Color Opponency with a Single Kind of Bistable Opsin in the Zebrafish Pineal Organ

Color opponency with a single kind of bistable opsin in the zebrafish pineal organ

Seiji Wadaa, Baoguo Shena, Emi Kawano-Yamashitaa, Takashi Nagataa, Masahiko Hibib,c, Satoshi Tamotsud, Mitsumasa Koyanagia,e, and Akihisa Terakitaa,e,1

aDepartment of Biology and Geosciences, Graduate School of Science, Osaka City University, 558-8585 Osaka, Japan; bDivision of Biological Science, Graduate School of Science, Nagoya University, 464-8602 Nagoya, Japan; cLaboratory of Organogenesis and Organ Function, Bioscience and Biotechnology Center, Nagoya University, 464-8601 Nagoya, Japan; dDepartment of Chemistry, Biology, and Environmental Science, Faculty of Science, Nara Women’s University, 630-8506 Nara, Japan; and eThe Osaka City University Advanced Research Institute for Natural Science and Technology, Osaka City University, 558-8585 Osaka, Japan

Edited by King-Wai Yau, Johns Hopkins University School of Medicine, Baltimore, MD, and approved September 12, 2018 (received for review February 12, 2018) Lower vertebrate pineal organs discriminate UV and visible light. sensitive parietopsin, expressed in a single photoreceptor (5, Such color discrimination is typically considered to arise from 7). Interestingly, color opponency is generated through two dis- antagonism between two or more spectrally distinct opsins, as, tinct intracellular signal transduction cascades driven by pinopsin e.g., human cone-based color vision relies on antagonistic relation- and parietopsin, respectively. Thus, in the lizard parietal eye, ships between signals produced by red-, green-, and blue-cone color opponency is a property of single cells rather than neural opsins. Photosensitive pineal organs contain a bistable opsin (para- networks, as is the case for color vision. pinopsin) that forms a signaling-active photoproduct upon UV Pineal organs of lower vertebrates such as lamprey and most exposure that may itself be returned to the signaling-inactive teleosts also show color opponency, discriminating UV and vis- “dark” state by longer-wavelength light. Here we show the spec- ible light. This property has been recorded at the level of pineal trally distinct parapinopsin states (with antagonistic impacts on sig- ganglion cells (that receive light information from several pineal naling) allow this opsin alone to provide the color sensitivity of this photoreceptor cells), with firing of these neurons suppressed by organ. By using calcium imaging, we show that single zebrafish UV and enhanced by visible light (6, 8). We previously found pineal photoreceptors held under a background light show re- that parapinopsin, an opsin first identified in catfish pineal and sponses of opposite signs to UV and visible light. Both such re- parapineal organs (9), is a UV-sensitive opsin and that it sup- sponses are deficient in zebrafish lacking parapinopsin. Expressing ports the UV reception involved in color opponency of the a UV-sensitive cone opsin in place of parapinopsin recovers UV lamprey pineal organ (10). Interestingly, parapinopsin is a responses but not color opponency. Changes in the spectral com- bistable opsin: exposure of dark-adapted parapinopsin to UV position of white light toward enhanced UV or visible wavelengths light produces a stable photoproduct that is itself maximally respectively increased vs. decreased calcium signal in parapinopsin- sensitive to visible light and, on light absorption, can regenerate sufficient but not parapinopsin-deficient photoreceptors. These data the original dark state, showing interconvertibility between the dark state and its photoproduct. Therefore, parapinopsin has two reveal color opponency from a single kind of bistable opsin estab- stable “color states” with their absorption maxima at largely lishing an equilibrium-like mixture of the two states with different separated wavelengths, unlike other vertebrate cone opsins signaling abilities whose fractional concentrations are defined by whose photoproduct is unstable (10, 11). However, it remains the spectral composition of incident light. As vertebrate visual color opsins evolved from a bistable opsin, these findings suggest that color opponency involving a single kind of bistable opsin might have Significance been a prototype of vertebrate color opponency. Color discrimination in animals is considered to require oppo- bistable opsin | pineal organ | color opponency | UV-sensitive opsin | nent processing of signals from two or more opsins sensitive to molecular evolution different parts of the spectrum. We previously reported that lower vertebrate pineal organs, which discriminate between UV and visible light, employ a bistable opsin called parapinopsin olor discrimination is achieved through antagonistic responses that has two stable photointerconvertible states, a signaling- to different wavelengths of light, known as color opponency. C inactive state maximally sensitive to UV and a visible light- Such color opponency has been described in many animal spe- sensitive signaling-active photoproduct. Here, we present evi- cies and has previously been considered to require antagonistic dence that the photoequilibrium between these two states in interaction between signals produced by two or more opsins each the zebrafish pineal organ is dependent on the spectral compo- sensitive to a different portion of the spectrum. The best-known sition of incident light, setting the opsin’s signaling activity instance of color opponency originates in the retina and is well according to color to allow parapinopsin alone to generate color known as color vision. Thus, in humans, trichromatic vision ari- opponency between UV and visible light at the level of single ses from red-, green-, and blue-sensitive cone opsins, which are pineal photoreceptor cells. exclusively present in different photoreceptor cells (1–4). These

photoreceptor cells transmit light information to the brain via a Author contributions: S.W., M.K., and A.T. designed research; S.W. and B.S. performed complex neural network involving neurons that subtract outputs research; E.K.-Y., T.N., M.H., and S.T. contributed new reagents/analytic tools; and S.W., from different photoreceptor cells to generate color opponency. M.K., and A.T. wrote the paper. Such operations enable humans to recognize an enormous The authors declare no conflict of interest. number of colors. Many vertebrates also have extraocular pho- This article is a PNAS Direct Submission. toreception, and color opponency in these systems has also been This open access article is distributed under Creative Commons Attribution-NonCommercial- reported in the context of sensing dawn/dusk (5, 6). Mechanisms NoDerivatives License 4.0 (CC BY-NC-ND). for such extraocular color opponency can be quite different from 1To whom correspondence should be addressed. Email: [email protected]. those of color vision. In lizard parietal eyes, blue vs. green dis- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. crimination is achieved at the level of single cells by comparing 1073/pnas.1802592115/-/DCSupplemental. the activity of two opsins, blue-sensitive pinopsin and green- Published online October 15, 2018.

