Horizontal cells expressing melanopsin x are novel photoreceptors in the avian inner retina

Significance

In the vertebrate retina, three types of photoreceptors—visual photoreceptor cones and rods and intrinsically photosensitive retinal ganglion cells (ipRGCs)—converged through evolution to detect light and regulate image- and nonimage-forming activities. ipRGCs express the photopigment melanopsin (OPN4), encoded by two genes: the Xenopus (Opn4x) and mammalian (Opn4m) orthologs. In the chicken retina, both OPN4 proteins are found in ipRGCs. Opn4x is also present in retinal horizontal cells (HCs) connecting with visual photoreceptors. We show that HCs displaying intrinsic photosensitivity constitute photoreceptors requiring Opn4x expression and retinaldehyde, acting through an invertebrate-like phototransduction cascade and GABA release. These mechanisms may enable HCs to regulate nonvisual tasks with ipRGCs and lateral interactions with visual photoreceptors.

Abstract

In the vertebrate retina, three types of photoreceptors—visual photoreceptor cones and rods and the intrinsically photosensitive retinal ganglion cells (ipRGCs)—converged through evolution to detect light and regulate image- and nonimage-forming activities such as photic entrainment of circadian rhythms, pupillary light reflexes, etc. ipRGCs express the nonvisual photopigment melanopsin (OPN4), encoded by two genes: the Xenopus (Opn4x) and mammalian (Opn4m) orthologs. In the chicken retina, both OPN4 proteins are found in ipRGCs, and Opn4x is also present in retinal horizontal cells (HCs), which connect with visual photoreceptors. Here we investigate the intrinsic photosensitivity and functioning of HCs from primary cultures of embryonic retinas at day 15 by using calcium fluorescent fluo4 imaging, pharmacological inhibitory treatments, and Opn4x knockdown. Results show that HCs are avian photoreceptors with a retinal-based OPN4X photopigment conferring intrinsic photosensitivity. Light responses in HCs appear to be driven through an ancient type of phototransduction cascade similar to that in rhabdomeric photoreceptors involving a G-protein q, the activation of phospholipase C, calcium mobilization, and the release of the inhibitory neurotransmitter GABA. Based on their intrinsic photosensitivity, HCs may have a key dual function in the retina of vertebrates, potentially regulating nonvisual tasks together with their sister cells, ipRGCs, and with visual photoreceptors, modulating lateral interactions and retinal processing.

In the vertebrate retina, photoreceptors can be classified according to their function as canonical or noncanonical. The first group comprises specialized ciliary retinal neurons, cones, and rods in the outer retina, which participate in the image-forming processes associated with day/night vision. The second group is composed of intrinsically photosensitive retinal ganglion cells (ipRGCs) in the inner retina, which preferentially participate in the processing of photic inputs related to nonimage-forming tasks (photic synchronization of circadian rhythms, pupillary light reflexes, inhibition of pineal melatonin, etc.) (1–7). The photopigment melanopsin (OPN4) (1) confers photosensitivity to ipRGCs as clearly shown through later experiments with OPN4 knockout mice or OPN4 heterologous expression in nonretinal cells for loss or gain of function, respectively (8–11). Through evolution, OPN4 appears to have been encoded by at least two genes in vertebrates: Opn4x and Opn4m, the Xenopus and mammalian ortholog genes, respectively (12). Interestingly, the first-appearing vertebrates, nonmammals including fish, amphibians, and birds, possess both Opn4 genes whereas mammals only have Opn4m. Evidence suggests that during the course of evolution, mammals lost some visual opsins and Opn4x as they entered the nocturnal niche. In the chicken retina, different laboratories have reported the expression of Opn4 genes in outer nuclear, inner nuclear, and ganglion cell layer (GCL) cells (12–21). At the protein level, OPN4M was shown to be restricted to ipRGCs, whereas OPN4X was found in the GCL at embryo day 8 (E8) but mostly in PROX1 (+) horizontal cells (HCs) by E15 (21); the homeobox gene PROX1 is a universal HC marker that is expressed from E8 forward (22, 23). These and other experiments raised a crucial question: Can Opn4x (+) HCs intrinsically respond to light? HCs are retinal interneurons lying adjacent to the outer plexiform layer and implicated in visual perception and the regulation of signaling between cones, rods, and bipolar cells, enhancing image contrast, color discrimination, and light adaptation. HCs are involved mainly in the lateral interactions of the outer retina (24, 25). Once determined, retinal HC precursors migrate, differentiate, and express a set of specific markers; the final inner nuclear positioning occurs around E15 (23, 26), the same time at which a very strong expression of OPN4X was observed (21). Molecular comparative biology indicates that rods and cones have evolved from a common ciliary photoreceptor precursor, whereas RGCs, amacrine cells, and HCs may have evolved from a common precursor with rhabdomeric photoreceptors (27). Two studies provide strong evidence of photoresponses in HCs in teleosts (28, 29), shown to express vertebrate ancient (VA) opsin and OPN4 and to intrinsically respond to light well beyond rod and cone responses (28–31). In this work we investigated the intrinsic photosensitivity of HCs from embryonic avian retina, the role of the nonvisual photopigment Opn4x in these photic responses, and the neurochemical features of the events triggered by the light stimuli.

