Altered Functional Responses of the Retina in B6 Albino Tyr^c/c Mice

Abstract

Purpose: Mammals with albinism present low visual discrimination ability and different proportions of certain retinal cell subtypes. As the spatial resolution of the retina depends on the visual field sampling by retinal ganglion cells (RGCs) based on the convergence of upstream cell inputs, it could be affected in albinism and thus modify the RGC function.

Methods: We used the Tyr^c/c line, a mouse model of oculocutaneous albinism type 1 (OCA1), carrying a tyrosinase mutation, and previously characterized by a total absence of pigment and severe visual deficits. To assess their retinal function, we recorded the light responses of hundreds of RGCs ex vivo using multi-electrode array (MEA). We estimated the receptive field (RF)-center diameter of Tyr^+/c and Tyr^c/c RGCs using a checkerboard stimulation before simultaneously stimulating the center and surround of RGC RFs with full-field flashes.

Results: Following checkerboard stimulation, the RF-center diameters of RGCs were indistinguishable between Tyr^c/c and Tyr^+/c retinas. Nevertheless, RGCs from Tyr^c/c retinas presented more OFF responses to full-field flashes than RGCs from Tyr^+/c retinas. Unlike Tyr^+/c retinas, very few OFF-center RGCs switched polarity to ON or ON-OFF responses after full-field flashes in Tyr^c/c retinas, suggesting a different surround suppression in these retinas.

Conclusions: The retinal output signal is affected in Tyr^c/c retinas, despite intact RF-center diameters of their RGCs. Adaptive mechanisms during development are probably responsible for this change in RGC responses, related to the absence of ocular pigments.

Albinism is a recessive disease involving 22 mutations identified to date,¹ participating in melanin biosynthesis and melanosome biogenesis. Therefore, albino mammals exhibit variable hypopigmentation of the eyes, skin, and hair. All of them exhibit various functional and structural impairments of the visual system. People with albinism suffer from reduced visual acuity,^2,3 extreme sensitivity to light, and nystagmus (involuntary eye movements). Similarly, albino mice^4,5 present lower visual sensitivity than pigmented mice.⁶ These deficits are most likely the result of various developmental defects affecting the composition and organization of retinal cells in albino mice.^7–11 Accordingly, their number of ipsilateral retinal ganglion cells (RGCs) is reduced,^12–14 as it is in other albino mammals.^15–17 In addition, some studies suggest that opsin expression is affected in cone photoreceptors,^18,19 and the number of rod photoreceptors⁸ or their rhodopsin content seem reduced in albino rodents.²⁰ We speculate that these abnormalities are likely to alter light processing by the retina, thereby affecting RGC responses, the output signal from the retina to the brain. Indeed, the RGC responses to light reflect the convergence and integration processes that occur in the upstream layers of the retina, containing photoreceptors, bipolar cells (BCs), amacrine cells (ACs), and horizontal cells (HCs). The interconnections between these cells govern the responses of RGCs^21–23 to stimuli presented in a given area, called the receptive field (RF).^24,25 The RF comprises (1) an excitatory center, where the cell responds to either increases (ON polarity) or decreases (OFF polarity) of light,²⁶ and (2) an inhibitory surround, with an antagonistic response resulting from lateral inhibition exerted by ACs and HCs.^22,27,28 From a functional point of view, Wässle and colleagues have suggested that the size of the RF of RGCs is an important determinant of their resolving power,^29,30 which derives from their mosaic-like distribution across the retina.³¹ Besides, in primates, the RF diameter of the foveal RGCs is smaller than the RF diameter of peripheral RGCs, notably due to a high density of photoreceptors,³² which confers to this region a greater convergence among photoreceptors, BCs, and RGCs.^33,34 Interestingly, the foveal region does not develop properly in people with albinism (foveal hypoplasia), possibly explaining their poor visual acuity.³⁵ Yet, despite lacking a fovea, albino mice also show limited spatial discrimination abilities.^4,5,36 Although these visual deficits may result from downstream processing, such as in retinal axon targets, we hypothesized that they could be attributable to changes in intra-retinal wiring, possibly affecting RGC RF diameters. 

Here, we recorded the light responses of hundreds of ex vivo RGCs, to study the spatial resolution properties of albino retinas through RGC responses to checkerboard stimulation.³⁷ We did not detect a change in their RF-center diameter, nor in the proportion of ON- or OFF-center responses. However, when we assessed RGC responses to full-field flashes, as in Baden et al. (2016),³⁸ stimulating both the RF center and surround, the Tyr^c/c RGCs showed more OFF responses than the Tyr^+/c RGCs. Alongside, a reduced proportion of OFF-center RGCs switched polarity to an ON or ON-OFF response after full-field stimulation in albino Tyr^c/c retinas. This suggests altered antagonistic surround for RGCs from Tyr^c/c retinas, compared to Tyr^+/c ones. 

