Calcium regulation of neural rhythms, memory and Alzheimer's disease

Abstract

Alzheimer's disease (AD) begins with a decline in cognition followed by neuronal cell death and dementia. These changes have been linked to a deregulation of Ca²⁺ signalling caused by a progressive increase in the resting level of Ca²⁺, which may influence cognition by interfering with the rhythm rheostat that controls the sleep/wake cycle. The rise in resting levels of Ca²⁺ may not alter the processes of memory acquisition during consciousness (gamma and theta rhythms), but may duplicate some of the events that occur during the slow oscillations responsible for the twin processes of memory consolidation and memory erasure that occur during sleep. The persistent elevation in the resting level of Ca²⁺ induced by an accumulation of amyloid β (Aβ) oligomers duplicates a similar small global elevation normally restricted to the period of slow oscillations when memories are erased during sleep. In AD, such a rapid erasure of memories soon after they are acquired during the wake period means that they are not retained for consolidation during sleep. The Aβ deregulates Ca²⁺ signalling through direct effects on the neurons and indirectly by inducing inflammatory responses in the microglia and astrocytes. Some of these deleterious effects of Aβ may be alleviated by vitamin D.

Introduction

Neurons have highly developed Ca²⁺ signalling systems (Berridge, 2012) responsible for regulating neural functions such as brain rhythms, information processing, learning and memory. Remodelling of these Ca²⁺ signalling pathways that create inappropriate Ca²⁺ responses have been linked to many major neural diseases (Khachaturian 2013; LaFerla 1989; Stutzmann 2005; Thibault et al. 1992; Bezprozvanny & Mattson, 2012a; Berridge 1998, 2010,2011 b). In the case of Alzheimer's disease (AD), Ca²⁺ levels are set too high and this has an impact on many neural functions and particularly memory formation and consolidation. Why subtle deregulation of Ca²⁺ signalling can have such a marked effect on memory remains somewhat mysterious. Memory formation is a complex process that depends on memory acquisition during consciousness followed by memory consolidation and erasure during sleep. To obtain a better understanding of AD, therefore, the first part of this review explores the way Ca²⁺ signalling participates, in not only controlling brain rhythms, but also how it generates memories that occur during the markedly different brain rhythms that characterize the sleep–wake cycle. The remodelling of Ca²⁺ signalling in AD may explain the observation that this sleep–wake cycle is markedly disrupted in AD with an increase in wakefulness (Roh et al. 2008). The second part of the review explores the possibility that the deregulation of Ca²⁺ signalling that occurs in AD results in the continuous activation of a Ca²⁺-dependent memory erasure mechanism, which is normally restricted to the early phase of sleep. Such a mechanism could explain why memories are rapidly lost soon after they are acquired during periods of wakefulness.

Michael Berridge is an Emeritus Babraham Fellow at the Babraham Institute in Cambridge. He received his B.Sc. in 1960 from the University of Rhodesia and Nyasaland in Salisbury, Rhodesia and was awarded his Ph.D. from the University of Cambridge in 1964. His current research interest is focused on the role of inositol trisphosphate (InsP₃) in controlling the spatiotemporal aspects of Ca²⁺ signalling with particular emphasis on neural signalling and neural diseases such as Alzheimer's disease, bipolar disorder and schizophrenia.

Neural rhythms and neural functions

The brain is highly rhythmical and the different neural rhythms that occur during the sleep–wake cycle regulate its multiple functions. These rhythms can be divided into the fast gamma (20–80 Hz), alpha (8–12 Hz) and theta (6–10 Hz) oscillations, which occur during the awake state, and the slower delta (1–4 Hz) and slow oscillations (<1 Hz) that occur during sleep (Fig. 1). These different oscillatory modes are regulated by the ascending arousal system that consists of a heterogeneous population of neurons that project their axons throughout the brain where transmitters such as orexin, acetylcholine, noradrenaline, 5-hydroxytryptamine, histamine and dopamine are released on to the excitatory and inhibitory neurons that constitute functional neural circuits (Pace-Schott & Hobson 2007; Datta 1996). These transmitters then act on receptors coupled to signalling pathways to adjust the level of the tonic excitatory drive that regulates these different oscillatory states (Fig. 1).

