Highly sensitive visual restoration and protection via ectopic expression of chimeric rhodopsin in mice

Summary Photoreception requires amplification by mammalian rhodopsin through G protein activation, which requires a visual cycle. To achieve this in retinal gene therapy, we incorporated human rhodopsin cytoplasmic loops into Gloeobacter rhodopsin, thereby generating Gloeobacter and human chimeric rhodopsin (GHCR). In a murine model of inherited retinal degeneration, we induced retinal GHCR expression by intravitreal injection of a recombinant adeno-associated virus vector. Retinal explant and visual thalamus electrophysiological recordings, behavioral tests, and histological analysis showed that GHCR restored dim-environment vision and prevented the progression of retinal degeneration. Thus, GHCR may be a potent clinical tool for the treatment of retinal disorders.


INTRODUCTION
Inherited retinal degeneration (IRD) is a major cause of vision loss.More than 2 million people worldwide are blind due to IRD, 1 and few effective treatments exist.3][4][5][6][7][8][9] In addition, clinical trials are under way to investigate the effects of introducing channelrhodopsin 2 (RST-001, ClinicalTrials.govIdentifier: NCT01648452) and Chrim-sonR (GS-030, ClinicalTrials.govIdentifier: NCT03326336) into retinal ganglion cells (RGCs) via gene transduction achieved by intravitreal injection of recombinant adeno-associated virus (rAAV).The first clinical case report on optogenetic therapy was recently reported. 102][13] They cannot restore vision in dimly lit environments, such as indoors or at night, and strong light irradiation can promote retinal degeneration. 14,15Physiological photoreception mediated by mammalian rhodopsin, however, relies on amplification through G protein activation.Although the introduction of vertebrate opsin improved photosensitivity in mice, 9,16 it is unclear how the chromophore retinal is metabolized in the retina where the visual cycle is broken.Animal rhodopsin also causes toxicity if all-trans retinal is not properly metabolized, 17,18 and is, thus, hampered by safety and stability concerns in terms of clinical application.
Because of the previous limitations of animal visual opsins, one attempt to circumvent them is the chimeric rhodopsin of melanopsin and G protein-coupled receptor (GPCR). 8,19Melanopsin is a non-visual opsin, and despite being an animal opsin, it is not easily photobleached.However, it has a ''bistable'' photo-cycle and requires different wavelengths of light for conformational change, which may result in unnatural appearance. 20,21herefore, a chimeric rhodopsin of microbial opsin and GPCR, [22][23][24] is not photo-bleached and is a monostable pigment like visual opsin, but may be able to achieve highly sensitive visual restoration via G protein stimulation.
In this study, to achieve light sensitivity, stability, and safety, we attempted to restore vision in mice using Gloeobacter and human chimeric rhodopsin (GHCR). 23,24

Design of GHCR
Although there is no sequence identity between microbial and animal opsin, both possess similar chromophore (retinal) and protein (seventransmembrane helix) structures.As we previously reported, 24 to generate GHCR, we replaced the second and third intracellular loops of Gloeobacter rhodopsin with human sequences and introduced the E132Q mutation (Figure S1).Previous work has shown that GHCR induces G protein activation in vitro. 24storing light-evoked activity in the retina with GHCR We injected a viral vector (rAAV-DJ or rAAV-2) containing the GHCR coding sequence under the control of the hybrid promoter comprising the CMV immediate-early enhancer, CBA promoter, and CBA intron 1/exon 1, known as the CAGGS promoter, (CAGGS-GHCR; Figure 1A) into the vitreous humor of 10-week-old rd1 mice.We adopted the rAAV-DJ vector to achieve more efficient, widespread gene transfer, 25,26 and used rAAV-2 as a benchmark, as it has already been used in the clinic. 27The retinas were harvested 2-4 months later.Enhanced green fluorescent protein (EGFP) reporter gene expression was observed in both the ganglion cell layer and the inner nuclear layer (Figures S2A and  S2B).To evaluate the function of ectopically expressed GHCR in the mouse retina, we performed multi-electrode array (MEA) recording to record the extracellular potential of RGCs.As a result of photoreceptor degeneration, the untreated control retina showed no RGC response as detected by MEA (Figure 1B).In contrast, the treated retinas showed obvious light-induced responses down to 10 14 photons/cm 2 /s of white light-emitting diode (LED) irradiation (Figure 1C).
Next, to create a stable vector for human gene therapy, we designed a codon-optimized version of GHCR (coGHCR) and fused the ER2 endoplasmic reticulum (ER) export signal to its C-terminus to increase gene expression levels.Immunolabeling revealed expression across the whole retina, including in the bipolar cells, of treated rd1 mice (Figure 1D).As a result, the firing rate increased significantly, and a photoresponse was confirmed down to 10 13 photons/cm 2 /s, which had not observed before optimization (Figures 1E, 1F, and S2C).The retinas of WT mice were highly responsive to all light stimulus levels under dark-adapted conditions, but under light-adapted conditions, the firing rate was also modulated in response to light stimulus intensity, and coGHCR response was similar to the light-adapted conditions in WT mice (Figure S2D).No photoresponse to any light stimulus level was obtained from control untreated mice.Moreover, the number of firing cells per unit area also increased significantly (Figure 1G).Since rhodopsin shows selectivity for Gi/o class G proteins upon heterologous expression, [28][29][30][31] we measured Gi/o activation with a homogeneous time-resolved fluorescence (HTRF) cyclic adenosine monophosphate (cAMP) assay.We observed a 5-fold increase in activation in coGHCR-treated compared with GHCR-treated mice (Figure 1H).The maximum spectral sensitivity of retinas treated with coGHCR was around 500 nm, and a photoresponse was obtained even upon stimulation with light with a wavelength > 600 nm (Figure 1I).