11310–11315 | PNAS | October 30, 2018 | vol. 115 | no. 44 www.pnas.org/cgi/doi/10.1073/pnas.1802592115 Downloaded by guest on September 23, 2021 unclear how the stable photoproduct of parapinopsin is involved A C 405 588 in the pineal color opponency. PP upstream ROI 1 䢢ROI 4 Bistable opsins are also employed in visual cells of many in- ~5.3kbp SV40-pA ROI 2 䢢ROI 5 vertebrates (12). Interestingly, under specific light conditions, the ROI 3 ROI 6 bistability can give rise to a form of “color opponent” behavior. GCaMP6s Drosophila B Thus, for example, visual cells contain a blue-sensitive 7 rhodopsin whose photoproduct shows green/yellow sensitivity. The 5 ROI 7 0.2 ΔF/F blue-sensitive opsin is not signaling-active, but its photoproduct is, 2 4 6 Average of 5 s ROIs 1-6 driving depolarization of the photoreceptor cell. In white-eyed 1 3 Drosophila, a bright blue light stimulus generates prolonged depolarization after light-off. This is known as prolonged depolariz- D 0.2 ing afterpotential (PDA), which occurs because the photoproduct 588 405 588 588 405 588 405 405 ΔF/F (signaling active metarhodopsin) out-titrates the arrestin required 5 s to inactivate it. A subsequent long-wavelength (yellow or orange) light stimulus photoreisomerizes signaling-active metarhodopsin to signaling-inactive rhodopsin and results in suppression of the PDA, thereby generating an effective “hyperpolarization” (13). The op- Fig. 1. Color opponency in the calcium response of PP cells. (A) Construct used to express GCaMP6s in PP cells. (B) Fluorescence image of GCaMP6s in posite effects of blue and green/yellow light on photoreceptor – polarization in this paradigm reveal the potential for bistable op- PP cells by two-photon excitation. ROIs 1 7 were used to calculate change rates in fluorescence intensity in C. (Scale bar: 20 μm.) (C) Color opponency in sins to support a sort of color opponency but this does not translate the calcium response of each ROI containing a part of single PP cells (ROIs 1– to more physiological lighting conditions. 6) or all PP cells (ROI 7) to 405-nm (violet line) and 588-nm (yellow line) light Recently, we found that parapinopsin in the zebrafish pineal stimuli. The averaged profile of ROIs 1–6 clearly shows color opponency. organ exhibits a bistable nature identical to that observed for Error bars indicate SE. (D) Repeated antagonistic response to 405- and 588- lamprey parapinopsin (SI Appendix, Fig. S1). We also obtained the nm light stimuli. The light intensities of 405- and 588-nm light stimuli in C promoter sequence of zebrafish parapinopsin (14). Here, we in- and D were ∼3.2 × 1014 and ∼5.4 × 1017 photons per cm2·s, respectively. The vestigated the light response of parapinopsin-expressing cells (PP durations of both stimuli were ∼450 ms. ΔF/F values are change rates of cells) by using genetic procedures. We found that parapinopsin normalized fluorescence intensity with the averaged intensity of 10 points alone can support color opponency at the single photoreceptor before initial light stimuli. level. Color opponency in the zebrafish pineal organ has been considered to have little relevance to light regulation of pineal melatonin secretion (14, 15), and there is no evidence directly type) and Go-mediated signal transduction cascades, leading to demonstrating its involvement in light entrainment of the circadian hyperpolarization and depolarization, respectively (7). Moreover, rhythms in zebrafish. Here, we also discuss physiological relevance we previously found the coexpression of parapinopsin with pari- of the zebrafish pineal wavelength discrimination on the basis of etopsin in photoreceptor cells of the parietal eye in iguana (18). light conditions that enable parapinopsin-based color opponency. These findings, together with the existence of a parietopsin gene in the zebrafish genome, encouraged us to investigate whether Results parietopsin is expressed in PP cells to mediate visible light sensi- Antagonistic Chromatic Responses in a Single Pineal Photoreceptor tivity in color opponency, in combination with the UV-sensitive Cell. To investigate responses to light in PP cells, we performed parapinopsin. RT-PCR and in situ hybridization analyses revealed calcium imaging in transgenic zebrafish PP cells expressing the that parietopsin was completely coexpressed with parapinopsin in calcium indicator GCaMP6s (16) by using a two-photon excitation the pineal organ (Fig. 2 A and B and SI Appendix,Fig.S3A). Like microscope at 930 nm (Fig. 1 A and B). The excitation light for lizard and Xenopus parietopsins, zebrafish parietopsin is a green- GCaMP6s is expected itself to activate photoreceptors and can sensitive opsin with an absorption maximum at ∼520 nm (7, 18, therefore be considered equivalent to a continuous blue light 19) (Fig. 2C). This observation and the localization of Go in PP stimulation (∼465 nm) throughout the experiment. After approxi- cells (SI Appendix,Fig.S3B) suggested that parapinopsin and mately 120 s of imaging, fluorescence reached a plateau (SI Ap- parietopsin mediate color opponency of the pineal single photo- pendix,Fig.S2), allowing us to investigate the calcium level changes receptor cell through a mechanism similar to that in parietal eye in PP cells upon additional stimulation with different wavelengths photoreceptor cells. oflight.Ineachregionofinterest(ROI;ROIs1–6inFig.1B) To investigate the contributions of the two opsins to the color containing part of a single PP cell, stimulation with 405-nm light opponency observed by calcium imaging, we established para- causedadecreaseinfluorescenceintensity. This response is con- pinopsin- and/or parietopsin-KO fish by CRISPR/Cas9-mediated sistent with parapinopsins’ known ability to show light-dependent genome editing (SI Appendix,Fig.S4). Surprisingly, fish lacking −/− interactionwithtransducin,whichisexpectedtoleadtohyperpo- parietopsin (PT ) still showed an antagonistic response to 405- and 588-nm light stimuli, similar to the response of the WT (Fig. 2 larization of the PP cell and closure of voltage-sensitive calcium − − channels (17). In contrast, stimulation with 588-nm light elevated D and E). When the parapinopsin gene was disrupted (PP / ), the the fluorescence intensity, demonstrating the antagonistic effect of responses to 405- and 588-nm light stimuli largely disappeared (Fig. 2F and SI Appendix,Fig.S5). Double-KO fish showed a similar UV and visible light on the calcium in individual PP cells to UV − − − − and visible light (Fig. 1C). The fluorescence changes between the response (Fig. 2G,PP / /PT / ). Slight residual calcium increases – upon 588-nm light stimuli were observed in parapinopsin-deficient average of ROIs 1 6 and ROI 7 encompassing the whole pineal − − − − − − organ were almost identical, suggesting that the color opponency PP cells (PP / and PP / /PT / ) but were nearly negligible com- observed in the whole pineal organ was based on the antagonistic pared with the increase in PP cells containing parapinopsin. This chromatic response of a single PP cell. Such antagonistic responses suggests that, under the imaging conditions, parapinopsin alone NEUROSCIENCE were reproducibly observed by repeated stimulation with UV and mediated color opponency of the pineal photoreceptor cells, irre- visible light (Fig. 1D). This represents color opponency in a single spective of the presence of parietopsin, enabling us to focus on the photoreceptor cell of the teleost pineal organ. bistable nature of parapinopsin in further investigations. To investigate the contribution of bistability, we generated Color Opponency Depending on Molecular Property of Parapinopsin. transgenic fish expressing SWS1 opsin, a UV-sensitive cone op- Color opponency in a single photoreceptor cell has previously sin that shares parapinopsin’s transducin signaling cascade but is − − been reported in the parietal eyes of the side-blotched lizard (5). not bistable (17, 20), in PP cells (SWS1/PP / ). In the SWS1/ − − In that photoreceptor cell, two different opsins, blue-sensitive PP / mutant, SWS1-expressing PP cells clearly exhibited de- pinopsin and green-sensitive parietopsin, drive gustducin- (Gt- crease of calcium levels in response to 405-nm light stimulation

