A metabotropic glutamate receptor agonist enhances visual signal fidelity in a mouse model of retinitis pigmentosa

Many inherited retinal diseases target photoreceptors, which transduce light into a neural signal that is processed by the downstream visual system. As photoreceptors degenerate, physiological and morphological changes to retinal synapses and circuitry reduce sensitivity and increase noise, degrading visual signal fidelity. Here, we pharmacologically targeted the first synapse in the retina in an effort to reduce circuit noise without sacrificing visual sensitivity. We tested a strategy to partially replace the neurotransmitter lost when photoreceptors die with an agonist of receptors that ON bipolars cells use to detect glutamate released from photoreceptors. In rd10 mice, which express a photoreceptor mutation that causes retinitis pigmentosa (RP), we found that a low dose of the mGluR6 agonist l-2-amino-4-phosphonobutyric acid (L-AP4) reduced pathological noise induced by photoreceptor degeneration. After making in vivo electroretinogram recordings in rd10 mice to characterize the developmental time course of visual signal degeneration, we examined effects of L-AP4 on sensitivity and circuit noise by recording in vitro light-evoked responses from individual retinal ganglion cells (RGCs). L-AP4 decreased circuit noise evident in RGC recordings without significantly reducing response amplitudes, an effect that persisted over the entire time course of rod photoreceptor degeneration. Subsequent in vitro recordings from rod bipolar cells (RBCs) showed that RBCs are more depolarized in rd10 retinas, likely contributing to downstream circuit noise and reduced synaptic gain, both of which appear to be ameliorated by hyperpolarizing RBCs with L-AP4. These beneficial effects may reduce pathological circuit remodeling and preserve the efficacy of therapies designed to restore vision.

In murine models of RP, light-independent circuit noise can be reduced by blocking gap junctions and/or ionotropic glutamate receptors in inner retinal circuitry (Biswas et al., 2014;Busskamp et al., 2010;Cross et al., 2022;Simon et al., 2020;Toychiev et al., 2013;Tun et al., and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 5 5 2023).Since degeneration originates in rods, we hypothesized that targeting type-6 metabotropic glutamate receptors (mGluR6) at rod-RBC synapses might improve signal fidelity in degenerating rd10 retinas.We first characterized the developmental time course over which rod degeneration in rd10 retinas degrades visual signaling.We recorded in vivo electroretinograms (ERGs) to measure photoreceptor and bipolar cell activity, then measured light-evoked excitatory postsynaptic currents (EPSCs) from ON alpha ganglion cells (ONα RGCs) in an in vitro whole-mount retina preparation to evaluate visual signals and noise within the circuit.Rod degeneration reduced light responses and increased circuit noise, effects that were partially ameliorated by mimicking tonic glutamate input to RBCs with the mGluR6 agonist L-2-amino-4-phosphonobutyric acid (L-AP4).A low concentration of L-AP4 reduced noise in ONα RGCs without altering EPSC amplitude.Perforated-patch recordings indicated that RBCs rest at more depolarized potentials in rd10 retinas, likely reducing the gain of RBC synaptic output to downstream targets (Oesch & Diamond, 2011).Our results indicate that L-AP4 reduces circuit noise and preserves synaptic gain by hyperpolarizing rd10 RBCs.
They suggest a broadly applicable therapy that could ameliorate deleterious downstream effects of photoreceptor degeneration, maintaining visual signaling and potentially preserving the efficacy of subsequent restorative therapies.
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 6 6

Animals
In all experiments, animals were treated in accordance with National Institutes of Health guidelines, as approved by the National Institute of Neurological Disorders and Stroke Animal Care and Use Committee (ASP-1344).Wild-type (C57Bl6) and rd10 mice (B6.CXB1-Pde6b rd10 /J, Jackson Labs) of either sex were used and were housed on a 12:12 light-dark cycle and provided free access to food and water.

