIP₂.

We first tested whether IP₃ or IP₃-mediated Ca²⁺ release was sufficient to activate TRPA1 by using a control internal solution, allowing for an increase in [Ca²⁺]_ic (see Materials and methods), internal solutions containing 100 µM IP₃ to stimulate IP₃Rs, or 1 mg/ml heparin to block IP₃Rs, with both treatments occluding subsequent UVR-induced IP₃R-mediated Ca²⁺ release. UVR (240 mJ/cm²) exposure after dialysis for 5 min with each of the three solutions elicited retinal-dependent photocurrents with similar amplitudes (Fig. S3, A and B; I_UVR/Ctrl = 4.01 ± 0.33 pA/pF, I_UVR/IP3 = 4.04 ± 0.47 pA/pF, I_UVR/heparin = 5.05 ± 0.56 pA/pF), which suggests that UVR-induced activation of whole-cell currents is not mediated by IP₃ or IP₃R-mediated Ca²⁺ release.

We next examined the contribution of DAG to retinal-dependent UVR photocurrents. Bath application of the PC-PLC (10 U/ml), which generates DAG in the plasma membrane (Oancea et al., 1998), or of the DAG analogue OAG (100 µM), did not elicit a significant current in HEMs, and 5 min of incubation with the respective treatments did not affect the retinal-dependent photocurrents elicited by UVR (Fig. S3, C and D; I_PC-PLC = 0.10 ± 0.10 pA/pF, I_OAG= 0.20 ± 0.09 pA/pF, I_UVR/PC-PLC = 4.31 ± 0.70 pA/pF, I_UVR/OAG = 4.13 ± 0.48 pA/pF). Because increasing DAG levels failed to elicit whole-cell currents and had no effect on UVR photocurrents, we concluded that DAG does not modulate TRPA1 downstream of UVR.

PIP₂ hydrolysis is a key regulator for many ion channels (Suh and Hille, 2005), including TRPA1 (Dai et al., 2007; Karashima et al., 2008; Kim et al., 2008). To test if the presumed decrease in PIP₂ levels caused by PLCβ-mediated hydrolysis affects UVR photocurrents, we attempted to maintain elevated PIP₂ levels by dialyzing HEMs with the PIP₂ analogue diC8-PIP₂ (20 µM; Karashima et al., 2008; Kim et al., 2008). We found that after 5 min of patch pipette dialysis to allow diC8-PIP₂ to diffuse into cells, UVR stimulation (240 mJ/cm²) elicited significantly smaller photocurrents at all voltages compared with UVR photocurrents measured immediately after break-in (Fig. 4, A, B, and D). This finding suggests that increased PIP₂ prevents UVR-induced TRPA1 activation and raises the question of whether PIP₂ hydrolysis is required for the UVR photocurrent.

To test if the dialysis with diC8-PIP₂ inhibited TRPA1, or another component of the UVR phototransduction cascade, we compared the whole-cell currents elicited by UVR and the TRPA1 agonist cinnamaldehyde (CA) in HEMs dialyzed with diC8-PIP₂ or GPAnt-2a (Fig. S4). We found that dialysis with both diC8-PIP₂ and GPant-2a inhibited UVR photocurrents, whereas only diC8-PIP₂ inhibited currents elicited by CA (Fig. S4). These data suggest that GPAnt-2a inhibits an important component of the UVR signaling cascade, Gα_q/11, while diC8-PIP₂ directly inhibits TRPA1 activity, which is required for both the UVR- and CA²⁺-elicited increase in whole-cell current.

To test if a decrease in PIP₂ levels is also sufficient to cause TRPA1 activation in HEMs, we dialyzed cells with polylysine (polyK; 50 mg/ml), which binds and sequesters PIP₂ (Lukacs et al., 2007; Klein et al., 2008; Ufret-Vincenty et al., 2011), preventing it from acting on the TRPA1 channels. A subsaturating UVR dose (160 mJ/cm²; Bellono et al., 2013) evoked a submaximal UVR photocurrent in HEMs dialyzed with control internal solution. PolyK alone, when included in the pipette solution and allowed to diffuse into the cell and sequester PIP₂, did not lead to an increase in whole-cell current; however, it significantly enhanced UVR photocurrents elicited by the same subsaturating UVR dose (Fig. 4, C and E; I_PolyK = 0.07 ± 0.03 pA/pF, I_UVR/Ctrl = 2.51 ± 0.07 pA/pF, I_UVR/PolyK = 4.90 ± 0.44 pA/pF). The polyK-modulated UVR photocurrent was inhibited by the TRPA1 antagonist HC-030031 (HC; 100 µM; Fig. 4 C, right; and Fig. 4 E; I_UVR/PolyK+HC = 0.27 ± 0.12 pA/pF), which suggests that the enhanced photocurrent was mediated by TRPA1.

