Melanopsin, a Canonical Light Receptor, Mediates Thermal Activation of Clock Genes

Melanopsin (OPN4) is a photo-pigment found in a small subset of intrinsically photosensitive ganglion cells (ipRGCs) of the mammalian retina. These cells play a role in synchronizing the central circadian pacemaker to the astronomical day by conveying information about ambient light to the hypothalamic suprachiasmatic nucleus, the site of the master clock. We evaluated the effect of a heat stimulus (39.5 °C) on clock gene (Per1 and Bmal1) expression in cultured murine Melan-a melanocytes synchronized by medium changes, and in B16-F10 melanoma cells, in the presence of the selective OPN4 antagonist AA92593, or after OPN4 knockdown by small interfering RNA (siRNA). In addition, we evaluated the effects of heat shock on the localization of melanopsin by immunocytochemistry. In both cell lines melanopsin was found in a region capping the nucleus and heat shock did not affect its location. The heat-induced increase of Per1 expression was inhibited when melanopsin was pharmacologically blocked by AA92593 as well as when its protein expression was suppressed by siRNA in both Melan-a and B16-F10 cells. These data strongly suggest that melanopsin is required for thermo-reception, acting as a thermo-opsin that ultimately feeds the local circadian clock in mouse melanocytes and melanoma cells.

the ability of thermo-discrimination [10][11][12] , which can be rescued by the targeted expression of mouse melanopsin 10 . In addition, melanopsin has been recently reported in murine blood vessels 13 , in which it mediates blue-light dependent photo-relaxation. Since the vascular physiology is under circadian control 14 , one may suggest that melanopsin could act as sensor that ultimately feeds the local temporal controlling system. Following this line, our group has shown that melanopsin and rhodopsin are expressed in murine melanocytes and melanoma cells where they may participate in a photo-sensitive system 15 . Based on these findings, we questioned whether an opsin could also function as a thermo-sensor in mammalian cells, conveying temperature information to the clock gene machinery of cutaneous melanocytes, cells known to be exposed to cycles of environmental light and temperature 16 .

Results and Discussion
We evaluated the effect of a heat stimulus (39.5 °C) on clock gene expression in cultured Melan-a melanocytes and B16-F10 melanoma cells. Cells were maintained for three days under constant darkness and temperature (37 °C), a situation in which each cell displays its own rhythm of clock gene expression, usually leading to undetectable rhythm of the cell culture 17 . Melan-a cells exposed to 1 h heat pulse (39.5 °C) showed no difference from control cells in Per1, and Bmal1 expression 0, 1 and 2 h after the stimulus (Supplemental Fig. 1). Because no effect was found in non-synchronized melanocytes, our next step was to repeat the same assay in cells synchronized by two medium changes 18,19 . In fact, 24 h after cell synchronization, heat shock led to increased Per1 (0 and 1 h after the stimulus, Fig. 1A) but not Bmal1 (Fig. 1B) expression. On the other hand, heat shock induced Per1 increase in B16-F10 cells in constant dark condition 1 h after the end of the stimulus (Fig. 1C). Again, Bmal1 was irresponsive to heat shock (Fig. 1D). Melan-a or B16-F10 cells were kept for 3 days in constant dark and temperature (37 °C). In the beginning of the 4 th day, Melan-a cells were synchronized by two medium changes and after further 24 hours they were heat-stimulated (39.5 °C) during 1 h. B16-F10 cells were heat stimulated in the beginning of the 4 th day. Total RNA was extracted immediately and 1 h after the end of the stimulus for Melan-a and B16-F10 cells. Boxplots show the median, quartiles, maximum, and minimum expression values of each gene transcript normalized by 18S ribosomal RNA (for Melan-a cells) and Rpl37a (for B16-F10 cells), and expressed relative to the minimal value at 37 °C (N = 4-6). Statistical analysis was performed by Two-way ANOVA followed by Bonferroni post-test.
