The neurons that drive infradian sleep-wake and mania-like behavioral rhythms

Abstract

Infradian mood and sleep-wake rhythms with periods of 48 hr and beyond have been observed in bipolar disorder (BD) subjects that even persist in time isolation, indicating an endogenous origin. Here we show that mice exposed to methamphetamine (Meth) in drinking water develop infradian locomotor rhythms with periods of 48 hr and beyond which extend to sleep length and mania-like behaviors in support of a model for cycling in BD. This cycling capacity is abrogated upon genetic disruption of DA production in DA neurons of the ventral tegmental area (VTA) or ablation of nucleus accumbens (NAc) projecting, dopamine (DA) neurons. Chemogenetic activation of NAc-projecting DA neurons leads to locomotor period lengthening in clock deficient mice, while cytosolic calcium in DA processes of the NAc was found fluctuating synchronously with locomotor behavior. Together, our findings argue that BD cycling relies on infradian rhythm generation that depends on NAc-projecting DA neurons.

Keywords: bipolar disorders; circadian rhythms; dopamine; infradian rhythms; methamphetamine; nucleus accumbens; psychopathology; rapid cycling; tyrosine hydroxylase; ventral tegmental area.

Figure 1.. Variability and extent of infradian rhythm induction by methamphetamine

(A) Experimental regimen of running wheel activity monitoring in response to escalating concentrations of Meth in drinking water. (B) Representative actograms displaying running wheel activity patterns of two single housed animals exposed to Meth. Meth concentrations applied are indicated by color shading in correspondence to A). Lomb-Scargle (LS) periodograms on the left and continuous wavelet transform (CWT) heatmaps on the right of each actogram show the emergence and evolution of infradian locomotor rhythms. Green line in periodograms demarcates the significant threshold for rhythmicity. Black traces in heatmaps are the wavelet ridges reporting the instantaneous peak period. (C) Evolution of the peak period across animals identified by the wavelet ridge of the CWT. Time course of Meth treatment as indicated in B. (D) Composite heatmap of normalized periodograms from the final 14 days of Meth treatment suggests profound period flexibility. (E) Binned display of highest peaks derived from periodograms shown in D. Shown are means ± SEM. (F) Representative actograms of Meth-treated mice plotted modulo according the peak period from the last 14 days of the 60-day actogram. The modulo actograms demonstrate that mice can adopt robust rhythmicity in the near to far infradian range upon Meth exposure via drinking water. Greyed areas indicated data loss.

Figure 2.. 48 hr rhythmic mice show cycling in sleep and mania-associated behaviors.

(A) Regimen to record sleep during 48 hr rhythmicity. (B) PIR-derived spontaneous locomotor activity (left) and sleep (right) of a 48 hr rhythmic mouse. (C) Rhythmic strength (Scale-averaged spectral power) in the circadian (22–26 hr) and infradian (44–52 hr) range prior to Meth exposure and during 48 hr cycling. (D) Daily total sleep prior to Meth and on active and inactive days while cycling. (E) Daily sleep (average across four consecutive days) prior to Meth and during 48 hr cycling. (F) Experimental design for open field test (OFT) in 48 hr and 24 hr-rhythmic mice. (G, H) Actogram (G) of a 48 hr cycling mouse used for OFT analysis. Time of testing during active/inactive days indicated. (H) Representative locomotor traces derived from 10 min open field video recording on inactive and active days, as indicated in (G). (I) Total distance travelled during OFT of 48 hr cycling mice during their active and inactive days. (J) Spatial d in 48 hr cycling mice during inactive and active days. (K) Representative actogram of Meth-naïve animals displaying 24hr rhythms in locomotor activity. (L) Total distance travelled during OFT conducted during the active (ZT12-18) and inactive (ZT0-6) phases of 24 hour rhythmic, Meth-naïve mice. (M, N) Total distance travelled on active (M) and inactive (N) days by 48 hr cyclers compared to the inactive phase of Meth-naïve mice. (O) Spatial d in Meth-naïve mice during their active vs inactive phase. (P, Q) Spatial d on active (P) and inactive (Q) days by 48 hr cyclers compared to the inactive phase of Meth-naïve mice. Mean ± SEM. n= 8 (C-E) and 7 (I, J, L-Q). Mann-Whitney test, C-E and L-Q. Wilcoxon’s test, I and J. ns, not significant. *P<0.05; **P<0.01; ***P<0.001. See also Figure S1

Figure 3.. Infradian rhythm generation capacity is abrogated upon ^VTADA neuron ablation.