Figure 1.

Tonic excitatory drive and the control of neuronal rhythms. Neurons of the ascending arousal systems release transmitters such as orexin, acetylcholine (ACh), 5-hydroxytryptamine (5-HT), dopamine (DA) and noradrenaline (NA), which induce signalling systems to control neural rhythms that occur during the sleep/wake cycle. The tonic excitatory drive mechanism depends on membrane depolarization driven by different signalling mechanisms. Hydrolysis of the phospholipid PtdIns4,5P₂ has two effects. First, it closes the K_V 7.2/7.3 channels responsible for the M current. Switching off this M current depolarizes the membrane to increase neuronal activity. Secondly, the formation of inositol 1,4,5-trisphosphate (InsP₃) releases Ca²⁺ that stimulates the Ca²⁺-activated non-selective cation (CAN) channel. The CAN channel can also be activated by Ca²⁺ entering through the voltage-operated Ca²⁺ (VOC) channel. The NA and DA act through the cyclic AMP signalling pathway to enhance the activity of the hyperpolarizing-activated cyclic nucleotide-gated (HCN) channel responsible for the depolarizing I_h current.

Variations in the activity of this tonic excitatory drive functions much like a rhythm rheostat in that it controls a hierarchy of rhythms with the lowest frequencies occurring during sleep that are then switched to the higher frequency rhythms of the wake state (Fig. 1). The Ca²⁺ signalling mechanisms control the rhythm rheostat and have to operate within the parameters of the ongoing rhythms to regulate memory formation, consolidation and erasure. Memory acquisition occurs during fast gamma rhythms during consciousness, while the slow oscillations mediate memory consolidation and erasure during sleep. The sleep–wake cycle is markedly disrupted in AD with an increase in wakefulness associated with a decrease in the slow oscillation responsible for non-rapid eye movement (NREM) sleep rhythms (Roh et al. 2008).

Gamma and theta rhythms and calcium signalling mechanisms

The fast theta and gamma oscillations, which are highly synchronous throughout the brain, are generated by a typical network oscillator consisting of fast spiking inhibitory interneurons and excitatory neurons that interact with each other through a positive/negative feedback loop (Fig. 2). Each inhibitory interneuron controls the activity of a large array of excitatory neurons all of which send axon collaterals back to the inhibitory interneuron. Most information is available for the gamma rhythms where the interneurons fire an action potential on each cycle and this induces a brief hyperpolarization in all the excitatory neurons that occurs synchronously over extensive areas of the brain (Hájos & Paulsen 2013). In contrast to this continuous firing of the inhibitory neurons, the excitatory neurons fire much less frequently with each action potential occurring within a narrow time window towards the end of each pacemaker depolarization (Fig. 2). This gamma oscillation synchronization provides a precise timing mechanism that enables excitatory neurons located in different parts of the brain to communicate with each other by firing together as part of a circuit as they process sensory information. This action potential coincidence is also crucial for triggering the input-specific Ca²⁺ transients responsible for memory formation.

Figure 2.

Neuronal network gamma oscillations. Most neural circuits consist of fast spiking inhibitory interneurons (red) and excitatory neurons (green) interacting with each other through a positive/negative feedback loop (see inset at the bottom). Each interneuron controls the activity of many excitatory neurons (red arrows) all of which send axon collaterals back to the inhibitory interneuron (green arrow). The interneuron fires an AP on each gamma cycle and this serves to induce synchronous hyperpolarizations in all the excitatory neurons. The excitatory neurons fire much less frequently towards the end of the pacemaker depolarization. The inhibitory interneuron registers each of these APs as a small EPSP that sum to activate the interneuron to fire an AP to initiate another gamma oscillatory cycle. ACh, acetylcholine; AP, action potential; EPSP, excitatory postsynaptic potential.

