Abstract
Myelination depends on maintenance of oligodendrocytes that arise from oligodendrocyte precursor cells (OPCs). We show that the dynamic nature of oligodendroglia and myelination are regulated by the circadian transcription factor BMAL1. Bmal1 knockdown in OPCs during development – but not adulthood – decreases OPC proliferation, whereas BMAL1 regulates OPC morphology throughout life. OPC-specific Bmal1 deficiency impairs remyelination in an age-dependent manner, suggesting that age-associated decrements in circadian regulation of oligodendroglia may contribute to the deficient remyelination potential in demyelinating diseases like multiple sclerosis (MS). This oligodendroglial dysregulation and dysmyelination increase sleep fragmentation in OPC-specific Bmal1 knockout mice, and sleep fragmentation is causally associated with MS. These findings have broad mechanistic and therapeutic implications for numerous brain disorders that include both myelin and sleep phenotypes.
One-Sentence Summary BMAL1 regulates the homeostatic maintenance of oligodendroglia and myelin, that subsequently controls sleep architecture.
Main Text
Myelin ensheathes axons to facilitate efficient transduction of electrical signals and metabolic support of neurons (1, 2). Myelin-forming oligodendrocytes arise from oligodendrocyte precursor cells (OPCs), which are evenly distributed throughout the entire central nervous system (CNS). OPCs are highly dynamic as the most mitotic cells in the CNS (3, 4) with elaborate morphology that promotes motile filopodia and migration. OPC proliferation is stimulated when oligodendroglial loss initiates adjacent OPCs to divide (5), in response to neuronal activity in some neural circuits (6), and through biophysical and spatial constraints of their microenvironment (7). Even though OPCs are a functionally, spatially, and temporally heterogenous precursor population (8–10), they have a remarkable ability to maintain this consistent homeostatic density throughout the brain. How this unique harmony between prolific cellular self-renewal and population level homeostasis is achieved in normative brain health and disrupted in myelin-associated diseases like multiple sclerosis (MS) remains to be fully determined.
From cyanobacteria to humans, temporally dynamic mechanisms are imperative to the maintenance of homeostatic states and behaviors. This is afforded to organisms through the evolution of the circadian system which allows for biological processes to occur at the proper time of day. At the cellular level, circadian rhythms are generated ubiquitously throughout the body by a molecular transcriptional/translational feedback loop that has a period of ∼24 hours. Briefly, the products of the core clock genes—Clock and Bmal1—heterodimerize and drive the transcription of the clock genes families Period (Per) and Cryptochrome (Cry). Accumulated levels of PER and CRY within the cytoplasm feed back into the nucleus, displacing the CLOCK and BMAL1 heterodimer and consequently disrupting their own transcription (11). This transcriptional machinery regulates cytoskeletal factors (12), cell cycle (13), and metabolism (14) in numerous cell populations. OPCs proliferate on a circadian cycle (4) and sleep deprivation negatively impacts OPC proliferation and differentiation (15). The potential role this circadian clock plays in regulating the homeodynamic nature of oligodendroglial lineage cells remains unknown. We posit that the molecular circadian system driven by the transcription factor BMAL1 regulates oligodendroglial cells and myelination, which contribute to the maintenance of systems-level homeostatic processes such as sleep.