(A) Experimental design: DA neuron ablation in the VTA via Casp3 and subsequent behavioral monitoring using running wheels. (B, C) Bilateral injection of AAV-DIO-taCasp3-TEVp into the VTA of DAT-Cre mice led to a loss of TH immunofluorescence in the VTA (C) compared to saline injected mice (B). IPN, interpeduncular nucleus; SN, substantia nigra compacta. Scale bars, 500 μm. (D, E) Representative actograms showing running wheel activity of ^VTACasp3 mice and controls in constant darkness in response to escalating levels of Meth in drinking water. LS-periodograms below the actograms are computed for the time window indicated by the blue bar at the bottom right of the actograms. Greyed areas indicate data loss. (F) Composite display of periodograms from individual animals computed from the final 2 weeks of recording under Meth treatment. Periodograms are normalized to its peak amplitude. Grey shading indicates absence of significant amplitudes. (G, H) Periodogram-derived amplitude spectral density (ASD) of significant periodicities in the circadian (20–27 hr) and infradian (27–96 hr) period range before (G) and during the final 2 weeks of Meth-exposure (H). (I, J) Periodogram-derived highest peak in the 20–96 hr range before (I) and during the final 2 weeks of Meth-treatment (J). (K) Total locomotor activity (wheel revolutions) prior to (Water) and during Meth treatment (Meth, final 2 weeks of treatment), Mean ± SEM. n= 5–7. Mann-Whitney’s test, I and J. Two-way ANOVA with Bonferroni’s multiple comparison test, G, H and K. ns, not significant; *P<0.05; **P<0.01; ****P<0.0001. See also Figure S2.

Figure 4.. Selective disruption of Th in the VTA abrogates infradian rhythm generation capacity.

(A) Schematic illustrating the Meth-mediated elevation of extracellular DA at a DA neuronal terminal. DDC, L-dopa decarboxylase; Tyr, tyrosine. (B) Strategy for selective elimination of DA production in DA neurons of the VTA. (C, D) TH immuno-labeling of midbrain sections from saline (Ctrl, C) and AAV-GFP-Cre injected animals (THKO, D). Scale bars, 500 μm. (E, F) Representative actograms showing running wheel activity of control and ^VTATHKO mice in constant darkness in response to escalating levels of Meth in drinking water. Greyed areas indicate missing data. LS-periodograms correspond to time spans indicated by blue bar. (G) Composite display of normalized periodograms from individual animals corresponding to the final 2 weeks of Meth treatment. (H, I) Periodogram-derived ASD% of significant periodicities in the circadian and infradian period ranges before (H, Pre-Meth) and at the end of Meth-treatment (I, Meth). (J, K) Periodogram-derived highest peak before (J, Pre-Meth) and at the end of the Meth-treatment regimen (K, Meth). (L) Total locomotor activity prior to (Water) and during Meth treatment (Meth). Mean ± SEM. n= 7–8. Mann-Whitney’s test, J and K. Two-way ANOVA with Bonferroni’s multiple comparison, H, I, and L. ns, not significant; *P<0.05; **P<0.01; ****P<0.0001. See also Figure S3.

Figure 5.. Selective disruption of Vmat2 in the VTA does not abrogate the capacity for infradian rhythm generation.

(A) Strategy of Vmat2 disruption leading to loss of DA vesicular uptake and release selectively in DA neurons of the VTA. (B, C) Bilateral injection of AAV5-GFP-Cre into the VTA of Vmat2^fl/fl mice leads to loss of Vmat2 message in the VTA but not the SN. Shown are representative images of brain sections after situ hybridization with a Vmat2 specific riboprobe of viral (C, VmatKO) and saline-injected mice (B, Ctrl). Blue precipitate indicates Vmat2 message. Scale bars, 500 μm (SN), 250 μm (VTA). (D, E) Representative actograms displaying running wheel activity of control (D) and ^VTAVmat2KO mice (E) in constant darkness in response to Meth in drinking water. LS-periodograms shown below correspond to the time span indicated by blue bar. (F) Composite display of normalized periodograms computed from the final 2 weeks of recording under Meth treatment. (G, H) Periodogram-derived PSD% of significant periodicities in the circadian and infradian period ranges before (G) and at the end of Meth-treatment (H). (I, J) Periodogram-derived highest peak in the 20–96 hr range before (I, Pre-Meth) and at the end of Meth-treatment (J, Meth). (K) Total locomotor activity prior to (Water) and during Meth treatment (Meth). Mean ± SEM. n= 5–6. Mann-Whitney’s test, I and J. Two-way ANOVA with Bonferroni’s multiple comparison G, H, and K. ns, not significant; ****P<0.0001. See also Figures S4 and S5.

