The sleep-wake cycle regulates brain interstitial fluid tau in mice and CSF tau in humans

See allHide authors and affiliations

Science  22 Feb 2019:
Vol. 363, Issue 6429, pp. 880-884
DOI: 10.1126/science.aav2546

Sleep may protect the brain from AD

Two main proteins accumulate in the brain in Alzheimer's disease (AD), β-amyloid (Aβ) and tau. Aβ appears to instigate AD, but tau appears to drive brain damage and cognitive decline. Sleep deprivation is known to increase Aβ acutely and chronically. Now, Holth et al. show that chronic sleep deprivation strongly increases tau acutely over hours and also drives tau pathology spreading in the brains of mice and humans (see the Perspective by Noble and Spires-Jones). Thus, sleep appears to have a direct protective effect on a key protein that drives AD pathology.

Science, this issue p. 880; see also p. 813


The sleep-wake cycle regulates interstitial fluid (ISF) and cerebrospinal fluid (CSF) levels of β-amyloid (Aβ) that accumulates in Alzheimer’s disease (AD). Furthermore, chronic sleep deprivation (SD) increases Aβ plaques. However, tau, not Aβ, accumulation appears to drive AD neurodegeneration. We tested whether ISF/CSF tau and tau seeding and spreading were influenced by the sleep-wake cycle and SD. Mouse ISF tau was increased ~90% during normal wakefulness versus sleep and ~100% during SD. Human CSF tau also increased more than 50% during SD. In a tau seeding-and-spreading model, chronic SD increased tau pathology spreading. Chemogenetically driven wakefulness in mice also significantly increased both ISF Aβ and tau. Thus, the sleep-wake cycle regulates ISF tau, and SD increases ISF and CSF tau as well as tau pathology spreading.

Tau is an abundant cytoplasmic protein in neurons. In diseases called tauopathies—which include Alzheimer’s disease (AD), progressive supranuclear palsy, Pick’s disease, and chronic traumatic encephalopathy—tau becomes hyperphosphorylated and aggregates in structures such as neurofibrillary tangles and neuropil threads (1). Tau aggregation in the brain significantly correlates with neuronal and synaptic loss (2). There is substantial evidence that once tau aggregation occurs, it can spread from one synaptically connected region to another (37).

Although tau is predominantly cytoplasmic, it is normally released by neurons into the extracellular space. This release is increased through excitatory neuronal activity (8, 9). When neuronal activity is increased chronically, it can increase tau propagation and pathology (10). Because neuronal synaptic strength and connectivity are higher during wakefulness than during sleep (11, 12), we asked whether tau levels in the brain interstitial fluid (ISF) varied with the sleep-wake cycle. We measured tau levels in the hippocampal ISF of wild-type mice. ISF tau levels were low during the light period when, in our experience, wild-type mice generally sleep ~60% of the time. After the transition to the dark period, when we have observed that mice are awake ~70% of the time, ISF tau levels increased almost twofold (Fig. 1, A and B, and tables S1 and S3). Neuronal activity directly regulates lactate concentration in vivo (12, 13), and like ISF tau, ISF lactate was also higher during wakefulness and lower during sleep (Fig. 1, D and E), as previously observed (12, 14). The change in ISF tau between light and dark (~90%) is greater than what we have previously observed with ISF β-amyloid (Aβ) (~30%) (15). Given this change, we asked whether acute sleep deprivation alters ISF tau. Three hours after the beginning of the light period, mice were kept awake through manual stimulation. Sleep deprivation induced a significant twofold increase in ISF tau (Fig. 1, A and C, and tables S1 and S3). This increase was paralleled by an increase in ISF lactate (Fig. 1, D and F). In mice subjected to sleep deprivation during which neuronal activity was attenuated by infusion of tetrodotoxin (TTX) by means of reverse microdialysis, there was no increase in ISF tau or lactate (Fig. 1, A, C, D, and F, and tables S1 and S3).

Fig. 1 ISF tau exhibits diurnal fluctuation and increases after manual sleep deprivation (SD) but not in the presence of TTX.