Results
Functional knock down of circadian gene Bmal1 dysregulates OPC proliferation and morphology
To determine the role of the BMAL1-driven molecular clock in regulating OPC biology, we first confirmed expression of the complete molecular clock machinery throughout the oligodendroglial lineage using the RNAseq transcriptome database of cell type-specific gene expression in mouse brain (16, 17) (Fig. S1), and we confirmed BMAL1 expression in mouse OPCs (Fig. 1A-B). We next developed a conditional clock gene knockout (Bmal1fl/fl) and cell type-specific Cre driver mouse model (NG2::Cre) to constitutively eliminate functional Bmal1 from OPCs during embryonic development (Fig. 1C). BMAL1 levels were significantly decreased in OPCs isolated from postnatal day (P)6 NG2::Cre+;Bmal1fl/fl (OPC-Bmal1-KO) compared to NG2::Cre-;Bmal1fl/fl (OPC-Bmal1-WT) mice (Fig. 1D). We hypothesized that BMAL1 loss in OPCs would disrupt OPC dynamics given that it regulates cell cycle, proliferation and cytoskeletal factors in other cells (12, 13). Mice were injected with the thymidine analogue 5-ethynyl-2’-deoxyuridine (EdU) on P18-20 to mark newly proliferated OPCs. EdU+/PDGFRα+ OPCs were significantly reduced in the corpus callosum of P21 OPC-Bmal1-KO compared to OPC-Bmal1-WT mice (Fig. 1E-F). Bmal1 loss led to a significant decrease in OPC density of OPC-Bmal1-KO compared to OPC-Bmal1-WT mice (Fig. 1G-H) that persisted into adulthood (P63; 5527 ± 221.7 vs 4289 ± 355 cells/mm3; n = 4-6 mice; P < 0.05). These cellular differences did not result in corpus callosum volume changes (Fig. S2A). We also identified a significant reduction in OPC morphological complexity by evaluating the length, volume, and branching points of processes in OPC-Bmal1-KO compared to OPC-Bmal1-WT mice (Fig. 1I-J).
OPC-specific Bmal1 knock down dysregulates oligodendrocytes and myelination
Given the striking effects of BMAL1 loss on OPCs, we next tested how loss of Bmal1 affects myelinating oligodendrocytes. Because a decrease in OPC density impacts the progenitor pool available for oligodendrogenesis and considering oligodendrocytes also lack functional BMAL1 in our model due to embryonic knock down, we evaluated the effect of OPC-specific Bmal1 loss on oligodendrogenesis. OPC-Bmal1-KO mice exhibit a significant decrease in density of CC1+ oligodendrocytes and myelin basic protein (MBP) content in the corpus callosum at P21, an age that corresponds to the end of developmental myelination (Fig. 2A-C). Using transmission electron microscopy (TEM), we observed thinner myelin sheaths quantified as an increase in the g-ratio (g-ratio = inner axon diameter/total fiber diameter) of axons projecting from the cortex into the corpus callosum of the premotor circuit in OPC-Bmal1-KO compared to control mice (Fig. 2D-F, Fig. S3). This decrease in myelin sheath thickness was found in small and medium but not large caliber axons (Fig. 2E). As NG2 is a proteoglycan that is not only expressed in OPCs but also in pericytes, Bmal1 loss in NG2+ cells could potentially compromise the blood-brain barrier (BBB), leading to brain inflammation and neurotoxicity promoting oligodendroglia and myelin loss. Brain-wide loss of Bmal1 has previously been linked to BBB hyperpermeability associated with pericyte dysfunction and dysregulation of platelet-derived growth factor receptor β (18). To rule out this possibility, we tested the integrity of the BBB through administration of sodium fluoresceine and found no differences in brain permeability between genotypes at P21 (Fig. S2B).
Loss of Bmal1 in oligodendroglial lineage cells disrupts motor and cognitive function
We next evaluated if these cellular and myelin decrements contributed to alterations in motor and cognitive behaviors known to be affected by changes in frontal lobe white matter (19, 20). The identified decrements in OPC-Bmal1-KO mice produced a significant dysregulation of motor behavior at P35, including a decrease in paw swing speed and stride length assessed using the CatWalk gait analysis system (Fig. 2G-H). We also tested cognition using a modified novel object recognition test (NORT) that places greater emphasis on attention and short-term memory instead of hippocampal-dependent long-term memory (19, 20). OPC-Bmal1-WT mice spent more time investigating the novel over familiar object while OPC-Bmal1-KO mice did not discriminate between the objects during the testing phase, suggesting deficits in white matter-associated cognition (Fig. 2I-J).
