Adaptive Thermogenesis in Mice Is Enhanced by Opsin 3-Dependent Adipocyte Light Sensing.

Almost all life forms can detect and decode light information for adaptive advantage. Examples include the visual system, in which photoreceptor signals are processed into virtual images, and the circadian system, in which light entrains a physiological clock. Here we describe a light response pathway in mice that employs encephalopsin (OPN3, a 480 nm, blue-light-responsive opsin) to regulate the function of adipocytes. Germline null and adipocyte-specific conditional null mice show a light- and Opn3-dependent deficit in thermogenesis and become hypothermic upon cold exposure. We show that stimulating mouse adipocytes with blue light enhances the lipolysis response and, in particular, phosphorylation of hormone-sensitive lipase. This response is Opn3 dependent. These data establish a key mechanism in which light-dependent, local regulation of the lipolysis response in white adipocytes regulates energy metabolism.

INTRODUCTION

The detection of photons by animals has been exploited for adaptive advantage in many ways. The visual sense–irradiance detection by photoreceptors in the retina and formation of virtual images in the brain–is the most obvious example, because it is a component of our conscious existence. However, functioning in parallel in many types of animals are various non-visual ocular photoreceptors. In mammals, the best characterized are the retinal ganglion cells expressing melanopsin (Opsin 4 [OPN4]) and neuropsin (OPN5) that function in negative phototaxis (Johnson et al., 2010), circadian clock entrainment (Panda et al., 2002; Buhr et al., 2015), the pupillary light reflex (Hattar et al., 2002), and eye development (Nguyen et al., 2019; Rao et al., 2013).

Photoreceptors that function outside the eye are found throughout the animal kingdom. They exist as chromatophores in the skin of frogs (Moriya et al., 1996; Provencio et al., 1998), within the pineal organs that produce melatonin (Okano et al., 1994), and as deep brain photoreceptors that regulate seasonal breeding responses in avian species (Nakane et al., 2010). Extraocular photoreceptors were assumed to be absent from mammals until expression domains outside the eye were defined for OPN3, OPN4, and OPN5. Coupled with studies demonstrating light-dependent signaling (Kato et al., 2016; Kojima et al., 2011; Koyanagi et al., 2013; Yamashita et al., 2010) mammalian extraocular photoreception was proposed. We now know that OPN5 photoentrains the circadian clock in skin (Buhr et al., 2019) and that OPN4 can acutely regulate blood vessel dilation (Sikka et al., 2014, 2016). It has also been suggested that adipocyte function might be modulated by light stimulation of OPN4 (Ondrusova et al., 2017). Although attempts to express mammalian OPN3 have proven difficult, studies on its vertebrate ortholog from pufferfish suggest that it may function as a photosensitive opsin (Koyanagi et al., 2013). Accumulating evidence points to extraocular photoreception via OPN3 in both mouse and human (Barreto Ortiz et al., 2018; Buscone et al., 2017; Castellano-Pellicena et al., 2019; Regazzetti et al., 2018; Yoshimoto et al., 2018).

Mammals employ three types of adipocytes (Giralt and Villarroya, 2013). White adipose tissue (WAT) consists primarily of white adipocytes and is the major energy storage site. Brown adipose tissue (BAT) is made up exclusively of brown adipocytes, which generate heat via non-shivering thermogenesis (NST) (Giralt and Villarroya, 2013). Under conditions of cold exposure, WAT can differentiate into “brite” adipocytes that have functional UCP1, although its capacity is at most a third of that of BAT (Rosenwald et al., 2013; Shabalina et al., 2013; Wu et al., 2012). The process of lipolysis releases free fatty acids (FFAs) and glycerol from WAT for systemic use (Zechner et al., 2012). BAT then uses FFAs for the generation of heat by oxidative uncoupling via UCP1 (Ikeda et al., 2017; Kazak et al., 2015, 2017a). Thus, WAT and BAT both have important functions in the regulation of energy balance. Though it was originally believed that only newborn humans had significant depots of brown fat, it is now understood to be present in the adult (Cypess et al., 2009; van Marken Lichtenbelt et al., 2009; Nedergaard et al., 2007; Virtanen et al., 2009). Gathering evidence suggests that activation of BAT might be valuable in protecting against metabolic syndrome (Harms and Seale, 2013; Seale, 2013).

Here we describe an extraocular function for OPN3 in the light-dependent regulation of adipocyte function. When cold challenged, mice with an adipocyte-specific deletion of Opn3 fail to defend their body temperature normally, show an attenuated induction of cold-induced genes in BAT, and use less fat mass when fasted. Many of these phenotypes are reproduced in mice that are raised without the blue light wavelengths that normally stimulate OPN3. Furthermore, we show that blue light has an adipocyte-specific, acute stimulatory effect on thermogenesis. These metabolic perturbations appear to be explained by the OPN3 and blue light dependence of the lipolysis response, a pathway that normally provides fatty acid fuel for thermogenesis. These data identify an unanticipated mechanism for light information decoding for energy homeostasis.

RESULTS

OPN3 Is Expressed in Adipocytes

No reliable antibodies for murine OPN3 are presently available, so to assess expression of Opn3, we took advantage of three alleles (Figure S1): Opn3 (in combination with the tdTomato reporter Ai14), and Opn3-eGFP, an expression reporter transgene based on a bacterial artificial chromosome (GENSAT 030727-UCD). The interscapular adipose tissue (iAT) depot comprises interscapular subcutaneous white adipose tissue (iscWAT) and interscapular brown adipose tissue (iBAT). X-gal labeling of control cryosections from postnatal day (P) 16 iAT showed no background labeling in wild-type mice (Figure 1A) but intense labeling in the Opn3 iscWAT (Figure 1B). Labeled adipocytes were not detected in control iBAT (Figure 1C). In iBAT from Opn3, X-gal-labeled cells were not readily apparent, but at higher magnification and bright transillumination, a subset expressing brown adipocytes was detected (Figure 1D).

Figure 1.

Expression of Opn3 in iAT and inWAT

(A–D) X-gal-labeled wild-type (A and C) and Opn3 (B and D) cryosections of iAT, including iscWAT (A and B) and iBAT (C and D) at P16.

(E–H) X-gal-labeled wild-type (E and G) and Opn3 (F and H) cryosections of inWAT, including white adipocytes (WAd) and brite adipocytes (BrAd). (G and H) Higher magnification of inWAT in wild-type (G) and Opn3 (H) cryosections.

(I–K) Detection of tdTomato (red, grayscale) in Opn3 mice for iAT showing positive cells in iscWAT (I, enlarged region in K) and iBAT (J, enlarged region in K). iscWAT and iBAT are separated by a leaflet of muscle (m). Labeling of nuclei with Hoechst 33258 is presented in green.

Scale bars: (A, B, E, F, and K) 100 μm and (C, D, and G–J) 50 μm.

Neonatal inguinal white adipose tissue (inWAT) has a high content of brite adipocytes (Figures 1E and 1F, BrAd). In control P16 inWAT, neither the large unilocular white adipocytes nor the smaller brite adipocytes were X-gal labeled (Figures 1E and 1G). By contrast, the large, unilocular white adipocytes from Opn3 mice were X-gal positive (Figures 1F and 1H). When we used the Opn3 allele (Figure S1) to convert the tdTomato reporter Ai14 (Figures 1l–1K), cryosections showed that almost all adipocytes within iscWAT were positive (Figures 1l and 1K). In iBAT, a subset of brown adipocytes was positive (Figures 1J and 1K). The Opn3-eGFP reporter confirmed expression of Opn3 in most white and brown adipocytes (Figure S2). Finally, Genotype-Tissue Expression (GTEx; accession phs000424.v7.p2, accessed on 02/21/2018) data report low to moderate expression of OPN3 in human subcutaneous and omental adipose tissue (2.9 and 3.1 tags/million, respectively) (Figure S2). These data indicate that both mouse and human adipose tissues express Opn3 and raise the possibility of direct light responsiveness.

Photon Flux within iscWAT and iBAT Is Sufficient for Opsin Activation

To measure photon flux within the iAT of a pigmented mouse (Figure 2A), we fabricated the Holt-Sweeney microprobe (HSM) (Figure 2B) from a light-shielded optic fiber and attached a transparent spherical collecting tip, permitting omnidirectional measurements of scalar irradiance under constant angular sensitivity (Holt et al., 2014). The microprobe is mounted within a pulled Pasteur pipette and lowered into the iAT with a stereotaxic frame. Photon flux measurements across the 350–800 nm spectral range were taken every 0.5 mm up to a 2.5 mm total depth (Figure 2C). At the λmax of 480 nm for OPN3 (Figure 2D), the measured photon flux was 5 × 1014 photons cm−2s−1 at 0.5 mm (deepest point within iscWAT) (Figure 2C) and 2 × 1013 photons cm−2s−1 at 2.5 mm (deepest point within iBAT) (Figure 2C). Surface illumination was controlled to 1% of clear sky sunlight intensity (direct sunlight intensity was measured to be 2 × 1017 photons cm−2s−1). Total light attenuation ranged from less than one log quanta at 0.5 mm to just over two log quanta at 2.5 mm. Extrapolating for full sunlight, iscWAT photon flux would be approximately 5 × 1016 photons cm−2s−1. Signaling thresholds for atypical opsins are as low as 1010 photons cm−2s−1 (Wong, 2012). Thus, these data indicate that iscWAT and iBAT photon flux is sufficient for opsin stimulation.

