Opsin 5 mediates violet light-accelerated wound healing in mammalian corneas

SUMMARY

The cornea is the transparent tissue at the ocular surface that generates most of the refractive power of the eye. Due to its exposed location, the cornea is uniquely in danger of injury. Rapid and efficient healing is required for high-acuity vision. Here, we show that epithelial corneal wounding induces Opn5, an opsin G-protein-coupled receptor (GPCR) sensitive to ultraviolet/violet light, at the cornea’s surface in both mice and primates. The expression of corneal Opn5 is coincident with the direct light sensitivity of multiple pathways, including circadian rhythms, damage-response genes, and immune modulators. The presence of a violet-light:dark cycle substantially accelerates epithelial wound closure both in vivo and ex vivo. Corneas lacking Opn5 have markedly impaired wound healing. Violet light also accelerates wound healing in non-human primate corneas and induces gene expression similar to mice. These results demonstrate that corneal wounding induces direct photosensitivity via Opn5, which accelerates wound healing in the cornea.

In brief

Díaz et al. demonstrate that mechanical injury to the surface of the cornea induces temporary expression of an opsin photoreceptor, Opn5, in mice and primates. The induction of this photoreceptor is associated with accelerated healing in the presence of violet light.

Graphical Abstract:

INTRODUCTION

The cornea and lens of the eye constitute an optical system that focuses light onto the retina. The cornea is the dome-shaped, transparent tissue at the surface of the eye, and it contributes about 2/3 of refractive power for the visual system in most mammals.^1,2 The cornea consists of three cellular layers: the epithelium, stroma, and endothelium. The epithelium is the outermost layer of cells and undergoes constant renewal from progenitors in the limbus, a ring of proliferative cells located at the circumference of the cornea. Due to its location at the anterior pole of the eye, it is in direct contact with the external environment and is uniquely prone to injury; it is estimated that 1.5% of individuals experience a corneal abrasion annually, and 3% of all emergency room visits are thought to be due to eye trauma, primarily corneal injury.^3–5 When surface damage occurs to the cornea, dead cells are sloughed off and replenished by both cells from the limbus and cells dividing within the basal epithelium.^6–8 Basal and reparative cell proliferation occurs with a near-24 h circadian rhythm.^9–11 Rapid and complete healing of the cornea is essential to prevent infection of the underlying stroma, which can result in opacification and blindness. In humans, corneal scarring remains a leading cause of blindness worldwide.¹²

The non-visual opsins, OPN3, OPN4, and OPN5, are opsin-family receptors that are activated by light but that subserve non-visual functions.^13–18 We have previously shown that the local molecular circadian clocks in cells of the murine retina and skin display an Opn5-dependent violet-light-induced photoentrainment, which is independent of the phase of behavioral rhythms.^14,18 A similar photoentrainment of the local molecular circadian clocks is observed within the cornea.¹⁴ Interestingly, Opn5 is not constitutively expressed in healthy mouse corneas in vivo but is induced upon tissue culture.¹⁹ In addition to the regulation of local circadian clocks, Opn5 expression in the retina is necessary for violet-light-dependent suppression of defocus-induced myopia,^20–22 influences the regulation of body temperature,¹⁷ and, in conjunction with Opn4, regulates the light-dependent timing of vascular development.^23,24 Opn4 has also been identified in the trigeminal sensory neurons that innervate the cornea, and it plays a role in photo-oculodynia.^25,26

OPN5 integrates light cues over relatively long timescales (minutes or longer) compared with the millisecond-level responses required for visual signaling.²⁷ In addition to neural expression in the retina, OPN5 and the other non-visual opsins are also expressed in several non-neuronal cell types. We have identified Opn5 in epithelial cells and keratocytes in the cultured murine cornea and in melanocytes of dermal hair follicles.^18,19 Opn4 and Opn3 have also been identified in cultured mammalian melanocytes.^28–30 Opn3 is also expressed in white adipocytes of scapular and inguinal fat pads, where it performs a role in light-induced lipolysis.³¹ While the diverse roles of non-visual opsins are still being defined, their expression in non-neural cell types appears to involve slow light response kinetics and often the release of paracrine factors.

In the current work, we demonstrate that Opn5 is transiently expressed in corneas after physical injury in vivo. This expression is coincident with the induction of direct corneal light sensitivity and light-accelerated healing. Corneas from Opn5^−/− mice show a loss of light sensitivity and show deficiencies in the speed and integrity of healing after epithelial injury. The most extreme phenotypes of Opn5^−/− corneas can be rescued by diffusible signals from wild-type tissues, suggesting a light-sensitive paracrine mechanism. We also demonstrate that non-human primate corneas show similar mechanical-injury-induced Opn5 expression, light-dependent acceleration of healing, and photoinduction of the same molecular targets as in mice.

RESULTS

Opn5 is induced in mouse corneas after epithelial de-epithelialization

To investigate injury-induced corneal opsin expression in vivo, we performed mechanical de-epithelialization using an Algerbrush as a mild form of corneal injury, comparable to a corneal abrasion injury in humans.³² We quantified Opn5 transcript levels and the number of fluorescent cells in the corneas of Opn5^Cre/+; Ai14 (tdTomato) mice over a 3-day, post-abrasion time course in vivo. tdTomato-positive cells were observed as soon as 6 h post-abrasion, with peak levels at around 12 h (Figures 1A and 1B). The number of Opn5-positive cells then waned steadily to near zero by 72 h post-abrasion. Because the flox-based reporter system permanently activates tdTomato expression within positive cells, their steady disappearance after 12 h indicates that the cells are dying or migrating or have sloughed off. Opn5 transcript was also detected in corneas within 3 h of de-epithelialization and followed an induction and resolution similar to those for the tdTomato reporter (Figure 1C). As has been observed in cultured cornea,¹⁹ Opn5-induced cells were largely localized initially in the basal epithelium. Subsequently, Opn5-expressing cells were found throughout the epithelium and upper stroma before disappearing (Figures 1D and 1E). The lack of replenishment of tdTomato-positive and Opn5-transcript-expressing cells indicates that the mechanical wound induces Opn5 expression only transiently.

