Proteomic analysis of mitochondrial differences in the rat DRN across sleep
The DRN is a center of serotonergic regulation, providing neurotomical evidence that 5-HT serves as “sleep pressure” 3, 4. Proteomic analysis was performed in rat DRN tissues that were collected from three groups of rats: sleep onset (SO) rats that spent at least 80% of the 12 h dark period spontaneously awake and were sacrificed at Zeitgeber time 0 (ZT 0), sleep period (SP) rats that spent at least 80% of the first 6 h of the light period asleep and were sacrificed at ZT 6, and sleep termination (ST) rats that spent at least 70% of the 12 h light period asleep and were sacrificed at ZT 12. Sleep pressure was maximum in SO rats, decreased in SP rats, and reached a minimum in ST rats, consistent with the notion that homeostatic sleep drive becomes stronger during wakefulness and decreases during sleep.
To characterize protein abundance in the rat DRN at different sleep pressure, we examined the expression of 5656 proteins based on liquid chromatography-dual mass spectrometry (LC-MS/MS) quantitative analysis. We detected significant differences in 264 proteins across sleep (Extended Data Fig. 1). Further subcellular localization analyses of differential proteins indicated that most differentially expressed proteins in subcellular structures were found in mitochondria (8.37% [38]; Figs. 1a, b), second only to the cytoplasm and cell membrane (Fig. 1b). Gene Ontology (GO) analysis (Fig. 1c) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation (Fig. 1d) were performed for all the identified proteins. We noted multiple energy metabolism-related terms. Furthermore, the k-means clustering algorithm showed that most DRN proteins changed in various sleep-dependent patterns (Figs. 1e, f). Proteins in the rat DRN showed four patterns: (1) ST higher than SO and SP, (2) significantly higher in SP, (3) significantly lower in SP, and (4) progressively lower with sleep progression. Differentially expressed proteins that were localized in mitochondria were consistent with the four patterns of variation (8, 10, 7, and 13, respectively). Combined with the major differences in mitochondria, we propose that certain biological components that are associated with mitochondria in the rat DRN are able to detect sleep pressure.
Mitochondrial function in the rat DRN is altered with sleep time architecture
Levels of ATP support neuronal activity and the execution of brain function. Aerobic respiration in mitochondria is the most efficient and dominant way to produce ATP. The NADH pathway, in which electrons enter the Q junction via complex I (CI), is more efficient in ATP production, whereas the FADH2 pathway, in which electrons enter the Q junction via complex II (CII), is mainly responsible for supplementary energy synthesis and usually activated in various nonphysiological states that are associated with high energy consumption, such as cancer and stress14.
To further explore whether core biological processes of mitochondria and energy supply in the rat DRN are coupled with sleep pressure, rat DRN tissues were collected at different time points of sleep pressure to assess mitochondrial proton leakage (LEAK), oxidative phosphorylation (OXPHOS) capacity, electron transfer (ET) capacity, and ATP levels. The oxygen consumption rate (OCR) was measured by OROBOROS O2K, reflecting the activity of enzymes in the respiratory chain (Figs. 2a-c).
Basal OCR levels (basal, MT − A, P > 0.05) in DRN tissues remained stable in different sleep states (Fig. 2d). In SP and ST rats, NADH-pathway LEAK (PML, PM − MT, P < 0.05) and OXPHOS capacity (PMP, D − PM, P < 0.01), supported by pyruvate and malate and ET capacity (NE, U − R, P < 0.05), significantly increased compared with SO rats, indicating an overall higher NADH pathway capacity after sleep (Fig. 2d). Upon the further addition of glutamic acid and succinate (G and S, substrates of the FADH2 pathway) and uncoupler CCCP (U), respectively, increases in the OXPHOS (PMGSP, S − MT) and ET (NSE, U − A) respiratory capacities were observed only in SP rats (P < 0.05). Moreover, after inhibiting CI with rotenone (R), OXPHOS capacity (Sp, S − D, P > 0.05) and ET capacity (SE, R − A, P > 0.05) of the FADH2 pathway showed no differences among the three groups (Fig. 2d). These results indicated that mitochondrial OXPHOS (PMGSP) and ET (NSE) respiratory capacities increased with a reduction of sleep pressure, which mainly relied on activation of the NADH pathway in the middle of sleep.
