Calcineurin governs baseline and homeostatic regulations of non–rapid eye movement sleep in mice

Significance

The sleep regulatory functions of protein phosphatases have not been characterized in mice. Calcineurin, also known as PP2B or PPP3, is a conserved Ca²⁺/calmodulin-dependent phosphatase that removes phosphorylation from serine/threonine (Ser/Thr) residues of substrate proteins to modulate their functions. In mice, gain or loss of calcineurin activity in the mouse brain neurons dramatically increase or decrease daily sleep amount, respectively. Moreover, depletion of calcineurin in the adult mouse brain prevents the homeostatic sleep response—the increased amount and intensity of homeostatic recovery sleep—following sleep deprivation. Thus, calcineurin acts as an important regulator of daily sleep amount and homeostatic sleep response in the mouse brain.

Abstract

Sleep need accumulates during waking and dissipates during sleep to maintain sleep homeostasis (process S). Besides the regulation of daily (baseline) sleep amount, homeostatic sleep regulation commonly refers to the universal phenomenon that sleep deprivation (SD) causes an increase of sleep need, hence, the amount and intensity of subsequent recovery sleep. The central regulators and signaling pathways that govern the baseline and homeostatic sleep regulations in mammals remain unclear. Here, we report that enhanced activity of calcineurin Aα (CNAα)—a catalytic subunit of calcineurin—in the mouse brain neurons sharply increases the amount (to ~17-h/d) and delta power—a measure of intensity—of non–rapid eye movement sleep (NREMS). Knockout of the regulatory (CnB1) or catalytic (CnAα and CnAβ) subunits of calcineurin diminishes the amount (to ~4-h/d) and delta power of baseline NREMS, but also nearly abrogates the homeostatic recovery NREMS following SD. Accordingly, mathematical modeling of process S reveals an inability to accumulate sleep need during spontaneous or forced wakefulness in calcineurin deficient mice. Moreover, calcineurin promotes baseline NREMS by antagonizing wake-promoting protein kinase A and, in part, by activating sleep-promoting kinase SIK3. Together, these results indicate that calcineurin is an important regulator of sleep need and governs both baseline and homeostatic regulations of NREMS in mice.

In the “two-process” model, process C regulates circadian timing of sleep, whereas process S regulates sleep amount/intensity and maintains sleep homeostasis through the waxing and waning of sleep need (1–6). Regulation of sleep homeostasis is traditionally studied by sleep deprivation (SD) in almost of all experimental model systems (4, 5, 7–12). In mice, the homeostatic sleep response to SD consists of the increased amount and intensity of recovery non–rapid eye movement sleep (NREMS) as well as increased amount of recovery REMS (13). The intensity of NREMS can be measured by the delta (1 to 4 Hz) power density or slow wave activity (SWA) of electroencephalogram (EEG) signals (7, 14, 15). Because the delta power/SWA of NREMS increases in proportion to prior waking time and decreases exponentially during sleep, it has been used for mathematical modeling of process S in mice and humans (14, 16–18).

Genetic studies in mice have identified multiple sleep-promoting serine/threonine (Ser/Thr) protein kinases, including the adenosine monophosphate (AMP)-activated protein kinase (AMPK)-related kinase SIK3 and its upstream kinase LKB1, Ca²⁺/calmodulin-dependent kinase II (CaMKII), and mitogen-activated protein kinase (MAPK)/ERK (12, 19–25). While knockout mice for any of these kinases moderately reduce NREMS amount by ≤3.5-h/d (19–24), gain-of-function of SIK3 or CaMKII markedly increases NREMS amount by 4 to 5-h/d (19, 23). Additionally, loss-of-function studies reveal a partial involvement of SIK3 and CaMKII in the homeostatic regulation of NREMS (12, 21). On the other hand, the sleep regulatory functions of Ser/Thr protein phosphatases have not been characterized in mice because many of these phosphatases are essential for mouse embryogenesis. Calcineurin/PP2B/PPP3 (hereafter calcineurin) is a conserved Ca²⁺/calmodulin-dependent phosphatase, which has been shown to regulate baseline sleep amount, but not homeostatic sleep response, in Drosophila (26, 27).

In this study, we characterized the sleep regulatory functions of the catalytic and regulatory subunits of calcineurin in mice by adeno-associated virus (AAV)-mediated adult brain chimeric (ABC)-expression or knockout (KO) method (22, 28). Because this somatic genetics approach skips mouse development and minimizes genetic crosses, it facilitates rapid one-step analysis of sleep phenotypes, especially for essential and redundant genes, in adult mice (22, 28). We showed that ABC gain or loss of function of calcineurin in the mouse brain neurons markedly increased or decreased daily NREMS amount, respectively. Moreover, ABC knockout of calcineurin nearly abrogated the homeostatic recovery NREMS following SD owing to inability to accumulate sleep need. Furthermore, our genetic and biochemical results indicate that calcineurin promotes baseline NREMS by antagonizing protein kinase A (PKA) and, in part, by activating SIK3 via dephosphorylation of S551. Therefore, calcineurin governs both baseline and homeostatic regulation of NREMS in mice.

Results

Enhanced CNAα Activity Transiently Increases the Amount and Intensity of NREMS.

In mammals, calcineurin consists of the regulatory subunit (CNB1, 2) and catalytic subunit (CNAα, β, γ), both of which are encoded by essential and/or redundant genes (29–31). CNAα/β/γ contains a carboxyl (C)-terminal autoinhibitory domain that blocks access to the catalytic center of the phosphatase domain (29). Typically, calcineurin is activated by relieving this autoinhibition through conformational changes triggered by Ca²⁺ binding of CNB1/2 and calmodulin (32). To study the gain-of-function phenotype of calcineurin in mice, we used the Tet-on system to perform inducible (i)ABC-expression of constitutively active mutant CNAα/β/γ^CA lacking the autoinhibitory domain (28). Specifically, we conducted retro-orbital injection of C57BL/6J adult mice with dual AAV-PHP.eB viruses expressing rtTA from the panneuronal human synapsin (hSyn) promoter and CNAα/β/γ^CA from tetracycline response element (TRE) (28, 33), respectively (Fig. 1 A–C).

Fig. 1.

Calcineurin is a key regulator of daily sleep amount. (A) Schematic of wild-type and constitutively active mutant CNAα/β/γ^CA. (B) Schematic of Dox-inducible (i)ABC-expression of CNAα/β/γ^CA. (C) Immunoblotting of HA-tagged CNAα/β/γ^CA in iABC-CNAα/β/γ^CA mouse brain lysates. (D) Video-based analysis of daily sleep amount in iABC-CNAα/β/γ^CA mice before and after Dox treatment (n = 6/group). (E) EEG/EMG analysis of daily NREMS (Top) or REMS (Bottom) amount in iABC-EGFP and iABC-CNAα^CA (n = 6/group) mice. (F–H) Quantification of NREMS/Wake/REMS (NR/W/R) amount (F), hourly plot of relative NREMS delta power (G) and NREMS EEG power spectra analysis (H) of iABC-EGFP and iABC-CNAα^CA (n = 11/group) mice on day 3 after Dox treatment. (I and J) Immunoblotting (I) and quantification (J) (n = 4) of CNAα protein in the brain lysates of AAV-hSyn-EGFP (ABC-Ctrl) or AAV-hSyn-Cre (ABC-CnAα^KO) injected CnAα^flox/flox mice. (K–M) Quantification of NREMS/Wake/REMS (NR/W/R) amount (K), hourly plot of relative NREMS delta power (L) and NREMS EEG power spectra analysis (M) of ABC-Ctrl and ABC-CnAα^KO (n = 6/group) mice. Data are mean ± SEM. Two-way ANOVA with Sidak’s test (F and K); unpaired t test, two-tailed (H, J, and M); mixed-effects model (G and L). All experiments used biological replicates and not statistically significant (NS, P > 0.05) is not shown in all figure panels.

