Chronobiological Approaches to Alzheimer’s Disease

Abstract

Dynamic circadian rhythms contribute to memory formation, and the hormonal and neurochemical changes that follow circadian patterns are frequently dysregulated with aging. The effect of aging on circadian rhythms is a double-edged sword; on one hand, poor sleep quality compromises neuronal structure and function in regions that support cognition, and on the other hand, perturbation of central and peripheral oscillators changes the hormonal milieu, with consequences for neuroplasticity. In the current review, recent developments surrounding the circadian regulation of memory formation are described, with reference to how mechanisms that support temporal coding might change with advancing age. The cognitive consequences of changes in sleep patterns are also discussed. New roles for the circadian clock genes period-1, period-2, and bmal1 in memory formation are discussed in the context of age-related cognitive decline. The potential for chronobiological approaches to the treatment and prevention of Alzheimer’s disease merits further exploration from a pharmacotherapeutic perspective, as the timing of drug delivery could potentiate or diminish treatment efficacy.

Circadian rhythms are endogenously generated, entrainable systems that depend on light for periodic oscillations. Light stimulation activates pacemaker neurons, which then signal via output pathways to induce physiological and behavioral activity. Temporal organization of behavior is classically attributed to neurons of the hypothalamic suprachiasmatic nucleus (SCN), which exhibit oscillatory firing and show rhythmic alterations in gene transcription. While fluctuations in motor activity and ingestive behavior occur over a twenty-four hour cycle and are controlled by the SCN, the hippocampus participates in the temporal microstructure of behaviors related to learning. While rats with SCN lesions show disorganization of their activity rhythms, rats with hippocampal lesions are hyperactive, but exhibit intact activity rhythms [1]. These data from lesioned animals support a role for the hippocampus in constraining the amplitude of behavioral rhythms (Figure 1).

Figure 1. Effects of hippocampal or suprachiasmatic nucleus lesions on motor activity patterning.

Nocturnal rodents exhibit high levels of voluntary wheel running at night. Hippocampal lesions increase motor activity, particularly during the night phase (Whishaw and Jarrard, 1995). Suprachiasmatic nucleus lesions disrupt the temporal organization of motor behavior without influencing the overall amount of activity (Moore, 1996).

Anatomical connections point towards a role for hippocampal neuronal activity in the regulation of SCN firing. The ventral subiculum of the hippocampus sends excitatory afferent projections to the dorsal portion of the SCN [2]. These projections are not reciprocated, as the SCN does not send efferents back to the hippocampus [3]. However, hippocampal plasticity is responsive to changes in SCN output, on the basis of several lines of research documenting impairment of hippocampal plasticity following constant lighting or genetic manipulation of circadian gene expression. This review explores the regulation of hippocampal function across the circadian cycle, and discusses recent advances that are relevant to aging and Alzheimer’s disease (AD). The relative contributions of clock genes, sleep-wake cycling, and stress are also discussed with regard to their role as modulators of hippocampal function. Perturbation of activity rhythms in AD patients and AD mouse models are reviewed in light of a possible relationship between altered sleep-wake cycling and disease progression. Finally, we propose that chronobiology could be harnessed to enhance the efficacy of existing AD pharmacotherapeutics.

Circadian fluctuations in hippocampal structure and function

Regions such as the SCN are well-characterized with regard to changes in neuronal structure, function, and biochemistry over a twenty-four hour period [4]. Other brain regions, such as the hippocampus, have received less attention with regard to circadian rhythms and neuronal plasticity. An early study demonstrated clear circadian variability in long-term potentiation (LTP; [5]), a form of plasticity thought to contribute to associative learning. In this study, LTP in hippocampal area CA1 was strongest in the inactive (light) phase, and weakest in the active (dark) phase. The pattern was inverted in the hippocampal dentate gyrus, suggesting that the synaptic mechanisms supporting associative learning might be under the control of endogenous oscillators in the hippocampus.

LTP is supported by structural changes at the synapse. Repeated pairings of pre- and postsynaptic activity first expand the size of the dendritic spine, then with prolonged coincident activation, leads to the de novo formation of new spines and synapses [6]. While the exact time course surrounding structural mechanisms that support potentiation are still being characterized, the general consensus is that active synapses survive, and repeated stimulation leads to formation of new synaptic contacts. The presence of circadian variability in synaptic function suggests that the number and morphology of hippocampal spines and synapses might also exhibit some circadian variability, but this possibility has yet to be addressed in the literature. In hibernating rodents, hippocampal spines and synapses are rapidly lost upon hibernation, and regained in a temperature-dependent fashion upon awakening [7]. Despite this indication that the sleep-wake cycle might be associated with differences in synapse number in the hippocampus, there have been no published studies exploring this possibility. The observations in hibernating rodents do suggest that the structure and function of hippocampal neurons is, to some extent, controlled by factors associated with the sleep-wake cycle.