Results

Characterization of HC Primary Cultures Expressing OPN4x.

HCs isolated from the 2.5% phase of a BSA gradient obtained from disaggregated embryonic chicken retinas at E15 displaying positive immunoreactivity for the typical HC marker PROX1 (≥85%) after flow cytometry (Fig. S1) were cultured for 2–3 d under constant illumination conditions and immunostained for OPN4x. Fig. 1 shows OPN4x (+) immunoreactivity in most cells of the 2.5% phase. These cells also expressed PROX1 after subsequent immunopurification by an anti-OPN4x antibody (Fig. S2 A and B). Cultured cells also expressed mRNAs for another HC marker, Islet-1, and for Gq, the member of the G-protein family involved in the rhabdomeric-like photocascade (Fig. S2C). By contrast, only trace inmmunoreactivity associated with RETP1 (≤6%) and NF-200 (≤5%), markers of photoreceptor cells and RGCs, respectively, was observed in the cultures (Fig. S2), and only when cultures were maintained for 5 d or longer were a few Muller glial cells seen using the specific marker glutamine synthase.

Fig. S1.

(A) Schematic diagram showing the discontinuous BSA gradient used to isolate Opn4x (+) HC highly enriched retinal cell populations after immunopurification with a specific Opn4x antibody. Cells were dissociated from the embryonic retinal tissue at E15 as described in SI Materials and Methods. After dissociation, the cell suspension was subjected to a discontinuous BSA gradient prepared with BSA solutions of 1%, 2.5%, 3%, and 4%. A highly enriched HC population was collected mainly from the 2.5% BSA phase and further immunopurifed with an anti Opn4x antibody by immunopanning and cultured for several days under the described conditions. (B) A representative experiment of flow cytometry for total disaggregated chicken embryonic retinal cells (Left) or 2.5% gradient phase cells (Right) at day 15 displaying positive immunostaining for PROX1, the specific horizontal cell marker (labeled with Alexa-Fluor-488 A) in the total retina (30.8%) and in the 2.5% BSA phase (85%).

Fig. 1.

Characterization of primary cultures of embryonic retinal cells obtained from the different phases of a discontinuous BSA gradient at day 15. (Left) Cell cultures from the different phases (2.5%, 3%, 4%) show positive immunoreactivity for Opn4x (green) in cell somas and processes; propidium iodide (PI, red) staining denotes cellular nuclei. (Right) Magnification (200×) of microphotographs from cell cultures from the 2.5–3% BSA gradient phases showing immunofluorescence colocalization (merge) between PI and Opn4x in cell somas and processes. OPN4 (+) immunoreactivity demarks the typical morphology of neuronal cells with shorter or longer processes as indicated in the selected square further magnified on the right.

Fig. S2.

Immunostaining of primary cultures of chicken retinal cells at E15 from the 2.5% phase of a discontinuous BSA gradient after 5 d in culture. (A) DAPI staining of cellular nuclei (blue, Right) and positive immunocytochemistry (green, Left) (magnification: 200×) for PROX1, a specification marker for HCs and Opn4x in most culture cells. No positive immunoreactivity was observed for NeuN, a neuronal marker of RGCs, nor for NF-200, denoting neuro-filaments. (B) Merging of DAPI, PROX1, and Opn4x immunostaining of individual cells in culture as shown in A (magnification: 600×). (C) Detection of mRNAs by RT-PCR in HCs or total retina. HCs from the 2.5% phase of the BSA gradient display detectable levels of transcripts for the HC specification markers Prox-1 and Islet-1, as well as for components of the rhabdomeric-like photocascade (G-protein q and Opn4x) but not for a RGC specification marker (Brn3) or α-transducin (α-trans), a typical marker of visual photoreceptors cells, by RT-PCR in embryonic HC primary cultures. Total retina was used as a positive control and GAPDH as a housekeeping gene.

Retinal cells from the 3% and 4% phases of the gradient also expressed OPN4x and upon magnification showed much longer neurites than those in HC cultures, likely denoting OPN4x (+) RGCs (Fig. 1).

Endogenous Levels of Retinaldehyde Isomers and Light Responses in HCs.

To function as a photopigment, OPN4 requires the presence of sufficient retinaldehyde isomers as the active chromophore (6–11). Results shown in Fig. 2 C and D indicate that HC cultures contained detectable basal levels of all-trans retinal. Nevertheless, to evoke significant light responses, 0.6 µM of exogenous all-trans retinal were added to the cultures in most experiments. Light response was then assessed by Ca²⁺ fluorescent imaging with Fluo4 in cells from disaggregated embryonic retinas at E15 kept in culture for several days (Fig. S3). When photic responses were evaluated in HC cultures, pulses of white light (1,000 lx) of different durations ranging from 5 to 50 s were able to promote significant increases in relative Ca²⁺ levels compared with basal fluorescence seen in the dark (P < 0.001 by ANOVA) (Figs. 2 and 3). Responses were further increased with the addition of exogenous all-trans retinal (Fig. 2B). After light stimulation of different durations, relative levels of Ca²⁺ increased 50% from 5 s of exposure to 20–50 s (Fig. 3), and the time to peak revealed that the highest response in each experiment did not attain saturation along the photic stimuli. Moreover, when HCs were exposed to prolonged illumination stimuli of longer duration (5–10 min), sustained responses were observed in intracellular Ca²⁺ levels, peaking after 1 min of light onset (ΔF = 1.34 ± 0.06), decreasing by 50% after 3 min, and returning to basal levels minutes later. The decline in light responses was probably due to fluorescent indicator saturation after continuous light exposure. When two brief pulses of bright light were given to the cultures, separated by an interval of 5 min in the dark to allow for cell recovery, Ca²⁺ levels peaked immediately after the photic stimulation for both pulses (Fig. 3E). In addition, when light pulses of different wavelengths were tested in the Opn4x (+) HC cultures, using LEDs of similar intensities at 395–400 nm, 460 nm, and 625 nm for 5 s, the highest responses were visualized with blue light (Fig. S3D).