(A) ISF tau levels normalized to baseline (06:00 to 09:00) over the 24-hour analysis period. Manual SD and TTX infusion occurred from 09:00 to 15:00 (shaded); control animals were undisturbed. (B) Average ISF tau is significantly increased during dark (wake) compared with light (sleep) in control animals, demonstrating diurnal fluctuation (n = 8 mice; paired t test). (C) Average ISF tau (normalized to baseline) during SD (09:00 to 15:00) was significantly increased in sleep-deprived mice compared with controls or mice with SD in the presence of TTX [n = 8 mice; one-way analysis of variance (ANOVA), P = 0.007; Bonferroni post-hoc]. (D) ISF lactate over the 24-hour analysis period. (E and F) As with ISF tau, average ISF lactate was significantly (E) increased during dark compared with light in control animals and (F) increased during SD [n = 8 mice; (E) paired t test, (F) Kruskal-Wallis]. All data represent mean ± SEM. All mice were 3 to 5 months old. All conditions, three female, five male. *P < 0.05, **P < 0.01.

Because sleep deprivation increased ISF tau in the mouse brain, we wondered whether similar changes would be seen in the cerebrospinal fluid (CSF) of humans. We recently studied CSF in a group of adults, 30 to 60 years of age, who were monitored with lumbar catheters during one night of normal sleep and one night of sleep deprivation, with sleep sessions randomized and separated in time (16). We found that sleep deprivation significantly increased CSF Aβ by 30% (16). Using samples collected from the same participants, we found that CSF tau was increased to an even greater extent by more than 50% (Fig. 2, A and C, and tables S2 and S4), and CSF tau levels significantly correlated with CSF Aβ levels (Fig. 2D and fig. S1). We also assessed three additional proteins: neuronal neurofilament light chain (NfL), synuclein, and astrocytic glial fibrillary acidic protein (GFAP). Synuclein, a presynaptic protein whose levels are increased after neuronal activity (17), was increased by sleep deprivation (Fig. 2, B and C, and tables S2 and S4). Unlike tau and synuclein, there was no increase in NfL or GFAP in CSF after sleep deprivation, suggesting some specificity in sleep–protein level interaction (Fig. 2E, fig. S2, and tables S2 and S4).

Fig. 2 CSF tau levels increase and are correlated with Aβ in sleep-deprived human subjects, and chronic SD in mice increases tau spreading.

(A) Human CSF tau and (B) synuclein (α-syn) levels normalized to baseline (07:00 to 19:00). SD began at 21:00, and CSF is compared from 01:00 to 11:00 (shaded). (C) Tau levels during SD are significantly increased by 51.5% and α-syn levels increased by 36.4% compared with undisturbed sleep [n = 6 participants (tau), n = 4 to 6 participants (α-syn); GLMM, first-order autoregressive]. (D) Total CSF Aβ is significantly correlated with CSF tau in control and SD conditions during the SD time period (n = 6 participants; Pearson’s correlation). (E) CSF NfL is unchanged by SD (n = 6 participants). (F) Ipsilateral hippocampal AT8 p-tau staining in grid control and chronic SD P301S male mice with unilateral hippocampal tau fibril injection. (G) SD does not alter p-tau staining in the ipsilateral hippocampus (n = 14 to 16 mice). (H) The ipsilateral LC/hippocampus AT8 ratio is increased in SD mice (n = 13 to 16 mice; Mann-Whitney). (I) AT8 staining in the LC of SD and control hippocampal-seeded P301S mice. Scale bars, (F) and (I) 125 μm. (J) p-Tau is significantly increased in the ipsilateral LC (n = 13 to 16 mice; unpaired t test) and (K) trended toward an increase in the contralateral LC (n = 14 to 16 mice; Mann-Whitney) of SD compared with control animals. All data represent mean ± SEM. **P < 0.01, ***P < 0.001.