Electroretinograms
In vivo ERG responses were recorded following published protocols from the NEI Vision Core (Li et al., 2020) with an Espion E2 Visual Electrophysiology System (Diagnosys, Lowell, MA, USA).Both WT and rd10 animals were dark adapted overnight (~10 hours) and anesthetized by i.p. injection of ketamine (100 mg/kg)/xylazine (6 mg/kg) mixture under dim red light.Pupils were dilated with 1% tropicamide and irritation was reduced with 0.5% phenylephrine.Animals were placed on a heating plate maintained at 37°C.Responses were recorded with a gold loop wire electrode placed at the center of the cornea, a reference electrode in the mouth, and a ground electrode in the tail.ERGs were recorded after a further 3 minutes dark adaptation using 20 ms flashes (10 -5 , 10 -4 , 10 -3 , 10 -2 , 1, and 10 cd•s/m 2 ).

Retina dissection for in vitro experiments
Animals were dark-adapted overnight, deeply anaesthetized with isoflurane (Baxter), and euthanized via cervical dislocation.After bilateral enucleation, the eyes were submerged in and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 7 7 bicarbonate-buffered Ames medium (∼32°C, Sigma Millipore, 285-295 mOsm) continuously equilibrated with carbogen (95% O 2 /5% CO 2 ).The cornea, lens, and iris were removed using small surgical scissors and forceps under a dissecting microscope (Zeiss) and infrared LED illumination (940 nm, ThorLabs), visualized through the microscope eyepieces with infrared image converters (BE Meyers).Retinas were either used immediately or left in their eyecups and stored for up to 6 hours at room temperature in light-proof chambers in carbogen-equilibrated Ames solution.

Whole-mount retina recordings
To mount the intact retina, the vitreous was removed from the eyecups, and the retina was carefully separated from the pigment epithelium.The retina was mounted photoreceptor-side down on a poly-L-lysine-coated microscope slide (12mm diameter, Corning BioCoat Cellware) that was secured to the bottom of recording chamber with vacuum grease (Dow Corning).The chamber and tissue were superfused continuously in carbogen-equilibrated Ames solution (7-9 mL/min, ~32°C).was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint excitatory synaptic input, ON RGCs were held at the estimated reversal potential for excitatory input of −60 mV.

Retinal slice recordings (paired and perforated patch recordings)
For both whole-cell recordings from RBCs and AII amacrine cells and RBC perforated patch recordings, dissections were performed in room-temperature bicarbonate-based AMES media (285-295 mOsm) equilibrated with carbogen.A section of isolated retinal tissue was embedded in low gelling temperature agarose (3% in HEPES-based AMES media) and then submerged in ice-cold HEPES-based AMES media (285-295 mOsm, NaOH-adjusted to pH 7.4).Transverse slices (200 μm thick) were cut on a vibratome (Leica VT1000S), stored in a light-proof carbogen-equilibrated container, and used for 4-6 hours after slicing.Slices were placed in a recording chamber under a harp (ALA Scientific) and continuously superfused with carbogen-equilibrated Ames medium as above.
For perforated patch recordings, RBCs from P20-25 WT and rd10 animals, dark-adapted overnight, were identified by their distinct somatic shape and light response characteristics, and confirmed via confocal imaging after breaking into whole cell at the conclusion of perforated patch recordings.Current clamp recordings were conducted using the same glass electrodes as above, filled with internal solution containing (in mM): 125 K-aspartate, 10 KCl, 10 HEPES, 5 N-methyl glucamine-HEDTA, 0.5 CaCl2, 1 ATP-Mg, 0.2 GTP-Mg, and 0.1 Alexa 488/Alexa 594 hydrazide at 265-270 mOsm, adjusted to pH of 7.3 with N-methyl-D-glucamine (NMG)-OH.Beta-Escin (25 μM) was used as a perforating agent.Membrane currents were filtered at 300 Hz and sampled at 10 kHz.