Additionally, oligomycin (O, an inhibitor of ATP synthase) was added after the addition of whole substrate. The difference in OCR before and after adding O reflected ATP synthase activity (ATP synthase, U − O, P > 0.05, Fig. 2d). The results showed that ATPase activity was significantly higher in SP than in SO and ST, which was consistent with ATP levels in the rat DRN among the three groups (P < 0.001, Fig. 2e).
Mitochondrial function was terminated by antimycin (A), an inhibitor of complex III (CIII), when the rate of oxygen consumption represented nonmitochondrial respiration. The nonmitochondrial OCR (A, P < 0.05, Fig. 2c) in the rat DRN was significantly lower than in SO and ST, indicating that respiratory efficiency of the rat DRN was higher during the state of lower sleep pressure. These results indicate that the efficiency of energy metabolism in rat DRN neurons significantly improved during sleep periods.
Mitochondrial morphology in the rat DRN is altered with sleep pressure
The function of mitochondria is closely related to morphology. We used transmission electron microscopy (TEM) to visualize high-resolution mitochondrial ultrastructures in DRN regions from SO, SP, and ST rats (Fig. 3a). More than 11,000 mitochondria, entirely labeled in ~ 55 DRN slides (20 µm ⋅ 15 µm), were measured in n = 4 rats/group (SO, 11,519 mitochondria from 58 slides; SP, 11,038 mitochondria from 54 slides; ST, 11,740 mitochondria from 53 slides). The amounts of mitochondria and their morphological changes across sleep were characterized by calculating the mitochondrial total area and density in each slide and measuring the width, length, and area in individual mitochondria.
High-mitochondrial density slides (> 0.8/µm2) were much more abundant in SP (n = 39) and ST (n = 35) rats with significant increases in mitochondrial density compared with SO rats (n = 5, Fig. 3b), suggesting an increase in the amount of mitochondria during sleep state .The higher percentage of the total area of mitochondria relative to tissue indicated more powerful mitochondrial function, which is associated with high-energy-demanding subdivisions of neurons, such as somatodendrites and synapses15. As sleep pressure decreased, total mitochondrial areas significantly decreased among the lower mitochondrial area percentage slides (< 8%) but increased among the higher mitochondrial area percentage slides (> 8%, Fig. 3c). These results suggest that mitochondria in the rat DRN tend to be divided and proliferate with a more efficient profile across sleep.
Individual mitochondrial areas after 12 h of light-period asleep decrease by an average of 16.1% relative to spontaneous wakefulness (P < 0.05, Fig. 3d). There was also a leftward shift of the mitochondrial population toward smaller sizes in SP and ST rats compared with SO rats (Fig. 3e). This shift occurred preferentially in mitochondria of small sizes (< 0.1 µm2) in SP groups, which together accounted for ~ 60% of all mitochondria. However, the ST group exhibited an overall shift to the left relative to the SO group, suggesting that the decrease in mitochondrial sizes during sleep followed a fission dynamic trend.