Figure 2.

Measurement of Photon Flux within iBAT and iscWAT

(A) Schematic describing set up for measuring intra-tissue photon flux. Collimated photons from a plasma source are directed toward an anesthetized mouse, into which the fiber probe is guided via a stereotaxic frame. Spectra are measured by an OceanOptics spectrometer.

(B) Holt-Sweeney microprobe (scale bar: 100 μm) is an optic fiber with a transparent spherical tip that accepts photons over approximately 4π steradians.

(C) Measurement depths along iAT.

(D) Absolute photon flux within interscapular adipose color coded for depth. The uppermost blue trace is surface flux and, at the λmax for OPN3, is about 2 × 1015 photons cm−2s−1. At the maximum 2.5 mm depth (brown trace), the flux at the OPN3 λmax is approximately 1 × 1013 photons cm−2s−1. Each trace is averaged data from n = 3 mice. Color shading is ±SEM.

Transcriptome Analysis Suggests that OPN3 Regulates Metabolism

To address OPN3 function, we performed a microarray-based transcriptome analysis on P16 control and Opn3 germline null mice. iAT and inWAT were harvested alongside liver (low-expression control). Using the AltAnalyze suite (Emig et al., 2010; Salomonis, 2012), we identified Z score significant clustering of differentially regulated transcripts (red, upregulated; blue, downregulated) into select WikiPathway models (Figures S3A–S3C), with a more detailed schematic included (Figures S3D–S3H). As expected, the Opn3 transcript showed the highest negative fold change in Opn3 null iAT (3.0-fold down, p = 1.5 × 10−4) and inWAT (5.6-fold down, p = 7.6 × 10−3) but was not significantly changed in liver. Overall, transcriptome analysis indicated that OPN3 activity was required for normal regulation of metabolism, as evidenced by deregulation of the adipocyte extracellular matrix, lipid, glucose, and energy generation pathways.

In inWAT, Opn3-dependent, differentially regulated transcripts cluster within the peroxisome proliferator-activated receptor (PPAR) pathway and the mitochondrial electron transport chain (ETC) (Figure 3A). The PPAR pathway regulates adipocyte size, as well as lipid metabolism and energy generation (Fan and Evans, 2015; Liu et al., 2007). This pathway regulates energy generation in part because multiple components of the lipolysis pathway, including HSL (hormone-sensitive lipase), ATGL (adipose triglyceride lipase) (Leone et al., 1999; Rhee et al., 2003; Vega et al., 2000), and perilipin (PLIN) (Arimura et al., 2004), directly or indirectly depend on the transcriptional co-activator PGC1a for their expression. Furthermore, UCP1 is downregulated, presumably as a response to deregulation of its transcription factors PGC1a and RXRa/b (Puigserver et al., 1998). Notably, the transcript for lipoprotein lipase (Lpl), an enzyme with a role in extracellular lipolysis (Olivecrona, 2016) is deregulated in the liver of Opn3 null mice (Figure S3). Finally, inWAT from Opn3 null mice showed a striking cluster of 17 downregulated ETC transcripts (Figure 3A). Combined, these data suggest deregulated energy metabolism in Opn3 null mice.

Figure 3.

Opn3 Null and Minus Blue Reared Mouse inWAT Phenotype

(A) Schematic of clustered Opn3-dependent transcript changes in PPAR (WP2316), lipid uptake, and mitochondrial ETC (WP295) pathways (red, upregulated; blue, downregulated).

(B) UCP1 (green) labeling of inWAT in control and Opn3 animals at P16.

(C) Immunoblot detecting UCP1 and β-tubulin (TUBB) in P16 inWAT from Opn3 and Opn3 mice.

(D–L) Adipocyte size distribution (D and I) in inWAT comparing control and Opn3 (D) and full spectrum versus minus blue raised (I) mice at P16. Data are presented as mean ± SEM, n = 3 for each genotype. Direct comparisons between genotypes at each interval were performed with Student’s t test, *p < 0.05, **p < 0.01, ***p < 0.001.

(E and J) Hematoxylin staining of histological sections of P16 inWAT from Opn3 (E) and full spectrum (380, 480, and 630 nm) reared versus minus blue (380 and 630 nm) reared (J) mice.

(F and K) Total NAD levels in inWAT and liver for P16 Opn3+/+ and Opn3 mice (F, n = 4) or for mice reared either in full spectrum or minus blue lighting (K, n = 3). p values calculated using Student’sttest.

(G and L) Immunoblots detecting multiple components of the ETC (ATP5A, COX1, SDHB, NDUFB8, and UCP1) in P16 iAT for Opn3, and minus blue reared mice.

(H) TEM showing abnormal mitochondrial morphology in the Opn3 null iBAT at P28.

Scale bars: (B) 500 μm, (E and J) 100 μm, and (H) 2 μm.

Detection of UCP1 in inWAT either by immunofluorescence (Figure 3B) or by immunoblot (Figure 3C) confirmed the comparatively low levels in Opn3 null mice. inWAT cell-size assessment showed that Opn3 null mice had, on average, larger adipocytes (Figure 3D). Hematoxylin inWAT staining also show lower proportions of the smaller brite adipocytes (Figure 3E), consistent with adipocyte size assessment.

We then measured total NAD content in Opn3 null mice to assess mitochondrial function (Peek et al., 2013). Quantification showed no change in liver NAD in Opn3 null mice but a reduction in inWAT (Figure 3F), consistent with reduced mitochondrial content. Though transcriptome analysis for Opn3 null iBAT did not cluster the ETC, we were prompted by the effects of Opn3 deficiency on inWAT to assess mitochondrial status in iBAT. Immunoblotting for the ETC components ATP5A (Complex V), COX1 (Complex IV), SDHB (Complex II), NDUFB8 (Complex I), and UCP1 revealed some variability in the presence of SDHB in Opn3 null mice but a consistently low level of both NDUFB8 and UCP1 (Figure 3G). Correspondingly, iBAT transmission electron microscopy (TEM) at P28 shows disorganized organellar cristae (Figure 3H), similar to that of Ucp1 null mice (Kazak et al., 2017b), implying in part, Opn3-dependent changes in mitochondrial maintenance and/or organization.

To assess OPN3 function as a photoreceptive opsin, we raised C57BL/6J mice in “minus blue” lighting conditions from embryonic day (E) 16.5 (see STAR Methods) and compared them with mice raised under full spectrum conditions. Subtracting 480 nm wavelengths reduces total photon flux, but because opsins are sensitive to specific wavelengths, this lighting paradigm ensures that any non-blue opsin receives unchanged stimulation. Minus blue lighting resulted in inWAT with quantifiably larger adipocytes (Figure 3I), lower brite adipocyte content (Figure 3J), and lower total NAD (Figure 3K). Furthermore, minus blue iBAT showed lower levels of NDUFB8 and UCP1 (Figure 3L). When compared with Opn3 null mice, the minus blue phenotype is milder. This might be explained by residual activation of OPN3 resulting from low-efficiency absorption of violet and red photons. Altogether, these data support the hypothesis that OPN3 functions as a light sensor that regulates adipose tissue development.

Adaptive Thermogenesis in Mice Is Promoted by Blue Light in an OPN3-Dependent Manner

In mice, body temperature is partly maintained by heat that is generated by skeletal muscle shivering or within BAT via NST pathways that employ UCP1, creatine metabolism, and calcium cycling (Ikeda et al., 2017; Kazak et al., 2015, 2017a). The energy for thermogenesis is provided partly by the oxidative metabolism of FFAs that are stored in adipocytes. The process of lipolysis liberating FFAs is thus crucial for normal NST (Himms-Hagen, 1972). Furthermore, it has been shown that lipolysis in white adipocytes is directly required to fuel NST (Schreiber et al., 2017; Shin et al., 2017). Several features of our Opn3 null and minus blue mice suggested defects with NST.

When Opn3 and Opn3 neonatal mice were exposed to 4°C over the course of 3 h under full spectrum lighting, core body temperatures (CBTs) of Opn3 null mice were lower than those of wild-type mice (Figure 4A). To account for possible thermogenic influences caused by the mixed genetic background of Opn3 mice, we generated a new Opn3 loss-of-function allele on a pure C57BL/6J background by CRISPR (Opn3) and repeated the analysis. Opn3 mice also show lower defended CBTs (Figure 4B), confirming the reduction in NST to be Opn3 dependent. We could again test the assertion that OPN3 functions as a light sensor by assessing NST in minus blue mice. After 3 h of cold exposure, minus blue mice similarly showed lower defended CBTs than mice reared under full spectrum lighting (Figure 4C). These data show that the absence of 480 nm light that normally stimulates OPN3 during development mimics genetic loss of function.

Figure 4.

Opn3 Is Required for Light-Dependent Enhancement of the Thermogenesis Response

CBT assessments over a time course after a 4°C cold exposure for P21–P24 (neonatal) or adult mice of the indicated genotypes. Lighting conditions during cold exposure are indicated by the colored lines above the chart’s horizontal axis.

(A) CBTs during cold exposure in Opn3 and control Opn3 in full spectrum (380 + 480 + 630 nm) lighting.

(B) CBTs during cold exposure in Opn3 and control Opn3 C57BL/6J background mice in full spectrum lighting.

(C) CBTs during cold exposure for C57BL/6J mice raised either in full spectrum (gray trace) or in minus blue (380 + 630 nm, blue trace) lighting.