In the absence of doxycycline (Dox), there was no difference in the daily sleep amount among these iABC-CNAα/β/γ^CA mice as shown by video-based sleep/wake analysis (28). On the second and third day after drinking Dox-containing water, panneuronal expression of CNAα^CA nearly doubled daily sleep amount, while induction of CNAβ^CA or CNAγ^CA did not change sleep amount (Fig. 1 C and D). Interestingly, the sleep amount of iABC-CNAα^CA mice returned to normal on the fourth day of Dox treatment (Fig. 1D), suggesting the existence of a potential negative feedback mechanism to compensate for the calcineurin hyperactivity-induced hypersomnia. EEG/electromyogram (EMG) recording confirmed that iABC-CNAα^CA mice transiently increased NREMS amount to ~15 to 20-h/d, accompanied with elevated NREMS delta power, and decreased the amount of REMS and wakefulness (Fig. 1 E–H and SI Appendix, Fig. S1 A–C). It should be noted that panneuronal induction of CNAα^CA, CNAβ^CA or CNAγ^CA did not appear to impair the physiology or health of these mice during 7 d of Dox treatment (Table 1). These results indicate that enhanced calcineurin (CNAα) activity in the mouse brain neurons sharply and transiently increases the amount and intensity of NREMS.

Table 1.

Physiological and sleep phenotypes of calcineurin mutant mice

	Summary table
	Knockout of CnB1				Overexpression of CNAαCA
	hSyn	CaMKII	mDlx	Ple94	hSyn	Vglut2	Vgat
Body weight	↑	↑	−	−	↓	−	−
Lethality	+	+	−	−	−	−	−
NREMS	↓	↓	−	↓	↑	↑	↑
REMS	↑	↑	−	−	↓	↓	↓
Recovery NREMS	↓	↓	−	−	NA	NA	NA

^“−”indicates no change from control; “NA” indicates not applied. Please note that ABC-CnB1^hSyn-KO and CnB1^CaMKII-KO mice exhibited lethality approximately 4 wk after AAV injection, while CnB1^mDlx-KO and CnB1^Ple94-KO mice remained healthy for at least 3 mo post-injection. Overexpression of CNAα^CA did not affect survival rate during 7 d of Dox treatment.

Knockout of CnAα Diminishes the Amount and Intensity of NREMS without Affecting Circadian Rhythm.

We performed ABC-KO of CnAα in the mouse brain neurons by injecting CnAα^flox/flox adult mice with AAV-hSyn-Cre (28). Immunoblotting showed that the level of CNAα protein was reduced by ~77% in the brain lysates of Cre-expressing (ABC-CnAα^KO) relative to EGFP-expressing (control) CnAα^flox/flox mice (Fig. 1 I and J). Notably, ABC-CnAα^KOmice, relative to the control mice, diminished NREMS amount to ~4 to 6-h/d, accompanied with reduced NREMS delta power and increased duration of wakefulness (Fig. 1 K–M). The ABC-CnAα^KOmice exhibited normal circadian rhythm as shown by the wheel-running assay (34) (SI Appendix, Fig. S2 A and B). Moreover, ABC-CnAα^KOmice appeared healthy and showed normal motor coordination, but exhibited hyperactivity and impaired learning and memory in a series of behavioral tests (SI Appendix, Fig. S2 C–H).

We also performed ABC-KO of CnAα, CnAβ, or both via triple-target CRISPR by injecting constitutively Cas9-expressing mice with AAV-sgRNA viruses expressing a set of three single-guide RNAs targeting CnAα or CnAβ (28, 35) (SI Appendix, Fig. S3 A and B). In contrast to the severe insomnia phenotype of ABC-CnAα^KO mice, ABC-CnAβ^KO mice exhibited only a modest increase of REMS (SI Appendix, Fig. S3 C–E). Relative to ABC-CnAα^KO mice, ABC-CnAα/β^DKO mice further diminished the amount and delta power of baseline NREMS (SI Appendix, Fig. S3 C–E). Moreover, consistent with low CNAγ expression in the mouse brain (31), homozygous CnAγ^KO mice exhibited no sleep phenotype (SI Appendix, Fig. S3 F–I). Taken together, these results suggest that CNAα plays a major role, whereas CNAβ may have a minor yet redundant role in sleep regulation.

Knockout of CnB1 Diminishes the Amount and Intensity of NREMS, but Also Alters Circadian Distribution of NREMS.

Since CNB1, but not CNB2, is highly expressed in the mouse brain (30), we generated ABC-CnB1^KO mice by retro-orbital injection of CnB1^flox/flox mice with AAV-hSyn-Cre. Ablation of CNB1 expression also led to instability of CNAα/β/γ proteins (30) (Fig. 2 A and B), rendering a more severe deficiency of calcineurin in the mouse brain neurons. ABC-CnB1^KO mice appeared obese and hyperactive, but otherwise healthy during the first 3.5 wk following AAV injection (SI Appendix, Fig. S4 A and B and Movies S1 and S2). However, they became sick afterward, rapidly lost weight, and began to die for unknown reasons. Thus, we examined the sleep phenotypes of ABC-CnB1^KO mice between the second and third week after AAV injection.

Fig. 2.

Knockout of CnB1 diminishes the amount and intensity of baseline NREMS. (A) Immunoblotting of brain lysates from AAV-hSyn-EGFP (ABC-Ctrl) or AAV-hSyn-Cre (ABC-CnB1^KO) injected CnB1^flox/flox mice. (B) Quantification of mRNA or protein levels of CNAα/β/γ and CNB1 in ABC-Ctrl and ABC-CnB1^KO mouse brains (n = 5/group). (C–E) Quantification of NREMS/Wake/REMS (NR/W/R) amount (C), hourly plot of relative NREMS delta power (D) and NREMS EEG power spectra analysis (E) in ABC-Ctrl (n = 13) and ABC-CnB1^KO (n = 11) mice under LD cycle. (F–H) Quantification of NREMS/Wake/REMS (NR/W/R) amount (F), hourly plot of relative NREMS delta power (G) and NREMS EEG power spectra analysis (H) in ABC-Ctrl and ABC-CnB1^KO (n = 7/group) mice under constant darkness (DD). Data are mean ± SEM. Unpaired t test, two-tailed (B, E, and H); two-way ANOVA with Sidak’s test (C and F); mixed-effects model (D and G).

ABC-CnB1^KO mice exhibited severe insomnia by diminishing NREMS amount to ~2 to 5-h/d, accompanied with decreased NREMS delta power (Fig. 2 C and E and SI Appendix, Fig. S4 C and D). The wild-type mice exhibited a normal V-shaped hourly curve of NREMS delta power, which corresponded to the dissipation and accumulation of sleep need during the 24-h cycle. By contrast, ABC-CnB1^KO mice exhibited a relatively flat hourly curve of NREMS delta power (Fig. 2D), indicative of abnormal dissipation or accumulation of sleep need during the sleep/wake cycle (see below).