Mitochondria are highly dynamic organelles that undergo fission and fusion to maintain their size and shape. With regard to observing dynamic changes in mitochondrial morphology across sleep, individual mitochondria were defined as one of three types: small (S type, < 0.1 µm2, a consequence of the mitochondrial fission process with immature function), large with a round shape (LR type, > 0.1 µm2 and width/length > 0.5, presenting robust function based on continuity of the inner membrane16, 17), and large with a long shape (LL type, > 0.1 µm2 and width/length < 0.5, which appear to be in a dynamically active state with a lower capacity to generate ATP than the LR type). The proportion of small mitochondria in the DRN gradually increased from 58% (at SO) to 64% (at ST) with the prolongation of sleep (Fig. 3g). Consequently, the proportion of large mitochondria in the DRN decreased with a reduction of sleep pressure, but with different trends in mitochondrial shapes between the first and second 6 h of light-period sleep. During the first 6 h, the reduction of large mitochondria mainly occurred in the LL type (Fig. 3g), implying that mitochondrial fission was augmented during sleep. In contrast, during the second 6 h, the reduction of large mitochondria mainly occurred in the LR type, with a continuous increase in the S type (Fig. 3g), suggesting that activation occurred in both mitochondrial fission and fusion. These results indicated that sleep duration and sleep pressure were highly integrated with the mitochondrial dynamic process.
Mitochondrial dynamics-related protein expression in the rat DRN is coupled with sleep pressure
Mitochondrial fission and fusion dynamics processes are both mediated by large guanosine triphosphatases that belong to the dynamin family. Dynamin-related protein 1 (Drp1) interacts with fission 1 (Fis1) to constrict mitochondria, resulting in the division of a mitochondrion into two separate organelles. Mitofusin 1 (Mfn1) and Mfn2 are responsible for the fusion of outer mitochondrial membranes, and optic atrophy 1 (OPA1) is responsible for the fusion of inner mitochondrial membranes16, 17. Previous studies showed that there is a circadian rhythm in the expression of Drp1 in the mouse brain18. The OXPHOS and ATP production of mitochondria in the brain are coupled by regulating mitochondrial dynamics.
We investigated whether these mitochondrial dynamics-related proteins in the DRN change as a coupled phenomenon of sleep homeostasis (Fig. 4a). These proteins in the rat DRN were detected every 3 h from ZT 0 to ZT 21, and sleep homeostasis was assessed by electroencephalographic (EEG) delta (0.5-4 Hz) power density in 1-h blocks across 24 h. Mitochondrial dynamics-related proteins in the rat DRN showed time-of-day-dependent variations. Specifically, Drp1 levels gradually increased and maintained a maximum during the first and second 6 h of light-period sleep, followed by a gradual decline during 12 h of dark-period wakefulness, showing an inverted U-shape curve (P < 0.001, Fig. 4b). Similarly, the expression curve of Fis1 resembled an inverted V shape, with a maximum level at ZT 6 (corresponding to SP, P < 0.01, Fig. 4c). In contrast, Mfn1 (P < 0.01, Fig. 4d) and Mfn2 (P < 0.01, Fig. 4e) levels expressed a J-shaped curve, with a gradual decline during the first period of sleep followed by a dramatic increase that reached a maximum at ZT 21 (corresponding to wake termination). OPA1 levels did not exhibit circadian rhythms but showed a similar increase as Mfn1/2 during the dark-period of wakefulness (P > 0.05, Fig. 4f).
The delta power of EEG during NREMS is characterized by physiological indicators of sleep pressure19. Consistent with previous reports, delta power density during NREMS, which was analyzed in 1-h blocks, gradually decreased during light-period sleep and reached a minimum at ZT 11–12 (corresponding to ST, P < 0.01, Figs. 4g, h) as a result of a reduction of sleep pressure with a prolongation of sleep time. It then gradually rebounded during the dark period of wakefulness, suggesting that sleep pressure increased as a prolongation of wakefulness. This fluctuation in delta power density across sleep homeostasis exhibited a sharp V-shaped curve. Interestingly, the V-shaped curve of sleep pressure was temporally coupled with inverted V- or U-shaped curves of Drp1 and Fis1 and J-shaped curve of Mfn1 and Mfn2.
