Enhancing Experience-Dependent Plasticity Accelerates Vision Loss in a Murine Model of Retinitis Pigmentosa

Abstract

Purpose

Retinitis pigmentosa (RP) causes retinal degeneration and the progressive loss of vision. Here, we evaluated whether enhancing experience-dependent plasticity prolongs vision in a mouse genetic model of autosomal dominant RP.

Methods

First, we quantified the loss of visual acuity under both mesopic and photopic conditions for mice heterozygous for the P23H mutation in the Rhodopsin gene (Rho^P23H/+) and littermate controls. Then, we compared acuity to the magnitude of the b-wave measured from full-field electroretinograms (ffERG) or the response properties of excitatory neurons in layer 2/3 of primary visual cortex calculated from in vivo calcium imaging, under both mesopic and photopic conditions for each mouse. Last, we repeated these measurements and comparisons for mice that also lacked a functional gene for the Nogo-66 receptor (Ngr1), a gene required to close the critical period for visual plasticity.

Results

Acuity progressively declined under mesopic conditions followed by photopic conditions. Acuity deficits only broadly correlated with the retinal response measured by the electroretinogram. In contrast, acuity deficits were consistent with the percent of cortical excitatory layer 2/3 neurons responsive to higher spatial frequency visual stimuli. Acuity deficits were more severe at earlier ages in Ngr1^−/−;Rho^P23H/+ mice, and no mice possessed sufficient acuity at 10 to 12 months of age to be tested on the visual water task.

Conclusions

The Rho^P23H/+ mouse model of RP displays progressive loss of visual acuity first under mesopic conditions and then under photopic conditions. The severity of acuity impairment correlates only broadly with deficits in retinal function measured with the ffERG but is consistent with the percentage of neurons in primary visual cortex responsive to higher spatial frequency visual stimuli measured with calcium imaging. Contrary to our hypothesis, enhancing neural plasticity is deleterious for visual function in Rho^P23H/+ mice.

Enhancing neural plasticity is pursued as a therapeutic intervention for neurodevelopmental disorders, neural injury, and neurodegenerative conditions.^1–3 In the visual system, sustaining or reactivating plasticity otherwise confined to a developmental critical period promotes recovery of acuity in rodent models of amblyopia.⁴ Whether similar plasticity can prolong useful vision and acuity during retinal degeneration is unclear. Testing this possibility may provide insight into the potential for modulating plasticity as a treatment for degenerative eye conditions, as well as other forms of neurodegeneration.

Retinitis pigmentosa (RP) is a group of inherited retinal disorders that result in progressive loss of vision and has a worldwide prevalence of 1 in 4000.⁵ RP preferentially disrupts the function of rod photoreceptor cells, and RP patients are typically diagnosed in late adolescence to early adulthood following the onset of night blindness and impaired peripheral vision.⁶ In the years following, vision continues to deteriorate and can result in low vision or complete vision loss.⁷ Longitudinal studies of patients with RP reveal that the loss of visual acuity is significantly slower than the deterioration of photoreceptors.^8–10 This finding has led to the hypothesis that visual plasticity may play a role in preserving vision.^11,12

Approximately 30% of cases of RP are associated with an autosomal dominant (adRP) mutation.¹³ More than 25 different genes and over 1000 mutations are associated with adRP, but the P23H point mutation in the rhodopsin gene (Rho) is the most prevalent among North Americans, accounting for more than 10% of all cases.^13,14 The characteristic features of human P23H adRP are conserved in mice heterozygous for the Rho-P23H mutation.¹⁵ These mice exhibit a progressive decline in the ffERGs that is associated with thinning of the outer nuclear layer (ONL), where photoreceptors reside. The typical thickness of the ONL for wild-type (WT) adult mice is ∼50 µm.^16,17 In Rho^P23H/+ mice, the thickness of the ONL is reduced to 20–30 microns at 4 months (mo.) of age, 10–20 microns at 6 mo., and 10 microns or less at 10 mo. and older.^16–26

Visual acuity is constrained by the density of cones and retinal ganglion cells (RGCs) but also shaped by processing of visual information by visual cortex.^27,28 In the mouse, visual acuity can be measured with the visual water task.²⁹ This two-alternative forced choice test reveals that the typical visual acuity of WT adult mice under photopic conditions (photopic, ∼100 cd/m²) is between 0.4 and 0.5 cycles/deg (cpd).^29,30 This acuity requires primary visual cortex (V1).²⁸ A small fraction of excitatory neurons in V1 respond to sinusoidal gratings at spatial frequencies (SFs) slightly beyond behavioral estimates of visual acuity.^31,32 Other studies have measured the optomotor response, a reflex in rodents mediated by the superior colliculus, as a surrogate for visual acuity.²⁰ The spatial frequency thresholds measured for the optomotor reflex are lower than cortical-dependent visual tasks and mature faster than acuity measured with the visual water task.^30,33 Overall, the relationship between impaired acuity, retinal function, and cortical function, has not been triangulated for any disorder of vision including RP.

Nogo-66 receptor 1 (NGR1) is one of several factors that limit plasticity in the adult brain.³⁴ NGR1 is enriched at excitatory synapses.^35,36 Adult knock-out mice lacking a functional Ngr1 gene (Ngr1^−/−) retain visual plasticity otherwise confined to the critical period and recover normal visual acuity in a mouse model of amblyopia, a prevalent developmental visual disorder.^30,37–39 Here, we established a framework for registering acuity to retinal function and cortical function for Rho^P23H/+ mice and then tested if enhancing visual plasticity by eliminating Ngr1 expression would better sustain acuity during progressive retinal degeneration.

Methods and Materials

This study did not generate new unique reagents. Please refer to the Table for the key reagents used in this study.

Table.