In agreement with these findings a 15-min white light pulse (WLP) applied to desynchronized Melan-a cells did not alter clock gene expression in comparison to cells kept in constant dark (DD) condition. Interestingly, in B16-F10 cells the expression of Per1, Per2, and Bmal1 was upregulated in response to WLP 15 . Although the effects of heat shock or light pulse were not observed in desynchronized Melan-a cells, we cannot rule out that this stimulus may have affected the clock machinery of single cells, which would remain uncoupled to each other, and therefore, the overall oscillation would be unnoticed in the cell population. Another point of view is related to morpho-physiological differences between normal and malignant melanocytes, the latter showing denser dendritic projections among cells, a feature that probably allows more efficient cellular coupling 15 .
Temperature cycles have been shown to alter rhythmic parameters of clock genes in peripheral tissues 20 , and are a strong zeitgeber in normal murine keratinocytes 21 and fibroblasts 22 . To our knowledge, this is the first report that a short heat pulse affects clock gene machinery in murine melanocytes and melanoma cells. In fact, in non-mammalian vertebrates, we have already shown that heat shock increases the expression of clock genes in the photosensitive teleost ZEM-2S cell line, but only when cells were synchronized by light/dark (LD) cycles 23 .
Previous studies using Melan-a and B16-F10 cells 15 and human and mouse spermatozoa 24 showed immunolabeling of melanopsin in regions capping the nucleus. We have shown that a 15-min WLP promoted OPN4 translocation from nucleus region to the cytoplasm and cell membrane in B16-F10 cells 24 h after the stimulus 15 . Based on these findings, we investigated whether heat shock was also capable of inducing melanopsin translocation in Melan-a and B16-F10 cells. Our results demonstrate that the cytoplasm and the nucleus capping location of OPN4 in both cell lines was not altered 24 h after the heat shock of 39.5 °C ( Fig. 2A-D), suggesting that OPN4 is not required to be inserted into the membrane to detect heat stimulus. In another line of thought, one may consider that the basal level of membrane-bound OPN4 may be enough to detect heat and trigger heat-induced responses in murine melanocytes and melanoma cells.
We then pharmacologically inhibited melanopsin with the antagonist AA92593, shown to be specific because it competes with retinaldehyde for the melanopsin retinal binding site which is very distinct from other opsins. Its administration to mice in vivo specifically and reversibly modified melanopsin-dependent light responses including the pupillary light reflex and light aversion 25 . Our data show that the increase of Per1 level induced by heat in Melan-a and B16-F10 cells was significantly reduced in the presence of the melanopsin antagonist ( Fig. 3A,C) whereas Bmal1 expression was not affected (Fig. 3B,D). Surprisingly, in Melan-a cells the group incubated with the antagonist and kept at 37 °C showed a statistically significant increase of Per1 transcript when compared (2) group in constant dark and exposed to 1 h heat stimulus (39.5 °C). Twenty-four hours later the medium was removed and the cells were fixed with 4% paraformaldehyde. DAPI stained nuclei in blue and OPN4 immunopositivity (1:500 antiserum), revealed with a Cy3-labeled secondary antibody, in orange. Photomicrographies were taken with Axiocam MRm camera (Zeiss) and pseudocolored with Axiovision software (Zeiss). Scale bar 50 μm (200x magnification).
to DMSO-treated control group (p = 0.0478, Fig. 3A); a phenomenon that showed no statistical significance (p = 0.08) in B16-F10 cells. The apparent intrinsic effect of AA92593 on control cells kept at 37 °C could be due to the following reasons: I) melanopsin acts as a thermosensor, and when inhibited, the cell would lose its ability to sense temperature, and the response would resemble the heat-evoked behavior; II) another possibility lies on a partial agonistic activity of AA92593 which, although not reported by Jones and co-workers 25 , could explain the response found in the control group treated with AA92593; III) since AA92593 is a competitive melanopsin antagonist, its presence in the retinal-binding pocket of melanopsin leads to the displacement of retinal, which could trigger a downstream signaling that would ultimately result in Per1 increased expression.