Restoration of visual cortex responses by GHCR
To investigate whether retinal light responses were transmitted to the visual cortex, we then examined visual evoked potentials (VEPs) generated by the visual cortex (Figure 2A).The output from the RGCs is sent through their axons (optic nerve) to the lateral geniculate nucleus (LGN) of the thalamus, which is a region of the diencephalon, then from the LGN to the primary visual cortex in the occipital lobe of the cerebral cortex.For these experiments, we used rd1 mice in which both eyes had been treated with the AAV-DJ-CAGGS-GHCR, AAV-DJ-CAGGS-coGHCR, or control EGFP (AAV-DJ-CAGGS-EGFP) vectors.Significant VEPs were not detected in the control or GHCR-treated mice.In contrast, VEPs were observed in coGHCR-treated mice (Figure 2B).In response to 3 cd s/m 2 light stimulation, the average VEP amplitude in coGHCR-treated mice was significantly higher (56.4 mV; n = 6) than those in GHCR-treated mice (22.1 mV; n = 8) and control mice (17.9 mV; n = 6) (Figure 2C).Based on this result, all subsequent experiments were performed using coGHCR.

Characterization of the in vivo responses restored by GHCR transduction
Next, light-dark transition (LDT) testing was performed to investigate whether ectopic expression of coGHCR in degenerating retinas led to behavioral changes due to vision restoration (Figure 3A).Rodents with intact vision tend to stay in dark places as they are nocturnal and feel uneasy in bright environments, whereas blind rodents spend roughly half of their time in bright places.The coGHCR-treated mice spent significantly less time in the bright area compared with the untreated rd1 mutant mice (Figure 3B), thereby confirming vision restoration via behavioral analysis.And the visual restoration effect was still maintained after two years (Figure 3C).Furthermore, in order to directly compare the effects of coGHCR with genes in clinical trials, we treated rd1 mice with chimeric rhodopsin (AAV-6-CAGGS-coGHCR), microbial opsin (AAV-6-CAGGS-ChrimsonR 32 ), animal rhodopsin (AAV-6-CAGGS-human rhodopsin), or the control EGFP (AAV-6-CAGGS-EGFP) vector.At an illuminance of 3,000 lux, a significant reduction in the time spent in the bright half of the observation area was noted for coGHCRtreated mice (0.32; n = 6) compared with control mice (0.50; n = 8) (Figure 3D).A similar tendency was observed in ChrimsonR-treated mice

Restored object recognition function upon GHCR gene therapy
LDT testing measures only light and dark discrimination.Visual recognition testing (VRT) was performed to evaluate whether the mice could recognize an object with the restored level of vision.4][35] We examined mice in a place preference apparatus with a tablet showing a fighting video (Figure 3F).The ratio of the time spent in the area with the fighting compared with the time spent in the control area (showing a video of an empty cage with the same illuminance) over 15 min was measured.The coGHCR-treated (AAV-DJ-CAGGS-coGHCR) mice spent significantly more time in the fighting video half of the apparatus (0.55, n = 33) than the untreated rd1 mice (0.50, n = 30).On the other hand, microbial opsin-treated (AAV-DJ-CAGGS-C1V1 36 ) mice spent roughly equivalent time in each half (0.49, n = 20) (Figure 3G).