(D) CBTs during cold exposure in the same cohorts of mice shown in (A) except in minus blue lighting.

(E) As in (D) except for adult mice (2 months).

(F) CBTs during cold exposure in Opn3 and control Opn3 in full spectrum (380 + 480 + 630 nm) lighting for 180 min and minus blue (480 nm withdrawn) lighting for a further 120 min.

Data are presented as mean ± SEM.

At the same time, it was possible that OPN3 could mediate acute light responses. We tested this by cold exposing the same cohort of Opn3 wild-type and null mice (Figure 4A) under minus blue lighting. Interestingly, the CBTs of control and Opn3 null mice were statistically indistinguishable (Figure 4D). This suggested that blue light can, via OPN3, acutely promote adaptive NST and that the activity of OPN3 is necessary for this effect.

Neonatal mice have a beige adipocyte content that is higher than it is in adult mice (Sanchez-Gurmaches et al., 2012; Xue et al., 2007), and this may reflect special NST requirements given their low mass-to-surface area ratio. Thus, we also sought to establish whether adult mice showed a blue-light-promoted, OPN3-dependent NST response. For this analysis, we performed two types of experiments. In the first, we assessed CBTs in cohorts of adult control and Opn3 null mice in the minus blue condition and showed that they were indistinguishable (Figure 4E). Then, with the same cohorts of mice, we repeated the assessment in full spectrum lighting and showed that over 3 h of cold exposure, wild-type mice CBTs were higher than those of Opn3 null mice (Figure 4F, to minute 180). This shows that as in neonatal mice, adult NST is promoted by light and by OPN3 activity. At minute 180, we turned off the 480 nm light and observed a decrease in CBTs of wild-type mice to become indistinguishable from that of Opn3 null mice. This provides further evidence that blue light can acutely regulate adaptive thermogenesis in an OPN3-dependent manner and that this can occur in both neonatal and adult mice.

White Adipocyte Opn3 Is Required for a Normal Thermogenesis Response

To determine whether we could attribute aspects of the germline null phenotype to adipocyte Opn3, we conditionally deleted Opn3 with pan-adipocyte Adipoq-cre (Eguchi et al., 2011). The inWAT of neonatal Adipoq-cre; Opn3 mice shows low beige content (Figures 5A and 5B) and larger adipocyte size distribution (Figure 5C) similar to Opn3 and minus blue mice. This is consistent with the hypothesis that OPN3 functions as a light sensor within adipocytes to regulate adipose tissue development.

Figure 5.

White Adipocyte Opn3 Is Required for a Normal Thermogenesis Response

(A and B) Hematoxylin staining of Opn3 (A) and Adipoq-cre; Opn3 (B) inWAT at P16.

(C) Adipocyte size assessment in Opn3 and Adipoq-cre; Opn3 inWAT at P16. * = p < 0.05 by Student’s t test.

(D–H) CBT assessments over a time course after a 4°C cold exposure for adult mice of the indicated genotypes. Lighting conditions during cold exposure are indicated by the colored lines above the chart’s horizontal axis.

(D and E) CBTs of Opn3 and Ucp1-cre; Opn3 mice (D) under minus blue conditions and (E) in full spectrum lighting for 180 min and then for a further 120 min in minus blue.

(F) As in (E) except for cohorts of neonatal Adipoq-cre; Opn3 and control Opn3 mice.

(G) As in (D) except for cohorts of adult Adipoq-cre; Opn3 and control Opn3 mice.

(H) As in (F) except for adult mice.

(I) Relative expression of transcripts for the thermogenesis pathway genes Ucp1, Pgc1a, Prdm16, Dio2, Cidea, and Pparg in iBAT from mice of the indicated genotypes. iBAT was harvested from control mice in ambient temperature (24°C) and those exposed to 4°C for 3 h.

Data are presented as mean ± SEM. Scale bars: 100 μm.

Because both brown and white adipocytes express Opn3, it was possible that the reduced NST in Opn3 null mice could be explained by either BAT or WAT defects (or both). Thus, we measured CBTs in cohorts of neonatal mice in which Opn3 was conditionally deleted either from only brown adipocytes with Ucp1-cre (Kong et al., 2014) or from all adipocytes with Adipoq-cre. We first confirmed efficient cre-mediated deletion of the floxed region of Opn3 in both Ucp1-cre; Opn3 and Adipoq-cre; Opn3 mice by PCR (Figure S5A). Cold-exposed Ucp1-cre; Opn3 mice had CBTs indistinguishable from those of control Opn3 mice in any lighting condition (Figures 5D and 5E). This indicated that brown adipocyte Opn3 was not required for normal NST.

By contrast, neonatal Adipoq-cre; Opn3 mice in full spectrum lighting showed a more limited ability than Opn3 control mice to defend their body temperature (Figure 5F). When 480 nm light was withdrawn at minute 180, CBTs of Opn3 mice rapidly dropped to the level of the conditional null (Figure 5F). Repetition of cold exposure experiments in adult Opn3 and Adipoq-cre; Opn3 null mice showed a similar thermogenesis deficit (Figures 5G and 5H) in which the absence of blue light, either throughout the cold exposure (Figure 5G) or acutely at minute 180 (Figure 5H), could mimic the conditional loss of adipocyte Opn3. The absence of an NST deficit in Rx-cre; Opn3 mice (Figure S5B), in which Opn3 is conditionally deleted from retinal neurons, confirms that retinal OPN3 is not necessary for this pathway. Furthermore, unchanged tail temperature measured via infrared thermography in Adipoq-cre; Opn3 mice (Figure S5C) indicates that the observed CBT differences are unlikely to be a consequence of cutaneous vasodilation and conductive heat loss. Finally, qPCR assessment of a set of thermogenesis pathway transcripts (Ucp1, Pgc1a, Prdm16, Dio2, Cidea, and Pparg) in iAT showed, consistent with transcriptome analysis and the NST deficit, that loss of adipocyte Opn3 resulted in diminished expression of Ucp1, Pgc1a, and Prdm16 (Figure 5I). Combined, these data suggest that Opn3 expression in white adipocytes, but not in retina or BAT, is required for the light-dependent component of the thermogenesis response.

Loss of Opn3 Results in Decreased Energy Expenditure

To determine how Opn3 contributes to longitudinal energy balance and homeostasis, we performed indirect calorimetry in 4-month-old Opn3 (n = 5) and Opn3 (n = 10) male mice. Animals were acclimated and individually housed in metabolic chambers (TSE Systems, PhenoMaster Cages), to assess energy expenditure at 22°C, 16°C, 10°C, and 30°C (Figure 6A). On average, Opn3 mice showed lower oxygen consumption (Figure 6B) and carbon dioxide expiration (Figure 6C) than controls, with the differences becoming significant at the lowest ambient temperatures (p = 0.03 and p = 0.04 for O2, separate repeated-measures ANOVA conducted across a 5-h period during lights off and lights on, respectively). These differences disappeared once the ambient temperature was restored to thermoneutrality (30°C), suggesting that these differences represent adaptive, not pathological, responses. Correspondingly, Opn3 animals show lower energy expenditure (Figure 6D), with significant differences at 10°C. Unnormalized energy expenditure data (Figures S4A–S4C) also argue for differences in energy expenditure between Opn3 and Opn3 mice that are exacerbated by decreasing ambient temperature.

Figure 6.

Loss of Opn3 Alters Energy Metabolism

(A) Schematic detailing the ambient temperature throughout the experiment and the durations of measurement intervals.

(B–D) VO2 (B), VCO2 (C), and energy expenditure (EE, D) measurements by indirect calorimetry (TSE Systems, PhenoMaster Cages) were obtained for 22°C, 16°C, 10°C, and 30°C from Opn3 (gray trace, n = 5) and Opn3lacz/lacz (blue trace, n = 10) animals. Each graph shows a 24-h period of averaged data ± SEM for the corresponding ambient temperature. The lighting conditions were maintained at a standard 12L:12D cycle, depicted as lights off 6PM–6AM (gray shaded region) followed by lights on 6AM–6PM (yellow shaded region).

(E) Respiratory exchange ratio (RER) was calculated by the ratio VCO2/VO2.

(F) Spontaneous locomotor activity (XY) was measured by infrared beam breaks.

(G and H) 24-h average food (G) and water consumption (H) were measured by differential weight-based sensors and plotted per ambient temperature.

(I) Postmortem fat depot masses in iAT, inWAT, and pgWAT. All statistics performed on the data in (B)–(F) are repeated-measures ANOVA across 5-h time intervals: 7PM–12AM, 12AM–5AM, 7AM–12PM, and 12PM–5PM.

Statistics performed on the data in (G)–(I) are 2-way ANOVAs with p values reported from Holm-Šídák corrected multiple comparisons.

The respiratory exchange ratio (RER) was estimated by calculating the ratio of VCO2/VO2 and was not significantly changed between Opn3 and Opn3 animals throughout the experiment. This suggests the absence of substrate utilization preference and reflects the overall decreased metabolic demand caused by the loss of Opn3. This is supported by the lack of locomotor activity differences between Opn3 and Opn3 animals (Figure 6F), which decouples the observed changes in energy expenditure from gross activity levels. In addition, locomotor activity data strongly suggest the absence of change in circadian phasing between Opn3 and Opn3 animals, implying that loss of Opn3 does not result in an altered activity cycle. Finally, Opn3 animals consumed less food (Figure 6G) and water (Figure 6H) compared with controls, consistent with their lower energy expenditure. Postmortem fat depot masses were higher in the Opn3 animals, perhaps suggesting diminished fat mobilization (Figure 6I). Altogether, these data argue for the importance of OPN3 for adaptive thermogenesis, with direct consequences for whole-organism energy storage and utilization.