Figure 1. Epithelial abrasion in vivo induces the opsin Opn5 in mouse corneas and induces temporary light sensitivity.

(A) Opn5^Cre/+; Ai14 (tdTomato) mice underwent de-epithelialization with an Algerbrush, and corneas were collected at the indicated times post-de-epithelialization. Blue: Hoechst nucleic acid stain; magenta: native tdTomato. Scale bar represents 200 μm unless otherwise noted. White dashed lines indicate area remaining to be re-epithelialized. Yellow dashed rectangles indicate area shown in detail in bottom images. Gray dashed lines on 12–48 h indicate estimate of original abrasion, based on density of Hoescht nuclear stain.

(B) Number of tdTomato-positive cells/mm² in Opn5^Cre/+; Ai14 mouse corneas for the indicated hours post-de-epithelialization. n = 5 each time point. Blue points: individual corneas; the orange line connects each mean value with a straight line. *p > 0.05 in ANOVA with Tukey post hoc analysis.

(C) Similar to (B) but transcript of Opn5 measured relative to β-actin using ΔΔCt RT-PCR. n = 4 each point. *p < 0.05 in ANOVA with Tukey post hoc analysis.

(D) Images of fluorescence from tdTomato cells (magenta) or Hoechst stain (blue) of a whole-mount Opn5^Cre/+; Ai14 cornea 24 h post-de-epithelialization.

(E) Similar to (D) but at higher magnification to show details of 3 cells. (E^′) shows the side view of the same volume as in (E) to demonstrate the location between the stroma and basal epithelium of the tdTomato+ cells at 24 h post-de-epithelialization.

(F) Two cohorts of C57bl/6J mice were given a de-epithelialization of a single cornea and then maintained either in darkness or given a 30 min, 415 nm, 5 × 10¹⁴ photon/cm²/s light pulse at either 24 or 48 h post-de-epithelialization. de-epithel., de-epithelialized.

(G) Western blot of phosphorylated or total ERK 1/2 protein from mice described in (F). n = 4 each group. *p < 0.05 in ANOVA with Tukey post hoc analyses of phosphorylated/total ERK ratio.

(H and I) Western blots of phosphorylated and total ERK from corneas of Opn4^−/−; Pde6b^rd1/rd1 (H) or Opn5^−/− (I) mice 24 h post-de-epithelialization. n = 4 each group. *p < 0.05 in ANOVA with Tukey post hoc analyses of phosphorylated/total ERK ratio.

As OPN5 is known to encode a functional photopigment that can activate mitogen-activated protein kinase (MAPK) pathway signaling,²⁷ to assess if its transient expression is correlated with the induction of light sensitivity in the cornea, we measured the phosphorylation of MAPK (ERK 1/2) proteins in the corneas of violet-light-exposed mice after an in vivo de-epithelialization.²⁷ One eye was de-epithelialized and the other left intact as a contralateral control (Figure 1F). 24 or 48 h after abrasion, a 30 min violet (415 nm, 5 × 10¹⁴ photons/cm²/s) light exposure was given, and phosphorylated ERK 1/2 (p-ERK1/2) was measured in both the de-epithelialized and fellow eyes (Figure 1G). We used 415 nm light since it is well within the absorption spectrum OPN5 but not so short in wavelength to cause ionizing damage at the intensities utilized.^14,33 In wounded eyes, we observed significant light-induced phosphorylation of ERK 24 h post-abrasion but not after 48 h (Figure 1G). The corneas of fellow, unwounded eyes showed no photo responses at any time point.

To test for the contribution of visual or melanopsin-based photoreception from the retina, we repeated the experiment in Opn4^−/−; Pde6b^rd1/rd1 mice, which lack rods, cones, and melanopsin, are visually blind, and show a complete lack of behavioral circadian photoentrainment.^13,34 Following corneal de-epithelialization, wounded corneas displayed normal light-induced ERK phosphorylation, demonstrating that light activation of p-ERK does not require input from visual photoreceptive or melanopsin retinal ganglion cell pathways (Figure 1H). This also demonstrates that local expression of OPN4 in the trigeminal nerve is not necessary for these photic effects.^25,26 Crucially, we de-epithelialized corneas of the Opn5^−/− mice and found that 415 nm light did not induce ERK phosphorylation despite the presence of rods, cones, and melanopsin (Figure 1I). Thus, Opn5 is induced by corneal wounding and appears necessary for the induced photosensitivity, as evidenced by local phosphorylation of ERK in response to violet light.

Light sensitivity in the wounded cornea regulates multiple molecular pathways

We next defined wound- and light-dependent changes in the corneal transcriptome. Corneal mRNA was harvested for bulk RNA sequencing (RNA-seq) 12 h after de-epithelialization and a 60 min violet light pulse (415 nm, 5 × 10¹⁴ photons/cm²/s). We identified ~300 differentially expressed transcripts in the light- vs. dark-treated corneas, the majority of which were upregulated by light (Figure 2A). Network pathways that were regulated by light (as identified by WikiPathways and Panther analyses) included circadian clock genes, extracellular matrix regulating pathways, and genes related to immediate immune responses (Figures 2B and S1A). Of note, the transcription factor EGR1, which has been implicated in wounding response pathways in the cornea and in eye growth, was identified as a light-regulated transcript.^22,35 We confirmed by quantitative RT-PCR and western blot that the Egr1 transcript and protein were more abundant in response to a 415 nm light pulse in wounded corneas than in darkness (Figures 2C and 2D). Also of note was the light inducibility of genes related to antigen response, such as the Nod-like receptor Nlrc5.³⁶

Figure 2. Extensive transcriptional changes by light in wounded corneas.