Important Ca²⁺ signalling events occur when excitatory neurons, which are connected together as part of a neuronal circuit, fire action potentials that are synchronized by the gamma oscillation rhythm. Such action potential coincidence (Fig. 2) is important for memory formation through a process of spike time-dependent plasticity, which favours the induction of long-term potentiation (LTP) (Caporale & Dan 2010). Depolarization of a neuron while it receives a pulse of glutamate from another neuron serves to open the NMDA receptors (NMDARs) on the activated spines to allow rapid entry of Ca²⁺ to trigger LTP (Fig. 3A). A form of NMDAR-independent synaptic plasticity is induced by Ca²⁺ entering through Ca_V1.2 channels (Moosmang et al. 2013). The brief high concentration spike of Ca²⁺, which is highly localized within the activated spine, induces at least three immediate biochemical events: phosphorylation of the AMPA receptors (AMPARs), exocytosis of vesicles containing new AMPARs and activation of actin polymerization resulting in a change in the shape of the spine. These biochemical events are the basis of new memories (Citri & Malenka 2008) that are then retained until either consolidated or erased during the slow oscillations that occur during NREM sleep. The mechanisms responsible for memory formation seem to operate normally in AD, but are not retained long enough to be consolidated during sleep. A clue as to why these memories are rapidly erased in AD has emerged from analysing the mechanisms responsible for memory erasure that are normally restricted to the period of slow oscillations that occur during NREM sleep.

Figure 3.

Ca²⁺-induced synaptic plasticity during memory formation (A) and memory erasure (B). A, the opening of NMDA receptors (NMDARs) or Ca_v1.2 L-type Ca²⁺ channels generate large local increases in Ca²⁺ within specific spines (see inset). This localized Ca²⁺ signal induces long-term potentiation (LTP) that depends on AMPA receptor (AMPAR) phosphorylation by CaMKII, AMPAR insertion that increases the number of receptors in the postsynaptic membrane, actin remodelling that changes spine morphology and protein synthesis at polyribosomes located at the base of each spine. B, lower levels of Ca²⁺ in the spines, which may depend on Ca²⁺ diffusing in from the global dendritic Ca²⁺ waves of the slow oscillations that occur during sleep, activates long-term depression (LTD) that depends on calcineurin (CaN) acting to reverse many of the processes that occur during LTP.

Slow oscillation and calcium signalling mechanisms

The slow (<1 Hz) oscillations, which occur during NREM sleep, have been linked to both memory consolidation and erasure (Diekelmann & Born 2011). The onset of these endogenous slow oscillations, which occurs when the tonic excitatory drive is reduced at the onset of sleep (Fig. 1), has typical UP states when the membrane potential (V_m) is depolarized to approximately –65 mV that alternates with DOWN states where the membrane is hyperpolarized by 10–15 mV. These slow oscillations occur in near synchrony throughout the brain and this enables neurons to communicate with each other as they fire rapid bursts of action potentials on the crest of each UP state (Massimini et al. 2013). This synchrony is achieved through a slow wave propagation mechanism whereby the action potentials that occur at the start of the UP state in one neuron entrains neighbouring neurons to initiate their slow oscillation thereby a wave of excitation propagates through the brain travelling at approximately 3 m s⁻¹ in an anterior–posterior direction. The signalling mechanisms that occur during these slow oscillations are responsible for both memory consolidation and erasure.

A decline in the tonic excitatory drive that occurs during NREM sleep turns down the rhythm rheostat to set the stage for oscillations to occur (Fig. 1). A decrease in the activity of the metabotropic transmitters that stimulate the hydrolysis of PtdIns4,5P₂ enables this lipid to accumulate such that it can open the KV7.2/KV7.3 channels resulting in an increase in the M current that provides the membrane hyperpolarization that drives the DOWN state. The endogenous oscillator that drives the alternating UP and DOWN states depends on an interaction between different channels (Fig. 4) (Crunelli et al. 2008, 2007). During the DOWN state, there is steady pacemaker depolarization that depends on both a hyperpolarizing-activated cyclic nucleotide-gated 1 (HCN1) channel and a Ca²⁺-dependent non-selective cation channel (CAN). The latter is activated directly by Ca²⁺ whereas the HCN1 is modulated indirectly by Ca²⁺ acting on adenylyl cyclase (AC) to provide cyclic AMP. This cyclic AMP does not activate HCN1 directly, but it alters the open probability by shifting the activation curve towards more depolarizing potentials.

