Encephalopsin (OPN3) is required for normal refractive development and the GO/GROW response to induced myopia

Lead contact and materials availability

No new reagents or mouse lines were generated for this study. Further information and resource requests should be directed to the lead contact, Richard A. Lang, Division of Pediatric Ophthalmology, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Avenue, Cincinnati OH 45,229, Tel: 513–636–2700 (Office), 513–803–2230 (Assistant), FAX: 513–636–4317. Email: richard.lang@cchmc.org.

Experimental model and subject details

Lighting conditions

Animals were housed in standard fluorescent lighting (photon flux 1.62 × 10¹⁵ photons cm−2 s−1) on a 12L:12D light/dark cycle until weaning, when they were transferred to full-spectrum lighting. For full-spectrum lighting, LEDs were used to yield a comparable total photon flux of 1.68 × 10¹⁵ photons/cm²/sec. The spectral and photon flux information for the LED lighting is as follows: near-violet (λ_max=380 nm, 4.23 × 10¹⁴ photons/cm²/sec in the 370–400 nm range), blue (λ_max=480 nm, 5.36 × 10¹⁴ photons/cm²/sec in the 430–530 nm range), and red (λ_max=630 nm, 6.72 × 1014 photons/cm²/sec in the 600–700 nm range). Photon fluxes were measured at approximately 61 cm from the source through an empty standard mouse cage.

Tissue processing

Retinal tissue was processed as previously described [42]. Briefly, mice were anesthetized with isofluorane, followed by cervical dislocation. Then, eyes were enucleated and fixed for 30 min in 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS), rinsed, and stored at 4^◦C until use. Whole-mount retinae were processed as previously described, using the same method of marking the orientation with a mark at the ventral pole before dissection for topology studies. Sectioned eyes were processed similarly, also as previously described [43].

Staining

Similarly, retinae for immunofluorescence (IHC) were again incubated in 5% PBST and 50% acetone for antigen retrieval and then washed, blocked, and incubated for 3 to 5 days in primary antibody as previously described. Primary antibodies included those against chicken anti-GFP (1:1000, abcam ab290, RRID:AB_303395, Cambridge, UK, abcam.com), rabbit anti-dsRed (1:1000, TakaraBio 632,496, RRID:AB_10013483, Kusatsu, Shiga, Japan, takarabio.com), rabbit anti-Rbpms (1:500, Abcam ab152101), mouse anti-calbindin (1:300, Abcam ab82812, RRID:AB_1658451), rabbit anti-melanopsin (N15, a gift from Ignacio Provencio), and mouse anti-Calretinin (1:300, Sigma Aldrich MAB1568, St. Louis, MO) [43].

RGC overlap assessment

Flat-mount retinal preparations from Opn5^cre; Ai14; Opn3-eGFP mice were imaged using a Nikon A1R confocal microscope. High-resolution large-tiled images corresponding to the superior and inferior regions of the retina were acquired from three animals. Tiles were subdivided into multiple subfields for cell type analysis. Briefly, images were imported into ImageJ (v1.53), and cells were manually characterized into one of seven groups based on the co-expression of tdTomato (reporting the Opn5-lineage), GFP (reporting Opn3 expression), and OPN4 (melanopsin). Cell type assignment was performed using the CellCounter plugin in ImageJ.

Refractive assessment

All mouse lines were tested for normal refractive development and lens-induced myopia (LIM) and measured weekly to quantify refraction and ocular parameter changes. For our LIM model, mouse refraction and ocular parameters were measured at 3 weeks, and then a head pedestal was attached and goggles mounted the next day. Each head pedestal fits one −30D lens over the right eye and a 0D lens over the left eye and was generously provided by Toshihide Kurihara of Keio University [29].

On the day of refractive assessment, five minutes before assessment, the mice had their eyes dilated with 1% tropicamide and were anesthetized with either ketamine (60 mg/kg)/xylazine (7 mg/kg; mixed background) or ketamine (90 mg/kg)/xylazine (10 mg/kg; C57Bl/6J) depending on the strain background. To better preserve the essential tear layer but reduce mouse movement, mice were scruffed, and their refraction was quantified using infrared autorefractometry (Photorefractor, Stria-Tech, Tubingen, Germany). This was performed before full sedation but after the mice lost their equilibrium to prevent disturbing the tear film.