Notably, ABC-CnB1^KO mice also displayed a significant trough of NREMS at the dark-light transition, suggesting a circadian or light-dependent sleep phenotype (SI Appendix, Fig. S4 C and D). However, we observed similar sleep phenotypes of ABC-CnB1^KO mice under constant darkness (Fig. 2 F–H), eliminating the possibility of a light-dependent sleep phenotype. Ex vivo imaging of the Period2::Luciferase reporter in the SCN slices (36) revealed that ABC-CnB1^KO mice exhibited a longer circadian period than that of control mice (SI Appendix, Fig. S4 F–H), consistent with aberrant circadian distribution of NREMS in these mutant mice.

Calcineurin Is Required for the Homeostatic Recovery NREMS Following SD.

Next, we investigated the homeostatic sleep response of ABC-CnB1^KO mice following two different protocols of SD. In ABC-CnB1^KO mice, panneuronal knockout of calcineurin nearly abolished the increased amount of recovery NREMS, but also blunted the increase of NREMS delta power following 6-h of SD by gentle handling (Fig. 3 A and B and SI Appendix, Fig. S5 A–D). A similar deficiency in the homeostatic recovery NREMS was also observed in ABC-CnB1^KO mice following 6-h of SD by curling prevention with water (CPW) (37) (Fig. 3 D and E and SI Appendix, Fig. S5 F–I). Likewise, ABC-CnAα^KO mice were partially defective, whereas ABC-CnAα/β^DKO mice were deficient for the homeostatic recovery NREMS following SD by CPW (Fig. 3 G, H, J, and K and SI Appendix, Fig. S5 K–N and P–S). Therefore, calcineurin is also required for the homeostatic regulation of NREMS in mice, which contradicts previous finding that calcineurin is dispensable for the homeostatic sleep response in the fruit fly (26, 27).

Fig. 3.

Calcineurin is required for the homeostatic recovery NREMS following SD. (A–C) Hourly plot of accumulated recovery NREMS amount (A), quantification of NREMS delta power change (ZT7) (B) after 6-h SD by gentle handling, and quantification of Td and Ti during simulated process S (C) of ABC-Ctrl (n = 10) and ABC-CnB1^KO (n = 13) mice. (D–F) Hourly plot of accumulated recovery NREMS amount (D) and quantification of NREMS delta power change (ZT7) (E) after 6-h SD by CPW, and quantification of Td and Ti during simulated process S (F) of ABC-Ctrl and ABC-CnB1^KO (n = 7/group) mice. (G–I) Hourly plot of accumulated recovery NREMS amount (G) and quantification of NREMS delta power change (ZT7) (H) after 6-h SD by CPW, and quantification of Td and Ti during simulated process S (I) of ABC-Ctrl and ABC-CnAα^KO (n = 6/group) mice. (J–L) Hourly plot of accumulated recovery NREMS amount (J) and quantification of NREMS delta power change (ZT7-9 owing to no NREMS at ZT7-8) (K) after 6-h SD by CPW, and quantification of Td and Ti during simulated process S (L) of ABC-Ctrl (n = 8) and ABC-CnAα/β^DKO (n = 6) mice. Data are mean ± SEM. Unpaired t test, two-tailed (B, C, E, F, H, I, K, and L); mixed-effects model (A, D, G, and J).

Calcineurin Deficiency Renders a Defective Process S Owing to Inability to Accumulate Sleep Need.

In the “two-process” model, sleep is regulated by a circadian-dependent process C and a sleep-dependent process S, whereby sleep need accumulates during waking and dissipates through sleep to maintain sleep homeostasis (1, 2, 15). To study the dynamics of the accumulation/dissipation of sleep need in calcineurin knockout mice, we performed simulation of process S based on changes of NREMS delta power over a 48-h sleep/wake cycle including 6-h of SD as previously described (14, 16, 17) (SI Appendix, Fig. S5 E, J, O, and T). In ABC-CnAα^KO mice, the time constant (Td) for delta power decrease was normal, whereas the time constant (Ti) for delta power increase was twice of that in the wild-type control mice (Fig. 3I). Both ABC-CnB1^KO and CnAα/β^DKO mice exhibited a normal Td and an abnormally large Ti bordering the ceiling (~26-h) of simulation (Fig. 3 C, F, and L). These observations suggest that calcineurin knockout mice exhibit a defective process S owing to inability to accumulate sleep need during spontaneous or forced wakefulness. Thus, calcineurin is an important regulator of process S that governs both baseline and homeostatic regulations of NREMS.

Calcineurin Governs Different Aspects of Sleep Regulation in Different Mouse Brain Neurons.

Despite significant progress in elucidating the neural pathways that regulate sleep/wake behaviors in the mouse brain (38, 39), the specific neuron populations that govern daily sleep amount and homeostatic sleep response remain largely unknown. Thus, we took a gain-of-function approach to investigate the sleep phenotypes resulted from induction of CNAα^CA in either glutamatergic or GABAergic neurons by retro-orbital injection of Vglut2^Cre or Vgat^Cre mice (40) with AAV-hSyn-DIO-rtTA and AAV-TRE-CNAα^CA, respectively (Fig. 4A). Doxcycline-inducible expression of CNAα^CA in vGlut2⁺ or vGAT⁺ neurons markedly increased the daily amount and delta power of NREMS (Fig. 4 B–E and SI Appendix, Fig. S6 A and B). By contrast, induction of CNAα^CA in the vGAT⁺ or vGlut2⁺ neurons, respectively, increased or decreased the amount of REMS (Fig. 4 B and D). These observations suggest that calcineurin promotes baseline NREMS, but oppositely regulates REMS, in the excitatory and inhibitory neurons.

Fig. 4.

Calcineurin regulates baseline NREMS and recovery NREMS in distinct mouse brain neurons. (A) Schematic of generating iABC-CNAα^CA/Vglut2^Cre or iABC-CNAα^CA/Vgat^Cre mice. (B and C) Quantification of NREMS/Wake/REMS (NR/W/R) amount (B) and NREMS EEG power spectra analysis (C) of iABC-CNAα^CA/Vgat^Cre (n = 7) mice before and after (Day 5) Dox treatment. (D and E) Quantification of NREMS/Wake/REMS (NR/W/R) amount (D) and NREMS EEG power spectra analysis (E) of iABC-CNAα^CA/Vglut2^Cre (n = 8) mice before and after (Day 3) Dox treatment. (F) Schematic of cell type–specific ABC-CnB1^KO by injecting CnB1^flox/flox mice with AAV-CaMKII/Dlx/Ple94-Cre. (G–J) Quantification of NREMS/Wake/REMS (NR/W/R) amount (G), NREMS EEG power spectra analysis (H), hourly plot of accumulated recovery NREMS amount (I), and quantification of NREMS delta power change (ZT7) (J) after 6-h SD in AAV-hSyn-EGFP (Ctrl, n = 7), AAV-CaMKII-Cre (CnB1^CaMKII-KO, n = 7), AAV-Dlx-Cre (CnB1^Dlx-KO, n = 6), and AAV-Ple94-Cre (CnB1^Ple94-KO, n = 8) injected CnB1^flox/floxmice. Data are mean ± SEM. Paired t test, two-tailed (B–E); two-way ANOVA with Sidak’s test (G); unpaired t test, two-tailed (H); mixed-effects model (I); One-way ANOVA with Dunnett’s test (J).