Fig. 2.

Light-induced increase in Ca²⁺ levels measured by Fluo-4 AM fluorescent microscopy measured in HC cultures with (w/) or without (w/o) the addition of exogenous all-trans retinal (0.6 µM). (A) Intracellular increase in Ca²⁺ levels after light exposure for 10 s (yellow mark) without (w/o) exogenous all-trans retinal in individual cells (red) compared with levels with (w/) exogenous all-trans retinal (blue line) representing the average of photic responses in individual cells. (B) Graphical representation of relative fluorescent Ca²⁺ levels (ΔF) in HC cultures with (w/) or without (w/o) all-trans retinal (0.6 µM) administration in the culture medium. (C and D) Endogenous levels of all-trans retinal in primary cultures of HCs by HPLC determination (C) in both all-trans retinal stereoisomers, syn (peak at 357 nm) and anti (peak at 361 nm), by UV spectrum (D). Data are mean ± SEM (n = 3). Calcium fluorescent imaging was assessed as stated in Materials and Methods.

Fig. S3.

Light-induced changes in intracellular Ca²⁺ levels at E15 measured by Fluo-4 AM fluorescent microscopy in individual retinal cells from whole embryonic disaggregated retinas kept in culture (A–C) or in primary HC cultures (D): (A–C) Individual cells in culture exhibit differential responses to white light stimuli of 1,000 lx by increasing (A) or decreasing (B) intracellular Ca²⁺ levels whereas others display very weak responses (C) or no response at all (not shown) after subsequent light pulses. Cells were exposed to two brief white light pulses of 1,000 lx for 10 s each. Light duration is marked with a yellow box. (D) Schematic representation of the light-induced changes in the relative fluorescent Ca²⁺ levels (ΔF) in primary Opn4x (+) HC cultures subject to light stimuli of different wavelengths and similar intensities. Cell cultures were exposed to light pulses using LEDs of 395–400 nm, 460 nm, and 625 nm for 5 s The highest responses recorded were visualized with blue light. Data are mean ± SEM (n = 6 from two independent experiments); *P < 0.05 by Student t test.

Fig. 3.

Light-induced changes in intracellular Ca²⁺ levels measured by Fluo-4 AM fluorescent microscopy in HC cultures. (A–C and F) Individual cells in culture exhibit a significant response to light by increasing intracellular Ca²⁺ levels as seen in lighter color recordings. The red line represents the average response of all photosensitive cells assessed. Cells were exposed to a single brief white light pulse of 1,000 lx for 5 (A), 20 (B), or 50 s (C) or to two brief white light pulses of 1,000 lx for 20 s (E). Light duration is marked with yellow in the axis. Plots shown in A–C and E represent the F/F_o ratio for the changes in fluorescence levels between each relative intensity level measured after a light stimulus of 1,000 lx (F) and the mean of intensities of serial pictures before stimulation (F_o). (D) Graphical representation of relative fluorescent Ca²⁺ levels (ΔF) calculated as indicated in Materials and Methods after 5 s, 20 s, and 50 s of light stimulation compared with dark controls. Data are mean ± SEM (n = 25 from three independent experiments); *P < 0.05 by Student t test. (F) A typical HC (magnification: 600×) showing several processes grown after 3 d in culture loaded with Fluo-4 AM and displaying a significant increase in intracellular calcium fluorescence levels after light exposure, visualized in a pseudocolor scale with the highest response in red. Ca²⁺ levels were measured by Fluo-4 AM fluorescent microscopy as described in Materials and Methods.

OPN4X-Mediated Intrinsic Photosensitivity in HCs.

We next investigated the role of OPN4X in conferring HC intrinsic photosensitivity by knocking down its expression and assessing the effect of light-evoked responses on fluorescent Ca²⁺ levels. Fig. S4 shows that the specific shRNA for Opn4x significantly decreased OPN4X expression (Fig. S4 A, B, E, and F). Light increases in intracellular somatic Ca²⁺ levels in control cultures treated with a scrambled sequence RNA (ssRNA) (Fig. 4 A and B) were completely attenuated in those cells positively transfected with Opn4x shRNA (Fig. 4 A and B; Fig. S4 C and D). Fig. 4B indicates an almost 90% decrease compared with the control and a significant treatment effect (P < 0.001). In another series of experiments, when a chemically synthesized opsinamide was used to inhibit OPN4 activity as previously shown (32), no photic increases in fluorescent Ca²⁺ levels were found in HC cultures compared with cells treated with vehicle only (Fig. S5).

