Timekeeping in the hindbrain: a multi-oscillatory circadian centre in the mouse dorsal vagal complex

Abstract

Metabolic and cardiovascular processes controlled by the hindbrain exhibit 24 h rhythms, but the extent to which the hindbrain possesses endogenous circadian timekeeping is unresolved. Here we provide compelling evidence that genetic, neuronal, and vascular activities of the brainstem's dorsal vagal complex are subject to intrinsic circadian control with a crucial role for the connection between its components in regulating their rhythmic properties. Robust 24 h variation in clock gene expression in vivo and neuronal firing ex vivo were observed in the area postrema (AP) and nucleus of the solitary tract (NTS), together with enhanced nocturnal responsiveness to metabolic cues. Unexpectedly, we also find functional and molecular evidence for increased penetration of blood borne molecules into the NTS at night. Our findings reveal that the hindbrain houses a local network complex of neuronal and non-neuronal autonomous circadian oscillators, with clear implications for understanding local temporal control of physiology in the brainstem.

Fig. 1. Spatiotemporal variation in Per2/PER2 in the AP and NTS.

a Example traces of PER2::LUC expression over 7 days in culture at the whole tissue level in the AP, NTS and 4^thVep from one brainstem slice. b Cross-correlograms showing the ~2 h lag of NTS to AP (left panel; n = 8) and close to antiphasic relationship of 4^thVep to AP (right panel; n = 5) over 72 h. c Altered times of peak PER2::LUC bioluminescence (AP vs NTS ****p < 0.0001, n = 8, t test; 4^thVep vs AP and NTS ^###p < 0.001, n = 5, Tukey’s test). d Periods of whole area structures of the brainstem (^#p < 0.05, n = 5, Tukey’s test). e Reconstructed images from grid-based, normalised PER2::LUC activities in the brainstem slice showing phase of spatial PER2::LUC expression every 4 h over 3 days. Warm colours code toward peak phase. f (from left to right) The average phase of the oscillation for the first 24 h in culture (upper left) and the period for the first 48 h (lower left), the map of clustering based on the correlation of time series (middle), and the raster plot of normalised time series of the oscillation listed in cluster groups (each row represents one grid) (right). The first cluster (red) approximates the anatomical position of NTS, the second one (green) the 4^thVep, and the third cluster (blue) the AP (middle). g Traces of putative single cell PER2::LUC expression over 7 days in culture. h Rayleigh plots depicting phase clustering around CT 8.5 in AP and CT 11 in NTS at day 2, and subsequent desynchronisation of single cell oscillators at day 5. i Rayleigh r value of individual cells for peaks 2-5 (left panel) and the SD of individual cell lags relative to the phase of the main structure for peaks 2-5 (right panel) for n = 8 cultures. (****p < 0.0001, two-way ANOVA RM). J. Frequency histogram of single cell lags relative to the phase of the whole structure (****p < 0.0001, two-way ANOVA Interaction). k In vivo core clock gene expression of Per2 and Bmal1 in the AP and NTS relative to CT0. (*p < 0.05, **p < 0.01, ****p < 0.0001, one-way ANOVA).

Fig. 2. AP and NTS neurons exhibit robust 24 h and projected day-night variation in electrical activity.

a Example long-term recording profiles of spontaneous multi-unit activity (sMUA; bin = 600 s) in the AP (in blue) and NTS (in red) showing the peak of electrical activity at late projected day/early night. Grey boxes depict the projected dark phase. All recordings were started at ZT2. b The width of the 24 h sMUA profiles for individual recording sites (n = 4 slices from 3 mice) in the AP and NTS taken at 50% of the peak amplitude (****p < 0.0001, t test). c The projected ZT of the peak of sMUA of each electrode in the AP and NTS with an average of ~ZT9.5 for both structures. d For the short-term recording protocol (ZT3-4 vs ZT15-16) sMUA was detected at greater proportion of electrodes in the AP and NTS during the projected night (active spots; blue for AP and red for NTS) (*p < 0.05, ****p < 0.0001, Fisher’s tests). e Mirror heatmaps coding the levels of sMUA in the DVC complex at projected early day (above) and early night (below) obtained from short-term recordings. f sMUA from single recording electrodes localised in the AP (in blue; n = 192) and NTS (in red; n = 842); sMUA was significantly elevated early night in both structures (**p < 0.01, ****p < 0.0001, t tests).

