Compensation of Inner Retina to Early-Stage Photoreceptor Degeneration in a Rho^P23H/+ Mouse Model of Retinitis Pigmentosa

Abstract

Retinitis pigmentosa (RP) is an inherited retinal disorder characterized by the degeneration of photoreceptors. Rho^P23H/+ mice, which carry a Pro23His mutation in the RHODOPSIN (Rho) gene, are one of the most studied animal models for RP. However, except for the photoreceptors, other retinal neural cells have not been fully investigated in this model. Here, we record the temporal changes of the retina by optical coherence tomography (OCT) imaging of the Rho^P23H/+ mice, from early to midphase of retinal degeneration. Based on thickness analysis, we identified a natural retinal thickness adaption in wild-type mice during early adulthood and observed morphological compensation of the inner retina layer to photoreceptor degeneration in the Rho^P23H/+ mice, primarily on the inner nuclear layer (INL). Rho^P23H/+ mice findings were further validated via: histology showing the negative correlation of INL and ONL thicknesses; as well as electroretinogram (ERG) showing an increased b-wave to a-wave ratio. These results unravel the sequential morphologic events in this model and suggest a better understanding of retinal degeneration of RP for future studies.

1. Introduction

Retinitis pigmentosa (RP) is the most common inherited retinal degenerative disease and affects 1 in 3000-7000 people worldwide (Hartong et al., 2006; Soucy et al., 2023). An early symptom of RP includes impaired night vision, usually starting in childhood, as a consequence of the rod photoreceptors dysfunction in the retina (Lassoued et al., 2021; Portera-Cailliau et al., 1994). Disease progression results in the loss of peripheral vision and, ultimately, leads to the degradation of cone photoreceptors and subsequent degeneration of central vision. Currently, there are no effective treatments for RP. Thus, RP severely affects the patient’s quality of life. To date, over 80 genes have been linked with the pathogenesis of RP (https://web.sph.uth.edu/RetNet/) (Anasagasti et al., 2012). In North America, the Pro23His mutation in the RHO gene is the most commonly identified mutation associated with autosomal dominant RP (Berson, 1996; Dryja et al., 1990; Nakazawa et al., 2019; C.-H. Sung et al., 1991). Hence, the Rho^P23H/+ animal models are commonly studied to elucidate RP pathophysiology and develop novel therapeutic strategies (Chen et al., 2014; Drenser et al., 1998; Galy et al., 2005; Olsson et al., 1992; Roof et al., 1994, n.d.; Sakami et al., 2014, 2011; Tam and Moritz, 2006; Vats et al., 2022; Xi et al., 2022).

The retina includes two types of photoreceptors: rods and cones. These photoreceptors connect with the stratified neural retina and span the inner and outer segments (IS&OS), the retinal outer nuclear layer (ONL), and the outer plexiform layer (OPL). Rods constitute 95% of the photoreceptor population in both rodents and primates (Stojanovic and Hwa, 2002). Rhodopsin, the dim-light-sensing visual pigment, plays a crucial role in vision as the first step of the phototransduction cascade. Upon activation of rhodopsin by light, rods are hyperpolarized – an essential early step in the visual process (Nathans, 1992; Palczewski et al., 2000; Wald, 1938). Due to its abundance, the inherent structural instability of the mutant rhodopsin protein overwhelms the proteostasis system in rods, leading to rod death (C. H. Sung et al., 1991). Rod degeneration leads to observable thinning of the retinal ONL and IS&OS, as demonstrated in both RP patients and Rho^P23H/+ animal models through optical coherence tomography (OCT) imaging (Nakazawa et al., 2019; Vats et al., 2022). This structural degradation of the retina, confirmed by in vivo imaging, is consistent with immunohistochemistry (IHC), electroretinography (ERG), and transcriptomic analyses utilizing RNA-Seq data findings. Jointly, these findings delineate a comprehensive pathophysiological profile of the disease model (Chiang et al., 2015; Leinonen et al., 2022; Sakami et al., 2014, 2011).

Yet, it is not fully understood how RP affects other retinal layers. In both humans with RP (Fariss et al., 2000; Jones et al., 2016; Marc et al., 2007) and rodent RP models (Chua et al., 2009; Leinonen et al., 2020), neuronal remodeling occurs within the retinae. Notable previous research indicate changes in the inner retina, including a significant thickening of the inner retinal layers (Oh et al., 2020). For instance, Aleman et al. reported findings of increased inner retinal thickness in human patients with X-linked RP, caused by a mutation in the RPGR gene (Aleman et al., 2007). Similar thickening of the inner retinal layers have been reported in an autosomal dominant RP mouse model, with the RHO I370N mutation (Stefanov et al., 2020). Intriguingly, studies on Rho^P23H/+ knock-in mice reveal a compensatory increase in the bipolar to photoreceptor response ratio, attributed to synaptic remodeling between rods and bipolar cells following rod cell loss (Leinonen et al., 2020). Nevertheless, to date, comprehensive morphological examinations encompassing the full depth of the retina, particularly the inner layers, are lacking in Rho^P23H/+ mice.