Visual stimulation and analysis
Full field light stimuli (500 m diameter spot) were presented using a customized 912 × 1140pixel digital projector (DLPLCR4500; Texas Instruments) (Franke et al., 2019) driven by a 405 nm LED (ThorLabs) at a frame rate of 60 Hz.Spatial stimuli patterns were created with MATLABbased software (https://github.com/Schwartz-AlaLaurila-Labs/sa-labs-extension).Photon flux was attenuated to desired levels using a motorized neutral density filter wheel (FW102C, Thorlabs) and routed through the microscope (Scientifica Hyperscope) condenser, which was adjusted so that and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 10 10 images were in focus at the plane of the photoreceptor outer segments.Photoisomerization rates were calculated based on a collecting area of 0.85 μm 2 for rods (Govardovskii et al., 2000;Lyubarsky et al., 2004).Stimuli were centered relative to the recorded cell and focused on the photoreceptor layer.Irradiance (W/m 2 ) was converted to photoisomerization rate (R*/rod/s) using the estimated collecting area of rods and cones (0.5 and 0.37 μm 2 , respectively), the 405-nm LED (ThorLabs) emission spectrum, and the photoreceptor absorption spectra (Govardovskii et al., 2000).
For voltage clamp experiments, EPSC charge was calculated by integrating the averaged current responses (average of 5 trials) over the stimulus time (1 second) window.Noise (variance) was calculated over the 1 second interval prior to light stimulus; all data for variance and charge calculations were baseline corrected.
Electrophysiological data were analyzed in MATLAB using a custom written open-source package (http://www.github.com/SchwartzNU/SymphonyAnalysis).Figures were constructed in

Estimating the RRP size
During paired RBC-AII recordings, a step depolarization in the RBC elicited an EPSC in the AII comprising transient and sustained components (Singer & Diamond, 2003).When evoked by large presynaptic voltage steps, the transient component reflects the release of the entire readily-releasable pool (RRP) (Oesch & Diamond, 2011;Singer & Diamond, 2003).AII EPSCs were integrated and a linear fit to the sustained component (50-100 ms after step onset) was extrapolated back to the time and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 11 11 that the step was initiated (Neher, 2015;Singer & Diamond, 2006), thereby yielding an accurate measure of the RRP size (Singer and Diamond, 2006).

Experimental design and statistical analysis
Unless indicated otherwise, normally distributed data are reported as mean ± SD with p-values calculated using 2x3 between subjects ANOVA followed by unpaired two-tailed t-tests for comparisons between WT and rd10, and mixed ANOVA followed by paired t-test for comparisons between control and L-AP4 treated retinas.Contrast response curves graphed in figures 5E and 6Giiii are shown as median ± upper and lower quartiles, as those data were not normally distributed (Jarque-Bera test).These p-values were determined using permutation and the Wilcoxon Rank Sum test.Resting membrane potentials shown in figure 8C are reported as mean ± SE with p-values calculated using 2x3 between subjects ANOVA followed by unpaired two-tailed t-tests for comparisons between WT and rd10, and paired t-test for comparisons between control and L-AP4 treated RBCs.
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 12 12

In vivo ERG measurements reveal the time course of retinal dysfunction in rd10 mice
We first characterized the timeline of rod signal degeneration in dark-adapted rd10 mice (P19-45) using in vivo electroretinograms (ERGs).Rd10 mice typically do not exhibit detectable scotopic (i.e., rod-driven) ERGs after ages P30-35 (Gargini et al., 2007;Jae et al., 2013), although some light sensitivity may remain at later stages of RP (Rodgers et al., 2023;Scalabrino et al., 2023).The awave component of the ERG waveform reflects photoreceptor activation, whereas ON bipolar cell responses are contained within the b-wave (Li et al., 2020;Perlman, 1995;Pinto et al., 2007).Both components were smaller in rd10 mice than in WT even prior to significant rod degeneration (fig. 2A,C,E), likely reflecting the effects of the PDE6 mutation on rod signaling (Bowes et al., 1990;Chang et al., 2007;Gargini et al., 2007;Kuehlewein et al., 2021).At P19, a-waves were not detectable in response to flashes below 0.01 cd•s/m 2 in WT, and below 0.1 cd•s/m 2 in rd10 (fig.2C).
This likely reflects the sensitivity limits of the in vivo ERG to detect a-waves, as the larger b-waves were recorded in response to stimuli as dim as 10 -4 cd•s/m 2 (fig.2E) in both WT and rd10 retinas.In response to brighter flashes, both a-and b-waves in rd10 retina progressively diminished such that no a-waves were detected by P45 (fig.2B,D,F), when rods have degenerated completely (Gargini et al., 2007;Puthussery et al., 2009).
To examine the age-dependence of signal loss more closely, we delivered a range of flash strengths to P19-45 WT and rd10 animals (fig.2G-J).In response to flashes activating only rods (0.1 cd•s/m 2 ), or both rods and cones (10 cd•s/m 2 ), WT a-wave amplitudes were larger than those in rd10 across all measured ages (n = 8, p = 0.0001; fig.2G,H).Rd10 a-waves could not be resolved at any age in response to dimmer flashes, whereas WT a-wave amplitudes increased with age.In responses and is also made available for use under a CC0 license.