Next, we performed gene knockdown assays using the endoribonuclease-prepared siRNAs (esiRNA), which consist of a heterogeneous mixture of siRNAs against murine melanopsin mRNA. These siRNAs selectively suppress gene expression with low off-target effects 26 . In fact, we show a significant reduction of Opn4 transcription as well as melanopsin protein level 48 h after transfection in Melan-a cells (Fig. 4A,B,E). In B16-F10 cells, although the mRNA levels of melanopsin did not decrease 48 h after transfection (Fig. 4F), a fact that could be due to a faster mRNA turnover, OPN4 protein level was drastically reduced (Fig. 4C,D). Then, our next step was to heat shock Melan-a and B16-F10 cells with reduced melanopsin expression, and the results demonstrated that the heat-induced increase of Per1 was significantly reduced in both cell lines (Fig. 5A,C) while no effect on Bmal1 expression was found (Fig. 5B,D). We have demonstrated that an opsin in a skin cell type can be activated not only by photons 15 but also by thermal energy. In fact, thermo-isomerization of rhodopsin and cone opsins has been shown to occur and it requires about half the energy necessary to photo-activate the photo-pigment 27 . Therefore, taken together our results clearly demonstrate-for the first time -the involvement of melanopsin in thermo-responses of mammalian cells.
According to Colin Pittendrigh Escape from Light theory 28 , higher temperatures are found during the photo-phase of the day and, therefore, temperature and light are environmental entities exerting simultaneous selective pressures on the organisms. It is not surprising, therefore, that light and temperature may be perceived by the same conserved proteins, having in mind that photo-and thermo-sensitive systems probably co-evolved during evolution. Considering that the skin is constantly exposed to both physical stimuli 16 , it is relevant to better understand how the skin perceives heat and light. In fact, our data add another layer of complexity for this system: an opsin, which is classically a light sensor, also acts as a thermo-sensor that ultimately feeds the local circadian clock. Interestingly, the opsin-mediated clock gene activation is conserved in malignant melanocytes, which warrants further investigation whether this event was pro-or anti-tumorigenic. Within this line, it has been recently shown that clock gene machinery of melanoma cells is suppressed 15,29 , but clock gene activation by dexamethasone, forskolin, or heat shock results in reduced melanoma proliferation in vitro and in vivo without leading to cell death 29 . Taken altogether, our data bring convincing evidence of a new role for a canonical mammalian light sensor, which challenges the current paradigm that mammalian opsins exclusively function as photo-pigments.  Experimental Design. In all experiments, Melan-a and B16-F10 cells were maintained in the medium described above but FBS was reduced to 2% (experimental medium). Cell manipulation during the experiments was carried out under red dim light (7 W Konex bulb and Safe-Light filter GBX-2, Kodak, Rochester, NY, USA). For all protocols described below, Melan-a and B16-F10 cells were kept in constant dark and temperature (37 °C) during 3 days. At the beginning of the 4 th day the medium of Melan-a cells was changed twice with a 2-hour interval, after which the cells were kept in DD during further 24 h. This procedure has been shown to synchronize clock genes in this cell line 31 .

Effect of Heat Stimulus on Clock Gene Machinery of Melan-a and B16-F10 Cells. Melan-a or
B16-F10 cells were seeded at the density of 10 6 and 10 5 cells respectively in 25 cm 2 flasks. In the beginning of the 4 th or 5 th day (24 h after medium changes), B16-F10 and Melan-a cells were, respectively, heat-stimulated (39.5 °C) during 1 h while the control group remained at 37 °C. Total RNA was initially extracted immediately, 1 and 2 h after the end of the heat stimulus, and the time points showing maximal response of clock gene expression were adopted for subsequent assays. Melan-a or B16-F10 cells were kept during three days in constant dark and temperature (37 °C). At the beginning of the 4 th day, Melan-a cells were synchronized by two medium changes, and after further 24 h cells were transfected with esiRNA against melanopsin or EGFP (both at 10 nM) using Lipofectamine 3000 transfection kit. B16-F10 cells were transfected with esiRNA, as described above, at the beginning of the 4 th day. In both experimental scenarios, 48 hours after transfection, cells were divided into four groups: (1) control group at 37 °C in the presence of esiRNA EGFP (10 nM); (2) heat-stimulated (39.5 °C) group in the presence of esiRNA EGFP (10 nM); (3) group at 37 °C in the presence of esiRNA melanopsin (10 nM); (4) heat-stimulated (39.5 °C) group in the presence of esiRNA melanopsin (10 nM). Boxplots show the median, quartiles, maximum, and minimum expression values of each gene transcript normalized by Rpl 37a and expressed relative to the minimal value of the esiRNA EGFP group at 37 °C (N = 5-11). Total RNA was extracted immediately and 1 h after the end of the stimulus for Melan-a and B16-F10 cells, respectively. Statistical analysis was performed by One-way ANOVA followed by Tukey's test.