After the mice were fully anesthetized and no longer responsive to a toe pinch, biometric measurements of the mouse eye were taken using a spectral domain optical coherence tomography (SD-OCT) imager (Envisu R-Class, Bioptogen-Leica Microsystems, Wetzlar, Germany, leica-microsystems.com). Eye drops were provided as needed throughout imaging and during recovery, and the mice were kept warm using a heating pad. Ocular parameters included corneal thickness, depth of aqueous, lens thickness, vitreous chamber depth, retinal thickness, axial length, and choroidal thickness. After imaging, the anesthetic was reversed using 5mg/kg atipamezole.

The Opn3 germline and the mRx-cre; Opn3^fl/fl lines had slightly different backgrounds due to the method with which they were generated (see “Mice” for more detail). Previous studies have established line-dependent differences in refractive development [44]. This may explain the slight difference in baseline refraction between Opn3^fl/fl and Opn3^+/+ mice.

Head pedestal surgery

The mice were anesthetized with ventilated isofluorane (induction: 3%, maintenance: 1%–2%), and the top of their heads were shaved. The exposed skin was then cleaned with 3% hydrogen peroxide, and a 1 cm by 1 cm region of skull was exposed. The region was again sterilized with 3% hydrogen peroxide, and the exposed fascia and periosteum were removed. The area was then dried, and dental acrylic (Ortho-Jet, Lang Dental, Wheeling, IL) was used to bond the head pedestal to the mouse to which lenses could be attached. The mice were provided carprofen-medicated diet gel water for three days for analgesia.

GO/GROW RT-qPCR Analysis

The animals were anesthetized with isofluorane followed by cervical dislocation as a secondary form of euthanasia at the same time of day and within a two-hour period to minimize circadian variability. The retina and RPE were then harvested and immediately snap frozen and homogenized in TRI Reagent (Invitrogen, Waltham, MA) using 2.0mm RNase-free zirconium oxide beads in the TissueLyser II sample disrupter (Qiagen, Hilden, Germany). Phase separation was achieved using chloroform, and RNA in the aqueous phase was precipitated using ethanol. The resultant product was then purified using a modified GeneJET RNA purification column (ThermoFisher Scientific #K0732) protocol and eluted into RNase-free water. The RNA was then treated with RNase-free DNase I (ThermoFisher Scientific #EN0521), and cDNA was synthesized using a Verso cDNA synthesis kit (ThermoFisher Scientific AB1453/B). Quantitative RT–PCR was performed with Radiant SYBR Green Lo-ROX qPCR mix (Alkali Scientific Inc.) using a ThermoFisher QuantStudio 6 Flex Real-Time PCR system.

The GO/GROW genes list was adapted from a list developed by He et al. [45], which is a compilation of genes whose expression is changed in a variety of myopia-induction animal models. Validated primers for the relevant GO/GROW signature genes were then obtained from the Harvard PrimerBank [46]. The primer sequences are listed in the Table 2.

Tissue processing, sectioning, and immunofluorescence

The animals were anesthetized with isofluorane and cervically dislocated as a secondary form of euthanasia

Data analysis

All data processing was performed using GraphPad Prism v8 (San Diego, CA). Ocular measurements were quantified using MouseSwimmerV2, a MatLab program generously provided by Machelle Pardue of Georgia Tech/Emory University. Then, a two-way ANOVA with Šidák multiple comparisons was performed to compare Opn3 mutant animals to controls, with and without goggling, at multiple time points. For the refractive development analysis, the two eyes were averaged, as both eyes received the same treatment. For the goggling experiments, the effect of −30 diopter lenses was calculated by determining the “shift” in the difference in measurement between the −30 diopter (right) and the 0 diopter (left) eye. To eliminate inter-subject variability, all ocular and refractive measurements were normalized to their initial baseline values before calculating the shift.