GHCR protective effects against retinal degeneration
We employed another mouse model of retinal degeneration using Rho P23H/+ mice with the P23H RHO mutation, referred to as P23H mice. 37P23H mice were selected to evaluate the protective effect because they have slower retinal degeneration than rd1 mice.We subretinally delivered AAV DJ-CAGGS-coGHCR and the control (AAV DJ-CAGGS-EGFP) vector into postnatal day (PND) 0-1 Rho P23H/+ mouse retinas, targeting the outer retina, and quantified the protective effects of the vector via morphological and electrophysiological examination.Subretinal injection of AAV-DJ efficiently induced gene expression in the murine outer retina (Figure 4A).Optical coherence tomography (OCT) showed that the outer retinal thickness (ORT), which is the thickness from the outer nuclear layer (ONL) to the rod outer segment (ROS), of coGHCRtreated mice (50.0 mm; n = 13) was significantly greater than that of the control mice (42.7 mm; n = 10) at PND 30 (Figures 4B and 4C).The ORT of the treated mice remained significantly greater than that of control mice until PND 50 (Figure S3).
Electroretinography (ERG) revealed that the treated mice had larger rod, mixed, and cone response amplitudes (141.2 mV, 271.4 mV, and 159.0 mV, respectively; n = 9) than the control mice (70.4 mV, 158.7 mV, and 99.1 mV, respectively; n = 14) at PND 30 (Figures 4D and 4E).All amplitudes in the control mice gradually decreased, whereas all amplitudes in the coGHCR-treated mice continued to increase until PND 42 (Figures S4A-S4C).Thereafter, the amplitudes in the treated mice also gradually decreased, although they remained significantly higher than those in the control mice until PND 66.
We also performed terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) to detect apoptosis in the retinas.The number of TUNEL-positive cells in the coGHCR-treated mouse ONL (67.3 cells; n = 3) was significantly lower than that in the control mouse ONL (289.7 cells; n = 3) at PND 31 (Figures 5A-5C).
To expand these observations, we obtained transmission electron microscopy (TEM) images of transverse sections from PND 31 mice.Consistent with the OCT results, the ONL (Figure 5D) and ROS (Figure 5E) of coGHCR-treated mice were relatively intact compared with those of controls, and the ROS structure was less disorganized (Figure 5F).In addition, coGHCR-treated mice had less swelling of their ER, a feature that is indicative of ER stress (Figure 5G).Since retinoid levels are known to affect ER stress and retinal degeneration, retinoid analysis of the treated eyes was performed.The amount of retinal was measured by HPLC using the retinal oxime method after 10 min of exposure to 1000 lux, a fluorescent lighting level assuming a normal indoor environment.The results showed that 11-cis retinal oximes was significantly elevated in the treated eyes (54.1 G 18.2 pmol/2 retinas; n = 9) versus controls (39.5 G 6.5 pmol/2 retinas; n = 9) (Figures 5H and 5I).The amount of all-trans-retinal oxime was also elevated in treated eyes; however, this elevation did not attain statistical significance (p = 0.22) (Figure 5J).