Reagents and Resources Employed in the Study

Reagent or Resource	Source	Identifier
Deposited data
Calculated tuning properties for all neurons	—	Not applicable
Experimental models: organisms/strains
Mouse: B6;DBA-Tg(tedO-GCaMP6s)2Niell/j	The Jackson Laboratory	RRID:ISMR_ JAX:024742
Mouse: B6;Cg-Tg(Camk2a-tTA)1Mmay/DboJ	The Jackson Laboratory	RRID:ISMR_ JAX:007004
Mouse: B6.129S6(Cg)-Rhotm1.1Kpal/J	The Jackson Laboratory	RRID:IMSR_JAX:017628
Mouse: Nogo-66 receptor (ngr1/rtn4r) constitutive mutant mice	—	Kim et al.42
Software and algorithms
MATLAB	MathWorks	https://www.mathworks.com/
Processing	Processing	https://processing.org/

Experimental Model and Subjects

All procedures were approved by University of Louisville and the University of Arizona Institutional Animal Care and Use Committees and were in accord with guidelines set by the U.S. National Institutes of Health. Mice were anesthetized by isoflurane inhalation and killed by carbon dioxide asphyxiation or cervical dislocation following deep anesthesia in accordance with approved protocols. Mice were housed in groups of 5 or fewer per cage in a 12-hour/12-hour light/dark cycle. Animals were naïve subjects with no prior history of participation in research studies.

Mice

Rho^P23H/+ mice (017628) were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) and crossed to double transgenic mice, Tg-CaMKII-tTA (007004) and Tg-TRE-GCaMP6s (024742), to express GCaMP6s in forebrain excitatory neurons.^40,41 Mice were genotyped with primer sets suggested by The Jackson Laboratory. These mice were crossed onto the Ngr1^−/− background.⁴² Mice were genotyped with custom primer sets.

Experimental Design

Visual acuity was measured through the right eye for all mice under light-adapted photopic conditions and subsequently in dark-adapted mesopic conditions (WT, n = 30, 14 males and 16 females; Rho^P23H/+, n = 44, 25 males and 19 females; Ngr1^−/−, n = 33, 14 males and 19 females; Ngr1^−/− Rho^P23H/+, n = 46, 21 males and 25 females). ERGs and histology were then performed on a subset of these mice that did not carry both transgenes for expression of GCaMP6s. Mice expressing GCaMP6s were directed to calcium imaging experiments.

Monocular Lid Suture

To measure visual acuity from the right eye, the left eyelid was sutured shut using a single mattress suture made of 6–0 polypropylene monofilament (Prolene 8709H; Ethicon, Somerville, NJ, USA) under brief isoflurane anesthesia (2%). The knot was secured with cyanoacrylate glue. The eyelid suture was removed when the visual water task was fully completed.

Visual Water Task

The visual water task is a two-alternative forced choice discrimination task that measures visual acuity mediated by visual cortex.²⁹ In brief, a monitor was positioned at the wide end of a custom-made trapezoidal-shaped tank behind clear acrylic. An opaque divider (46 cm long) divided the monitor and created a point where the mouse had to choose which side to swim toward. One side of the screen displayed a sinusoidally modulated grating at 95% contrast. The other side of the screen displayed an isoluminant gray screen. The spatial frequency of the grating was calculated from the screen to the choice point (46 cm). The tank was filled with room-temperature water, and a hidden platform was submerged below the surface of the water in front of the monitor displaying the grating.

Mice were divided into two groups (flights) of ≤6 mice per flight and a total of ≤12 mice in a cohort, with alternating rest and test blocks. Mice were trained to swim toward the monitor with the low-frequency (0.05 cpd) grating behind the hidden platform. The side on which the stimulus was displayed was pseudorandom. In a successful trial, the mouse swam past the midline divider to the monitor with the grating and the platform. In a failed trial, the mouse crossed the midline divider to the side with the isoluminant gray screen. Each mouse completed up to 50 trials/day in blocks of 10. To advance past training, mice had to achieve two consecutive blocks with 80% accuracy or better. Before testing, the left eye of each mouse was sutured shut (monocular deprivation), and an additional training period was repeated with the same criterion. The mouse had to complete two consecutive blocks with 8 out of 10 correct trials (80%). Typically, this was accomplished in 20 to 50 trials over 1 to 2 days.

Testing began with discrimination between a low-frequency grating (0.05 cpd) and an isoluminant screen. After three consecutive correct choices (3/3), the SF was increased by one additional cycle of the grating on the subsequent trials until the mouse could no longer discriminate between the stimulus and control. Validation of this measure included retesting at the failed SF, requiring either five consecutive correct trials (5/5) or 8 of 10 correct trials at the failed SF to advance. If unsuccessful (<8/10 correct trials) at the failed SF, retraining at half the SF was conducted to correct any possible side bias in their performance. The mouse would then resume testing at the SF two steps below the failed SF. Testing ended when mice failed three times at adjacent SFs (i.e., 0.42, 0.47, and 0.44). The three failed SFs were averaged and used to define the visual acuity threshold.

Photopic visual acuity was tested with monitor brightness at the choice point of 100 cd/m². Mesopic visual acuity was tested with monitor brightness at the choice point of 0.025 cd/m², which was achieved by lowering the monitor brightness to 25% and using neutral density (ND) filters (ND 0.9, ND 0.9, and ND 1.2) after 1 hour of dark adaptation with room luminance ≤ 0.01 cd/m². All equipment emitting light was covered; the experimenter was dark adapted for 20 minutes and worked under dim red light.