Figure 4.

This hypothetical scheme describes the main ionic channels and signalling mechanisms that have been implicated in generating the slow oscillations characterized by periodic UP (red panels) and DOWN (green panels) states. The CaV3 T-type channels, which are located on the dendrites, are switched on at the beginning of an UP state to provide a low-threshold Ca²⁺ spike(s), but then inactivate to become a persistent inward current, also known as the T window current, to provide a tonic depolarization that helps to maintain the UP state. Opening of these T channels at the beginning of each UP state results in a global elevation of Ca²⁺ that spreads rapidly down the dendrites and into the spines where it may act to induce long-term depression (see Fig. 3B). ACh, acetylcholine; CaN, calcineurin; ER, endoplasmic reticulum; 5-HT, 5-hydroxytryptamine; InsP3, inositol 1,4,5-trisphosphate; HCN, hyperpolarizing-activated cyclic nucleotide-gated (channel).

The beginning of the UP state is characterized by the opening of Ca_V3 T-type channels that produce a low-threshold Ca²⁺ spike(s), but these channels then inactivate to become a persistent inward current, also known as the T window current, to provide a tonic depolarization that helps to maintain the UP state (Crunelli et al. 2008, 2007). The opening of these T type channels, which are located on the soma and dendrites, results in a global elevation of Ca²⁺ that not only spreads through the dendrites, but it also diffuses into the spines (Errington et al. 2011). During the course of a slow oscillation, therefore, there are two important Ca²⁺ signalling events taking place that may have a major impact on memory formation. First, the global elevation that spreads into all of the spines may function to activate long-term depression (LTD) to erase temporary memories (Fig. 3B). Secondly, it is likely that there are localized pulses of Ca²⁺ within specific spines during rapid action potentials that occur on the crest of each slow oscillation as memories that are reactivated in the hippocampus are redistributed to the cortex where they are consolidated into longer-term memories during sleep.

It has been proposed that memories acquired during the wake period are stored temporarily in the synaptic connections of relevant brain circuits located mainly in the hippocampus. Some of these temporary memories represent novel information that will be retained, but many other memory traces are erased during sleep to avoid cluttering up the brain with irrelevant information. This synaptic homeostasis hypothesis proposes that the redundant information stored in those potentiated synaptic connections are returned to baseline levels during sleep (Tononi & Cirelli, 2006; Vyazovskiy et al. 2009). The way in which some memories are consolidated and retained while others are erased is not known. The following speculation suggests that these two processes may run concurrently, even within individual neurons, due to a spatial separation of the Ca²⁺ signalling events during slow oscillations.

The fast action potentials that occur during the UP state of the slow oscillation (Fig. 4) probably reflects the neural activity responsible for memory consolidation. The spindles and ripples that appear in EEG recordings during sleep are thought to reflect the hippocampal–cortical dialogue as labile information stored in the hippocampus is transferred for more permanent storage in the cortex (Ji & Wilson 2013). The ripples represent high-frequency bursts of action potentials as memories are reactivated in the hippocampus, whereas the spindles result from the oscillatory firing of thalamocortical neuronal loops as this information is received and consolidated in the cortex (Diekelmann & Born 2011). Activation and transfer of these memories may depend on the same spike time-dependent plasticity mechanisms described earlier (Fig. 3A). The large and persistent pulsing of Ca²⁺, restricted to activated spines, will contribute to consolidation by stimulating protein synthesis by polysomes located at the base of each spine.

At the same time that novel memories are being consolidated, other memories are being deleted and it is conceivable that these processes may run concurrently in individual neurons. The erasure mechanism is not fully understood, but Errington et al. (2011) have suggested that the global elevation in Ca²⁺ that occurs during the slow oscillation might play a role in the mechanism of ‘homeostatic synaptic plasticity’. The most probable explanation for this is that the global elevation of Ca²⁺, which is known to invade the spines, will activate LTD. The specific proposal is that Ca²⁺ in the dendrites diffuses into the spine where it may reach levels in the 300–500 nm range that will activate LTD resulting in memory erasure (Fig. 3B). This Ca²⁺-dependent erasure depends on activation of calcineurin that reverses the three processes that occurred during LTP, i.e. AMPARs are dephosphorylated, AMPARs are retrieved from the spine surface through endocytosis and the actin filaments are depolymerized. In this way, the conundrum as to how memory consolidation and erasure can run concurrently can be resolved if high levels of Ca²⁺ necessary for LTP are localized to those spines undergoing consolidation, whereas all redundant memories can be erased en masse through the global elevation of lower levels of Ca²⁺ that pervade the spines to activate LTD. The loss of memory in AD may result from a similar erasure being activated continuously during the wake period when memories are acquired.