Next, we attempted to dissect underlying interactions between mitochondrial dynamics and sleep homeostasis based on the present findings. During the first 6 h of light-period sleep, as sleep pressure decreased, Drp1 and Fis1 gradually increased and Mfn1/2 gradually decreased in the rat DRN, which together promoted mitochondrial division and proliferation. Consequently, the proportion of small-size mitochondria and total area of mitochondria increased, which were accompanied by the activation of mitochondrial respiratory chain enzymes and ATP synthesis. The electrical activity of serotonergic neurons in the DRN20 and release of 5-HT in many brain regions21 decrease during sleep and increase during wakefulness. From the perspective that mitochondria are highly dynamic to allow adaptive energy production22, functional and morphological changes in mitochondrial during the first 6 h of light-period sleep appeared to be dissociated with a low-energy-demanding condition in the DRN. Therefore, we speculate that these changes in mitochondria are preparing for high-energy expenditure for upcoming conditions (e.g., microarousals or wakefulness), which also can be explained as a mitochondrial dynamic process that drives sleep homeostasis.
During the second 6 h of light-period sleep, because Drp1 and Fis1 are still maintained at a high level, mitochondria remain in a trend of fission as a result of the proportion of small-size mitochondria further declining in ST rats. These mitochondria also conserved the high capacity of NADH-pathway ET and OXPHOS. Therefore, we presumed that this mitochondrial dynamic fission during this period still drives sleep homeostasis and causes the reduction of sleep pressure. Additionally, Mfn1/2 levels began to rebound from ZT 6 to ZT 9 and induced the mitochondrial fusion process. Indeed, we observed a higher proportion of LL-type mitochondria in ST rats compared with SP rats. Notably, these mitochondrial changes might be associated with wakefulness because total sleep time that was analyzed in 1-h blocks showed that it began to decrease during ZT 11–12 (P < 0.01, Extended Data Fig. 2b), which also caused the reduction of ATP in ST rats (Fig. 2e). We further predicted that mitochondrial fusion might be a more predominant dynamic trend in the rat DRN during dark-period wakefulness because Drp1 and Fis1 decreased and Mfn1/2 increased. These mitochondrial fusion processes might drive the accumulation of sleep pressure during wakefulness.
Microinjection of Mdivi-1 promoted mitochondrial fusion and FADH2-pathway-dependent capacity in the rat DRN
The increases in mitochondrial division and OXPHOS capacity in the DRN are accompanied by a decrease in sleep pressure, suggesting that mitochondrial morphology and function in the DRN may perceive sleep pressure. Consequently, a factor that causes alterations of mitochondrial morphology and function in the DRN may affect sleep and wakefulness. To confirm this possibility, mitochondrial division inhibitor 1 (Mdivi-1), an inhibitor of Drp1, was microinjected in the rat DRN at ZT 0. We observed the effects of Mdivi-1 on mitochondrial morphology (Fig. 5) and function (Fig. 6) 3 h after the microinjection (ZT 3).
The microinjection of Mdivi-1 in the rat DRN at a low dose (0.5 ng) did not significantly affect mitochondrial morphology compared with vehicle (Figs. 5a-g). However, at moderate and high doses (2 and 5 ng) of Mdivi-1, a progressive mitochondrial fusion dynamic trend was observed in the DRN as a result of reductions of mitochondrial densities (i.e., amounts, Fig. 5b) but increases in total (Fig. 5c) and individual (Fig. 5d) mitochondrial areas. Consequently, there were overall rightward shifts of the mitochondrial population toward larger sizes in the Mdivi-1-treated groups (2 and 5 ng) compared with the vehicle group (Fig. 5e).
Furthermore, the intra-DRN administration of Mdivi-1 (2 and 5 ng) reduced the proportion of S-type mitochondria by 5–11% compared with vehicle. These changes were accompanied by increases in the proportion of large-type mitochondria, particularly the LR type, especially in the moderate dose (2 ng) Mdivi-1 group, in which this proportion was up to 37% and dramatically higher than in the vehicle group (26%). Considering that LR-type mitochondria present robust functions, we presumed that these mitochondrial dynamics changes that were induced by Mdivi-1 might serve in high energy-demanding conditions in the DRN.