For the RT-qPCR experiments, the relative expression of GO/GROW signature genes was calculated using the ^ΔΔCt method by first normalizing to a control gene (Tbp) run on the same plate and then determining the relative expression of the target gene versus the geometric mean of the control eyes. Next, paired t tests were used to assess differences between goggled and non-goggled eyes. All data are expressed as an average ± standard error of the mean, with the significance labeled in figure legends. A p-value <0.05 was considered statistically significant.

Opn3 is expressed in retinal ganglion cells and overlaps with OPN4 expression

We used the Opn3-eGFP reporter line to assess the expression of the Opn3 gene in the mouse eye. This reporter was validated by cross-referencing the expression pattern in neural tissue [35] with that of the Opn3^::mCherry knock-in reporter [47], in which mCherry is fused to the C-terminus of OPN3. High correspondence indicates that Opn3-eGFP faithfully reports Opn3 transcript expression. Although the direct fluorescence of eGFP from this reporter was detectable, we used an anti-eGFP antibody to increase the signal strength.

Retinal ganglion cells (RGCs) were observed to express Opn3-eGFP (Figure 1A–C). At birth, Opn3 expression is observed throughout the ganglion cell layer (identified with the ganglion cell marker RBPMS; Figure 1A, P0.5). However, by P8.5 and P12.5, the proportion of retinal ganglion cells that express Opn3 had diminished and was restricted to a few cells (Figure 1B,C). Labeling for eGFP at P28 in flat-mount retinas showed a 98.1% overlap with the pan-RGC marker RBPMS (Figure 1D,F). Calbindin and calretinin were found in 42.2% and 20.3% of Opn3-eGFP expressing cells, respectively (Figure 1E-G). Interestingly, a significant expression overlap was observed between OPN4 (by antibody labeling) and Opn3-eGFP (Figure 1F,G,H), with 23.9% of Opn3-eGFP cells immunolabeling for OPN4. OPN4 RGCs have six morphological subtypes with distinct electrophysiological properties and functions [48-53]. Analysis of the morphology of OPN4 immunoreactive cells and the overlapping expression of OPN4 and Opn3-eGFP revealed that the M1 subtype that plays a primary role in circadian entrainment and the pupillary light reflex [54] does not express Opn3 (Figure 1I). However, there is substantial overlap of Opn3 expression with the M2 and the M4 subtypes of OPN4 RGCs (Figure 1I). The population of OPN4 RGCs, and the M2 subtype specifically (as well as M1 and M3 populations where Opn3 is not expressed), have been implicated in refractive development and induced myopia responses [27,28]. Outside of the RGCs, Opn3-eGFP also showed that Opn3 is expressed in a distinct layer of cells just outside the retinal pigmented epithelium (RPE, Figure 1A,B,J). These cells are within the choriocapillaris, a vascular component of the choroid that supports the RPE and photoreceptors. Opn3 choriocapillaris expression increases with development from P0 to P8.5.

Opn3 expression in RGCs does not significantly overlap with expression of neuropsin/Opn5

It has been shown previously that neuropsin/Opn5 is expressed in retinal ganglion cells [14] and, like OPN4, is involved in the induced myopia response in mice [55]. For this reason, we assessed the relationship between the expression of Opn3 and Opn5 in the retina. Opn5 is expressed at low levels but has been detected successfully using an Opn5^cre allele to convert a cre recombinase reporter. Here, we used mice of the genotype Opn5^cre; Rosa26^Ai14; Opn3-eGFP and labeled flat-mount retinas for eGFP and OPN4 (Figure 2A-J). We then assessed 3306 RGCs (n=3 animals) for expression overlap between the three opsins at P28 (Figure 2K,L). This revealed that, of the opsin-expressing RGCs, Opn5 cells were the most abundant (Figure 2K, L). Opn5 RGCs rarely showed expression overlap with Opn3 or OPN4 (Figure 2K,L). However, the significant expression overlap between Opn3 and OPN4 was confirmed (Figure 2K,L). These data suggested that Opn5 RGCs were likely to function as a distinct population but that we would need to consider whether the co-expression of Opn3 and OPN4 might have functional consequences.