(A) Volcano plot of 331 differentially abundant transcripts in corneas 24 h post-de-epithelialization after a 1 h, 415 nm light pulse in vivo measured by bulk RNA-seq. Transcripts (243) upregulated in light are shown in red and downregulated (87) are shown in blue. n = 4 for each group.

(B) Individual read counts of transcripts of selected major regulated pathways.

(C) RT-PCR of Egr1 in 6 de-epithelialized corneas after darkness or 1 h, 415 nm light pulse in vivo. *p < 0.05 in Student’s two-tailed t test.

(D) Western blot of EGR1 or total ERK (loading control) in de-epithelialized corneas exposed to 415 nm light in vivo for the indicated duration. Right image shows normalized quantification of EGR1 to total ERK ratio. n = 4 each group. *p < 0.05 in ANOVA with Tukey post hoc analyses of EGR1/total ERK ratio.

We next compared transcripts from Opn5^−/− and wild-type wounded corneas using bulk RNA-seq, and ~2,700 transcripts were differentially expressed (Figure 3A). Surprisingly, the extent of the altered transcript profile in Opn5^−/− corneas was much broader than the light-response pathways identified in wild-type corneas, indicating that the absence of Opn5 is associated with changes to non-photic pathways as well (Figures 3A and 3B). A stronger presence of the immune response and general wound response activation was present among the Opn5-linked transcripts. 84 of the light-induced transcripts in wild-type wounded tissue (from Figure 2A) were dysregulated in Opn5^−/− corneas. Pathway analyses suggest that these transcripts belong to pathways regulating integrin signaling, circadian rhythms, and wound responses (WikiPathways and PantherDB; Figure S1). These data indicate that Opn5 gates the light regulation of multiple pathways but also contributes a developmental, constitutive, or damage-induced role in darkness.

Figure 3. Transcriptional changes in Opn5^−/− wounded corneas.

(A) Volcano plot of 2,677 differentially abundant transcripts in wild-type vs. Opn5^−/− corneas 24 h post-de-epithelialization after a 1 h, 415 nm light pulse in vivo measured by bulk RNA-seq. n = 4 for each group.

(B) Individual read counts of transcripts of selected major regulated pathways.

Light and Opn5 regulate proper wound healing in murine corneas

The identification of wound response pathways, collagen genes, and extracellular matrix genes in RNA-seq analyses led us to investigate corneal wound healing in Opn5^−/− mice. To isolate the local light effects on corneas, we first utilized an ex vivo wounding assay.^37,38 Corneas were de-epithelialized at the time of euthanasia, and whole anterior segments (without lens) were maintained as organotypic cultures. To monitor wound healing, we imaged fluorescein-stained corneal surfaces daily for 1 week post-culture/de-epithelialization (Figure 4A). Wild-type and Opn5^−/− anterior segments were maintained in a 12 h:12 h light:dark cycle of violet light (415 nm, 5 × 10¹⁴ photons/cm²/s) or in total darkness (except for the < 1 min blue light for imaging) for the duration of the experiment. Wild-type corneas fully re-epithelialized within 1 week, and this occurred approximately 2 days faster in the presence of a light:dark cycle than in constant darkness (p < 0.05 ANOVA; Figure 4B). Surprisingly, the corneas of Opn5^−/− mice failed to heal in either condition (Figures 4A and 4B). Thus, OPN5 is necessary for, and light enhances, the normal healing of corneas in an ex vivo wounding model.

Figure 4. Delayed healing in Opn5^−/− corneas after epithelial abrasion.

(A) Images of fluorescein-stained corneas in ex vivo assays of anterior segments as organotypic cultures de-epithelialized on day 0.

(B) Percentage of de-epithelialization zone remaining in (A) shown as mean ± SEM. n = 5 each group, *p < 0.05 ANOVA with Tukey post hoc analyses. “Light” is 12 h:12 h light:dark cycle of violet light (415 nm, 5 × 10¹⁴ photons/cm²/s). “Dark” is darkness except one daily blue light exposure for fluorescein imaging for <1 min.

(C) De-epithelialized corneas were placed directly on top of anterior segments with similarly de-epithelialized corneas. Only the top cornea was stained with fluorescein (green in schematic). The percentage of de-epithelialization zone remaining in Opn5^−/− corneas after epithelial abrasion is shown as mean ± SEM. n = 5 each group. *p < 0.05 ANOVA with Tukey post hoc analyses comparing Opn5^−/−: Opn5^−/− (light green hatched line) with Opn5^−/−: wild type (dark green solid line). All co-cultured corneas were maintained in a light:dark cycle of 415 nm, 5 × 10¹⁴ photons/cm²/s.

(D) Corneas of K14; Ai14 mice were de-epithelialized, and anterior segments were maintained ex vivo for 1 week with a fluorescence image collected every 2 h. A 2 h, 415 nm, 5 × 10¹⁴ photons/cm²/s light was administered where indicated by orange arrows. The percentage of remaining de-epithelialization is shown as mean ± SEM for light exposed (orange) or dark controls (gray). n = 5 each group, small points show individual runs. p < 0.05 comparison of area under curve compared by Student’s two-tailed t test.