We observed robust mitochondrial function that was induced by Mdivi-1 (2 and 5 ng) in the rat DRN by assaying OXPHOS capacity, ET capacity, and ATP levels (Figs. 6a-d). The microinjection of a moderate dose of Mdivi-1 (2 ng) in the rat DRN significantly increased basal OCR levels (basal, P < 0.05), the ET ability of the FADH2 pathway (SE, P < 0.05), total ET capacity (NSE, P < 0.05), and ATP levels (P < 0.05) compared with vehicle (Fig. 6d). In the high-dose Mdivi-1 group, a significant increase in NADH-pathway LEAK was found (NL, P < 0.05), suggesting a facilitatory effect on NADH pathway-dependent OXPHOS reserve capacity. However, FADH2 pathway-dependent ET capacity and ATP levels in these rats were not as prominent as in the moderate-dose Mdivi-1 group.
Altogether, these results suggest that the mitochondrial division process during the light-phase sleep was inhibited by the microinjection of Mdivi-1 (2 and 5 ng) in the rat DRN. Furthermore, the moderate dosing (2 ng) of Mdivi-1 promoted mitochondrial proliferation and fusion and a significant improvement in FADH2 pathway ET capacity. We presumed that these mitochondrial morphological and functional changes that occurred in the DRN could accumulate sleep pressure and finally affect sleep-wake behavior.
Microinjection of Mdivi-1 in the rat DRN changed sleep structure and pressure
Next, using EEG spectral analysis, we monitored rats’ sleep-wake stages for 6 h (ZT 0–6) immediately after Mdivi-1 was microinjected in the DRN. Compared with vehicle, microinjections of moderate and high doses of Mdivi-1 in the DRN significantly increased the time spent in wakefulness (W, P < 0.001) and decreased total sleep (TS) time (P < 0.001, Fig. 7a), indicating that inhibition of the mitochondrial fission process in the DRN during light-phase sleep promoted wakefulness and suppressed sleep.
We then analyzed sleep stages and EEG spectra. Intra-DRN Mdivi-1 (2 or 5 ng) administration slightly decreased NREMS time (P < 0.05, Fig. 7a) but increased delta power density during NREMS, particularly in the high-dose group (P < 0.05, Figs. 7c, d). These increases in delta power density during NREMS were maintained during the entire 6-h recordings, reflected by delta power density-time the curve in the Mdivi-1-treated groups (2 and 5 ng) that were completely above the curve in the vehicle-treated group (P > 0.05 or P < 0.05, Fig. 7e). NREMS was divided into light sleep (LS, < 70%) and slow-wave sleep (SWS, ≥ 70%) based on the delta power ratio of EEG. Rats that were treated with the moderate dose of Mdivi-1 spent more time in SWS (P > 0.05, Fig. 7a), which further caused the percentage of SWS relative to TS to significantly increase (P < 0.05, Fig. 7b). Delta power during NREMS and SWS are the best characterized physiological indicators of sleep pressure19. Therefore, these results support the possibility that sleep pressure accumulates when mitochondrial fission was inhibited by microinjection of Mdivi-1 in rat DRN.