Rod degeneration degrades scotopic signal fidelity in ONα RGCs
To measure physiological circuit output in dark-adapted WT and rd10 retinas, we recorded from RGCs in the in vitro whole-mount retina across the same age range as the in vivo experiments.
Whereas ERG recordings yield population light responses, whole-cell voltage-clamp recordings of excitatory postsynaptic currents (EPSCs) in RGCs report downstream signals and noise in individual neurons.We focused on sustained ONα RGCs, which have been well characterized physiologically and anatomically (Bleckert et al., 2013;Goetz et al., 2022;Krieger et al., 2017;Laboissonniere et al., 2019).We grouped recorded RGCs into three age ranges: P16-20, the beginning stages of rod degeneration in rd10; P21-25, over which time more than 50% of rd10 rods die; and P26-30, when rod degeneration is mostly complete (Barhoum et al., 2008;Jae et al., 2013;Kim et al., 2018;Li et al., 2018;Puthussery et al., 2009).We first recorded synaptic dark noise and light responses to 1 s light steps (40R*/rod/s; fig.3A).Cells were filled through the patch pipette with Alexa 488 to identify any morphological differences between WT and rd10 RGCs, or between age groups; no qualitative differences were observed across ages or between WT and rd10 (fig.3B).
WT ONα RGCs exhibited synaptic dark noise that decreased with age (fig.3C,D).By contrast, dark noise was relatively low in P16-20 rd10 RGCs and increased as rod degeneration progressed and is also made available for use under a CC0 license.

Low dose of L-AP4 suppresses circuit noise under scotopic conditions
The results thus far indicate that rd10 rod degeneration decreases visual responses and increases circuit noise, degrading visual signal fidelity.We hypothesized that a low dose of mGluR6 agonist L-AP4 (binding affinity ~2 μM) (Naples & Hampson, 2001;Thomsen, 1997) might stabilize RBC activity without completely blocking responses to rod input, thereby enhancing visual signaling in rd10 animals.Consistent with this prediction, 50 nM L-AP4 lowered dark noise in both WT and rd10 ONα RGCs across all age groups (fig.4A-C).The effect of 50 nM L-AP4 on ONα RGC EPSC amplitudes was highly variable between cells in both WT and rd10, leading to an insignificant effect overall (fig.4D-F).Taken together, these results suggest that a low dose of L-AP4 improves signalto-noise characteristics in the degenerating retina.

Rod degeneration degrades mesopic contrast signals in ONα RGCs
In the experiments presented thus far, dim light stimuli were delivered from darkness to evoke primarily rod-mediated responses.Even at night, however, animals navigating their visual world must distinguish the relative luminance of objects from their immediate surroundings, conditions that are best described in terms of contrast.To examine the effects of rod degeneration on contrast encoding, we delivered stimuli upon a 500 R*/rod/s full-field background, mesopic conditions under and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 15 15 which contrast responses in RGCs are mediated by both the rod and cone pathways (Nath et al., 2023).Similar to our results under scotopic conditions (fig.3C,D), WT ONα RGCs exhibited background noise at P16-20 that decreased with age (fig.5A,B).Rd10 ONα RGCs exhibited the opposite trend, with low noise levels early in development that increased significantly by P26-30 (p = 0.046; fig.5B), likely in response to severe rod loss (Barhoum et al., 2008;Pennesi et al., 2012;Puthussery et al., 2009).