To better understand how Opn5 was exerting its effect on wound healing, cell autonomously or through local paracrine signaling, we co-cultured wounded Opn5^−/− corneas with wild-type anterior segments so that the Opn5^−/− cornea was laid directly on top of the wild-type (Figure 4C). We then measured the amount of the wound remaining on the Opn5^−/− corneal surface in the presence of a light:dark cycle. When in direct contact with wild-type anterior segments, the Opn5^−/− corneas re-epithelialized, albeit more slowly than wild-type corneas in similar conditions. The co-cultured Opn5^−/− corneas did not heal when co-cultured with another Opn5^−/− anterior segment. Conversely, wild-type corneas co-cultured with another wild-type tissue did not show altered healing rates. Despite the slower rate of healing, the co-cultured Opn5^−/− corneas showed significantly improved healing compared with Opn5^−/− corneas without co-culture (Figure 4C). These results suggest that Opn5-dependent corneal wound healing is dependent on intercellular signaling, most likely mediated by diffusible signals. The inability of Opn5^−/− corneas to re-epithelialize is thus likely due to an absence of one or more secreted factors and not a developmental or cell-intrinsic healing deficiency.

In order to understand if shorter durations of light exposure were sufficient to accelerate healing, 2 h durations of daily violet light exposure (415 nm, 5 × 10¹⁴ photons/cm²/s) were administered on the background of darkness ex vivo. For these experiments, we imaged whole murine anterior segments from K14-Cre; Ai14 mice every 2 h in organotypic culture. The light needed to image the fluorescence of the tdTomato reporter (Ai14) was brief (<100 ms) and of a long enough wavelength (550 nm) so as not to activate OPN5. Wounded K14-Cre; Ai14 corneas re-epithelialized within 6–7 days in darkness, similar to experimental imaging with fluorescein (Figures 4B and 4D). When a daily 2 h light pulse (415 nm, 5 × 10¹⁴ photons/cm²/s) was administered, corneas re-epithelialized within 2–3 days post-abrasion (p < 0.05, Student’s t test; Figure 4D). Strikingly, a brief violet light exposure was as effective in accelerating re-epithelialization as a full light:dark cycle (Videos S1 and S2).

The role of Opn5 in corneal wound healing in vivo

To extend these findings in vivo, we performed corneal de-epithelialization on anesthetized wild-type and Opn5^−/− mice. The corneas of wild-type mice had re-epithelialized by about 24 h post-de-epithelialization (Figure 5A). By contrast, the corneas of Opn5^−/− mice failed to heal until the fourth day post-de-epithelialization (p < 0.05, ANOVA; Figures 5A and 5B). In order to assess the spectral specificity of the light, which accelerates corneal healing, we compared healing rates of wild-type mice in 12 h:12 h light:dark cycles of violet (415 nm, 5 × 10¹⁴ photons/cm²/s), green (525 nm, 5 × 10¹⁴ photons/cm²/s), or red (625 nm, 5 × 10¹⁴ photons/cm²/s) light to constant darkness. Consistent with previous reports, the presence of a light:dark cycle yielded faster re-epithelialization within the 24 h healing window (Figure 5C).³⁹ However, in our analyses, only violet light promoted healing acceleration compared with darkness and green and red wavelengths (p < 0.05, ANOVA; Figure 5C). Violet wavelengths are consistent with the spectral sensitivity of OPN5-mediated photoreception in mammals.^14,18,33

Figure 5. Opn5 is necessary for corneal wound healing and light responses.

(A) Photographs of fluorescein-stained corneas that were de-epithelialized on day 0.

(B) Percentage of initial wound remaining after epithelial removal in vivo for wild-type (blue) and Opn5^−/− (green). Shown are mean ± SEM, n = 4, *p < 0.05 two-way ANOVA with Tukey post hoc analyses.

(C) Wild-type mice received a corneal de-epithelialization 2 h prior to lights turning on in a light: dark cycle or at the same corresponding phase in constant darkness. The fluorescein-stained wound remaining was measured at 12, 16, 20, and 24 h post-de-epithelialization in either darkness, violet (415 nm), green (525 nm), or red (625 nm). All light was at 5 × 10¹⁴ photon/cm²/s. Shown are mean ± SEM, n = 4, *p < 0.05 two-way ANOVA with Tukey post hoc analyses.

(D) Immunofluorescence of keratin 14 (magenta) and DAPI (green) in wild-type (top) or Opn5^−/− corneas 24 h post-de-epithelialization. Scale bar represents 200 μm.

(E and F) Immunofluorescence of keratin 12 (magenta) and DAPI (white) in wild-type (E) or Opn5^−/− (F) corneas 24 h post-de-epithelialization.

(G and H) DAPI stain to compare fellow, untreated cornea (G) to a de-epithelialized cornea (H) 24 h post de-epithelialization.

We also collected corneas for histological assessment 24 h post-abrasion. The healed corneas of wild-type mice showed minimal change, with healthy epithelial and stromal layers, compared with unwounded corneas in cross-sections (Figures 5D and 5E). However, the wound zone was readily identifiable in Opn5^−/− corneas, as re-epithelialization was incomplete, and the stroma showed weakness during cryosectioning in the wounded area (Figures 5D and 5F). In regions in which epithelial layers had reformed, we identified abnormal densities in the underlying stroma only in wounded Opn5^−/− corneas (Figure 5F, red arrow). Interestingly, these Opn5-dependent mechanisms of corneal structure are only present post-wounding in Opn5^−/− corneas and not in fellow untreated eyes (Figures 5H and 5I). This suggests that Opn5 is not necessary for corneal development and homeostasis but is crucial for normal wound healing in adult animals.