Electroretinogram

Mice were dark adapted overnight, anesthetized with a solution of ketamine (80 mg/kg) and xylazine (16 mg/kg) administered via intraperitoneal (IP) injection, and prepared for ffERG recordings under dim red light. Pupils were dilated and accommodation relaxed with the application of eye drops, with 2.5% phenylephrine hydrochloride ophthalmic solution (NDC 82260-102-10; Bausch + Lomb, Bridgewater, NJ, USA) and 0.5% tropicamide ophthalmic solution (NDC 24208-590-64; Bausch + Lomb) for 30 seconds. Eyes were then rinsed three times with sterile irrigating solution (balanced salt solution; Alcon, Geneva, Switzerland). A contact lens with a gold electrode (LKC Technologies, Germantown, MD, USA) was placed on the cornea, and its position was maintained with a hypromellose ophthalmic solution (VISTA Gonio Eye Lubricant; Keeler, Malvern, PA, USA). Ground and reference needle electrodes were placed in the tail and on the midline of the forehead, respectively. Body temperature was maintained using a feedback-controlled electric heating pad (37°C).

Mesopic responses were measured at two test flash intensities (0.025 and 0.05 cd·s/m²) after 10 minutes of adaptation to a rod-saturating background (20 cd/m²). Photopic responses were measured at two test flash intensities (10 and 100 cd·s/m²). For each stimulus, a-wave and b-wave amplitudes and b-wave implicit times were analyzed with a custom MATLAB code (MathWorks, Natick, MA, USA). The a-wave was measured from baseline (recorded 10–20 ms before stimulation) to the negative trough. The b-wave was measured from the a-wave trough to the b-wave peak. The flash intensities in the protocol for this experiment—mesopic flash of 0.025 cd·s/m² and photopic flash of 100 cd·s/m²—matched the visual water task screen luminance, allowing for direct comparison between behavioral vision acuity and retinal function. ERGs were performed within 2 weeks after concluding measuring acuity with the visual water task.

Cranial Windows

Widefield epifluorescent calcium imaging and two-photon calcium imaging were performed though a cranial window as previously described.^43,44 In brief, mice were administered carprofen (5 mg/kg) and buprenorphine (0.1 mg/kg) for analgesia and anesthetized with isoflurane (4% induction, 1%–2% maintenance). The scalp was shaved. Mice were mounted on a stereotaxic frame with palate bar and their body temperature maintained at 37°C with a heat pad controlled by feedback from a rectal thermometer (TCAT-2LV; Physitemp Instruments, Clifton, NJ, USA). The scalp was resected, the connective tissue removed from the skull, and a custom aluminum headbar affixed with C&B Metabond (Parkell, Brentwood, NY, USA). A circular region of bone 3 mm in diameter centered over left visual cortex was removed using a high-speed drill (Foredom, Danbury, CT, USA). Care was taken not to perturb the dura. A sterile 3-mm circular glass coverslip was sealed to the surrounding skull with cyanoacrylate (Pacer Technology, Ontario, CA, USA) and dental acrylic (Ortho-Jet; Lang Dental, Wheeling, IL, USA). The remaining exposed skull was likewise sealed with cyanoacrylate and dental acrylic. Mice recovered on a heating pad and returned to standard housing for at least 2 days prior to two-photon imaging. Implanting the cranial window and imaging were performed within 2 weeks after concluding measuring acuity with the visual water task.

Widefield Epifluorescent Calcium Imaging

After implantation of the cranial window and before two-photon imaging, the binocular zone of visual cortex was identified with widefield calcium imaging similar to our method for optical imaging of intrinsic signals.^45,46 In brief, mice were anesthetized with isoflurane (4% induction), provided a low dose of the sedative chlorprothixene (C1761, 0.5 mg/kg IP; Sigma-Aldrich, St. Louis, MO, USA) and secured by the aluminum headbar. The eyes were lubricated with a thin layer of ophthalmic ointment (Puralube; Dechra Pharmaceuticals, Northwich, UK). Body temperature was maintained at 37°C with a heating pad regulated by rectal thermometer. Visual stimulus was provided through custom-written software. A monitor was placed 25 cm directly in front of the animal and subtended +40 to −40 degrees of visual space in the vertical axis. A horizonal white bar (2° high × 20° wide) centered on the 0° azimuth drifted from the top to bottom of the monitor with a period of 8 seconds. The stimulus was repeated 60 times. Cortex was illuminated with blue light (Semrock Optical Filter, 475 ± 30 nm with 475/35 nm BrightLine single-band bandpass filter; IDEX Health & Science) from a stable light source (Intralux DC-1100; Volpi Group, Dietikon Switzerland). Fluorescence was captured utilizing a green filter (HQ620/20, IDEX Health & Science) attached to a tandem lens (50-mm lens; Computar, Carlsbad, CA, USA) and camera (Manta G-1236B; Allied Vision, Stadtroda, Germany). The imaging plane was defocused to approximately 200 µm below the pia. Images were captured at 10 Hz as images of 1024 × 1024 pixels and 12-bit depth. Images were binned spatially 4 × 4 before the magnitude of the response at the stimulus frequency (0.125 Hz) was measured by Fourier analysis.

Visual Stimuli and Two-Photon Calcium Imaging

Visual stimulus presentation and image acquisition were both performed according to our published methods which were modified from previous studies.^43,47–49 In brief, a battery of static sinusoidal gratings was generated in real time with custom software (Processing and MATLAB). Stimulus presentation was synchronized to the imaging data by time stamping the presentation of each visual stimulus to the image acquisition frame number with a transistor–transistor logic (TTL) pulse generated with the Arduino platform at each stimulus transition. Orientation was sampled at equal intervals of 15° from 0° to 150° (12 orientations). SFs were sampled in 10 steps on a logarithmic scale at half-octaves from 0.028 to 1.02 cpd. An isoluminant grey screen (blank) was included as a ninth step in the SF sampling as a control. Spatial phases were equally sampled at 45° intervals from 0° to 315° for each combination of orientation and SF. Gratings with random combinations of orientation, SF, and spatial phase were presented at a rate of 4 Hz on a monitor with a refresh rate of 60 Hz. Imaging sessions were 20 minutes (4800 gratings presented in total). Consequently, each combination of orientation and SF was presented 33 times on average (range, 17–56). The monitor was centered on the zero azimuth with elevation 35 cm away from the mouse and subtended 45° (vertical) by 80° (horizontal) of visual space.