Wada et al. PNAS | October 30, 2018 | vol. 115 | no. 44 | 11311 Downloaded by guest on September 23, 2021 To test this hypothesis, we analyzed calcium level changes A 405 588 Eye Brain Pineal under 860-, 930- and 1,000-nm two-photon excitations, which are Parietopsin 0.2 ∼ Parapinopsin ΔF/F considered equivalent to violet, blue, and green light ( 430, β-actin 5 s ∼465, and ∼500 nm) excitations, respectively, and are therefore B D expected to generate a different “basal photoequilibrium.” When WT (n=41) imaging was performed with two-photon excitation with 860-nm Parietopsin light (i.e., “violet”), the 588-nm light stimulus increased the d E calcium levels, whereas the 405-nm light stimulus did not sig- r PT−/−(n=28) nificantly decrease calcium levels (Fig. 3E). In contrast, in two- “ ” Parapinopsin photon excitation with 1,000-nm light (i.e., green ), the 405-nm F light stimulus decreased the calcium level, whereas the 588-nm F PP−/−(n=40) light stimulus did not evoke an obvious increase (Fig. 3 ). These Merge G results, together with the response profile of two-photon excitation D −/− −/− at 930 nm (Fig. 3 ), strongly support that the photoequilibrium of C 1.0 PP /PT (n=27) ~520 parapinopsin is fundamentally important to generate color oppo- 0.0 H nency to light stimuli. -1.0 Parietopsin SWS1/PP−/− (n=27) We next confirmed whether color opponency based on a single

Rel. Diff. Abs. 400 500 600 parapinopsin is generated under environmental light conditions. Wavelength (nm) In vitro spectroscopic analysis revealed that, under white light with spectral distribution similar to that of sunlight in the early after- Fig. 2. Contribution of opsins to color opponency in PP cells under two- photon imaging. (A) RT-PCR analyses of parietopsin and parapinopsin ex- noon, zebrafish parapinopsin formed a photoequilibrium similar pression in the eye, brain, and pineal organ of zebrafish. (B) In situ hybrid- to that under the sunlight conditions wherein the ratio of the dark ization analyses of two opsins in the zebrafish pineal organ using the double state to the photoproduct was ∼4:1 (SI Appendix, Fig. S1, curve 4). fluorescence method. Yellow arrowheads in B indicate coexpression of two The establishment of such a photoequilibrium allows the possi- opsins in the zebrafish pineal organ. Orientations marked with “d” and “r” bility that PP cells exhibit color opponency based on parapinopsin indicate the dorsal and rostral sides, respectively. The white dotted traces alone under sunlight. We investigated the calcium level changes indicate the landmarks of the pineal organ. (Scale bars: 100 μm.) (C)Relative difference absorption spectrum of zebrafish parietopsin before minus after light irradiation. The spectrum is shown as an average of three measurements. (D–H) Calcium level changes upon 405- or 588-nm light stimuli in PP cells of −/− −/− A Under 930 nm BC405 stimulus 588 stimulus WT (D; n = 41), parietopsin-KO (E;PT , n = 28), parapinopsin-KO (F;PP , (“blue light condition”) − − − − = / / = Dark state Photoproduct n 40), double-KO (G;PP /PT , n 27), and SWS1 opsin-expressing/para- Shift pinopsin-KO fish (H; SWS1/PP−/−, n = 27). Error bars indicate SE. The light intensities of 405- and 588-nm light stimuli in D–H were ∼3.2 × 1014 and ∼5.4 × 1017 photons per cm2·s, respectively. The durations of both stimuli were ∼450 ms. Activation Shift ΔF/F values are change rates of normalized fluorescence intensity with the av- of Gt2 0.2ΔF/F eraged intensity of 10 points before initial light stimuli. Statistical evaluation of 405 588 the differences in amplitudes of calcium level changes among the WT and D 5 s mutants are shown in SI Appendix,Fig.S5A and B. “blue light condition” WT (n=26) Shift