Dysregulation of Ca²⁺ signalling and Alzheimer's disease

The development of AD is driven by the accumulation of amyloid β (Aβ) oligomers, which are a neuron-derived pathogenic factor that brings about the loss of memory and neuronal cell death that characterizes the progression of AD. The Ca²⁺ hypothesis of AD suggests that these deleterious effects of Aβ depend on a dysregulation of Ca²⁺ signalling (Khachaturian 2013; LaFerla 1989; Stutzmann 2005; Thibault et al. 1992; Bezprozvanny & Mattson 2012a; Stutzmann & Mattson 2012; Berridge 2013, 1998, 2011b). The basic idea is that abnormal amyloid metabolism induces an upregulation of neuronal Ca²⁺ signalling that is responsible for the initial decline in memory and subsequent apoptosis. When Ca²⁺ is measured in the spines and dendrites of cortical pyramidal neurons of transgenic mice, there was a higher than normal resting level in those neurons located close to amyloid deposits (Kuchibhotla et al. 2007). Similarly, the resting level of Ca²⁺ in the cortical neurons of triple transgenic AD animals was 247 nmol l⁻¹, which was twice that found in the non-transgenic controls (110 nmol l⁻¹) (Lopez et al. 2011). In addition, there is increasing evidence that Aβ also acts on neighbouring microglial cells and astrocytes (Abramov et al. 2004; Saijo & Glass, 2008) to induce local inflammatory responses that contributes to Ca²⁺ signalling deregulation. The following sequence of events attempts to explain these multiple actions of Aβ on neurons, microglia and astrocytes to induce the upregulation of Ca²⁺ signalling that may be responsible for AD (the numbers on Fig. 5 correspond to those outlined below):

Figure 5.

Ca²⁺ hypothesis of AD. The development of AD is induced by the accumulation of Aβ oligomers that have a number of actions. The Aβ can act directly on the neurons to bring about elevations in Ca²⁺ that have been linked to the initial phase of memory loss and the subsequent increase in apoptosis that characterizes the development of AD. The Aβ can also induce inflammatory responses in neighbouring microglia and astrocytes that activate processes that enhance this dysregulation of Ca²⁺ signalling. Vitamin D3 may alleviate the development of AD by inhibiting the inflammatory responses and by increasing the expression of processes that reduce the elevation of Ca²⁺. Aβ, amyloid β; AD, Alzheimer's disease; GSH, glutathione; IL, interleukin; InsP₃, inositol 1,4,5-trisphosphate; mTOR, mammalian target of rapamycin; PLC, phospholipase C; PMCA, plasma membrane Ca²⁺-ATPases; ROS, reactive oxygen species; RYR, ryanodine receptor; TNF, tumor necrosis factor; VDR, vitamin D receptor; VDRE, vitamin D response element.

1. Neuronal Ca²⁺ signalling deregulation may depend on changes in both the entry of external Ca²⁺ and its release from internal stores. The Aβ oligomers that accumulate outside diseased neurons can bring about an elevation in Ca²⁺ through different mechanisms. They can be inserted into the membrane to form channels (Demuro et al. 2010) or they can activate the Ca²⁺-sensitive receptor (CaSR) to increase the level of InsP₃ (Ye et al. 2006; Chiarini et al. 2008; Armato et al. 2004). The CaSR is coupled to phospholipase C through the G protein G_q, which is inhibited by the regulator of G protein signalling 4 (RGS4). The level of RGS4 is reduced in the human AD brain and this may further enhance the generation of InsP₃ (Emilsson et al. 2010). Injection of Aβ into Xenopus oocytes stimulated the production of InsP₃ through a G protein-dependent activation of phospholipase C (Demuro & Parker 2007) providing further evidence that some of the actions of Aβ might be mediated by an increase in the InsP₃/Ca²⁺ signalling pathway.