Fig. S4.

Knockdown of Opn4x protein expression in HC cultures. (A–D) Expression of Opn4x in HC primary cultures (magnification: 200×) was substantially knocked down by specific Opn4x shRNA. Cells in culture were stained with DAPI (blue, nuclear staining) and treated with the Opn4x shRNA (B) or ssRNA (control) (A) as described in Materials and Methods. A significantly decreased OPN4X immunolabeling (red) was found in Opn4x shRNA-treated cells (B) compared with control cells treated with the ssRNA (A). A significant number of cells in the cultures were positively transfected with the shRNA as shown in C (red) and loaded with the fluorescent indicator Fluo-4 AM (green); merging (yellow) is visualized in some cells (D). The decrease in OPN4X expression was confirmed by immunocytochemistry (B–D) and by RT-PCR for the specific shRNA used compared with ssRNA. GADPDH was determined as a housekeeping gene (E). (F) Histograms showing relative levels of Opn4x mRNAs clearly indicate a significant decrease of ∼55% in Opn4x mRNA expression in HC cultures after shRNA treatment compared with ssRNA controls.

Fig. 4.

Calcium fluorescent imaging of HC cultures treated with the Opn4x shRNA. (A) Significant light-induced changes in intracellular Ca²⁺ levels found in HC cultures treated with the control ssRNA (red line) were drastically affected in cell cultures treated with the specific Opn4x shRNA (blue line). (B) Graphical representation of relative fluorescent Ca²⁺ levels (ΔF) after a brief white light pulse of 1,000 lx for 5 s in Opn4x shRNA-treated cell cultures compared with ssRNA-treated cells (control). Data are mean ± SEM (n = 15 from three independent experiments); ***P < 0.001 by Student t test.

Fig. S5.

Effect of melanopsinergic (opsinamide) and visual photocascade inhibitors on the intracellular Ca²⁺ after light exposure in HC cultures. Significant light-induced changes (F/F_o) in intracellular Ca²⁺ levels visualized in HC cultures treated with the vehicle (control) (B) were drastically affected in cell cultures treated with opsinamide (10 µM) (A). Graphical representation of relative fluorescent Ca²⁺ levels (ΔF) calculated as indicated in SI Materials and Methods after a brief white light pulse of 1,000 lx for 5 s in cell cultures treated with the effector (opsinamide) compared with controls treated with the vehicle (C). Data are mean ± SEM (n = 10 from 2 independent experiments); ***P < 0.001 by Student t test. By contrast, when specific visual signaling inhibitors such as the pertussis toxin (500 nM, preincubated for 24 h) (D) or zaprinast (0.1 M, preincubated for 4 h) (E) was tested, light responses were not significantly affected in the cultures.

Light-Triggered Events in HCs.

To investigate the molecular events triggered by light downstream of OPN4X activation, we treated Opn4x-immunopurifed HC cultures with different Ca²⁺ chelators, retinal bleachers, or Gq protein and phospholipase C (PLC) inhibitors to examine their effect on fluorescent Ca²⁺ levels after light stimulation. The elevation in somatic Ca²⁺ levels by light observed in cultures treated with vehicle (Fig. 5 F and G) were completely abolished by treatment with U73122, a PLC inhibitor (Fig. 5 A and G), or with hydroxylamine, a retinal bleacher that strongly affects photopigment activity (Fig. 5 B and G). Suramin analogs were shown to act as direct antagonists of heterotrimeric G proteins (33) and reported to impair light responses in ipRGCs (11). We examined the effect of 100 μM suramin on HCs applied for 15–20 min before light stimulation and found that light responses were substantially abolished after treatment (Fig. 5 C and G). On the contrary, when specific inhibitors of the visual phototransduction cascade were used, such as the pertussis toxin (500 nM) that disrupts signaling through the Gi/o protein α-transducin or zaprinast (0.1 M), a specific phosphodiesterase 6 (PDE6) inhibitor (6, 17), light responses were not affected in the cultures (Fig. S5 D and E).

Fig. 5.

Effect of retinal bleachers and nonvisual phototransduction cascade inhibitors on intracellular Ca²⁺ after light exposure in HC cultures. Significant light-induced changes (F/F_o) in intracellular Ca²⁺ levels visualized in HC cultures treated with the vehicle (control) (F) were drastically affected in cell cultures treated with the PLC inhibitor U73122 (5 µM) (A); the retinal bleacher hydroxylamine (Hyd, 30 mM) (B); the G-protein q inhibitor suramin (Sur, 100 µM) (C); the blocker of the IP₃-induced Ca²⁺ release 2-APB (100 µM) (D); or a chelating agent, 1 mM EGTA (E). (G) Graphical representation of relative fluorescent Ca²⁺ levels (ΔF) after a brief white light pulse of 1,000 lx for 5 s in cell cultures treated with the effectors compared with vehicle-treated controls. Data are mean ± SEM (n = 20 from three independent experiments); ***P < 0.001 by Student t test.