In the present study, we systemically monitored the retinal architecture in Rho^P23H/+ and wild-type (WT) mice over time, using a spectral domain OCT (SD-OCT) system. Retinal boundaries were segmented from cross-sectional OCT images, allowing us to measure the thickness of various retinal layers, including inner retinal layers and photoreceptors. We delineated the temporal progression of both normal retinal thinning and early adulthood degenerative phase in Rho^P23H/+ mice through comparative analyses between the Rho^P23H/+ and WT mice, furthermore, we correlated our in vivo imaging with histological examinations and ERG functional recordings.

2. Materials and methods

2.1. Animals

The C57BL/6J mice (WT) (stock no. 000664) and (Rho^P23H/P23H) knock-in mice (stock no. 017628) were procured from the Jackson Laboratory. Experimental Rho^P23H/+ mice were generated by crossbreeding of WT and Rho^P23H/P23H mice. An equal number of male and female animals were included in this study. The animals were housed in the South Biomedical Science Tower (SBST) Animal facilities at the University of Pittsburgh under a 12 h light cycle (illuminated at 50 lux), followed by a 12 h dark cycle (with environmental illumination <10 lux). All procedures were performed in strict adherence to the guidelines set forth in the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. Furthermore, all animal experiments underwent rigorous examination and received approval from the University of Pittsburgh Institutional Animal Care and Use Committee (IACUC).

2.2. OCT Imaging Acquisition

A total of six Rho^P23H/+ mice (N=12 eyes) and six WT mice (N=12 eyes) were imaged using a high-resolution SD-OCT (Bioptigen Envisu R-Class, Leica, Wetzlar, Germany). This OCT system offers an axial resolution of approximately 3 μm (in tissue) with a depth range of 1.56 mm (in tissue). Before imaging, mice were anesthetized via intraperitoneal injection of ketamine/xylazine cocktail containing 80 mg/kg body weight (bw) of ketamine and 7 mg/kg bw of xylazine. Additionally, the pupils of the animals were dilated by 1% tropicamide (Akorn, Gurnee, IL, USA, NDC17478-102-12) and 2.5% phenylephrine (Akorn, NDC 17478-201-02). Subsequently, the mice were carefully positioned on a stage that allowed precise manipulation for optimal image quality.

The within-subject OCT B-scan images were acquired at the region centered by the optic nerve head (ONH) at postnatal time points: P28, 35, 41, and 69. These images covered the temporal to the nasal direction of the eye. Each B-scan was comprised of 640 pixels (1.6 mm) in the X-direction and 480 pixels (1.2 mm) in depth. To enhance image quality, five B-scans were acquired during each session and subsequently merged using accompanying software. To be noted, Rho^P23H/+ and WT mice underwent imaging on the same day at each time point to minimize systemic errors. The entire experimental procedure for each animal, including anesthesia, animal manipulation, and OCT imaging, was completed in approximately 15 minutes. Frequent application of eye gel (~every 5 min) was employed to ensure the cornea remained moist throughout the procedure.

2.3. Retinal Thickness Measurement

The merged B-scan images were exported from the OCT system for further processing. Initially, the eight boundaries of the retina (Fig. 1) were automatically segmented using a customized graphical user interface (GUI) software developed through MATLAB App Designer. These boundaries included the following: (1) the inner limiting membrane; (2) the interface of the retinal nerve fiber layer (NFL) and inner plexiform layer (IPL); (3) the interface of IPL and inner nuclear layer (INL); (4) the interface of INL and OPL; (5) the interface of OPL and ONL; (6) external limiting membrane; (7) upper border of the retinal pigmented epithelium (RPE); and (8) Bruch’s membrane (BM). Manual corrections were carried out by three independent graders. Segmentations were then smoothed using a moving average filter, which removes curve inconsistencies resulting from local gradient changes. The mean absolute difference and standard deviation were found to be less than 1 pixel among all graders. Moreover, the correlation coefficients exceeded 0.99 between the graders, validating consistency in boundary identification. In cases where segmentation differences exceeded 5 pixels, consensus was reached among the three graders. The independent grader segmentations were averaged to determine each retinal layer’s thickness.

Fig. 1.

