Opn5L1 is a retinal receptor that behaves as a reverse and self-regenerating photoreceptor

Most opsins are G protein-coupled receptors that utilize retinal both as a ligand and as a chromophore. Opsins’ main established mechanism is light-triggered activation through retinal 11-cis-to-all-trans photoisomerization. Here we report a vertebrate non-visual opsin that functions as a Gi-coupled retinal receptor that is deactivated by light and can thermally self-regenerate. This opsin, Opn5L1, binds exclusively to all-trans-retinal. More interestingly, the light-induced deactivation through retinal trans-to-cis isomerization is followed by formation of a covalent adduct between retinal and a nearby cysteine, which breaks the retinal-conjugated double bond system, probably at the C11 position, resulting in thermal re-isomerization to all-trans-retinal. Thus, Opn5L1 acts as a reverse photoreceptor. We conclude that, like vertebrate rhodopsin, Opn5L1 is a unidirectional optical switch optimized from an ancestral bidirectional optical switch, such as invertebrate rhodopsin, to increase the S/N ratio of the signal transduction, although the direction of optimization is opposite to that of vertebrate rhodopsin.

The N-and C-termini modified Opn5L1 opsin employed here is also shown to bear the very same properties of the wild protein, thus ruling out possible interferences due to those modifications.
Finally, let me fairly stress that I am not an expert in the experimental methods/protocols employed in this study and this report is not intended to disregard major issues that may be possibly raised by experimental experts in this field and that should be taken into high consideration.
That said, I warmly recommend publication of this work on NatureComm. This study eventually provides a new paradigm for activation-deactivation of vertebrate rhodopsins, showing an alternative (reverse!) evolutionary route for the design of unidirectional photoreceptors, a necessary key step for increasing S/N ratio in signal transduction as achieved in vertebrates.
I have only a very minor suggestion. I would suggest to move the following sentence at page 7: "It should be noted that Opn5L1 bearing native N-and C-termini also showed molecular characteristics indistinguishable from those of Opn5L1 with modified N-and C-termini (Supplementary Fig. 4a-d)." backward to page 5, right AFTER the following sentence: "we replaced the N-and C-termini of chicken Opn5L1 with those belonging to other Opn5 proteins with higher yields (Supplementary Fig. 2)." Reviewer #3 (Remarks to the Author): The manuscript of Keita Sato belongs to the best manuscripts I have ever reviewed.
It describes a rhodopsin that belongs to a special uncharacterized melanopsin subfamily. The team demonstrated very elegantly that this rhodopsins undergoes photo-and thermal reactions that are very different from all other rhodopsins described so far. As an invertebrate rhodopsin it binds all-trans-retinal -unusual by itself -, it is active in the darkalso unusual, and it photo-isomerized into 11-cis and becomes inactive. But, most surprisingly the 11-cis photoproducts reacts with the retinal polyene backbone under formation of a thioadduct. This is a spectacular finding because this has been discussed > 40 years ago for animal rhodopsins but turned out to be wrong. However, such thioadduct formation is know for flavin-based photoreceptors and bacterial phytochromes ( which I would like to be mentioned because ist underpins the great flexibility of nature ).
I miss a little the theoretical background in the discussion and an argumentation why this reaction is not unlikely and the authors my reconsider the polarity of axcited states. An open question is the nature of thioadduct formation. Is it a thermal reaction or a secondary photochemical reaction similar to LOV-proteins. But, many interesting questions about the mechanism could be discussed that would be by far go beyond the scope of this primary report. A basic publication is this respect is the one from Mathies and Stryer (PNAS 1976, 73, 2169-73. The manuscripts reflects great rhodopsin knowledge, clean spectroscopy, and unusual analytical skills that are -unfortunately -not found any more very often in these days. The prove of the thioadduct is rock solid ( spectroscopy, mutagenesis and Mass spectroscopy), which is the mayor clue.
I have just one question. Is the protein really active in the dark or could it function as a sensor for retinal ? Could it be that it is an apoprotein most the time and only when the retinal concentration rises rhodopsin is formed and this binding process is the start for a signaling process ?
Reviewer #4 (Remarks to the Author): This is a well written and comprehensive study of the photochemistry and molecular characteristics of OPN5L1 in vitro. The authors convincingly show that this opsin binds all-trans retina; that this has the effect of allowing G-protein interaction; and that light isomerizes to 11-cis retinal at the same time switching off the receptor. They also continue to a convincing demonstration that 11-cis retinal isoform is retained by the opsin and thermally relaxes to the active all-trans isoform over hours. These characteristics are unique for opsins and therefore represent a very interesting photochemical description.
The major weakness of this paper is the lack of a functional perspective. Assuming that these characteristics are recapitulated in vivo, what physiological function can the authors conceive for Opn5L? Some speculation is required, but some information is also lacking. Where is Opn5L expressed in the chicken? This could provide some clues as to function.
The authors propose that Opn5L can act as an optical switch, but does this have any physiological relevance in vivo? Can the authors provide an estimate of the rate of Opn5L 'photobleach' in its native environment under real world conditions? A related question that the authors can address is whether applying all-trans retinal can recover 'bleached' Opn5L? If all-trans regenerates Opn5L pigment readily in the dark, then its unlikely that light will have a marked effect on its signaling in vivo and Opn5L could signal retinal concentrations, alternatively if it does not then this indicates light is indeed the physiologically relevant stimulus.