2. An increase in the formation of InsP₃ will enhance the amount of Ca²⁺ being released from the endoplasmic reticulum by the InsP₃ receptors (InsP₃Rs). Indeed, a feature of AD is an increase in the activity of InsP₃Rs (Cheung et al. 2013; Müller et al. 2006). This elevation in InsP₃-dependent Ca²⁺ signalling, which needs to be turned down for slow oscillations to occur during sleep, may explain the observation that the sleep–wake cycle is markedly disrupted in AD with an increase in wakefulness (Roh et al. 2008). In transgenic mice, an increase in amyloids may induce global elevations of Ca²⁺ through a burst firing-mediated mechanism that depends on glutamate activating an InsP₃/Ca²⁺ signalling pathway (Czarnecki et al. 2006). Expression of the Ca_v1.2 L-type Ca²⁺ channel, which has been implicated in LTP induction (Moosmang et al. 2013) (Fig. 3A), is induced by Aβ (Webster et al. 2012; Dursun et al. 2009) and this will enhance the release of Ca²⁺ from ryanodine receptors (RYRs). Such an action would be enhanced further by the increased expression of the RYR, particularly the RYR3 isoform (Supnet et al. 2007). Neuronal levels of the Ca²⁺ buffer calbindin-28 k are known to be reduced in AD (Sutherland et al. 2011). In addition, Aβ may also reduce Ca²⁺ extrusion from the cell by inhibiting both the plasma membrane Ca²⁺-ATPase (PMCA) and the Na⁺/K⁺-ATPase that maintains the Na⁺/Ca²⁺ exchanger (NCX) (Mark et al. 2011). Thus, there are a number of mechanisms that could contribute to the upregulation of Ca²⁺ signalling to account for persistent elevation in the resting level of Ca²⁺ (Kuchibhotla et al. 2007; Lopez et al. 2011).

3. This dysregulation of neuronal Ca²⁺ signalling seems to be exacerbated by Aβ-induced neuroinflammation that occurs during AD (Saijo & Glass 2008). Aβ-induced Ca²⁺ signals can enhance microglial inflammatory responses by increasing the release of cytokines and reactive oxygen species (ROS) (Farber & Kettenmann 2006). The Aβ also acts through CaSRs to produce InsP₃ that then releases Ca²⁺ from internal stores (Lee et al. 2008). Depletion of these stores then triggers store-operated Ca²⁺ entry through the Orai1 channel (Ohana et al. 2011) that is maintained by the hyperpolarization induced by the calcium-activated potassium channel KCa3.1. Microglial-dependent neurotoxicity could be reduced in vivo by inhibiting these KCa3.1 channels with triarylmethane-34 (Kaushal et al. 2007; Maezawa et al. 2008) thus emphasizing the significance of Ca²⁺ in regulating neuroinflammation. Drugs thought to reduce inflammation such as Ro5-4864, which is a ligand for the translocator protein, reduce both the accumulation of Aβ and the decline in cognition in transgenic AD mice (Barron et al. 2000).

4. One of the consequences of the Aβ-dependent elevation of microglial Ca²⁺ is activation of the inflammasome. The oligomers that are taken up in the phagosome vesicles enter the cytosol to increase NLRP3 activity resulting in stimulation of caspase-1 that cleaves pro-interleukin-1β (IL-1β) to form IL-1β (Fig. 5). The inflammasome inhibits phagocytic clearance of Aβ and has been strongly implicated in AD (Heneka et al. 2008). The triggering receptor expressed in myeloid cells 2 (TREM-2), which functions as a negative regulator of innate immunity, suppresses the ability of the microglia to release inflammatory mediators such as tumour necrosis factor (TNF)α. TREM2 is a transmembrane glycoprotein that associates with DNAX-activating protein 12. A variant of TREM2, which reduces the anti-inflammatory role of TREM2, is associated with a markedly increased risk of developing AD (Guerreiro et al. 2002; Jonsson et al. 2013).

