Extraretinal Spike Normalization in Retinal Ganglion Cell Axons

Abstract

Spike conduction velocity characteristically differs between myelinated and unmyelinated axons. Here we test whether spikes of myelinated and unmyelinated paths differ in other respects by measuring rat retinal ganglion cell (RGC) spike duration in the intraretinal, unmyelinated nerve fiber layer and the extraretinal, myelinated optic nerve and optic chiasm. We find that rapid spike firing and illumination broaden spikes in intraretinal axons but not in extraretinal axons. RGC axons thus initiate spikes intraretinally and normalize spike duration extraretinally. Additionally, we analyze spikes that were recorded in a previous study of rhesus macaque retinogeniculate transmission and find that rapid spike firing does not broaden spikes in optic tract. The spike normalization we find reduces the number of spike properties that can change during RGC light responses. However, this is not because identical spikes fire in all axons. Instead, our recordings show that different subtypes of RGC generate axonal spikes of different durations and that the differences resemble spike duration increases that alter neurotransmitter release from other neurons. Moreover, previous studies have shown that RGC spikes of shorter duration can fire at higher maximum frequencies. These properties should facilitate signal transfer by different mechanisms at RGC synapses onto subcortical target neurons.

Keywords: axonal spikes; cell type-specific differences; normalization of duration; optic chiasm; optic tract; retina.

Figure 1.

Duration, width, and shape of RGC spikes. A, B, Mean of cardinal spikes elicited by injection of pseudorandomly fluctuating current and recorded with whole-cell patch-clamp electrodes. A, Top and bottom traces show V_m and second-order time derivative of V_m (d²V_m/dt²), respectively, of an RGC soma versus time. Duration measured from c1 (base of depolarizing phase) to c2 (end of repolarizing phase). Red line plots d²V_m/dt². Green dashed lines before and after spike peak show Gaussian fits used to calculate c1 and c2, respectively, and are plotted over the red line. B, Top and bottom traces show V_m and dV_m/dt, respectively, versus time. Different RGC soma than in A. Points labeled inf, p1, zc, and p2 mark moments when spike rises, depolarizes most quickly, reaches the peak of depolarization, and hyperpolarizes most quickly, respectively. Each time point calculated as explained in Materials and Methods. C, Spike widths, and parts thereof, calculated from inf–p2 (red dots), p1–p2 (blue dots), and inf–zc (green dots), and plotted versus spike duration (c1–c2). Each dot is the mean of the widths of all spikes recorded from a single soma. Lines are linear regressions. The y-intercepts range between −20 and 20 μs. D, E, Spikes recorded on MEA and identified as somatic by biphasic waveform (D) and as axonal by triphasic waveform (E). F, Spike recorded by tungsten-in-glass microelectrode from optic chiasm. Each trace (D–F) is the mean of all spikes recorded from each unit. G–I, Best fittings of sum-of-three-Gaussians function to spike waveforms. Examples of fittings that are rejected (G, red) versus accepted (H, light blue; I, green) as fits to unaveraged, MEA-recorded axonal C spikes (dark blue). The error of fit in each panel was calculated as described in Materials and Methods, and accepted if it was <0.500. Fits to the second outward peak of some spikes were interrupted by follower spikes (e.g., H).

Figure 2.

Activity-induced broadening of intraretinal, but not extraretinal, F1 spikes. A, E, I, Averages of C spikes (green) and of F1 spikes (brown) recorded from an RGC soma (A), nerve fiber layer axon (E), and optic chiasm axon (I), and analyzed as in Figure 1D–F, respectively. A, E, I, Mean spikes aligned horizontally to p1 peaks. B, F, J, Expanded views of p1–p2 from A, E, and I, respectively. Lines in I and J are dashed to show the overlap of C and F1 spikes. C, G, K, Scatter plot of F1 spike width versus C spike width for all intraretinal somata (C; n = 166), all intraretinal axons (G; n = 201), and all optic chiasm axons (K; n = 38) recorded from. Unity line plots where C and F spike durations would match, and insets are histograms of the percentage increase in F1 spike duration. The dashed vertical line in each inset corresponds to the unity line in the scatter plot; line height corresponds to 70% of the total count of spiking units at each recording locus. D, H, L, Bar plots (mean ± SEM) compare C (green) and F1 (brown) spike widths plotted in C, G, and K, respectively, and were measured by the indicated time points. Asterisks are centered above the mean widths that differ by statistically significant amounts.

Figure 3.

Activity-induced broadening of electrically elicited nerve fiber layer spikes. Stimulating and recording MEA electrodes separated by ≥400 μm. A, C (green) and F1 (brown) spikes elicited by current pulses separated by 8 ms. Stimulus artifact (blue) after suppressing spikes by repeated stimulation. B, Mean ± SEM of differences between C and F1 spike widths as a function of interstimulus interval. Spike widths were calculated as the time between p1 and p2 (e.g., dots on spike waveforms in A after the stimulus artifact has faded). Asterisks positioned above statistically significant differences. C, F1 spike widths as a function of C spike widths at interstimulus intervals ranging from 3 to 500 ms. Each data point plots the mean C and F1 spike widths of an individual axon.