Given these results, we next tested the potential effects of longer periods of sleep deprivation on tau seeding and spreading, a process believed to be mediated by extracellular tau (18). Recombinant P301S human tau fibrils were produced and injected unilaterally into the hippocampus of 8- to 9-week-old P301S male mice, before the onset of tau pathology, as described (5). Two days later, mice were subjected to 28 days of sleep deprivation or control conditions. Sleep deprivation did not alter tau seeding in the hippocampus; however, it significantly increased tau spreading to a region of the brain synaptically connected to the hippocampus and involved in maintenance of wakefulness, the locus coeruleus (LC), as assessed with p-tau staining with the antibody AT8 (Fig. 2, F to K). Increased AT8-positive staining was also observed in the entorhinal cortex (fig. S3, C and D). We saw a similar significant increase in tau spreading to the LC in sleep-deprived mice with the anti-tau antibody MC1, which recognizes a conformationally abnormal tau epitope (fig. S4). We also separately subjected 7.5- to 10-week-old P301S mice to 28 days of sleep deprivation or control conditions without injecting tau fibrils. There was very little AT8-positive p-tau staining in the LC that did not differ between groups, indicating that the increased staining in the LC after tau fibril injection and its further increase after sleep deprivation are due to spreading and not simply due to enhancement of the tau pathology that develops in these mice (fig. S3, A and B). One possible explanation for why 28 days of sleep deprivation did not increase tau pathology in naïve, noninjected mice may relate to recent findings that soluble tau overexpression results in suppression of neuronal activity (19).

Physical sleep deprivation increases both Aβ and tau in ISF and CSF. Because physical sleep deprivation increases wakefulness but also influences additional processes such as sleep rebound, we wanted to specifically manipulate neuronal activity in a brain region that controls wakefulness. To do so, we expressed the excitatory chemogenetic system (hM3Dq), also known as designer receptors exclusively activated by designer drugs (DREADDs), in glutamatergic neurons (VGLUT2+) of the mouse supramammillary nucleus (SuM), a region recently established as a key arousal-promoting node. Administration of clozapine-N-oxide (CNO) to SuMVglut2-hM3Dq mice has been shown to result in sustained wakefulness for up to 12 hours (20). We bred APPswe/PS1δE9 transgenic mice that produce human Aβ (21) with mice that express Cre in excitatory glutamatergic VGLUT2-positive neurons (Vglut2-ires-Cre mice) (Fig. 3A). We then placed stereotaxic injections of an adeno-associated virus (AAV) vector expressing Cre-dependent hM3Dq into the SuM at 2.5 to 3 months of age, resulting in expression of hM3Dq in glutamatergic cells of the SuM of transgenic Vglut2-ires-Cre:APPSwe/PS1δE9 mice (Fig. 3B), and monitored CNO wake-activated and control animals (Fig. 3C). Mice first underwent sleep-wake assessment by means of continuous electroencephalography (EEG)/electromyography (EMG) monitoring as well as in vivo microdialysis to measure Aβ and lactate for 24 hours undisturbed. Both groups were awake ~40% of the time during the light phase and ~65% during the dark phase (Fig. 3, D and E, and fig. S5). During the second 24-hour period, mice were injected intraperitoneally with either CNO (0.3 mg/kg) or 0.9% NaCl (vehicle) 3 hours after the onset of the light period. Treatment of mice with CNO resulted in a marked increase in wakefulness for 9 hours (more than 95% wakefulness, hours 1 to 5) and a significant decrease in nonrapid eye movement (NREM) and REM sleep without any signs of sleep deprivation–induced sleep rebound. (Fig. 3, D and F to H, and fig. S5). In mice treated with vehicle, there was no effect on wakefulness, NREM, or REM sleep (Fig. 3E and fig. S5). There was also no effect of CNO or its parent compound clozapine on wakefulness in non–AAV phosphate-buffered saline–injected Vglut2-ires-Cre:APPSwe/PS1δE9 mice (fig. S6).

Fig. 3 Chemogenetic (hM3Dq) activation of glutamatergic supramammillary neurons drives sustained wakefulness without inducing sleep rebound.

