Cholinergic enhancement reduces spatial spread of visual responses in human early visual cortex

Abstract

Animal studies have shown that acetylcholine decreases excitatory receptive field size and spread of excitation in early visual cortex. These effects are thought to be due to facilitation of thalamocortical synaptic transmission and/or suppression of intracortical connections. We have used functional magnetic resonance imaging (fMRI) to measure the spatial spread of responses to visual stimulation in human early visual cortex. The cholinesterase inhibitor donepezil was administered to normal healthy human subjects to increase synaptic levels of acetylcholine in the brain. Cholinergic enhancement with donepezil decreased the spatial spread of excitatory fMRI responses in visual cortex, consistent with a role of acetylcholine in reducing excitatory receptive field size of cortical neurons. Donepezil also reduced response amplitude in visual cortex, but the cholinergic effects on spatial spread were not a direct result of reduced amplitude. These findings demonstrate that acetylcholine regulates spatial integration in human visual cortex.

Figure 1. Responses to visual stimulation have a bimodal phase distribution in early visual cortex

A checkerboard annulus stimulus was presented in block alternation with a gray screen. A sinusoid with a frequency equal to the frequency of block alternation (0.0521 Hz) was fit to the fMRI time course of each pixel in a computationally-flattened patch of early visual cortex. The phase of this sinusoid relative to the stimulus cycle was color coded according to the color map above panel B and plotted on the flat map of the right hemisphere of Subject #4. No spatial smoothing or statistical thresholding was performed. A, Spatial distribution of response phases in cortical areas V1, V2, and V3. The shape of half of the stimulus annulus can be seen in the pattern of responses in early visual cortex. Due to the contralateral organization of the visual pathways, the left side of the stimulus annulus is represented in this right hemisphere flat map. At the foveal confluence and in regions representing peripheral visual field locations, the phases are mainly blue, consistent with a negative BOLD (inhibitory) response to the stimulus. In between, the phases are mainly yellow and orange, consistent with a positive BOLD response. Average data are shown here to demonstrate the spatial distribution of positive and negative BOLD responses, but all the quantitative analysis of donepezil effects was performed on single 5-minute fMRI runs without averaging. The borders between the positive and negative BOLD responses are indicated by white lines and correspond to the inner and outer boundaries of half of the stimulus annulus. B, Pixel histogram of the data shown in panel A. The distribution of response phases was bimodal, with one mode corresponding to positive BOLD responses and the other corresponding to negative BOLD responses. This phase distribution was well fit by the two-Gaussian model.

Figure 2. Correlation of fMRI time courses with models based on expected time courses of neural activity

A, step function at bottom indicates the timing of stimulus presentation and the expected time course of activity for neurons excited by the stimulus. Black line, convolution of this step function with a canonical hemodynamic response function (HRF). Red line, actual mean fMRI time series from pixels in the positive BOLD phase window in the histogram shown in Figure 1B. The actual and modeled time courses are very similar. B, step function shows expected response for neurons that are inhibited by stimulus presentation. Black line, convolution of this step function with the HRF. Blue line, mean fMRI time series from pixels in the negative BOLD phase window in the histogram in Figure 1B. Again, the two curves are similar. In addition, they are both 180 degrees out of phase relative to the positive BOLD responses shown in panel A. C, step function as in panel A. Black line, convolution of step function with the HRF. The purple and green lines are average fMRI time series from the population of pixels that were classified as having a positive BOLD response under placebo (red in left panel of Figure 3) and a negative BOLD response under donepezil (blue in right panel of Figure 3). This demonstrates that the same population of pixels can respond either positively (purple) or negatively (green), depending on the level of cholinergic enhancement.

Figure 3. The spatial spread of the positive BOLD response to visual stimulation is reduced following donepezil administration

Each cortical flat-map pixel in the left hemisphere of Subject #4 was assigned a color based on the phase of its response relative to the stimulus cycle. Red, pixels with a positive BOLD (excitatory) response to the checkerboard stimulus. Blue, pixels with a negative BOLD (inhibitory) response. Yellow, boundary between the positive BOLD region and the surrounding negative BOLD region, drawn based on the placebo condition. Donepezil reduced the number of pixels with a positive BOLD response and increased the number of pixels with a negative BOLD response to the stimulus.

Figure 4. Donepezil reduced spatial spread and amplitude of excitatory visual responses in early visual cortex

For each 5-minute fMRI run, a phase histogram like that shown in Figure 1B was generated and fit with the two-Gaussian model. This allowed the categorization of each pixel as either positive or negative BOLD. The fraction of positive BOLD responses was computed separately for left and right hemispheres in each visual area. In addition, the amplitude of the best-fit sinusoid was computed for each pixel that was classified as having a positive BOLD response to the stimulus. *, p < 0.05; **, p < 0.001.

Figure 5. The effects of donepezil on spatial spread and on the amplitude of excitatory visual responses were uncorrelated

For each fMRI run in the donepezil sessions, the proportion of positive BOLD pixels (spatial spread) and the response amplitude of these positive BOLD pixels were measured and normalized by the corresponding mean values in the placebo condition.

Figure 6. Effects of donepezil on the BOLD response to breath holding

Subjects alternated blocks of breath holding and paced breathing, and a sinusoid with a period equal to the period of block alternation (48 seconds) was fit to the fMRI time series for each visual cortical flat-map pixel. The mean response amplitude (blue) and phase (red, corresponding to the temporal delay of the BOLD response) are shown. For the group of five subjects, donepezil had no effect on mean amplitude and decreased mean phase relative to placebo in areas V1 and V3. *, p < 0.05.