To address the specificity of the phospholipids that modulate the UVR photocurrents, we used recombinant pleckstrin homology (PH) domains from phospholipase C δ1 (PLCδ1-PH) that selectively bind PIP₂ and from the general receptor for phosphoinositides type 1 (GRP1-PH) that selectively binds PI(3,4,5)P₃ (PIP₃; Klein et al., 2008). Dialysis of recombinant PH domains in HEMs will bind and sequester the phosphoinotides, resulting in decreased cellular levels. When PLCδ1-PH was included in the pipette, a significantly enhanced photocurrent was elicited by a subsaturating UVR dose (160 mJ/cm²), compared with boiled PLCδ1-PH or GRP1-PH (Fig. 4, F and G; I_PLCδ1-PH = 3.01 ± 0.36 pA/pF, I_{Boiled PLCδ1-PH} = 1.32 ± 0.26 pA/pF, I_GRP1-PH = 1.34 ± 0.24 pA/pF). These results suggest that sequestering PIP₂, but not PIP₃, levels in HEMs leads to increased TRPA1 activity in response to UVR, leading us to hypothesize that UVR-mediated PIP₂ hydrolysis releases the PIP₂-mediated inhibition of TRPA1.

Retinal-dependent UVR-induced Ca²⁺ responses triggered downstream of Gα_q/11 signaling are mediated by two sources of Ca²⁺: (1) efflux from intracellular thapsigargin-sensitive stores and (2) influx via TRPA1 at the plasma membrane (Bellono et al., 2013). Because Gα_q/11/PLCβ signaling generates the second messenger IP₃, we hypothesized that IP₃Rs mediate Ca²⁺ release from intracellular stores. To test this hypothesis, we treated HEMs with the IP₃R antagonist XeC (25 µM; Gafni et al., 1997; Oka et al., 2002) and found that UVR-induced Ca²⁺ responses were significantly reduced (Fig. 5, A and B). To study the contribution of each Ca²⁺ source (IP₃R vs. TRPA1) to the overall response, we measured Ca²⁺ responses in HEMs preincubated with XeC in order to inhibit IP₃R, or HC-030031 (HC; 100 µM) in order to inhibit TRPA1 (Fig. 4 A). Ca²⁺ responses elicited by 240 mJ/cm² UVR were reduced by 71% in the presence of XeC, by ∼45% in the presence of HC, and by ∼90% in the presence of both antagonists, as compared with vehicle-treated cells (Fig. 4 B; fluorescence increase over baseline: ΔF/F_o = 3.21 ± 0.27 for vehicle, 0.94 ± 0.16 for XeC, 1.77 ± 0.29 for HC, and 0.34 ± 0.07 for XeC + HC). These results suggest that UVR phototransduction evokes a rise in [Ca²⁺]_ic via intracellular Ca²⁺ release from IP₃R and Ca²⁺ influx through TRPA1 ion channels.

We next sought to distinguish the contribution of each Ca²⁺ source (IP₃R and TRPA1) to the biphasic nature of the transient UVR-induced Ca²⁺ response. To do that, we measured the initial slope and the decay time constant of Ca²⁺ responses in HEMs treated with XeC or HC, compared with vehicle. The mean initial slope of UVR-induced Ca²⁺ responses (measured during the first 30 s after the beginning of UVR stimulation) was reduced by ∼76% with XeC treatment and by ∼35% with HC treatment when compared with vehicle-treated HEMs (Fig. 5, C and D; ΔF/s = 0.100 ± 0.009 for vehicle, 0.024 ± 0.003 for XeC, and 0.064 ± 0.01 for HC). These results suggest that both sources contribute to the initial rising phase of the Ca²⁺ response, but IP₃-mediated Ca²⁺ release has a significantly greater contribution.