The authors should put the all-trans retinal concentration required to activate Opn5L into context, are these physiological levels? This study reports photochemical and structural properties of Opn5L1, an opsin-type protein whose nominal ligand (i.e., ligand in darkness) is all-trans retinal. The data show that this all-trans retinal is a light-regulated chromophore. Upon illumination, the Opn5L1-bound all-trans retinal undergoes isomerization specifically to 11-cis retinal, and this transition is accompanied by a reduction in Giprotein activation. This all-trans to 11-cis photoisomerization is followed by a slow reversion, in darkness, of the 11-cis retinal to all trans retinal through a reaction promoted by the thiol group of a specific cysteine in Opn5L1 (cys188).
As noted by the authors, identifying and understanding the nature of new members within the multiple families of opsin-type proteins are topics of major interest and importance to the fields of cell biology and physiology. The manuscript reports highly interesting information on Opn5L1, a member of the Opn5 group with a novel retinal "photocycle" that is remarkably distinct from, e.g., those of the retinal-bleaching visual opsins of vertebrates and the bidirectional (reversibly photoconvertible) visual opsins of invertebrates. Also as noted in the text, the occurrence of Opn5s in multiple non-retinal tissues supports the logic of the notion that Opn5L1 may function in circadian-associated, photoentrainment, or other non-visual light-dependent signaling pathways. Major overall strengths of the study are the logic of the experimental designs, and the thoroughness and presentation clarity of the biochemical analyses. The combined results obtained from spectrophotometry, the characterization of retinal/protein states at differential temperatures, liquid chromatography/mass spectrometry analysis, substitution of a threonine for cys188, and from other peptide preparations/analyses (notably, the experiment demonstrating all-trans retinal's bridging of cys188 to the retinal binding site at lys296) (Supplementary Figure 8) convincingly establish Opn5L1's action on the retinal ligand in light (all-trans to 11-cis) and in darkness (11cis to all-trans). My only significant criticism concerns the limited information provided on the downregulating action of illuminated Opn5L1, i.e., of all-trans to 11-cis retinal photoisomerization, on Gi-protein activation (Figures 1c, 1j, and Supplementary Figures 3 and 4d). The reported data show that photoisomerization reduces activation of the Gi present. However, the text lacks information that could enable evaluation of how robust is this physiological activity. Because the paper is the first photo-characterization of Opn5L1, and because Opn5L1's potential roles in cell signaling pathways and metabolism remain to be investigated, extensive characterization of the Gi-protein effectspecifically, the magnitude of Opn5L1's action on Gi-protein-targeted downstream biomolecules as a function of the number of Opn5L1s photoisomerized -is probably beyond the scope of the paper. However, a main conclusion drawn by the authors is that Opn5L1 is a physiological "photoreceptor" that in reverse-acting fashion controls some cellular process via Gi downregulation. Thus, even in a first report, greater attention should be given to clarifying how potent is this Gi deactivation signal. First, the authors should analyze their spectrophotometric and related data to derive the quantum efficiency with which a photon absorbed by all-trans Opn5L1 converts the protein to 11-cis Opn5L1. Second, under conditions of their present assays, they should determine how many Gi-proteins are de-activated by a given photoisomerized Opn5L1, and the time course with which this occurs after a brief light flash of calibrated intensity. Third, they could determine how the physiological specific activity of Opn5L1 (i.e., the number of Gi-proteins deactivated per photoisomerized Opn5L1) compares with the physiological specific activity of some other representative retinal-binding opsin (number of G-proteins activated per photoisomerization of, e.g., a visual opsin). Information on these points, although not firmly establishing or ruling out a physiological role of Opn5L1's Gi-protein deactivation, would provide a better grounding of the present findings in the context of photoisomerization efficiency and G-protein regulation by other retinal-binding opsins. Overall, given the relevance specifically of the 11-cis isomer to physiological functions of retinalbinding opsins, the paper's clear evidence that Opn5L1 illumination specifically generates the 11cis isomer is immediately intriguing and suggestive of the "photoreceptor" function concluded by the authors. However, it is conceivable that Opn5L1 acts, alternatively, as (merely) a binding protein for all-trans retinal that scavenges or sequesters this widely distributed retinoid, and that Gi-protein regulation by Opn5L1 illumination is a weak side-effect of negligible physiological importance. Further information of the types noted in points 1-3 above could bear on the evaluation of this alternative hypothesis. Minor points: What is the effect of reducing retinal's Schiff base linkage to Opn5L's lys296 using borohydride? What are the binding constants for Opn5L (high-affinity) binding of all-trans and (low-affinity) binding of 11-cis retinal? In Figure 1, the semilog plot of the "single exponential functions" in Fig. 1 is appropriate. However, the sigmoid nature of these semilog plots might be confusing to some readers. For clarity, the Not all opsins are GPCR's, as stated in Abstract and Introduction. RGR-opsin and retinochrome are likely photoisomerases that do not interact with a G protein.
The authors should change the text to say that 'most' opsins are GPCR's.