Scientific RepoRts | 7: 13977 | DOI:10.1038/s41598-017-13939-3 Effect of Heat Stimulus on OPN4 Localization in Melan-a and B16-F10 Cells. Immunocytochemistry assays were performed as previously described 15 . A peptide comprised by the 15 N-terminal amino acid sequence of mouse melanopsin (Genbank accession NP_038915) with an appended C-terminal cysteine (MDSPSGPRVLSSLTQC) (Uniformed Services University of the Health Biomedical Instrumentation Center, Bethesda, MD, USA) was conjugated to keyhole limpet hemocyanin and used to immunize rabbits (Covance Labs, Denver, PA, USA). The antisera were used with no further purification 1 . Previous studies showed that increasing the concentration of the antigenic peptide led to the loss of immunoreactivity by pre-absorption in a dose-dependent manner, and that retinas of OPN4 knockout mice showed lack of immunoreactivity 32 .
Melan-a or B16-F10 cells were seeded (10 4 /well) into 8-chamber slides in the experimental medium as described above. The cells were kept in DD at 37 °C for 3 days and at the beginning of the 4 th day they were divided into 2 groups: the control remained in DD at 37 °C while the experimental group was exposed to 1 h of heat stimulus (39.5 °C). Twenty-four hours later the medium was removed and the cells were fixed in 4% paraformaldehyde as described below.
To verify the effectiveness of the Opn4 silencing Melan-a or B16-F10 cells were seeded (10 4 /well) into 8-chamber slides and kept in DD for 24 hours. On the 2 nd day, the cells were transfected with esiRNA, and immunostaining of OPN4 was performed 48 h after transfection. The cells were incubated in the primary antibody anti-melanopsin (1:500, Covance Laboratories, Denver, PA, USA) overnight at 4 °C. A Cy3-labeled anti-rabbit secondary antibody (1:500, Jackson Immunolab, West Grove, PA, USA) was applied for 1 h at room temperature. Melanopsin Knockdown by Endoribonuclease Small Interfering RNA (esiRNA). We used esiRNA as gene silencing tool (Sigma-Aldrich, St. Louis, MO, USA) that targets mouse Opn4 variants 1 and 2 (access numbers NM_001128599.1 and NM_013887.2). The esiRNA results from the cleavage of long double-stranded RNA (dsRNA). This process generates a heterogeneous mixture of siRNAs, all of which target the same mRNA of interest. This methodology provides highly selective gene suppression with lower off-target effects than single or pooled siRNAs 26 . As a control for our experiments, we used an esiRNA that targets the mRNA of Enhanced Green Fluorescent Protein (EGFP), which can be used as a negative control in systems that lack this protein.
Melan-a or B16-F10 cells were seeded (10 5 and 5 × 10 4 /well respectively) in a 12-well plate, and after the procedure described in the Experimental Design section they were transfected with esiRNA against melanopsin or EGFP (both at 10 nM) using Lipofectamine 3000 transfection kit (ThermoFisher, Waltham, MA, USA) according to the manufacturer's instructions.
To verify the functional role of melanopsin in perceiving heat, Melan-a or B16-F10 cells were divided into four groups 48 hours after transfection: (1) control group transfected with esiRNA EGFP (10 nM) at 37 °C; (2) heat-stimulated (39.5 °C) group transfected with esiRNA EGFP (10 nM); (3) group transfected with esiRNA melanopsin (10 nM) at 37 °C; (4) heat-stimulated (39.5 °C) group transfected with esiRNA melanopsin (10 nM). Total RNA was extracted immediately after the end of heat stimulus and gene expression of Opn4 (melanopsin encoding gene) was assessed by qPCR as previously described 15 in Melan-a and B16-F10 cells transfected with esiRNA against EGFP (control group) or against melanopsin mRNA. Lipofectamine 3000 displayed no effect per se on clock gene expression (data not shown).