(A) Diagram depicting Cre-dependent recombination of AAV-DIO-hM3Dq-mCherry in Vglut2-ires-Cre:APPSwe/PS1δE9 mice. (B) Labeling of mCherry-tagged hM3Dq by using an antibody to DsRed (counterstain cresyl violet). Scale bar, 500 μm. (C) Timeline of hM3Dq-mediated manipulation of the SuM. (D) EEG data showing significantly increased wakefulness over 9 hours after CNO injection compared with the nontreated day (n = 8 mice; four female, four male; two-way repeated measures ANOVA; interaction, P < 0.0001; Bonferroni post-hoc). (E) NaCl injection in AAV-hM3Dq mice showed no effect on percent wakefulness (n = 6 mice; one female, five male). (F to H) Analysis of (F) percent wakefulness, (G) percent NREM sleep, and (H) percent REM sleep from 06:00 to 09:00 (before CNO), 09:00 to 18:00 (after injection of CNO), and 18:00 to 06:00 (following night) showed significantly increased wakefulness and decreased sleep after CNO injection (red) compared with the nontreated day (black) but no differences before CNO injection. Analysis of the dark period (18:00 to 06:00) showed similar levels of percent wakefulness and sleep in both groups demonstrating a lack of NREM or REM sleep rebound after sustained wakefulness over a period of 9 hours (two-way repeated measures ANOVA; interaction, P < 0.0001; Bonferroni post-hoc). All data represent mean ± SEM. **P < 0.01, ***P < 0.001.

We next asked how ISF Aβ was influenced by the sleep-wake cycle and chemogenetic induction of wakefulness. As we have seen previously, ISF Aβ was higher during the dark phase versus the light phase in the 24 hours before treatment in both groups (Fig. 4, A and B, and tables S1 and S5). After treatment of Cre-expressing APPswe/PS1δE9 mice with CNO, there was a strong and significant increase of both ISF Aβ, which peaked at ~35% above the nontreated mice (Fig. 4, A and C, and tables S1 and S5), and ISF lactate (Fig. 4, E and G). In mice treated with vehicle, there were no changes in ISF Aβ (Fig. 4, B and D, and tables S1 and S5) or lactate (Fig. 4, F and H). In addition, there was no effect of CNO on Aβ in non–AAV-injected APPswe/PS1δE9 mice (fig. S7). To assess ISF tau using the chemogenetic approach, we injected Vglut2-ires-Cre mice with AAV-hM3Dq at 2.5 to 3 months of age. Before assessing the mice by means of microdialysis, we used activity tracking to determine that CNO increased wakefulness (fig. S8). Mice underwent in vivo microdialysis 5 to 7 weeks after injection, and ISF tau was assessed when treated with either CNO or NaCl. CNO treatment significantly increased ISF tau by ~40% and lactate by ~90% (Fig. 4, I to L, and tables S1 and S6) compared with NaCl-treated mice.

Fig. 4 Chemogenetic activation of glutamatergic supramammillary neurons increases hippocampal ISF Aβ, tau, and lactate levels.

(A) ISF Aβ levels normalized to baseline (06:00 to 09:00) during a 24-hour untreated period (black) and 24-hour period with CNO (red; n = 8 mice; four female, four male) or (B) NaCl injection (blue; n = 6 mice; one female, five male). (C) Average ISF Aβ was significantly increased after injection of CNO (09:00 to 18:00) compared with the untreated day (n = 8 mice, paired t test). (D) NaCl control injection did not alter average ISF Aβ (n = 6 mice). (E) Normalized ISF lactate levels before and after CNO (n = 7 mice; four female, three male) and (F) NaCl injection (n = 6 mice). (G) Average ISF lactate was significantly increased after CNO injection (09:00 to 18:00) compared with the untreated day (n = 7 mice; Wilcoxon signed rank). (H) Average ISF lactate after control NaCl injection (09:00 to 18:00) was unchanged. (I) ISF tau levels normalized to baseline (06:00 to 09:00) in CNO-treated mice (n = 9 mice; three female, six male) and NaCl-injected controls (n = 8 mice; four female, four male). (J) ISF tau after CNO treatment (09:00 to 18:00) was significantly increased compared with NaCl controls (n = 8 to 9 mice; unpaired t test). (K) Normalized ISF lactate levels of CNO (n = 8 mice; three female, five male) and NaCl-treated mice (n = 8 mice; four female, four male). (L) Average ISF lactate after CNO treatment (09:00 to 18:00) was increased compared with NaCl controls (n = 8 mice; Mann-Whitney). All data represent mean ± SEM. *P < 0.05, ***P < 0.001.