White Adipocyte OPN3 Is Required for Normal Lipolysis during Cold Exposure

The long-standing belief that BAT lipolysis is essential for NST has been challenged by recent analyses (Schreiber et al., 2017; Shin et al., 2017) showing that inhibiting lipolysis in BAT does not compromise defended CBTs as long as WAT or cardiac muscle lipolysis is intact. Because this distinction mimicked the CBT differences between our Ucp1-cre; Opn3 and Adipoq-cre; Opn3 mice, we asked whether Opn3 in white adipocytes was necessary to provide thermogenic fuel during cold exposure. This was addressed with two complementary approaches: First, we performed an in vivo fasting experiment aimed at augmenting the use of fat reserves via lipolysis during cold exposure (Shin et al., 2017; Schreiber et al., 2017).

In this analysis, cohorts of control (Opn3) and experimental (Adipoq-cre; Opn3) mice were either fed ad libitum or fasted overnight and then exposed to 4°C for 3 h (Figure 7A). At the termination of the experiment, fat depots were dissected and weighed. After subtracting fat mass differences between fed and fasted animals, we found that Adipoq-cre; Opn3 animals mobilized significantly less fat mass than controls (380 versus 230 mg iAT, 380 versus 180 mg inWAT, and 300 versus 33 mg perigonadal white adipose tissue [pgWAT]) (Figure 7C). This was consistent with the hypothesis that adipocyte Opn3 is required for a normal utilization of fat mass. Fat mass changes under conditions of fed versus fasting state have been documented previously (Syamsunarno et al., 2014) and illustrate the need for lipolysis under conditions of nutritional deprivation.

Figure 7.

Opn3-Dependent Fat Mass Utilization In Vivo and Light- and Opn3-Dependent Lipolysis Activation In Vivo and In Vitro

(A) Schematic describing the timeline of a fasting-cold exposure experiment.

(B) CBT of fed and fasted Opn3 and Adipoq-cre; Opn3 mice during 180 min of cold exposure.

(C) Fat mass used by Opn3 and Adipoq-cre; Opn3, fed and fasted, cold-exposed mice.

(D) Serum glycerol levels from the same fed and fasted Opn3 and Adipoq-cre; Opn3 mice as in (C) after the 180 min of cold exposure.

(E) Chart showing that Opn3 mice have elevated cAMP levels in lysates of inWAT during the light phase versus the dark phase, while the cAMP levels in the inWAT of Opn3 mice are similar in both phases, comparable to levels observed in the Opn3 dark phase and significantly lower than those seen in the Opn3 light phase.

(F) Cultured in vitro differentiated adipocytes show 475 nm light-dependent, dose-response elevation of cAMP.

(G) Two examples of immunoblots showing light-dependent and Opn3-dependent induction of phospho-660-HSL in cultured in vitro differentiated adipocytes. Each set of immunoblots (experiment 1 and experiment 2) was performed using white adipocytes isolated from separate mice.

(H) Quantification of phospho-HSL induction in wild-type control (Opn3, n = 5 mice) and Opn3 loss-of-function (Opn3, n = 3 mice) white adipocytes.

(I and J) Quantification of glycerol released from in vitro differentiated Opn3 (I) and Opn3 (J) cells in response to 2 h of blue light stimulation compared with darkness.

Lipolysis is initiated by β-adrenergic receptor activation of Gαs and adenyl cyclase (Carey, 1998). This elevates cyclic AMP (cAMP) and engages targets of protein kinase A (PKA), including HSL, PLIN, and cAMP response element-binding protein (CREB), liberating glycerol and FFAs from stored triglycerides (Rogne and Taskén, 2014). Fasted mice showed significantly elevated serum glycerol compared with fed mice, but this difference was diminished in Adipoq-cre; Opn3 mice compared with controls (Figure 7D). Assessment of cAMP from inWAT lysates revealed that control mice show significantly elevated cAMP compared with Adipoq-cre; Opn3 mice during the light phase of a 12:12 h light-dark (12L:12D) cycle (Figure 7E). However, this difference was abolished during the dark phase. Thus, both measurements of serum glycerol and cAMP in inWAT support the hypothesis that OPN3 is required for a light-dependent regulation of the lipolysis response.

In our second approach, we asked whether adipocytes could respond to light in isolation. Therefore, we differentiated adipocytes from the stromal vascular fraction (SVF) of inWAT from Opn3 and Opn3 mice. cAMP measurements from these white adipocytes follow a dose-dependent response to photon flux (Figure 7F). We then exposed cultured adipocytes of both genotypes to 480 nm light and compared levels of phosphorylated PKA substrates to unexposed cultures. We observed consistently elevated phosphorylated HSL (phospho-HSL) in light-stimulated Opn3 adipocytes compared with dark ones (Figure 7G). However, this light-dependent elevation was absent from Opn3 adipocytes. Quantification of this response shows significantly higher light-induced phospho-HSL in Opn3, but not Opn3, cultured adipocytes (Figure 7H). Light-induced Opn3-dependent changes in other PKA targets were also observed (Figure S5D). We also measured free glycerol in culture media from differentiated white adipocytes and found a significant elevation of free glycerol in blue-light-stimulated control adipocytes (Figure 7I), but not Opn3 null adipocytes (Figure 7J). These data argue for direct light responsiveness in isolated adipocytes that requires OPN3.

It has previously been suggested that OPN4 can mediate light responses in cultured primary adipocytes (Ondrusova et al., 2017). We therefore explored the possibility of OPN3-OPN4 interaction in light-mediated adipocyte function. Assessment of Opn4 expression in iscWAT and iBAT from Opn4 (Ecker et al., 2010) mice revealed no lineage marking in adipocytes (Figures S5F and S5G), despite a robust signal in the retina (Figure S5E). To address a potential role for OPN4 in NST, we measured CBTs in cold-exposed cohorts of Opn4 wild-type and null mice but found no significant differences (Figure S5H). Finally, we stimulated differentiated adipocytes from Opn4 and Opn4 null inWAT with blue light and showed a robust induction of phospho-HSL in both (Figure S5I). These data were inconsistent with an in vivo role for adipocyte OPN4 in local light responses. We also tested the involvement of Rhodopsin (Opsin 2 [OPN2]), the photopigment-mediating vision under dim light, in defending CBTs on cold exposure. We used a mouse with a P23H mutation in the Opn2 gene that produces a nonfunctional Rhodopsin protein (Sakami et al., 2011) (Figures S5J and S5K). Assessment of defended CBTs during cold exposure revealed no significant differences (Figure S5L). These data argue against a nonvisual function of OPN2 in NST.

DISCUSSION

We present an assessment of OPN3 (Blackshaw and Snyder, 1999; Koyanagi et al., 2013) (encephalopsin) function in the mouse, in which adipocytes use OPN3-dependent light sensing to regulate metabolic physiology. Extraocular photoreception has been described in many species, including vertebrates such as fish (Kojima and Fukada, 1999; Sato et al., 2016) and birds (Nakane et al., 2010). To date, however, there are only a few documented examples of extraocular light reception in mammals. It has been argued that non-canonical opsins function within adipocytes (Ondrusova et al., 2017) and within the skin (Buhr et al., 2019; Buscone et al., 2017; Castellano-Pellicena et al., 2019; Regazzetti et al., 2018) and that they can mediate a vasorelaxation response (Sikka et al., 2014, 2016) and induce autophagy in human colon cancer cells (Yoshimoto et al., 2018). In this report, our data argue that adipocyte OPN3 has an important role in regulating lipid homeostasis.

OPN3 Activity Mediates a Light-Dependent Pathway that Regulates Energy Metabolism

OPN3 has all the crucial molecular characteristics of the opsin family of light-responsive G protein-coupled receptors (Koyanagi et al., 2013). Thus, one hypothesis to explain the Opn3 phenotype places OPN3 as the candidate detector for decoding light information to regulate energy homeostasis. We tested this by raising C57BL/6J mice in minus blue conditions that exclude 480 nm wavelengths known to stimulate OPN3 homologs in other vertebrates (Koyanagi et al., 2013). Remarkably, minus blue mice show the same abnormal WAT histology and low NAD, reduced iAT ETC complexes, low UCP1, and NST deficits characteristic of Opn3 null mice. This outcome is consistent with the existence of an OPN3-dependent, light-decoding metabolic regulation pathway. It suggests that OPN3 functions during development to establish the histological and functional characteristics of metabolic tissues.

A characteristic of non-canonical opsins is that they can mediate acute responses to light. OPN4 mediates the pupillary light reflex (Hattar et al., 2002) and light-aversive behavior in neonatal mice (Johnson et al., 2010). We showed that OPN3 could mediate light responses over a similar timescale by demonstrating light- and OPN3-dependent changes in CBTs during cold exposure. When we withdrew blue light, we showed that wild-type mice rapidly reduced their CBTs to abnormally low Opn3 null levels. In this assessment, CBTs of Opn3 null and wild-type mice were indistinguishable in minus blue conditions, indicating that OPN3 activity is necessary for the acute enhancement of body temperature by blue light. However, because minus blue reared animals also show deficits in NST, even under blue light stimulation (Figure 4C), OPN3 likely possesses additional developmental roles that are not addressed in the current study.