The time constant for the decay phase (τ_off) of the UVR-induced Ca²⁺ response was measured by monitoring cellular Ca²⁺ levels for ≥200 s after the peak of the response in cells treated with XeC, HC, or vehicle. HEMs treated with XeC had Ca²⁺ responses with mean τ_off values ∼79% higher than vehicle-treated cells, which suggests that the IP₃-mediated response decays on a fast time scale. Interestingly, treatment with HC resulted in UVR-evoked Ca²⁺ responses that decayed ∼38% faster than vehicle-treated cells (Fig. 5, E and F; τ_off = 225.90 ± 37.44 s for XeC, 78.42 ± 12.68 s for HC, and 126.20 ± 10.61 s for vehicle). Hence, TRPA1 activation extends the duration of the Ca²⁺ responses, whereas IP₃R activation reduces the duration. Collectively, our results suggest that IP₃R-mediated Ca²⁺ release is important to ensure a fast rising phase of the response. In contrast, consistent with our previous findings (Bellono and Oancea, 2013; Bellono et al., 2013), TRPA1 activation is slow but persistent, contributing to a prolonged Ca²⁺ response (Fig. 6 B).

We have recently discovered that primary human melanocytes are capable of rapidly detecting UVR by first increasing intracellular Ca²⁺ and later producing more melanin (Wicks et al., 2011). The Ca²⁺ response is retinal dependent and is in part due to calcium release from intracellular stores and in part to calcium influx through TRPA1 ion channels (Bellono et al., 2013). Both components of the response require heterotrimeric G proteins and PLCβ activation (Wicks et al., 2011; Bellono et al., 2013). Here we investigated the identity of the G protein subunit that mediates the response and found that reducing the expression of both Gα_q and Gα₁₁, as well as using a peptide that inhibits Gα_q/11 signaling, significantly decreased both components of the Ca²⁺ response (Figs. 2 and 3). These results suggest that the UVR-activated pathway is mediated by Gα_q/11, which, in turn, activates PLCβ.

Signal transduction pathways associated with G protein activation often result in modulation of TRP ion channels. For example, in Drosophila, light stimulation of rhodopsin results in activation of Gα_q/11 and PLCβ, which hydrolyzes PIP₂ to produce DAG and IP₃, to regulate TRP ion channels (Hardie, 2001). We reasoned that one of these messengers (IP₃, DAG, or PIP₂) might regulate the UVR-activated TRPA1 photocurrent and tested our hypothesis by exogenously altering the levels of these messengers (Figs. 4 and S3). We found that reagents used to manipulate the levels of IP₃ and DAG had no effect on the photocurrent, whereas reagents that affect PIP₂ levels did. Allowing diC8-PIP₂ to diffuse into the cell blocked the current in response to a UVR dose that evokes a maximal response, which suggests that decreasing PIP₂ levels is a necessary step in TRPA1 activation. But is the decrease in PIP₂ sufficient to activate TRPA1?

Sequestration of PIP₂ using polyK did not elicit a significant current in the absence of UVR, but did potentiate TRPA1 currents evoked by UVR. Furthermore, using PLCδ1-PH to specifically sequester PIP₂ enhanced UVR photocurrents, whereas sequestration of PIP₃ by GRP1-PH had no effect. These results suggest that a decrease in PIP₂ can modulate the photocurrent, but may not be sufficient for TRPA1 activation. Therefore, other messengers or proteins that contribute to the phototransduction pathway could be involved. Our results add to the already controversial role of PIP₂ in modulating TRPA1. PIP₂ was found to potentiate agonist-activated TRPA1 currents (Karashima et al., 2008), but also to inhibit TRPA1 activity (Dai et al., 2007; Kim et al., 2008). GPCRs that activate Gα_q/11 (bradykinin, PAR2, and Mrgprs) can regulate TRPA1, but the mechanism is unclear (Bandell et al., 2004; Dai et al., 2007; Wang et al., 2008; Wilson et al., 2011). It remains to be determined in future experiments whether PIP₂ directly interacts with TRPA1 to modulate its activity and what other cellular messengers contribute to TRPA1 activation in response to UVR and to Gα_q/11-coupled receptors.