DISCUSSION
Because the phenotype of retinal degeneration is common across cases of RP, regardless of genotype, the strategy of optogenetic therapy has great potential as a universal therapeutic approach.It aims to target non-photoreceptive surviving neurons in the retina, such as RGCs and bipolar cells, and convert them to photoreceptive.
In this study, we demonstrated that ectopic expression of coGHCR is an effective method of optogenetic vision restoration in mice with retinal degeneration.MEA revealed that photoresponses were maintained for retinal irradiance levels as low as 10 13 photons/cm 2 /s.This is consistent with the response of the treated mice to 10 lux illumination in the behavioral test, and represents a significant improvement in sensitivity compared with that observed in previous studies of vision restoration with microbial opsins (threshold: 10 14 to 10 17 photons/cm 2 /s), [2][3][4][5][6][7] LiGluR/ MAG photoswitches (threshold: 10 15 -10 16 photons/cm 2 /s), 38,39 or photoactivated ligands (AAQ threshold: 10 15 photons/cm 2 /s 40 and DENAQ threshold: 4 3 10 13 photons/cm 2 /s 41 ).Although some vectors restored greater sensitivity, such as human rhodopsin, 9 cone opsin, 16 and Opto-mgluR6 (10 12 photons/cm 2 /s), 8 our LDT results at 3,000 lux (similar to a cloudy outdoor environment) suggest that photobleaching of rhodopsin like these does not work in bright environments.coGHCR is adaptable to a light environment ranging from at least 10 lux (similar to a night light levels with streetlights) to 3,000 lux, and is, thus, a suitable single-opsin vision restoration tool.
Furthermore, the typical channelrhodopsins have a spectrum limited to blue light, 42 which limits their use as a visual restoration tool.On the other hand, GHCR has a spectrum peak around 500 nm and facilitates responses to red light.Irradiation of high-energy light such as blue light can cause phototoxicity and cell death due to generation of free radicals. 43Therefore, there are concerns about phototoxicity in optogenetic tools that operate under blue light, such as channelrhodopsin, and long wavelength-shifted opsins have been developed. 44In this regard, the GHCR has the advantage of being highly sensitive and having a peak at intermediate (green) wavelengths, making it responsive to short and long wavelengths and less likely to exceed safe limits of light intensity. 45In addition, behavioral tests showed that coGHCR enabled responses to both sustained and transient stimulation lasting 10 ms.These findings suggested that coGHCR gene therapy can restore sensitivity to multiple light environments encountered in daily life.
The ERG amplitudes in coGHCR-treated mice continued to increase until PND 42, likely because the coGHCR-mediated signal was additive with the innate amplitude.This is consistent with the fact that gene expression of the AAV-DJ vector peaks at approximately 1.5 months after administration. 25We observed no apparent changes in the shapes of the ERG waveforms in the coGHCR-treated mice.The visual restoration effect was also maintained for two years, which shows promise for long-term pharmacological effects and safety.
coGHCR has Gt activity derived from rhodopsin. 24Gt is also known to be cross-linked with Gi/o, 46 and this was confirmed (Figure 1H).Although this study used a ubiquitous promoter, which cannot be fully confirmed, Gi/o is generally expressed specifically in ON-type bipolar cells, 47,48 where the light-responsive signal is likely to have been generated.When coGHCR is expressed ectopically in ON bipolar cells, it is expected to inhibit responses.However, the restored responses observed by MEA were all ON responses.In addition, the electrophysiological and behavioral results were similar to physiological responses, and no reversal reaction observed.In rd1 mice, photoreceptors are mostly lost by 4 weeks after birth and no optical response is obtained after 7 weeks at the latest. 49,50Therefore, responses from residual photoreceptors are unlikely in this study.A similar phenomenon has been confirmed in previous studies; the excitatory response is hypothesized to result from disinhibition of inhibitory amacrine cells. 6,8,9he safety of ectopic expression of opsins, such as channelrhodopsin 2, has been previously reported. 3,51,52To our knowledge, this is the first report of their protective effects against retinal degeneration.In vitro studies have shown that the P23H opsin is misfolded and retained in the ER. 535][56][57] Our results suggest that expression of coGHCR in the retinal outer layer suppressed photoreceptor apoptosis, which led to protection against degeneration.The lack of 11-cis-retinal induces cytotoxicity during the development of ROS in P23H mice. 58In fact, the amount of cis-retinal in the retina was significantly elevated after coGHCR treatment.Since coGHCR uses all-trans retinal as a chromophore, like microbial opsin, it does not consume cis-retinal and is free from photobleaching.Therefore, the expressed coGHCR may suppress cis-retinal consumption via photoreceptor substitution.If this hypothesis is correct, the protection effect of coGHCR may not be applicable to patients with all IRD genotypes.However, there are more than 140 known RP-linked rhodopsin mutations, and those that result in protein misfolding and retention in the ER are the most prevalent. 59,60n summary, the coGHCR vector has the advantages of both animal and microbial opsin as a vision regeneration tool.It restores sensitivity and an action spectrum that enables vision in lighting ranging from levels found outdoors to those in dimly lit indoor environments via G protein stimulation without the risk of bleaching; it can also be expected to protect against the progression of retinal degeneration in the majority of IRD patients.These results suggest that coGHCR is worthy of consideration for clinical application as a gene therapy for IRD.  were dark-adapted for 12 h and prepared under dim red illumination.At the time of the measurement, the mice were anesthetized again with the same doses.Visual stimuli were generated by a white LED flashes (3 cds/m 2 ).Signals were acquired and analyzed with a PuREC acquisition system (Mayo).Signals were low-pass filtered at 300 Hz and averaged over the 60 trials.