Opn5 is induced in non-human primate corneas and is associated with local light sensitivity

To determine if primate corneas also upregulate Opn5 after wounding, we collected anterior segments from the eyes of macaques (Macaca nemestrina) less than 3 h after harvest. We de-epithelialized the cornea and maintained anterior segments (without the lens) as organotypic cultures (Figure 6A). Using quantitative RT-PCR, we observed an approximately 100-fold increase in Opn5 after 24 h of culture. This level receded toward baseline within 3 days (Figure 6B). We did not observe changes in the expression of Opn3 or Opn4 during the 3 days of organotypic culture and could not detect the transcript for Opn1sw or Opn2 (rhodopsin) (data not shown). As with the analysis of mouse corneas, we measured the phosphorylation of ERK after 15–20 min of violet light exposure (Figure 6C). To characterize the spectral sensitivity of this photoreception, we applied 5 × 10¹⁴ photons/cm²/s of 400 (violet), 415 (violet), 475 (blue), or 525 (green) nm light for 30 min. Of the wavelengths tested, only violet wavelengths (400 and 415 nm) induced phosphorylation of ERK proteins within 30 min (Figure 6D). We also analyzed the induction of phospho-CREB, a downstream transcription factor activated in ERK pathways, in violet-light-exposed macaque corneas. We observed a marked increase of phospho-CREB-positive cells, particularly in regions around the corneal limbus (Figure 6E).

Figure 6. Photosensitivity of macaque corneas de-epithelialized ex vivo.

(A) Macaque anterior segments were maintained as organotypic cultures and received an approximately 1.5 cm diameter circular epithelial removal on the day of culture.

(B) Abundance of Opn5 transcript compared to β-actin using ΔΔCt RT-PCR in macaque corneas collected at indicated times. n = 6 each time point. Blue points show individual data points, and means are connected by an orange line. *p < 0.05 in ANOVA with Tukey post hoc analyses.

(C) Western blot of phosphorylated ERK or α-tubulin protein from corneas exposed to the indicated duration of 415 nm light.

(D) Ratio of phosphorylated ERK or α-tubulin in western blots from corneas exposed to 5 × 10¹⁴ photons/cm²/s of the indicated wavelength. n = 5 at each point. *p < 0.05 ANOVA and Tukey post hoc comparing time points to 0 min.

(E) Cryosections of macaque corneas near the corneal limbus 24 h post-de-epithelialization stained with DAPI nuclear stain (blue) and phospho-CREB (green) after a 30 min, 415 nm light pulse.

(F) Abundance of indicated transcript compared to β-actin using ΔΔCt RT-PCR in macaque corneas collected 24 h post-de-epithelialization after 1 h, 415 nm light pulse. n = 6 each time point. Points show individual data points, and bars show mean ± SEM. *p < 0.05 one-way ANOVA corrected for multiple comparisons.

(G) Macaque corneas de-epithelialized and stained with fluorescein were maintained in constant darkness or a 12 h light:12 h dark cycle of 415 nm light (5 × 10¹⁴ photons/cm²/s). Right image shows the percentage of wound remaining each day in darkness (blue solid line) or light cycle (green hatched line). Shown are the mean connected with straight line ± SEM, n = 4, *p < 0.05 two-way ANOVA with Tukey post hoc analyses.

We next analyzed the transcript levels of selected light-regulated mouse cornea genes in macaque cornea RNA. Using quantitative RT-PCR in macaque corneas exposed to 60 min of 5 × 10¹⁴ photons/cm²/s of 415 nm light or maintained in darkness, we observed that Per2 and Egr1 transcripts were upregulated in light in macaque, as in murine corneas (Figure 6F). Ifr8, Nlrc5, and Col6a2 levels were also higher after light exposure. These results confirm that after wounding, both murine and primate corneas transiently express Opn5 and become directly photosensitive.

Finally, we performed an ex vivo re-epithelialization assay using macaque corneas, comparing constant darkness to a 415 nm 12 h light:12 h dark cycle. The de-epithelialized zone was measured daily using fluorescein, as with experiments performed using mouse corneas. In the presence of a light:dark cycle, macaque corneas re-epithelialized completely within 2 days (Figure 6G). The cohort of corneas maintained in constant darkness required 4 days for complete restoration of the epithelial layer.

DISCUSSION

In this study, we identify OPN5, a violet-light-sensitive opsin-class G-protein-coupled receptor (GPCR), as essential for rapid corneal wound healing in mice and non-human primates. Opn5 is nearly undetectable in healthy corneas; only after wounding does the cornea express this opsin and confer light sensitivity. Opn5 induction is coincident with violet light activation of MAPK signaling, induction of light response genes, including Per2 and Egr1,^40–42 and light-dependent acceleration of re-epithelialization after abrasion. These effects are evident in both rodent and primate wounded corneas, and both are selectively sensitive to short-wavelength light. These results suggest that mechanical wounding results in transient light sensitivity of the cornea mediated by Opn5, which allows light to accelerate epithelial wound healing. Future studies will determine if other superficial injuries (such as chemical burns and radiation injury) and deeper stromal wounds also induce Opn5 and utilize light-accelerated healing mechanisms. It is unclear at present if Opn5 plays a role in recurrent corneal ulcers or scarring.

In Opn5^−/− mice, we observed extensive dysregulation of transcriptional pathways and a defective wound-healing response beyond what is observed in wild-type animals in darkness. This suggests that OPN5 has activity in darkness beyond its function as a photoreceptor. All opsins are GPCRs, and non-visual opsins have the potential to regulate G-protein signaling regardless of the lighting environment.^29,43–47 While OPN5 has been demonstrated to act as a bona fide photoreceptor in heterologous systems,²⁷ it is possible that it adopts both dark signaling and light signaling states in healing tissue. Interestingly, Opn5 is the most highly conserved among mammalian opsins.⁴⁸ A role for OPN5 in wound healing may partly account for the evolutionary constraint that explains this high degree of conservation.