Acute light stimulation enhances body temperature in humans (Cajochen et al., 2000; Dijk et al., 1991), and this response is mediated by 460 nm light, but not 550 nm light (Cajochen et al., 2005). Though OPN4 has been implicated due to known circadian regulation of CBTs, the current analysis suggests the alternative hypothesis that OPN3-dependent light responses are central to this physiology. Analysis in Drosophila, demonstrating that acute light exposure causes elevated temperature preference (Head et al., 2015), suggests that this configuration of light information decoding is deeply conserved. Humans differ from mice in that we are a diurnal species, and the metabolic interaction between OPN3 and human circadian clock remains an open question. Even so, it is very likely that the activity of OPN3 in the light-dependent regulation of metabolic pathways and body temperature will be tightly integrated with OPN4-dependent circadian and ocular photic input pathways that also regulate this physiology (Rupp et al., 2019).

White Adipocytes Are a Site of OPN3 Metabolic Activity

Many cell types express Opn3 (Blackshaw and Snyder, 1999; Nissilä et al., 2012; Regard et al., 2008; Sikka et al., 2016), and this raises the possibility that extraocular light reception in mammals is commonplace. In the current study, we have shown that white adipocytes are a crucial site of OPN3 function for NST. Although we excluded brown adipocyte Opn3 activity from involvement in NST, there are likely to be adipocyte-independent activities of OPN3 not identified by the current analysis.

Prompted partly by transcriptome analysis, we showed that lipolysis in cultured white adipocytes is enhanced by blue light in an OPN3-dependent manner. As illustrated by the lower-than-normal body temperature that results when lipid mobilization enzymes are compromised (Schreiber et al., 2017), lipolysis is an essential component of a normal thermogenesis response in mice. Blue-light-stimulated white adipocytes show elevated cAMP and, importantly, dramatic elevation of phospho-HSL, the rate-limiting enzyme in the lipolysis pathway (Langin et al., 1996), a response lost in Opn3 null adipocytes. Because lipolysis is an essential response for normal body temperature maintenance (Shin et al., 2017, 2018) and the resulting FFAs are required for the activation of UCP1 (Cannon and Nedergaard, 2004; Divakaruni and Brand, 2011; Fedorenko et al., 2012; Gonzalez-Hurtado et al., 2018), these findings provide, at least partly, a mechanistic explanation for the OPN3-dependent deficit in NST. Although we still do not fully understand the biochemistry of OPN3 receptor coupling and signaling in adipose tissue, the reduced ability of Adipoq-cre; Opn3 mice to use fat mass in response to fasting and cold exposure is consistent with a role for OPN3 in enhancing lipolysis in vivo. It would thus be crucial for future work to elucidate the specific OPN3-dependent signaling mechanisms in adipose tissue, enabling a better understanding of the direct link between blue-light-sensing OPN3 and lipolytic enzymes.

There is a growing body of evidence that OPN3 mediates light-dependent regulation of cellular physiology in mice and diverse human cell types (Barreto Ortiz et al., 2018; Buscone et al., 2017; Regazzetti et al., 2018; Yoshimoto et al., 2018). We now provide key evidence that OPN3 can regulate physiology at the organismal level, at least in the mouse. Both the primary amino acid sequence and the expression pattern of OPN3 are highly conserved. If the light-OPN3 adipocyte pathway exists in humans, there are potentially broad implications for human health. Our modern lifestyle subjects us to unnatural lighting spectra, exposure to light at night, shift work, and jet lag, all of which result in metabolic disruption (Fonken and Nelson, 2014; Fonken et al., 2013; Laermans and Depoortere, 2016; Opperhuizen et al., 2017). Based on the current findings, it is possible that insufficient stimulation of the light-OPN3 adipocyte pathway is part of an explanation for the prevalence of metabolic deregulation in industrialized nations where unnatural lighting has become the norm.

STAR★METHODS

LEAD CONTACT AND MATERIALS AVAILABILITY

This study did not generate new unique reagents. Mouse lines generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement. Further information and request for resources should be directed to and will be fulfilled by the Lead Contact, Richard A. Lang (richard.lang@cchmc.org).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

All experiments were approved by Cincinnati Children’s Hospital Medical Center, The University of Michigan Ann Arbor, and The University of Washington Institutional Animal Care and Use Committees and were in accordance with the National Institute of Health guidelines. Ages of mice used in this study include P16 (immunohistochemical analyses and microarray), P21-P24 (neonatal cold exposure), P28 (transmission electron microscopy), 2 months (adult cold exposure), and 3-4 months (fiber radiometry and indirect calorimetry). Male and female mice were used for all studies unless otherwise stated.

Mice

Animals were housed in a pathogen-free vivarium in accordance with institutional policies. Genetically modified mice used in this study were: B6;FVB-Tg(Adipoq-cre)1Evdr/J (Eguchi et al., 2011)(Jax stock #010803), Ai14 (Madisen et al., 2010)(Jax stock #007914), Opn4 (Panda et al., 2003), and Tg(Opn3-EGFP)JY3Gsat (MMRRC stock number 030727-UCD). The Ucp1 mouse line used in the thermoregulation assay studies was obtained from Jackson Laboratories: B6.FVB-Tg(Ucp1-cre)1Evdr/J (Jax stock #024670). The Opn4 mouse line was generously donated by Kwoon Y. Wong from the University of Michigan Ann Arbor (Ecker et al., 2010). Opn3 were generated from C57BL/6N ES cells obtained from EUCOMM (ES clone ID: EPD0197_3_E01). The ES cells harbor a genetic modification wherein the lacz-Neomycin cassette is flanked by FRT sites and a loxp site separates lacz from the neomycin coding region. Loxp sites also flank exon 2 of Opn3 allowing multiple mouse lines that can serve as reporter nulls, conditional floxed and null mice. The Opn3 reporter null line was created by crossing Opn3 mice to FVB/N-Tg(EIIa-cre)C5379Lmgd/J mice (Lakso et al., 1996)(Jax stock #003314). Opn3 line was created by crossing the Opn3 mice to 129S4/SvJaeSor-Gt(ROSA)26Sor(Farley et al., 2000)(Jax stock #003946) to remove the lacZ cassette. This means that Opn3 mice are of mixed C57Bl6/6N, FVB/N background and that Adipoq-cre; Opn3 mice are of mixed C57Bl6/6N, 129S4/Sv, B6;FVB background. Littermate control animals were used for all experiments with the exception of C57BL/6J mice reared under different lighting conditions.

The Opn3 was generated in-house using CRISPR-Cas9 technology. Four gRNAs that target exon 2 of Opn3 were selected to knock in the Cre cassette. Plasmids containing the gRNA sequence were transfected into MK4 cells (an in-house mouse cell line representing induced metanephric mesenchyme undergoing epithelial conversion). The editing efficiency of gRNA was determined by T7E1 assay of PCR products of the target region amplified from genomic DNA of transfected MK4 cells. The sequence of the gRNA that was subsequently used for the transfection is TACCGTGGACTGGAGATCCA. Sanger sequencing was performed to validate the knock-in sequence of founder mice.

The Opn3ΔEx2 allele was generated in-house using CRISPR-Cas9 technology as above. Four gRNAs that target exon 2 of Opn3 were selected. The sequences of the gRNAs are: for the 5′ end: TAGCAACGAATGCAAAGGTA GGG and ATCCACATGTTCTGCCCAGGAGG. For the 3′ end: GCGCTATGTTGGTAAGGTGT GGG and TGTGGTTTTAATCAGCACAGGGG. Out of the 6 pups derived from one injection, the founder animal had a 2203 bp deletion that was also confirmed by Sanger sequencing. The proximal breakpoint of this deletion is intron 1 (bp 175,667,424) to intron 2 (bp175,665,054) thus deleting the entirety of exon 2.

Mice were placed on normal chow diet (NCD: 29% Protein, 13% Fat and 58% Carbohydrate kcal; LAB Diet #5010) ad libitum with free access to water.

Genotyping

Primer sequences and pairs for genotyping each of the alleles in this study are listed in the table below:

OPN3 Mouse Lines

All OPN3 mouse lines used in this study, along with the scientific rationale, are detailed in the table below:

METHOD DETAILS

Lighting Conditions

Animals were housed in standard fluorescent lighting (photon flux 1.62x1015 photons/cm2/sec) on a 12L:12D cycle except where noted. For full spectrum lighting, LEDs were used to yield a comparable total photon flux of 1.68 × 1015 photons/cm2/sec. Spectral and photon flux information for LED lighting: near violet (λmax = 380 nm, 4.23 × 1014 photons/cm2/sec in the 370-400 nm range), blue (λmax = 480 nm, 5.36 × 1014 photons/cm2/sec in the 430-530 nm range), and red (λmax = 630 nm, 6.72 × 1014 photons/cm2/sec in the 590-660 nm range). Photon fluxes were measured at approximately 24” from source and through an empty standard mouse cage. For wavelength restricted experiments, C57BL/6J animals were housed in a 12L:12D cycle starting late gestation (embryonic day E16) either in full spectrum (380 nm + 480 nm + 630 nm LEDs) or in “minus blue” (380 nm + 630 nm LEDs) lighting.