Deletion of retinal Opn3 does not influence the number of ipRGCs or OPN4 expression

Our goal of assessing the function of OPN3 in refractive development and induced myopia could be confounded if deletion of Opn3 in RGCs had an impact on the number of intrinsically photosensitive retinal ganglion cells (ipRGCs) or their expression of OPN4. To assess this, we first confirmed that mRx-cre was an effective retinal cre driver by crossing it with the cre reporter Ai14 and assessing the distribution of tdTomato expressing cells. This showed the expected broad, cre-dependent conversion in retinal neurons (Appendix 1). We then analyzed OPN4 expression in mice with a retina-specific knockout of Opn3 (Opn3^fl/fl; mRx-cre). In retinae from these animals, we quantified the ipRGC number based on OPN4 immunolabeling in the ganglion cell layer as well as displaced cells in the inner nuclear layer (Appendix 1). A comparison between control and Opn3 conditional deletion retinae revealed no difference in OPN4 immunoreactivity (Appendix 1) or in the numbers of ipRGCs (Appendix 1). This indicated that despite the existence of Opn3/Opn4 double-positive RGCs, Opn3 could be studied independently of Opn4.

Opn3 null mice have abnormal eye growth that leads to mild myopia

To assess whether Opn3 might play a role in refractive development of the eye, we performed ocular biometry and refractive assessment on mice that were Opn3 germline null (Opn3^ΔEx2/ΔEx2). This was performed over a time course between postnatal weeks 3 and 9. Using spectral domain optical coherence tomography (SD-OCT), we measured the depth of the aqueous (Figure 3A), pole-to-pole lens thickness (Figure 3B), corneal thickness (Appendix 2), retinal depth (Appendix 2), choroidal thickness (Appendix 2), vitreous chamber depth (Appendix 2), and the axial length from the outer corneal surface to the RPE–choroid boundary (Figure 3C) [55]. The eyes of both control and Opn3 null mice are increasing in size over this period, commensurate with postnatal growth.

Biometry assessments showed that the Opn3 germline null mice had both a reduced aqueous depth (Figure 3A) and a thinner lens (Figure 3B). These dimensional changes accounted completely for the reduced axial length that was also apparent in the Opn3 null (Figure 3C). Consistent with this, the Opn3 null did not show any significant change in the thickness of the cornea (Appendix 2), retina (Appendix 2), or choroid (Appendix 3) or in the depth of the vitreous chamber (Appendix 2). The changes to the Opn3 null are summarized in the chart. These measurements document a role for OPN3 in the development of the anterior eye.

Assessment of the refractive performance of the Opn3 germline null eye revealed mild myopia (Figure 3D). From 3 to 4 weeks of the time course, the refraction of control eyes increased from −5 to +13 diopters (Figure 3D). In contrast, the refraction of Opn3 germline null mice increased only slightly from 3 to 5 weeks and then became more negative thereafter. Over the length of the time course, on average, the Opn3 null showed a reduction in refraction of about −10 diopters by weeks 5–8 (Figure 3D). The mild myopia of the Opn3 null eye was unexpected because a reduced axial length would normally lead to hyperopia.

The myopia of Opn3 null mice cannot be attributed to retinal OPN3

The process of emmetropization that drives refractive development in the eye is known to be dependent on light detection by photoreceptors [56]. In all cases assessed so far, it is the retinal photoreceptors that mediate light sensing [24,26,27,55,57]. To determine whether the retina was the domain of Opn3 expression crucial for the phenotype described in the Opn3 germline null, we generated retinal conditional null mice by combining the mRx-cre driver with the Opn3^fl allele. These generated cohorts of control (Opn3^fl/fl) and experimental (mRx-cre; Opn3^fl/fl) mice were subjected to biometry and refraction as described above. None of the measured features showed any statistically significant change (Figure 3E-H, Appendix 2). This suggests that RGC Opn3 is not the expression domain that explains the phenotype of the Opn3 germline null.