Imaging was performed with a resonant scanning two-photon microscope controlled by Scanbox image acquisition and analysis software (Neurolabware, West Hollywood, CA, USA). The objective lens was fixed at vertical for all experiments. Fluorescence excitation was provided by a tunable wavelength infrared laser (Ultra II; Coherent, Saxonburg, PA, USA) at 920 nm. Images were collected through a 16× water-immersion objective (0.8 NA; Nikon, Tokyo, Japan). Images (512 × 796 pixels, 520 × 740 µm) were captured at 15.5 Hz at depths between 150 and 400 µm. Eye movements and changes in pupil size were recorded using a Genie Nano-M1280 camera (Teledyne DALSA, Waterloo, ON, Canada) fitted with a 50-mm f/1.8 lens (Computar) and 800-nm long-pass filter (Edmunds Optics, Barrington, NJ, USA). Imaging was performed on alert mice positioned on a spherical treadmill by the aluminum headbar affixed to the skull. The visual stimulus was presented to the contralateral eye by covering the fellow eye with a small custom occluder. Neutral density filters were positioned in front of the monitor presenting visual stimuli for imaging of neuronal activity under mesopic conditions. Mice were dark adapted for 30 minutes or longer prior to imaging. Mice were imaged within 2 weeks of having their visual acuity measured with the visual water task.

Image Processing

Image processing was performed as described previously.^47,49 Imaging series for each eye were motion corrected with the SbxAlign tool. Regions of interest (ROIs) corresponding to excitatory neurons were selected manually with the Scanbox sbxsegmenttool following computation of pixel-wise correlation of fluorescence changes over time from 350 evenly spaced frames (∼1%). ROIs for each experiment were determined by correlated pixels the size similar to that of a neuronal soma. The fluorescence signal for each ROI and the surrounding neuropil were extracted from this segmentation map.

Image Analysis

Image analysis was performed as described previously with minor modifications.⁴⁷ The fluorescence signal for each neuron was extracted by computing the mean of the calcium fluorescence within each ROI and subtracting the median fluorescence from the surrounding perimeter of neuropil.^49,50 An inferred spike rate (ISR) was estimated from the adjusted fluorescence signal with the Vanilla algorithm.⁵¹ A reverse correlation of the ISR to stimulus onset was used to calculate the preferred stimuli.^47–50 Neurons that satisfied three criteria were categorized as visually responsive: (1) The ISR was highest with the optimal delay of 4 to 9 frames following stimulus onset; this delay was determined empirically for this transgenic GCaMP6s mouse.^47,49 (2) The signal-to-noise ratio was greater than 1.3; the signal was the mean of the spiking standard deviation at the optical delay between 4 and 9 frames after stimulus onset and the noise at frames −2 to 0 before the stimulus onset or 15 to 18 frames after it.^48,49 (3) The neuron responded to at least 13% of the presentations of the preferred stimulus. Visual responsiveness for every neuron was determined independently for each eye. The visual stimulus capturing the preferred orientation and SF was determined from the matrix of all orientations and SFs presented as the combination with highest average ISR.

The preferred orientation for each neuron was calculated as

$$Orientation=\frac{arctan\left({{\sum }_{n}O}_n*e^{i*2*\pi *{\theta }_n/180}\right)}{2}$$

where O_n is a 1 × 12 array of the mean z-scores associated with calculation of the ISR at orientations Q_n (0°–150° spaced every 15°). Orientations calculated with this formula were expressed in radians and converted to degrees. The tuning width was the full width at half maximum of the preferred orientation.

The preferred SF for each neuron was calculated as

$$SF={10}^{\frac{{\sum }_{k}S{f}_{\kappa }*\mathrm{lo}{\mathrm{g}}_{{10}^{{\omega }_{\kappa }}}}{{\sum }_{\kappa }S{f}_{\kappa }}}$$

where Sf_k is a 1 × 10 array of the mean z-scores at SFs ω_k(10 equal steps on a logarithmic scale from 0.028 to 1.02 cpd). Tails of the distribution were clipped at 25% of the peak response. The tuning width was the full width at half maximum of the preferred SF in octaves. The percent visually responsive neurons with significant responses at each SF was determined by comparing the distribution of ISR values at each SF versus the stimulus blank with a Kruskal–Wallis test with Dunn's correction for 10 comparisons. Neurons with P < 0.01 for a given SF were considered significant responses at that SF.³²

Retinal Dissection

Mice were euthanized with a solution of ketamine (80 mg/kg) and xylazine (16 mg/kg) administered via IP injection. After a toe pinch evoked no reaction, cervical dislocation was performed as a secondary form of euthanasia. The superior portion of the right eye was marked using a pen (Securline Sterile Surgical Skin Marker; Aspen Surgical, Caledonia, MI, USA), and the eye was enucleated. A hole punched with a #11 scalpel and a circumferential cut separated the cornea and lens from the eyecup. The retina was removed after cutting the optic nerve and separating it from the posterior eye cup. One deep cut was made at the mark on the superior retina, and two more minor cuts were made at the nasal and temporal sides of the inferior retina. This allowed the retina to be flattened on a glass slide, where it was fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) solution (pH 7.4) for 20 minutes, and washed with PBS. An ascending sucrose series was used to cryoprotect the retinal whole mounts (5%, 10%, and 15%; 1 hour each), followed by 20% sucrose/PBS overnight. The whole-mount retina was placed in 2:1 Tissue-Tek O.C.T. Compound (4583; Sakura Finetek, Torrance, CA, USA) and 20% sucrose for 1 hour, embedded in the same solution, and frozen in a 2-methyl butane liquid nitrogen-cooled bath. Retinas were sectioned at 14 µm using a Leica CM1850 cryostat (Leica Biosystems, Deer Park, IL, USA), mounted on Fisherbrand Superfrost Plus Microscope Slides (12-550-15; Fisher Scientific, Waltham, MA, USA), dried for 60 minutes on a heating plate (37°C), and then stored at −80°C.

