GABA from vasopressin neurons regulates the time at which suprachiasmatic nucleus molecular clocks enable circadian behavior

Abstract

The suprachiasmatic nucleus (SCN), the central circadian pacemaker in mammals, is a network structure composed of multiple types of γ-aminobutyric acid (GABA)-ergic neurons and glial cells. However, the roles of GABA-mediated signaling in the SCN network remain controversial. Here, we report noticeable impairment of the circadian rhythm in mice with a specific deletion of the vesicular GABA transporter in arginine vasopressin (AVP)-producing neurons. These mice showed disturbed diurnal rhythms of GABA_A receptor-mediated synaptic transmission in SCN neurons and marked lengthening of the activity time in circadian behavioral rhythms due to the extended interval between morning and evening locomotor activities. Synchrony of molecular circadian oscillations among SCN neurons did not significantly change, whereas the phase relationships between SCN molecular clocks and circadian morning/evening locomotor activities were altered significantly, as revealed by PER2::LUC imaging of SCN explants and in vivo recording of intracellular Ca²⁺ in SCN AVP neurons. In contrast, daily neuronal activity in SCN neurons in vivo clearly showed a bimodal pattern that correlated with dissociated morning/evening locomotor activities. Therefore, GABAergic transmission from AVP neurons regulates the timing of SCN neuronal firing to temporally restrict circadian behavior to appropriate time windows in SCN molecular clocks.

Keywords: GABA; arginine vasopressin; circadian rhythm; suprachiasmatic nucleus.

Fig. 1.

AVP neuron-specific deletion of Vgat reduces the frequency of miniature GABAergic synaptic currents in both AVP neurons and non-AVP neurons during the daytime. (A) Vgat expression in the SCN of Avp-Vgat^−/− mice was drastically reduced specifically in AVP neurons. In situ hybridization chain reaction was performed to detect Vgat mRNA (green dots) on coronal brain sections prepared from control (Upper) and Avp-Vgat^−/− mice (Lower) crossed with Rosa26-LSL-tdTomato reporter mice. AVP neurons were identified as tdTomato(+) cells. The locations of the magnified images are indicated by white rectangles in the low-power images. (Scale bars: 100 µm and 20 µm in the left and magnified images, respectively.) (B) Amplitude–frequency histograms showing that the mGPSC frequency in AVP neurons from Avp-Vgat^−/− mice (red line) was significantly reduced in 10- to 30-pA amplitude bins compared to control mice (blue line) during the daytime (Left, ZT2 to ZT10, n = 29 and 30 from 5 Avp-Vgat^−/− and 6 control mice, respectively) but not during the nighttime (Right, ZT14 to ZT22, n = 28 and 26 from 5 Avp-Vgat^−/− and 5 control mice, respectively). (C) The histograms show significant reduction of the mGPSC frequency in non-AVP neurons from Avp-Vgat^−/− mice across multiple amplitude bins during the daytime specifically (daytime, n = 18 and 17 from 4 Avp-Vgat^−/− and 4 control mice, respectively; nighttime, n = 21 and 16 from 4 Avp-Vgat^−/− and 2 control mice, respectively). (Inset) Samples of mGPSCs recorded in control (blue line) and Avp-Vgat^−/− mice (red line) in each condition. (Scale bars: 0.5 s and 40 pA.) Values are mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001 by two-way repeated-measures ANOVA followed by post hoc pairwise comparisons.

Fig. 2.

Avp-Vgat^−/− mice show lengthening of the activity time. (A) Representative locomotor activity of two control and two Avp-Vgat^−/− mice. Animals were initially housed in 12:12-h LD conditions and then transferred to DD. Gray shading indicates the time when lights were off. (B) Averaged daily profile of locomotor activity in LD or DD. (C) The mean free-running period and the activity time in DD. Values are mean ± SEM; n = 13 for control, n = 15 for Avp-Vgat^−/− mice. ***P < 0.001 by two-tailed Student’s t tests; ns, not significant.

Fig. 3.

Circadian gene expression is not altered in the SCN of Avp-Vgat^−/− mice. (A) Representative images of Avp and Per1 mRNA expression in the SCN of control and Avp-Vgat^−/− mice at the time indicated. Coronal brain sections were hybridized in situ to an antisense probe for each gene. (Scale bar, 100 µm.) (B) Circadian expression of each gene in the middle SCN along the rostrocaudal axis. Expression was quantified for the entire SCN (Avp) or separately for the shell and core (Per1) and expressed in arbitrary units (AU). Statistically significant difference was not detected in the expression of either gene. Representative regions defined as the shell and core SCN are indicated by yellow circles in A. Values are mean ± SEM; n = 6. No statistically significant effects of genotype were detected by two-way repeated-measures ANOVA followed by post hoc pairwise comparisons (ns).