Response:
We added "most" as you suggested.

Comment:
The name Opn5L1 is inappropriate. The hybrid protein with N-and C-termini from Opn5 should not have the same name as parent Opn5L1 protein. Possible alternative names for the hybrid protein are Opn5L1.5 or Opn5L1.2.

Response:
Thank you for your suggestion. According to your comment, we referred to chicken Opn5L1 having the N-terminus of Xenopus tropicalis Opn5m as Opn5L1N and chicken Opn5L1 having both the N-and C-termini of X. tropicalis Opn5m as Opn5L1NC in the revised manuscript.

Comment: 2 / 22
The white sections of the bars in Fig. 2B indicating atRAL are difficult to see. Use another fill color.

Response:
According to your comment, we filled the bars with cyan.

Comment:
I have only a very minor suggestion. I would suggest to move the following sentence at page 7: "It should be noted that Opn5L1 bearing native N-and C-termini also showed molecular characteristics indistinguishable from those of Opn5L1 with modified N-and C-termini (Supplementary Fig. 4a-d)." backward to page 5, right AFTER the following sentence: "we replaced the N-and C-termini of chicken Opn5L1 with those belonging to other Opn5 proteins with higher yields ( Supplementary Fig. 2)."

Response:
Thank you for your comment. We moved the text in the "Results" section as you suggested. But, many interesting questions about the mechanism could be discussed that would be by far go beyond the scope of this primary report. A basic publication is this respect is the one from Mathies and Stryer (PNAS 1976, 73, 2169-73.

Response:
It was reported that a flavin-cysteine adduct in LOV protein is formed in the The major weakness of this paper is the lack of a functional perspective.
Assuming that these characteristics are recapitulated in vivo, what physiological function can the authors conceive for Opn5L? Some speculation is required, but some information is also lacking. Where is Opn5L expressed in the chicken?
This could provide some clues as to function.

Response:
In response to your comment, we prepared a new figure (Fig. 5)  "We next investigated the tissue distribution of Opn5L1 by performing in situ hybridization. Hybridization signals of Opn5L1 were detected in multiple regions in chicken brain and retina (Fig. 5). In chicken telencephalon, hybridization signals of Opn5L1 were broadly distributed in the mesopallium, but not in the hyperpallium or nidopallium (Fig. 5b-e, Supplementary Fig. 13a, b). In addition, we found hybridization signals in the paraventricular nucleus (PVN) in the chicken diencephalon (Fig. 5f, g, Supplementary Fig. 13c). Furthermore, hybridization signals were observed in the inner side of the inner nuclear layer in chicken retina, which indicates that a subpopulation of amacrine cells express Opn5L1 (Fig. 5h, i, Supplementary Fig. 13d)." (page 12, lines 13-22) "Opn5L1 rapidly loses its Gi activation ability on absorption of a photon and then gradually recovers this ability in the dark. This molecular property could be utilized as a molecular timer to measure the time elapsed since the last illumination, e.g., after sunset. To obtain insight into Opn5L1's physiological function, we investigated the localization of Opn5L1 in chicken brain and retina.
The results showed that Opn5L1 is expressed in the mesopallium and PVN in the chicken brain. It was reported based on experiments using birds that light can penetrate to the relatively superficial mesopallium and even the hypothalamus 43 , suggesting that Opn5L1 works as a reverse photoreceptor "To obtain insight into the physiological roles of Opn5L1, we first investigated the photosensitivity of Opn5L1NC. As Opn5L1NC is completely "bleached" upon absorption of visible light at 0 ºC, we successively irradiated Opn5L1NC with 500 nm-light and recorded the absorption spectrum upon each irradiation. Then, we plotted the relative peak absorbance of Opn5L1NC against irradiation time on a semi-logarithmic scale and estimated the photosensitivity by linear regression of these plots (Supplementary Fig. 12a-c). The photosensitivity of Opn5L1 thus obtained was 0.73 relative to that of bovine rhodopsin, which is comparable to the photosensitivities of other opsins such as cone pigments and melanopsin 26,27 .
Using this value together with the molar extinction coefficient (44,700 at 510 nm, Supplementary Fig.12d, e), the quantum yield of Opn5L1 was calculated to be 0.42, which is equivalent to that of the photoreaction from acid metarhodopsin to the dark state in octopus rhodopsin 28 . Thus, the photochemical parameters of

