Sustained Responses in the OFF Channel Driven by Crossover Signals from the ON Channel

(A) View of the inner plexiform layer showing synaptic terminals of bipolar cells expressing SyGCaMP2. Yellow traces indicate the layers in the IPL. Field of view is 100 micrometers across.

(B) Spatial distribution of contrast-activated and contrast-suppressed ON bipolar terminals as a function of layer. The depth of the terminal in the IPL was measured from the photoreceptor side (layer 1) to ganglion cells (layer 6). Contrast-activated ON bipolar terminals showed the highest density in layer 5 and 6, whereas contrast-suppressed cells were mostly localized in layer 3 and 6.

(C) Spatial distribution of each OFF group as a function of layer. OFF terminals in Group 1 (low-pass) were at highest density in layer 6, whereas terminals in Group 3 (band-pass) were predominantly localized in layer 1. OFF bipolar terminals in Group 2 stratified throughout IPL with the highest density in layer 6.

(D) Histogram showing the distribution of cutoff frequencies (fc) in a population of 445 OFF terminals in 5 fish. Light transmission through ON pathway was inhibited by an intraocular injection of the mGluR6 agonist L-AP4 (100 µM estimated final concentration). Control is shown in black and L-AP4 in red. Note that OFF bipolar terminals in Group 1 (low-pass) are almost absent in presence of L-AP4.

(E) Plot of response amplitude as a function of frequency averaged across all OFF terminals, before (black trace) and after (red trace) L-AP4. Dashed lines represent the average cutoff frequency value (fc). Note that blocking signals through the ON pathway decreased the amplitude of responses in the OFF pathway across all range of frequencies.

See also Figure S4.

Crossover Inhibition Converts Band-Pass Terminals to Low-Pass

(A) A field of view showing the same population of bipolar cell terminals before and after the injection of L-AP4 into the eye of a zebrafish.

(B) Example of frequency tuning curves from three individual terminals from Group 1 (ROIs 1, 2, and 3 in A) and three from Group 2 (ROIs 4, 5, and 6) before and after L-AP4.

(C) Summary of the cutoff frequency values from all the OFF bipolar terminals in each group before and after L-AP4 (n = 34 terminals from 1 fish). Groups in control conditions were determined by K-means clustering (see Experimental Procedures). The cutoff frequency from the individual terminals was calculated as in Figure 1. Solid lines connect responses from the same terminals before and after L-AP4. Red dots represent mean ± SEM. p < 0.001.

Voxel-Based Analysis of Calcium Signal in Amacrine Cells Reveals Diversity in Temporal Tuning

(A) Left: view of the IPL showing amacrine cells expressing SyGCaMP3 (left) and the respective pixel-mask (right). Scale bar represents 20 µm. Right: raster plot showing the relative change in fluorescence for 6,210 pixels during a “forward” frequency seep. Only OFF voxels are shown, as defined by the responses to steps of light.

(B and C) K-means clustering revealed two major types of temporal tuning in amacrine cells, “low” band-pass (B, peak transmission at 4.6 ± 0.2 Hz and fc = 9.8 ± 0.11 Hz) and “high” band-pass (C, peak transmission at 9.9 ± 0.3 Hz and fc = 13.9 ± 0.2 Hz). Voxels were then separated further into ON (green), OFF (red), and ON-OFF (blue). Results were collected from five fish. The left-hand plots show averaged SyGCaMP3 responses of the three groups classified as low and high band-pass from a total of between 5,207 and 8,540 voxels from five fish. The right-hand plots show response amplitude as a function of frequency.

(D) Spatial distribution within the IPL of low band-pass (filled gray regions) and high band-pass (solid lines) voxels as a function of dendrite stratification in the IPL for ON (left), OFF (middle), and ON-OFF (right) pixels. Stratification is plotted such that 0% is the boundary with the ganglion cell layer and 100% the boundary with the inner nuclear layer.