Moreover, we retro-orbitally injected CnB1^flox/flox mice with AAV expressing Cre from the CaMKII (predominantly excitatory neurons) (41), distalless homeobox (Dlx, inhibitory interneurons) (33, 42), or Ple94/GPR88 (43) promoter (Fig. 4F). Retro-orbital injection of AAV-Ple94-Cre in the Ai14 mice resulted in Cre-dependent mCherry expression in the cortex, hypothalamus, striatum, and part of brainstem (43) (SI Appendix, Fig. S6C). While CnB1^CaMKII-KO mice exhibited sleep phenotypes similar to those of ABC-CnB1^KO mice, CnB1^Dlx-KO mice displayed no sleep phenotype (Fig. 4 G–J and SI Appendix, Fig. S6 D–H). By contrast, CnB1^Ple94-KO mice diminished baseline NREMS amount without affecting the homeostatic recovery NREMS after SD (Fig. 4 G–J and SI Appendix, Fig. S6 D–H).

An overview of the physiological conditions and sleep phenotypes of calcineurin mutant mice is presented in Table 1. Notably, global knockout of calcineurin across the mouse brain (or excitatory) neurons in ABC-CnB1^KO (or CnB1^CaMKII-KO) mice resulted in a multitude of phenotypes, including reduced NREMS, increased REMS, impaired homeostatic sleep response, obesity, and lethality after 1-mo post-AAV injection. In contrast, selective knockout of calcineurin in ABC-CnB1^Ple94-KO mice specifically caused a decrease of baseline NREMS amount without obesity or lethality. These results suggest that calcineurin may regulate different aspects of physiological or sleep phenotypes in distinct neuron populations.

Calcineurin Promotes Baseline NREMS by Antagonizing Wake-Promoting Kinase PKA.

It is well known that calcineurin antagonizes PKA functions in multiple physiological processes including synaptic plasticity (44–46). Since PKA functions as a wake-promoting kinase in the fly brain (47, 48), we hypothesized that calcineurin promotes NREMS by antagonizing PKA functions in the mouse brain. Consistent with this hypothesis, immunoblotting with anti-PKA phospho-motif antibodies detected an increase in global phosphorylation of PKA substrates in the ABC-CnB1^KO mouse brain lysates (Fig. 5 A and B).

Fig. 5.

Calcineurin promotes NREMS by antagonizing PKA functions. (A and B) Immunoblotting (A) and quantification (B) of global phosphorylation of PKA substrates in ABC-Ctrl and ABC-CnB1^KO brain lysates (n = 6/group) using anti-PKA phosphor-motif antibodies. (C) Immunoblotting of PKAα and PKAβ in the brain lysates of Cas9 mice injected with AAV-sgRNA^NT (ABC-Ctrl), AAV-sgRNA^PKAα (ABC-PKAα^KO), or AAV-sgRNA^PKAβ (ABC-PKAβ^KO). (D–F) Quantification of NREMS/Wake/REMS (NR/W/R) amount (D), hourly plot of relative NREMS delta power (E), and NREMS EEG power spectra analysis (F) in ABC-Ctrl (n = 7), ABC-PKAα^KO(n = 7), and ABC-PKAβ^KO(n = 5) mice. (G–I) Quantification of NREMS/Wake/REMS (NR/W/R) amount (G), hourly plot of relative NREMS delta power (H), and NREMS EEG power spectra analysis (I) of C57BL/6J mice injected with AAV-Ple94-EGFP (ABC-Ctrl) (n = 8), AAV-Ple94-Cre/AAV-hSyn-DIO-PKAα (ABC-PKAα^Ple94) (n = 5), or AAV-Ple94-Cre/AAV-hSyn-DIO-dnPKA (ABC-dnPKA^Ple94) (n = 8). (J–L) Quantification of NREMS/Wake/REMS (NR/W/R) amount (J), hourly plot of relative NREMS delta power (K), and NREMS EEG power spectra analysis (L) of CnB1^flox/flox mice injected with AAV-Ple94-EGFP (Ctrl) (n = 6), AAV-Ple94-Cre (CnB1^Ple94-KO) (n = 6), and AAV-Ple94-Cre/AAV-hSyn-DIO-dnPKA (CnB1^Ple94-KO/dnPKA^Ple94) (n = 8). Data are mean ± SEM. Unpaired t test, two-tailed (B, F, I, and L); two-way ANOVA with Sidak’s test (D, G, and J); mixed-effects model (E, H, and K).

The inactive PKA holoenzyme is a heterotetramer that consists of two catalytic subunits (PKAα, β) and two inhibitory regulatory subunits (RIα, β, or RIIα, β). The regulatory subunits contain two cyclic adenosine monophosphate (cAMP)-binding sites which, upon cAMP binding, activate PKA by inducing conformational changes to release the catalytic subunits (49) (SI Appendix, Fig. S7A). To examine whether PKA also acted as a wake-promoting kinase in mice, we generated ABC-PKAα^KO, PKAβ^KO, or PKAα/β^DKO mice via triple-target CRISPR by injecting Cas9 mice with AAV-sgRNA targeting PKAα and/or PKAβ (Fig. 5C). While ABC-PKAα/β^DKO resulted in immediate lethality, ABC-PKAα^KO and PKAβ^KO mice were healthy, exhibited reduced wakefulness and increased amount of NREMS, but did not significantly change NREMS delta power or REMS amount (Fig. 5 D–F).

We also constructed dominant-negative (dn)PKA by mutating one (B site) of the two cAMP-binding sites of the PKA regulatory subunit RIα (RIαB), such that it could constitutively associate with PKA catalytic subunits and disrupt the ability of cAMP to activate the mutant holoenzyme (50) (SI Appendix, Fig. S7A). Accordingly, ABC-expression of RIαB significantly reduced wakefulness and increased NREMS amount and delta power (SI Appendix, Fig. S7 B–D). Collectively, these results indicate that PKA is also a wake-promoting kinase that suppresses NREMS in mice.

Next, we performed ABC-expression of PKAα or dnPKA from the Ple94 promoter by coinjecting C57BL/6J mice with AAV-Ple94-Cre and AAV-hSyn-DIO-PKAα or AAV-hSyn-DIO-RIαB, respectively. While ABC-dnPKA^Ple94 mice increased baseline NREMS amount, ABC-PKAα^Ple94 mice diminished the amount and delta power of NREMS similar to that in ABC-CnB1^Ple94-KO mice (Fig. 5 G–I and SI Appendix, Fig. S7E). Moreover, ABC-expression of dnPKA from the Ple94 promoter restored the daily amount and delta power of NREMS in ABC-CnB1^Ple94-KO mice (Fig. 5 J–L and SI Appendix, Fig. S7F), supporting the notion that calcineurin promotes baseline NREMS by antagonizing the wake-promoting kinase PKA.

Calcineurin Promotes Baseline NREMS Partly by Activating SIK3 via S551 Dephosphorylation.

