Locomotion modulates specific functional cell types in the mouse visual thalamus

Abstract

The visual system is composed of diverse cell types that encode distinct aspects of the visual scene and may form separate processing channels. Here we present further evidence for that hypothesis whereby functional cell groups in the dorsal lateral geniculate nucleus (dLGN) are differentially modulated during behavior. Using simultaneous multi-electrode recordings in dLGN and primary visual cortex (V1) of behaving mice, we characterized the impact of locomotor activity on response amplitude, variability, correlation and spatiotemporal tuning. Locomotion strongly impacts the amplitudes of dLGN and V1 responses but the effects on variability and correlations are relatively minor. With regards to tunings, locomotion enhances dLGN responses to high temporal frequencies, preferentially affecting ON transient cells and neurons with nonlinear responses to high spatial frequencies. Channel specific modulations may serve to highlight particular visual inputs during active behaviors.

Fig. 1

Experimental setup and behavioral paradigm. a Illustration of the linear treadmill assay. Full field, upward drifting sinusoidal gratings of different temporal (TF, 1,2,4,8,16 Hz) or spatial (SF, 0.01, 0.02, 0.04, 0.08, 0.16 cpd) frequencies were delivered to the right eye while animals ran on the treadmill. b Simultaneous multi-electrode recordings from dorsal lateral geniculate nucleus (dLGN, coordinates LM 2.1 AP 2.5) and primary visual cortex (V1, coordinates LM 2.5 AP 3.8). Coordinates in mm from bregma. c Locomotion speed (top) and pupil size (bottom) as a function of time. Scale bar, 20 cm/s. d Visual stimulation epochs (shaded areas) were categorized into locomotion (red) or stationary (black) trials based on locomotion speed measurements (black trace from c dashed lines).  Scale bar, 20 cm/s. e Distribution of the duration of locomotion and stationary bouts (TF experiments: N = 12 mice in 23 sessions). f Fraction of locomotion (red) and stationary trials (black) for each temporal frequency (average±s.e.m. across sessions). g Pupil size as function of time for locomotion (red) and stationary (black) trials (TF experiments: 11/23 sessions with pupil size data; average ± s.e.m., N = 335 and 2414 epochs)

Fig. 2

Locomotion related modulations of early visual responses. a Peri-stimulus time spike raster plots and histograms for two dLGN neurons in response to full field drifting gratings stimuli (temporal frequency: 2 Hz, left; 4 Hz, right). Responses in locomotion (red) and stationary (black) trials are plotted separately. Scale bars,  10 spikes/s and 10 trials. b Mean firing rate (F0) and response (F1) amplitude in stationary vs. locomotion epochs trials for the cells in (a) (36% increase and 18% decrease in firing rate; 41% increase and 19% decrease in response amplitude). c Spike raster plot and histogram for a V1 cell. d Same for a V1 cell (44% increase in firing rate; 28% increase in response amplitude)

Fig. 3

Similar response amplitude modulations in dLGN and V1. a Distribution of firing rate (F0) modulations for dLGN neurons (TF experiments: n = 232 cells). b Same as a for V1 neurons (TF experiments: n = 110 cells). c Distribution of response amplitude (F1) modulations for dLGN neurons. d Same as c for V1 neurons. e Cumulative distributions of firing rate (F0) modulations for cells with positive modulation index (MI > 0). f Same as panel e for negatively modulated cells (MI < 0). g Response (F1) modulations for cells with positive modulation index (MI > 0). h Same as g for negatively modulated cells (MI < 0)

Fig. 4

Weak impact of locomotion on response variability. a Normalized F1 responses  of dLGN neurons in locomotion vs. stationary epochs for cells with ≥4 (gray) or ≥10 locomotion trials (black) (TF experiments: n = 235 cells and 106/235 cells, respectively). b Same as a for V1 neurons (n = 103 cells and 42/103 cells, respectively). c Trial-to-trial variance in F1 responses of dLGN neurons  in locomotion vs. stationary trials (same cells as in a). d Same as c for V1 neurons (same cells as in b). e Fano-factor of firing rates responses  in locomotion vs. stationary trials in dLGN (same cells as in a). f Same as e for V1 neurons (same cells in b). All measures are means computed across temporal frequency stimuli

Fig. 5

Weak impact of locomotion on pairwise activity correlations. a Trial-to-trial spike count responses of a correlated dLGN cell pair. Units are spike/s.  b Example cross-correlogram (CCG) of a correlated dLGN cell pair. c R_CCG (see Methods) for the pair in a. Red line indicate spike count correlation R_SC; black line is the exponential fit. Note how the R_CCG converges to the spike count correlation (R_SC) for large window sizes. d Distribution of integration time from exponential fits to the R_CCG for a sample of dLGN and V1 cell pairs. e Distributions of spike count correlations of dLGN cell pairs (641 pairs) in locomotion (red) and stationary (black) trials. Spike counts calculated using 1 s windows, starting 0.5 s after stimulus onset. f Same as e for V1 cell pairs (199 pairs). g Distribution of spike count correlations of dLGN cell pairs using the entire 2 s stimulus windows. h Same as g for V1 cell pairs