Retinal cross-sectional images from OCT B-scan with definitions of segmentation boundaries (1-8), layers, and slabs. The center of the optic nerve head (ONH) (red dot, spanning 260 μm) was manually labeled and excluded from thickness analysis. The rest of the image was divided into eight subregions (four in nasal and four in temporal, 160 μm each) for spatial analysis. Note that the images in the left eyes were flipped horizontally to match the temporal and nasal directions in the right eyes. Scale bar, 200 μm.

We measured the NFL, IPL, INL, OPL, ONL, IS&OS of photoreceptor cells, RPE, the ganglion cell complex (GCC), inner retina, outer retina, and total retina. These measurements were obtained by subtracting the locations of corresponding boundaries of the layer (Fig. 1). The unit of thickness was then converted from pixels to microns by scaling it to the digital resolution (1.85 μm/pixel), with a refractive index of 1.35 for the retina. Additionally, the center of the ONH was manually labeled for each image to exclude the ONH region (~260 μm) from analysis.

For spatial analysis, the remaining image area was further divided into eight subregions, each spanning 160 μm (four in nasal and four in temporal direction) across each eye. It is important to note that the temporal and nasal sides are reversed for the right and the left eyes; thus, to conduct spatial dependency analysis, the images from the left eyes were flipped to align with those from the right eyes. Subregion specific layer/slab thickness was determined by averaging the results from all locations within said subregion. The thicknesses of all eight subregions were then averaged to obtain the value for the entire image. It is worth mentioning that the outermost edges (~60 μm) of the B-scan images were excluded from the analysis, due to the higher possibility of segmentation errors and variations in those areas.

2.4. Electroretinogram (ERG)

To access retinal functions, ERG recordings were obtained from both WT and Rho^P23H/+ mice using a Celeris system (Diagnosys). Mice were dark-adapted overnight (>8 h) before the ERG recordings. Mice were anesthetized with a ketamine/xylazine cocktail, pupils were dilated with 1% tropicamide eye drops, and corneal electrodes were meticulously positioned. Scotopic ERG responses were elicited using a series of flashes at varying intensities (0.001, 0.01, 0.1, 1, 10, 30 cd·s/m²). Each intensity was then averaged across sweeps (N=5 for intensities ≤0.01, N=3 for intensities >0.01) and flashes were interposed by intervals ranging from 10 to 30 seconds. We recorded the responses from both eyes and averaged them. The amplitude of a- and b-waves, along with their respective ratios, were quantified from the scotopic ERG data.

2.5. Histology

Following in vivo analyses, eyes from both mutant and WT mice were enucleated for histological examination. To maintain orientation, the superior side of the cornea-sclera area was marked via cautery burn. The eyes were then fixed in Hartmann’s fixative (Electron Microscopy Sciences, 64133-10) overnight and at room temperature. Ensuing fixation, the eyes were transferred to 70% ethanol for storage. Tissue processing was executed under vacuum in an automated tissue processor (Sakura Tissue-Tek, VIP-5), were allocated 20 minutes per station (with the exclusion of heat application, except during the paraffin stations). Eyes were then embedded in low-melting-point paraffin wax (Leica, Paraplast-X), with careful alignment of the cautery mark towards the upper left in each mold. The solidification process was conducted using an embedding center and cold plate (Leica, EG1160). Wax blocks were mounted and cut in 5 μm thickness sections via a manual rotary microtome (Leica, RM2235) equipped with disposable low-profile steel blades (Feather, Accu-Edge). When the central region of the eye was reached, featured by the pupil gap and inclusion of the optic nerve head apparent in the block, sections were collected, floated in a heated water bath (Science Instruments, TekBath), and mounted onto charged slides (Rankin Biomedical, RAN 203-W). The slides were dried at room temperature overnight, de-waxed, stained using Meyer’s progressive hematoxylin (Electron Microscopy Sciences, 26043-05) and alcoholic Eosin-Y with Phloxine B (Richard Allan Scientific, 71311), mounted with xylenes substitute mounting medium (Epredia Shandon-Mount, 1900331) under 1.5 thickness cover glass (Epredia, 22-050-244). To ensure even distribution of the mountant, the slides were laid flat for a minimum duration of 3 days prior to imaging on a light microscope attached to a color camera (Leica).

2.6. Statistics

Intraclass correlation coefficients were employed to access the concordance among individual grader-corrected segmentation outcomes. The significance of retinal thickness differences across respective regions, between Rho^P23H/+ mice and WT counterparts, were calculated with unpaired, two-tailed Student’s t-test. Significantly different subregion counts are shown only if the differences persisted throughout the entire image between the two cohorts. A linear mixed effects model was used to evaluate the temporal changes within the same group. All statistical analyses were performed using the MATLAB software (R2022b, MathWorks Inc., Natick, MA, USA).