5. Aβ can also induce microglial inflammation by stimulating toll-like receptors (TLR-2 and TLR-4). Polymorphisms in these receptors have been associated with an increased susceptibility and progression of AD (Minoretti et al. 2011; Yu et al. 2009, 1997). Activation of the TLR-2/4 receptors can have both beneficial and deleterious actions. With regard to the former, TLR-2/4 receptors stimulate phagocytosis that removes and destroys Aβ. The deleterious effects depend upon activation of the NF-κB signalling pathway that increase release of proinflammatory mediators such as TNFα and ROS, all of which can enhance neuronal Ca²⁺ signalling. Increased susceptibility and progression of AD have been linked to polymorphisms in TNFα (Di Bona et al. 2013; Yang et al. 2004). The monophosphoryl lipid A has an interesting property of being able to promote the beneficial phagocytic mechanism while not inducing the deleterious inflammatory response and thus may be an effective treatment for AD (Michaud et al. 2004). Omega-3 fatty acids can also enhance phagocytosis of Aβ to reduce the formation of proinflammatory cytokines (Hjorth et al. 2009).

6. The TNFα and IL-1β released from the microglia can have a number of actions. The TNFα binds to the TNF receptor to contribute to neuronal cell death by activating apoptosis. The TNFα can also enhance Ca²⁺ signalling by acting through the JNK signalling pathway to increase the expression of InsP₃R1 by phosphorylating the transcription factor specificity protein 1 (Sp1) (Park et al. 2009). Both TNFα and IL-1β can also influence memory by altering the electrophysiological correlates of LTP and LTD (Cunningham et al. 2005; Albensi & Mattson 2004).

7. The exogenous ROS, which diffuses into the neuron, will add to that being produced by the mitochondria and can enhance Ca²⁺ signalling in several ways. It can increase the sensitivity of both the InsP₃Rs and RYRs to increase the release of Ca²⁺ while inhibiting the PMCA Ca²⁺ pump (Lock et al. 2012). These two effects of ROS, which will increase the resting level of Ca²⁺, will potentially set up a positive feedback system in that the excess Ca²⁺ will increase mitochondrial ROS formation (Müller et al. 2006). Inhibition of ROS formation by a mitochondrial-targeted antioxidant MitoQ prevents the cognitive decline in a transgenic mouse model of AD (McManus et al. 1995).

8. Aβ acts on the astrocytes to induce an inflammatory response and the resulting increase in ROS decreases the level of the antioxidant glutathione (GSH), which has serious repercussions for the neuron as it receives its GSH from the astrocytes (Abramov et al. 2004). A decrease in neuronal GSH levels will enable ROS to have a greater impact on Ca²⁺ levels as outlined above.

9. A number of mechanisms have been proposed to describe how the elevation in the resting level of Ca²⁺ accounts for symptoms of AD. The characteristic loss of memory that occurs in the early stages of AD may be driven by the abnormal resting levels of Ca²⁺, which is manifest either as a persistent elevation in the level of Ca²⁺ in the dendrites and spines (Kuchibhotla et al. 2007) or as an increase in spontaneous Ca²⁺ transients (Busche et al. 2008, 2006). This increase in Ca²⁺ signalling, may disrupt cognition by activating LTD through the mechanism described earlier (Fig. 3B) (Berridge, 2013, 1998). As the disease progresses, elevation of Ca²⁺ will begin to rise further to a point where it will activate apoptosis resulting in the neuronal cell loss responsible for the final stages of dementia.