Opn5^−/− corneas required 4 days for re-epithelialized in vivo (with persistent abnormalities), and they failed to heal ex vivo for as long as we measured. This suggests an interplay between the OPN5-driven healing mechanisms within the cornea and the neural and humoral environment. Paracrine signaling mechanisms constitute a large portion of cell-cell communication during corneal healing in vivo.^49,50 This occurs between corneal cell types (keratocytes or myofibroblasts to epithelial cells, for example) and between the trigeminal nerve and corneal cells. Because all nerves were transected in our ex vivo experiments (mimicking a neurotrophic cornea), a loss of signals from healthy nerves may account for the difference in re-epithelialization rate ex vivo vs. in vivo in wild-type tissue. Improved healing of Opn5^−/− tissue in co-culture experiments suggests the presence of light-regulated paracrine factors among corneal cell types. These OPN5-dependent factors contribute to the milieu of paracrine healing factors in the cornea, likely in the initial phases of repair. Future work will delineate the pathways by which the local light response of healing tissue interacts with wound-healing responses mediated by the peripheral nervous system.

The current data add to the emerging literature describing opsin-mediated non-visual light reception in mammals. OPN4 has been extensively studied as a photoreceptor in the inner retina, where it regulates the photoentrainment of the behavioral circadian clock.^13,51–56 Interestingly, OPN4 was first identified in the melanophores of frog skin, where it allows light-driven expansion of melanosomes.^57–59 Opn4 has since been detected in mammalian skin, along with 2 other non-visual opsins, Opn3 and Opn5, where they are responsible for a variety of functions.^{18,29,33,60,61} For example, OPN5 in skin decodes light information to establish the phase of local molecular circadian rhythms.¹⁸ This is similar to the role it plays in the mouse retina, where it mediates photoentrainment of a molecular clock and functions independently of the OPN4-assisted behavioral clock.^14,34 Outside the eye and skin, non-visual opsins have been detected in deeper structures in the mammalian brain,^62–64 smooth muscle,^65,66 adipose tissue,³¹ testes,^62,63 and cranial nerves.²⁶ In the brain, Opn5 is expressed in the hypothalamus and regulates thermogenesis in scapular brown adipose tissue in response to violet light.¹⁷ Opn3-expressing white adipose tissue also modulates body temperature in direct response to blue light through lipolysis.^17,31 Smooth muscles of peripheral vasculature and pulmonary airway show acute relaxation in response to light through pathways that appear to require Opn3 and Opn4.^66–68 Finally, the trigeminal ganglia have been shown to respond to blue light using an Opn4-dependent pathway, potentially mediating acute photoallodynia.²⁶ Our current data reveal a dynamic function of OPN5 in light-regulated wound healing. A remarkable feature of the Opn5-dependent wound response system is its transient nature during the regenerative phases of healing. While the selective advantage conferred by this mechanism is not yet known, it is possible that this facultative photoreception is utilized to synchronize and temporally sequester healing mechanisms in coordination with sunlight cycles, which could be particularly damaging at certain phases of cell division.

Further understanding of the light- and OPN5-dependent healing mechanisms in the cornea may reveal new therapeutic targets and strategies for ameliorating ocular surface injury. The use of light has gained popularity in the study of wound healing,⁶⁹ and molecular circadian clocks clearly play a role in coordinating healing.^9,70 The identification of OPN5 as a photoreceptive protein target, as well as its spectral absorption properties, represents an important step toward realizing the potential of light- and circadian-driven therapeutics. The intensity of violet light used in our study was set to ensure a steady activation of OPN5 and may exceed what a nocturnal mouse would encounter daily in nature. However, the implementation of light and, specifically, OPN5 activation hold therapeutic potential for the treatment of ocular abrasions in any species that express the photoreceptor. The function of OPN5 can likely be harnessed, both chemically and photically, to modulate corneal surface wound healing and may provide insights into defective wound healing, which may result in blinding corneal scarring. Future experiments will determine the minimum duration and intensity of violet light to optimize healing acceleration with minimum exposure to short-wavelength light.

Limitations of the study

The cornea’s cellular physiology is under the control of the molecular circadian clock. There is a 24 h rhythm to basal mitosis and to wound response in the cornea.^9,71 In this study, we focused on wounds at a uniform circadian phase just before lights on, or “dawn.” In future studies, it would be useful to repeat these studies at alternate phases of the circadian cycle.

In mouse studies, we rely on a Cre-loxP-based reporter system for Opn5 expression, as there is not presently a specific mouse OPN5 antibody. A disadvantage of a Cre-loxP system is that cells are lineage marked and continue to express the reporter once it has recombined. However, the induction of Opn5 from an undetectable baseline in the cornea and the pairing with the detection of Opn5 transcript levels make us confident that the use of a Cre-loxP is appropriate and accurate for this study.

We have focused on mechanical epithelial wounds in this work. Future studies are necessary to determine if corneal issues such as chemical burns, radiation injury, and stromal ulcerations also involve light signaling and Opn5.

RESOURCE AVAILABILITY

Lead contact

Requests for materials and additional information should be directed to the lead contact, Ethan D. Buhr (buhre@uw.edu).

Materials availability

No unique resources or new materials were generated for this work.

Data and code availability

• Bulk RNA-seq data, including metadata, can be accessed at the Gene Expression Omnibus (GEO) by NCBI under the accession number GEO: GSE293197.

• Transcripts with p values adjusted for multiple comparisons at or below 0.05 can be found in Table S2.

• All data will be shared upon request by the lead contact.

• No bespoke computer code was generated for this work.

• Any additional information on the data or procedures reported in this paper will be provided upon request by the lead contact.