LDT recording
Mice were tested in a 30 3 45 3 30-cm box, containing equally sized light and dark chambers connected by a 5 3 5-cm opening via which mice could move freely.The bright half of the box was illuminated from above by a white LED.The illumination intensity of measured at the floor level.The animals were placed in the bright half and movement recorded (HD Pro Webcam C920, Logitech, Lausanne, Switzerland).A trial lasted 10 min, and then the testing apparatus was dismantled and cleaned with 70% ethanol.Videos were analyzed using ANY-maze tracking software and were validated by comparison with manual analysis.Time spent in the bright half was recorded.

VRT recording
Mice were tested in a 216 3 148 3 220-mm box, containing equally sized light and dark chambers connected by a 120 3 60-mm opening via which mice could move freely.The size of the tablet was 107 3 9.9 3 193-mm (B1-760HD, Acer Inc, New Taipei, Taiwan).The resolution of the display was 1280 3 720 pixels, and the resolution of the videos was 640 3 480 pixels.The luminance of all videos was 20 G 3 lux.All videos were presented without sound.The box was illuminated from above by a white LED with 10 lux.The illumination intensity of measured at the floor level.The animals were placed in the bright half and movement recorded (HD Pro Webcam C920, Logitech, Lausanne, Switzerland).A trial lasted 15 min, and then the testing apparatus was dismantled and cleaned with 70% ethanol.Videos were analyzed using Move-tr/2D tracking software (Library, Tokyo, Japan) and were validated by comparison with manual analysis.Time spent in the bright half was recorded.

Gi/o coupled GPCR activation assay
HTRF-based cAMP detections were conducted with cAMP Gi kit (Cisbio #62AM9PEB, Bedford, MA) according to the manufacturer's instructions.HEK293T cells were kept in DMEM (12-well plate) supplemented with 10% (v/v) fetal bovine serum in a humidified incubator at 37 C 5% CO2.HEK293T cells were seeded on a 12-well plate at 1x10 5 cells/well, and on the day 2, 1x10 6 vg/well/500 ml of AAV vector (AAV-DJ-GAGGS, AAV-DJ-CAGGS-GHCR, AAV-DJ-CAGGS-coGHCR) was added and transfected.Transfected cells were kept in the dark for 2 days.On the day 4, after seeding in a 384-well plate at 6,500 cells/well/5 ul and incubating for 4 hours in the dark.Photo-stimulation (525 nm LED 10 16 photons/cm 2 /s 1 minute) was performed.The signal was detected using plate reader Infinite M1000PRO (Tecan, Ma ¨nnedorf, Switzerland).

TEM
Eye cups were fixed with aldehyde/DMSO at 37 C for 2-4 h and then eye cups were cut in half on their dorsal-ventral axis and fixed again for several minutes.Ultrathin sections were cut with a diamond knife.Specimens were examined using a transmission electron microscope (JEM-1400Plus).

Preparation of cryosections of retinas
Enucleated eyes were fixed for 20 min in 4% paraformaldehyde (PFA) in PBS and then dissected as previously described. 61The obtained tissues were post-fixed overnight in 4% PFA in PBS and stored in methanol at -20 C. Cryosections of retinas (12 mm) were prepared as previously described, 62 after the eyeballs were immersed overnight in 4% PFA.The retinal sections were observed using a confocal microscope (LSM710; Carl Zeiss, Jena, Germany).

TUNEL assay
After the cryosection mentioned above, cell apoptosis was detected by TUNEL using ApopTag In Situ Apoptosis Detection Kits (Chemicon International, Darmstadt, Germany; cat.#S7165) according to the manufacturer's instructions.Nuclei were counterstained with DAPI.The retinal sections were observed using a confocal microscope (LSM710; Carl Zeiss, Jena, Germany).

OCT imaging
The thickness of the retina was analyzed by an SD-OCT system (Envisu R4310; Leica, Wetzlar, Germany) tuned for mice.The imaging protocol entailed a 3 mm33 mm perimeter square scan sequence producing a single en-face image of the retina through a 50-degrees field of view from the mouse lens, following mydriasis.The en-face image consisted of 100 B-scan tomograms with each B-scan consisting of 1000 A-scans.The retinal thickness of 150 mm from the optic disc of each quadrant was measured.