It is possible that calcineurin antagonizes PKA functions by dephosphorylating their common substrates. Interestingly, there are two “LxVP” calcineurin-binding motifs in SIK3, which is predicted to be a calcineurin substrate (51) (Fig. 6A). Mutations of both LxVP motifs in SIK3-L disrupted its association with CNAα in the transfected HEK293 cells (Fig. 6B).

Fig. 6.

Calcineurin promotes NREMS by activating SIK3 via S551 dephosphorylation. (A) Schematic of SIK3 protein (Top) and multisequence alignment (Bottom) showing phylogenetic conservation of the two LxVP motifs in SIK3. (B) Mutations of two LxVP motifs to AAAA in SIK3 (LxVP^m2) disrupt the SIK3–CNAα interaction in transfected AAVpro 293T cells. (C and D) Immunoblotting (C) and quantification (D) (n = 4/group) of SIK3 pS551 in the brain lysates of ABC-Ctrl and ABC-CnB1^KO mice. (E) In vitro phosphatase assay in ABC-CnB1^KO brain lysate supplemented with CNAα, CNB1, and SIK3-L recombinant proteins. (F) Immunoblotting of SIK3 pS551 in PNC with or without KCl treatment. (G) Immunoblotting of SIK3 pS551 in PNC after NMDA treatment with or without FK506 or CsA. (H) Immunoblotting of SIK3 pS551 in PNC with or without bicuculline (Bic) treatment. (I) Quantification of SIK3 pS551 level (n = 3/group) in PNC after bicuculline treatment with or without FK506 or CsA. (J) Schematic of epistasis analysis of ABC-CnB1^KO and Sik3^Slp mutations. (K and L) Quantification of NREMS amount (K) and hourly plot of accumulated recovery NREMS amount (L) in ABC-Ctrl (n = 6), ABC-CnB1^KO (n = 5), and ABC-CnB1^KO/Slp (n = 5) mice. (M) A model depicting that calcineurin governs the baseline and homeostatic NREMS regulations via distinct downstream signaling pathways. Data are mean ± SEM. Unpaired t test, two-tailed (D); One-way ANOVA with Dunnett’s test (I); two-way ANOVA with Sidak’s test (K); mixed-effects model (L).

SIK3 also contains a conserved inhibitory PKA phosphorylation site, S551, of which deletion or mutation results in Sleepy (SLP) mice (19, 52). We found that phosphorylation of SIK3 at S551 was increased in primary cultured neurons following treatment with calcineurin inhibitors, FK506, or cyclosporin A (53, 54) (SI Appendix, Fig. S8C). The level of SIK3 S551 phosphorylation was also increased in the brain lysates of ABC-CnB1^KO or ABC-CnAα^KO mice, but not ABC-CnAβ^KO mice (Fig. 6 C and D and SI Appendix, Fig. S8 A and B). Furthermore, in vitro phosphatase assay confirmed that recombinant CNAα and CNB1 could catalyze S551 dephosphorylation of SIK3 (Fig. 6E).

We observed that SIK3 S551 was dephosphorylated in primary cultured neurons following treatment with KCl, N-methyl-D-aspartate (NMDA), or GABA_A antagonist Bicuculline, all of which caused neuronal activation (Fig. 6 F–H). Moreover, NMDA or Bicuculline-induced S551 dephosphorylation of SIK3 could be blocked by FK506 or cyclosporin A (Fig. 6 G and I and SI Appendix, Fig. S8D), suggesting that neuronal activities induce calcineurin-mediated dephosphorylation of SIK3.

Next, we performed epistasis analysis by retro-orbital injection of Rosa26^Cas9/+;Sik3^Slp/+ mice (Materials and Methods) with AAV-gRNA^CnB1 (Fig. 6J). As controls, we injected Rosa26^Cas9/+;Sik3^Slp-flox/+ mice with AAV-gRNA^NT or AAV-gRNA^CnB1, respectively (Fig. 6J). Importantly, expression of SLP kinase partially restored the amount and delta power of baseline NREMS, but did not rescue the recovery NREMS after SD in ABC-CnB1^KO mice (Fig. 6 K and L and SI Appendix, Fig. S8 E–G). These genetic and biochemical results indicate that calcineurin promotes baseline NREMS, in part, by activating SIK3 via dephosphorylation of S551 (Fig. 6M).

Discussion

In this study, we showed that enhanced calcineurin (CNAα) activity in the mouse brain neurons sharply increased baseline NREMS amount to average ~17-h/d accompanied with elevated NREMS delta power. Conversely, genetic ablation of calcineurin in the adult mouse brain diminished baseline NREMS to average ~4-h/d accompanied with reduced NREMS delta power. These results indicate that calcineurin is a key determinant of daily NREMS amount, which also explains why insomnia is a frequently reported side effect of calcineurin inhibitors, FK506, or cyclosporin A, that are widely used as immunosuppressants in humans (53, 54). Moreover, ABC-CnB1^KO and CnAα/β^DKO mice exhibited a relatively flat hourly curve of NREMS delta power, but also were deficient for the homeostatic recovery NREMS after SD. Consistent with these observations, mathematical modeling of NREMS delta power revealed that mice lacking calcineurin exhibited a disrupted process S owing to inability to accumulate sleep need. Taken together, these results indicate that calcineurin is an important regulator of sleep need and governs both baseline and homeostatic regulations of NREMS in mice.

It should be noted that ABC-CnAα^KO mice exhibited normal circadian rhythm but severe insomnia, whereas ABC-CnB1^KO mice displayed a longer circadian period, but similar sleep phenotypes under both light–dark (LD) and DD cycles. These observations indicate that, although calcineurin also regulates circadian rhythm, the sleep phenotypes of calcineurin mutant mice are largely independent of circadian phenotypes. However, we could not exclude the possibility that circadian dysfunctions may contribute to some aspects of sleep phenotypes, such as aberrant circadian timing of NREMS in ABC-CnB1^KO mice.

We found that calcineurin could promote baseline NREMS, but oppositely regulate REMS, in the excitatory and inhibitory neurons by AAV-mediated expression of calcineurin in vGlut2⁺ or vGAT⁺ neurons. Moreover, ABC-CnB1^CaMKII-KO mice not only diminished baseline NREMS amount, but also abrogated recovery NREMS after SD. By contrast, ABC-CnB1^Ple94-KO mice diminished baseline NREMS amount, but showed normal recovery NREMS after SD. Thus, it is possible that calcineurin governs baseline NREMS amount and homeostatic recovery NREMS in distinct brain neurons. Our genetic and biochemical results indicate that calcineurin promotes baseline NREMS by antagonizing the wake-promoting kinase PKA. Notably, inhibition of PKA activity by expression of dnPKA (RIαB) from the Ple94 promoter restored the amount and delta power of NREMS in ABC-CnB1^Ple94-KO mice. This finding is consistent with two recent reports that calcineurin antagonizes PKA at the excitatory post-synapses to regulate baseline NREMS amount (55, 56). It should be noted that ABC-expression of PKAα from the hSyn or CaMKII promoter resulted in immediate lethality, which prevented us from examining whether hyperactivity of PKA impaired the homeostatic sleep response as in ABC-CnB1^KO mice. Thus, it remains uncertain whether PKA plays a critical role in the homeostatic regulation of sleep need following SD.