BMAL1-mediated regulation of OPCs proliferation and differentiation in vitro
To understand the molecular mechanism(s) through which BMAL1 controls OPCs, we first confirmed rhythmic clock gene expression in OPCs isolated from P6 Per2-Luciferase mice that express luciferin under the control of Per2. We synchronized OPCs in vitro using 100 nM dexamethasone and confirmed circadian synchronization by measuring luminescence every 4 hrs (Fig. 3A). The expression rhythmicity of the clock genes Bmal1, Per2, and Rev-Erbα (a transcriptional repressor of Bmal1) was dampened in OPCs isolated from OPC-Bmal1-KO (Fig. 3B) compared to rhythmic expression in OPC-Bmal1-WT mice, confirming global circadian dysregulation of clock gene expression in Bmal1 knockout OPCs. We then interrogated if the cellular effects of Bmal1 loss in OPCs were a consequence of decreased OPC number exclusively or potentiated by a dysregulation of their differentiation potential. In OPCs isolated from OPC-Bmal1-KO mice, we found a 14% decrease in proliferating EdU+/PDGFRα+ cells in Bmal1 knockout compared to control OPCs (Fig. 3C-D). We then induced OPC differentiation for 3 days and found a decrease in MBP+ oligodendrocytes when Bmal1 was knocked down (Fig. 3E-F) that recovered when differentiation was completed after 6 days in both genotypes (Fig. 3G-H).
As the Bmal1-driven circadian clock can regulate from 10-50% of the genome (21), it is possible that the effects observed in the OPC-specific Bmal1 knockout are due to a variety of downstream pathways controlled by BMAL1. BMAL1-associated OPC deficits are primarily related to decreased expression in genes linked with cytoskeletal regulation (Actb and Arpc2; Fig. 3I,) cell cycle (cMyc, Tp53 and Cdk1; Fig. 3J), and proliferation (Sox5; Fig. 3K), but not differentiation (Sox10, Olig2 and Id2; Fig. 3L).
Effect of BMAL1 disruption in OPCs during adulthood
As the two major waves of developmental OPC migration that begin OPC density establishment occur during mid-embryonic development (22) and developmental oligodendrogenesis is completed before early adulthood in mice, we hypothesized that Bmal1 disruption in OPCs in adulthood would result in less dramatic effects on OPC dynamics. To investigate this, we induced knock out of Bmal1 in OPCs using a conditional, inducible OPC-specific Cre mouse line PDGFRα::CreERT2 crossed with Bmal1fl/fl. PDGFRα::CreERT2+;Bmal1fl/fl (OPC-Bmal1-iKO) and PDGFRα::CreERT2-;Bmal1fl/fl (OPC-Bmal1-WT) mice were injected intraperitoneally with 100 mg/kg tamoxifen for 3 consecutive days at 3 months (Younger Adults) or 10 months (Older Adults) of age and brains were assessed 6 weeks later. We obtain approximately 80% recombination following this 3 day tamoxifen schedule (19), and found no evidence of ‘leaky’ Cre expression in neurons in this Cre driver (23). We found that OPC density is no longer decreased, but morphological complexity is still disrupted when Bmal1 knock out is induced in younger and older adults (Fig. 4A-B) rather than embryonically as above. Taken together, our findings illustrate a key role for Bmal1 in oligodendroglial lineage homeostasis and function during development and in adulthood.
Remyelination potential of oligodendroglial lineage cells lacking Bmal1
Deficits in OPC density and differentiation are underlying causes for limited remyelination in MS (24). Given that Bmal1 loss in OPCs results in pronounced cellular changes during developmental myelination, we hypothesized that the OPC circadian clock is necessary for proper cellular recovery following a demyelinating injury. We generated a unilateral focal demyelinating lesion in the cingulum of the corpus callosum with stereotactic injection of lysolecithin 6 weeks after tamoxifen injections at 3 or 10 months of age in OPC-Bmal1-iKO and OPC-Bmal1-WT mice (Fig. 4C-D). Lysolecithin-induced lesions progress from an active demyelination phase during the first 3 days, to OPC recruitment (days 3 to 7), differentiation (days 7 to 10), and active remyelination (days 10 to 21) (25). We first evaluated OPC proliferation and density in the lesion 5 days post-lysolecithin injection (dpi). When Bmal1 knockout and demyelination occur in young adulthood, intra-lesion OPC density is significantly lower in OPC-Bmal1-iKO mice than OPC-Bmal1-WT mice. This difference was not attributed to OPC proliferation changes (Fig. 4E-F). By 9 dpi the OPC density difference no longer exists between genotypes, but the number of lesion-associated oligodendrocytes compared to the non-lesioned hemisphere is significantly decreased in OPC-Bmal1-iKO compared to OPC-Bmal1-WT young adults (Fig. 4H-I). Twenty days after demyelination, remyelination of small caliber axons is reduced in OPC-Bmal1-iKO compared to OPC-Bmal1-WT (Fig. 4K-L). When Bmal1 loss and demyelination are induced in older adults, the significant differences between genotypes in OPCs at 5 dpi (Fig. 4G), oligodendrocytes at 9 dpi (Fig. 4J) and remyelination at 20 dpi (Fig. 4L) were not found, suggesting that the ability of OPCs to dynamically respond to a demyelinating lesion is impacted by both age and BMAL1 status.
