Colour and melanopsin mediated responses in the murine retina

Abstract

Introduction: Intrinsically photosensitive retinal ganglion cells (ipRGCs) integrate melanopsin and rod/cone-mediated inputs to signal to the brain. Whilst originally identified as a cell type specialised for encoding ambient illumination, several lines of evidence indicate a strong association between colour discrimination and ipRGC-driven responses. Thus, cone-mediated colour opponent responses have been widely found across ipRGC target regions in the mouse brain and influence a key ipRGC-dependent function, circadian photoentrainment. Although ipRGCs exhibiting spectrally opponent responses have also been identified, the prevalence of such properties have not been systematically evaluated across the mouse retina or yet been found in ipRGC subtypes known to influence the circadian system. Indeed, there is still uncertainty around the overall prevalence of cone-dependent colour opponency across the mouse retina, given the strong retinal gradient in S and M-cone opsin (co)-expression and overlapping spectral sensitivities of most mouse opsins. Methods: To address this, we use photoreceptor isolating stimuli in multielectrode recordings from human red cone opsin knock-in mouse (Opn1mwR) retinas to systematically survey cone mediated responses and the occurrence of colour opponency across ganglion cell layer (GCL) neurons and identify ipRGCs based on spectral comparisons and/or the persistence of light responses under synaptic blockade. Results: Despite detecting robust cone-mediated responses across the retina, we find cone opponency is rare, especially outside of the central retina (overall ~3% of GCL neurons). In keeping with previous suggestions we also see some evidence of rod-cone opponency (albeit even more rare under our experimental conditions), but find no evidence for any enrichment of cone (or rod) opponent responses among functionally identified ipRGCs. Conclusion: In summary, these data suggest the widespread appearance of cone-opponency across the mouse early visual system and ipRGC-related responses may be an emergent feature of central visual processing mechanisms.

Keywords: RGCs; colour opponency; intrinsically photoreceptive retinal ganglion cells; ipRGCs; melanopsin; mouse; retinal ganglia cells; silent substitution.

Figure 1

Detection of colour opponent units. (A) Categorisation of all units that were responsive to 67% Michelson contrast stimuli (n = 1,429) based on their response to L- and/or S-cone-selective contrast modulation (0.25 Hz square wave). (B) Spectral composition of the stimulus pairs used to isolate cone-based responses. Dashed filled line: low, solid line: high. (C) Relative photon absorbance (log₁₀ photons) and Michelson contrast for each photopigment under each condition. (D–G) Mean ± SEM responses of example units for each of the cone isolating stimuli shown in (B,C) (n = 100 trials/stimulus); (D) non-opponent S-ON response, (E) non-opponent L-OFF response, (F) S-ON/L-OFF colour-opponent cell, (G) L-ON/S-OFF colour-opponent cell. (H) The mean ± SEM population response of S-ON/L-OFF (n = 36) colour opponent cells.

Figure 2

Distribution of colour opponency across the retina. (A) Population level cone preference from three 256-channel MEA recordings (spanning 3 × 3 mm of retina). Black colour indicates electrodes where we did not detect significant modulations in multiunit firing to either L- or S-opsin isolating stimuli. (B) Interpolated map of retina cone-preference from experiments in (A) showing opsin transition zone modelled by linear fitting. (C) Spatial distribution of isolated units with strong S-opsin bias, no bias, or strong L-opsin bias relative to modelled opsin transition zone (n = 1,429 cells from 25 retinas). Left panel shows numbers of isolated cells, right panel shows same data as a proportion of total for that location. (D) Spatial distribution of isolated colour opponent units relative to modelled opsin transition zone (n = 36 S-ON/L-OFF and n = 9 L-ON/S-OFF cells). Left panel shows estimated locations of individual cells, right panel shows binned proportions of colour opponent cells relative to all cells isolated within that region (data analysed by χ²-test).