A low dose of L-AP4 decreases mesopic synaptic noise in ONα RGCs
The mGluR6 agonist, L-AP4, acts on ON cone bipolar cells as well as RBCs and may therefore affect noise and light responses in the cone pathway under mesopic conditions.We found that 50 nM L-AP4 had no significant effect on noise in WT ONα RGCs (fig.6A,B), but reduced noise in the two older age ranges in rd10 ONα RGCs (P21-25: 56 ± 25%, p = 0.024, P26-30: 59 ± 23%, p = 0.05) and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint ( 6A,C).EPSCs evoked by 750 R*/rod/s stimuli (+50% contrast steps) were not significantly altered by 50 nM L-AP4 in either WT (fig.6D,E) or rd10 ONα RGCs (fig.6D,F).
Lastly, we assessed the effects of 50 nM L-AP4 on responses to positive contrast steps up to +400% contrast (2500R*/rod/s) with L-AP4 (fig.6G).L-AP4 reduced EPSC amplitudes in rd10 P16-20 RGCs from 50% -400% contrast stimuli (p = 0.0002, fig.6Gi) but had no significant effect on contrast responses in the older age groups, nor any effect on WT contrast response amplitudes at any age (fig.6Gi-iii).L-AP4 slightly reduced the half maximum contrast sensitivity of P26-30 WT ONα RGCs from 47 ± 8% contrast to 40 ± 15% contrast (p = 0.014) but did not change contrast sensitivity at other ages (fig.6H).Rd10 RGCs showed no significant change in contrast sensitivity with low dose L-AP4 regardless of age (fig.6I).L-AP4 may not have affected mGluR6 synapses equally, and differing degrees of degeneration across rd10 rods likely contributed to the variability of L-AP4's effect on signal, but the significant decrease seen in rd10 RGC noise with L-AP4 was consistent (fig.6A,G).These results suggest that low-dose L-AP4 in rd10 retinas reduced noise without altering signal under mesopic luminance, thus improving retinal encoding even as rods continued to degenerate.This change was most evident in rd10 animals age P21-30 before the complete loss of rods.

Rd10 retina does not tolerate intravitreal injections without photoreceptor damage
To test whether L-AP4 could improve visual function over the time course of degeneration in RP, we intraocularly injected L-AP4 into rd10 mouse eyes with the intent of recording in vivo ERGs over the course of rod degeneration.L-AP4 injection (10M, 0.5 µL) did not affect a-waves in WT animals (n = 6; pre-injection vs. one day post-injection) and reduced b-waves evoked by 10 cd•s/m 2 flashes by 40 ± 25% (p < 0.0001; fig.7Ai-ii), an expected result from a saturating concentration of and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 17 17 L-AP4.In rd10 animals, however, L-AP4 injection reduced a-waves by 70 ± 28% (p = 0.0008; fig.7Bi).Similar results were observed upon injection of normal saline (p = 0.002; fig.7Bi), indicating that intraocular injections were poorly tolerated by rd10 photoreceptors, preventing further study of the longitudinal effects of L-AP4 on visual signaling in rd10 retinas using intraocular injections.