In an attempt to elucidate the calcium sources responsible for the light increases observed in somatic Ca²⁺ levels and because it is known that 2-aminoethoxydiphenyl borate (2-APB) is a membrane-permeable blocker of the inositol 1,4,5-trisphosphate (IP₃)-induced Ca²⁺ release from internal stores (34), we investigated the effect of 2-APB on light responses in HCs; the treatment with 100 µM of 2-APB before light exposure significantly inhibited light-mediated increases in intracellular Ca²⁺ levels (Fig. 5 D and G). Furthermore, photic responses were substantially blocked when HC cultures were treated with EGTA (1 mM), a known extracellular Ca²⁺ chelating agent, before light stimulation (Fig. 5 E and G).

HCs can regulate photoreceptor functioning upon depolarization, and GABA release may be implicated in lateral inhibition (35–37). To investigate its potential utilization as a neurotransmitter and its release in these cells under the different light conditions examined, HCs were loaded with ³H-GABA in the dark, the culture medium was washed out, and the cells were then exposed to bright light for 30 min or kept in the dark. HCs in culture take up radiolabeled GABA from the extracellular medium under control conditions (Fig. 6), and light stimulation significantly increased its release to the extracellular medium compared with dark controls (P < 0.001).

Fig. 6.

³H-GABA release in HC cultures after 30 min of light exposure. HCs were incubated for 30 min at 37 °C with ³H-GABA (0.5–1 mCi, specific activity 87.1 Ci/mmol) in the dark, the culture medium was washed out, and cells were exposed to bright light for 30 min or kept in dark. Graphical representation showing ³H-GABA release in primary HC culture after 30 min of light (1,000 lx) and dark controls. Data are mean ± SEM (n = 40 from three independent experiments); ***P < 0.001 by Student t test.

Discussion

Within the retina, HCs project laterally and integrate inputs from visual photoreceptor rods and cones, providing them with negative and positive feedback to generate an antagonistic, and to a lesser degree agonistic, center-surround receptive field (Scheme 1, Scenarios 1 and 3) (25). Each HC receives chemical synaptic inputs from many rods and cones and, in return, generates a feedback signal that alters HC neurotransmitter release. Through this neural network comprising HCs, rods, cones, and bipolar cells, light in the retina generates a representation of spatial contrast and visual detection of edges, color discrimination, and light adaptation (35, 36). HC feedback is critical for establishing the antagonistic receptive field-surround of the subsequent neuron in the visual path (the bipolar cell). HCs are thus responsible for lateral interaction and inhibition (24, 34, 35). In the dark, HCs are depolarized as a consequence of glutamate release by visual photoreceptors, which in turn causes the release of GABA by HCs onto cone synapses promoting a neurochemical inhibition, proton release, or ephaptic signaling (35). Conversely, when the light is on, cones and rods become hyperpolarized, stopping the release of glutamate and thus preventing the depolarization and subsequent HC response. In addition to their typical role in the retinal circuits, in nonmammalian vertebrates HCs may conserve vestiges of ancient photoreceptor cells (Scheme 1, scenario 2) as clearly demonstrated for ipRGCs (2, 6, 17, 18). In teleosts, it has been reported that retinal HCs expressing VA opsin or OPN4 can intrinsically respond to light (28, 29). OPN4 is found in brain, iris, and retinal cells of vertebrates (1, 3, 5, 6, 12, 17–19) and highly conserved through evolution (38). In mammals OPN4 is confined to ipRGCs; however, the distribution of multiple Opn4s (Opn4m and Opn4x) in diverse retinal cell types of nonmammalian vertebrates suggests a biological function of higher complexity (see review in ref. 6). In the chicken retina, HCs express Opn4 mRNA (14, 19) and Opn4x protein (21). OPN4 is an ortholog of the Gq-coupled invertebrate opsins (6), and the photocascade operating in the OPN4 (+) ipRGCs clearly involves the participation of PLC, Ca²⁺ mobilization, and TRP channel activation (18, 39–41). The embryological origin of these cells within the retina and other molecular features strongly suggest that ipRGCs and HCs are likely to derive from a common ancient photoreceptor cell. The present paper provides evidence that Opn4x-expressing HCs kept in culture in the presence of retinaldehyde intrinsically respond to light with significant and sustained increases in intracellular Ca²⁺ from internal IP₃ stores but also in part from the extracellular medium. The cascade depends on G-protein q and PLC activation and ultimately leads to the release of intracellular GABA. Light-sensitive HCs can provide a plus-local regulation of visual photoreceptor function, further modifying neurotransmitter release and thus light intensity perception, because under this illumination condition HCs may increase negative feedback with a much stronger lateral inhibition of neighbor photoreceptors. One possible function of OPN4 is depolarization of the melanopsin-containing HCs themselves, which may inhibit the hyperpolarization responses of HCs triggered by light-dependent hyperpolarization of photoreceptor cells and thus regulate the negative feedback to visual photoreceptors. Accordingly, the OPN4-mediated depolarization of HCs by light may contribute to membrane potential regulation in the photoreceptor cells, modulating visual function. For its part, negative feedback reduces the dynamic range of the cone synapse by diminishing the maximal release rate of cones. By boosting neurotransmitter release from cones, positive feedback may recover the dynamic range lost in the negative feedback. Jackman and colleagues (25) propose that positive and negative feedbacks spread out differentially in the retina. Their results suggest that positive feedback acts locally; because there is no evidence of factors preventing positive feedback from causing runaway, this crucial role could be taken on by HCs expressing melanopsin. HCs cooperate in image-forming pathways with feedback signaling to cones and feed forward signaling to bipolar cells (42, 43). Here we provide evidence that HCs in culture release ³H-GABA upon 30 min of light exposure. Several mechanisms for HC feedback to photoreceptors have been proposed. HCs contain GAD, the synthesizing GABA enzyme (44, 45), and exhibit a direct GABAergic synaptic connection to ON and OFF bipolar cells (46) and to visual photoreceptors (47), showing that vesicular GABA release from HCs is required for feedback inhibition of photoreceptors.