Both germline and retina conditional Opn3 null mice are susceptible to induced myopia

The mechanisms underlying myopia pathways can be assessed using induced myopia. Given the refractive development changes observed in Opn3 null mice, it was interesting to assess their susceptibility to LIM. This analysis was performed using external lenses mounted on a head post according to an established method [21,55,58]. For each mouse, a zero-diopter lens was mounted in front of the control eye (left) and a −30-diopter lens in front of the experimental (right) eye. Lenses were mounted at 3 weeks of age, and biometry and refraction measurements were subsequently taken over a 3-week time course. The analysis was performed for control and experimental cohorts of both germline null (Opn3^ΔEx2/ΔEx2) and retina conditional null (mRx-cre; Opn3^fl/fl) mice. Data were normalized to the 3 weeks of age time point (t=0) so that the control eye growth response and the experimental eye response to refractive mismatch could be compared.

This analysis showed that Opn3 germline and conditional mutant mice both responded to the myopiagenic stimulus, albeit with subtle differences. The axial length (Figure 4A, D), corneal thickness (Appendix 3), aqueous depth (Appendix 3), lens thickness (Appendix 3), retinal thickness (Appendix 3), refraction (Figure 4B,E), and choroidal thickness (Figure 3C,F) in cohorts of both Opn3 mutant genotypes were statistically indistinguishable from control cohorts according to a two-way ANOVA analysis. However, there were notable deviations from controls in some parameters, including a less robust refractive shift and reduced choroidal thinning in Opn3^ΔEx2/ΔEx2 mice (Figure 4B,C, compare blue and black dashed lines) and retinal change that was thinner in Opn3^ΔEx2/ΔEx2 mice but thicker in mRx-cre; Opn3^fl/f mice (Appendix 3 compare blue dashed in D with blue dashed line in H). In summary, loss of function in Opn3 does not prevent an induced myopia response but appears to subtly change some features of the response.

The myopia GO/GROW myopia gene signature is partly OPN3 dependent

Multiple studies with a variety of animal models have identified a gene signature associated with both form-deprivation myopia [59] and LIM [60]. The gene signature, referred to as GO/GROW, is evident in genes that are expressed in the retina [61], RPE [45,62], choroid [63], and sclera [64] and is thought to correspond to the cellular recognition of refractive mismatch and the response in ocular growth. Despite extensive characterization of this gene expression signature in chicks [65-67] and to a lesser degree in guinea pigs [68], tree shrews [69,70], and monkeys [71,72], GO/GROW has not been explored in mice or in the context of genetic models of myopia.

In evaluating the GO/GROW gene signature in this study, we had the dual goals of assessing whether this gene signature was useful in mice and whether any of the genes were Opn3 dependent. We chose to evaluate a selection of GO/GROW signature genes with a range of functions (transcription factors, secreted factors, membrane proteins; Table 2) at 24 h after myopia induction, a common time point in previous studies using other animal models. We performed this analysis comparing the GO/GROW gene signature in wild-type and Opn3 null mice, as this was the mutant that demonstrated a unique myopia phenotype and reduced myopic shift in response to LIM.

The analysis showed that some of the GO/GROW signature genes had a distinct response in the Opn3 null germline (Figure 5A). When compared to the zero-diopter eye, the retina from the −30-diopter eye of Opn3 null showed a significant downregulation of Ctgf/Ccn2 (Connective Tissue Growth Factor/Cellular Communication Network Factor 2), Cx43/Gja1 (Connexin 43/Gap Junction A1), and Egr1 (Early Growth Response 1), with the latter two genes showing a reversed polarity of gene expression change compared with controls (Figure 5A, compare blue and gray bars). An additional group of genes (Rarb, Penk and Fgf10) showed a statistical trend in expression change. This analysis indicates that a subset of GO/GROW signature genes are dependent on OPN3 activity for their expression change in response to refractive mismatch.

Refractive development and induced myopia

The minor role of GWAS-associated genetic variants [73] in the myopia boom [4] and documented environmental risk factors, including outdoor light exposure [74] and focus [75], in myopia development make opsins a prime candidate for the sensors of environmental lighting cues that mediate emmetropization. It is therefore unsurprising that refractive phenotypes have been documented in every opsin signaling pathway studied [23-25], including the non-canonical opsins. Loss of melanopsin/Opn4 function or of the ipRGCs that express Opn4 can result in early myopia and late hyperopia [27] or persistent hyperopia [28]. Neuropsin/Opn5-dependent violet light stimulation is protective against induced myopia [55] and can explain why sunlight is protective against myopia [74]. Encephalopsin/Opn3 is widely expressed [32,33,35] but only has a confirmed light-sensing function in adipocytes [17,27,76], with light responses in the colon [77], airway [37], and uterus [38] suggested to be OPN3 dependent. Until this study, the role of Opn3 in the eye had not been investigated.