Shift but did not significantly respond to 588-nm light (Fig. 2H), Dark state Photoproduct suggesting that the bistable nature of parapinopsin is responsible E for the visible light-dependent calcium increase. 0.5ΔF/F “violet light condition” 5 s WT (n=13) Color Opponency Based on Photoequilibrium of Parapinopsin Two F States. How the bistable nature underlies the antagonistic chro- “green light condition” 0.2ΔF/F WT (n=13) matic response in PP cells under imaging light conditions can be 5 s speculated as follows. As 930-nm light in two-photon excitation is considered to serve as blue light (∼465 nm) in one-photon excita- Fig. 3. Color opponency involving shift of photoequilibrium between the dark state and photoproduct of parapinopsin. (A–C) Schematic models of tion, the imaging conditions were considered to form a “photo- ” “ ” photoequilibrium between the dark state and photoproduct of parapinopsin equilibrium or semiphotoequilibrium (i.e., photoequilibrium-like and changes in the amount of its photoproduct upon different light stimuli mixture) of the dark state and photoproduct of parapinopsin (Fig. under two-photon imaging. Circles filled with purple and green indicate A – 3 ). For 10 20 s after starting imaging, fluorescence intensity de- molecules of the dark state and photoproduct of parapinopsin, respectively. creased because of hyperpolarization of the PP cells, which dem- Photoequilibrium of parapinopsin under two-photon excitation with 930-nm onstrates accumulation of the parapinopsin photoproduct under light (i.e., blue light) is shown with black arrows (A). The shifts in photo- excitation light (SI Appendix,Fig.S2). The 405-nm light stimulus equilibrium caused by stimuli of 405- and 588-nm light are shown as purple (B) increases the short-wavelength component of incident light and and orange (C) arrows, respectively. Red arrows indicate the amount of cone thus increases the fractional concentration of the signaling active transducin (Gt2) activated by the photoproducts, which is equivalent to the photoproduct from that produced by the imaging light alone, hyperpolarization level. Note that the circles and the arrows schematically leading to hyperpolarization (Fig. 3 B and D, calcium decrease). illustrated in A–D are not quantitatively drawn. The recovery of photo- Conversely, because the photoproduct is more sensitive to longer equilibrium to the original state after stimuli turn off under “blue light” condition could account for the antagonistic calcium response under two- wavelengths, the 588-nm light stimulus decreases its fractional – concentration, reducing the pool of signaling-active opsin, resulting photon imaging conditions. (D F) Calcium level changes upon 405- and 588- nm light stimuli in WT PP cells under two-photon excitation with 930-nm (D; in termination of hyperpolarization and increased calcium levels = = = C D n 26), 860-nm (E; n 13), and 1,000-nm (F; n 13) lights (i.e., blue, violet, (Fig. 3 and , calcium increase). After the stimuli are turned off, and green lights, respectively). Schematic models for shifts in photo- the photoproduct amount is considered to revert to the original equilibrium by light stimuli are also shown in D. Error bars indicate SE. The photoequilibrium level under “blue light” conditions (Fig. 3D). light intensities of 405- and 588-nm light stimuli in D–F were ∼3.2 × 1014 and Therefore, the 405-nm and 588-nm light stimuli may transiently ∼5.4 × 1017 photons per cm2·s, respectively. The durations of both stimuli were decrease and increase calcium levels by changing the fractional ∼450 ms. ΔF/F values are change rates of normalized fluorescence intensity concentration of photoproduct from the photoequilibrium level. with the averaged intensity of 10 points before initial light stimuli.

11312 | www.pnas.org/cgi/doi/10.1073/pnas.1802592115 Wada et al. Downloaded by guest on September 23, 2021 in photoreceptor cells at 10-s intervals to reduce the effect of the A two-photon imaging light, under continuous irradiation by white 4 3