Methods

Animals and Tissue Preparation

Mice were housed in a breeding colony maintained on a 12:12 light:dark cycle and fed standardly. All procedures were performed in accordance with the ARVO statement for the Use of Animals in Ophthalmic and Vision Research, and the animal care standards of Sorbonne University, approved by Charles Darwin Ethics Committee, protocol #19873. The initial albino model was the Charles River B6N-Tyr^c-Brd/BrdCrCrl mouse with a single nucleotide mutation in the tyrosinase gene (Tyr^c).³⁹ As the C57BL/6N background contains a mutation in the CRB1 gene (Rd8), causing retinal degeneration,⁴⁰ we eliminated it by backcrossing C57BL/6N with C57BL/6J mice. Littermates contained both Tyr^+/+ and Tyr^+/c and Tyr^c/c albino animals, the latter being phenotypically distinct (white) from their pigmented Tyr^+/+ and Tyr^+/c counterparts. Mice were genotyped as follows. Genomic DNA was extracted with proteinase-K lysis (buffer: 100 mM TRIS-HCl, pH 8,5, 0.2% SDS, 0.2 M NaCl, 0.5% proteinase K), incubated for 1 hour at 55°C, centrifugated for 10 minutes at 13,000 rpm, mixed with isopropanol (1:1), centrifugated for 10 minutes at 13,000 rpm, and resuspended in 40 µL of ddH2O after being incubated at 65°C for 1 hour. PCR was done using a GoTaq DNA polymerase (M7805; Promega, Madison, WI, USA), using primers for Tyr mutation (Forward: CTGTGCCTCCTCTAAGAACTTGT; Reverse: TCCGCAGTTGAAACCCATG) and for Rd8 mutation (Forward: GGTGACCAATCTGTTGACAATCC; Reverse: TCCGCAGTTGAAACCCATGA). After PCR and gel verification of the product size (235 bp for Tyr and 434 bp for Rd8), the sequencing platform of the Institut de la Vision performed Sanger sequencing (ABI 3730 Genetic Analyzer, Applied Biosystems, 48 capillaries using the BigDyeTerm version 1.1 CycleSeq kit; Applied Biosystems). ApE software (M. W. Davis) allowed identification of the point mutation. 

Here, we compared the homozygous tyrosinase mutants Tyr^c/c to their Tyr^+/c counterparts, indistinguishable from the Tyr^+/+ mice for albinism key markers. Retinal tissue was harvested from 6 to 11-week-old male and female mice (Tyr^+/c: n = 6; Tyr^c/c: n = 3). 

Recordings

Mice were euthanized by carbon dioxide inhalation followed by cervical dislocation. The eye was enucleated, and the retinal tissue was superinfused with AMES medium (A1420; Sigma-Aldrich, St. Louis, MO, USA) buffered with 0.19% sodium bicarbonate (S5761; Sigma-Aldrich, St. Louis, MO, USA), and bubbled with 95% oxygen and 5% carbon dioxide at room temperature. Dissection and excision of retinal tissue were performed ex vivo, under dim red illumination. A portion (approximately 1/8) of the ventro-temporal retina, was cut and separated from the retinal pigment epithelium and vitreous. The retina was placed onto a perforated nitrocellulose culture membrane, precoated with 0.1% poly-L-lysine (P8920; Sigma-Aldrich, St. Louis, MO, USA), and gently pressed with a micromanipulator onto a multielectrode array (MEA; 256MEA30/8iR-ITO; Multichannel systems, Reutlingen, Germany) in a customized Nikon Eclipse Ti inverted microscope (Nikon, Dusseldorf, Germany), mounted under the MEA system (Multichannel Systems, Reutlingen, Germany), with the ganglion cell layer facing the electrodes. The MEA recording chamber was filled with AMES medium, buffered, and bubbled as previously described, at a rate of 2.5 to 3.0 mL/min, at 35°C to 37°C. Voltage signals from the microelectrodes were amplified, sampled at 20 kHz and filtered by a second-order Butterworth 300 hertz (Hz)-high-pass filter before being stored (MC_Rack software). 

Visual Stimuli

The retina was dark-adapted for 30 minutes. Light stimuli were displayed using a digital mirror device (DMD; at a resolution of 1024 × 768; ViALUX, Chemnitz, Germany) coupled to an X-Cite exacte ultra-stable fluorescence light source (XCT10A; Lumen Dynamics, Mississauga, Ontario, Canada) on the photoreceptor plane, with standard optics. The light level corresponded to photopic vision: 4.9e4 and 1.4e5 isomerizations / photoreceptor/s for S cones and M cones, respectively (ND30A - Reflective Ø25 mm ND Filter, Optical Density: 3.0; Thorlabs, Newton, NJ, USA). 