Intra-adipose Tissue Radiometry

Fabrication of the Holt-Sweeney microprobe (HSM) is described as follows (Holt et al., 2014). The termination of one end of a 100 μm silica core fiber optic patch cable (Ocean Optics, Dunedin, FL, USA) was removed. The fiber’s furcation tubing and jacketing was stripped, and the fiber’s polyimide buffer was removed 5 cm from the fiber’s end using a butane torch. A 10 g weight was attached to the end of the fiber and then pulled upon heating with the butane torch, narrowing the diameter. The narrowed region of the fiber was then cut using carborundum paper, to yield a flat fiber end with a diameter of 30 – 50 μm. The sides of the narrowed fiber were painted with a film opaquing pen to prevent stray light from entering, while leaving a small transparent opening at the fiber tip. For structural support, this bare, tapered fiber was then secured in the tip of a pulled glass Pasteur pipette using a drop of cyanoacrylate glue, leaving only 6-9 mm of bare optical fiber protruding. A small light-scattering ball was added to the end of the tapered optical fiber for spectral scalar irradiance measurements. To do this, titanium dioxide was thoroughly mixed with a high-viscosity UV-curable resin, DELO-PHOTOBOND, GB368 (DELO Industrie Klebstoffe, Windach, Germany). The tip of a pulled fiber was quickly inserted and removed from a droplet of the resin and titanium dioxide mixture, resulting in a sphere with a diameter of approximately twice that of the tapered fiber. As all measurements from a given probe were normalized to the signal from the same probe in a gelatin blank, small variations in the probe diameter have no effect on our results. The sphere was cured for 12 h using a Thorlabs fiber coupled LED light source (M375F2, Thorlabs Inc, Newton, NJ, USA).

For intra-tissue radiometric measurements in mice, 4 month-old adult animals were anesthetized under ventilated isoflurane and placed in a mouse stereotaxic frame (Stoelting Co, Wood Dale, IL, USA). The hair overlying the intrascapular region was shaved and a small 10 mm rostrocaudal incision was made through the dorsal skin to expose the underlying tissue. A 21-gauge needle attached to the stereotaxic frame was first lowered through the intrascapular region to produce a pilot hole through the adipose tissue. Following, the Holt-Sweeney microprobe was affixed to the stereotaxic frame and lowered through the pilot hole. After the probe is in position, the dorsal skin was repositioned to cover as much of the incision site as possible without obstructing the probe’s descent. For broadband light illumination, a Thorlabs plasma light source (HPLS345, Thorlabs Inc, Newton, NJ, USA) was positioned above and in front of the mouse stereotaxic frame. The light was delivered to the animal via a 5 mm liquid light guide connected to a 2 in. collimating lens secured in a vice. The distance from the collimating lens to the animal was approximately 2 ft.

Scalar irradiance measurements as a function of wavelength were obtained at the surface of the adipose tissue and at probe depth increments of 0.5 mm up to 2.5 mm. Spectral irradiance data were collected using an Ocean Optics 200-850 nm spectrometer (JAZ series, Ocean Optics, Dunedin, FL, USA).

Immunohistochemistry and Tissue Processing

Animals were anesthetized under isoflurane and sacrificed by cervical dislocation. Adipose tissue depots (interscapular adipose tissue complex and inguinal WAT) from P16 male mice were harvested and fixed in ice cold 10% zinc formalin for 1 hour at 4°C. After washing in PBS, adipose tissue samples were prepared for cryosectioning as described previously (Berry et al., 2014). Gelatin-embedded tissues were sectioned at 16 μm in a cryostat and labeled with primary antibodies as previously described (Berry et al., 2014). Chicken antibodies to GFP (ab13970, 1 in 500), and rabbit antibodies to UCP1 (ab10983, 1 in 500), were purchased from Abcam. Alexa 488 conjugated isolectin (1 in 300) and Alexa 594 conjugated F-actin were purchased from Thermo Fisher Scientific. Alexa 488 conjugated secondary antibodies (1 in 300) were purchased from Jackson ImmunoResearch.

X-Gal Staining

For X-Gal labeling, tissue samples were fixed in X-Gal fixative (1% formaldehyde, 0.2% glutaraldehyde, 2 mM MgCl2, 5 mM EGTA, and 0.01% Nonidet P-40) for two hours at room temperature. Tissues were cryosectioned as described above and then labeled with X-Gal. The reaction was monitored closely and stopped when background started to appear in control (wild-type) tissues. Following two washes in PBS, cryosections were imaged using a bright field microscope.

Hematoxylin Labeling and Cell-Size Quantification

Gelatin-embedded frozen sections of inguinal WAT (as described above) were stained with hematoxylin and imaged under bright-field. Samples were imaged with a rhodamine filter to assess adipocyte size distribution. Using the free hand selection tool on ImageJ, adipocytes were outlined and the area measured in μm2. Cell size distribution was determined by quantifying 60 cells from at least 10 regions, for a total of approximately 600 cells per animal. Cell areas were binned into 200 μm2 intervals and the frequency of total cells (%) charted for each interval.

Pre-Adipocyte Differentiation and Light Induction

IngWAT dissociation and extraction of stromal vascular fraction was performed as described before (Liu et al., 2017). Briefly, the inguinal fat pads were collected in PBS and digested in 1.5 mg/ml Collagenase A in PBS with 4% BSA and penicillin/streptomycin at 37°C, with intermittent agitation over 40 minutes. The stromal vascular fraction was extracted by passing the enzymatically dispersed cells through a 100 μm cell strainer and cultured in basal media (DMEM containing 10% fetal bovine serum and penicillin/streptomycin). For differentiation, the stromal vascular cells were plated on day1 such that the cells reached confluency on day 3. On day 4, the basal media was replaced with induction media containing Insulin (100 nM), Rosiglitazone (1 μM), IBMX (0.5 mM) and Dexamethasone (2 μg/ml) in basal media. Thereafter, the differentiating cells were maintained in basal media containing insulin (100 nM) until the day of experimentation.

Light inductions to assay lipolysis responses were typically done on day 13 – day 15 of differentiation. For this, cultures were moved to a dark, 37°C incubator, protected from light, overnight. The next day, the cultures were serum-starved, under dim red light, where the complete basal medium was washed out using serum-free medium (at least three washes) and the cells were left in serum-free media for 3 hours, before light inductions. For light inductions, half the Opn3 and Opn3 cultures were left in the dark incubator, while the rest were moved to an adjacent incubator that housed a light set-up to deliver 5x1014 photons/cm2/sec of 480 nm wavelength. The culture conditions in the two incubators were comparable except for the lighting. Prior to stimulation, any movement between incubators was accomplished with care and within a matter of seconds so as to avoid any potential temperature shock. Wild-type controls were always processed alongside Opn3 null samples. Light inductions were carried out for 30 minutes, after which the cells were washed in PBS and snap frozen by immersing the culture plates in liquid nitrogen and frozen at −80°C until lysate preparations for western blotting.

@For the light-induced cyclic AMP response, in vitro differentiated Opn3 and Opn3 adipocytes were used between days 7 and 10 of differentiation. The cells were incubated with 9-cis-Retinal (5 μM) the day before the assay and one hour before the light induction, the cells were incubated in fresh DMEM without phenol red. The light pulses (465 nm) were delivered for 30 minutes with varying intensities as indicated in the results. The cells were then harvested to quantify cAMP levels by direct immunoassay (fluorometric kit by Abcam, ab138880) as per manufacturer’s instructions. For ex vivo cAMP quantification from harvested tissue, dissected inguinal white adipose tissue samples were homogenized by a pellet pestle in ice cold lysis buffer. Briefly, all samples and standards (50 ul each) were tested in duplicates, to which 25 μL of 1x HRP-cAMP was added. The plates were incubated at room temperature for 2 hours and after the washing steps, 100 μL of AbRed indicator was added. The plates were incubated for 1 hour and the fluorescence was measured at Ex/Em = 540/590 nm using a Biotek Synergy4 microplate reader.

Western Blotting

Western blots were performed using standard protocols. Adipose tissue lysates were made in NP40 lysis buffer: 150 mM NaCl, 1% NP40, 50 mM Tris 8.0 with phosphatase inhibitors. Lysates were prepared by sonication and the lysates were separated from overlaying fat layer by three rounds of centrifugation. After BCA method of protein quantification, lysates were boiled in Laemmli sample buffer (4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.004% bromophenol blue and 0.125 M Tris HCl, pH 6.8). Blots were incubated with OxPhos antibodies (Thermofisher 45-8099 1:1000) and UCP-1 (Abcam ab10983). HRP-conjugated secondary antibodies were used at 1:5000 dilution and detected by enhanced chemiluminescence (ThermoFisher Scientific).

Glycerol Assay

Glycerol assays were performed using free glycerol detection reagent (Sigma, F6428) as per manufacturer’s instructions. For serum glycerol, terminal blood collections were performed using cardiac puncture method and sera were frozen immediately at −80C until use. For glycerol detection, 1:20 ratio of sera to free glycerol reagent was used to perform the assay.

For in vitro differentiated adipocytes, wild-type and Opn3 null cells were dark adapted overnight on Day13 or Day14 of differentiation and serum starved for at least 3 hours on the day of the experiment. All cells were given fresh serum-free media before stimulating one set of wild-type and Opn3 null cells with blue light, while another set of cells were left in the dark incubator. Culture media was collected at the end of 2 hours of incubation in blue light or darkness and frozen on dry ice immediately for storage at −80C until use. For the glycerol assay, a 1:10 ratio of culture media to free glycerol reagent was used to determine the amount of glycerol released during the 2 hours of darkness or blue light stimulation.