3. Results

In the Rho^P23H/+ mouse model, a fast phase of photoreceptor degeneration is reported between P15-30, with the progression markedly slower post 1 month of age (Chiang et al., 2015; Sakami et al., 2011). Our study focused on recording retinal morphological changes occurring from P28 to 69, the period immediately following the rapid degeneration phase. Analysis of the retinal cross-sectional images obtained via OCT B-scans (Fig. 2), revealed that across all examined time points, the ONL and IS&OS regions were visibly thinner in the Rho^P23H/+ mice as compared to the WT controls. Moreover, the conjunction between the IS and OS was disrupted in the Rho^P23H/+ mice, characterized by the absence of a bright band in the ellipsoid zone. These observations are in agreement with published histological findings, that attribute the shorter OS in the Rho^P23H/+ mouse retina to a substantial, approximately 50%, decrease in rhodopsin levels relative to the WT mouse retina (Sakami et al., 2011). No significant difference was observed between the sexes.

Fig. 2.

Representative retinal cross-sectional images of the Rho^P23H/+ and wild-type (WT) mice at P28, P35, P41, and P69. Note that the Rho^P23H/+ mouse retina showed notable thinning of ONL (marked with stars) and disrupted IO/OS junction (marked with arrows) starting at P28 when compared to their age-matched WT controls.

3.1. Natural retinal thinning during early adulthood

In the C57BL/6J WT mice, a baseline layer thickness at P28 was measured by segmentation of all retinal layers (Fig. 3 and Table 1). Notably, a decline in the thickness of retinal layers was observed in these post-weaning mice. For example, the total retinal thickness exhibited a decrease from 200.5 μm to 187.7 μm from P28 to P69 (P-value < 0.05; see Fig. 3A). This reduction was predominantly attributed to the narrowing of the inner retinal layers (NFL, IPL, INL, and OPL; Fig. 3B), while the thickness of the outer retinal layers (ONL, IS&OS, and RPE) remained relatively unchanged (Fig. 3C). Within the inner retina, the thinning was specifically observed in the NFL, the INL, and the IPL, but not in the OPL (Fig. 3D-H). The INL, in particular, underwent significant thinning from 23.3 μm to 19.9 μm between P35 and P41, thereafter maintaining a relative consistency (Fig. 3G and Table 1). These results suggest that, alongside the synaptic pruning occurring within the NFL, indicative of retinal ganglion cell refinement; there may also be a slight reduction in the inner retinal cell number of the INL during early adulthood.

Fig. 3.

Thickness of individual retinal layers at P28, P35, P41, and P69 for the WT (blue) and Rho^P23H/+ (red) mice. Data and error bars are means±SD. * indicates the significance (p-value <0.05) between the animal groups at one time point (—) and two time points within one animal group (⎵) by one-way repeated-measures ANOVA, as well as longitudinal differences between two groups by two-way repeated-measures ANOVA. N=12 eyes for each group.

Table 1.

Retinal layer thicknesses (mean ± SD) within the scanning region and their ratios to ganglion cell complex (GCC) thickness were measured in WT and Rho^P23H/+ mice at indicated time points. N=12 eyes for each group.