HPLC analysis of retinal
After 15 hours of dark adaptation, mice were exposed to light adaptation at 1000 lux for 10 minutes.The mice were subsequently sacrificed, and the removed mouse retinas were homogenized.Hydroxylamine was added to the homogenized retinas for oximation.The retinal-oximes was dissolved in hexane to make a sample for HPLC analysis.All of these processes were performed under dim red lights.Two retinas were used per assay.
Retinal oximes were analyzed by HPLC (Shimadzu LC20A series, Japan) with a silica column (Ultrasphere 5um, SI 250 x 4.6mm, Avantor, USA).The mobile phase consisted 96.0%(v/v) hexane, 4.0%(v/v) ethyl acetate and the flow rate was 1.0mL/min.The column temperature was 35 C. Absorbance at 360 nm was monitored for retinal oximes.Each retinal isomer was quantified from the area of the corresponding peak based on a calibration retinal standard reagent.All trans-retinal (Sigma-Aldrich) and 11-cis retinal (Toronto research chemicals) were used as standard reagents.

Data and software availability
Raw MEA spike data were sorted offline to identify single units using Offline Sorter software (version 4.4.0)(Plexon).Spike-sorted data were analyzed with NeuroExplorer 5 software (version 5.115) (Nex Technologies).The data that support the findings of this study are available from the corresponding author on request.

QUANTIFICATION AND STATISTICAL ANALYSIS
All of the results are expressed as the mean G SEM.The averaged variables were compared using the unpaired t-test and the one-way ANOVA test.Tukey's test was used for multiple comparisons.P-values of less than 0.05 were considered statistically significant.All experiments were randomized.SPSS 26 (IBM Corporation, Armonk, NY) was used for statistical analysis.

Figure 1 .
Figure 1.Ectopic GHCR expression restores light responses in the rd1 mouse retina (A) DNA expression cassette schematic.The GHCR coding sequence is driven by the CAGGS promoter, flanked by inverted terminal repeats (ITR), and stabilized by a polyadenylation signal sequence (pA) and a woodchuck hepatitis posttranscriptional regulatory element (WPRE).(B, C, and E) Raster plots and peri-stimulus time histograms for light stimulation of control (AAV-DJ-CAGGS-EGFP) (B), GHCR-treated (AAV-DJ-CAGGS-GHCR) (C), and coGHCR-treated (AAV-DJ-CAGGS-coGHCR) mice (E).Responses to exposure to a white LED with varying light intensity for 1.0 s.Gray shading around the averaged traces represents the standard error of the mean (SEM).(D) Confocal image of a transverse rd1 mouse retina section 2 months after AAV-DJ-CAGGS-coGHCR intravitreal injection.Green, FLAG tag antibody signal (vector); red, PKCa signal (bipolar cells); blue, 4 0 ,6-diamidino-2-phenylindolenuclear (DAPI) counterstaining.Scale bar, 50 mm.(F) Quantitation of the firing rates of RGCs transduced with GHCR or coGHCR at the indicated light intensity.(G) Histogram showing the number of RGCs that responded to light per unit area (2.6 mm 2 ) of the retinas of GHCR-or coGHCR-treated mice (n = 3 each).(H) Changes in cAMP consumption in response to Gi/o-coupled G-protein-coupled receptor activation in HEK293T cells transfected with GHCR and coGHCR (n = 3 each).

Figure 5 .
Figure 5. coGHCR treatment suppressed retinal apoptosis and ER stress (A and B) TUNEL-stained transverse sections (A) and enlarged images of the white squares (B) of coGHCR-treated and control (AAV-DJ-CAGGS-EGFP subretinally injected) mouse retinas at PND 31.Red, TUNEL-positive cells; blue, DAPI nuclear counterstaining.Scale bar, 1,000 mm in (a) and 100 mm in (B).(C) Histogram of the number of TUNEL-positive cells in the ONLs of coGHCR-treated (n = 3) and control mice (n = 3) at PND 31.(D-G) (D) TEM images of transverse sections from coGHCR-treated and control mice at PND 31, showing the outer retinal layer (D), the outer segment at low magnification (E) and high magnification (F), and the inner segment (G).The arrowhead indicates swollen ER.Scale bar, 20 mm in (D), 5 mm in (E), 1 mm in (F), and 500 nm in (G).