We document a myopic phenotype in the germline Opn3 null that is not present in the retinal conditional null. This phenotype was characterized by short axial length, reduced aqueous compartment depth, and a thinner lens, but preserved vitreous depth, with a reduced axial length resulting from the anterior eye changes. Despite this short axial length, Opn3 null eyes demonstrate axial elongation in response to the refractive mismatch of LIM, although the resultant myopic shift is less robust than that of the control. The pattern of refractive development and response to LIM in the Opn3 germline null prompts comparison to refractive development of the Opn4 null. This been reported by Chakraborty et al. [27] to have a reduced anterior chamber depth abnormal lens placement and, in a Cre/Cre null, to have a shorter axial length [28]. Further, Liu et al. reported that their null had a reduced myopic shift, like the Opn3 germline null, but with impaired axial elongation.

However, there are also distinct differences between these opsin mutants. Chakraborty et al. [27] reported that the Opn4 null exhibits slowed axial growth (rather than stable axial length differences) and no differences in lens thickness, resulting in a different refractive phenotype of early myopia and subsequent hyperopia, while Liu’s [28] Cre/Cre null is hyperopic throughout development, neither of which match the pattern of the Opn3 germline null. Further, despite the overlapping expression of Opn3 and Opn4 in a proportion of M2 and M4 ipRGCs, the failure of the retina-specific deletion of Opn3 results in any refractive development phenotype indicates that the crucial domain of Opn3 expression for this response is not a retinal neuron. Many other domains of Opn3 expression can be assessed [32,33,35], including the choriocapillaris domain documented here.

Myopia has different classifications based on the source of refractive error. Refractive myopia results from changes in anterior power, while axial myopia results from axial elongation [1]. Many human myopes, however, manifest characteristics of both types [78]. The Reykjavik Eye Study [79] was one of the first studies to report that corneal and lens power characteristics as well as axial length constitute total refractive error, thus underlining the importance of examining myopic progression in its totality. Much of the animal model research on myopia has focused on genetic models or myopia induction models that result in eye elongation, and thus much attention has been focused on axial length. The Opn3 myopia model highlights the importance of examining ocular dimensions as a whole and provides an animal model for myopia where elongation is not the predominant source of myopic refractive error.

Opn3 germline null animals have a thinner lens, a change that can be associated with myopia development [79-82]. However, when the lens–myopia relationship is explored, an inverse relationship is often found between lens power (correlated with lens thickness) and refractive error/axial length changes in individuals who become myopic earlier in life [83]. For those who develop refractive error later in life, refractive changes are more often a result of changes in anterior focusing, and thus the relationship between lens power and refraction is usually direct [84]. This phenomenon of anterior hyperopia in young myopes is thought to be a counter to myopic axial elongation and has alternatively been explained as an active response to increasing axial length or as a passive mechanical stretching [85], which results in a compensatory reduction in lens power until the lens can no longer compensate [86,87].

However, there is not always a connection between lens thickness and power, as demonstrated by the “lens paradox” of increasing lens thickness over the human lifetime with no corresponding increase in lens power, underlining how changes to both the external structure of the lens (curvature and thickness) and internal refractive gradient influence lens power [84]. As a result, the refraction of a lens is more complicated than its thickness. Some ocular modeling studies have found that both a short anterior chamber [87] as well as, counterintuitively, a thinner, but not posteriorly displaced lens, are both predictive of myopic refraction in the anterior segment [88]. Nonetheless, it is important to consider that most of this research and modeling has been done assuming human eye optics, which may not be generalizable to mice. Further, corneal curvature was not assessed in this study and could be contributing to the overall myopic refraction of the eye.