Gaussian White Noise Stimulus

We displayed a binary checkerboard of 25 by 25 squares, 28 µm wide, on the recording area to estimate RGC RFs. The luminance of each square varied randomly at a frequency of 30 Hz. This stimulus was applied for 30 minutes. 

Full-Field Chirp Stimulus

In darkness, a repeated full-field chirp light stimulus was applied, similar to the one used by Baden et al. (2016),³⁸ to classify RGCs. After an initial 7500 ms flash, followed by a 7500 ms dim background, 2 sets of full-field flashes were applied. During the first one, the light flashes had fixed contrast and increasing frequency (from 250 ms to 12.5 ms duration). During the second one, they had fixed frequency and increasing contrast. In total, the whole stimulus consisted of 25 repetitions of a 75-second pattern. 

Analysis

Spike Sorting

We used the SpykingCircus software to sort the extracellular spikes from individual RGCs from the recordings of each electrode.⁴¹ RGC responses were analyzed with a customized open-source Matlab (MathWorks, Natick, MA, USA) analysis package (https://github.com/computational-neuroscience-lab/MEA_Clustering). We manually selected for further analyses only the putative cells which exhibited detectable RF and met the refractory period criteria. We obtained 1142 RGCs from Tyr^+/c retinas and 675 RGCs from Tyr^c/c retinas. 

Receptive Field

To estimate the RGC RF-center, we calculated the spike-triggered average (STA) of each cell's response to the white noise stimulus over a 700-ms interval (21 frames preceding the spike), as described in Trapani et al. (2022).⁴² The RF diameter has then been calculated as:  

\begin{equation*}R{{F}_{diameter}} = 2\sqrt {\frac{{are{{a}_{STA}}}}{\pi }} \end{equation*}

 

To compute the temporal component of the RF, we applied a spatial mask to the STA by considering only the squares within the center of the RF, and we computed the average across the space. 

Phasic Index

The temporal STA was then used to compute the phasic index (PhI), as described in Ravi et al., 2018.⁴³ Briefly, the PhI was determined from the temporal RF by computing the absolute value of the sum of positive and negative areas, which was then divided by the sum of their absolute values. The resultant PhI varies between zero and one: zero signifies a biphasic temporal RF, where the areas above and below zero are equivalent; and one indicates a monophasic temporal RF. 

Classification and Mosaic Analysis

To cluster the cells into different subtypes, we used the clustering algorithm described in Trapani et al. (2022).⁴² This clustering uses the mean response to the chirp stimulus, the temporal STA, and the RF diameter. The results presented in this paper are obtained for: size_min = 4, C¹_psth = 0.825, C¹_RF = 0.950, C²_psth = 0.600, and C²_RF = 0.900. Within a cluster, mosaic overlay of the cell’s RF exhibiting similar responses to chirp stimulus, confirmed correct classification. To quantitatively compare mosaics from Tyr^+/c and Tyr^c/c, we used a normalized nearest neighbor distance (NNND). This distance was computed as the ratio of twice the distance between a cell's RF center and its nearest neighboring cell's RF center over the sum of each RF’s radius. Two adjacent cells with touching RFs will have an NNND of 2. Values below 2 indicate overlapping receptive fields, whereas values above 2 indicate distant receptive fields. This provided a measure of the mosaic coverage.  

\begin{equation*}NNN{{D}_{1,2}} = \frac{{2{{d}_{1 \leftrightarrow 2}}}}{{{{r}_1} + {{r}_2}}}\end{equation*}

 

To compare the proportion of RGCs presenting ON, OFF, or ON-OFF responses to full field flashes, we excluded from the analysis RGCs not presenting a clear peri-stimulus time histogram (PSTH; 26 cells from the Tyr^+/c group and 1 cell from the Tyr^c/c group). 

Polarity Index

We calculated an ON-OFF PI using the first full-field flash of the chirp stimulus, presented 25 times. We defined a window of 700 ms after the onset and the offset to define, respectively, the “ON-step” and the “OFF-step.” The PI was calculated as the ratio between the number of spikes occurring in the ON-step window, divided by the total spike count in both windows. The index ranged from 1 (“pure” ON response) to 0 (“pure” OFF response).  

\begin{equation*}Polarity\ Index = \frac{{total\ spike{{s}_{ON}}}}{{total\ spike{{s}_{ON}} + total\ spike{{s}_{OFF}}}}\end{equation*}

 

All cells within 0.4 to 0.6 are considered ON-OFF cells. This represents an ON-step versus OFF-step spike count imbalance of less than 30%. 