The dentate gyrus of the hippocampus continuously generates new neurons, and the process of neurogenesis fluctuates over a twenty-four hour period. Rodents have greater levels of neurogenesis during the inactive (light) phase, relative to the active (dark) phase, as assessed through incorporation of the DNA synthetic marker bromodeoxyuridine (BrdU; [8]). The environmental induction of adult neurogenesis also exhibits circadian periodicity. Running is a well-recognized stimulator of hippocampal neurogenesis, but running during the active phase promotes neurogenesis more than running during the light phase [9]. In this respect, both basal neurogenesis and environmentally-induced changes in cell proliferation vary over the circadian cycle. Just as the capacity for functional alterations in synaptic strength differs based on the time of day, the process of generating new neurons is also under circadian control, reinforcing the importance of the light/dark cycle as a determinant of hippocampal structure and function.

Effect of aging on circadian rhythms in the brain

Aging alters circadian rhythms throughout the body, and the central nervous system (CNS) is no exception. Aging is dampens fluctuations in clock gene expression in the SCN [10], leading to changes in hormone levels throughout the body [11]. The hippocampal ventral subiculum sends afferent input to the SCN [12], but the functional relevance of these inputs has yet to be addressed. Because the hippocampus is involved in maintaining memories over time, it is possible that hippocampal inputs to the SCN master clock may contribute to circadian regulation (Figure 2). If hippocampal projections to the SCN participate in circadian rhythms, then loss of function among hippocampal neurons could potentially alter neuronal activity in the SCN. While this chain of events is merely hypothetical, it is anatomically possible for hippocampal neurons to modulate neuronal activity in the SCN master clock.

Figure 2. Connectivity between the hippocampus and hypothalamic regions involved in the regulation of circadian periodicity.

Abbreviations: DG, dentate gyrus; CA3, cornu ammonis 3; CA1, cornu ammonis 1; vSub, ventral subiculum; SCN, suprachiasmatic nucleus; BNST, bed nucleus of the stria terminalis; LatH, lateral hypothalamus.

The SCN constitutively expresses the circadian clock genes CLOCK and bmal1 (Figure 3). However, cyclic expression of clock genes has been reported outside of the SCN. In young mice, hippocampal subfields exhibit high levels of CLOCK and bmal1 during the light phase, relative to the dark phase [13]. Aging alters the phase of high CLOCK and bmal1 by shifting the increase in protein immunoreactivity towards the dark phase [13]. The presence of circadian fluctuations in CLOCK and bmal1 expression may be strain-dependent, as the above study reported cyclical changes in CLOCK and bmal1 expression in C57Bl6 mice over a twenty-four hour period [13], while another study in C3H/J mice reported no circadian variability in hippocampal CLOCK and bmal1 immunoreactivity [14]. Although genetic background may modulate the extent to which hippocampal clock gene expression is locked to specific circadian phases, it is clear that in older animals, the amplitude and phase of clock gene expression in the central nervous system is perturbed.

Figure 3. Core components of the molecular clock.

Over a twenty-four hour cycle, clock-bmal1 heterodimers activate transcription of period (Per) and cryptochrome (Cry). Period (Per) and cryptochrome (CRY) proteins form heterodimers in the cytoplasm, then enter the nucleus where they inhibit transcriptional activation of clock-bmal1 heterodimers. This negative feedback model serves as the basis for circadian timekeeping. Clock and bmal1 also regulate the expression of genes associated with metabolism, cytoskeletal regulation, and neuroplasticity.

Aged rodents respond less readily to light shifts, indicated both by changes in light-evoked immediate early gene expression in the SCN [15], and by temporal patterns of voluntary motor activity in a running wheel [16]. In aging humans, the SCN is one of the few regions that exhibits frank neuronal loss with aging [17–18]. Moreover, patterns of neuronal firing among the remaining neurons are dampened in aging rodents [19–20]. Photostimulation-induced resetting of firing rhythms among SCN neurons is also compromised during aging [21]. Given that neuronal function in the SCN is clearly attenuated in old age, it becomes possible to speculate that hippocampal dysfunction arises not from some endogenous mechanism, but may instead be secondary to circadian dysregulation following impairment of neuronal functioning in the SCN.

The hippocampus replays behaviorally relevant patterns of neuronal firing during sleep [22]. Sleep often becomes fragmented with aging [23]. In animal models, interfering with sleep is detrimental for hippocampal function at the behavioral, structural, and electrophysiological level [24]. In this regard, deficits in hippocampal function in old age may occur, in part, due to changes in sleep patterns.