s) ∼ · light with intensity 1% of that found in a sunny location in the 2 early afternoon and with a similar spectral distribution (Fig. 4A). 2 −/− In WT and PT , a decrease followed by a steady state of cal- 12 White light (Xenon) 10 cium level was observed when the white light was turned on (SI UV light-mixed white light (Xenon and LED) Appendix “ ” 7 , Fig. S6), similar to the case under imaging blue 6 Visible light-mixed white light (Xenon and LED) −/− −/− −/− lights (SI Appendix, Fig. S2). In contrast, PP and PP /PT 5 1% intensity of direct and indirect sunlight 4 did not show large decreases in calcium levels in response to 3 white light (SI Appendix, Fig. S6). After the calcium level Intensity (photon/cm 400 450 500 550 600 650 Wavelength (nm) reached a plateau (∼10 min), PP cells were exposed to “mixed white light” with UV or visible LED light, which has a broad B spectral distribution (UV light- or visible light-mixed white light; WT (n=24) Fig. 4A), to investigate whether color opponency based on parapinopsin is observed under natural light-like conditions. We found decreased and increased calcium levels following exposure PT–/– (n=23) to UV and visible light-mixed white light, respectively, in WT and − − − − − − − − PT / but not in PP / and PP / /PT / (Fig. 4B). Changes in 0.2ΔF/F calcium level were also observed in PP cells following exposure –/– 30 s to different intensities and durations of UV and visible light- PP (n=17) SI Appendix A–C mixed white lights ( ,Fig.S7 ). These observa- PP–/– /PT–/– (n=16) tions suggest that parapinopsin alone can allow pineal photoreceptors to respond to changes in the spectral distribution of incident light (i.e., color opponency) under physiological Fig. 4. White light spectra and color opponency of PP cells under white sunlight conditions. light conditions. (A) Spectra of the xenon white light (solid black trace), UV LED light-mixed white light (broken magenta trace), and visible LED light- Discussion mixed white light (broken green trace) are compared with those of 1% in- In this study, we describe a chromatically antagonistic response tensity of direct sunlight (solid gray trace) measured at 2:30 PM on April 16, of calcium in a single zebrafish pineal cell produced by a single 2018, in Osaka, Japan. The light intensities (360–780 nm) of white light, UV- kind of bistable opsin, parapinopsin. The mechanism relies upon mixed white light, visible light-mixed white light, and direct sunlight were ’ ∼9.2 × 1014, ∼9.3 × 1014, ∼9.9 × 1014, and 1.1 × 1017 photons per cm2·s. The parapinopsin s ability to form two photointerconvertible opsin ∼ × 13 ∼ × 13 states with different spectral sensitivity and signaling capacity. light intensities of UV and visible LED lights were 1.3 10 and 7.6 10 photons per cm2·s, respectively. (B) Changes in calcium level of PP cells in WT Under extended light exposure, an equilibrium between these − − − − (red open circle; n = 24), PT / fish (green closed circle; n = 23), PP / fish − − − − two states is formed in which the fractional concentration of the (purple open triangle; n = 17), and PP / /PT / fish (black closed triangle; n = signaling active state is dependent on the spectral composition of 16) under different light conditions shown in A. The shaded traces indicate incident light, with lights biased toward UV driving greater hy- calcium level changes of individual fish. PP cells of WT and mutants were perpolarization than those biased toward visible light. exposed to the white light (Top, white bars), UV light-mixed white light A previous study revealed a correlation of photoresponses (hatched magenta bars), and visible light-mixed white light (hatched green between calcium dynamics and electrophysiological outputs in bars). Note that the calcium levels decreased and increased transiently in photoreceptor cells (i.e., cones) of the zebrafish retina (21). response to exposure to UV and visible light-mixed white lights, respectively, − − Therefore, the antagonistic chromatic changes in calcium levels in WT and PT / . Error bars indicate SE. The durations of both stimuli were measured in this study qualitatively suggest electrophysiological ∼9.4 s. ΔF/F values are change rates of normalized fluorescence intensity outputs of color opponency in the PP cells of zebrafish. Para- with the averaged intensity of 10 points before UV-mixed white light ex- posures. Note that the relaxation of responses to UV and green LED light in pinopsin is found in a wide variety of lower vertebrates (9, 10, 14, − − 18), and its bistable nature might contribute to a single kind of WT and PT / were not observed as a result of time-lapse recoding with low opsin-based color opponency in pineal-related organs other than time resolution. the teleost pineal organ. Photoreceptor cells in many invertebrates such as insects, The duration of light exposure for maximum changes in in- crustaceans, and cephalopods employ bistable opsins (12). Pho- – SI Appendix toreceptor cells of white-eyed Drosophila show “opponent” re- crease of calcium level in PP cells (10 20 s; , Fig. S7), sponses to blue and orange light under specific artificial which is considered to be correlated to accumulation of the photoproduct, is shorter than the duration (more than 100 s) experimental conditions (13). Following intense blue light irra- ∼ diation, fly photoreceptors generate a depolarizing PDA that estimated to be required for 99% level of the photoequilibrium SI Appendix Supplementary Information Text persists in the dark as a result of insufficient shut-off of light- formation ( , ). To activated rhodopsin; orange light then causes “opponent” re- address this inconsistency, we roughly estimated a time evolution sponses (i.e., hyperpolarizations) because of the reverse photo- of the signaling active photoproduct after exposure to the white reaction of the accumulated “signaling-active metarhodopsin” light, based on the very simplified reaction scheme of the dark (arrestin-free) to the inactive dark-state rhodopsin (13). Be- state and photoproduct of parapinopsin according to that of cause, similarly to the mechanism of PDA, one might speculate bistable Drosophila rhodopsin and metarhodopsin (13) (SI Ap- that PP cells require abundant “active” or “not-inactivated” pendix, Supplementary Information Text and Fig. S9). The esti- photoproducts to generate the reverse responses of photore- mated profile suggests that the change in signaling active ceptor cells (i.e., depolarization), it is important to consider the photoproduct amount in PP cells reaches maximum at approxi- NEUROSCIENCE activity of biochemical deactivation mechanisms in this system. mately 20 s under exposure of PP cells to different white light (SI We therefore investigated the recovery kinetics of light response Appendix, Fig. S9C). The duration is similar to that for maximum via shut-off of light-activated parapinopsin after turning off the changes in calcium level in PP cells under exposure to white light white lights (i.e., in the dark). Interestingly, calcium levels re- (SI Appendix, Fig. S7C). The profiles estimated with different covered to that of the dark level by ∼20 s after turning off the values for reaction parameters suggest that the dark-inactivation white light (SI Appendix, Fig. S8), showing no prolongation of the of active photoproducts of parapinopsin under white light might light response, as in a typical PDA, which continues for hours in contribute to accumulation of active photoproducts (SI Appen- the dark. Therefore, it is of interest to discuss how the active dix, Fig. S9 C–E). Because the estimation is based on a very photoproduct accumulates under white light. simplified model, investigation of a detailed mechanism(s)