Figure 6.. Antiphasic calcium rhythms in ^NAcDA processes and period shortening of ^VTA->NAcDA-neuron-driven locomotor rhythms by antipsychotics.

(A) Circadian clock-deficient animal model producing ultradian (2–8hr) locomotor rhythms. (B) Experimental layout showing optic fiber targeting of calcium indicator expressing DA neurons and long-term recording of indicator fluorescence at 10 min intervals in constant darkness alongside locomotor activity by PIR. (C) TH and GCamp6s immunofluorescence in the NAc of a reporter mouse with path of recording fiber implant demarcated by dashed line. (D) Example traces of calcium indicator (GCaMP6s) fluorescence (green) and PIR counts (black) from NAc (top and middle) and dorsal striatum (bottom). Shown are smoothened data (10-min average moving window) and standard deviation of the GCaMP6s trace (light green). (E) left: The acrophases of the cosinor fits for the traces shown in c (top recording) are antiphasic; right: Group analysis confirming that the GCaMP6s and PIR signal fluctuations are antiphasic, i.e., shifted by Pi (n= 5, P > 0.87 for statistical difference from Pi, one sample t-test). (F) Correlation of GCaMP6s versus PIR traces from recording in c, top. Insert in e: r-values of animal cohort (n = 5, P < 0.0001 per animal, Wald Test). (G, H) Experimental regimen of locomotor rhythm monitoring in constant darkness and CNO treatment upon viral delivery of a chemogenetic actuator into DA neurons of the VTA. (I) Immunofluorescence images of VTA and NAc sections labeled for mCherry (red) and TH (green) from DAT-Cre x Bmal1^−/− mice injected with AAV-DIO-hM3Dq-mCherry in the VTA. Core and Shell indicate respective subregions of the NAc. Scale bars, 500 (NAc) and 250 μm (VTA). (J, K) Representative, modulo-plotted actograms showing locomotor responses to CNO in drinking water followed by CNO+Haldol. Periodograms are computed from indicated time windows. ‘Tro’ indicates the trough after the dominant periodogram peak during water treatment. The 0hr-Tro periodogram segment therefore captures the periodicities that dominate at baseline (water only treatment). Dashed line, significant threshold for rhythmicity. (L) % of total ASD allocated to 0hr-Tro and Tro-24hr, respectively. (M) Cumulative locomotor activity over 96 hr normalized to the sum of the three 96 hr time windows indicated by brackets. Mean ± SEM. n= 5. Two-way ANOVA with Bonferroni’s multiple comparison. ns, not significant; *P<0.05; **P<0.01.

Figure 7.. Loss of infradian rhythm generation capacity upon NAc-projecting DA neuronal ablation and two-oscillator model for BD cycling.

(A) 6-hydroxydopamine (6-OHDA) was bilaterally injected into the NAc to ablate DAergic processes. (B, C) TH immunolabeling of striatal and midbrain sections from saline (B) and 6-OHDA (C) injected animals. Scale bars, 500 μm. (D, E) Representative actograms showing running wheel activity of saline (D) and 6-OHDA-injected mice (E) in constant darkness in response to escalating doses of Meth in drinking water. LS-periodograms correspond to the time interval indicated (blue bar). (F) Composite display of normalized periodograms from individual animals computed from the final 2 weeks of recording under Meth treatment. (G, H) Periodogram-derived highest peak in the 20–96 hr range before (G, Pre-Meth) and at the end of Meth-treatment (H, Meth). (I) Model actogram showing daily wake periods reflective of a DO operating at a frequency non-harmonious to the SCN clock. (J) In-and-out of phase ‘beating’ of the DO and SCN clock when frequencies are non-harmonious resulting in mood cycling. (K, L) Model of sleep-wake and mood cycling at a DO period of 48 hr which is in harmony with the 24 hr SCN clock. (M) Upon ‘activation’, the DO influences/dominates sleep-wake and mood rhythms. Mean ± SEM. n= 7. Mann-Whitney’s test. ns, not significant; ***P<0.001. See also Figure S6.