Transiency Index

The transiency index was used to compare the duration of RGC responses during the full field flash stimulus. The first time-window we defined, representing the transient component, ranged from 0 to 500 ms after stimulus onset or offset for ON responses, or OFF responses, respectively. The second time-window represented the sustained component and extended from 500 ms to 2000 ms after light onset or offset for ON or OFF responses, respectively. The transiency index is defined as the number of spikes in the sustained window divided by the sum of spikes during both windows. A transiency index close to 1 indicates that the RGC is responding in a sustained manner. An index close to 0 indicates that the RGC is responding transiently.  

\begin{eqnarray*} && Transience\ Index \\ && = \frac{{total\ spike{{s}_{sustained}}}}{{total\ spike{{s}_{transient}} + total\ spike{{s}_{sustained}}}}\end{eqnarray*}

 

Statistics

Data presented in this paper are expressed as mean ± SEM when they follow a Gaussian distribution, or as median (interquartile range) when not. We used a mixed-effects model to compare the RF-center diameter of RGCs between the Tyr^+/c and the Tyr^c/c groups, unpaired two-tailed Mann-Whitney tests to compare medians, unpaired Kruskall-Wallis test followed by Dunn's multiple comparison to compare more than two groups. Statistical significance was defined as P < 0.05. 

Results

RF-Center Diameters are Maintained in Tyr^c/c Retinas, as is the Proportion of ON- and OFF-Center Responses

We tracked the response of hundreds of RGCs simultaneously, to provide an overview of their responses to light, and estimated the diameter of their RF-center, to test whether they could be altered in albinism. An RGC that increases its discharge rate following light increments in its RF-center is of ON polarity, or of OFF polarity if it responds to light decrements. We recorded the ventro-temporal retina, where the proportion of ipsilateral RGCs is reduced in albinism.¹² Light stimuli were presented at the photoreceptor level, whereas the extracellular activity of RGCs was recorded by a MEA (Fig. 1A). We first presented a white noise checkerboard, followed by full-field flashes of varying frequencies and contrasts^38,44 (Fig. 1C). After spike sorting⁴¹ (Fig. 1B), we characterized different features of these RGC responses to both stimuli. The checkerboard was used to estimate the size of the RF-center, by calculating the spatial STA of RGC responses (Fig. 1D). We were able to record RGCs with an RF-center of either ON or OFF polarity in both genotypes, without any major difference in their spatial or temporal STA (Fig. 1E). In addition, we did not find a significant difference in the RF-center diameter between all RGCs from Tyr^c/c retinas or Tyr^+/c retinas (Fig. 1F; Tyr^+/c: 150.9 ± 1.2 µm, n = 1142 RGCs; Tyr^c/c: 152.1 ± 1.2 µm, n = 675 RGCs, Mixed effect model P = 0.7902). In addition, the distribution of the RF-center diameters was particularly comparable between both genotypes (Fig. 1F), with a comparable number of recorded RGCs per retina between all experiments (on average, there were 207 ± 45 RGCs per retina in the Tyr^c/c group and 225 ± 43 in the Tyr^+/c group). Because RGCs from different functional subtypes differ in their RF-center diameter,³⁸ we also independently tested ON-center and OFF-center responses but did not find any significant difference between genotypes (Fig. 1G). 

Figure 1.

 