Wada et al. PNAS | October 30, 2018 | vol. 115 | no. 44 | 11313 Downloaded by guest on September 23, 2021 involved in the dark inactivation of the signaling-active photo- opponency would provide a clue to the physiological function of products, such as phosphorylation of and arrestin binding to parapinopsin-based color opponency as well as how putative parapinopsin photoproducts and their kinetics is important to parapinopsin-based and parapinopsin/parietopsin-based oppo- understand how a certain amount of the active photoproduct is nent mechanisms are used for the color opponency of PP cells, kept in PP cells under white light (SI Appendix, Supplementary depending on different situations. Information Text). In addition, this issue is also important to It has been suggested that the ancestral animal opsin had a investigate if activities of proteins, such as arrestin (22), are bistable nature (26). Therefore, a single kind of bistable opsin- regulated when the photoproducts revert to the dark state upon based color opponency in a single photoreceptor cell might have absorption of visible light. been a prototype of color opponency. As shown in Fig. 3, the We also observed slow elevation of calcium levels after these shift of the quantity ratio of signaling-inactive and -active (“dark” levels were greatly decreased under continuous blue light irra- and “photoproduct,” respectively) states corresponding to the diation (i.e., two-photon excitation) or xenon white light (SI wavelengths of light is a key to generating color opponency. Appendix, Figs. S2 and S6). Such slow elevation of calcium may Bistable opsins in which dark and light states have largely be related to the kinetics of signaling-active photoproduct (SI overlapping absorption spectra do not clearly exhibit their pho- Appendix, Fig. S9C), which could constitute a mechanism of light toequilibrium shift depending on the wavelengths of light, and adaptation. As calcium itself affects electrophysiological re- therefore such a bistable opsin is not suitable for a single kind of sponses by regulating several signal transduction proteins (23, opsin-based color opponency. In other words, the separation of 24), it is important to investigate such parapinopsin-based light spectral sensitivities between dark and light states could be − − adaptation by comparing WT and SWS1/PP / zebrafish. critical to detect the color information. Accordingly, the color We found coexpression of parapinopsin and green-sensitive opponency with a single kind of bistable opsin could have parietopsin, which underlies depolarization in the lizard parietal simply developed from a brightness detection system with a eye, in the photoreceptor cells of the zebrafish pineal organ. single kind of bistable opsin through molecular evolution of the Although parapinopsin-deficient PP cells exhibited a slight re- opsin, that is, amino acid mutations resulting in a bistable opsin sidual increase in calcium levels, suggesting depolarization (Fig. having two spectroscopically separated states, like para- − − − − − − 2), the residual responses in PP / and PP / /PT / were similar pinopsin. Taken together, such a single kind of bistable opsin- in amplitude (SI Appendix, Fig. S5B), suggesting that the residual based color opponency could be the first step for developing increase in the calcium level did not involve a large contribution varied mechanisms of color opponency including color vision by parietopsin. Additionally, following exposure to white light, with multiple opsins. no significant difference in calcium changes was observed be- − − − − − − tween PP / and PP / /PT / (SI Appendix, Fig. S6). Taken to- Materials and Methods gether, these observations suggest a limited contribution of Animals. Zebrafish (Danio rerio) were obtained from the Zebrafish In- parietopsin in regulating calcium levels under the white light ternational Resource Center and National BioResource Project Zebrafish. conditions used in this study. Such little involvement of parie- Zebrafish were maintained on 14-h light/10-h dark cycles at 28.5 °C. Embryos topsin under these light conditions may be explained by its mo- (0–7 d postfertilization) were maintained in E3 medium (5 mM NaCl, lecular property as a bleaching opsin (19). Under bright light 0.17 mM KCl, 0.33 mM CaCl2, and 0.33 mM MgSO4). The animal experiment conditions such as those in early to late afternoon (Fig. 4A and SI procedures were approved by the Osaka City University Animal Experiment Appendix, Fig. S10A), the amount of functional parietopsin de- Committee (no. S0032) and complied with the regulations on animal ex- creases because the photoproduct releases its retinal chromo- periments from Osaka City University. phore and loses its function after light absorption. In this ∼ circumstance, it remains possible that parietopsin contributes Generation of Transgenic Fish. The 5 kb upstream sequence of zebrafish to color opponency by activating Go under dim light conditions parapinopsin (PP1; DDBJ/EMBL/GenBank accession no. AB626966) was previously isolated and inserted into the Tol2 vector pT2AL200R150G (14, 27), (e.g., dusk and dawn), as suggested for the lizard parietal eye (5, 7). which contains the EGFP expression cassette. To introduce GCaMP6s into PP In contrast to the two-opsin system involving parietopsin, the cells, the EGFP sequence was replaced with the GCaMP6s sequence obtained chromatic mechanism involving parapinopsin alone in the from pGP-CMV-GCaMP6s (no. 40753; Addgene). To introduce SWS1 opsin zebrafish pineal organ may not be suitable for detecting dawn 2 (DDBJ/EMBL/GenBank accession no. D85863) into zebrafish PP cells, the and dusk. Light intensity at these times [e.g., irradiance of 0.1 W/m SWS1-P2A-mCherry sequence was used instead of the GCaMP6s sequence. or less (25), ∼3% brightness of the white light used in Fig. 4] is Plasmid DNA (∼25 pg) and Tol2 mRNA (∼25 pg) were coinjected into em- not sufficiently strong to form a photoequilibrium between the bryos during the one-cell stage. two states, which is essential for generating color opponency via this mechanism. The intensity and spectral distribution of natural Generation of Gene-KO Fish. Guide RNAs (SI Appendix, Fig. S4) were gener- sunlight varies depending on numerous factors, e.g., the time ated as described previously (28). The plasmids pT7-gRNA and pT3TS-nCas9n of day (e.g., early and late afternoon) and local conditions (e.g., were obtained from Addgene (nos. 46759 and 46757). The oligonucleotide sun or shade). Light intensity in the late afternoon is considered sets designed for guide RNAs were as follows: 5′-TAGGGTGGCGGTGTC- to be bright enough for parapinopsin-based color opponency (SI CTGGGTC-3′ and 5′-AAACGACCCAGGACACCGCCAC-3′ for zebrafish para- Appendix, Fig. S10A). Interestingly, the spectral distribution of pinopsin guide RNA and 5′-GGCGTATGAGCGCTATAACG-3′ and 5′-AAAC- light largely differs between the sunny and shady locations in the CGTTATAGCGCTCATACG-3′ for zebrafish parietopsin guide RNA. Annealed late afternoon; late-afternoon sunlight in a sunny location con- oligonucleotide sets were inserted into pT7-gRNA, which had been linearized tains abundant longer-wavelength components whereas shorter- by BsmB I. Guide RNAs were transcribed by using the MEGAscript T7 Tran- wavelength components are relatively increased in the shade. scription Kit. Cas9 mRNA was transcribed in vitro by using the mMESSAGE mMACHINE T3 Transcription Kit (Ambion). Guide RNA (∼50 pg) and Cas9 The photoequilibria of parapinopsin formed under sunny and ∼ shade sunlight conditions were different; the relative amounts of mRNA ( 150 pg) were coinjected into embryos during the one-cell stage. Mutations in the zebrafish parapinopsin and parietopsin nucleotide se- photoproduct in the photoequilibrium under the sunny and quences were confirmed by a heteroduplex mobility assay (29). Stop codons shady location conditions were ∼8% and ∼25%, respectively (SI −/− −/− Appendix B appear after Val224 and Ser135 in the PP and PT mutants, respectively. , Fig. S10 ). Therefore, such changes in wavelength Opsin(s)-deficient fish were obtained by mating the established KO strains. components between sunny and shady locations in the late afternoon may be detected by the mechanism involving para- Two-Photon Imaging. Zebrafish larvae (7 d postfertilization) were mounted pinopsin alone because late afternoon sunlight in the shade is dorsal side up in low-melting agarose gel (1.5% in E3 medium) on 35-mm approximately fivefold brighter than the white light used in this glass-bottom dishes (Iwaki). To prevent drying and immobilize the larvae, study, in which parapinopsin-based color opponency was gener- E3 medium containing 0.002% Tricaine (MS222; Sigma) was added to the ated (Fig. 4 A and B and SI Appendix, Fig. S10). Determining the dishes. Two-photon calcium imaging of transgenic fish was performed by intensity of white light that enables PP cells to generate color using a multiphoton laser scanning microscope (FVMPE-RS; Olympus). The