Histochemistry

Slides were thawed on a heating plate (37°C) for 50 to 60 minutes, and sections outlined on the slide with a hydrophobic barrier (PAP pen, H-4000; Vector Laboratories, Burlingame, CA, USA). Each slide was washed with ∼200 µL of 1× Invitrogen PBS (AM9625; Thermo Fisher Scientific, Waltham, MA, USA) for 5 minutes, and sections were permeabilized with 0.5% Triton X-100 (ICN807423; MP Biomedicals, Irvine, CA, USA) in 1× PBS. A blocking solution of 1× PBS with 5% normal donkey serum (S30, 100 mL; MilliporeSigma, Burlington, MA, USA) was applied, and slides were incubated for 1 hour at room temperature followed by a solution containing 4′,6-diamidino-2-phenylindole (DAPI; D1306, 1:2000; Thermo Fisher Scientific) to stain somas.

ONL Measurements

Images of the retinal sections were taken on a confocal laser scanning microscope (FLUOVIEW FV4000; Olympus, Tokyo, Japan). A stitched image of the entire section was acquired, and high-power images were acquired at 100-µm intervals. Across the section, at each area, a z-stack of 1-µm images (14 µm total) was acquired using a 40× objective with a 1.5× zoom. Olympus cellSens Dimensions imaging software measured the retinal layers at each area from a projection consisting of the middle five images of the z-stack, ∼5 µm in depth (the diameter of photoreceptor somas). Six evenly spaced measurements of the ONL were made across each image (superior to inferior). All measurements for a single retina (6 images × 6 measurements = 36 measurements) were averaged to define the average ONL thickness.

Statistics

No statistical methods were used to predetermine sample size. All statistical analyses were done using Prism 8 (GraphPad Software, Boston, MA, USA). Multiple comparisons were tested with analysis of variance (ANOVA) tests and paired tests with mixed-effects analysis for data not divergent from a parametric distribution and the non-parametric Kruskal–Wallis test for data that did not fit a parametric distribution.

Results

The visual water task measures functional vision. We modified this task to test and compare visual acuity under normal room light (photopic) and dark-adapted low light (mesopic) conditions. Photopic conditions were 100 cd/m². Mesopic conditions were 0.02 cd/m². These luminance conditions correspond to 1.13 × 10⁵ Rh*/rod/s and 30 Rh*/rod/s, respectively.⁵² This mesopic condition, although not rod isolating, was the lowest condition under which the experimenter could manage this task. In addition, we tested acuity under monocular viewing conditions using the right eye by temporarily suturing the left eye closed. This eliminated the potential compensation by binocular vision, precluded differences in acuity between the two eyes, and permitted direct comparisons of measurements of retinal and cortical function. Rods vastly outnumber cones in the mouse retina and represent nearly 97% of the photoreceptor population.⁵³ Thus, we anticipated that any deficits in visual acuity would be evident first under mesopic conditions where rod function predominates.

First, we measured photopic acuity (Fig. 1; Supplementary Fig. S1). WT mice at ages ranging in age from 1 to 3 mo. to 7 to 9 mo. displayed acuity in the typical range of 0.4 to 0.5 cpd.²⁹ WT mice at 10 to 12 mo. had slightly lower acuity (Fig. 1a). We then tested mesopic acuity and found that WT mice exhibited similar acuity under photopic and mesopic conditions at all ages (Fig. 1a and data not shown).

Figure 1.

Rho^P23H/+ mice display a progressive deficit in acuity under both normal and low light conditions. (a) Visual acuity of WT and Rho^P23H/+ mice under photopic (P) and mesopic (M) conditions. Mesopic conditions are indicated by the gray bars. Lines connect the acuity measurements for each mouse under both conditions within an age group. P values above the symbols compare the acuity of Rho^P23H/+ mice to age- and luminance-matched WT mice (Brown–Forsythe and Welch ANOVA). P values below the pairs of symbols compare age groups of Rho^P23H/+ mice between photopic and mesopic conditions (mixed-effects analysis). Tests were corrected for all 16 comparisons between the two tests. (WT P23H: 1–3 mo., n = 8 and 11; 4–6 mo., n = 5 and 16; 7–9 mo., n = 6 and 13; 10–12 mo., n = 8 and 13). Circles and triangles represent individual mice (b) Fraction of mice under both photopic conditions that were unable to reach threshold criterion for testing of photopic visual acuity (WT P23H: 7–9 mo., n = 0/6 and 2/15; 10–12 mo., 1/9 and 4/17).

By comparison, Rho^P23H/+ mice displayed significantly lower photopic acuity than age-matched WT mice at 7 to 9 mo. The lower photopic acuity of WT mice at 10 to 12 mo. mitigated the difference between genotypes (Fig. 1a). Rho^P23H/+ mice also had significantly lower mesopic acuity than WT mice at all ages tested. As expected, the photopic visual acuity of Rho^P23H/+ mice was significantly higher than mesopic visual acuity across age. In addition to lower acuity, a fraction of Rho^P23H/+ mice 7 to 9 mo. old and 10 to 12 mo. old were unable to reach the criterion for testing, consistent performance at better than 70% at 0.05 cpd under photopic conditions; therefore, their visual acuity could not be assessed (Fig. 1b). In some cases, mice could not even orient toward the side of the tank displaying the visual stimulus. We set the acuity of these mice at 0.01 cpd for subsequent comparisons.