Fig. 6

Unspecific impact on spatial and temporal frequency tunings. a Temporal frequency tuning of F1 responses of example dLGN cells in ocomotion (red) and stationary (black) epochs. Scale bar, 10 spikes/s. b Distribution of preferred temporal frequency of dLGN neurons (N = 143 cells) estimated from locomotion (red) and stationary (black) trials. c Distributions of temporal frequency bandwidth for the cells in b. d Temporal frequency tuning curves of F1 responses of example V1 cells estimated from locomotion (red) and stationary (black) trials. e Distributions of preferred temporal frequency of V1 neurons (N = 98 cells) in locomotion (red) and stationary (black) trials. f Distributions of temporal frequency bandwidth for cells in e. g Example spatial frequency tuning curves of dLGN cells. h Distributions of preferred spatial frequency of dLGN neurons (N = 128 cells). i Distibutions of spatial frequency bandwidth for cells in h. j Example spatial frequency tuning curves of V1 cells. k Distributions of preferred spatial frequency of V1 cells (N = 47 cells). l Distributions of spatial frequency bandwidth for the cells in k. Dashed lines in d, g and j denote F1 responses to gray screen. Continuous lines in a denote F1 of shuffled spike times

Fig. 7

Preferential modulations of dLGN neurons with nonlinear responses to high spatial frequencies. a Grouping of dLGN cells (N = 163) based on normalized cycle averages to the spatial frequency stimulus and blank (30–50 trials). Cells were separated in three groups (Group 1, N = 35 cells, in purple; Group 2, N = 86 cells, in blue; Group 3, N = 42 cells, in green). Right inlet represents the correspondence of neurons to each experiment (N = 15 experiments, N = 8 mice). b Mean of the cycle averages for the groups in a. c Average F0 (left) and F1 (right) responses of cells in each group to the spatial frequency stimulus (same cells as in b). Scale bar for cycle averages is 10 spikes/s. d Response (F1) modulation index of individual groups to the spatial frequency stimulus. e P-values for the differences between groups computed using two-sample Kolmogorov–Smirnov tests

Fig. 8

Preferential modulations of dLGN neurons with transient-ON responses. a Grouping of dLGN cells (N = 122) based on cycle averages in response to full-field contrast reversal stimulus (150–200 trials). Classification into four groups: transient ON, N = 43 cells (blue); sustained ON, N = 33 cells (light-blue); transient OFF, N = 24 cells (light red); and sustained OFF, N = 22 cells (red). Right inlet relates to which experiment each neuron was recorded (N = 8 experiments, N = 5 mice). b Mean of the normalized cycle averages over each group. c Cumulative distributions of response (F1) modulations of each group measured during spatial frequency sessions. d P-values for the difference between groups computed by K–S tests for F1 modulation

Fig. 9

dLGN response modulations following atropine application. a Pupil diameter without (black, ON cycle/OFF cycle, left/right) and with (green) atropine recorded in the same animal (separate sessions). b Mean firing rate (F0) and response (F1) amplitude of an example cell in stationary vs. locomotion trials for five different temporal frequencies measured with atropine (93% increase in firing rate: 78% increase in response amplitude). High modulations by locomotion are independent of pupil diameter. c Same as b but for spatial frequencies (53% increase in firing rate; 23% increase in response amplitude). d Response (F1) modulation index of cells in Fig. 3g, h (black-dashed, N = 232 cells), control (gray N = 42 cells) and with atropine (green, N = 51 cells) to the temporal frequency stimulus (left). Response (F1) modulation index of cells in Supplementary Fig. 3g-h (black-dashed, N = 164 cells), baseline (gray, N = 41 cells) and under atropine (green, N = 51 cells) to the spatial frequency stimulus (right). P-values are shown in the inlet (K–S test). e Superimposed responses of neurons in the three groups (N = 4 mice; ON, unclassified and OFF) from the same animals with (Atropine) and without (Control) atropine application (top row) to full-field contrast reversal stimulus. Population mean of locomotion (red) and stationary (black) responses for the control group (middle row, baseline: ON cells: N = 27, unclassified cells: N = 26, OFF cells: N = 42). Population mean of locomotion (red) and stationary (black) responses for the atropine group (bottom row, baseline: ON cells: N = 25, unclassified cells: N = 42, OFF cells: N = 21)