Layer Thickness (μm)	P28		P35		P41		P69
	WT	RhoP23H/+	WT	RhoP23H/+	WT	RhoP23H/+	WT	RhoP23H/+
NFL	18.7±3.0	17.6±4.1	16.6±2.3	15.6±2.8	16.0±2.3	15.0±3.3	16.6±4.0	14.8±1.2
IPL	42.3±2.2	44.5±3.7	41.4±2.3	43.3±2.9	40.1±1.3	41.5±1.3	40.0±3.4	42.3±1.3
GCC	61.0±3.1	62.1±3.5	58.0±3.3	58.9±1.3	56.1±1.5	56.5±2.4	56.6±2.4	57.1±1.7
INL	22.3±1.8	24.3±1.5	23.3±1.7	24.4±0.9	19.9±0.3	24.1±1.2	19.1±0.9	23.9±1.3
OPL	13.5±1.6	12.7±1.1	11.4±1.7	11.4±1.0	11.7±0.7	10.7±0.7	11.5±1.3	10.1±0.4
Inner Retina	96.8±2.8	99.0±2.7	92.7±2.9	94.8±1.8	87.7±2.3	91.2±2.5	87.2±2.9	91.2±1.3
ONL	49.6±1.3	29.5±1.1	50.1±1.3	25.5±1.5	46.9±1.4	25.4±1.2	48.1±1.3	19.1±1.9
IS&OS	41.4±1.8	22.2±1.7	39.9±1.4	20.8±2.4	36.5±2.7	24.1±1.6	37.3±2.6	20.5±1.5
RPE	12.7±1.1	12.1±1.3	9.8±2.6	9.3±1.1	12.9±1.8	11.1±1.3	15.0±1.7	11.4±2.2
Outer Retina	103.7±2.1	63.9±2.6	99.7±2.4	55.6±2.4	96.4±2.9	60.6±2.9	100.5±2.7	50.9±4.5
Total Retina	200.5±3.7	162.9±4.1	192.4±3.9	150.4±3.3	184.0±4.9	151.8±4.4	187.7±3.1	142.1±4.4
Layer Thickness Ratio (%)	P28		P35		P41		P69
	WT	RhoP23H/+	WT	RhoP23H/+	WT	RhoP23H/+	WT	RhoP23H/+
NFL/GCC	30.2±3.8	28.0±5.5	28.2±2.9	26.3±4.8	28.0±2.8	26.2±4.4	28.5±5.9	25.5±1.7
IPL/GCC	69.8±3.8	72.0±5.5	71.8±2.9	73.7±4.8	72.0±2.8	73.8±4.4	71.5±5.9	74.5±1.7
INL/GCC	37.4±4.2	39.9±4.2	41.1±5.0	42.0±1.9	36.2±0.8	43.2±3.4	34.4±2.5	42.7±3.3
OPL/GCC	22.3±3.1	20.6±2.0	20.1±3.2	19.5±1.5	21.0±1.1	19.1±1.4	20.4±2.1	18.0±1.2
ONL/GCC	82.4±4.4	48.2±2.8	87.5±5.4	43.8±2.3	84.9±3.3	45.5±3.2	86.4±4.6	34.0±3.5
IS&OS/GCC	68.5±4.4	36.1±3.1	70.1±6.4	35.7±4.8	65.8±4.7	43.1±3.0	66.7±4.8	36.3±2.5
RPE/GCC	21.0±1.9	19.7±2.3	17.1±4.3	15.8±1.7	23.2±3.0	19.8±2.4	26.7±3.6	20.0±4.1
Outer/GCC	171.9±8.4	104.0±6.3	174.7±11.6	95.4±4.6	174.0±4.2	108.3±6.7	179.7±9.1	90.3±8.4

3.2. Progressive loss of photoreceptors in Rho^P23H/+ mice

Previous characterizations have documented the longitudinal photoreceptor loss in Rho^P23H/+ mice (Sakami et al., 2011). We found that the total retinal thickness of Rho^P23H/+ mice decreased from 162.9 μm to 142.1 μm between P28 and P69, with the magnitude of reduction not surpassing that observed in WT retinae (Table 1). Aligning with earlier findings, the ONL thickness in Rho^P23H/+ mice was already only 59% of the WT’s at P28, as measured from OCT scans. This underscores the rapid initial-phase degeneration within the first postnatal month (Fig. 3I, Table 1). Between P28 and P69, ONL degeneration proceeded at a slower rate, diminishing from 29.5 μm to 19.1 μm. By P69, the ONL thickness was 40% that of the age-matched WT mice (Table 1). Statistical analyses revealed a consistently reduced ONL thickness in all eight subregions of the mutant mice, when compared to the control group, at every time point observed (Fig. 4 A). Notably, the ONL in Rho^P23H/+ mice was consistently and significantly thicker (by approximately 2-4 μm) on the nasal side compared to the temporal side. This asymmetrical pattern of ONL thickness, absent in WT mice (Fig. 5), illustrates the spatial asymmetry of photoreceptor degeneration characteristic of this RP model. Aligning with previous findings that RP patients, with the RHO-P23H mutation, typically exhibit a more pronounced initial visual dysfunction in the temporal superior visual field, as compared to other areas of the retina (Sakami et al., 2011). Correspondingly, histological analyses of Rho^P23H/+ mice have demonstrated accelerated thinning of the ONL on the ventral/inferior side of the mouse retina (Sakami et al., 2011).

Fig. 4.

Statistical analysis of Rho^P23H/+ vs. WT retinae. (A) Comparisons of retinal layer thicknesses of the Rho^P23H/+ mice to the WT mice. The value in each cell indicated the number of subregions showing significance (p-value < 0.05 by t-test) between the two animal groups and the color of cell indicated the direction of thickness change (blue: no change, yellow: thinning, orange: thickening) between the two groups. (B) – (C): Spatial distribution of significance analysis for thicknesses of (B) INL, and (C) INL normalized by GCC (INL/GCC) in the eight subregions and entire images. A cell in orange indicated a p-value <0.05 by t-test, and in blue indicated a p-value >0.05 by t-test. T1-4, subregions on the temporal side spanning close to far from ONH; N1-4, subregions similarly on the nasal side.

Fig. 5.