Furthermore, we showed that expression of mutant SLP/SIK3 kinase, which lacks S551 of SIK3—a substrate of calcineurin, partially restored baseline NREMS amount, but not the homeostatic recovery NREMS after SD. Thus, it is possible that calcineurin may modulate the activities of other kinases or phosphatases, such as ERK (20, 57) or PP1 (58), to regulate baseline NREMS amount and/or homeostatic recovery NREMS. Alternatively, calcineurin may directly dephosphorylate downstream effector proteins, such as the low-voltage activated Ca²⁺ channel (Ca_v3.2) (59, 60) and metabotropic glutamate receptor subtype 5 (mGluR5) (61, 62), to regulate baseline or recovery NREMS, respectively. Therefore, future studies are warranted to elucidate the downstream signaling pathways by which calcineurin governs the baseline and homeostatic regulations of NREMS in the mouse brain.

Materials and Methods

Animals.

Mice were provisioned ad libitum with food and water and housed in a controlled environment at 23 ± 1 °C and 50% ± 10% humidity with a 12-h light/dark photoperiod (lights on from 09:00 to 21:00), in accordance with the protocols approved by the Institutional Animal Care and Use Committee of the National Institute of Biological Sciences, Beijing (NIBS). CnAα^flox/flox mice were generated by mating CnAα^{tm1a(EUCOMM)Wtsi}/Cmsu (EPD0220_3_D06) mutant from Cambridge-SU Genomic Resource Center (CAM-SU GRC) with FLPeR mice (003946, JAX) from Jackson Laboratory. CnAβ^flox/flox(T018586) and CnAγ^−/− (T027948) mice were sourced from GemPharmatech. Rosa26^Cas9 (026179, JAX), Rosa26^LSL-Cas9 (024857, JAX), Vglut2^Cre (016963, JAX), and vGAT^Cre (016962, JAX) mice were purchased from Jackson Laboratory. Whereas CnB1^flox/flox mice were generated by flanking exon 2 of CnB1 with two loxP sites, Sik3^{Slp-flox/flox} mice were generated by flanking exon 13 of Sik3 with two loxP sites. Both Rosa26^Cas9/+;Sik3^Slp/+ and Rosa26^Cas9/+; Sik3^Slp-flox/+mice were generated by breeding Rosa26^Cas9/Cas9;Sik3^{Slp-flox/flox} with CMV-Cre/+ mice (006054, JAX).

Plasmids.

Mouse open reading frames (ORFs) of CnAα (MR227082), CnAβ (MR208424), CnAγ (MR224578), and PKAα (MR205322) were purchased from OriGene Technologies. pAAV-Tre-3HA-CnAα/β/γ^CA were constructed from pAAV-Tre-EGFP by replacing EGFP with ORF sequences of CnAα^CA (AA 1-398), CnAβ^CA (AA 1-408), and CnAγ^CA (AA 1-391) mutants lacking the C-terminal autoinhibitory domain. pAAV-hSyn-3HA-RIαB was constructed by subcloning RIα^G325D into pAAV-hSyn-EGFP (Addgene, 105539). pAAV-hSyn-DIO-3HA-RIαB was constructed by cloning RIα^G325D into pAAV-hSyn-DIO-hM3D(Gq)-mCherry (Addgene, 44361). Addgene supplied pAAV-CaMKII-Cre (182736), and pAAV-Ple94-Cre (49125). pAAV-Dlx-Cre was constructed from pAAV-mDlx-NLS-mRuby2 (Addgene, 99130) by replacing mRuby2 with Cre. pCS2-3HA-SIK3^LxVPm2 was constructed by mutating two “LxVP” motifs to “AAAA” by site-directed mutagenesis.

For ABC-KO of genes by triple-target CRISPR/Cas9, different sets of three sgRNAs sequences for CnAα, CnAβ, CnB1, PKAα, or PKAβ were designed on the CRISPRdirect website (https://crispr.dbcls.jp). Annealed oligonucleotides for individual sgRNAs were inserted into pAAV-U6-sgRNA-Cbh-Cre-KASH-WPRE (60229, Addgene) and followed by sequential subcloning to construct pAAV-sgRNA.

AAV Packaging and Purification.

AAV packaging and purification were conducted as previously described (22, 28). Briefly, AAVpro 293T cells (632273, Clontech) without Mycoplasma contamination were transfected with PHP.eB (Addgene, 103005), pHelper (Addgene, 112867), and specific AAV plasmids carrying target genes using the polyethylenimine MAX (24765, Polysciences). After 72 h, cells were collected and resuspended in 1× Gradient Buffer (10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 10 mM MgCl₂). The cells underwent lysis through five freeze-thaw cycles, accompanied by vortexing, and were treated with benzonase nuclease (E1014, Millipore) at 37 °C for 30 min. After centrifugation at 20,000 g at 4 °C for 30 min, the supernatant was carefully layered onto preformed discontinuous iodixanol gradients (15%, 25%, 40%, and 58%) and centrifuged at 288,000 g for 4 h at 4 °C. The virus-containing layer at 40% was extracted via a needle inserted approximately 1 to 2 mm below the interface between the 40% and 58% layers. The extracted AAVs were concentrated using 100 kDa Amicon filters (UFC801096, EMD) and formulated in phosphate-buffered saline (PBS) supplemented with 0.01% Pluronic F68 (24040032, Gibco). Finally, virus titers were determined by qPCR using a linearized AAV plasmid as a standard for calibration.

Retro-Orbital AAV Injection.

Mice were anesthetized with 4% isoflurane for 2 min and retro-orbitally injected with 10¹² viral genomes (vg) of AAV as previously described (22, 28).

EEG/EMG or Video-Based Sleep Analysis.

EEG/EMG surgeries were performed on 11- to 13-wk-old adult male mice by experienced technicians who are blind to the genotypes of test mice. The mice were anesthetized with isoflurane (4% for induction, 2% for maintenance). Once unconscious, their heads were shaved, and their skulls were exposed and cleaned. An electrical drill set the coordinates at the lambda point (0, 0, 0) and drilled four holes at the following coordinates: (−1.27, 0, 0), (−1.27, 5.03, 0), (1.27, 5.03, 0), and (1.27, 0, 0). EEG electrodes were implanted at a depth of 1.3 mm, and EMG wires were inserted into the neck muscles, both secured with dental cement. Before EEG/EMG recording, the mice were tethered to a counterbalanced arm (Instech Laboratories) for 1 wk to acclimate to the whole setup and the recording room environment. For baseline sleep and recovery sleep recording, 3 d of baseline sleep recording were conducted, followed by 6 h of SD and recovery sleep recording. For the Tet-on inducible system, 1 wk after viral injection, these mice underwent EEG/EMG recording for 1 d with baseline conditions using normal water, followed by 5 d of induction with doxycycline (Dox, 1.5 mg/mL) containing water.