11314 | www.pnas.org/cgi/doi/10.1073/pnas.1802592115 Wada et al. Downloaded by guest on September 23, 2021 Mai Tai HP DeepSee IR laser (Spectra-Physics) was used for two-photon ex- and fluorescein RNA labeling kit (Roche), respectively. Sections were pre- citation of GCaMP6s in PP cells. The wavelength and intensity of the IR laser treated with proteinase K and hybridized with each RNA probe in ULTRAhyb were controlled by using FLUOVIEW software (Olympus). The laser intensities Ultrasensitive Hybridization Buffer (Ambion). For double fluorescence la- at 860, 930, and 1,000 nm were 2.37, 1.49, and 0.74 W, and 2.5%, 4%, and 8% beling, sections hybridized with DIG-labeled probes were incubated with of each were used as experimental conditions, respectively. A Galvano HRP-conjugated anti-DIG antibody (Roche) and subsequently treated with scanner was used to acquire images for all procedures (∼450 ms in a 160 × the TSA plus DNP (HRP) system (Perkin-Elmer), followed by incubation 160-pixel image). Light stimulation using lasers was conducted by scanning with Alexa 488-conjugated anti-DNP antibody. Fluorescein-labeled probes with 405-nm (OBIS; Coherent) and 588-nm (Sapphire; Coherent) lasers with on the sections were detected by incubation with alkaline phosphatase- the same scanning areas as those of IR laser scanning. The light intensities of conjugated anti-fluorescein antibody (Roche) followed by a color reaction ∼ 2 lasers were calculated based on the size of the scanning area ( 0.0064 mm ). using the HNPP Fluorescent Detection Set (Roche). In imaging under two-photon excitation, the intensities of light stimuli using ∼ × 14 ∼ × 17 2· 405- and 588-nm lasers were 3.2 10 and 5.4 10 photons per cm s, Spectroscopic Analyses of Opsin-Based Pigment. Zebrafish parapinopsin and respectively, and the durations were ∼450 ms. Three types of white lights parietopsin were expressed and purified as previously described (30, 31). were used for white light experiments (Fig. 4A); the xenon lamp white light Expressed proteins were reconstituted by adding an excess of 11-cis-retinal, as a standard white light in this paper was mixed with UV (∼365 nm, ∼8.9 × extracted with 1% n-dodecyl β-d-maltoside (DM) in 50 mM Hepes buffer (pH 1012 photons per cm2·s) or visible (∼580 nm, ∼4.7 × 1013 photons per cm2·s) 6.5) containing 140 mM NaCl (buffer A) and purified by using 1D4-agarose LED light (pE-4000; CoolLED). White light was provided through the cold with buffer A containing 0.02% DM. Absorption spectra were measured filter (CLDF-50S; Sigma Koki) from a xenon light source. White-light irradi- with a UV2450 spectrophotometer (Shimadzu) at 0 °C. Full methods are ation was blocked at the time of image acquisition by using a shutter syn- described in SI Appendix, Materials and Methods. chronized with IR laser scanning. An XLPLN 25× WMP2 objective lens (NA, ∼1.05; Olympus) was used for all imaging procedures. The temporal changes in GCaMP6s fluorescence intensities in time-series images were analyzed by Measurement of Natural Light Spectra. A CL-500A illuminance spectropho- using FLUOVIEW. Subtraction of mean fluorescence intensity between each tometer (Konica Minolta) was used for all measurements. Spectrum mea- ROI containing PP cells and the nonfluorescent region outside PP cells (i.e., surement in the early afternoon was carried out at 2:30 PM on April 16, 2018, background) was carried out to calculate net fluorescence intensity. In Figs. in Osaka, Japan. Spectrum measurement of sunny and shady locations in the 1D,2D–H,3D–F, and 4B and SI Appendix, Figs. S2 and S5–S8, the mean late afternoon was performed at 6:00 PM on June 16, 2018, in Osaka, Japan. fluorescence intensity in an ROI containing all PP cells in one fish such as ROI The shade was made by shielding direct sunlight with a metallic case (height 7 in Fig. 1B was used to eliminate arbitrariness in selecting the ROI. The 30 cm × width 40 cm × depth 3 cm). Shade light, as indirect sunlight from the fluorescence changes in time-series images were shown as the rates of sky, was measured at a point 100 mm away from the metal case against the

change ΔF/F0 (normalized with initial point; dark) or ΔF/F (defined in each direction of the sun. The detector was oriented perpendicularly to the sky figure legend). and measured at the ground level (∼5cm).

RT-PCR. Total RNA from the pineal organ, brain, and eye of zebrafish was ACKNOWLEDGMENTS. We thank Robert J. Lucas (University of Manchester) purified with an RNeasy Mini kit (Qiagen) and reverse-transcribed into cDNA for his valuable comments on this manuscript and Robert S. Molday (Univer- by using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). sity of British Columbia) for supplying rho 1D4-producing hybridoma. This β-Actin was used as an internal reference gene. The primer sequences are work was supported by Japanese Ministry of Education, Culture, Sports, Science and Technology Grants-in-Aid for Scientific Research 15H05777 (to shown in SI Appendix, Materials and Methods. A.T.), 16KT0074 and 17H06015 (to M.K.), 14J40094 (to E.K.-Y.), and 13J05782 (to S.W.); Japan Science and Technology Agency (JST) Core Research for In Situ Hybridization. Preparation of RNA probes and in situ hybridization Evolutional Science and Technology (CREST) Grant JPMJCR1753 (to A.T.) and were carried out as previously described (17). Digoxigenin (DIG)- and JST Precursory Research for Embryonic Science and Technology (PRESTO) fluorescein-labeled antisense and sense RNA probes for zebrafish parapinopsin Grant JPMJPR13A2 (to M.K.); and Research Fellowships of the Japan Society and parietopsin mRNAs were synthesized by using the DIG RNA labeling kit for the Promotion of Science for Young Scientists (to E.K.-Y. and S.W.).