Microarray Analysis

Interscapular adipose tissue complex and inguinal white adipose tissue from P16 mice were harvested at one hour after lights on (ZT1) and snap frozen on dry ice. Tissue pieces were homogenized in TRIzol (TriReagent Invitrogen) using RNase-free Zirconium oxide beads (2.0 mm) in a TissueLyser II (QIAGEN). Phase separation was achieved using chloroform and RNA in the aqueous phase was precipitated using ethanol. RNA was purified by column method using GeneJET RNA purification kit (ThermoFisher Scientific #K0732) and eluted into RNase-free water. RNA quality was assessed using the Agilent 2100 Bioanalyzer and an RNA-integrity number cut-off of 7 was applied for selecting samples for microarray assay. RNA from biological triplicates were submitted for microarray assay (ClariomD, Affymetrix) to the Technology Center for Genomics and Bioinformatics, University of California, Los Angeles.

Data analysis including normalization, gene expression changes and gene-enrichment analysis was performed using AltAnalyze, developed by Nathan Salomonis at Cincinnati Children’s Hospital Medical Center. AltAnalyze uses the robust multi-array average method of normalization. Briefly, the raw intensity values are background corrected, log2 transformed and then quantile normalized. Next, a linear model is fit to the normalized data to obtain an expression measure for each probe set on each array. Gene expression changes greater than 1.1 fold were calculated using unpaired t test, where a p value < 0.05 was used as a cut-off.

Quantitative RT-PCR

Intrascapular adipose depots were harvested immediately following cold challenge assays. Snap frozen tissue was homogenized and processed for RNA as described above. RNA was treated with RNase-free DNase I (ThermoFisher Scientific #EN0521) and cDNA was synthesized using a Verso cDNA synthesis kit (ThermoFisher Scientific AB1453/B). Quantitative RT-PCR was performed with Radiant SYBR Green Lo-ROX qPCR mix (Alkali Scientific Inc.) in a ThermoFisher QuantStudio 6 Flex Real-Time PCR system. Primer information for quantitative PCR is included in the Table. Relative expression was calculated by the ΔΔCT method using Tbp (TATA binding protein) as the normalizing gene. Statistical significance was calculated by an unpaired t test, using a p value cut-off of < 0.05.

The primers used for the corresponding target gene are as follows:

Transmission Electron Microscopy

Freshly dissected adipose tissues from P28 male mice were collected and 1 mm samples from approximately similar areas were fixed in 2% glutaraldehyde, 1% paraformaldehyde in PBS for 1 hour at room temperature before processing and sectioning for transmission electron microscopy as described before(Cinti et al., 2001).

NAD/NADH Quantification

NAD levels were measured using NAD/NADH assay kit from Abcam (ab65348). Briefly, tissues samples (inguinal adipose tissue and liver) from P16 mouse pups were snap frozen in liquid nitrogen, homogenized in NADH/NAD extraction buffer and filtered through a 10kD spin column (ab93349) to remove enzymes. Assay procedure was followed per kit instructions and levels of NADH and NAD+ were determined normalized to tissue weight.

Thermoregulation Assay

Core body temperature assessment upon acute cold exposure was performed on control and experimental male and female mice with the Opn3 reporter null (Opn3 and Opn3), with the exon 2 deletion on the C57BL/6J background (Opn3), with pan-adipocyte conditional deletion of Opn3 (Opn3 and Adipoq-cre; Opn3), with brown adipocyte conditional deletion of Opn3 (Opn3 and Ucp1cre; Opn3), with retinal conditional deletion of Opn3 (Opn3 and Rx-cre; Opn3), and control and Opn4 null mice. In addition, C57BL/6J mice reared under wavelength restriction (with or without blue, as described previously) were subject to this assay. Littermates were separated from their home cage and individually housed in a home-built lighting chamber situated in an electronically monitored 4°C cold room for 3 or 5 hours depending on the assay. While the mouse was conscious, body temperature was measured rectally with a RET-3 Microprobe Thermometer (Kent Scientific) every 20 minutes for the duration of the assay. Food and water were available ad libitum for all mice except when Adipoq-cre; Opn3 mice were fasted overnight, where food withdrawal was maintained during the cold assay. The thermo probe operator was blinded to mouse genotype and prior temperature measurements throughout the study. At the end of the cold exposure, mice were euthanized and relevant tissues were collected. The 3-hour cold exposure assays subjected mice to either a red (630 nm) and violet (380 nm) LED illumination combination (RV), or a red (630 nm), blue (480 nm) and violet (380 nm) LED combination (RBV). For the 5-hour cold exposure assays, the entirety of the 3-hour assay was extended by 2 hours following withdrawal of the 480 nm wavelength LED illumination. Two different ages of animals, postnatal day 21 (P21) and 2 month-old adults, were selected for these cold exposure assays. The order of cage placement was randomized at this time, such that the thermo probe operator remained blinded. For all cold exposure assays involving fed or overnight fasted Adipoq-cre; Opn3 animals, intrascapular (iAT), inguinal (inWAT), and perigonadal (pgWAT) adipose tissues were harvested. Following animal euthanasia, the fat pads were manually dissected, and their weight recorded. For fat depots with left and right pads, both were harvested and weighed, and the average recorded per animal.

Indirect Calorimetry and Energy Expenditure

12–16-week-old Opn3 and Opn3 male mice were acclimated in metabolic chambers (PhenoMaster®, TSE Systems GmbH, Germany) for 3 days before the start of the study. Mice were continuously recorded for a total of 17 days with the following measurements taken every 15 minutes: gas exchange (O2 and CO2), food intake, water intake, and spontaneous locomotor activity (in the XY plane). Ambient temperature was adjusted via climate-controlled chambers that housed the metabolic chambers. VO2, VCO2, and energy expenditure (EE) were calculated according to the manufacturer’s guidelines (PhenoMaster® Software, TSE Systems GmbH, Germany), with EE estimated via the abbreviated Weir formula. The respiratory exchange ratio (RER) was calculated by the ratio VCO2/VO2. Where appropriate, values were normalized by body weight (mL/hr/kg for VO2 and VCO2, and kcal/hr/kg for EE). Food and water intake were measured by top-fixed load cell sensors, from which food and water containers were suspended into the sealed cage environment. For food consumption, mice demonstrating excessive food grinding behavior were excluded from statistical analyses. After 9 days of continuous recording, cages were replaced with fresh ones and sealed, and gas exchange re-equilibration completed all within 2 hours.

Tail Infrared Thermography (FLIR)

Adult Opn3 and Adipoq-cre; Opn3 animals were placed in a tubular mouse restraint (Kent Scientific, Torrington, CT). These restraints permitted respiration via a slotted nose cone but immobilized the animal while exposing the tail through a rear port. IR thermographic images were taken with a FLIR T530 infrared camera (FLIR® Systems, Wilsonville, OR). Tail temperatures were quantified by describing a pixel-averaged circular region of interest of consistent size and rostrocaudal distance from the base of the tail.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using GraphPad Prism version 4.00 (GraphPad Software), Microsoft Excel and MATLAB 2018a (Figure 6). Two-tailed distribution, two-sample unequal variance t-test was used to determine the statistical significance between two independent groups. Time series datasets between two groups (Figures 4, 5, and 6) were analyzed via one-way repeated-measures ANOVA. Datasets involving two or more factors (Figures 6G–6I) were analyzed by 2-way ANOVAs with Holm-Šídák corrected multiple comparisons.

DATA AND CODE AVAILABILITY

Whole-genome expression profiles are available at accession number GEO: GSE140757.

Allele	Primers	Pairs	bp
Opn3	F1: ACCCAGGCTTCTTTTGGTCT	F1R1 – Wild-type	1191
	R1: AGAGTCGTTGGCATCCTTGG	F1R1 – Opn3fl	1231
	F2: ACTATCCCGACCGCCTTACT	F1R2 – Opn3fl	1610
	R2: GAACTGATGGCGAGCTCAGA	F1R2 – Opn3cko	640
		F2R2 – Opn3lacz	701
Opn3cre	F1: TGCTGGCCTATGAACGTTATATCC	F1R1 – Wild-type	401
	R1: TCAGTTCTGGGTGACTAACTGATC	F1R2 – Opn3cre	390
	R2: CACTCGTTGCATCGACCGGTAATGC
Adipoqcre	F: GCATTACCGGTCGATGCAACGAGTGATGAG	FR – cre	450
Ucp1cre	R: GAGTGAACGAACCTGGTCGAAATCAGTGCG
Rxcre
Opn3ΔEx2	F1: CTCAGAACCCACAAAGTGCTGG	F1R1 – Wild-type	307
	R1: GTGGACTGCAATGTCCCATCTATC	F1R2 – Opn3ΔEx2	191
	R2: GTGGGCATCATAGCCCATTGCTAC
Opn4	F1: AGGAGTGTATAGAGCCGGAAGTCTG	F1R1 – Wild-type	520
	R1: CCAGTCCAGAAGCCTAGGGCATGCC	F2R2 – Opn4−	380
	F2: TGCTCCTGCCGAGTATCCATCATGGC
	R2: CGCCAAGCTCTTCATATCACGGGTAG
Opn4cre	F1: AGGCTGGATGGATGAGAGC	F1R1 – Wild-type	187
	R1: GTTGTGAAGCTGGGATCCTG	F2R2 – cre	184
	F2: CGACCAGGTTCGTTCACTCA
	R2: CAGCGTTTTCGTTCTGCCAA
Z/EG	F: CCCCTGCTGTCCATTCCTTA	FR – Z/EG	224
	R: TGACCATGATTACGCAAGC