Fig. 3. Daily changes in AP to NTS communication.

a Photograph showing the position of the brainstem section on the electrode array; black represents stimulation sites, green and orange represent electrodes where the example excitatory and inhibitory responses (shown in c) were recorded. Black lines delineate structures: the AP and the borders of DVC. b Schematic representation of the stimulation protocol. c Representative excitatory and inhibitory responses in the NTS shown in modified peri-stimulus time histograms (PSTHs) representing the sum of spike density function around 30 stimuli. d Amplitudes of nocturnal excitations were significantly higher than those recorded during the day (*p = 0.042, Mann–Whitney test), but inhibitory responses did not show this day-to-night increase (p = 0.336, Mann Whitney test). e Proportion of excitations (in green) to inhibitions (in orange) recorded at day and night.

Fig. 4. Mechanical and pharmacological disconnection reveals the circadian properties of DVC oscillators.

a False-coloured bioluminescence image of DVC explant with the unilateral NTS disconnected from the AP. Borders of three regions of interest (ROIs) were delineated and overlaid. White bar depicts 100 µm. b Example traces of wholearea PER2::LUC expression for the AP (in blue), NTS connected (NTSc, in red) and NTS disconnected from the AP (NTSd, in grey) from one hindbrain slice. c Damping rate, shown as relative amplitude to day 2, did not vary between NTSc and NTSd (p = 0.4172, two-way ANOVA RM). d Period of NTSd PER2::LUC rhythms was shorter than the AP (n = 5, **p = 0.0061, Tukey’s test) or NTSc (n = 5, **p = 0.0045, Tukey’s test). e,i Example PER2::LUC bioluminescence traces showing effects of tetrodotoxin (TTX, 0.5 µM) (AP – blue, NTS – red) and TTX washout (dotted box). f, j Single cell bioluminescence traces during TTX treatment with example Rayleigh plots of one day pre- and one day post-treatment (day 2 and 4, respectively) for both structures. g, k TTX had no effect on whole tissue damping rate in the AP (p = 0.8043, two-way ANOVA RM), but did in the NTS (day 3: **p = 0.0046, Sidak’s test). h, l Rayleigh r value showing synchrony amongst single cell oscillators from day 2 to 5. TTX did not reduce synchrony in the AP (p = 0.2380, two-way ANOVA RM), but desynchronised NTS cells (day 3: ***p = 0.0007, Sidak’s test). m Reconstructed normalised PER2::LUC time-lapse images show loss of the spatial phase organisation under TTX. n (upper) Correlation clusters maintain a canonical phase organisation before TTX. (lower) The phase relation between the AP and NTS under control conditions is lost with TTX elicited desynchrony. NTS regains synchronisation after wash. The small bars external to the circle indicate average phases of respective clusters. Black arrow direction indicates the average phase of all oscillators, and its length indicates the degree of synchronisation. o Neuronal constituents of the NTS and AP respond rapidly to TTX and desynchronise. p TTX has no effect on the synchrony of the non-neuronal constituents of the 4^thVep and the border cluster. The sync index is quantified by the Kuramoto order parameter.

Fig. 5. Increased permeability to Evans Blue dye in the NTS at early night.

a Representative images of the rostral-caudal DVC showing the merge of anti-vimentin and anti-GFAP staining (top row) for identification of the AP/NTS glial border along with annotations of the ROIs of the NTS taken for subsequent analysis. Evans Blue dye (EB) fluorescence imaging (middle & bottom rows) with added annotation of the AP/NTS border at ZT1 vs ZT13; note the increased penetration of EB into the adjacent NTS at ZT13. White bar depicts 100 µm. b Intensity of EB staining in the whole, medial and lateral areas of the NTS for the intermediate level sample slice during LD (top panel) or DD (bottom panel) (*p < 0.05, **p < 0.01, t tests or Mann–Whitney tests). c Gene expression relative to CT of the tight junction proteins Claudin-5 (Cldn-5) and Zona occludens-1 (Tjp-1) in the AP/glial border over 24 h (*p < 0.05, one-way ANOVAs). L lateral, M medial.

Fig. 6. Day-night variation in responsiveness of NTS neurons to metabolic factors.

a–d Multi-electrode array recordings conducted during the day (ZT4-6) or night (ZT16-18) reveal that NTS neurons increased nocturnal responsiveness to (a) CCK, (b) elevated glucose, (c) ghrelin, and (d) orexin A. Pie charts represent the proportion of NTS recording locations responding through activation (orange), inhibition (light green) or no change in firing activity (grey). Bars represent the amplitude of drug-evoked excitation or inhibition (*p < 0.05, ***p < 0.001 t test or Mann–Whitney test). Multi-unit activity traces showing representative responses to drug application during the day or night are presented below. e–g Temporal variation in gene expression for metabolic receptors in the NTS sampled at CT0, 6, 12, and 18. Data are plotted relative to the values at CT0. Orexin receptor gene expression (Hcrtr1 and Hcrtr2) varied across these time points. *p < 0.05, one-way ANOVAs or Kruskal–Wallis tests.