RBCs are more depolarized at rest in rd10 retinas
Experiments presented thus far show that a low dose of L-AP4 during rd10 rod degeneration reduced circuit noise but not light-evoked responses in RGCs.To dissect the mechanisms by which L-AP4 influenced circuit noise properties, we first measured the resting membrane potential (RMP) directly from P20-25 RBCs with perforated patch recordings (fig.8A).We hypothesized that at this intermediate stage of degeneration, progressive rod loss and consequent loss of glutamatergic input would depolarize the RMP of RBCs.Indeed, rd10 RBCs rested significantly more depolarized (RMP = -43 ± 2 mV, n = 8) compared to WT RBCs (RMP = -53 ± 1 mV, n = 8, p = 0.003); (Dunn & Rieke, 2006;Grimes et al., 2014;Herrmann et al., 2011;Oltedal et al., 2009).A saturating concentration of L-AP4 (10 M) re-hyperpolarized the RMP of rd10 RBC to -52 ± 2 mV (p = 0.026, n = 8), closer to that of WT RBCs (fig.8B,C).50 nM L-AP4, which reduced noise levels in RGCs (fig.4,6), exerted variable effects on the RMP of individual RBCs that were overall insignificant.These results suggest that recordings from downstream RGCs, which collect input mediated by many RBCs, provide a more sensitive indication of the subtle effects of 50 nM L-AP4 on circuit noise.As a result of photoreceptor death, postsynaptic changes in the rd10 glutamatergic signaling cascade occur where mGluR6 (Barhoum et al., 2008;Gargini et al., 2007;Puthussery et al., 2009)  This glutamatergic signaling reorganization, in addition to a loss of glutamate from degenerated rods, may also contribute to the depolarized RMP observed in our data.
Synaptic output from the RBC is reduced in early stages of retinal degeneration.
The size of the readily-releasable vesicle pool (RRP) at RBC synapses reflects a balance between vesicle release and replenishment that depends on RBC membrane potential (V RBC ) (Graydon et al., 2018;Oesch & Diamond, 2011).As V RBC depolarizes in response to increased ambient luminance, ongoing RBC release also increases, thereby reducing the number of vesicles available for release and, consequently, the synaptic gain in response to subsequent visual stimuli.These findings suggest that L-AP4, by hyperpolarizing the RMP in rd10 RBCs, may increase the synaptic gain between RBCs and postsynaptic AII amacrine cells.This could explain, at least in part, how L-AP4 reduces noise in RGCs without significantly decreasing EPSC amplitude (fig.6).This hypothesis presumes that the release characteristics at RBC-AII synapses in rd10 retinas are similar to those in WT.To test this, we made dual whole-cell voltage clamp recordings from synaptically coupled RBC-AII pairs in acute retinal slices (Singer & Diamond, 2003); fig.8D-G).Voltage steps delivered to the presynaptic RBC elicited EPSCs in the postsynaptic AII (Figure 8E).A step to -29 mV elicited a large EPSC, the transient component of which reflects release of the entire RRP (Singer & Diamond, 2006).Smaller steps evoked release and a consequent decrease in the response to a subsequent large step, demonstrating how V RBC influences the size of the RRP (Oesch & Diamond, 2011) (fig.8F).
EPSCs elicited by large steps were 24% smaller in rd10 pairs, although variability between pairs rendered this effect statistically insignificant (fig.8G).Importantly, the dependence of V RBC on RRP size in rd10 pairs was similar to that observed in WT (Figure 8F): The size of the RRP varied steeply at about -50 mV, near the RMP in WT RBCs, suggesting that the more depolarized RMP observed in and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 19 19 rd10 RBCs would significantly decrease the RRP and, consequently, the synaptic gain of RBC-AII synapses -an effect that would be partially ameliorated by L-AP4 (Figure 8C).

Discussion
The results presented here show that L-AP4 reduced RP-induced noise in rd10 circuitry without compromising light-evoked responses in ONα RGCs, thereby improving visual signal fidelity even as photoreceptors degenerate.These data suggest L-AP4 as a potential therapy during photoreceptor degeneration to replace glutamatergic input to ON bipolar cells that is lost when photoreceptors die.
During rod degeneration, rd10 RBCs rest at more depolarized levels than in WT, likely reducing the gain of their synaptic output to AIIs (fig.8).By re-hyperpolarizing rd10 RBCs, a low dose of L-AP4 reduced circuit noise and restored synaptic gain.This novel approach should work in retinas expressing any RP-causing mutation and could complement or precede other therapies, such as prosthetic photoreceptors (B.-Y.Wang et al., 2022) or gene therapy (Bennett et al., 1996;Pang et al., 2008;Scalabrino et al., 2023;Tun et al., 2023), to restore vision lost to this disease.Stabilizing visual signaling in the inner retina during photoreceptor degeneration also may delay or ameliorate circuit remodeling (T.Wang et al., 2019) and thereby preserve the efficacy of subsequent restorative therapies.