It is perhaps significant that the features of the Opn3 germline eye share similarities to myopia of prematurity. Myopia of prematurity results in a shallower anterior eye, a more anteriorly situated lens, and, depending on the study, a shorter axial length. However, myopia of prematurity also involves increased corneal curvature (which we did not assess) and a thicker lens, in contrast to the Opn3 germline’s thinner lens. It is believed that this congenital form of myopia may result from inhibition of anterior eye growth [89-91]. These similarities may suggest that the Opn3 pathway could be modified in myopia of prematurity.

The GO/GROW gene signature

The GO/GROW gene signature was identified as a response to both form deprivation and LIM in multiple species. In this study, we examined whether GO/GROW gene expression changes could be documented in mice and whether any changes might be dependent on the activity of OPN3. Notably, several the GO/GROW genes, Ctgf, Cx43, and Egr1, all showed significant OPN3-dependent changes in expression 24 h after initiation of a refractive mismatch. Identification of OPN3-dependent GO/GROW gene expression changes will ultimately allow us to better understand the mechanism of action of OPN3 in refractive development.

Ctgf is known to be expressed in the mouse retinal vasculature, specifically vascular endothelial cells, where it works as part of a Hippo pathway regulatory loop to maintain vascular development and blood–retinal barrier function [92] Further, Ctgf has also been investigated for its role in binding and inhibition of VEGF, forming what has been called the “fibro-angiogenic switch,” which regulates neovascularization after retinal injury [93]. Ctgf has been studied in the context of myopia and glaucoma in relation to choroidal neovascularization and the myofibroblastic phenotype with increased ECM deposition seen in severe disease [94,95]. In a previously published study on the retinal GO-GROW signature in tree shrews, Ctgf expression was found to be significantly downregulated in response to myopia induction after 6 h, following the trend of our Opn3 mutants [96].

The connection of CX43 to myopia has also been documented, particularly in a clinical context, as missense mutations in the gap junction gene are known to cause oculodentaldigital displaysia, which results in a spectrum of disease affecting the teeth, limbs, eye, and ocular adnexa [97]. Various ocular phenotypes have been reported with these mutations, including syndromic myopia with a short axial length, a shallow anterior chamber, and glaucoma [98]. Cx43 has also been a candidate gap junction in studies on the transmission of epigenetic signals through siRNA as a mediator of neuronal regeneration and modulation, and it was found to have a unique ability to transmit 12–24 nucleotide-long sequences when compared to Cx32 and Cx26 [99]. This has been suggested as a potential mode of transmission for cellular response to myopia development as well as other diseases, including retinal vascular disease, macular degeneration, and retinal damage in traumatic brain injury (TBI), the last of which is characterized by Cx43 downregulation [100].

Egr1/Zenk has long been the subject of interest in myopia research since it was first identified in chicks [101] as an immediate early gene response to goggling. Egr1/Zenk is upregulated under induced hyperopia and downregulated under induced myopia [102]. Later, along with the nob (Nyx null) mouse model [103], the Egr1 null became one of the first genetic models of mouse myopia [104]. Opn3 null downregulation of Egr1 therefore follows the induced myopia pattern seen in these previous studies as well as that reported after 6 h in the study by He et al. [45]

Conclusions

This analysis shows that the opsin family GPCR encephalopsin/OPN3 is required for normal refractive development of the eye and that a non-retinal domain of expression is involved. The analysis also shows that several of the GO/GROW refractive mismatch signature genes are changed in an OPN3-dependent manner. Thus, this work identifies the role of opsin family GPCR encephalopsin/OPN3 in refractive development and the response to refractive mismatch. This work adds OPN3 to the list of opsin family GPCRs (in the mouse, OPN4 [27,28], OPN5 [55], the cone opsin (OPN1SW/OPN1MW) pathways [26], and OPN2 [24]) that are implicated in emmetropization and myopia. The exclusion of retinal Opn3, a domain crucial for the refractive development phenotype, suggests a distinct mechanism of involvement of the other opsins. Future work could address this question as well as the molecular mechanism of action of OPN3, including an investigation of the light dependence of this phenotype and the developmental periods of refractive development that require OPN3.