Figure 4.

Spike widths (inf–p2) versus the timing of preceding and following spikes. Points and error bars plot the mean ± SEM of ordinate values measured in nerve fiber layer (gray; n = 56) and optic chiasm (black; n = 26) axons and were pooled over the indicated time bins. A, C spike widths as a function of the spike-free period preceding each C spike. B, Difference between C and F1 spike widths as a function of the interspike interval. Asterisks are adjacent to statistically significant differences.

Figure 5.

Lack of activity-induced broadening of optic tract spikes. A, Averages of C spikes (green) and of F1 spikes (brown) recorded from an optic tract axon, plotted by dashed lines and aligned horizontally to their peaks. The traces are indistinguishable in amplitude and time course. B, C (green) and F1 (brown) spike widths (mean ± SEM) measured by indicated time points for each optic tract axon and then averaged across all axons recorded from (n = 23). C, C and F1 spike widths for each optic tract axon recorded from. In several instances, dots plotting C and F1 values from multiple fibers overlapped. If only two data points overlapped, one was displaced by ±0.01 ms vertically and laterally. In the few cases where three data points overlapped, points were plotted at (x, y), (x + 0.01, y + 0.01), and (x − 0.01, y − 0.01). These displacements prevented the data from being moved relative to unity line and allowed the figure to show data from all fibers. D, The difference (mean ± SEM) between axonal C and F1 spike widths as a function of interspike interval. None of the mean differences were statistically significant.

Figure 6.

Light-induced broadening of intraretinal axonal C spikes. A, Spikes of transient On cell, identified by low spike frequency in darkness (e.g., single spike at t ∼ −300 ms in A) and brief increases in spiking following light onset (top trace, single unaveraged sweep). Gray and white background behind spikes show timing of dark and light, respectively. B, C, Average of spikes recorded before and after light offset (B) and before and after light onset (C), per color code. D, Spike widths calculated from indicated time points and averaged across all axons recorded from in dark versus light (mean ± SEM; n = 435 axons for inf–zc; n = 212 axons for p1–p2 and inf–p2). E, Mean spike inf–p2 width in light versus dark, where each dot plots values of a different axon. Unity line plots where C spike durations in light would match those in dark. Inset, Euclidean distances from unity line of points in E. F, Mean ± SEM of normalized spike widths before and after onset (blue) and offset (red) of light (n = 136 axons). Spike durations pooled over 500 ms time bins. G–J, Normalized spike width as function of time for On (G), Off (H), sustained (I), and transient (J) cells. Spike widths pooled over 250 ms time bins. F–J, Spike width of each axon during each time bin normalized to the mean of all C spike widths of that axon in the dark. Lines connect values in dark or in light, but not at transition between dark and light. Asterisks are next to statistically significant differences between dark and light values.

Figure 7.

Light-induced broadening of intraretinal axonal C spikes. A, Spikes of Off axons identified by the increase in spiking following light offset (top trace, continuous recording, single unaveraged sweep). B–E, Spikes recorded on MEA, identified as axonal by triphasic waveform. Format is as shown in Figure 6. B, C, Plots of mean of spike waveforms extracted by PCA from one axon before and after light offset (B) and before and after light onset (C), per color code. D, E, Plot mean of spike waveforms extracted by PCA from another axon for same transitions between dark and light, and formatted as in B and C.

Figure 8.

Absence of light-induced broadening in optic chiasm C spikes. A, Spikes of transient Off cell identified by spike burst after light offset (top trace, single unaveraged sweep, formatted as in Fig. 6). B, C, Averages of spike waveforms recorded before and after light offset (B) and before and after light onset (C). D, Spike widths averaged across all fibers recorded from. E, Single-fiber C spike widths in light versus dark. Scatter plot and inset constructed as in Figure 6. F, Change in mean C spike width as a function of time before and after light onset (blue) and offset (red). Values pooled across all optic chiasm axons recorded from (n = 19 for inf–zc, p1–p2, and inf–p2) and formatted as in Figure 6. Spike widths pooled over 250 ms time bins.

Figure 9.

C spike widths of RGC axons distinguished by light response. Mean spike waveform of each cell type calculated by aligning all C spikes of a given axon to their p1 time point and averaging them; repeating this for the C spikes of all axons of a given cell type; and then calculating the mean of all the averaged waveforms. A–D, Each panel superimposes the portion of the mean spike waveforms compared, from just before p1 until just after p2. These are color coded as indicated and aligned to their p1 time points (as in Fig. 2). E–H, Widths of all C spikes recorded from each axon calculated and averaged from time points listed below each pair of bars. Bars (color coded as indicated) plot the mean ± SEM of averages from all axons of On and Off units in nerve fiber layer (E) and optic chiasm (G), and of transient and sustained units in nerve fiber layer (F), and optic chiasm (H). Sample sizes are 85 On, 80 Off, 44 transient, and 110 sustained axons in nerve fiber layer, and 19 On, 19 Off, 15 transient, and 22 sustained axons in optic chiasm. Asterisks are above statistically significant differences.