Disruption of BMAL1 in oligodendroglia is associated with sleep fragmentation in mice
As the global elimination of Bmal1 markedly disrupts circadian and sleep processes in mice (26, 27) and given the robust effects of even small changes in myelination on brain-wide neural circuit dynamics (28), we tested the hypothesis that Bmal1 disruption of the oligodendroglial lineage may lead to systems-level dysregulation of homeostatic circuits regulating circadian and sleep processes. We thus assessed how constitutive Bmal1 loss in OPCs affects global circadian rhythmicity and sleep during adulthood. We subjected OPC-Bmal1-WT and OPC-Bmal1-KO mice to a 12::12 light/dark (LD) cycle for 7 days to assess circadian entrainment followed by constant darkness (DD) for 15 days to assess free-running circadian rhythms (Fig. 5A). OPC-Bmal1-KO mice exhibit normal circadian periods (tau) compared to controls (Fig. S4A) with no genotype effect on locomotor activity throughout the day (Fig. S4B-C). At 3.5 months of age, we implanted electroencephalogram (EEG) electrodes into the cortex of OPC-Bmal1-WT and OPC-Bmal1-KO mice to measure sleep waves across the hippocampus and evaluated baseline sleep recordings in light::dark cycles. In mice, sleep is polyphasic, and around two thirds of these short sleep periods occur during the light/rest cycle. Sleep is characterized by two main stages consisting of rapid eye movement (REM) and non-REM (NREM) sleep. The general pattern of sleep/wake is not affected in mice with Bmal1-disrupted OPCs, as the total time spent awake does not vary between genotypes during the light or dark phases (Fig. 5B left). However, when we evaluate sleep architecture, Bmal1 loss in OPCs leads to significant sleep fragmentation during the dark/active phase; OPC-Bmal1-KO mice have 29% shorter but 57% more frequent wake events (Fig. 5C-D left, H). During the dark/active phase, NREM events are also shorter but more frequent even though the total time spent in NREM does not differ (Fig. 5E-G left, H). REM sleep does not vary between genotypes (Fig. S5A-C). To further elucidate the effect of OPC-specific BMAL1 on sleep homeostasis, we subjected the mice to a 6-hr sleep deprivation challenge. The effect on sleep fragmentation is exacerbated in OPC-Bmal1-KO mice following sleep deprivation as wake events during the dark phase are 49% shorter and 95% more frequent than controls (Fig. 5B-D right). The total time spent in NREM during the dark/active phase is significantly increased in OPC-Bmal1-KO compared to OPC-Bmal1-WT mice and the events are shorter and more frequent, following the same pattern as baseline (Fig. 5E-G right). Taken together, these findings show that the genetic disruption of Bmal1 in oligodendroglial lineage cells leads to increased sleep fragmentation and a greater drive for restorative NREM sleep following sleep deprivation, illustrating a role for oligodendroglia in sleep function.
