Pannexin 1 sustains the electrophysiological responsiveness of retinal ganglion cells

Abstract

Pannexin 1 (Panx1) forms ATP-permeable membrane channels that play a key role in purinergic signaling in the nervous system in both normal and pathological conditions. In the retina, particularly high levels of Panx1 are found in retinal ganglion cells (RGCs), but the normal physiological function in these cells remains unclear. In this study, we used patch clamp recordings in the intact inner retina to show that evoked currents characteristic of Panx1 channel activity were detected only in RGCs, particularly in the OFF-type cells. The analysis of pattern electroretinogram (PERG) recordings indicated that Panx1 contributes to the electrical output of the retina. Consistently, PERG amplitudes were significantly impaired in the eyes with targeted ablation of the Panx1 gene in RGCs. Under ocular hypertension and ischemic conditions, however, high Panx1 activity permeated cell membranes and facilitated the selective loss of RGCs or stably transfected Neuro2A cells. Our results show that high expression of the Panx1 channel in RGCs is essential for visual function in the inner retina but makes these cells highly sensitive to mechanical and ischemic stresses. These findings are relevant to the pathophysiology of retinal disorders induced by increased intraocular pressure, such as glaucoma.

Figure 1

RGCs have the highest levels of Panx1 expression in the retina. (A) Real -time PCR in purified primary cells shows significant enrichment of Panx1 in RGC (red bar) vs. whole retina (green bar) and Muller glia (Muller GL, blue bar), *P ≤ 0.05: n = 5, Student’s t-test; (B) Representative micrographs of in situ RNA hybridization of Panx1 transcripts (red puncta indicated by arrows on the insert) using RNAscope technique. Insert (zoom, right panel) shows Panx1 transcripts in magnified ganglion cell layer (GCL) region, where RGCs are located; nuclei labeling: DAPI (blue); Scale bar, 25 µm. (C) Representative micrographs of immunostaining in retina sections: the highest level of Panx1 labeling (red) in the GCL co-localized with Brn3a-positive RGCs (green), as indicated by asterisks. The lower panel shows control staining in Panx1 knockout tissue. Scale bar, 25 µm. (D) Representative retinal flat-mounts co-immunostained for Panx1, and RGCs markers TUJ1 (magenta), and Brn3A (green). The level of Panx1 labeling (red) varied significantly among RGCs with higher than average levels detected in about one-third of all Brn3A-positive neurons (asterisks). The remaining RGCs (Brn3a-positive, no asterisks) showed significantly lower levels of the Panx1 protein. Scale bar, 25 µm.

Figure 2

Voltage patch clamp recordings of Panx1 currents from inner retinal neurons. Activation of Panx1 currents in individually patched neurons in the GCL was triggered by applying depolarizing voltage steps (depicted in detail in Fig. 6C). (A) Representative micrographs of morphometric cell type identification, reconstructed from confocal z-stack datasets of Alexa-568 labeled RGCs. Neuron phenotyping was performed by the presence of an axon (arrows, left panels). Scale bar, 25 µm. (B) Representative current responses to the applied pulse protocol. Cells were treated with a Panx1 agonist cocktail (Agnst, 3 mM ATP, 20 mM KCl) or antagonist (Pbcd, 300 µM probenecid. (C) Representative traces of current-voltage relationship in single cell recordings. Experimental data points were obtained by plotting the current amplitude elicited by a given voltage step under the control condition, after application of the agonist or antagonist. (D) Quantitation graphs showing maximum currents (nA) evoked by a +80 mV voltage step in control and experimental conditions for each of the two RGC and one AC subpopulations with statistical analysis. Data are presented as means ± SE; n_{responsive WT RGCs} = 11; n_{WT ACs} = 7; n_{Panx1 KO RGCs} = 6. **P < 0.01; *P < 0.05; significance: Kruskal-Wallis test and Dunn’s Multiple Comparison test. Color coding: control bath solution, black; agonists, red; antagonist, green.

Figure 3

Panx1 channel activity in three functional subtypes of RGCs. (A) RGCs morphological types identification, performed by 3D confocal reconstruction of cells labeled with Alexa-568 dye; RGC types were defined by stratification of their dendritic tree in ON and/or OFF sublamina of the inner plexiform layer. Scale bar, 25 µm. (B) Functional typing of RGCs by spiking activity in response to the light stimulus. (C) Pannexin1 channel-activated currents in whole-cell mode in response to the depolarizing voltage steps from −80 to 60 mV with holding potential of −60 mV in control solution (black), with Panx1 agonist cocktail (Agnst, 20 mM K⁺ 0.1 mM ATP, red) or Pbcd (300 μM, green) conditions. (D) Representative I/V relationship for ON (n = 20), OFF (n = 20) and ON/OFF (n = 11) RGCs. (E) Mean values for the Panx1 channel currents; significance: Kruskal-Wallis test and Dunn’s Multiple Comparison test *P < 0.05, ns, non-significant color coding: control bath solution, black; agonists, red; antagonist, green.