STAR★METHODS

METHOD DETAILS

All mouse experiments were done in accordance with the Institutional Animal Care and Use Committee at the University of Washington oversight and directives. All animal procedures conformed to the regulatory standards of that committee. Opn5^Cre; Ai14 mouse line was generated as described in Nguyen et al. 2019.²³ Opn4^−/−; Pde6b^rd1/rd1 mice are as described in Diaz et al. 2020.¹⁹ Opn5^−/− mice were generated as described in Buhr et al. 2015.¹⁴ Male and female mice were used in all experiments and were between 3 weeks and 6 months of age. For RNA sequencing experiments, only male mice were used to minimize transcript differences from sex-related chromosomes, x-inactivation, and because male mice contain both y and x chromosomes. Mice were of the strain C57Bl/6J, backcrossed more than 8 generations from founder lines of 129S1/SvImJ in the case of Opn4^−/−. All animals were group housed and randomly assigned to experimental groups.

All macaque (Macaca nemestrina) tissue was received from the University of Washington Primate Center through their tissue distribution program. No direct experiments were performed on live macaques for this study, and tissue was donated as collateral tissue from other studies.

In most experiments except where noted, 415-nm light is used from an LED array. The following durations were chosen for individual experimental assays.

Measurement parameter	Light duration
Fast molecular event (phosphorylation)	30 min
Slow molecular event (production of nascent RNA or protein)	60 min
Healing of epithelium	2–12 h daily

In vivo epithelial abrasion of the cornea

Mice were anesthetized with isoflurane inhalation and given a subcutaneous injection of 10 mg/kg Carprofen. A drop of NeoPoly (neomycin, polymyxin B, and gramicidin; Bauch and Lomb) ophthalmic solution was applied to each eye. An Algerbrush II with a 0.5 mm burr (Precision Vision) was used to de-epithelialize the cornea of one eye of each mouse. Briefly, a circular abrasion of the corneal epithelium was achieved by 5–6 passes of a spinning Algerbrush burr over a circular region of ~2 mm diameter on the ocular surface. Sterile artificial tears were applied to each eye to prevent dehydration. Animals were observed regularly prior to tissue harvest for signs of discomfort such as scratching or tending to the eye, but none of these behaviors were observed.

For fluorescein imaging, a 0.1% (w/v) solution of fluorescein in Hank’s Balanced Salt Solution (HBSS, Life Technologies) was applied as an eye-drop to the mouse eye and rinsed with sterile artificial tears (Alcon). Eyes were immediately imaged under blue illumination (465-nm LED exposure for approximately 1–2 s) with a digital camera attached to a dissection microscope. An image of the freshly de-epithelialized cornea was taken for each eye and immediately following euthanasia by CO₂ asphyxiation at the indicated time. The size of the fluorescein-stained area for each cornea at terminal collection was compared to the image taken at time of de-epithelialization of the same eye (time 0). Procedures were carried out under dim red-light illumination unless otherwise noted.

Ex vivo epithelial abrasion of the cornea

For ex vivo experiments, mouse corneas were de-epithelialized as described above immediately postmortem after CO₂ asphyxiation. The anterior segments of the eyes were dissected in HBSS, and the lens carefully removed. They were cultured on cell membrane inserts (PICMORG50, Millipore) in phenol-red-free DMEM (Gibco) with 10% FBS, 25 U/mL penicillin, and 25 μg streptomycin (Life Technologies) and maintained in an incubator at 36°C with 5% CO₂. A 3 mm linear incision was made to cell membrane inserts with a scalpel to allow media to fill the interior of the anterior chamber from beneath and to cover approximately 2/3 of the external cornea. The incubator is equipped with an LED array for which individual LED types (400 nm, 415 nm, 475 nm, and 525 nm) can be controlled independently for intensity and duration. Light measurements were made using an SRI-2000 UV spectrophotometer (Allied Scientific). Culture dishes were removed individually daily under dim red light for fluorescein imaging by a digital camera under blue (465 nm LED for approximately 1–2 s).

For fluorescence imaging, K14-Cre; Ai14 mouse corneas were de-epithelialized immediately postmortem. The anterior segment, excluding the lens but retaining the iris/ciliary body, was cultured epithelium-down under a cell culture membrane insert (PICMORG50, Millipore) in phenol-red-free DMEM (Gibco) with B27-supplement (Gibco,17504044), 25 U/mL penicillin, and 25 μg streptomycin (Life Technologies) and maintained in a microscope-stage incubator at 36°C in a dish sealed with parafilm. A fluorescence image was captured every 2 h with a 100-ms, 550-nm excitation pulse. Area of de-epithelialization zone was measured for each image using ImageJ (NIH). A 2-h, 415-nm, 5 × 10¹⁴ photons/cm²/s light was administered once a day for the first 4 days in the light-treatment group.

Macaque anterior segments were kept in ice-cold HBSS, and corneas were de-epithelialized in a zone approximately 1.5 cm in diameter. They were imaged using fluorescein staining (0.1% in HBSS) under blue LED light prior to placement in phenol-red-free DMEM (Gibco) with 10% FBS, 25 U/mL penicillin, and 25 μg streptomycin (Life Technologies) and maintained in an incubator at 36°C with 5% CO₂. The volume of media was adjusted so that the level of media was approximately 2/3 the height of the anterior segment (Figure 5A). The corneas were maintained in darkness or dim red light for handling unless otherwise noted.

For ex vivo light:dark cycles, both mouse and macaque tissue cultures received 12 h daily of 5 × 10¹⁴ photons/cm²/s, 415 nm light.

Quantitative RT-PCR

Whole mouse corneas or 5-mm biopsy punch of macaque corneas were placed in TriReagent (Life Technologies) and frozen at −80° C for storage. RNA was extracted using 1-bromo-3-chloropropane separation followed by isopropanol and 75% ethanol RNA precipitation. RNA was solubilized in molecular-grade water. RNA was converted to cDNA using High Capacity RNA-to-cDNA Master Mix (Applied Biosystems). PCR was performed on an ABI 7500 Fast machine (Applied Biosystems) using a 2-step cycle. Transcript abundance was compared to β-actin or Hrpt1 as endogenous controls using 2^−ΔΔCt method.