Comparable diameters of RGC receptive fields (RF), with similar proportions of ON-center and OFF-center responses in Tyr^c/c and Tyr^+/c retinas. (A) MEA set-up. Retinal diagram showing the focus position of the light stimuli onto the photoreceptors (PR) and the electrode position facing the retinal ganglion cell (RGC) layer. The RF of one RGC is schematized by a light blue triangle. Microscope photography showing the array of 256 electrodes, scale bar = 100 µm. (B) Schematic representation of the analysis pipeline. Ex vivo ventro-temporal (VT) retinas from Tyr^+/c and Tyr^c/c adult mice were recorded with an MEA. Spyking Circus algorithm was used to sort the recorded activity and obtain individual RGC activity over time. Analyzed RGC functional properties included RF center diameter, response polarity, transiency, and phasic indices (PhI). (C) Schematic of the displayed stimulation sequence, first a random checkerboard sequence at 30 Hz, then a full-field chirp consisting of a series of flashes of varying frequency and contrast. (D) Schematic of the spatial Spike Triggered Average (STA) calculation. The STA corresponds to the average stimulus preceding a cell's spike, in a defined time window, where the RF-center stands out from the background. (E) Temporal (left) and spatial (right) RF components of two representative RGCs, from the Tyr^+/c and Tyr^c/c groups. The temporal STA shows the standardized polarity (over SD) of the preferred stimulus, light increments for the upper two cells, and light decrements for the bottom two cells. Spatial STA shows the RF-center and surround, white ellipses highlight the RF center, scale bar = 100 µm. (F) RF center diameter of all RGCs, from Tyr^+/c (n = 1142 cells) and Tyr^c/c (n = 675 cells) retinas, showing no significant difference between both groups (P > 0.05). (G) RF-center diameters according to their polarity, showing no significant differences between both groups. ON RGCs: Tyr^+/c n = 567, median RF center of 140.3 µm, interquartile range = 121.3 to 163.9 µm; Tyr^c/c n = 311, RF of 140.8, interquartile range = 124.9 to 168.6 µm, P = 0.2206. OFF RGCs: Tyr^+/c n = 575, RF of 162.1, interquartile range = 139.3 to 183.5 µm; Tyr^c/c n = 364, RF of 157.5, interquartile range = 136.9 to 178.2 µm, P = 0.0529. For F and G, the dark lines indicate the median and interquartile range, whereas the cross indicates the mean (ns = nonsignificant, P > 0.05, unpaired two-tailed Mann-Whitney test).

No Difference in Mosaic Analysis of Subtypes of RGCs

Once we had compared the central RF diameters of the recorded RGCs, we compared their responses to full-field light flashes, strongly stimulating both their RF-center and surround (see Fig. 1C). To this end, we analyzed PSTH, representing the mean firing rate of RGCs in time windows of 0.033 seconds that span the entire stimulus duration (Figs. 3A, 3B, 3C). Using responses to checkerboard and full-field stimuli, we performed a clustering analysis to group RGCs into subtypes and determined whether changes between genotypes could be identified (Fig. 2). We identified subtypes with similar polarity responses to checkerboard and full-field stimulus: ON-center ON full-field RGCs (see Figs. 2A, 2B) or OFF-center OFF full-field RGCs (see Figs. 2C, 2D). RGCs presenting an ON or OFF response to a stimulus may also respond to a stimulus of opposite polarity if it sufficiently stimulates its suppressive surround. Indeed, other subtypes instead presented a polarity switch between the checkerboard and full-field stimulus: OFF-center and ON or ON-OFF full-field RGCs (see Figs. 2E, 2F) and ON-center OFF or ON-OFF full-field RGCs (see Figs. 2G, 2H). We did not find any major difference across genotypes in the RGC subtypes identified. For the most part, comparable subtypes with similar STA and PSTH were identified between Tyr^c/c and Tyr^+/c retinas. Furthermore, we did not detect major difference in mosaic arrangement of their RF in mosaics, measured by the spacing of their RF using an NNND (see Fig. 2). Nevertheless, in albino Tyr^c/c retinas, we could not find an RGC subtype with an OFF-center and pure ON full-field response as found in the Tyr^+/c retinas (see Fig. 2E). Instead, the responses rather resembled an RGC subtype with OFF-center and ON-OFF full-field response (see Fig. 2F). However, this could be due to an insufficient number of representative cells of the OFF-center and pure ON full-field response to identify a subtype in albino retinas, especially because the number of experiments with Tyr^c/c is lower. 

Figure 2.

 

RGC subtypes with their STA, PSTH, mosaic distribution and spacing in Tyr^c/c, and Tyr^+/c retinas. (A–H) Examples of different RGC functional subtypes are shown with the normalized temporal spike triggered average (STA, top left), the normalized peri-stimulus time histograms (PSTH; top right, averaged traces in color, all traces in gray), the receptive field (RF) mosaic by experiment (bottom left) and the normalized nearest neighbor distance (NNND; bottom right). Each subtype is defined by its response to checkerboard stimulation (ON-center or OFF-center) and to full-field stimulation (ON, ON-OFF, or OFF). This figure only represents an example of the few RGC subtypes identified. Most RGC subtypes are comparable between Tyr^c/c and Tyr^+/c retinas. However, we could not find an equivalent to the OFF-center ON full-field RGC subtype (E) in Tyr^+/c retinas and found only an OFF-center ON-OFF full-field RGC subtype in Tyr^c/c retinas (F). Pigmented Tyr^+/c RGCs are in blue-green and Tyr^c/c RGCs in magenta-orange. Scale bar = 100 µm. (A, B) ON-center - ON full-field RGCs. (C, D) OFF-center - OFF full-field RGCs. (E) OFF-center ON full-field RGCs. (F) OFF-center - ON-OFF full-field RGCs. (G, H) ON-center - ON-OFF full-field RGCs.