Allele	Mouse Lines	Rationale & Data
Opn3IKMC		Parent allele used to derive other modular alleles.
Opn3lacZ	Opn3lacZ/lacZ	Reporter null allele. Key data presented for this mouse line includes Figures 1,3, 4, 6, and 7, where we show adipocyte expression, adipocyte morphometric differences, NAD, ETC, body temperature, gene expression differences, and adipocyte specific light responses.
Opn3fl	Adipoqcre; Opn3fl/fl	Conditional deletion of Opn3 from adipocytes. Key data for this mouse line are presented in Figures 5 and 7 where we directly measure markers of lipolysis that are OPN3- and light-dependent. This allele is crucial because it permits us to focus on analyses that explore adipocyte-specific OPN3 functions. Additional data on tail temperature from Opn3fl/fl and Adipoqcre; Opn3fl/fl animals are presented in Figure S5C.
	Ucp1cre; Opn3fl/fl	Conditional deletion of Opn3 from brown adipocytes. Key data for this mouse line are presented in Figure 5E showing no difference between Opn3fl/fl and Ucp1cre; Opn3fl/fl adult mice in defended body temperature following cold exposure.
	Rxcre; Opn3fl/fl	Conditional deletion of Opn3 from retinal progenitors. Key data for this mouse line are presented in Figure S5B showing no difference in defended core body temperature between cold exposed Opn3fl/fl and Rxcre; Opn3fl/fl adult mice.
Opn3cre	Opn3cre; Ai14	In house developed and validated constitutive Cre driver line used in combination with Ai14 (Jax #007908) to characterize OPN3 expression in iscWAT and iBAT via the fluorescent tdTomato reporter. Key data presented for this line are in Figures 1l–1K.
Opn3ΔEx2	Opn3ΔEx2/ΔEx2	In house generated CRISPR-Cas9 mediated 2203 bp deletion of exon 2 of the Opn3 locus. Key data presented for this mouse line is in Figure 4B, showing persistence of defended core body temperature differences between controls and Opn3 loss-of-function animals on a congenic C57BL/6J background.

Target	Forward primer	Reverse primer
Dio2	CAGTGTGGTGCACGTCTCCAATC	TGAACCAAAGTTGACCACCAG
Prdm 16	CAGCACGGTGAAGCCATTC	GCGTGCATCCGCTTGTG
Pgc1a	CCCTGCCATTGTTAAGACC	TGCTGCTGTTCCTGTTTTC
Cidea	TGCTCTTCTGTATCGCCCAGT	GCCGTGTTAAGGAATCTGCTG
Ucp-1	ACTGCCACACCTCCAGTCATT	CTTTGCCTCACTCAGGATTGG
Pparγ	GTGCCAGTTTCGATCCGTAGA	GGCCAGCATCGTGTAGATGA
Hprt1	TCAGTCAACGGGGGACATAAA	GGGGCTGTACTGCTTAACCAG
Tbp	GAAGCTGCGGTACAATTCCAG	CCCCTTGTACCCTTCACCAAT

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal anti-UCP1	Abcam	ab10983; RRID: AB_2241462
Chicken polyclonal anti-GFP	Abcam	ab13970; RRID: AB_330798
Rabbit polyclonal anti-TUBB3	BioLegend	802001; RRID: AB_2564645
Rabbit polyclonal anti-Recoverin	Millipore Sigma	AB5585; RRID: AB_2253622
Phalloidin (Alexa Fluor 594)	Thermo Fisher Scientific	A12381; RRID: AB_2315633
Mouse monoclonal anti-Rhodopsin	Abcam	ab3267; RRID: AB_303655
Isolectin IB4 (Alexa Fluor 488)	Thermo Fisher Scientific	I21411; RRID: AB_2314662
Anti-Phospho-PKA (pThr197)	Cell Signaling Technology	5661S; RRID: AB_10707163
Anti-PKA C-a	Cell Signaling Technology	5842S; RRID: AB_10706172
Anti-Phospho-PKA Substrate (RRxpS/T)	Cell Signaling Technology	9624S; RRID: AB_331817
Anti-Phospho-HSL (pSer660)	Cell Signaling Technology	4126S; RRID: AB_490997
Anti-HSL	Cell Signaling Technology	18381T; RRID: AB_2798800
Anti-Phospho-PLIN1 (pSer522)	Vala Sciences	4856
Anti-Phospho-CREB (pSer133)	Cell Signaling Technology	9198S; RRID: AB_2561044
Anti-phospho-ERK (p44/42)	Cell Signaling Technology	9102S; RRID: AB_330744
Chemicals, Peptides, and Recombinant Proteins
DMEM High Glucose	Thermo Fisher Scientific	11965–092
Insulin	Millipore Sigma	I2643
Rosiglitazone	Millipore Sigma	R2408
Dexamethasone	Millipore Sigma	D1756
IBMX	Millipore Sigma	I5879
9-cis-Retinal	Millipore Sigma	R5754
Critical Commercial Assays
cAMP Assay Kit (Competitive ELISA, Fluorometric)	Abcam	ab138880
OxPhos Rodent WB Antibody Cocktail	Thermo Fisher Scientific	45–8099; RRID: AB_2533835
Free Glycerol Reagent	Millipore Sigma	F6428
GeneJET RNA Purification Kit	Thermo Fisher Scientific	K0732
Verso cDNA Synthesis Kit	Thermo Fisher Scientific	AB1453
NAD/NADH Assay Kit (Colorimetric)	Abcam	ab65348
Deposited Data
Gene Expression Dataset	GEO NCBI	GEO: GSE140757
Experimental Models: Organisms/Strains
Mouse: Adipoq-cre: B6.FVB-Tg(Adipoq-cre)1Evdr/J	The Jackson Laboratory	010803; RRID: IMSR_JAX:010803
Mouse: Ucp1-cre: B6.FVB-Tg(Ucp1-cm)1Evd/J	The Jackson Laboratory	024670; RRID: IMSR_JAX:024670
Mouse: Ai14: B6;129S6-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J	The Jackson Laboratory	007908; RRID: IMSR_JAX:007908
Mouse: Opn2: B6.129S6(Cg)-Rhotm1.1Kpal/J	The Jackson Laboratory	017628; RRID: IMSR_JAX:017628
Mouse: C57BL/6J	The Jackson Laboratory	000664; RRID: IMSR_JAX:000664
Mouse: Rx-cre: B6.Cg-Tg(Rax-cre)1Zkoz/Ph/J	Klimova et al., 2013	N/A
Mouse: Opn3lacZ/lacZ	This paper	N/A
Mouse: Opn3fl/fl	This paper	N/A
Mouse: Opn3ΔEx2/ΔEx2	This paper	N/A
Mouse: Opn3cre	This paper	N/A
Mouse: Opn3-eGFP: B6-Tg(Opn3-EGFPJY3Gsat/Mmucd)/J	GENSAT	030727-UCD; RRID: MMRRC_030727-UCD
Mouse: Opn4: Opn4tm1Skay	KOMP Repository	MGI: 2449695
Mouse: Opn3: Opn3tm2a(EUCOMM)Wtsi	EuMMCR Repository	MGI: 4434308
Mouse: Opn4cre; Z/EG	Ecker et al., 2010	N/A
Oligonucleotides
Dio2 F: CAGTGTGGTGCACGTCTCCAATC	This Paper	N/A
Dio2 R: TGAACCAAAGTTGACCACCAG	This Paper	N/A
Prdm16 F: CAGCACGGTGAAGCCATTC	This Paper	N/A
Prdm16 R: GCGTGCATCCGCTTGTG	This Paper	N/A
Pgc1α F: CCCTGCCATTGTTAAGACC	This Paper	N/A
Pgc1α R: TGCTGCTGTTCCTGTTTTC	This Paper	N/A
Cidea F: TGCTCTTCTGTATCGCCCAGT	This Paper	N/A
Cidea R: GCCGTGTTAAGGAATCTGCTG	This Paper	N/A
Ucp1 F: ACTGCCACACCTCCAGTCATT	This Paper	N/A
Ucp1 R: CTTTGCCTCACTCAGGATTGG	This Paper	N/A
Pparγ F: GTGCCAGTTTCGATCCGTAGA	This Paper	N/A
Pparγ R: GGCCAGCATCGTGTAGATGA	This Paper	N/A
Hprt1 F: TCAGTCAACGGGGGACATAAA	This Paper	N/A
Hprt1 R: GGGGCTGTACTGCTTAACCAG	This Paper	N/A
Tbp F: GAAGCTGCGGTACAATTCCAG	This Paper	N/A
Tbp R: CCCCTTGTACCCTTCACCAAT	This Paper	N/A
Software and Algorithms
GraphPad Prism 8.2.1	GraphPad	GraphPad.com
MATLAB 2018a	Mathworks	Mathworks.com
ImageJ/Fiji 2.0.0	ImageJ	ImageJ.nih.gov
AltAnalyze	Salomonis, 2012	AltAnalyze.org
FLIR Tools Version 2.1	FLIR	Flir.com
Other
Infrared Thermal Camera	FLIR	T530
Rectal Thermometer Probe	Physitemp	RET-3
TSE PhenoMaster	TSE Systems Inc.	N/A
JAZ Spectrometer	Ocean Optics	JAZ