Spatial distributions of thicknesses for (A) ONL, and (B) ONL normalized to GCC (ONL/GCC) showed the ONL was significantly and consistently thicker (2 ~ 4 μm) in the nasal vs. the temporal side at every time point for the Rho^P23H/+, but not in the WT mice. N=12 eyes for each group.

In contrast to the approximate 3 μm reduction in the IS&OS layers observed in WT mice, the Rho^P23H/+ mice IS&OS thickness remained unchanged from P28 to P69 (Fig. 3 J, Table 1). Spatial asymmetry was also observed in the IS&OS, with the nasal side thicker than the temporal side (data not shown). Our finding that the IS&OS did not exhibit progressive thinning is consistent with previous studies indicating that at P15 - when rod photoreceptors are fully differentiated - the outer segments (OS) are already shorter in Rho^P23H/+ mice; due to the presence of only about 50% of the properly folded rhodopsin compared to WT mice (Sakami et al., 2011, Chiang et al., 2015).

3.3. Inner retinal layers were gradually affected in Rho^P23H/+ mice

With similar inner retina (NFL+IPL+INL+OPL) thickness at P28 compared to the WT control (Figs. 3B & 4A, Table 1), the Rho^P23H/+ mice demonstrated significantly greater inner retina layer thickness at P35, P41, and P69 (Figs. 3B and 4A&C, Table 1). This suggests that the inner retinal thinning observed in healthy retinal growth did not occur in the Rho^P23H/+ mice. Detailed assessments of each inner retinal layer pinpointed the INL as the primary contributor to the disparity observed between the two animal groups; rather than the GCC (composed of NFL and IPL) or the OPL (Fig. 3D-H). Specifically, the INL in Rho^P23H/+ mice did not exhibit the thinning observed in WT mice. Hence, at P41 and P69 the INL of Rho^P23H/+ mice was considerably thicker than age-matched WT controls (Figs. 3G and 4, Table 1). A modest significance in thickness reduction suggests that the OPL in Rho^P23H/+ mice was marginally thinner than that of the WT at P41 and P69 (Figs. 3H and 4A, Table 1). The observed OPL thinning may be attributed to the loss of rod synapses, along with the reorganization of synapses in the inner retinal neurons; a phenomenon consistent with observations in both affected RP human retinal tissue and RP rodent models (Chua et al., 2009; Fariss et al., 2000; Jones et al., 2016; Leinonen et al., 2020; Marc et al., 2007).

3.4. Morphological and functional evidence for compensatory responses in inner retinal neurons responding to rod degeneration

Histological assessments confirmed a thicker INL in Rho^P23H/+ mice at P71 compared to WT controls (Fig. 6A-B). Furthermore, an inverse correlation between the within-animal thicknesses of the INL and the ONL was observed in Rho^P23H/+ mice, using retinal cross-sectional images at P71 (Fig. 6C). This suggests that within the Rho^P23H/+ cohort, individuals with a thinner ONL tended to have a thicker INL, and vice versa. This observed inverse correlation between INL and ONL thicknesses may suggest a compensatory effect characterized by the preservation of neuron numbers in the INL, in response to photoreceptor loss. Functional changes accompanying this morphological adaptation were examined through ERG (Fig. 7). Both scotopic a- and b-waves were diminished in Rho^P23H/+ mice at P28 and P70 compared to age-matched WT controls (Fig. 7B), corroborating the occurrence of degeneration within this timeframe. Notably, the scotopic b-: a-wave ratio in Rho^P23H/+ mice increased significantly from P17 to P28. This increased scotopic b-: a-wave ratio was maintained through P70 – where it measured over twice as high as that in age-matched WT mice (Fig. 7 C-D). A higher b-: a-wave ratio indicates an amplified bipolar cell response to rod hyperpolarization. This signifies a functional compensation by the inner retinal neurons for the attenuated rod response resulting from rod degeneration. Interestingly, the functional adaptation by the inner retinal neurons, as evidenced by the increased b-: a-wave ratio, appeared to precede the morphological compensation indicated by the preservation of the INL in Rho^P23H/+ mice. The photopic b-wave didn’t show significant change in the Rho^P23H/+ mice as it is known that cones are not degenerating within the monitoring time frame (Fig. 7E).

Fig. 6.

Histology validation of INL compensation to photoreceptor degeneration in Rho^P23H/+ mice at P71. Representative histological retinal cross-section image from WT (A) and Rho^P23H/+ (B) mice. (C) The associations of INL thickness (red squares) and total retina thickness (green circles) to ONL thickness from Rho^P23H/+ mouse retinae at P71 were measured in histology images. Note that a negative correlation of ONL:INL thickness is not seen between ONL and total retinal thickness.