Effects of degeneration vary within and between individual retinas
Photoreceptor degeneration in RP progresses at varying rates depending on the mutation (S.Daiger et al., 2013;S. P. Daiger et al., 2007): In humans, some RP patients present visual deficits by and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 20 20 10 years of age (Y.N. Kim et al., 2020), whereas others do not exhibit symptoms until middle age (Kuehlewein et al., 2021;McLaughlin et al., 1995).In the rd10 mouse model, rod degeneration begins at ~P16 in the central retina and progresses gradually to the periphery (Gargini et al., 2007).Single-cell and population recordings indicate that the physiological impact of disease progression varies between neighboring regions of the same retina (Puthussery et al., 2009;Stasheff et al., 2011).
We also observed high cell-cell variability in noise levels and light responses in rd10 RGC recordings (fig.3-6), likely due to the variable impact of rod degeneration on individual upstream RBCs.

Reducing noise in the primary rod pathway could enhance cone-mediated vision
The rod and cone pathways interact at several points within the retinal circuit.Rods contact cones directly through gap junctions (Schneeweis & Schnapf, 1995), and AII amacrine cells relay rod pathway signals to cone bipolar cell terminals (Völgyi et al., 2004).In low light, the high-gain primary rod pathway amplifies both signals and noise; as luminance levels increase, the gain of this pathway is reduced, clearing the way for smaller cone pathway signals to pass through.Rod degeneration in rd10 increases noise levels in the electrically-coupled ON cone bipolar cell to AII network (Goo et al., 2015;Jae et al., 2013;Stasheff et al., 2011;Trenholm & Awatramani, 2015), effects that are eliminated by blocking ionotropic glutamate receptors and gap junctions (Biswas et al., 2014) (Naples & Hampson, 2001;Thomsen, 1997) is typically applied at high concentrations (≥ 10 μM) to silence ON pathways (Slaughter & Miller, 1981).The 50 nM dose delivered in our experiments likely bound only a small fraction of mGluR6 receptors, yet it effectively reduced pathological noise without blocking visual signals from healthy rods (fig.6).
RBCs are depolarized substantially following loss of glutamatergic input at just a fraction of rod synapses (fig.8); a low dose of L-AP4 appears to reduce this pathological activity while still allowing RBCs to detect light-evoked changes in glutamate release at healthy synapses and transmit visual information to downstream targets.L-AP4 likely decreases the gain of healthy rod-RBC synapses, but by hyperpolarizing RBCs it also increases the gain of their synaptic outputs to AIIs (fig.8).Consequently, signals originating in surviving rods are transmitted effectively through the circuit to RGCs atop reduced background noise (fig.6).50 nM L-AP4 reduced contrast responses in the youngest rd10 RGCs, but no similar effect was seen in WTs.This result may be due to an unknown effect of manipulating glutamatergic input while rod degeneration is just beginning.
Effects of 50 nM L-AP4 were difficult to detect in RBCs (data not shown) but were clearly evident in ONα RGCs (fig.6).Each ONα RGC collects signals deriving from ~500 upstream RBCs (Dunn & Rieke, 2006) and thereby averages out variable signals to reveal the effects of 50 nM L-AP4.
Following photoreceptor death, mGluR6 (Gargini et al., 2007) and TRPM1channels (Gayet-Primo & Puthussery, 2015) eventually migrate within the plasma membrane towards the soma.Their functionality post-mislocalization remains unclear, a further argument that the optimal treatment window that best preserves retinal structural integrity is prior to total rod loss (Scalabrino et al., 2023).
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint
Clinical trial studies (Fenner et al., 2023;Sahel et al., 2021) and optogenetic approaches in mouse models of RP (Rodgers et al., 2023) suggest that vision may still be restored after total photoreceptor loss.Morphological remodeling, however, continues even after photoreceptor loss (Pfeiffer, Marc, et al., 2020), and would further impede therapies in later stages of the disease.Pharmacological therapy approaches that enhance circuit-wide inhibition or block gap junctions effectively reduce pathological activity (Biswas et al., 2014;Toychiev et al., 2013), but they interfere with other components of retinal circuitry that may cause off-target effects.Our approach suggests a broadly applicable treatment that reduces pathological noise without compromising surviving visual signals.
Unfortunately, we were unable to test the longitudinal effects of L-AP4 on morphological remodeling during RP because rd10 retinas did not tolerate intraocular injections, even at injection volumes below 0.5 µL (fig.7) (Hombrebueno et al., 2014;Miki et al., 2009).We suspect that the generally fragile state of the rd10 retina largely contributed to its inability to withstand mild trauma and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 23 23 caused by the injection.The small hole required to insert the syringe needle (30G) reduced the intraocular pressure, a change that was well tolerated by WT eyes but not by rd10 eyes; species with larger vitreal volumes may better tolerate injections.Alternatively, less mechanically traumatic delivery methods, e.g., nanoparticles, ocular drug implants, or eyedrops (Batabyal et al., 2020;Meza-Rios et al., 2020;Sun et al., 2021), may preserve functional tissue.
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint 24 24    was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.; https://doi.org/10.1101/2024.04.30.591881 doi: bioRxiv preprint noise during early rod degeneration (P16-20), peak rod degeneration (P21-25), and end of rod degeneration (P26-30) before and after application of 50 nM L-AP4 (lighter shades).L-AP4 reduced noise in all age groups in both strains.(B, C) Summary of baseline noise in WT (B) and rd10 (C) ONα RGCs at 3 age groups (± s.d., n = 10 cells/age/strain) showing noise at each age range before (x-axis) and after application of 50 nM L-AP4 (y-axis).L-AP4 reduced noise in all ages of both WT (p = 0.024, 0.015, and 0.022) and rd10 RGCs (p = 0.048, 0.02, and 0.042 respectively in increasing age groups).(D) Example EPSCs evoked by 40R*/rod/s light steps in WT and rd10 ONα RGCs during early rod degeneration (P16-20), peak rod degeneration (P21-25), and end of rod degeneration (P26-30) before and after application of 50 nM L-AP4 (lighter shades).(E, F) Summary of EPSC charge transfer in WT (E) and rd10 (F) ONα RGCs at all 3 age groups before and after application of 50 nM L-AP4 (lighter shades).No significant changes observed with L-AP4 in either WT or rd10 ONα RGC response amplitudes within each age group.and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024.and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted April 30, 2024. ;https://doi.org/10.1101https://doi.org/10. /2024.04.30.591881 doi: bioRxiv preprint .04.30.591881 doi: bioRxiv preprint