The prominent preclinical measurement of retinal dysfunction is the ffERG.⁵⁴ The b-wave of the ERG represents the aggregate activity of ON bipolar cells that are postsynaptic to both rod and cone photoreceptors.⁵⁵ By 6 mo. of age, the photopic b-wave in Rho^P23H/+ mice is less than a third of the typical amplitude of WT mice, and the mesopic b-wave is severely attenuated.¹⁵ We measured the b-wave amplitude in response to full-field flashes for WT and Rho^P23H/+ mice at photopic and mesopic conditions that matched the parameters used to test visual acuity (Fig. 2). The Rho^P23H/+ mice at 7 to 9 mo. exhibited reductions in the photopic and mesopic b-waves²⁰ (Figs. 2a–d). These deficits were exacerbated at 10 to 12 mo., we could not record a photopic (4/15) or mesopic (11/15) b-wave in many Rho^P23H/+ mice (Figs. 2c, 2d).

Figure 2.

Reduced b-wave amplitude in Rho^P23H/+ mice is associated with a range of acuities. (a, b) Average ± SEM ERG traces of WT (gray/black lines) and P23H/+ mice (blue/purple lines) at 7 to 9 mo. (solid lines) and at 10 to 12 mo. (dashed lines) recorded for photopic flashes (16 dB) (a) and mesopic flashes (−20 dB) (b) (WT, 7–9 mo., n = 9; WT, 10–12 mo., n = 15; P23H, 7–9 mo., n = 11; P23H, 10–12 mo., n = 15). (c, d) Corresponding column plots of mean b-wave amplitude per mouse for photopic flashes (WT, gray; P23H, blue) and mesopic flashes (WT, black; P23H, purple). Circles represent individual mice at 7 to 9 mo., and triangles represent mice at 10 to 12 mo. Horizonal lines indicate the mean. (e, f) Scatterplots of visual acuity and b-wave amplitude for the subset of WT and P23H/+ mice with both measurements for photopic (e) and mesopic (f) conditions. Circles represent mice at 7 to 9 mo., and triangles represent mice at 10 to 12 mo. The vertical line represents mice with b-wave amplitudes less than 8 µV (WT, 7–9 mo., n = 5; WT, 10–12 mo., n = 7; P23H, 7–9 mo., n = 7; P23H, 10–12 mo., n = 9).

Next, we compared photopic and mesopic acuity to the mean amplitude of their b-waves in a subset of mice for which both were measured (Figs. 2e, 2f). There was a clear separation between the distributions for WT and Rho^P23H/+ mice. Unexpectedly, the b-wave amplitude was a poor predictor of acuity. Rho^P23H/+ mice at 7 to 9 mo. and at 10 to 12 mo. yielded photopic b-wave amplitudes that were ∼10% or less of the average for WT mice yet possessed acuities ranging from unable to reach threshold for testing up to 0.31 cpd, which overlapped with WT mice (Fig. 2e). A similarly broad range of acuities was associated with mesopic b-wave amplitudes (Fig. 2f).

Then, we compared photopic and mesopic visual acuity for WT and Rho^P23H/+ mice to the response properties of excitatory neurons in V1 measured with two-photon calcium imaging that were evoked by visual stimuli presented under the same luminance conditions (Fig. 3; Supplementary Fig. S2). We identified more than 3000 visually responsive excitatory neurons in layer 2/3 (L2/3) from WT and Rho^P23H/+ mice and calculated the orientation tuning and SF tuning for each. In WT mice, excitatory neurons in L2/3 spanned the full range of possible preferred orientations, yet most preferred low SFs many octaves below acuity thresholds.^31,32,56 Tuning for orientation was unremarkable, as Rho^P23H/+ mice with impaired acuity exhibited distributions for preferred orientation and orientation selectivity index similar to adult WT mice (Figs. 3a, 3b). In contrast, SF tuning was shifted to lower SFs in Rho^P23H/+ mice (Fig. 3c). In Rho^P23H/+ mice that did not reach criteria for testing on the visual water task, we were unable to identify populations of visually responsive neurons (7 visually responsive neurons from 434 segmented ROIs and 3 mice).

Figure 3.

Cortical tuning to lower spatial frequencies is associated with impaired acuity in Rho^P23H/+ mice. (a) Preferred orientation in degrees, (b) Half-maximum orientation tuning width (degrees), (c) Preferred SF in cycles per degree for excitatory neurons in L2/3 of V1 for WT (gray/black circles) and P23H/+ mice (blue/purple circles) under photopic (P) and mesopic (M) conditions. Circles represent individual neurons (P WT, n = 800; S WT, n = 676; P P23H, n = 823; S P23H, n = 729). Kruskal–Wallis test was used for c. (d–f) Examples of SF response curves presenting the percent of visually responsive neurons for each SF tested for a WT mouse (d) and two Rho^P23H/+ mice (e, f) for both photopic (blue) and mesopic (purple) conditions. Arrows indicate the acuity measured for each mouse for the same luminance conditions as the calcium imaging. (g) SF response curves averaged across subjects for WT mice (n = 5) and P23H/+ mice (n = 6) under photopic and mesopic conditions. Error bars represent SEM. Two-way ANOVA compared WT and P23H/+ mice at each luminance condition. (h) Interpolated percent of visually active neurons for each mouse in panel g responding at the SF corresponding to the measured visual acuity as in panels d to f. Values for WT mice are presented between dashed vertical lines that bound the range of acuities for WT mice at 1 to 3 mo. presented in Figure 1a. Circles represent individual mice at 7 to 9 mo., and triangles represent mice at 10 to 12 mo. for P23H/+ mice. (i) Column plot for the data presented in h (unpaired two-tailed t-test).