10. A marked feature of AD is a decline in autophagy (Son et al. 2011), which seems to be associated with an increase in the activity of the mammalian target of rapamycin (mTOR) (Caccamo et al. 2008, 2010). The InsP₃R is known to play a role in autophagy by assembling a complex containing regulators such as Beclin-1, Bcl-2 and hVps34 (Criollo et al. 2009; Vicencio et al. 2006). The level of Beclin-1, which is a key component of the autophagy complex, is known to be reduced in AD (Pickford et al. 2009). The decline in autophagy in AD may be related to an increase in InsP₃ that disrupts the autophagic complex by binding to the InsP₃R. The drug Li⁺, which is known to reduce the risk of developing AD (Nunes et al. 2005), can reduce this inhibitory effect by lowering the level of InsP₃ (Sarkar et al. 2012). Autophagy may also be reduced in AD by the elevated levels of Ca²⁺ that can disrupt the complex by activating hVps34 (Gulati et al. 2011). The activation of hVps34 may also account for the increase in mTOR (Gulati et al. 2011) that could explain the decline of autophagy in AD. The cognitive decline in mouse models of AD is reduced by rapamycin, which inhibits the activity of mTOR (Caccamo et al. 2008, 2010). Another role for mTOR is to phosphorylate Tau to increase its pathological role in AD. The elevation of Ca²⁺ can also stimulate CaMKK2 to increase the activity of AMPK that then enhances the phosphorylation of Tau thus contributing to the symptoms of AD (Mairet-Coello et al. 2013).

11. There is considerable evidence for a link between vitamin D deficiency and the onset of AD and other neurodegenerative diseases such as multiple sclerosis and Parkinson's disease (Garcion et al. 2012; Tuohimaa et al. 2007; Berridge 2011b; Wang et al. 2009; Lu'o'ng & Nguyên 2011). Expression of the vitamin D receptor (VDR) is reduced in the hippocampus of patients with AD (Sutherland et al. 2011) and VDR polymorphisms have been identified as risk factors for AD (Lehmann et al. 2002; Wang et al. 2009).There is evidence that Aβ acts to both reduce the expression of VDR while increasing the expression of the Ca_v2.1 Ca²⁺ channel.

1. Administration of vitamin D can reverse many of the changes induced by Aβ. In cultured primary neurons, vitamin D acts to increase the expression of the VDR and it reduces the expression of the Ca_v1.2 L-type Ca²⁺ channel (Brewer et al. 2012b; Taniura et al. 2006; Dursun et al. 2009; Gezen-Ak et al. 2006). In the intestine, vitamin D is known to increase the expression of proteins such as PMCA, NCX1 and Ca²⁺ buffers such as calbindin-28 k and parvalbumin (Wasserman 2008; Pérez et al. 2002). It seems reasonable to propose that vitamin D may reduce the risk of AD by promoting the expression of those proteins, such as PMCA and NCX1, that act to lower the level of intracellular Ca²⁺ (Berridge 2011b). Vitamin D can also have a beneficial effect by dampening down the microglial inflammatory responses to reduce the formation of TNFα.

In summary, the build-up of Aβ oligomers during the onset of AD has a profound effect on the activity of the local community of cells in the brain. The inflammatory response in both the microglia and astrocytes contribute to dysregulation of neural Ca²⁺ signalling that seems to be one of the major factors in the development of AD. It is argued that in the early stages of AD, this alteration in signalling is manifest as a persistent elevation of the resting level of Ca²⁺ that results in memories acquired during the wake period being rapidly erased before they can be consolidated during sleep. Vitamin D may play a critical role in memory retention by regulating the expression of the Ca²⁺ components necessary to maintain low resting levels of Ca²⁺.

Glossary

AD

Alzheimer's disease

Aβ

amyloid β

CAN

Ca²⁺-activated non-selective cation channel

CaSR

Ca²⁺-sensitive receptor

GSH

glutathione

HCN1

hyperpolarizing-activated cyclic nucleotide-gated 1

InsP₃

inositol 1,4,5-trisphosphate;

LTD

long-term depression

LTP

long-term potentiation

mTOR

mammalian target of rapamycin

NCX

Na⁺/Ca²⁺ exchanger

NREM

non-rapid eye movement

NLRP

NACHT, LRR and PYD domains-containing protein

PMCA

plasma membrane Ca²⁺-ATPases

PtdIns4,5P₂

phosphatidylinositol 4,5-bisphosphate

RGS4

regulator of G protein signalling 4

ROS

reactive oxygen species

RYR

ryanodine receptor

TREM-2

triggering receptor expressed in myeloid cells 2

TNFα

tumour necrosis factor α

VDR

vitamin D receptor

Competing interests

None declared.