Figure 3

Synaptic blockade (DNQX and DL-AP4) to confirm ipRGCs. (A) Spectra and Relative photon absorbance (log₁₀ photons) for each photopigment for the stimuli used here to check for extrinsic responses (rod or cone driven, top) and intrinsic responses (bottom). (B,C) Two example ipRGCs. (B) Rod/Cone mediated responses to a 500 ms light step (ISI: 500 ms; Violet or Violet and Red light) before and during the drug application which abolishes the RGC responses from the outer photoreceptive layer. Left raster plot, red lines denote the first and last 200 trials that were used to produce the before and after histograms to the right. The red line on the histograms denotes the 99% confidence interval that the response is above baseline firing rate. (C) Corresponding response to a 10 s light step (Blue light; 90 s ISI) in the presence of the synaptic blockade. Above Perievent histogram, red line denotes the 99% confidence interval that the response is above baseline firing rate. Below is the corresponding heatmap showing the trial by trial response. (D–F) Combined data from all ipRGCs classified this way. (D) Proportion of all light responsive units that showed a light response following synaptic blockade. (E) Mean ± SEM population response to a 10 s light pulse whilst under synaptic blockade (n = 30, 2.5 s smoothing). (F) The Mean ± SEM cone responses from this ipRGC population in the absence of synaptic blockade (using stimuli illustrated in Figures 1B,C).

Figure 4

Identification of melanopsin responses via cone-isoluminant light steps. (A) Stimuli presented to the mice. Melanopsin High (blue) and Low (red) stimuli were presented as 10 s light steps from darkness (80 s ISI). Respective spectra and relative photon absorbance (log₁₀ photons) for each photopigment is shown below. Michelson contrast between stimuli shown in black. (B) Proportions of all cells passing our criteria for ipRGC classification based on response under synaptic blockade or comparison between Mel High and Low steps in Opn1mw^R (left) and Opn1mw^R; Opn4^−/− (right) mice. Data were collected in Opn1mw^R mice (C,E) and animals which lacked functional melanopsin expression (Opn1mw^R; Opn4^−/−; D,F). (C,D) Mean ± SEM response of units that were deemed melanopsin positive from the Mel high vs. low steps from darkness. (C: n = 66/1,735, D: n = 4/291). (E,F) Mean ± SEM response under synaptic blockade for putative ipRGCs to a 10 s light pulse (2.5 s smoothing; stimuli and spectra Figure 3A bottom). (E) Mean ± SEM response of cells shown in (C). Red trace is the mean response of ipRGCs identified under synaptic block Figure 3E. (F) All cells passing either of our ipRGC classification criteria in red cone MKO mice (n = 6).

Figure 5

Cone mediated responses in the second subpopulation of ipRGCs. (A) Distribution of S- vs. L- cone preference across putative non-M1 ipRGCs isolated using Mel High vs. Low steps. (B) Mean ± SEM cones responses of the putative non-M1 ipRGC population to various cone isolating stimuli (n = 66).

Figure 6

Putative Rod opponency. (A) Paired light stimuli used to detect Rod opponency. Above: Relative photon absorbance (log₁₀ photons) and Michelson contrast for each photopigment. below spectra, dashed filled line: low, solid line: high. (B) An example unit showing evidence of rod-opponency, where responses to L+S contrast were greater than the response to equivalent contrast applied to both rods and cones. (C) Mean ± SEM of all cells with evidence of rod opponency (n = 18/1,429). (D) Proportion of units that responded to the 67% Michelson contrast steps that were classified as rod opponent. (E) Cone preference of putative Rod Opponent cells (n = 18). (F) Percentage of cells that are classed as rod opponent at each location. (G,H) Population data from all units displaying rod opponency. (G) Response to Mel high and Mel low steps from darkness (Mel high: blue, Mel low: red). (H) Response to a 10 s light step (Monochromatic Blue light, 90 s ISI) in the presence of synaptic blockade (spectra and irradiance as used in Figure 3).