Response:
Thank you for your valuable suggestion. According to your comment, we refer to the report in our manuscript as follows: "This strongly suggests that Opn5L1 does not bind 11-cis-but instead binds all-trans-retinal, which could be formed through isomerization of 11-cis-retinal My only significant criticism concerns the limited information provided on the downregulating action of illuminated Opn5L1, i.e., of all-trans to 11-cis retinal photoisomerization, on Gi-protein activation (Figures 1c, 1j, and Supplementary   Figures 3 and 4d). The reported data show that photoisomerization reduces activation of the Gi present. However, the text lacks information that could enable evaluation of how robust is this physiological activity. Because the paper is the first photo-characterization of Opn5L1, and because Opn5L1's potential roles in cell signaling pathways and metabolism remain to be investigated, extensive characterization of the Gi-protein effect -specifically, the magnitude of Opn5L1's action on Gi-protein-targeted downstream biomolecules as a function of the number of Opn5L1s photoisomerized -is probably beyond the scope of the paper. However, a main conclusion drawn by the authors is that Opn5L1 is a physiological "photoreceptor" that in reverse-acting fashion controls some cellular process via Gi downregulation. Thus, even in a first report, greater attention should be given to clarifying how potent is this Gi deactivation signal.
First, the authors should analyze their spectrophotometric and related data to derive the quantum efficiency with which a photon absorbed by all-trans Opn5L1 converts the protein to 11-cis Opn5L1.

Response:
Thank you for your valuable comment. We investigated the photosensitivity and quantum yield of Opn5L1 and compared them to those of bovine rhodopsin. The "To obtain insight into the physiological roles of Opn5L1, we first investigated the photosensitivity of Opn5L1NC. As Opn5L1NC is completely "bleached" upon absorption of visible light at 0 ºC, we successively irradiated Opn5L1NC with 500 nm-light and recorded the absorption spectrum upon each irradiation. Then, we plotted the relative peak absorbance of Opn5L1NC against irradiation time on a semi-logarithmic scale and estimated the photosensitivity by linear regression of these plots (Supplementary Fig. 12a-c). The photosensitivity of Opn5L1 thus obtained was 0.73 relative to that of bovine rhodopsin, which is comparable to the photosensitivities of other opsins such as cone pigments and melanopsin 26,27 .
Using this value together with the molar extinction coefficient (44,700 at 510 nm, Supplementary Fig.12d, e), the quantum yield of Opn5L1 was calculated to be 16 / 22 0.42, which is equivalent to that of the photoreaction from acid metarhodopsin to the dark state in octopus rhodopsin 28 . Thus, the photochemical parameters of Opn5L1 are quite similar to those of other opsins that function as an optical switch." (page 11, lines 22-24; page 12, lines 1-12)

Comment:
Second, under conditions of their present assays, they should determine how many Gi-proteins are de-activated by a given photoisomerized Opn5L1, and the time course with which this occurs after a brief light flash of calibrated intensity.

Response:
In our assay conditions shown in figure  behave like other diffusible ligand-bound GPCRs in the cells. We added the following sentences in the revised manuscript, which also incorporate our response to your next comment: "Furthermore, we compared the G protein activation efficiency of Opn5L1 with those of chicken Opn5m and bovine rhodopsin ( Supplementary Fig. 5). The Gi activation efficiency of Opn5L1 was about 5-fold higher than that of chicken Opn5m and 100-fold lower than that of bovine rhodopsin. This efficiency of Opn5L1 is similar to that of bistable opsin, which has a 50-fold lower activation efficiency than bovine rhodopsin 10 ." (page 7, lines 7-12)

Comment:
Third, they could determine how the physiological specific activity of Opn5L1 (i.e., the number of Gi-proteins deactivated per photoisomerized Opn5L1)

