Figure 1.
Correlated Spikes of Retinal GCs Occur on Varied Timescales.
(A) Panel shows exemplary spontaneous spikes of a GC pair recorded simultaneously. (B) Cross-correlation functions (CCFs) of spontaneous spikes were fit with a Gaussian function from which values for amplitude and width were computed. (C) CCF with a bimodal structure consisting of two peaks separated by a central trough at time zero (inset). Note that the narrow bimodal component is superimposed on a somewhat broader profile. (D and E) CCFs displaying unimodal peaks with different Gaussian widths (W), reflecting differences in the temporal precision of unimodal spike correlations. (F) Scatterplot demonstrating an inverse relationship between the Gaussian amplitudes for unimodal CCFs and the inter-somatic distance of the recorded GC pairs. The dashed line indicates a linear regression fit of the data.
Figure 2.
Blockade of Chemical Synaptic Activity Abolishes Broad Unimodal CCFs, but Not Medium Unimodal or Bimodal CCFs.
(A) A cocktail of neurotransmitter antagonists effectively eliminate the light-evoked responses of both ON (bottom) and OFF (top) GCs. Presentation of the light stimulus (I = 3000 R*/rod/sec) is indicated by the grey bar. (B and C) Application of the cocktail does not significantly alter spike correlations reflected in bimodal CCFs or unimodal CCFs with relatively small (medium) widths. (D) In contrast, spike correlations reflected in unimodal CCFs with relatively broad widths are largely eliminated when chemical synaptic activity is blocked. (E) Example of a unimodal CCF in which application of the cocktail of antagonists abolished the broad profile revealing a surviving narrower component of medium width. (F) Summary of the effects of cocktail on spike correlations reflected in unimodal CCFs. The correlations indicated by broad CCFs are readily blocked by cocktail treatment, whereas those reflected by more narrow unimodal CCFs are largely retained.
Figure 3.
Blockade of Gap Junctions Eliminates All Types of GC Spike Correlations.
(A) Application of 18β-GA (25 µM) effectively blocks the Neurobiotin tracer coupling of GCs. Micrographs show elimination of tracer coupling of an ON G6 cell, which shows coupling to ACs under control conditions, and an OFF G3 cell, which shows coupling to GC and AC neighbors under control conditions. Scale bar = 100 µm. (B) The blockade of gap junctions with 18β-GA does not significantly alter the full-field light responses of ON (bottom) and OFF (top) GCs. Presentation of the light stimulus (I = 3000 R*/rod/sec) is indicated by the grey bar. (C,D, and E) Blockade of gap junctions effectively eliminated all spike correlations reflected in narrow, medium, and broad CCFs.
Figure 4.
Broad Spike Correlations are Absent in the Cx36 KO Mouse Retina.
(A–D) GC pairs in the Cx36 KO mouse retina displayed either narrow bimodal (A) or medium (C) unimodal spike correlations. Both narrow bimodal and medium spike correlations were abolished following application of 25 µM 18β-GA (B and D). (E) Bar graph summarizing changes in the relative frequency of unimodal spike correlations when Cx36-expressing gap junctions are ablated in the KO mouse retina. (F–H) Light adaptation of the WT mouse retina with a photopic (I = 3000 R*/rod/sec) background stimulus does not alter narrow (F) or medium (G) spike correlations, but eliminates the broad (H) correlations.
Figure 5.
Neighboring OFF α-GCs Exhibit Both Narrow and Medium Spike Correlations.
(A) Tracer coupling pattern revealed after iontophoresis of Neurobiotin into a single OFF α-GC (asterisk). The OFF α-GC is coupled to an array of neighboring α-GCs (white arrows) and a cohort of nearby ACs (white arrowheads). Scale bar = 100 µm. (B and C) The CCF generated for the spontaneous spike activity of a pair of OFF α-GCs exhibits a characteristic superposition of two components: a unimodal component with medium Gaussian width (B) upon which a narrow, bimodal component (C) is superimposed. (D) Neither narrow nor medium spike correlations of OFF α-GC pairs were eliminated by the blockade of chemical synaptic activity.
Figure 6.
OFF α-GCs Express Cx36 In AC-To-GC, but Not GC-To-GC, Gap Junctions.
(A) Example of the morphology of a single Neurobiotin injected GFP-positive GC (red) in the CB2-GFP retina corresponding to that of an OFF α-GC (41). Scale bar = 50 µm. (B) Relative gene expression of Cx36, Cx57, Cx32, Cx37, GABA receptor α2 subunit (GABAR- α2) and tyrosine hydroxylase (TH) in CB2+ OFF α-GCs. (C) A Neurobiotin-injected OFF α-GC in the WT mouse retina displaying the characteristic tracer coupling to neighboring GCs and ACs. Scale bar = 50 µm. (D) Single optical section (thickness = 0.5 µm) of a confocal image showing the colocalization of Cx36 immunolabeled puncta (green) with dendritic crossings of Neurobiotin-injected OFF α-GC (blue) and tracer coupled ACs (red). Scale bar = 10 µm. (E) No significant colocalization of Cx36 puncta (green) are found at dendritic crossings of Neurobiotin coupled neighboring OFF α-GC dendrites (red).
Figure 7.
GC-to-GC Coupling Underlies Narrow Spike Synchrony, Whereas AC-To-GC Coupling Mediates the Medium Spike Correlations.
(A) Video image of a PTE recording from a pair of OFF α-GCs in the living mouse retina. Scale bar = 50 µm. (B) Schematic of the image in panel A showing how the mosaic of somata in the GCL was outlined to determine the identity of the GCs recorded in pair wise recordings in which one cell was injected with Neurobiotin. (C) Neurobiotin-injected OFF α-GC in the Cx36 KO mouse retina shows only homologous coupling to OFF α-GC (arrows). Scale bar = 100. (D) Spontaneous spiking of a pair of OFF α-GCs in the Cx36 KO mouse retina produces a CCF with only a narrow, bimodal component. Red line indicates 99% confidence level. (E) A Neurobiotin-injected G17 GC (asterisk) shows coupling to neighbor G17 GCs (arrow). Scale bar = 100 µm. (F) Spike correlations of G17 GC pairs are reflected by narrow, bimodal CCF components. bimodal CCFs. (G) Neurobiotin-injected G6 cell (asterisk) shows the characteristic tracer coupling to neighboring ACs (arrows). Scale bar = 150 µm. (H) Correlated spikes of a pair of G6 cell neighbors are reflected by a CCF with a unimodal profile with medium Gaussian width. (I) Neurobiotin-injected G1 cell (asterisk) in the WT mouse retina shows characteristic tracer coupling to multiple cohorts of neighboring ACs (arrows). Scale bar = 150 µm. (J) Correlated spikes of a pair of G1 cell neighbors in the WT are reflected by a CCF with a unimodal profile with medium Gaussian width. (K) Neurobiotin-injected G7 cell (asterisk) in the WT mouse retina shows characteristic tracer coupling to neighboring GCs (arrowheads) and ACs (arrows) (left). Scale bar = 150 µm. (L) Correlated spikes of a pair of G7 cell neighbors in the WT are reflected by a CCF with medium unimodal profile. (M) Neurobiotin-injected G7 cell (asterisk) in the Cx36 KO retina shows no tracer-coupling. Scale bar = 150 µm. (N) No significant spike correlations between G7 cell neighbors in the Cx36 KO retina are reflected by a CCF with a flat profile.
Table 1.
Coupling Pattern and Spike Correlations of GC Subtypes.