Western blot

Corneas were quickly frozen on dry ice and stored at −80°C. Tissue was disrupted in Tris lysis buffer with 0.1% SDS and 1% Triton X-. 4–12% Bis-Tris SDS-PAGE gels (Life Technologies) were run with 20 μg of total protein for each sample. Proteins were transferred to nitrocellulose membranes in Tris-glycine transfer buffer containing 20% methanol. Nitrocellulose membranes were stained with phospho-p44/42 ERK 1/2 (Cell Signaling #9101), total ERK1/2 (Cell Signaling #4695), α-tubulin (Cell Signaling #3873). Note, α-tubulin was used as a loading control for macaque tissue because the total ERK 1/2 antibody used for mice produced a non-specific band in monkey tissue. 2° antibodies consisted of anti-rabbit or anti-mouse fluorescent antibodies (LiCor). Fluorescence was imaged on an Odyssey CLx (LiCor) blot scanner. Expression levels were measured using ImageJ software (NIH) and phospho-ERK levels were compared to total-ERK or α-tubulin from the same protein samples.

Bulk RNA-seq

Corneas were collected 12-h after de-epithelialization in DNA/RNA Shield (Zymo Research R1200–25). RNA was extracted using Quick-DNA/RNA Microprep Plus Kit (Zymo Research D7005), and quality was assessed with Biolanalyzer 2100 and RNA 6000 Nano (Agilent) to ensure only samples with RIN >8 were accepted. Library preparation and RNA sequencing was processed by DNBseq (BGI, Shenzhen China). RNA sequences obtained were analyzed using Galaxy online tools.⁷² The analysis workflow used is the following: sequences were aligned using HISAT2, mapped and assembled with StringTie, differential expressions were analyzed using DESeq2 with a false discover rate <0.5.

Immunofluorescence

Tissues were fixed in 4% para-formaldehyde overnight before transfer to a 30% sucrose solution in phospho-buffered saline (PBS). Tissues were embedded in OCT and sliced on a cryostat slicer at −20°C. Slices were washed with PBS and stained with 1° antibodies of the indicated type overnight at 4°C. They were then washed with PBS and stained with 2° antibodies for 1 h. They were imaged with a Leica SP8 confocal system on a Leica DM6000 microscope.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistics were performed using Sigma Plot 11.0 (Systat Software).

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116045.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Phospho-ERK 1/2 p44/42	Cell Signaling	Cat#9101; RRID:AB_331646
ERK1/2	Cell Signaling	Cat#4695; RRID:AB_390779
α-tubulin	Cell Signaling	Cat#3873; RRID:AB_1904178
EGR1	Proteintech	Cat#55117-1-AP; RRID:AB_2881272
Anti-rabbit 800 CW	LiCor	Cat#926-32211
Chemicals, peptides, and recombinant proteins
OCT	TissueTek	Cat#4583
Donkey serum	Jackson Immuno	Cat# 017-000121
DMEM	Gibco	Cat#90-013-PB
B27 Supplement	Thermo Fisher	Cat#0080085SA
D-Luciferin potassium salt	Biosynth	Cat#L-8220
Hanks Balanced Salt Solution	Thermo Fisher	Cat#14025076
9-cis-retinaldehyde	Simga	Cat#R5754
all-trans-retinaldehyde	Sigma	Cat#R2500
RNAlater	Qiagen	Cat#76104
Tri reagent	Thermo Fisher	Cat#AM9738
Critical commercial assays
High Capacity RNA to cDNA kit	Applied Biosystems	Cat#4387406
Absolute Blue QPCR mix	Thermo Fisher	Cat#AB4163A
Experimental models: Organisms/strains
Period2::Luciferase, B6.129S6-Per2tm1Jt/J	Jackson Laboratories	Stock # 006852
Opn5−/−	Díaz et al.19	N/A
Opn5Cre; Ai14	Rollag et al.57	N/A
K14Cre	Jackson Laboratories	Stock # 018964
B6; 129S6-Gt(ROSA)26Sortm14(CAG-tdTomato)	Jackson Laboratories	Stock # 007908
Opn4−/−;Pde6brd1/rd1	Sugiyama et al.27	N/A
Oligonucleotides
Primers for RT-PCR see Table S1	This paper	N/A
Software and algorithms
ClockLab	Actimetrics	https://www.actimetrics.com/products/clocklab/
Lumicycle Analysis	Actimetrics	https://www.actimetrics.com/products/lumicycle/lumicycle-32/
ImageJ	NIH	https://imagej.nih.gov/ij/
Sigma Plot 11.0	Systat	systatsoftware.com
Deposited data
RNA sequence data	NCBI-GEO	GSE293197
Other
Zeiss LSM 700 confocal microscope	Zeiss	LSM 700
Abi7500 Fast Real-time PCR machine	Abi	7500 Fast
Quantum radiometer	Macam	Q203
Lumicycle luminometer	Actimeterics	Lumicycle 32
Retiga Lumo CCD camera	Q-imaging	Retiga Lumo
Microscope stage incubator	Bioscience tools	Cat#TC-MWPHB
Millicell cell culture inserts	Millipore	Cat#PICMORG50

Highlights.

• The corneas of mice and primates express an opsin photoreceptor, Opn5, in response to injury

• Cells within the cornea become temporarily sensitive to short-wavelength light during healing

• Opn5-dependent pathways regulate wound healing

• Corneal wound healing is accelerated by violet light