Association of sleep fragmentation with multiple sclerosis in humans
Given that dysmyelination in OPC-Bmal1-KO mice is associated with increased sleep fragmentation, could changes in sleep fragmentation be associated with myelin disorders in humans? We evaluated the causal association of changes in sleep on the risk of MS. We performed Mendelian randomization (MR) analyses on lead variants identified in a GWAS of sleep fragmentation (defined as number of sleep episodes) in 85,723 UK Biobank individuals (29). We evaluated their association with MS risk by leveraging results from a GWAS of MS in the International Multiple Sclerosis Genetics Consortium (N = 47,429 MS patients and 68,374 controls) (Table S1) (30). Eleven variants associated with number of sleep episodes/sleep fragmentation were suitable for MR analysis after exclusions (Table S1). We identified that sleep fragmentation is causally associated with an increased risk of MS (Inverse Variance Weighted OR = 1.11 [1.01-1.23], P = 0.034, per unit increase in the number of sleep episodes, 95%; Fig. 5I). Analysis using statistically weaker methods provided consistent causal and statistically significant effect at an α of 0.05 with Weighted median method (OR = 1.17 [CI = 1.03-1.32], P = 0.012, Fig. 5I). Furthermore, MR Egger analysis, which estimates the total pleiotropic effect of the instruments used, had a causal effect size with consistent direction of effect to the other four methods indicating that there was no significant pleiotropy detected (MR Egger Intercept P = 0.32; Fig. 5I). Further analysis using all variants without correcting for winner’s curse provided consistent causal and significant effect at an α of 0.05 with Weighted median method (OR = 1.14 [CI = 1.03-1.27], P = 0.016, Fig. S6). Taken together, these findings indicate a previously underappreciated association between sleep fragmentation and myelin disease.
Discussion
The results presented here demonstrate a critical role for BMAL1 in modulating OPC dynamics and myelination throughout development, adulthood, and in disease. We show that BMAL1 is not only necessary for the dynamic nature of oligodendroglial lineage cells and myelination but also for the homeostatic regulation of sleep architecture. By eliminating Bmal1 from OPCs, we find a significant deficit in density, proliferation, and morphological complexity, leading to a decrease in oligodendrocytes. These cellular deficits translate into thinner myelin sheaths and decrements in motor and cognitive functions associated with white matter structures (20). Importantly, our results suggest this oligodendroglial decrease is related to precursor population depletion more than altered differentiation, exemplified by diminished gene expression of cell cycle and proliferation regulators. This observation is in agreement with the known role of BMAL1 in regulation of cell cycle checkpoints in other cells (13). Previous work identified increases in transcripts that control OPC proliferation during the sleep/light phase (15), a period when Bmal1 peaks in mice. As OPCs are the most consistently mitotic cells in the CNS (3, 4), even small changes in proliferation can disrupt their homeostatic density. The oligodendrocytes that arise from these OPCs will also putatively lack functional BMAL1 because of its constitutive loss in OPCs during embryonic development. While our in vitro studies indicate that BMAL1 regulates OPC proliferation and division more so than differentiation, future studies specifically targeting knock down of Bmal1 in oligodendrocytes will further stratify the role of BMAL1 in myelination.
Even though BMAL1 continues to regulate OPC morphology throughout life, the effect on OPC density is abrogated when BMAL1 elimination is initiated during adulthood. As BMAL1 regulates the OPC cell cycle, and the mitotic rate of OPCs decreases with age (9), this finding further supports the role of BMAL1 in controlling OPC division. OPCs are morphologically dynamic with surveillance of their microenvironment. This allows them to not only establish and maintain their equal distribution but also to respond to loss of adjacent OPCs (5). The decrease in Actb expression in OPCs following BMAL1 loss suggests that BMAL1 contributes to the regulation of cytoskeletal factors and complexity in OPCs, similar to other glial cells (12, 31). Collectively, these data suggest that BMAL1 loss in OPCs could be associated with accelerated aging as both declining OPC proliferation and density and changes in morphology are linked to aging and aging-related brain disorders (32, 33). It should also be noted that although BMAL1 acts as a transcription factor coordinating clock-controlled genes expression (21), the majority of BMAL1 targets do not show rhythmic transcription patterns (34). While our in vitro findings support that OPCs are circadian dysregulated, whether BMAL1 function in OPCs is strictly circadian is an open area of investigation. Future work focused on other clock genes will distinguish non-circadian BMAL1-regulated pathways in oligodendroglia.