Circadian regulation of memory formation

The hippocampus is essential for memories over time and locations in space [25]. Given this role, and the amenability of these functions to behavioral testing in rodent models, there have been a number of studies exploring the circadian regulation of hippocampus-dependent learning. The main findings of this line of research are as follows: 1), nocturnal rodents learn most efficiently during their active phase [26]; 2), perturbation of clock gene expression using transgenic models or constant lighting impairs hippocampus-dependent memory [14]; 3) mutant mice that show deficits in pulsatile hormone release and abnormal activity rhythms [27] also exhibit impairment of learning and memory [28]. These observations have been documented using behavioral paradigms that recruit the hippocampus for both spatial and temporal demands.

Mnemonic consequences of disrupted sleep patterns

Sleep deprivation impairs hippocampal structure and function via multiple mechanisms that are especially relevant to the aging brain. Aging alone is associated with reduced sleep quality in humans [29], and within aged populations, activity rhythm fragmentation is correlated with cognitive decline [30]. Indeed, there appears to be a linear relationship between the integrity of activity rhythms and cognition across the spectrum of aging and AD [31]. In rodents, adult neurogenesis, a form of hippocampal structural plasticity, was attenuated following sleep deprivation in a glucocorticoid-dependent manner [32]. However, sleep deprivation may also constrain adult neurogenesis through glucocorticoid-independent mechanisms, since methods of sleep deprivation that do not elevate corticosterone levels also suppress adult neurogenesis [33]. In addition to the suppression of hippocampal cell proliferation, sleep deprivation or sleep fragmentation also reduces the number of hippocampal dendritic spines [34]. In this regard, structural plasticity is constrained or attenuated by lack of sleep.

During sleep, neurons fire sequentially in a spatial series of patterns that recapitulates the patterns evoked by recent experience, a phenomenon known as ‘replay’ [22]. Replay of neuronal firing patterns is thought to contribute to memory, via mechanisms that are still being explored. The importance of sleep for associational learning is also reflected in gain-of-function studies demonstrating that sleep immediately after training improves performance [35]. Therefore, it is possible that altered sleep patterns contribute to deficits in neuronal function with aging and dementia.

Clock gene contributions to learning

Circadian timing regulates the expression and activity of genes and proteins in the central nervous system. Transcription and translation act in an autoregulatory fashion to mediate timekeeping in SCN neurons over a twenty-four hour period (Figure 3). A number of genes have been identified and characterized with regard to their importance for circadian timekeeping. Mammalian bmal1, clock, cry1, cry2, per1, and per2 all participate as oscillatory components of the circadian clock.

Perturbation of genes implicated in circadian regulation, whether through altered lighting or transgenic loss-of-function, has a negative impact on hippocampus-dependent memory. In one study, Siberian hamsters were subject to circadian dysregulation through altered lighting, and this paradigm compromised object recognition memory [36]. In another study, changing the light-dark schedule interfered with contextual fear conditioning in mice [37]. Likewise, unpredictable variability in lighting patterns disrupts spatial memory in rats [38–39]. A critical issue in these investigations is whether clock genes themselves contribute to memory, or if altered sleep patterns mediate memory deficits. In control experiments, Ruby et al., [36] partially addressed this point by depriving light-exposed hamsters of the opportunity to sleep after training. Preventing the light-exposed hamsters from sleeping did not alter their performance on an object recognition task, suggesting that for this task, in this species, circadian dysregulation rather than sleep deprivation interfered with performance. However, more work remains to fully parse the relative contributions of clock gene dysregulation and sleep deprivation.

Hippocampal function is also impaired in genetic models with disrupted circadian periodicity. Spatial learning in the radial arm maze was impaired in Per1 homozygous knockout mice, which have altered activity rhythms [14]. Similarly, in flies lacking Per1, long-term memory is also impaired [40]. Based on these observations, it is clear that genetic or environmental interference with circadian oscillators impairs hippocampal neuroplasticity.

Some, but not all, of these negative effects may be mediated by prolonged exposure to elevated glucocorticoids, which has a detrimental influence on hippocampal neuroplasticity [41]. Methods of sleep deprivation that induce stress, such as forced treadmill running [42] and the disc-over-water method [43], also interfere with memory. Exposure to elevated glucocorticoids at levels comparable to those observed following stress impairs memory [44], and maintaining normal physiological levels of glucocorticoids through adrenalectomy and low-dose corticosterone replacement prevents the effects of sleep deprivation on hippocampal structure [32]. Therefore, it is possible that stressful conditions that disrupt sleep compromise hippocampal function in a glucocorticoid-dependent fashion.