Figure 4

Panx1 deficiency results in changes in RGC function. PERG responses were compared between C57Bl/6 wild-type (WT) and Panx1^−/− mice as described in Materials and Methods. (A) Box graphs of PERG responses in germline Panx1^−/− (Px1 KO). Left panel: PERG amplitude; right panel: PERG latency. n_WT = 12; n_KO = 12. (B) PERG response in RGCs with conditional Panx1 gene ablation. The recordings were performed 8 weeks after infection of the AAV2-GFP-Cre construct via intravitreal injection; n = 5. (C) Averaged waveforms of responses recorded from WT (white boxes), germline Panx1^−/− (Px1 KO, red), AAV2-EGFP-injected Panx1^fl/fl controls (AAV2-GFP, light blue) and conditional AAV2-Cre-EGFP –induced Panx1^RGC−/− (AAV2-Cre, dark blue) eyes. ***P < 0.001; **P < 0.01; significance: Mann Whitney test.

Figure 5

Panx1 inactivation protects RGCs in ischemia-reperfusion. The rate of survival following ischemia-reperfusion injury in the retinas with germline (Px1 KO, red, n = 11) and neuron-specific (Px1-CKO, green, n = 5) Panx1 ablation, WT C57Bl/6 control retinas (WT, black, n = 10) and wild type retinas from the probenecid-treated mice (WT + Pbcd, blue, n = 5). The data show percentage of NeuN-positive cells left after treatment relative to control. Data show values determined for individual animals; the bars represent means ± SD; **P < 0.01; *P < 0.05; significance: one-way ANOVA and Tukey test for multiple comparisons.

Figure 6

Panx1 forms functional channels in stably transfected N2a clones. (A) mRNA in situ hybridization analysis in stable N2a cell lines with (Px1-C1) and without (GFP-B2) Panx1 expression. The cells were labeled with the Panx1-specific fluorescent probe (red) to visualize transcripts and DAPI (blue) to visualize nuclei. Scale bar, 25 µm. (B) Western blot analysis of Panx1 protein expression levels in cellular extracts from three stable N2a clones, expressing either EGFP alone (GFP-B2) or both Panx1 and EGFP (Px1-C1 and Px1-C3). Loading: 3 or 10 µg of total protein per lane, as indicated. Panx1 (Px1) isoforms formed characteristic multiple bands 43–47 kDa, as indicated by the arrows. (C) Preconditioning (depolarizing) voltage ramp protocol to stimulate Panx1 channel activity, applied during electrophysiologial recordings. (D) I/V relation from N2a-Panx1-C1 (Px1-C1; n = 7) with and without 1.0 mM probenecid (Pbcd; n = 7) application; N2a-EGFP-B2 (GFP-B2; n = 8) expressing cells. (E) Data analyses and quantitation of evoked membrane currents response elicited at +100 mV from D. (F) Quantitation of evoked membrane currents response elicited at +100 mV from N2a-Panx1-C1 cells (n = 7) treated with 50 μM ATP (n = 7) and 300 μM Pbcd (n = 5); I/V relation graph is not shown. Data were normalized to N2a-Panx1-C1 membrane currents responses at +100 mV, set to 100%. Each bar represents mean ± SEM; **P < 0.01; *P < 0.05; significance: one-way ANOVA and Tukey test for multiple comparisons.

Figure 7

Overexpression of Panx1 increased N2a plasma membrane permeability to small molecules. (A) Ethidium bromide (EtBr) uptake by stably transfected N2a clones after 1 h incubation with and without IAA94 agonist. Cells were imaged by fluorescent microscopy at low (10x) magnification. (B–E) Time courses of EtBr uptake with or without pharmacological treatments. (B) Time course without any treatments (baseline). (C) Time course for EtBr uptake in the presence of 1.5 mM ATP. (D) Time course for EtBr uptake in the presence of 200 µM IAA94. (E) Dose-response curve for IAA94 in N2a-Panx1-C1 cells. Measurements were taken 30 min after application of IAA94; the basal value of EtBr uptake in the absence of IAA94 was subtracted from the recorded values; n = 9 per condition/time point.

Figure 8

Overexpression of Panx1 increases susceptibility of N2a cells to OGD injury. Plasma membrane permeability to extracellular EtBr (uptake) and intercellular EGFP (release) in stably transfected N2a lines was quantified under the following conditons. (A) Normoxia for 2 h with or without 1.5 mM ATP or 0.2 mM IAA94. (B) Normoxia or OGD for 1.5 h. EGFP release under OGD conditions indicate cells death by membrane breakdown. (C) Time course of EGFP release under OGD conditions. (D) EGFP release under the OGD conditions with or without apyrase or 1.5 mM ATP. Values are percentage of the maximum value observed after treatment with 0.03% saponin. **Time points with statistically significant (P < 0.01) increase in EGFP release by Panx1-C1 cells in the presence of ATP; significance calculated using one-way ANOVA and Tukey test for multiple comparisons. Error bars for most values are smaller than symbols, n = 9 per condition/time point.