Previous work demonstrated that changes in astrocyte circadian clock within a demyelinating lesion signal suppression of subventricular zone BMAL1 in neural precursor cells, driving them towards the oligodendrocyte lineage (35). However, this work did not investigate the specific effect of BMAL1 loss on oligodendroglia in remyelination. The data presented here imply that BMAL1 controls OPC recruitment rather than proliferation of existing OPCs, a critical step for remyelination (36) that is deficient in MS (37). Decreased expression of the migratory factors Arpc2 (38) and Sox5 (39) and aberrant morphology following OPC-specific BMAL1 loss further supports its role in migration. BMAL1-intact oligodendroglia from older adult mice become phenotypically similar to BMAL1-disrupted oligodendroglia following demyelination (Fig. 4G, J, L). This suggests that disruptions in the circadian system of oligodendroglia may contribute to the limited remyelination potential associated with progressive MS and aging (40). OPCs from aged rats lose the ability to differentiate into oligodendrocytes, but reversing DNA damage in aged OPCs restores remyelination potential (41). The normal process of aging is associated with declining circadian function through changes in circadian gene expression (42) and BMAL1 deficiencies lead to premature aging (43). Whether OPCs become circadian dysregulated with age remains to be fully elucidated.
Having established that Bmal1 expression is necessary for proper OPC lineage maintenance and myelin formation early in life, and global Bmal1 knockouts exhibit severe circadian (26) and sleep phenotypes (27), we aimed to discern if OPC-specific BMAL1 dysregulation and consequent dysmyelination affects these systems-level processes. Mice lacking functional BMAL1 in OPCs exhibit fragmented sleep architecture during the active phase, and this fragmentation is exacerbated by sleep deprivation. Importantly, these mice do not display the gross sleep deprivation that is linked to decreased myelin sheath thickness (44). These findings provide a hitherto underappreciated link between myelination and healthy sleep function, a connection that may be relevant to a range of neurological diseases. In mice, sleep fragmentation increases with age (45) and excessive sleep fragmentation is common in age-related disorders like Alzheimer’s disease (46), a disorder associated with deficits in oligodendrocytes and myelin (47). Numerous lines of evidence support that altered circadian biology is also significantly associated with MS prevalence. Up to 60% of individuals with MS report sleep and circadian rhythm abnormalities (48). Our finding that sleep fragmentation is causally associated with MS prevalence further supports the conclusion that myelination may be an underappreciated regulator of sleep processes. Whether these deficits are driven by changes in BMAL1 in OPCs, OPC population integration (8) into neural circuitry that controls sleep, or by alterations to myelin within those circuits remains to be determined. By defining the role of the circadian system in oligodendroglial lineage cell homeostasis, we have identified new mechanistic insights into myelin and sleep regulation that may provide therapeutic targets for brain disorders.
Funding
The U.S. Department of Defense (W81XWH-21-1-0846, EMG)
The National Multiple Sclerosis Society (PP-1907-34759, EMG)
The Weintz Family COVID-19 Research Fund (EMG)
The Department of Psychiatry and Behavioral Sciences, School of Medicine, Stanford University 2021 Innovator Grant Program (EMG)
The Brain and Behavior Research Foundation NARSAD (AW905644, EMG)
The Maternal and Child Health Research Institute Postdoctoral Fellowship (DR)
National Science Foundation Graduate Research Fellowship (DGE-1656518, CAG)
Ford Foundation Predoctoral Fellowship (CAG)
The National Cancer Institute, DHHS (PHS CA09302, LCM)
BioX Institute Fellowship (JG, MEG, JR)
The Instrumentarium Science Foundation (HMO)
NIH Shared Instrumentation Grant (1S10RR02678001; Electron Microscopy Core at Stanford University Cell Sciences Imaging Facility)
Author contributions
Conceptualization: EMG, DR, AB, LDC. Methodology: EMG, DR, AB, LDC, SK, JG, EE, CAG, LCM, BY, SEJ, NS, HMO, SN. Investigation: EMG, DR, AB, LDC, SK, JG, EE, CAG, LCM, MEG, RS, JR, SEJ, NS. Visualization: EMG, DR, AB, SK, JG, EE. Funding acquisition: EMG. Writing: EMG, DR.
Competing interests
Authors declare that they have no competing interests.
Data and materials availability
All data are available in the main text or the supplementary materials.
Acknowledgments
We would like to thank Drs. Richard Daneman and Caterina Profaci for their helpful guidance related to the assessment of the blood-brain barrier permeability.