RGC Responses to Full-Field Chirp Stimulus are Shifted Toward Their OFF Component in Tyr^c/c Compared to Tyr^+/c Retinas

Because comparing RGC subtypes between genotypes is not straightforward, we analyzed the PSTH responses of the general population of RGCs across experiments between genotypes. The mean firing rate following the extinction of a full-field flash of light (OFF polarity), was significantly increased for Tyr^c/c compared to Tyr^+/c RGCs (P < 0.0001; see Fig. 3A). This enhanced firing rate was consistently found following flashes of increasing frequency (see Fig. 3B) and was more pronounced for stimuli of higher contrast (see Fig. 3C). We further established a PI that reflects the polarity “bias” of each cell (see Fig. 3D). An index close to one, indicated an RGC that discharged predominantly following light onset (ON responses), whereas an index close to zero corresponded to an RGC that predominantly responded to light offset (OFF responses). We identified a significant shift of the PI toward OFF responses for Tyr^c/c RGCs compared to the Tyr^+/c RGCs (P < 0.0001; see Fig. 3D). 

Figure 3.

 

RGC responses to full-field chirp stimulus are shifted toward their OFF component in Tyr^c/c compared to Tyr^+/c retinas. (A, B, C). Mean peri-stimulus time histogram (PSTH) of all Tyr^c/c RGCs (magenta, n = 675 cells) and all Tyr^+/c RGCs (blue, n = 1142 cells) responding to a full field chirp stimulus. The PSTH represents the mean firing rate collected through 0.033 seconds time bin. The dark sinusoid shows the temporal profile of the chirp stimulus. (A) Mean PSTH for the first long-lasting flash of light. Tyr^+/c OFF response: median = 16.8, interquartile range = 3.6 to 47.4 Hz; Tyr^c/c OFF response: 50.4, interquartile range = 20.4 to 90.0 Hz, P < 0.0001. (B) Mean PSTH for the set of flashes of increased frequency. (C) Mean PSTH for the set of flashes of increased contrast. (D) Polarity indices for all RGCs from each group, estimated from their responses to the first full-field flash. An index close to 1 indicates an ON cell, close to 0.5 an ON-OFF cell, and close to 0 an OFF cell. Tyr^c/c RGCs are significantly shifted toward OFF responses (P < 0.0001). (E) Transiency indices for all RGCs from each group. An index close to 1 indicates a sustained response whereas an index close to 0 indicates a transient response. Tyr^c/c and Tyr^+/c RGCs are not statistically different in terms of transiency index. (F) Proportion of RGCs by PhI. A reduced proportion of RGCs is biphasic in albino Tyr^c/c retina, compared to Tyr^+/c retinas. For D and E, the dark lines indicate the median and interquartile range, whereas the cross indicates the mean, ns: P > 0.05; ****: P < 0.0001.

Besides, RGCs exhibit two types of temporal pattern of response: either a burst of action potentials triggered just after the change in light polarity (transient), or phasic action potentials throughout the stimulation (sustained). Therefore, we calculated an index of RGC response transiency (see Fig. 3E). An index close to one corresponds to a sustained response mode, whereas an index close to zero indicates a cell with a transient response mode. Here, we did not find any significant difference in the transiency index between Tyr^+/c and Tyr^c/c RGCs, with similar distribution of response patterns in both groups (P = 0.2705; see Fig. 3E). 

We also measured the PhI based on the temporal STA determined using the checkerboard stimulus (see Fig. 3F). An RGC with a biphasic temporal RF indicates a response characterized by a light response comprising two distinct phases, where the cell exhibits both positive and negative phases of activity following stimulation. Conversely, a monophasic temporal RF denotes a single-phase response, reflecting a uniform reaction of the RGC to light, devoid PhI of distinct positive-negative transitions. The proportion of cells exhibiting a biphasic response (phasic response <0.3) is reduced in albino Tyr^c/c (21.8%, n = 147 cells) compared to pigmented Tyr^+/c retinas (30.9%, n = 354 cells; see Fig. 3F), consistent with the change in PI that we observed. 