Fig. 7.

ERG results measured from Rho^P23H/+ mice (P23H) at P17, P28 and P70, as well as WT mice at P28 and P70. (A) Representative original trace with flash intensity of 1 cd·s/m². (B) Amplitudes of scotopic a- (green circles) and b-waves (red squares). (C) and (D): the corresponding scotopic b-: a-wave ratios at (C) flash intensity of 1 cd·s/m² and (D) at all flash intensities >0.1 cd·s/m². (E) Amplitude of photopic b-waves from Rho^P23H/+ mice and WT mice at P70. Data and error bars are means±SD. Raw data points were overlaid as black scatters. *, p-value<0.05 by unpaired, two-tailed Student’s t-test. N=6-8 mice.

4. Discussion

In this study, we used SD-OCT to thoroughly characterize the temporal changes in the Rho^P23H/+ mouse model of RP; focusing on the entire retinal depth, including the outer retina and inner retina. Conventional measuring of retinal layer thickness requires researchers to manually sample 6-8 evenly spaced positions along OCT B-scans, in order to generate a spidergram. In this study, we improved the methodology by averaging thickness at every x-pixel location with customized software, thereby significantly minimizing error caused by image noise. While retinal development completes approximately by P15, the morphology of the retina is often thought to be stable during the young adulthood of healthy mice. Interestingly, we found that young, healthy retinal layers continue changing until over 2 months of age; suggesting a long remodeling phase after retinal neurons are fully differentiated. From P28-P69, we observed a gradual and notable reduction of retinal layer thicknesses in WT mice. Our finding is supported by previous reports on C57 mice (Ferdous et al., 2021) and rats (Ryals et al., 2017) that also noted the progressive retinal thinning from 2-30 months of age. The progressive thinning of total retina has been attributed to normal aging factors in WT mice (Ferdous et al., 2021). However, as we followed retinal changes in the early adulthood of the animals, we suggest this general thinning of the whole retinal layers may be a result of natural retinal adaptation to the daily visual functional needs, via programmed apoptosis and retinal neural processing. Our results from WT mice between P28-69 support our hypothesis as the time-dependent thinning was more prominent in the inner retina than the outer retina. Additionally, the thinning of GCC and INL occurred at different time points (Fig. 3). These results suggest a pruning of different retinal neurons as reflected by the dynamic retinal morphological changes that occur as long as 2 months postnatally.

The RHO-P23H mutation causes rhodopsin to misfold and accumulate in the endoplasmic reticulum for degradation via proteasome and autophagy pathways (Wu et al., 1998). Due to the high biosynthesis rate of rhodopsin, misfolded rhodopsin is continuously produced and degraded by the rods. The overwhelmed proteolytic load of continually degrading misfolded rhodopsin, eventually, leads to rod cell death (Mendes et al., 2005). By comparing the thickness of the RP Rho^P23H/+ mouse model with age-matched WT controls, we were able to delineate the events occurring under this pathological condition throughout different areas of the retina. Firstly, we confirmed the shorter IS&OS in the mutant retina, as well as the progressive degeneration of the ONL follows a similar timeline as previously reported (Chiang et al., 2015; Sakami et al., 2011). The observed temporal changes of these two layers demonstrate that the thinning of the IS&OS and ONL are not correlated; suggesting two individual thinning events occurring, despite both layers involving photoreceptors. The relatively static but shorter IS&OS thickness over time in the Rho^P23H/+ mice is a result of reduced rhodopsin levels in the OS that remain stable during the observation time window. As rhodopsin is the main protein component in the OS, total rhodopsin protein amount determines OS length. For example, Rho^+/− animals do not develop retinal degeneration, but their retinae show shorter OS due to reduced rhodopsin protein levels (Liang et al., 2004). Additionally, our observed progressive decrease of ONL from P28-P69, is coherent to histological data showing a two-phase photoreceptor degeneration with 41% loss of ONL thickness from P15-P28; which decreased a further 20% from P28-P69 (Chiang et al., 2015).

Inner retinal neurons connect with the degenerating photoreceptors; thus, their morphological changes are secondary to ONL degeneration (Fig. 3H&I). The thinning of OPL is partially due to the death of rod photoreceptors and consequential diminished rod synapses. Subsequently followed by remodeling of the rod-connecting synapses of inner retinal neurons. Indeed, inner retinal remodeling is observed and reported by retinal histology, and supported by transcriptome analyses showing altered expression of genes that encode both synapse proteins and neurite modality regulatory proteins (Leinonen et al., 2020) in Rho^P23H/+ mice (as early as 1 month of age).