Fig. 4.

Phase relationships between SCN molecular clocks and the locomotor activity rhythm are altered in Avp-Vgat^−/− mice. (A) Representative images of PER2::LUC expression. Coronal SCN slices of the midrostrocaudal region were prepared from adult control and Avp-Vgat^−/− mice with a luciferase reporter (Per2::Luc) housed in DD. Detrended bioluminescence expressed in relative light units (RLU) within the shell and core regions indicated in the images were plotted on the right. Gray vertical lines labeled “Day 1” indicate around CT0 (16 h after the start of image recordings around CT8). (B) Amplitude of the three peaks defined in A and periods of PER2::LUC oscillations for two cycles. (C) The first peak phases in CT according to behavioral free-running period before slice preparation were shown as Rayleigh plots. Dots indicate the first peak phases of each slice and arrows indicate mean resultant vectors of them. The peak phase was significantly delayed in the SCN Avp-Vgat^−/−;Per2::Luc mice. (D) Representative amplitude, period, and first peak phase maps of PER2::LUC oscillation at the pixel level of SCN slices shown in A. Amplitude, period, and peak phase of PER2::LUC oscillations in the individual pixels covering the SCN were calculated by cosine curve fittings for data representing 72 h from the first CT6 after slice preparation. The dorsal 40% and ventral 30% regions of the SCN were regarded as the shell and core, respectively. Representative regions defined as the SCN shell or core are indicated. (E) Mean amplitude, mean period, SD of period, and SD of first peak phase of individual pixels’ PER2::LUC oscillations obtained for each region of individual slices are collected and compared between groups. (F) Rose plots of pixel’s first peak phases for each region in representative slices. The circular histograms in light brown and numbers in each circle indicate the distribution of individual pixels’ peak phases. Arrows indicate the mean resultant vectors. (G) Rayleigh plots of mean first peak phases of individual pixels’ PER2::LUC oscillations. Mean first peak phases of individual pixels’ PER2::LUC oscillations were delayed in both SCN shell and core of Avp-Vgat^−/−;Per2::Luc mice. Values are mean ± SEM; n = 6 for control, n = 7 for Avp-Vgat^−/−;Per2::Luc mice. **P < 0.01; ***P < 0.001 by two-way repeated-measures ANOVA followed by post hoc pairwise comparisons (B and E), or by Harrison–Kanji test followed by Watson–Williams test (C and G). P values of Rayleigh test were <0.001 for all circular data.

Fig. 5.

Temporal relationship between the in vivo intracellular Ca²⁺ rhythm of SCN AVP neurons and locomotor activity is disturbed in Avp-Vgat^−/− mice. (A) Representative plots of the in vivo jGCaMP7s signal of SCN AVP neurons (green) overlaid with locomotor activity (black) in actograms. Control (Left) and Avp-Vgat^−/− (Right) mice were initially housed in LD (LD1 to LD5) and then in DD (DD1 to DD10). The dark periods are represented as gray shaded areas. (B) Plots of locomotor activity onset (orange), activity offset (light blue), and the peak phase of the GCaMP signal (green) in individual control (Left) and Avp-Vgat^−/− (Right) mice. Identical marker shapes indicate data from the same animal. (C) Plots of the phase difference between activity onset and GCaMP peak (Left) or activity offset and GCaMP peak (Right). Positive values in ΔPhase (GCaMP peak – activity onset/offset) indicate that the GCaMP peak is earlier than the activity onset/offset. Red, Avp-Vgat^−/−; dark blue, control (Avp-Cre); light blue, control (Avp-Cre;Vgat^wt/flox). The average of values during DD8 to DD10 was compared between groups in bar graphs; mean ± SEM; n = 4. *P < 0.05 by two-tailed Student’s t test.

Fig. 6.

In vivo MUA of the SCN exhibits daily bimodal rhythm in Avp-Vgat^−/− mice. (A) Representative actograms of in vivo MUA rhythm in the SCN of a control and two Avp-Vgat^−/− mice. Animals were initially housed in LD (LD1 to LD3) and then in DD (DD1 to DD5). The dark periods are represented as gray shaded areas. The MUA (Left), wheel running (WR) activity (Center), and overlaid (Right; red: MUA, blue: WR) are shown. (B) Serial plots of MUA in the SCN (Upper) and WR (Lower) in the last 48 h of LD or DD. Ten-minute average counts were normalized with 48 h average. (C) Representative actograms of in vivo MUA rhythm outside SCN of a control and two Avp-Vgat^−/− mice. (D) Serial plots of MUA outside SCN (Upper) and WR (Lower) in the last 48 h of LD or DD.