EEG/EMG recording data was visualized and analyzed using a custom-designed, C++ language-based semiautomated sleep/wake staging software: First, EEG signals were subjected to fast Fourier transform analysis for 1 to 30 Hz with 1 Hz bins. Second, each 20-s epoch was automatically staged into NREMS (high amplitude delta [1 to 4 Hz] frequency EEG and low EMG tonus), REMS (high amplitude theta [6 to 9 Hz] frequency EEG and EMG muscle atonia) and wake (low amplitude fast EEG and high amplitude variable EMG signal). Finally, the automated sleep/wake staging results were manually inspected and corrected. The total time of NREMS, REMS, and wake of each mouse was averaged over the 3 d of EEG/EMG recording. NREMS/REMS/wake time hourly plots were calculated by the sum of all 20-s epochs staged as each state per hour, averaged over the 3 d of EEG/EMG recording. The hourly plot of relative NREMS delta power was calculated by averaging the relative NREMS delta power density in each hour during the 3 d of EEG/EMG recording. Values were normalized relative to the last 4 h of the light phase (i.e., ZT 8 to 12 defined as 100%) at baseline. The EEG power spectrum analysis was conducted by calculating the ratio of each 1 Hz bin EEG power to total (1 to 30) Hz EEG power in NREMS/REMS/wake state during the 24-h cycle and averaged for the 3 d recording period. To examine the sleep/wakefulness behavior under constant darkness, ABC-Ctrl and ABC-CnB1^KO mice were placed into a constant-dark room for 1 wk. After adaptation, sleep patterns were recorded by EEG/EMG in constant darkness for 3 d.

For video-based sleep analysis, after acclimation in the recording room for ≥3 d, mice were subjected to continuous video recording for at least 3 d. The video recording data were analyzed by a custom-designed automatic software named SleepV, which used an AI-augmented pattern recognition algorithm to classify the sleep/wake states based on the activity/inactivity of the mouse (28).

SD.

For SD by gentle handling, we kept mice awake by tapping the cage or touching them with a chopstick (Fig. 3A). For SD by curling prevention with water (CPW), we placed individual mouse (with electrode protected with parafilm) in a cage with shallow water at a height of 1.5 cm (37). After 6-h of SD (ZT0-6), mice were subjected to EEG/EMG recording for the subsequent 18 h (ZT7-24).

Accumulated recovery NREMS was calculated as the cumulative changes of NREMS amount following SD minus the baseline NREMS duration at the same ZT time. The increase of NREMS delta power following SD was calculated by the ratio of absolute NREMS delta power at ZT7-the first hour after SD—relative to that of baseline recording.

Simulation of Process S.

The dynamic changes of NREMS delta power of each mouse during a 48 h period, including 24 h baseline sleep and 6 h SD followed by 18 h recovery sleep, was used to simulate process S using Python 3.9.7 with code provided by Masashi Yanagisawa and Hiromasa Funato (17). The time course of process S was calculated iteratively by assuming that it increases according to an exponential saturating function (Eq. 1) during epochs scored as wakefulness or REMS and decreases according to an exponential function (Eq. 2) during epochs scored as NREMS (14, 16).

$$S_{\left(t+1\right)}=\text{UA}-\left(\text{UA}-S_t\right)·e^{\left(-dt/Ti\right)},$$ [1]

$$S_{\left(t+1\right)}=\text{LA}+\left(S_t-\text{LA}\right)·e^{\left(-dt/Td\right)}.$$ [2]

S_(t+1) and S_t are values of S for consecutive 20-s epochs. Ti is the time constant of the increasing exponential saturating function with an upper asymptote (UA), and Td is the time constant of the decreasing exponential function with a lower asymptote (LA). dt is the time step of the iteration (20 s). UA was set to the 99% level of delta power during NREMS, and LA was set to the intercept of delta power distribution of REMS and NREMS by use of Kernel density estimation (14, 16).

The initial value of S (S₀ at lights on of baseline) was determined as the median delta power of NREMS epochs in the first 20 min of recording. The median delta power from NREMS epochs in every 20 min was calculated and used for determining the parameters (Ti, Td) to minimize the mean square of the difference between empirical data and simulation. To save computational effort, initially, the best combination of Ti and Td was selected for a range of Ti (1 to 25 h, 1 h step) and Td (1 to 7 h, 0.25 h step), and then further optimized with finer steps (0.1 h step for both Ti and Td) within the 2 h around the initial combination of time constants. The hourly average of process S was normalized by dividing absolute power in every hour with the mean power during the last 4 h of the light phase of the basal recording.

Behavior Assays.

Test mice were individually housed in a 12-h light/dark cycle (lights on from 09:00 to 21:00), with all behavioral tests starting at ZT1. Behavioral assays were completed within 2 mo, with a rest interval of 4 to 7 d between experiments.

Rotarod test (63): Mice were trained twice before being tested. On the first day of training, mice were placed on rotating rods at 4 rpm for 5 min and, if mice fell off, they were immediately put back on the rotating rods. On the second day of training, the mice were placed on rotating rods that were accelerated continuously from 4 rpm to 40 rpm within 5 min. If mice fell off during this period, the training was stopped and they were returned to the home cage. In the final test, the same acceleration setting for the rotating rods was used, and the latency for each mouse to fall off was recorded three times, with a 15-min rest interval between trials.

Fear conditioning test (64): On the conditioning day, mice were placed into the chamber and allowed to explore freely for 2 min. Cue lights (20 s) and auditory tones (20 s, 7.5 kHz, 85 dB) were played simultaneously four times to exclude the interference of light and sound. After habituation, mice received five light/auditory cues coterminating with a scrambled foot shock (1 s, 0.7 mA). Inter-light/auditory intervals were randomly set between 60 and 120 s. On the test day, mice were placed in the conditioning chamber and allowed to explore freely for 6 min. Five hours later, mice were placed into a nonconditioned chamber with a different floor and background, and allowed to explore freely for 180 s for environment habituation, followed by 180 s of light/auditory stimulus to test cue memory. Contextual and cued fear memory was measured by the time proportion of freezing. The freezing behavior was defined as the absence of movements lasting for ≥2 s, except for those associated with breathing. The video data were analyzed by FreezeFrame 5.

Open field test (65): Mice freely explored a 30 cm × 30 cm × 30 cm square open field for 10 min, with no ground barriers. A camera suspended 10 cm above the arena recorded their path. The total distance of movement and time spent in the central area (15 cm × 15 cm in the center) were calculated by EthoVision XT14 from the video.

Novel object recognition test (66): Mice freely explored a 30 cm × 30 cm test area for 10 min in three consecutive days before the experiment to minimize environmental interference. On the test day, the mice habituated for 2 min. Next, two identical objects (A) were placed diagonally in the exploration area, and mice investigated them for 10 min to familiarize themselves. After a 5-h interval, one object A was replaced with object B. Mice were then allowed to explore for another 10 min. Object exploration was recorded by a camera, and the time spent exploring each object was manually counted when the mouse touched the object. The discrimination ratio was calculated as [(Time spent on object B – Time spent on object A)/(Time spent on object B + Time spent on object A)].

Tail suspension test (67): Tape was used to suspend the tail of the mice from a bracket, positioning the mouse’s nose approximately 15 cm above the apparatus floor. A camera recorded the mice’s movement state 6 min after tail suspension. Immobility time was manually analyzed, including small movements limited to the front legs and oscillations resulting from struggle without limb movements, which were also counted as immobility.

Forced swim test (68): Individually, mice were placed in an acrylic cylinder (20 cm diameter × 30 cm high) filled with water (15 cm deep, 23.5 ± 2.0 °C). Their movement was recorded using a camera for 6 min. Immobility during the last 4 min was manually analyzed, including small movements for balance and drift due to momentum, which were counted as immobility.

Circadian Analysis.