1. Nathans J, Thomas D, Hogness DS (1986) Molecular genetics of human color vision: 17. Kawano-Yamashita E, et al. (2015) Activation of transducin by bistable pigment The genes encoding blue, green, and red pigments. Science 232:193–202. parapinopsin in the pineal organ of lower vertebrates. PLoS One 10:e0141280. 2. Schnapf JL, Kraft TW, Baylor DA (1987) Spectral sensitivity of human cone photore- 18. Wada S, Kawano-Yamashita E, Koyanagi M, Terakita A (2012) Expression of UV- ceptors. Nature 325:439–441. sensitive parapinopsin in the iguana parietal eyes and its implication in UV- 3. Merbs SL, Nathans J (1992) Absorption spectra of human cone pigments. Nature 356: sensitivity in vertebrate pineal-related organs. PLoS One 7:e39003. 433–435. 19. Sakai K, et al. (2012) Photochemical nature of parietopsin. Biochemistry 51: 4. Dacey DM, Lee BB (1994) The ‘blue-on’ opponent pathway in primate retina origi- 1933–1941. – nates from a distinct bistratified ganglion cell type. Nature 367:731 735. 20. Lamb TD (2013) Evolution of phototransduction, vertebrate photoreceptors and 5. Solessio E, Engbretson GA (1993) Antagonistic chromatic mechanisms in photore- retina. Prog Retin Eye Res 36:52–119. – ceptors of the parietal eye of lizards. Nature 364:442 445. 21. Leung YT, Fain GL, Matthews HR (2007) Simultaneous measurement of current and 6. Uchida K, Morita Y (1994) Spectral sensitivity and mechanism of interaction between calcium in the ultraviolet-sensitive cones of zebrafish. J Physiol 579:15–27. inhibitory and excitatory responses of photosensory pineal neurons. Pflugers Arch 22. Eichel K, Jullié D, von Zastrow M (2016) β-arrestin drives MAP kinase signalling from – 427:373 377. clathrin-coated structures after GPCR dissociation. Nat Cell Biol 18:303–310. 7. Su CY, et al. (2006) Parietal-eye phototransduction components and their potential 23. Pugh EN, Jr, Nikonov S, Lamb TD (1999) Molecular mechanisms of vertebrate pho- evolutionary implications. Science 311:1617–1621. toreceptor light adaptation. Curr Opin Neurobiol 9:410–418. 8. Dodt E (1962) [Comparative studies on the adaptive behavior of the pure cone retina. 24. Yau KW (1991) Calcium and light adaptation in retinal photoreceptors. Curr Opin Citellus citellus, Sciurus vulgaris]. Pflugers Arch Gesamte Physiol Menschen Tiere 275: Neurobiol 1:252–257. 561–573. German. 25. Spitschan M, Lucas RJ, Brown TM (2017) Chromatic clocks: Color opponency in non- 9. Blackshaw S, Snyder SH (1997) Parapinopsin, a novel catfish opsin localized to the image-forming visual function. Neurosci Biobehav Rev 78:24–33. parapineal organ, defines a new gene family. J Neurosci 17:8083–8092. 26. Koyanagi M, Kawano-Yamashita E, Wada S, Terakita A (2017) Vertebrate bistable 10. Koyanagi M, et al. (2004) Bistable UV pigment in the lamprey pineal. Proc Natl Acad pigment parapinopsin: Implications for emergence of visual signaling and neo- Sci USA 101:6687–6691. 11. Terakita A, et al. (2004) Counterion displacement in the molecular evolution of the functionalization of non-visual pigment. Front Ecol Evol 5:23. rhodopsin family. Nat Struct Mol Biol 11:284–289, and erratum (2004) 11:384. 27. Urasaki A, Asakawa K, Kawakami K (2008) Efficient transposition of the Tol2 trans- NEUROSCIENCE 12. Koyanagi M, Terakita A (2008) Gq-coupled rhodopsin subfamily composed of in- posable element from a single-copy donor in zebrafish. Proc Natl Acad Sci USA 105: – vertebrate visual pigment and melanopsin. Photochem Photobiol 84:1024–1030. 19827 19832. 13. Stavenga DG, Hardie RC (2011) Metarhodopsin control by arrestin, light-filtering 28. Jao LE, Wente SR, Chen W (2013) Efficient multiplex biallelic zebrafish genome ed- – screening pigments, and visual pigment turnover in invertebrate microvillar photo- iting using a CRISPR nuclease system. Proc Natl Acad Sci USA 110:13904 13909. receptors. J Comp Physiol A Neuroethol Sens Neural Behav Physiol 197:227–241. 29. Ota S, et al. (2013) Efficient identification of TALEN-mediated genome modifications 14. Koyanagi M, et al. (2015) Diversification of non-visual photopigment parapinopsin in using heteroduplex mobility assays. Genes Cells 18:450–458. spectral sensitivity for diverse pineal functions. BMC Biol 13:73. 30. Koyanagi M, Takada E, Nagata T, Tsukamoto H, Terakita A (2013) Homologs of ver- 15. Ziv L, Tovin A, Strasser D, Gothilf Y (2007) Spectral sensitivity of melatonin suppression tebrate Opn3 potentially serve as a light sensor in nonphotoreceptive tissue. Proc Natl in the zebrafish pineal gland. Exp Eye Res 84:92–99. Acad Sci USA 110:4998–5003. 16. Chen TW, et al. (2013) Ultrasensitive fluorescent proteins for imaging neuronal ac- 31. Sugihara T, Nagata T, Mason B, Koyanagi M, Terakita A (2016) Absorption charac- tivity. Nature 499:295–300. teristics of vertebrate non-visual opsin, Opn3. PLoS One 11:e0161215.

Wada et al. PNAS | October 30, 2018 | vol. 115 | no. 44 | 11315 Downloaded by guest on September 23, 2021