Arousal Modulates Retinal Output

Abstract

At various stages of the visual system, visual responses are modulated by arousal. Here, we find that in mice this modulation operates as early as in the first synapse from the retina and even in retinal axons. To measure retinal activity in the awake, intact brain, we imaged the synaptic boutons of retinal axons in the superior colliculus. Their activity depended not only on vision but also on running speed and pupil size, regardless of retinal illumination. Arousal typically reduced their visual responses and selectivity for direction and orientation. Recordings from retinal axons in the optic tract revealed that arousal modulates the firing of some retinal ganglion cells. Arousal had similar effects postsynaptically in colliculus neurons, independent of activity in the other main source of visual inputs to the colliculus, the primary visual cortex. These results indicate that arousal modulates activity at every stage of the mouse visual system.

Keywords: arousal; locomotion; retina; superior colliculus; vision.

Figure 1

Visual Responses of Retinal Boutons Imaged in SC (A) SyGCaMP6f is injected into one eye. (B) Confocal images showing expression of SyGCaMP6f in synaptic boutons in the contralateral SC. (C) Mice were head-fixed on a treadmill surrounded by three monitors. (D) Positioning of the implant (circle) over superior colliculus (SC; purple), view through the implant showing SC and inferior colliculus (IC), and field of view of a typical two-photon imaging session of retinal boutons (rectangle). (E) Average frame of one two-photon imaging session (E1). Planes were imaged with 2-μm spacing to track boutons even during brain movements (E2). (F) Receptive fields of six boutons (i–vi), mapped with sparse sequences of white and black squares. Positions are in visual degrees relative to the front of the mouse and the height of the eyes. Ellipses outline the receptive fields at half height. (G) Distribution of ON/OFF indices across all boutons. For ease of description, we defined boutons as “ON” if their ON/OFF index was >0.5, “OFF” if it was <−0.5, and “ON+OFF” if it was intermediate. Triangles show ON/OFF indices of examples in (F). (H) Average (gray) and fitted (black) calcium responses of three boutons (1–3) in response to four sinusoidal gratings drifting in four directions. Gray shades show the times of stimulus presentation. (I) Direction tuning (mean ± SEM) and fitted tuning curves (solid lines) of the three boutons in (H) in trials where the pupil was small. (J) Distribution of maximum amplitudes in response to gratings across boutons. Boutons that are suppressed by gratings have negative maxima. Black bars represent boutons that are tuned to direction; white bars represent boutons that are not tuned. Triangles show maximum responses of examples in (H) and (I). (K) Orientation and direction selectivity indices of boutons that are selective only for orientation (light green), only for direction (blue), or for both orientation and direction (dark green). Circles show examples 1 and 2 in (H) and (I). (L) Distribution of preferred directions of boutons that are direction selective (gray bars) or only orientation selective (white bars). Fourth harmonic (light gray line) was fit using Fourier decomposition. Triangles show examples 1 and 2 in (H) and (I).

Figure 2

Activity in Retinal Boutons Varies with Arousal (A) Data from one experimental session recorded simultaneously during complete darkness showing running speed (yellow), calcium traces of retinal boutons (gray scale), and traces of two boutons (red and blue) that have positive and negative correlation with running. Calcium traces in gray scale were each Z scored and then sorted by mean correlation with running speed during darkness and during presentation of gratings (in B). (B) Data recorded from same retinal boutons as in (A) during presentation of gratings. In addition to (A), pupil size (green) and stimulus onsets (tick marks) are shown. (C) Correlation of retinal bouton calcium traces with running speed in darkness. Gray shade represents 2.5th to 97.5th percentile interval for null distribution generated by randomly shifting running trace against calcium traces. Dots show correlation strengths of example neurons in (A) and (B). (D) As in (C), but during presentation of gratings. (E) Correlation with running speed in darkness versus during visual stimulation for each bouton. Histograms show marginal distributions (same data as in C and D) for boutons with significant (black bars) and nonsignificant (white bars) correlations. (F) Correlation with pupil size during presentation of gratings. (G) Direction tuning (mean ± SEM) of four boutons during small (black) or large (red) pupil. Solid lines represent fitted tuning curves. Dotted lines represent 0 ΔF/F. Examples 1–3 are the same as in Figures 1H and 1I. (H) Distribution of arousal modulation for responses to gratings during small versus large pupil. Gray shade shows the 2.5th to 97.5th percentile interval of null distribution. Dots show modulations of examples in G (filled dots have significant modulations). Triangles (top) show mean values for boutons with significant positive or negative response modulations. (I) Direction selectivity index (DSI) during small versus large pupil for each bouton. Red line represents linear regression (linear mixed-effects model without intercept). Color of dots represents density in scatterplot.

Figure 3

Effect of Arousal Is Present in Firing Rates of Retinal Ganglion Cells (A) ON and OFF receptive fields of example retinal axons 1 (A1) and 2 (A2). (B) Spike waveforms of axon 1 (B1) and axon 2 (B2) on multiple neighboring channels of probe when the animal was running (red) or stationary (black) (recorded in darkness). (C) Traces of running speed (top) and firing rate (bottom) of axon 1 (C1) and axon 2 (C2) recorded in darkness. Grey shades represent periods of running (≥1 cm/s). (D) Cross-correlograms between firing rate and running speed for axon 1 (D1) and axon 2 (D2). A positive lag denotes that firing rate is lagging running speed. Grey shade shows the 2.5th to 97.5th percentile interval of time-shifted data. (E) Correlation strengths (measured at lag zero) with running speed in darkness versus the p value of correlation.

Figure 4

Visual Responses and Effect of Arousal on Neurons in SC (A) Average frame of two-photon imaging data shown in (H). (B) Distribution of ON/OFF indices across SC neurons. (C) Distribution of maximum amplitudes in response to gratings for excitatory (C1) and inhibitory neurons (C2). Dark bars represent neurons tuned to direction; light bars represent untuned neurons. Triangles show responses of examples in F. (D) Distribution of preferred directions of neurons that are direction selective (gray bars) or only orientation selective (white bars). Fourth harmonic (light gray line) was fit using Fourier decomposition. Triangles show preferred directions of examples 1–3 in (F). (E) Data from one experimental session recorded during presentation of gratings. Calcium traces in gray scale were each Z scored and then sorted by correlation with pupil size in the first half of data; only the second half is presented to show the robustness of correlations. (F) Direction tuning (mean ± SEM) of four neurons during trials where the pupil was small (black) or large (red). (G) Correlations of SC neurons with pupil size during presentation of gratings. Shaded area (G2): 2.5-97.5 percentile interval of null distribution. Brown line shows correlations of retinal boutons (same as in Figure 2F). Dots show correlations of example neurons in E. (H) Distribution of response modulation (measured and null). Dots: values of examples in F. (I) Direction tuning (mean ± SEM) of SC neurons during trials where the pupil was small (black) or large (red). Tuning curves were measured during control conditions (I1) and V1 inactivation (I2). Dotted lines show baseline firing rates. (J) Distribution of response modulations during control conditions (J1) and V1 inactivation (J2). (K) Response modulation during control conditions versus V1 inactivation for neurons with nonsignificant changes in response modulation (p < 0.05, permutation test, gray dots) and neurons with significant changes (black dots). Red dot marks values of neuron in (I).