We investigated why sleep pressure is perceived by mitochondrial dynamic changes in the DRN, which might be associated with serotonergic neuron-activating conditions that depend on energy supply from mitochondria. These neurons are wake-active, showing higher activity during wakefulness over sleep states20. During dark-phase wakefulness, the mitochondrial fusion process is predominant for supplying ATP to maintain the high activity of neurons in the DRN, which is associated with the accumulation of sleep pressure in rats. These rats were supposed to undergo a sleep state during the upcoming light phase, but this transition was dramatically disturbed by intra-DRN Mdivi-1 administration. Treatment with Mdivi-1 promoted mitochondrial fusion and augmented mitochondrial function by improving the FADH2 pathway, which could facilitate continuous neuronal activation at a high level. Consequently, these rats spent more time in wakefulness, with a consequent buildup of sleep pressure, with finally an enhanced NREMS intensity when sleep was permitted. However, Speijer reported evidence that a high FADH2/NADH ratio during aerobic respiration is a crucial determinant of reactive oxygen species formation14, 23. This indicates that this stressful increase in mitochondrial function may cause oxidative stress-induced damage while increasing energy supply.
Although the rats in both the moderate- and high-dose Mdivi-1 groups exhibited wake-promoting and NREMS-enhancing effects, sleep structure in these two groups was different. The rats in the high-dose group exhibited a shortened sleep latency (P < 0.01, Fig. 7a), indicating these rats immediately fell asleep after Mdivi-1 was microinjected in the DRN. This might be attributable to the dramatically high level of sleep pressure that was caused by the acute inhibition of mitochondrial fission. Additionally, rapid-eye-movement sleep (REMS) time (P < 0.01, Fig. 7a) and the percentage of REMS relative to TS significantly decreased in these rats (P < 0.01, Fig. 7b). This might suggest that serotonergic neurons, which are recognized as REM-off neurons20, were much more active by consuming large amounts of ATP when they were exposed to high-dose Mdivi-1.
Mdivi-1-induced sleep changes occurred through the serotonergic system
To confirm the effects of mitochondrial dynamics on the serotonergic system, we detected serotonergic neuron activity and serotonin metabolism in Mdivi-1-treated rats using double-staining immunofluorescence and high-performance liquid chromatography with electrochemical detection (HPLC-ECD, Fig. 8). Brain tissues of the DRN and its projecting regions were collected 3 h after Mdivi-1 was microinjected in the DRN. Tryptophan hydroxylase 2 (TPH2) expression is a marker of serotonergic neurons in the brain24, 25. C-Fos expression26, 27 is often considered an index of neuronal activation (Figs. 8a, b). Microinjections of Mdivi-1 (2 or 5 ng) in the DRN slightly but nonsignificantly increased the density of serotonergic neurons (P > 0.05, Fig. 8c) and significantly increased the c-Fos-positive ratio of serotonergic neurons (P < 0.01, Fig. 8d), suggesting that the mitochondrial fusion process was associated with the activation of serotonergic neurons in the DRN. The hyperactivity of serotonergic neurons also promoted 5-HT production, with a significant increase in 5-HT content in the DRN in Mdivi-1-treated (2 or 5 ng) rats (P < 0.01, Fig. 8e).
Serotonergic neurons in the DRN project to the HT, hippocampus (HP), basal forebrain (BF), and VLPO, together forming the critical serotonergic system in the regulation of arousal2, 28. Microinjections of Mdivi-1 (2 or 5 ng) increased 5-HT in the HT (P < 0.05), HP (P < 0.05), VLPO (P < 0.01), and BF (P < 0.01), and this effect was more significant in high-dose Mdivi-1-treated rats (Fig. 8f). When the mitochondrial fusion process was augmented by high-dose Mdivi-1, its potential effects on the serotonergic system appeared to not only be restricted to the DRN but also spread throughout other brain regions that regulate sleep-wake states.
Altogether, the present study provided evidence that mitochondrial dynamics in the DRN drive sleep homeostasis by modulating serotonergic function in endogenous sleep-wake-regulating pathways. The mitochondrial fission process in the DRN is associated with a reduction of sleep pressure during light-phase sleep. When the fission process is inhibited and turns into a moderate fusion process, which is also accompanied by improvements in mitochondrial OXPHOS, wake-promoting and NREMS-enhancing effects occur through the activation of serotonergic function in endogenous sleep-wake-regulating pathways.