On the other hand, sleep deprivation methods that do not elevate glucocorticoids have also been shown to impair hippocampal neuroplasticity. Interfering with sleep through cage tapping or gentle handling does not substantially increase levels of adrenal steroids, and yet these manipulations interrupt hippocampal function [45–46].

Altered activity rhythms in AD model mice

Home cage and running wheel activity patterns are an established behavior index of circadian integrity. Just as AD patients exhibit alterations in their activity rhythms, following a pattern known as ‘sundowning,’ AD model mice also show disrupted circadian patterns of activity. In the TGCRND8 mouse model of AD, levels and rhythmicity of home cage activity were altered before the onset of plaque deposition [47]. Likewise, in a double-transgenic mouse model overexpressing the human amyloid precursor protein and presenilin-1 genes, age-dependent changes in home cage activity were also reported, with younger APP/PS1 mice exhibiting reduced home cage activity, and older APP/PS1 mice showing higher levels of motor activity than normal controls [48]. Again, alterations in the diurnal rhythm of motor activity occurred before plaque deposition in this model [48]. In a third mouse model of AD arising from overexpression of human amyloid precursor protein, perturbation of activity rhythms over the lifespan were also noted in APP23 mice [49]. These convergent findings from three different groups using three different models strongly suggest that circadian alterations play a role in AD pathology. In AD mouse models, pathology typically starts along the entorhinal-hippocampal circuitry, and spreads to other brain regions with advancing age. No studies have addressed the possibility that aging might ‘set the stage’ for AD pathology by altering activity patterns via the SCN master clock.

Chronobiological approaches to Alzheimer’s pharmacotherapy

Cholinesterase inhibitors are currently the most widely prescribed treatment for AD. The hippocampus follows a circadian pattern with regard to cholinergic neurotransmission [50]. Hippocampal acetylcholine (ACh) release is strongest during the active phase in rodents (Masuda et al., 2005). Expression of the ACh synthesizing enzyme choline acetyltransferase (ChAT) is strongest during the active phase, and levels of the ACh degrading enzyme acetylcholinesterase (AChE) are lowest during the active phase, leading to an increase in the bioavailability of ACh [51]. In order to reinstitute normal physiological rhythms of ACh in AD patients, it might be advantageous to administer cholinesterase inhibitors in the morning, coincident with circadian acrophase in humans. More detailed studies of circadian influences on the pharmacokinetics of cholinesterase inhibitors are warranted, as published data surrounding time-of-day effects on the bioavailability of cholinesterase inhibitors are scarce.

Memantine, an NMDA receptor antagonist, is also prescribed for AD patients. While data are sparse regarding potential circadian rhythmicity of glutamatergic neurotransmission in the hippocampus, there is some indication that this might be the case. Exposure to elevated glucocorticoid hormones increases hippocampal glutamate accumulation [52], and glucocorticoids are released in a pulsatile manner. Therefore, there is the possibility that glucocorticoid rhythms drive circadian fluctuations in glutamatergic neurotransmission in the hippocampus. If we assume that hippocampal glutamate levels follow the same pattern as circadian fluctuations in corticosterone, then glutamatergic neurotransmission in healthy adults would be strongest during the active phase. By contrast, glutamate neurotransmission in AD patients, who exhibit severe circadian dysregulation, might be elevated at night. In order to normalize glutamatergic rhythms, memantine might be most efficacious when administered in the evening. However, as noted above, further assessment of circadian effects on metabolism and excretion of memantine are necessary in order to determine whether time of day influences memantine pharmacokinetics.

Summary and Conclusion

Hippocampal neuroplasticity fluctuates over the twenty-four hour cycle. Perturbation of circadian rhythms impairs hippocampal function, but the extent to which hippocampal plasticity deficits arise from changes in clock gene expression, or are secondary to sleep deprivation, remains poorly understood. Sleep deprivation also compromises hippocampal function, and sleep after a learning experience enhances retention, possibly by facilitating replay of behaviorally relevant sequences of neuronal firing. The SCN master clock is particularly vulnerable to aging, and exhibits flattening of rhythmic neuronal firing, and reduced neuroplasticity in response to light shifts. These deleterious changes may set the stage for neuropathological alterations in AD and other forms of dementia. AD model mice exhibit circadian abnormalities that resemble those observed in AD patients, and these abnormalities precede the onset of neuropathology. Overall, these data suggest that circadian dysregulation contributes to AD onset and progression, although more work remains to fully clarify the underlying mechanisms for this relationship.