Scheme 1.

Schematic representation of three different scenarios within the retina involving the synapses between visual photoreceptors and HCs after light stimulation. (Scenario 1) Upon light stimulation, classic HCs project laterally and integrate inputs from visual photoreceptor cones and rods, providing critical negative feedback for establishing an antagonistic receptive field. (Scenario 2) Intrinsically photosensitive HCs depolarize upon light stimulation, releasing GABA on the ephaptic synapse with visual photoreceptors, causing further hyperpolarization and a decrease in glutamate release. (Scenario 3) HCs may also transmit a positive feedback signal to cone terminals, elevating intracellular Ca²⁺ and accelerating neurotransmitter release.

Despite basic similarities, cell number, layer length, photopigment localization, and physiological responses differ significantly in the inner retina of birds compared with that of primates; in fact, the avian retina is especially rich in HCs and amacrine cells and presents higher complexity of interconnections and intraretinal visual processing. The complex processing occurring in higher forebrain areas in primates is achieved at a lower level in birds, within the inner retina perhaps indicating one of the crucial differences between primate and avian visual systems.

Our findings demonstrate that HCs are retinal photoreceptors expressing a retinaldehyde-based Opn4x photopigment that renders cells intrinsically photosensitive. In nonmammalian vertebrates these retinal interneurons may have a dual function, somehow regulating nonvisual tasks such as those in which ipRGCs are involved or lateral interactions with visual photoreceptors such as typical HCs (Scheme 1). Our findings leave many open questions as to the way in which this intrinsic photosensitivity affects in retinal circuits or why these sensitive interneurons were evolutionarily preserved in some species. The findings provide evidence of intrinsically photosensitive retinal HCs in birds that significantly increases the complexity of retinal neural circuits for light perception and visual processing.

Materials and Methods

For complete information on materials and methods, see SI Materials and Methods.

Animal Handling.

All experiments were performed in accordance with the Use of Animals in Ophthalmic and Vision Research of the Association for Research in Vision and Ophthalmology, approved by the local animal care committee (School of Chemistry, Universidad Nacional de Córdoba; Exp. 15–99-39796).

Primary Cultures of Embryonic Horizontal Cells.

HCs were purified from chicken neural retinas at E15 by a BSA discontinuous gradient of concentrations ranging from 1 to 4 % (wt/vol) as previously reported (48). Highly enriched HC cultures were characterized by specific HC markers as previously shown. See SI Materials and Methods for further information.

RNA Isolation and RT-PCR.

Total RNA from HC primary cultures was extracted using the TRIzol kit for RNA isolation (Invitrogen) (7, 48). See SI Materials and Methods for further information.

An initial denaturation step of 1 min at 94 °C, 25 cycles of 60 s at 94 °C, 50 s at 60°–65 °C, 90 s at 72 °C, and a final 5-min elongation step at 72 °C. Amplification products (Table S1) were separated by 2% agarose gel electrophoresis and visualized by ethidium bromide staining.

Table S1.

RT-PCR oligonucleotide primers

Name	Forward sequence	Reverse sequence	Size (pb)
Prox-1	CCAAATGTACTCCGCAAGC	TGCCACCATTTTTGTTCATGT	121
Islet-1	TATCCAAGGGATGACAGGAAC	GCTGTTGGGCGTATCTGGGAG	256
αTrans	AAGGACCTCAACTTCAGGATGT	CAGTCCTTGAGGTTCTCCTTG	479
Brn3	CCATCCTGCACGAGCCCAAGTA	GCCCCGTAGCAAGGTCTCATCAA	128
Gq	TCAAAACATCTTCACTGCCATG	TCAAAACATCTTCACTGCCATG	255
GAPDH	AGGCGAGATGGTGAAAGTCG	TCTTGCCCATTTGATGTTGCT	268
Opn4x	TGCTTTGTCAACAGCTTGCACAGAG	CAGCAATAATCTGTATGGTGCGCTTC	199

Opn4x Knockdown.

The shRNA sequence to suppress Opn4x (GAAGCTAACTGGAGTGATG) and a scrambled DNA target sequence (ACTCCATACTCGACACCTA) were created using the siRNA Wizard InvivoGen. cDNAs encoding shRNAs were inserted in a discistronic vector pSuper.neo+GFP (pSuperRNAi System-OligoEngine) under the control of the H1 RNAIII polymerase promoter. The transfection marker GFP was under the control of the PGK promoter. The resulting plasmids were referred to as Opn4x shRNA and ssRNA. The plasmids were mixed with Lipofectamine 2000 and added to the retinal cultures 12 h after plating.

