Memory,Sleep,and Tau Function

INTRODUCTION

Tau, the main component of aberrant intracellular aggregates in several tauopathies, including Alzheimer’s disease (AD), may be a potential therapeutic target in these neurodegenerative disorders [1, 2]. However, tau has an important function (for an example, see [3]) and therefore it is not advisable to eliminate this protein or greatly decrease the amount present. In this brief review, we indicate the contribution of tau to episodic memory and sleep. Indeed, sleep disturbances have been associated with tauopathies like AD and mild cognitive impairment [4–6]. Also, amyloid-β has been linked to altered sleep patterns in AD [7].

CONSOLIDATION OF EPISODIC MEMORY

The role of the hippocampus in human episodic memory was suggested after an epileptic patient lost their memory after partial removal of bilateral hippocampal regions [8]. Later, animal models were used to demonstrate the involvement of the hippocampal CA1 (and CA3) region in the recollection of episodic memory [9, 10]. However, the hippocampus plays a short-term role in the recollection of episodic memory, and it was later postulated that sleep consolidates this memory through the transmission of information from the hippocampus to the brain cortex [11].

Although memory consolidation is mainly sleep-dependent, it can also be sleep-independent [12–14]. Nevertheless, here we focus on sleep-dependent memory consolidation.

SLEEP PHASES AND MEMORY CONSOLIDATION

Electroencephalography (EEG) analysis reveals that the large wave activity found in wakefulness decreases in some phases of sleep. There are two main sleep phases, namely NREM and REM, which also show EEG differences. NREM is characterized mainly by slow wave oscillations (<1 Hz) and δ waves (∼3 Hz) the in neocortex and hippocampus, although a θ wave burst precedes and spindles follow cortical and thalamic down states in human NREM sleep [15]. Also, sharp wave ripples have been reported in distinct regions of the hippocampus [16]. Slow wave oscillations promote long-range effective connections [17], which may take place from the hippocampus to cortical regions. Given that serotonin signaling has been related to the intensity of δ-wave sleep [18, 19], the source of serotonin has been discussed. Serotonin could derive from the raphe nucleus. Indeed, the destruction of neurons from this nucleus with aging induces insomnia [20]. On the other hand, serotonin signaling occurs in the gastrointestinal tract [21], and a link between serotonin secretion and gut-microbiota and the gut-brain axis has been proposed [22].

In REM, a higher frequency of θ waves is present [23, 24], and θ stimulation facilitates actin polymerization and the formation of dendritic spines in hippocampal neurons [25].

In the sleep-wake cycle, higher wave activity takes place in wakefulness than in memory processes, which may result in a long increase (potentiation) of synaptic efficiency [26]. However, in NREM, there is a general weakening of synapses [11], which could lead to a decrease (depression) in synaptic efficacy [27], known as LTD. Nevertheless, sleep may promote the specific formation of dendritic spines (in distal dendrites), which could be related to learning (memory). However, a huge amount of information could be lost during LTD [28]. Indeed, electron microscopy analyses of the brain cortex of mice after sleep [29] indicate a decrease in synaptic efficacy, thereby suggesting that synaptic potentiation in the sleep-wake cycle should occur primarily during wakefulness, a period for learning, whereas depression should occur during sleep. During these changes in synaptic activity, the role of interneurons (somatostatin, parvalbumin, or vasoactive intestinal peptide types) can play an important role, mainly by reducing wave activity during NREM [23], a phase in which specific hippocampal-cortical transmission could take place [24]. In addition, in REM, the inhibitory activity of vasoactive intestinal peptide to block somatostatin activity may facilitate cortical-cortical interaction and integration to consolidate episodic memory [24]. In summary, learning for episodic memory occurs in the hippocampus during wakefulness, and there is hippocampal cortical transmission during NREM, in which slow δ waves could play a role. In contrast after the NREM-REM transition, cortical-cortical integration, which involves θ waves, occurs (Fig. 1).

OPEN IN VIEWER

Fig. 1

Role of tau in the sleep-wake cycle. In the absence of tau, there is a decline in δ power (NREM), whereas overexpression of mutated (phosphorylated) tau correlates with changes in θ power [36, 45].

MOLECULAR MARKERS FOR SLEEP PHASES

The pathways that modulate sleep duration have been analyzed using genomics and transcriptomics [30]. However, little is known about molecular markers for sleep phases. It has been suggested that the protein Cacna2d1 participates in the transition from NREM to REM sleep [31]. This protein may favor the calcium flow present in NREM. Cacna2d1 is involved in voltage-gated calcium channels, and some loss of function variants may cause the development of epigenetic encephalopathy [32]. However, tau protein also binds to the NMDA receptor complex and may therefore facilitate the entry of calcium into neurons [33, 34].

ROLE OF TAU PROTEIN IN THE SLEEP-WAKE CYCLE

The tau k.o. mouse raised by Dawson et al. [35] revealed that tau protein participates in the sleep-wake cycle. These mice show an increased period of wakefulness and decreased NREM sleep. These alterations could be related to a significant decline in δ power, together with an increased density of sleep spindles during NREM sleep [36]. Also, tau k.o. mice show slight slowing of hippocampal θ waves [37]. Other studies using tau k.o. mice suggested that tau protein plays a role in memory and anxiety-related behavior [38] and short memory deficits in mice lacking tau [39]. Also, deficits in fear memory have been described in these animals [40]. In addition, tau-driven degeneration of sleep and wake-regulating neurons in AD has been reported [41].

The hippocampus encompasses a region, namely the dentate gyrus, that hosts adult hippocampal neurogenesis. Related to hippocampal-dependent learning [42], this is a process in which adult neuronal stem cells differentiate into mature neurons, which can then be incorporated into hippocampal circuity [43]. Using tau k.o. mice, it was shown that tau is involved in the morphological differentiation and synapsis integration of newborn granule neurons in vivo [3].

Neurodegeneration and disorders associated with tau could arise as a result of tau loss of function, as shown for tau k.o. mice, or by gain of a toxic effect, as proposed for tau modified by phosphorylation (see [44] for example). In this regard, in a transgenic mouse overexpressing phosphorylated tau, phosphorylation was found to correlate with changes in hippocampal θ oscillations, thereby yielding reduced excitability [45, 46]. Additionally, tau phosphorylated at serine 396 (nomenclature for the largest tau isoform in the CNS) is required for LTD [47]. Although further work is needed, our current working hypothesis is that δ waves are involved in tau-dependent NREM, whereas θ waves are present in tau-dependent REM.

EARLY TAU PATHOLOGY IN PRODROMIC AD IS PRESENT AT THE LOCUS COERULEUS (LC). ROLE OF THE LC IN THE SLEEPING BRAIN

Phosphorylated tau is present in the early tau pathology associated with AD [48–50]. Also, it has been described that abnormal LC sleep activity could be related to changes in memory consolidation, by reducing NREM δ power and REM θ power [51]. In summary, LC could play a role in the sleeping brain (for review, see [52]).

CONCLUSIONS

Given that tau pathology widely affects the brains of AD patients, the removal of tau protein has been proposed as a potential approach to prevent this pathology (for examples, see [1, 53]). However, a huge decrease or elimination of tau protein may affect some brain functions, as described in this review. Given these considerations, the removal of toxic phosphorylated or aggregated tau through proteins like numb [54], and the maintenance of a suitable amount of non-toxic tau in neurons emerge as alternative approaches to tackle tauopathies, like AD [55].