At the population level, the preferred SF tuning distributions were similar for photopic and mesopic conditions for Rho^P23H/+ mice, although each Rho^P23H/+ mouse had lower acuity under mesopic conditions (Figs. 1a, 3c). This is likely the consequence of the considerable overlap of the range of acuities under the two luminance conditions for Rho^P23H/+ mice older than 7 mo. of age (Fig. 1a). Therefore, to determine with greater precision how acuity is related to the SF tuning of populations of V1 neurons, we plotted the fraction of visually responsive neurons at each SF tested for individual WT and Rho^P23H/+ mice with their acuity for both photopic and mesopic conditions (Figs. 3d–h). Representative examples of WT and Rho^P23H/+ mice revealed SF response curves shifted toward lower SFs in Rho^P23H/+ mice with lower acuity (Figs. 3d–f). Averaging these SF response curves across mice revealed significant differences between genotypes (Fig. 3g). We also estimated by linear interpolation from these SF response curves for each mouse the percent of neurons responsive at the acuity measured behaviorally (Figs. 3d–f, 3h, 3i). The scatterplot reveals that the estimated fraction of neurons responsive at the acuity threshold resides near 10% of the visually responsive population for both adult naïve WT mice and Rho^P23H/+ mice (Figs. 3h, 3i).

Last, given this relationship between acuity and visual cortical function, we tested the hypothesis that increased plasticity in visual circuitry would sustain acuity in this mouse model of adRP. We generated Ngr1^−/−;Rho^P23H/+ mice and measured their acuity followed by measurements of retinal function with the ERG, the tuning properties of neurons in V1 by calcium imaging, and the thickness of the retinal outer nuclear layer by histology at 7 to 9 mo. and at 10 to 12 mo. of age (Fig. 4). The following results refuted this hypothesis.

Figure 4.

Enhancing visual plasticity accelerates loss of vision in Rho^P23H/+ mice. (a) The visual acuity of Ngr1^−/− and Ngr1^−/−;Rho^P23H/+ mice under photopic (P) and mesopic (M) conditions. Mesopic conditions are indicated by the gray bars. Lines connect the acuity measurements for each mouse under both conditions within each group. P values above the symbols compare Ngr1^−/−;Rho^P23H/+ mice to age- and luminance-matched Ngr1^−/− groups (Brown–Forsythe and Welch ANOVA). P values below the symbols compare Ngr1^−/−;Rho^P23H/+ mice between P and M luminance conditions (mixed-effects analysis). Tests were corrected for all six comparisons between the two tests. Ngr1^−/− ± Rho^P23H/+: 4 to 6 mo., n = 4 and 6; 7 to 9 mo., n = 5 and 12; 10 to 12 mo., n = 5 and 7. (b) Fraction of mice under photopic conditions that were unable to reach threshold criterion for testing. WT and P23H/+ are reproduced from Figure 1a. Ngr1^−/− ± Rho^P23H/+: 7 to 9 mo., n = 0/5 and 8/20; 10 to 12 mo., n = 1/6 and 7/7. (c, d) Scatterplots of visual acuity and b-wave amplitudes for the subset of Ngr1^−/− and Ngr1^−/−;Rho^P23H/+ mice with both measurements for photopic (c) and mesopic (d) conditions. Circles represent mice at 7 to 9 mo., and triangles represent mice at 10 to 12 mo. The vertical line represents mice with b-waves amplitudes less than 8 µV (Ngr1^−/−, 7–9 mo., n = 4; Ngr1^−/−, 10–12 mo., n = 3; Ngr1^−/−;Rho^P23H/+, 7–9 mo., n = 10; Ngr1^−/−;Rho^P23H/+, 10–12 mo., n = 6). (e) Column plot of acuity values for mice with b-wave amplitudes less than 8 µV. WT and P23H are reproduced from Figures 1e and 1f (Kruskal–Wallis test). (f) Nissl-stained histological cross-sections of retina from WT, Rho^P23H/+, and Ngr1^−/−;Rho^P23H/+ mice at 10 to 12 mo. The positions of the ONL, outer plexiform layer (OPL), and inner nuclear layer (INL) are shown. Scale bar: 20 µm. (g, h) Scatterplots of visual acuity and ONL thickness for the subset of Ngr1^−/− and Ngr1^−/−;Rho^P23H/+ mice with both measurements for photopic (g) and mesopic (h) conditions. Circles represent mice at 7 to 9 mo., and triangles represent mice at 10 to 12 mo. (Ngr1^−/−, 7–9 mo., n = 4; Ngr1^−/−, 10–12 mo., n = 2; Ngr1^−/−;Rho^P23H/+, 7–9 mo., n = 6; Ngr1^−/−;Rho^P23H/+, 10–12 mo., n = 4).

The maturation of acuity by Ngr1^−/− mice is normal.³⁰ The acuity of Ngr1^−/− mice at 10 to 12 mo. was similar to that of WT mice (Fig. 4a; Supplementary Figs. S3a, S3b). Ngr1^−/−;Rho^P23H/+ mice displayed a reduction in acuity under mesopic conditions similar to Rho^P23H/+ mice at 4 to 6 mo. of age (Fig. 4a). However, the deficits in acuity were significantly greater in Ngr1^−/−;Rho^P23H/+ mice than in Rho^P23H/+ mice at 7 to 9 mo. and older (Fig. 4a; Supplementary Figs. S4a, S4b). In addition, the fraction of mice unable to reach the criteria for testing on the visual water task doubled for Ngr1^−/−;Rho^P23H/+ mice relative to Rho^P23H/+ mice at 7 to 9 mo. (Fig. 4b). Critically, none of the Ngr1^−/−;Rho^P23H/+ mice examined could reach criterion at 10 to 12 mo. This differed dramatically from the performance of Rho^P23H/+ mice at this age. Thus, enhancing plasticity accelerated the loss of vision in this model of RP.