Calcium Imaging by Fluo-4 AM Fluorescence Microscopy.

Cells were grown in an eight-well Lab-Tek recording chamber (NuncTM) in a colorless DMEM (GIBCO) containing 0.1% of Pluronic acid F-127 and 5 µM Fluo-4 AM (Invitrogen-Molecular Probes) Ca²⁺ indicator dye for 60 min at 37 °C. The fluorescence imaging technique was performed as described (7) with modifications by using Ca²⁺-sensitive indicator Fluo-4 AM excited at 515 nm (25.8 µW laser intensity) with a laser coupled to a confocal microscope (Olympus FluoView-300). Changes in fluorescence levels were quantified as the ratio between each relative intensity level measured after a light stimulus of 1,000 lx (F) and the mean of intensities of serial pictures before stimulation (Fo). ΔF was calculated as the difference between the maximal (peak) and minimal values of relative fluorescence in light responses for each assessment. See SI Materials and Methods for further information.

Retinoid Analysis.

Primary cultures were homogenized in phosphate buffer containing 200 mM hydroxylamine and processed for retinoid analysis by HPLC as reported (7). See SI Materials and Methods for further information.

³H-GABA Release.

HCs were incubated for 30 min at 37 °C with ³H-GABA (0.5–1 μCi) in 500 μL of buffer containing 140 mM NaCl, 5 mM KCl, 2.5 mM CaCl₂, 1 mM MgCl₂, 10 mM Hepes, and 10 mM glucose, adjusted to pH 7.4 with Tris base. The primary cultures were washed several times in fresh buffer to remove the excess ³H-GABA and exposed for 30 min to a 1,000-lx light stimulus. Supernatant was recovered, and radioactivity in the medium was determined in a scintillation counter. Fractional release was calculated as the ratio of radioactivity released/total radioactivity taken up by the culture.

SI Materials and Methods

Materials.

All reagents were of analytical grade. The secondary antibodies used for immunohistochemistry were DyLight 488-conjugated Donkey Anti-Mouse IgG and Dye Light Fluor 549 Donkey Anti-Rabbit IgG (dilution: 1:100; Jackson Immuno Research Laboratories). PROX1 Policlonal Antibody, Rabbit (dilution 1: 2,500, Millipore); Neurofilament of 200 kDa (NF-200) Mouse (1/10, Millipore); NeuN monoclonal antibody, Mouse (1/100, Millipore); DAPI, papain suspension in 0.05 M sodium acetate, and laminin were from Sigma Aldrich. Aqueous mounting medium (FluorSave) was from Calbiochem.

Immunocytochemistry.

Cultured cells were fixed for 30 min in 4% paraformaldehyde in phosphate buffered saline (PBS), and coverslips were washed in PBS and treated with blocking buffer (PBS supplemented with 0.1% BSA, 0.1% Tween 20, and 0.1% NaNO3 with the respective antibodies as described). They were then rinsed in PBS and incubated with goat anti-rabbit IgG AlexaFluor 488 or goat anti-mouse IgG Alexa Fluor546 (monoclonal antibodies) (1:1,000) for 1 h at room temperature. Coverslips were finally washed thoroughly and visualized by confocal microscopy (FV1000; Olympus).

Primary Cultures of Embryonic HCs.

HCs were purified from chicken neural retinas at E15 as previously reported (48). Briefly, eyes were dissected free from the head at the level of the ora serrata and sectioned in ice-cold Ca²⁺- Mg²⁺-free Hank’s buffered saline solution containing 25 mM glucose (CMF-HBSS). The retina was then cut into six to eight pieces and incubated with CMF containing 3 U/mL of papain for 25 min at 37 °C and then kept on ice until use. Cell suspension was subjected to a BSA discontinuous gradient of concentrations ranging from 1 to 4% and centrifuged at 300 × g for 15 min. Phase 2.5% was collected and poured into petri dishes pretreated with 2.5 µg/mL protein A, followed by incubation at 37 °C for 30 min with an anti-chicken Opn4x polyclonal antibody (Bio-Synthesis Inc.). After being washed exhaustively, identical aliquots of the remaining bound HCs were harvested in L15 containing B27 (Life Technologies, Invitrogen, GIBCO; dilution: 1/50 vol/vol) and seeded in petri dishes previously treated with 10 µg/mL polylysine and 5 µg /mL laminin. Primary cell cultures were incubated at 37 °C under constant 5% CO₂− (Life Technologies, Invitrogen, GIBCO; dilution: 1/500 vol/vol). The cell cultures were highly enriched in HCs according to specific HC markers as previously shown.

Flow Cytometry.

Retinas were washed in HBSS medium without Mg+ and Ca+ (HBSS−, Invitrogen) and incubated with 3 U/mL of papain for 25 min at 37 °C; this was followed by addition of DNaseI (Sigma) to 50 mg/mL and trituration to obtain single-cell suspensions. Three volumes of media were then added, and the tissue was triturated 5–10 times with a P1000 pipette (Rainin). Tissue was pelleted at 1,500 × g for 30 s, the supernatant removed, and cells were resuspended by flicking. This was done to obtain single-cell suspensions. Cells were fixed in 2% paraformaldehyde in PBS for 30 min and washed with HBSS+, followed by 0.1% Triton-X in HBSS+ incubation before staining with primary and secondary antibodies (PROX1 and Alexa fluor 488, respectively) with washes in between.