Figure 1 :
Figure 1: Working model of rod and ON-cone bipolar cell pathways in WT and rd10 retina.

Figure 2 :
Figure 2: In vivo ERG measurements reveal the time course of retinal dysfunction in rd10

Figure 4 :
Figure 4: Low dose of L-AP4 suppresses circuit noise under scotopic conditions.

Figure 8 :
Figure 8: RBCs are more depolarized at rest in rd10 retinas.
is also made available for use under a CC0 license.wasnot certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 and . Limiting pathological signals in RBCs would reduce noise in AIIs, thereby preventing contamination of cone pathway signals.By decreasing noise without reducing light responses, the low-dose L-AP4 treatment described here may improve visual signaling in degenerating retinas across light levels.and is also made available for use under a CC0 license.wasnot certified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 and is also made available for use under a CC0 license.wasnotcertified by peer review) is the author/funder.This article is a US Government work.It is not subject to copyright under 17 USC 105 V hold = -70 mV) in WT and rd10 ONα RGCs during early rod degeneration (P16-20), peak rod degeneration (P21-25), and end of rod degeneration (P26-30).(D)Summary of EPSC charge transfer evoked by +50% contrast steps in WT and rd10 ONα RGCs at all 3 age groups under mesopic (500 R*/rod/s) background luminance.WT ONα RGC responses were larger than those of rd10s in the two younger age groups (± s.d., n = 10 cells/age/strain, p < 0.0001 and p = 0.039).(E)ONαRGC responses of WT and rd10 retina across contrasts up to +400% on 500R*/rod/s background.WT ONα RGCs display larger responses than rd10s in the youngest age group (P16-20: p = 0.003).WT ONα RGCs reached saturation at +100% to +150% contrast, but rd10 RGC responses did not saturate until +250% to +300% contrast (median ± upper and lower quartile, n = 10 cells/age/strain).(F) WT ONα RGCs consistently reach half maximum light response at 50% contrast across ages.However, rd10 ONα RGCs reach half maximal contrast increased with age from +80% to +150% (± s.d., n = 10 cells/age/strain; P16-20 C 50 p = 0.024).