OFF-Center RGCs of Tyr^c/c Retinas Exhibited Less ON or ON-OFF Responses After Full-Field Stimuli Compared With Those of Tyr^+/c Retinas

We did not find any major difference in the proportion of ON or OFF-center responses after checkerboard stimulation between albino Tyr^c/c and pigmented Tyr^+/c retinas (see Figs. 4A, 4E). However, the RF properties of RGCs are dynamic and depend on the type of stimulus.^25,45–47 Therefore, we compared the RF-center polarity of the RGCs following the checkerboard stimulus with their polarity response to full-field flashes. In Tyr^+/c and Tyr^c/c mice, we find that individual RGCs presenting ON-center RF (see Fig. 4A) to checkerboard stimulus showed either ON responses (see Fig. 4B), ON-OFF (see Fig. 4C), or OFF responses (see Fig. 4D) when stimulated with full-field flashes. These three types of responses are also observed for OFF-center RGCs after full-field flashes (see Figs. 4F, 4G, 4H) in Tyr^+/c retinas (see Fig. 4E). However, in Tyr^c/c retinas, most RGCs with OFF center responses to checkerboard stimulation presented OFF responses to full-field flashes and much fewer presented ON or ON-OFF responses compared to Tyr^+/c RGCs (see Fig. 4E). By contrast, the ON-center RGCs following checkerboard stimulation mostly displayed ON responses to full-field flashes, in both Tyr^+/c and Tyr^c/c retinas with comparable proportions of cells with ON, ON-OFF, and OFF responses (see Fig. 4A). Thus, the main change observed in the albino retina is the decrease in the proportion of OFF-center RGCs with ON and ON-OFF responses to full-field flashes (see Fig. 4E). This decrease probably explains why overall, there is a reduction in the proportion of ON responding RGCs shown with the polarity ratio (see Fig. 3D) and in the proportion of biphasic responding RGCs (see Fig. 3F). This is also in line with the decrease in the mean firing rate of ON responses and increase in the mean firing rate of OFF responses in albino Tyr^c/c retinas (see Figs. 3A, 3B, 3C). 

Figure 4.

 

OFF-center RGCs of Tyr^c/c retinas exhibited fewer ON responses after full-field stimuli compared with those of Tyr^+/c retinas. (A) Distribution of RGC polarity responses (ON, ON-OFF, or OFF) to a full-field set of flashes, for RGCs showing an ON-center RF following a checkerboard stimulation, in Tyr^+/c and Tyr^c/c retinas. (B, C, D) Representative PSTH for RGCs with ON-center RF showing ON, ON-OFF, or OFF responses to full-field flashes, from Tyr^+/c (first row) and Tyr^c/c (second row) group. (E) Distribution of RGC polarity responses (ON, ON-OFF, or OFF) to full-field flashes, for RGCs presenting an OFF-center RF following a checkerboard stimulation, in Tyr^+/c or Tyr^c/c retinas. The proportion of RGCs with OFF-center RF that switch their polarity after full-field stimulation to ON or ON-OFF is reduced in Tyr^c/c retinas compared to Tyr^+/c retinas. (F, G, H) Representative PSTHs for RGCs with OFF-center RF presenting ON, OFF, or ON-OFF responses to full-field flashes, from Tyr^+/c (first row) and Tyr^c/c (second row) group. The temporal profile of the stimulus is indicated by the black line above. Error bars in A and E represent the standard deviation of the cumulative proportions between categories as a measure of variability between experiments.

Discussion

Here, we compare for the first time the light responses of RGCs from pigmented Tyr^+/c and albino Tyr^c/c mice. As RGCs are distributed in mosaics across the retina according to their subtypes,^48,49 we obtained a representative sampling of most RGC populations. Our RF measurements using a checkerboard stimulus (Gaussian white noise) are consistent with those reported in the literature (e.g. 154 µm, on average⁵⁰), but we did not detect a difference between the RF-center diameter of RGCs from Tyr^+/c or Tyr^c/c retinas. Therefore, the loss of spatial discrimination abilities in albino mice^4,5 is not due to a lack of the spatial resolution of RGC responses related to their RF diameter. The mosaic arrangement of RGC RFs can also contribute to determine the spatial resolution of the retina,^51,52 but we did not observe any striking difference in the organization of the mosaics of specific subtypes nor in their spacing (see Fig. 2). Further studies including larger number of RGCs are required, particularly to determine whether the RF surround properties²⁵ or its nonlinear aspects⁵³ are affected in albinism. 