Wheel running analysis (69): Mice were individually housed in a cage with a running wheel. The cage was positioned in a light-proof chamber. After being entrained to a 12-h light:12-h dark cycle for 14 d, the mice were released into constant darkness for 3 wk. Data collection utilized ClockLab data acquisition software (ClockLab, ACTIMETRICS), and subsequent analysis was performed using ClockLab analysis software.

Primary Neuron Culture (PNC).

Cortical neurons obtained from C57B6/J mice at P0-1 were plated onto poly-D-lysine (354210, Corning) coated 12-well tissue culture dishes at a density of 200,000 cells/well. They were grown in neurobasal media (21103-049, Gibco) supplemented with 2% B-27 serum-free supplement (175044-044, Gibco), 2 mM GlutaMAX (35050-61, Gibco), and 50 U/mL PenStrep (151401-22, Gibco). Cultured neurons were fed twice per week. At DIV14, cortical neurons were treated with bicuculline (20 μM, 2 h) (HY-N0219, MCE), KCl (50 mM, 1 h) (60128-250G-F, Sigma), or NMDA (50 μM, 30 min) (HY-17551, MCE) to induce SIK3 S551 dephosphorylation. Fk506 (50 μM, 2 h) (S5003, Selleck) or cyclosporin A (50 μM, 2 h) (HY-B0579, MCE) was either solely added for 2 h or included in the media 10 min before bicuculline or NMDA treatment. After treatment, cells were harvested with SDS-loading buffer for subsequent western blotting analysis.

Immunoblotting.

Mouse brains were quickly dissected and frozen in liquid nitrogen. A whole brain was homogenized by dounce homogenizer with 4 mL of HEPES-buffered sucrose solution (0.32 M sucrose in 4 mM HEPES, pH 7.4), freshly supplemented with protease inhibitors (04693132001, Roche), and phosphatase inhibitors (4906837001, Roche). Immunoblotting was performed according to standard protocols with the following antibodies: anti-HA (1:1,000, H6533, Sigma), anti-CNAα (1:10,000, 13422-1-AP, Proteintech), Anti-CNAβ (1:1,000, 13340-1-AP, Proteintech), Anti-CNAγ (1:1,000, 13422-1-AP, Proteintech), anti-β-actin (1:5,000, AF003, Beyotime), anti-GAPDH (1:5,000, 60004-1-Ig, Proteintech), Anti-CNB1 (1:500, PA529939, Invitrogen), Anti-DNM1 (1:1,000, 18205-1-AP, Proteintech), Anti-DNM1 S774 (1:1,000, ab55324, Abcam), Anti-SIK3 (1:1000, ab227044, Abcam), Anti-SIK3 pS551 (1:1,000, ab225634, Abcam), Anti-phospho-PKA substrate motif (RRXS*/T*) (1:1,000, #9624, CST), Anti-PKAα (1:1,000, 4782S, CST).

Coimmunoprecipitation.

AAVpro 293T cells were transfected with either pCS2-3HA-SIK3 or pCS2-3HA- SIK3^LxVPm2 and cultured in 6-well plate with polyethylenimine MAX. Two days after transfection, cells were harvested with 1 mL of ice-cold PBS and centrifuged at 500 g to obtain cell pellets. The pellets were resuspended in 500 μL of lysis buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 2 mM MgCl₂, 15 mM NaF, 10 mM Na₄P₂O₇), freshly supplemented with protease inhibitors (04693132001, Roche), and phosphatase inhibitors (4906837001, Roche). The following IP steps are similar to the above described.

In Vitro Phosphatase Assay.

Recombinant Flag-CNAα, Flag-CNB1, and phosphorylated HA-SIK3 were overexpressed in AAVpro 293T cells and purified with anti-HA (SA068005, Smart-Lifesciences) or Flag (SA042005, Smart-Lifesciences) antibody-conjugated affinity beads (cells were treated with forskolin if necessary). Mouse brain lysate was extracted with phosphatase reaction buffer (20 mM HEPES, pH 7.5, 10 mM MgCl₂, 2 mM dithiothreitol, 1 mM CaCl₂) supplemented with an additional 0.5% Triton X-100. Approximately 0.7 µg of phosphorylated HA-SIK3, 1.1 µg of Flag-CNAα, and 1.1 µg of Flag-CNB1 were incubated for 30 min at 30 °C in 30 µL of phosphatase reaction buffer freshly supplemented with mouse brain lysate, 100 nM Calyculin A (9902S, CST), and protease inhibitors. Reactions were stopped by adding SDS-loading buffer and boiling at 95 °C for 10 min. Samples were immediately resolved by SDS-PAGE followed by western blotting.

Immunohistochemistry.

Mice were anesthetized with isoflurane and transcardially perfused with ice-cold 10% sucrose in Milli-Q water, followed by ice-cold and 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer saline (PBS), pH 7.4. Dissected brains were postfixed overnight in 4% PFA and then transferred to 30% sucrose in PBS at 4 °C. Then, brains were frozen in Tissue-Tek O.C.T (Sakura) and stored at −80 °C. Brain slices were prepared using a cryostat (Leica Biosystems). After washing with PBS three times, brain slices were treated with permeabilization solution (1% Triton in PBS) at room temperature for 3 h, followed by incubation with blocking buffer (10% Blocking One from Nacalai Tesque with 0.3% Triton X-100 in PBS) at room temperature for 1 h. Brain slices were incubated with the primary antibody in blocking buffer at 4 °C overnight. After further washing with PBS three times, brain slices were incubated with secondary antibody in blocking buffer at 4 °C overnight. First antibodies: 1:2,000 Rat anti-GFP (Nacalai tesque, 04404–84); Second antibodies: 1:500 Donkey anti-rat Alexa Fluor 488 (Jackson ImmunoResearch, 712-545-153).

Statistics and Reproducibility.

All experiments were independently replicated at least twice, with a minimum of three biological replicates. Sample sizes were not predetermined using statistical methods. Randomization and blinding protocols were not implemented, except in cases where technicians performing EEG/EMG-based sleep phenotype analysis were blinded to either the genotypes of mice or the types of AAVs injected. Mean values were presented ± SEM. The number of animals or samples used in each experiment is indicated in the figure legends. Statistical analyses were conducted using appropriate methods: one-way ANOVA for comparisons involving a single variable (e.g., genotype), two-way ANOVA for comparisons involving two independent variables (e.g., genotype and time, phase, or frequency), two-way ANOVA with Sidak's test for multiple independent comparisons, and two-way ANOVA with Dunnett's test for comparisons against a control mean. Hourly plots of NREMS time and NREMS delta power density were analyzed using a mixed-effects model, with matched values from different zeitgeber times stacked. Statistical analyses were performed using Prism 8.0.2 (GraphPad) software.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Movie S1.

Sleep recording of AAV-hSyn-EGFP infected CnB1^floxflox mice. After 2.5 weeks of AAV-hSyn-EGFP injection in CnB1^floxflox mice, sleep/wake state was recorded by video system. This video showed that mice stayed mostly asleep at ZT0.5-ZT1 (9:30-10:00).

Movie S2.

Sleep recording of AAV-hSyn-Cre infected CnB1^floxflox mice. After 2.5 weeks of AAV-hSyn-Cre injection in CnB1^floxflox mice, sleep/wake state was recorded by video system. This video showed that mice stayed mostly awake at ZT0.5-ZT1 (9:30-10:00).

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.