Examining the relative contribution of UVR-induced IP₃R- and TRPA1-mediated Ca²⁺ responses revealed that IP₃-mediated Ca²⁺ release is rapid, but also declines fast. In contrast, TRPA1-mediated influx is slower to increase intracellular Ca²⁺, as suggested by the slow time course of the photocurrent activation, and slow to decay, allowing intracellular Ca²⁺ to remain elevated after UVR exposure, a necessary step for melanin production (Bellono et al., 2013). However, there is a discrepancy between the time course of photocurrents and that of Ca²⁺ responses. UVR photocurrents recorded under voltage-clamp conditions peak during or shortly after light stimulation, whereas Ca²⁺ responses measured in intact cells, in which the membrane voltage could change, peak seconds after irradiation. We recently found that UVR phototransduction depolarizes the plasma membrane of melanocytes to delay TRPA1 inactivation and prolong Ca²⁺ responses (Bellono and Oancea, 2013), a finding consistent with the TRPA1-mediated influx kinetics found in this study. The discrepancy between the time course of the response measured by the two methods is likely to be caused by the significant effects of membrane depolarization on the TRPA1 channel and consequent Ca²⁺ signaling dynamics. Our analyses of Ca²⁺ response dynamics also revealed the possibility of cross-talk between the Ca²⁺ release and influx pathways. Inhibition of IP₃R resulted in a significantly slower decline of the TRPA1-mediated Ca²⁺ response, which suggests that the IP₃-mediated response decays considerably faster than the overall response. However, when TRPA1 was inhibited, the decay of the IP₃-mediated Ca²⁺ responses was only slightly (although significantly) faster, which suggests that Ca²⁺ release may accelerate TRPA1 inactivation.

Based on our data, we propose that UVR exposure of HEMs stimulates a retinal-dependent Gα_q/11-coupled receptor, which activates PLCβ. Active PLCβ hydrolyzes plasma membrane PIP₂ into DAG and IP₃. Soluble IP₃ binds IP₃Rs in the ER, resulting in Ca²⁺ release, whereas the decrease in PIP₂ levels modulates TRPA1 activation and Ca²⁺ influx (Fig. 6). Our model for UVR signal transduction in human melanocytes resembles visual phototransduction in Drosophila photoreceptors (Hardie, 2001) and nonvisual phototransduction in the mammalian retina (Berson et al., 2002; Graham et al., 2008). The melanocyte UVR pathway also shares many similarities with a recently described UVR-activated signaling mechanism in Drosophila larvae, which is mediated by Ca²⁺ signals resulting from Gα_q/11 and TRPA1 activation (Xiang et al., 2010). The receptor for the Drosophila larvae UVR pathway appears to be a gustatory GPCR (Gr28b); how light can activate such a receptor remains unknown.

One of the remaining questions for the UVR phototransduction cascade in melanocytes is the identity of the receptor. The retinal dependence and G protein involvement suggests the involvement of an opsin GPCR. We previously found that rhodopsin expression contributes to UVR-induced Ca²⁺ responses (Wicks et al., 2011). However, the differences in spectral sensitivity and G protein coupling of rhodopsin and of the UVR pathway suggest that rhodopsin might work in conjunction with a different, possibly unidentified, UVR-sensitive receptor to mediate UVR phototransduction in melanocytes. Future experiments will identify the UVR receptor and determine the molecular mechanism of TRPA1 activation. Understanding this pathway might uncover new photoprotection strategies for human skin, thus lowering the incidence of skin cancer.

We thank Dr. Sharona Gordon for generously providing the recombinant PH domains and Dr. Anita Zimmerman for helpful discussion.

This work was funded by grants from Brown University, NIH RI-INBRE 8P20GM103430-12, subaward No. 120912/0003223 (to E. Oancea), a National Science Foundation Graduate Research Fellowship (to N.W. Bellono), and National Institutes of Health Predoctoral Fellowship (NIAMS-NIH [F31] No. AR056587 to J.A. Najera).

The authors declare no competing financial interests.

Author contributions: J.A. Najera, N.W. Bellono, and E. Oancea designed the research in the paper; J.A. Najera and N.W. Bellono performed the research; E. Oancea contributed new reagents/analytic tools; J.A. Najera and N.W. Bellono analyzed data; and J.A. Najera, N.W. Bellono, and E. Oancea wrote the paper.

Edward N. Pugh Jr. served as editor.