Besides, albinism could affect other functional properties of RGCs, as we revealed through RGC responses to full-field stimuli, stimulating both the excitatory center and suppressive surround of the RF. RGC response to their RF-center is primarily governed by their excitatory inputs from bipolar cell axons.⁵⁴ However, their antagonistic RF-surround arises from the lateral inhibition of HCs onto BCs,⁵⁵ and from the lateral inhibition of ACs onto BCs and RGCs.^26,27,56 Interestingly, we recorded a higher proportion of OFF responses in Tyr^c/c retinas when stimulating both RF-center and surround, as well as a higher average discharge rate after full-field flash extinction. Furthermore, in Tyr^c/c retinas, very few of the OFF-center RGCs recorded (checkerboard stimulation) displayed an ON response after full-field flashes, unlike Tyr^+/c retinas. Accordingly, 95% of RGCs from Tyr^c/c retinas showed both an OFF-center RF following checkerboard stimulation and an OFF response following full-field flashes instead of 58% in Tyr^+/c retina. Interestingly, in wild-type in vivo mice, approximately 40% of the RGCs with OFF-center RF switched their response to ON or ON-OFF upon full field stimulation,⁵⁷ which is equivalent to the 41.6% we observed in pigmented control retinas (see Fig. 3E). Furthermore, we found similar proportions of ON-center or OFF-center RGC responses after checkerboard stimulation, and revealed a difference only upon sufficient surround stimulation, specifically for OFF center RGCs. This suggests an imbalance between the networks driving the RF surround or center responses or an inability of the albino Tyr^c/c retina to adapt to higher luminance. Perhaps some RGC subtypes are missing from albino retinas, although not evident when we observed RGCs clustered according to their functional response into subtypes. Yet, we know that the proportion of ipsilateral RGCs is reduced^13,14 following neurogenesis defects in albino mice,^14,58 although its consequences on retinal function are unknown. In addition, changes in photoreceptor proportion,⁸ rhodopsin distribution,²⁰ or S- and M-opsins have also been suggested in non-isogenic albino rodent retinas.^18,19,59 Thus, we cannot exclude that change in certain features of retinal cells upstream of RGCs might also be affected, potentially leading to impaired intraretinal connectivity. However, our experiments did not allow us to investigate these, as we only recorded responses from RGCs, the retinal output signal resulting from light processing by all upstream cells (PRs, BCs, HCs, and ACs). How could these altered responses in albino retina occur? 

First, adaptative mechanisms could affect intraretinal circuit function in albinism. Recently, a study revealed that dopamine, known to play a pivotal role in the antagonistic surround response, affects different subtypes of RGCs: the blockade of D1-R led to surround activation revealing opposite polarity responses (ON responses) to large stimuli in OFF-center RGCs.⁶⁰ Thus, alteration of dopamine signaling could maybe explain the loss of the response switch in OFF-center RGCs observed in the albino retina. Indeed, dopaminergic ACs play a central role in presynaptic inhibition to adjust gain in local circuits,⁶¹ through volumetric release.⁶² It represents an interesting candidate for further pharmacological studies of intraretinal circuit function in albinism. 

Second, homeostatic plasticity, the ability of a system to adjust to changes to maintain its internal stability,⁶³ could also be involved. Indeed, it could occur after changes in the proportion of retinal cell types in the albino retina, perhaps driving the aforementioned changes in RGC responses. Alternatively, it could occur from development onward as an adaptation of the retinal circuit to attenuate the excessive light exposure associated with the absence of pigment in the albino eye. Development is closely linked to neuronal activity,⁶⁴ and light adaptation⁶⁵ is a prerequisite for RGCs to faithfully encode light contrast through luminance changes.⁶⁶ Mechanistically, homeostatic plasticity may operate at the level of intrinsic excitability, regulation of synapse number, of the excitation/inhibition balance at existing synapses,⁶⁷ and refinement of neurite arborization.⁶⁸ Moreover, spontaneous or evoked activity during development has been linked to differential refinement of ON and OFF responses.^69–72 

Our study has demonstrated for the first time a difference in the polarity response to full-field stimuli in albino RGCs compared to pigmented ones. It will pave the way for future studies of retinal activity during development or in adulthood in albino mice, while manipulating neuronal excitability. This will shed light on the retina’s ability to adapt to eye hypopigmentation and, ultimately, understand the basis of visual disorders in these mice.^4,5 

Acknowledgments

The authors thank Alain Chédotal and all the team members for helpful discussions and inputs. We also thank the members of the animal facility at the Institut de la Vision. 

Supported by Genespoir, Retina France to A.R., by a doctoral fellowship from the ED3C doctoral program of Sorbonne Université to V.C., by ERC Consolidator grant DEEPRETINA (101045253), ANR grants (ANR-18-CE37-0011 – DECORE, ANR-20-CE37-0018-04- Shooting Star, Chaire Industrielle MyopiaMaster, project NUTRIACT, project PerBaCo, project RetNet4EC), a grant from AVIESAN-UNADEV, and one from Retina France to OM. Furthermore, INSERM, Sorbonne Université, LabEx LIFESENSES (ANR-10-LABX-65) and IHU FOReSIGHT (ANR-18-IAHU-01) supported our research at the Institut de la Vision. 

Disclosure: V. Chotard, None; F. Trapani, None; G. Glaziou, None; B.S. Sermet, None; P. Yger, None; O. Marre, None; A. Rebsam, None