/ 22
compares with the physiological specific activity of some other representative retinal-binding opsin (number of G-proteins activated per photoisomerization of, e.g., a visual opsin). Information on these points, although not firmly establishing or ruling out a physiological role of Opn5L1's Gi-protein deactivation, would provide a better grounding of the present findings in the context of photoisomerization efficiency and G-protein regulation by other retinal-binding opsins. Overall, given the relevance specifically of the 11-cis isomer to physiological functions of retinal-binding opsins, the paper's clear evidence that Opn5L1 illumination specifically generates the 11-cis isomer is immediately intriguing and suggestive of the "photoreceptor" function concluded by the authors. However, it is conceivable that Opn5L1 acts, alternatively, as (merely) a binding protein for all-trans retinal that scavenges or sequesters this widely distributed retinoid, and that Gi-protein regulation by Opn5L1 illumination is a weak side-effect of negligible physiological importance. Further information of the types noted in points 1-3 above could bear on the evaluation of this alternative hypothesis.

Response:
According to your comment, we compared the activation efficiency of Opn5L1 with those of visual opsin, bovine rhodopsin and a non-visual opsin, chicken Opn5m ( Supplementary Fig. 5). In our assay condition, the Gi activation efficiency of Opn5L1 was about 100-fold lower than that of bovine rhodopsin and "Furthermore, we compared the G protein activation efficiency of Opn5L1 with those of chicken Opn5m and bovine rhodopsin ( Supplementary Fig. 5). The Gi activation efficiency of Opn5L1 was about 5-fold higher than that of chicken Opn5m and 100-fold lower than that of bovine rhodopsin. This efficiency of Opn5L1 is similar to that of bistable opsin, which has a 50-fold lower activation efficiency than bovine rhodopsin 10 ." (page 7, lines 7-12)

Comment:
What is the effect of reducing retinal's Schiff base linkage to Opn5L's lys296 using borohydride?

Response:
In supplementary Fig. 10, we show that we obtained evidence for adduct formation between Cys188 and the retinal by a western blotting-based method.

/ 22
In the experimental procedures, we needed to block the dissociation of retinal-thio adduct between retinal and Cys188 and the hydrolysis of Schiff base linkage between retinal and Lys296. Reductive conversion of this Schiff base linkage to secondary amine could prevent these reactions and tighten the linkage between Cys188 and Lys296. Sodium borohydride is one of the widely used reducing agents for such reductive amination. However, sodium borohydride also cleaves disulfide bonds. In the procedures, we used factor Xa to cleave the recognition site inserted into intracellular loop3. Because functional factor Xa forms a dimer bridged by a disulfide bond, sodium borohydride could prevent the protease activity of factor Xa. Thus, we instead used 2-picoline borane, which reduces Schiff base linkage selectively but not the disulfide bond in the present study. We revised the sentence as follows to show our intention more clearly: "Additionally, we irradiated an Opn5L1 samples in the presence of a boron-based reducing agent to stabilize the linkage between Cys188 and Lys296 by amination of the Schiff base, because the presence of a reactive protonated Schiff base can result in dissociation of the retinal-thio adduct or hydrolysis of the Schiff base ( Supplementary Fig. 10b)." (page 11, lines 7-11)

Comment:
What are the binding constants for Opn5L (high-affinity) binding of all-trans and (low-affinity) binding of 11-cis retinal? 21 / 22

Response:
As shown in supplementary figure 4, a titration experiment of Opn5L1 with all-trans-retinal indicated that the activity significantly increased in the range from 10 -8 to 10 -6 M of all-trans-retinal, and EC 50 for retinal was estimated to be 6.2 × 10 -7 M. However, as shown in Figure 1, we could not observe the direct binding of 11-cis-retinal to Opn5L1, probably because of its much lower affinity compared to that of all-trans-retinal. Thus, we could not determine the affinity of 11-cis-retinal for Opn5L1. We added the following sentence in the revised manuscript: "EC 50 for retinal was estimated to be 6.2 x 10 -7 M." (page 6, lines 17-18)

Comment:
In Figure 1, the semilog plot of the "single exponential functions" in Fig. 1 is appropriate. However, the sigmoid nature of these semilog plots might be confusing to some readers. For clarity, the Figure 1 legend (manuscript lines 562-563) should indicate the full equations for single exponential functions being used.

Response:
In response to your comment, we revised the legend as follows; "(g) The absorbance at 495 nm (black circles) and 270 nm (red triangles) plotted against the time after flash irradiation. The time profiles were fitted by single exponential functions y=y 0 -b×exp(-t/τ) with the same time constant (solid curves,