Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex

Abstract

Rapid variations in cortical state during wakefulness have a strong influence on neural and behavioural responses and are tightly coupled to changes in pupil size across species. However, the physiological processes linking cortical state and pupil variations are largely unknown. Here we demonstrate that these rapid variations, during both quiet waking and locomotion, are highly correlated with fluctuations in the activity of corticopetal noradrenergic and cholinergic projections. Rapid dilations of the pupil are tightly associated with phasic activity in noradrenergic axons, whereas longer-lasting dilations of the pupil, such as during locomotion, are accompanied by sustained activity in cholinergic axons. Thus, the pupil can be used to sensitively track the activity in multiple neuromodulatory transmitter systems as they control the state of the waking brain.

Figure 1. The pupil tracks rapid fluctuations in ACh and NE cortical projections.

(a, left) ChAT projections from BF (orange) and DBH projections from LC (blue). (right) A GCaMP6s-expressing axon traversing long distances in layer 1 of V1. (b) Simultaneous recording of (clockwise from upper left) treadmill velocity, pupil size, CNiFERs (see text) and axonal calcium activity. (c) Stimulation of the LC, but not immediately adjacent tissue, results in a large, rapid dilation of the pupil. (d) Activity in NE axons precedes small, rapid pupil dilations during stillness. (e) Activity in ACh axons also tracks rapid pupil dilations during stillness, but to a lesser extent. (f) At the beginning of walking, strong NE activity occurs along with pupil dilation. (g) ACh activity tracks the large, long-lasting dilation of the pupil that occurs around walking. (h) NE activity is coherent with fluctuations in the pupil over a broad range of infra-slow frequencies (blue). ACh activity is also coherent with pupil, particularly at the lowest frequencies, such as occur around walking (orange). Error bands represent 68% bootstrap confidence interval.

Figure 2. Pupil dilation during stillness is preceded by sequential activation of NE and ACh axons.

(a,b) Mean activity of NE axons (a, blue) and ACh axons (b, orange) aligned to the onset of dilation. (c) NE axons (blue) and ACh axons (orange) aligned to one canonical cycle of dilation and constriction derived from the Hilbert transform (see Methods). No modulation was observed for control auto-fluorescent blebs (grey). (d,e) Median cross-correlation between ACh, NE or bleb traces and the pupil (d) or pupil derivative (e). (f) The peak in cross-correlation for NE leads ACh with respect to pupil dilation (left). Lag-corrected correlation coefficients between NE, ACh or bleb traces and the pupil diameter or pupil derivative (right). These results suggest that NE activity drives rapid pupil dilations—or is tightly controlled by a separate driver—and that pupil diameter tracks both NE and ACh axonal activity during stillness. Error bands and bars are a 68% bootstrap confidence interval.

Figure 3. NE and ACh activity display different time courses around walking.

(a–c) Mean calcium trace for NE axons (a), ACh axons (b) or blebs (c), aligned to the start (left) and end (right) of running. Pupil diameter (grey) and treadmill velocity (dark green) are plotted on the same time base. (d) A large increase in activity during the first second of walking is apparent in NE and ACh axons, but not blebs. During the last second of walking, NE activity has reduced significantly to near baseline, whereas ACh activity remains high. Error bands and bars are a 68% bootstrap confidence interval.