We have shown that ISF tau is regulated by the sleep-wake cycle and that both ISF tau in mice and CSF tau in humans are strongly increased by sleep deprivation. ISF tau was increased with physical and chemogenetic induction of wakefulness. The half-life of tau in the brain is long (~10 days in mice and more than 20 days in humans) (22, 23). However, once tau reaches the extracellular space in the ISF or CSF, its half-life is very short (1 to 2 hours) (24). Thus, changes in the sleep-wake cycle can result in rapid changes in ISF and CSF tau. Because TTX blocks the sleep deprivation–induced increase in tau, it seems likely that the mechanism by which tau increases with wakefulness and sleep deprivation is linked with elevated neuronal metabolism and synaptic strength, causing enhanced tau release. Although the exact mechanism of tau release is unclear, the changes observed here represent free, non–vesicle- or non–exosome-associated tau. It is also possible that decreased ISF tau clearance during wakefulness could contribute to the change observed. Seeing in human CSF changes in tau and synuclein but not all proteins after sleep deprivation argues that the changes are more likely due to increased release of certain proteins rather than changes in global ISF clearance.

We and others have found that the sleep-wake cycle influences Aβ levels both acutely and chronically, which can influence the pathogenesis of AD (15, 25). Although there is a lot of evidence that Aβ aggregation initiates AD pathogenesis, including driving neocortical tau aggregation and spreading, tau accumulation appears to drive neurodegeneration. The acute increases in monomeric ISF tau through wakefulness and sleep deprivation may play a normal role in cell signaling, whereas increased release of pathological species may play a role in seeding, spreading, and neurodegeneration. The observation that increased wakefulness can increase ISF and CSF tau, tau spreading, and tau aggregation over longer periods of time (26) in mice provides evidence for its key role in regulating tau pathology. Thus, optimization of the sleep-wake cycle should be an important treatment target to test in the prevention of AD and other tauopathies.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

Tables S1 to S6

References (2731)

References and Notes

Acknowledgments: We thank F. Stewart for her assistance with microdialysis, C. Saper and Q. H. Ha for assistance learning SuM DREADD, R. DeMattos at Eli Lilly for providing m266 and 3D6 antibodies for enzyme-linked immunosorbent assay, and P. Davies for providing the MC1 antibody. Funding: This work was supported by a grant from the BrightFocus Foundation (A2017114F, S.K.F.), funding from the Deutsche Forschungsgemeinschaft (3625/1-1, S.K.F.), NIH F32 NS089381 (J.K.H.), NIH K08NS105929 (N.P.P.), NIH R01NS073613 (P.M.F.), R01NS092652 (P.M.F.), NIH P01NS074969 (D.M.H.), the JPB Foundation (D.M.H.), the Tau Consortium (D.M.H.), UL1 TR000448 and KL2 TR000450 (B.P.L.), R03 AG047999 (B.P.L.), K76 AG054863 (B.P.L.), P50 AG05681 (B.P.L.), P01 AG26276 (B.P.L.), and the McDonnell Center for Systems Neuroscience at Washington University School of Medicine (B.P.L.). Author contributions: J.K.H., S.K.F., N.P.P., J.R.C., J.C.G., P.M.F., B.P.L., and D.M.H. designed the research studies. J.K.H., S.K.F., C.W., T.E.M., M.B.F., and M.M. conducted experiments and acquired data. J.K.H., S.K.F., and D.M.H. wrote the first version of the manuscript. J.K.H., S.K.F., C.W., N.P.P., J.R.C., T.E.M., M.B.F., M.M., J.C.G., P.M.F., B.P.L., and D.M.H. edited the manuscript. Competing interests: D.M.H. cofounded and is on the scientific advisory board of C2N Diagnostics. D.M.H. is on the scientific advisory board of Denali, Genentech, and Proclara. D.M.H. consults for AbbVie. None of the authors report competing interests. Data and materials availability: All data are available in the main text or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article