Interestingly, apart from OPL thinning, we observed the preservation of a thicker INL in the Rho^P23H/+ mice, as compared to WT control (Fig. 3G). Similar INL thickening was also reported in patients with retinal degeneration (Huang et al., 2014). This result suggested a greater number, an enlarged size, or altered morphology of inner retinal neurons were preserved, which, we hypothesize may be a compensatory response to rod cell death in order to maintain visual function. As seen in the histology images, the diameter and shape of INL nuclei were not changed comparing the Rho^P23H/+ mice vs. WT control, and thus we think it is more possible that the thicker INL is due to more nucleus count in the Rho^P23H/+ mice. Our ERG data showed a growing scotopic b-: a-wave ratio in the Rho^P23H/+ mice from P17-70 (time period of rod degeneration), as compared to WT; suggesting a functional compensation of inner retinal neurons to reduced rod function. Leinonen et al. reported that Rho^P23H/+ mice at 2-3 months of age retained similar visual contrast sensitivity, comparable to WT controls when researchers genetically blocking cone phototransduction by knocking out Gnat2; suggesting that the closed-to-normal rod-derived visual contrast sensitivity, despite only 40% remaining rods is maintained by the compensatory increase in the ratio of bipolar: rod cell response upon illumination (Leinonen et al., 2020). They also showed that Rho^P23H/+ mouse retinae express higher levels of genes encoding postsynaptic receptors including Grm6, Trpm1, and Gpr179, suggesting higher bipolar synaptic sensitivity. They claimed that these changes are due to the higher number of receptors per bipolar synapse. However, their immunohistochemistry data showed that the 5-month-old Rho^P23H/+ mouse retina has thicker INL compared to the WT control. Their data echo our observation and indicate that the functional induction of bipolar response in the Rho^P23H/+ mice may not only be due to higher postsynaptic receptors per bipolar cell, but also more bipolar cells preserved along with aging. Collectively, our results of OCT, histology, and ERG suggest the compensatory increase in bipolar: rod cell response may not only be due to neurite remodeling but also the preservation of more bipolar/inner retinal neurons.

Why there can be more INL nucleus count in the Rho^P23H/+ mice vs WT control at P68. It is known that after postmitotic retinal cell differentiation is completed at P14, the neural retina also undergoes physiological cell death, a process essential during the formation of the functional retinal neural network. While photoreceptors do not reduce in numbers, featuring the reduction of RGCs and INL cell count. This pruning process is currently known to be facilitated by the microglia under a noninflammatory environment (Schafer et al., 2012; Vecino et al., 2016, 2016). While Rho^P23H/+ mice undergo pathological photoreceptor degeneration, the microglia in these animals are in a proinflammatory status, migrating to the injury site, the out retinal layers, to phagocytose the dead cells (Newton and Megaw, 2020; Yu and Saban, 2023). These microglia may lose the retinal pruning functionality under healthy physiological conditions, which may lead to a higher number of neurons in the INL.

Distinctively, the temporal changes of retinal ganglion cells (NFL+IPL=GCC, Fig. 3D-F) observed in WT mice were unaffected in Rho^P23H/+ mice. This can be anticipated as RGCs are not directly connected to the degenerative photoreceptors. However, previous studies on the physiological characteristics of RGCs of Rho^P23H/+ transgenic rats showed several alterations in terms of spontaneous firing rates, spike amplitudes, as well as the receptive field (Sekirnjak et al., 2011). Furthermore, OCT studies on RP patients (33.1 ± 15.9 years of age) showed that NFL thickness was significantly thicker than age-matched normal controls (Hood et al., 2009). These discrepancies may be due to the different time window of our study compared to others. For Rho^P23H/+ mice at P17, the OCT images were not as good due to the incomplete opening of the eye lids. However, at a later timepoint (such as 1 year), the thickness of retinal layers was not measurable due to fully degeneration. These facts emphasize the need for future long-term investigations with additional comprehensive morphological and functional evaluations.

5. Conclusion

We profiled the natural thinning of retinal layers in the normal retinae of young WT adult mice by recording temporal retinal morphology using OCT and subsequent accurate retinal layer segmentation. Moreover, we confirmed the degeneration of the ONL and IS&OS thinning, as we previously documented in the Rho^P23H/+ mouse model. With significantly improved fidelity of thickness measurements, we also observed the thinning of OPL and preservation of INL thickness, while the RGCs remained unaffected. The compensatory effects of inner retinal neurons were not only observed morphologically by OCT and histology, but were also seen functionally by ERG. These results set a foundation for the feasibility of visual restoration therapies targeting the inner retinal neurons.

Quantitative measurements of progressive photoreceptor degeneration

Quantification of the thinning of RPE layer

Characterization of gradual alteration of inner retinal layers

Histology validation of INL compensation to with histology

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.