RNA Isolation and RT-PCR.

Total RNA from HC primary cultures was extracted using the TRIzol kit for RNA isolation (Invitrogen) (7, 48). RNA integrity was checked in 1.5% agarose gel and quantified by UV spectrophotometry (Gene Quant spectrophotometer, Amersham Biosciences). Finally, 1–2 µg of total RNA was treated with DNase (Promega) to eliminate contaminating genomic DNA. cDNA was synthesized with M-MLV (Promega) using oligo(dT).

Oligonucleotides were designed using the Vector NTI Advance 10 program to generate 100- to 400-bp products. Sequences were as shown in Table S1.

An initial denaturation step of 1 min at 94 °C, 25 cycles of 60 s at 94 °C, 50 s at 60°–65 °C, 90 s at 72 °C, and a final 5-min elongation step at 72 °C. Amplification products were separated by 2% agarose gel electrophoresis and visualized by ethidium bromide staining.

Opn4x Knockdown.

The shRNA sequence to suppress Opn4x (GAAGCTAACTGGAGTGATG) and a scrambled DNA target sequence (ACTCCATACTCGACACCTA) were created using the siRNA Wizard InvivoGen. cDNAs encoding shRNAs were inserted in a discistronic vector pSuper.neo+GFP (pSuperRNAi System-OligoEngine) under the control of the H1 RNAIII polymerase promoter, and the transfection marker GFP was under the control of the PGK promoter. The resulting plasmids were referred to as Opn4x shRNA and ssRNA. The plasmids were mixed with Lipofectamine 2000 and added to the retinal cultures 12 h after plating.

Calcium Imaging by Fluo-4 AM Fluorescence Microscopy.

Cells were grown in an eight-well Lab-Tek recording chamber (Nunc) in a colorless DMEM (GIBCO) containing 0.1% of Pluronic acid F-127 and 5 µM Fluo-4 AM (Invitrogen-Molecular Probes) Ca²⁺ indicator dye for 60 min at 37 °C. The fluorescence imaging technique was performed as described with modifications by using Ca²⁺-sensitive indicator Fluo-4 AM excited at 515 nm (25.8 µW laser intensity) with a laser coupled to a confocal microscope (Olympus FluoView-300). The emitted fluorescence was captured every 2 s, using a PlanApo N 60×Uplan SApo oil-immersion objective (N.A.: 1.42; Olympus). The 12-bit 4 × 4 binned fluorescence images for each photo were used to quantify fluorescence levels in the cells using the Fluoview 10.1 software; the mean fluorescence intensity in each cell was background-corrected by subtracting the mean fluorescence of an area with no cells. The mean intensity over a particular area of cells in a selected field was measured in each captured image series. Changes in fluorescence levels were quantified as the ratio between each relative intensity level measured after a light stimulus of 1,000 lx (F) and the mean of intensities of serial pictures before stimulation (Fo). Values of F/Fo are not linearly related to changes in [Ca+2]i but are intended to provide a qualitative indication of variations in [Ca+2]i. No significant vehicle effects or changes in focus were detected. Responses were considered significant when the ratio at the peak differed from the baseline levels by at least 20%. In addition, ΔF was calculated as the difference between the maximal (peak) and minimal values of relative fluorescence in light responses for each assessment.

Retinoid Analysis.

Primary cultures were homogenized in phosphate buffer containing 200 mM hydroxylamine. One milliliter of ethanol was added, and retinoids were extracted twice with 3 mL of hexane. The samples were centrifuged at 3,000 × g for 5 min. The organic phases were collected, dried under a stream of argon gas, and redissolved in 200 μL of hexane. Hexane solutions were analyzed by normal-phase HPLC using 0.5% dioxane for retinaldehyde-oxime, all-transretinol, and 0.1% all-transretinyl palmitate in hexane at a flow rate of 1.5 mL/min on a silica column (Agilent-Zorbax- Rx Sil 5 μm, 150 × 4.6 mm; retinaldehyde-oxime and all-trans-ROL/Zorbax-Sil 5 μm, 250 × 4.6 mm; all-trans retinyl palmitate) in an Agilent model 1260 liquid chromatograph equipped with a photodiode-array detector (Agilent Technologies). The identity of retinoids (including the syn- and anti-oximes of each retinaldehyde) was confirmed by on-line spectral analysis and coelution with authentic retinoid standards.

³H-GABA Release.

HCs were incubated for 30 min at 37 °C with ³H-GABA (0.5–1 μCi) in 500 μL of buffer containing 140 mM NaCl, 5 mM KCl, 2.5 mM CaCl₂, 1 mM MgCl₂, 10 mM Hepes, and 10 mM glucose, adjusted to pH 7.4 with Tris base. The primary cultures were washed several times in fresh buffer to remove the excess ³H-GABA and exposed for 30 min to 1,000 lx light stimulus. After this, supernatant was recovered and radioactivity in the medium was determined in a scintillation counter. Fractional release was calculated as the ratio of radioactivity released/total radioactivity taken up by the culture.