This loss of vision could not be attributed to a commensurate decline of retinal function in Ngr1^−/−;Rho^P23H/+ mice. The b-wave amplitude was only slightly less than that of age-matched Rho^P23H/+ mice under both photopic and mesopic conditions. Mice with b-wave amplitudes near the noise level spanned the same broad range of acuity between genotypes (Figs. 4c, 4d; Supplementary Figs. S3c, S3d). Plotting the photopic and mesopic acuity for mice with b-wave amplitudes less than 10% of the control mean (∼8 µV) did not support a significant difference between groups (Fig. 4e).

Adult Ngr1^−/− mice display tuning for orientation and spatial frequency indistinguishable from WT mice.⁵⁶ We performed calcium imaging on Ngr1^−/−;Rho^P23H/+ mice that expressed GCaMP6s; however, these functionally blind mice did not yield enough visually responsive neurons for analysis (46 visually responsive neurons from 848 ROIs and 3 mice) (Supplementary Fig. S4).

The number and function of photoreceptors dictate the upper limit of vision in RP. The amplitudes of ERG b-waves reflect the cumulative signaling from photoreceptors to ON bipolar cells, and differences in the number of photoreceptors might be masked in the b-wave amplitude. Therefore, we measured and compared the thickness of the ONL in Ngr1^−/− and Ngr1^−/−;Rho^P23H/+ mice 7 to 12 mo. old (Figs.4f–h). The Ngr1^−/−;Rho^P23H/+ ONL matched the range of thickness reported for Rho^P23H/+ mice at 6 mo. of age and older.^15,19,20,23 Some mice with ONLs that had an average thickness corresponding to a single layer of photoreceptors (or less) possessed photopic and mesopic acuity better than 0.20 cpd, whereas others with equally thin ONLs were functionally blind (Figs. 4g, 4h). Thus, we conclude that an increased loss of photoreceptors is not the primary cause of vision loss in Ngr1^−/−;Rho^P23H/+ mice.

Discussion

In summary, we triangulated the progressive loss of visual acuity with retinal and cortical function in the Rho^P23H/+ mouse model of RP.¹⁵ Deficits in acuity were greater under mesopic conditions across all ages for Rho^P23H/+ mice. The severity of acuity impairment correlated only broadly with deficits in retinal function measured with ffERG but were consistent with the percentage of neurons in primary visual cortex responsive to higher SFs measured with calcium imaging. Contrary to our hypothesis, enhancing neural plasticity was deleterious for visual function in Rho^P23H/+ mice.

The ffERG is an effective tool for identifying retinal dysfunction but lacks the sensitivity to discriminate the severity of visual deficit. The pattern ERG may provide a better estimate of the loss of responsiveness to higher SFs, but the magnitude of the response is only a fraction of the flash ERG (less than 8 µV at 0.1 cpd) and suffers from the same limitations in the threshold for detection, particularly for mice with photoreceptor degeneration.⁵⁷ Cortical visual evoked potentials (VEPs) have also been employed to estimate visual acuity.⁵⁸ However, VEPs represent a complex combination of synchronous aggregate subthreshold and spiking activity.⁵⁹ In rodent models of amblyopia, acuity estimated by VEPs can be more than two octaves higher than acuity measured with the visual water task.⁶⁰ By comparison, we calculated the percent of visually responsive neurons in L2/3 of V1 for Rho^P23H/+ mice with a range of acuity deficits and determined that ∼10% of neurons responded at the SF corresponding to the behavioral acuity limit. This percent of visually responsive neurons may reflect the size of the cortical population required for perception. It would be interesting to ascertain if increasing the excitability of neurons tuned to higher SF might improve acuity, as demonstrated for neural ensembles in head-fixed assays of contrast perception.⁶¹

How might enhancing plasticity in visual circuitry accelerate the loss of acuity in this model of retinal degeneration? Expression of the Ngr1 gene is not detectable in photoreceptors but is present along the visual pathway in RGCs, visual thalamus, and visual cortex.^38,62 RGCs compensate during photoreceptor degeneration to sustain function and can maintain cone-mediated signaling even when almost all rods have been lost.^63,64 Perhaps the loss of Ngr1 expression in RGCs impairs this compensation or attenuates the synaptic drive of the retinogeniculate inputs to visual thalamus. Expression of Ngr1 in thalamus limits plasticity that permits recovery of acuity in a murine model of amblyopia upon the restoration of normal (binocular) vision after a month of monocular deprivation.³⁸ This plasticity may operate in reverse during progressive retinal dysfunction to decimate acuity. Alternatively, cortical plasticity may be the culprit. Recently, we discovered that adult Ngr1^−/− mice retain the high degree of representational drift in visual cortex that is otherwise confined to the critical period.⁵⁶ An increased capacity for adult excitatory cortical neurons to adapt their tuning properties to recent visual experience may permit cortical neurons to adopt non-visual inputs to more rapidly replace degrading feed-forward visual drive as their major source of excitation, thereby accelerating the loss of functional vision. Future work will be required to discriminate between these possibilities.

The visual system is a useful model for investigating the role of plasticity in neurodegeneration. Both behavioral performance and neuronal function can be readily evoked with quantitative visual experience. Retinal circuitry is well characterized, and many of the tools employed for measuring retinal function clinically are also available for rodent models. Neuronal function can be measured independently along several stages of the visual pathway. Vision is not required for viability, so even severe dysfunction can be observed. We conclude that enhancing plasticity is deleterious for vision in the context of retinal degeneration. Based on these findings, we propose that experience-dependent plasticity is predominantly maladaptive for neural circuits during neurodegeneration.