A wide range of Ca2+ signalling systems deliver the spatial and temporal Ca2+ signals necessary to control the specific functions of different cell types. Release of Ca2+ by InsP3 (inositol 1,4,5-trisphosphate) plays a central role in many of these signalling systems. Ongoing transcriptional processes maintain the integrity and stability of these cell-specific signalling systems. However, these homoeostatic systems are highly plastic and can undergo a process of phenotypic remodelling, resulting in the Ca2+ signals being set either too high or too low. Such subtle dysregulation of Ca2+ signals have been linked to some of the major diseases in humans such as cardiac disease, schizophrenia, bipolar disorder and Alzheimer's disease.
Calcium (Ca2+) is a highly versatile intracellular signal capable of regulating many different processes . To achieve this versatility, the signalling system operates in many different modes, thus enabling it to function over a wide dynamic range. It can trigger exocytosis at synaptic endings within microseconds and muscle contraction in milliseconds, whereas, at the other end of the scale, it can operate over minutes to hours to drive processes such as gene transcription and cell proliferation. This versatility depends on each cell type having a specific Ca2+ signalling mechanism that is assembled from a very extensive Ca2+ signalling toolkit [1,2]. As cells differentiate, they express a unique set of toolkit components to create Ca2+ signalling systems with widely different spatial and temporal properties. Such Ca2+ signalling systems are not set in stone, but are constantly being remodelled to adapt to changing circumstances to ensure that each specific cell type continues to deliver the Ca2+ signals that characterizes its unique function. If the spatiotemporal properties of this output signal change due to a loss or defect of a key component, compensatory mechanisms come into play to restore the normal output signal. This remodelling process implies that the quality of the output signal is under constant review, with Ca2+ itself playing a critical role in this internal assessment mechanism. The Ca2+-sensitive protein phosphatase CaN (calcineurin) activates the transcription factor NFAT (nuclear factor of activated T-cells), which plays a critical role in this remodelling process. A number of major diseases such as heart disease, schizophrenia, BD (bipolar disorder) and AD (Alzheimer's disease) may result from abnormal Ca2+ signalling remodelling.
Ca2+ signalling toolkit
The Ca2+ signalling toolkit
There are a large number of receptors [GPCRs (G-protein-coupled receptors) and PTKRs (protein tyrosine kinase-linked receptors)] that are coupled to different PLC (phospholipase C) isoforms that generate the Ca2+-mobilizing second messenger InsP3 (inositol 1,4,5-trisphosphate).
The channels that introduce Ca2+ into the cytoplasm during cell activation are located either in the plasma membrane or on various organelles. The VOCs (voltage-operated channels), which are divided into three families: CaV1 (L-type), CaV2 (N-, P/Q- and R-type) and CaV3 (T-type), have specific functions in different cell types. There also are channels activated by agonists operated from outside or by various stimuli originating from inside the cell.
Release from internal stores represents a major source of signal Ca2+ for many cells. The major Ca2+ store is the extensive ER (endoplasmic reticulum) network that contains both the RyR (ryanodine receptor)  and InsP3R (InsP3 receptor) families . These channels are activated by Ca2+, enabling these release channels to excite each other to create intracellular Ca2+ waves during the globalization of Ca2+ signals. There also is an acidic store that contains the family of TPCs (two-pore channels) that are activated by NAADP (nicotinic acid–adenine dinucleotide phosphate) .
Ca2+ pumps and exchangers
Ca2+ pumps and exchangers are responsible for pumping Ca2+ out of the cell or back into the ER. These pumps and exchangers operate at different times during the recovery process. The Na+/Ca2+ exchangers have low affinities for Ca2+, but have very high capacities and this enables them to function at the beginning of the recovery process to rapidly remove large quantities of Ca2+ and are especially evident in excitable cells. On the other hand, the PMCA (plasma membrane Ca2+-ATPase) and SERCA (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase) pumps have lower capacities, but their higher affinities mean that they can complete the recovery process and can continue to pump to maintain the low resting levels of Ca2+ at approximately 100 nM.
Cells express a large number of Ca2+-binding proteins  that function as buffers both in the cytoplasm [PV (parvalbumin), CB (calbindin D-28k) and calretinin] and within the lumen of the ER [calsequestrin, calreticulin, GRP (glucose-regulated protein) 78 and GRP94]. The spatial and temporal properties of Ca2+ signals are shaped by their rapid binding to the cytosolic buffers. The ER buffers enable the store to accumulate the large amounts of Ca2+ necessary for rapid cell signalling. The mitochondria also play an important buffering role in that they express a newly discovered MCU (mitochondrial calcium uniporter) [7,8] that takes up large amounts of Ca2+ whenever the cytosolic levels rise and this is then extruded by a Na+/Ca2+ exchanger during the recovery period. The uptake of Ca2+ stimulates the oxidative processes that produce ATP and can also generate ROS (reactive oxygen species) that contributes to the redox signalling pathway. An alteration in this normal ebb and flow of Ca2+ through the mitochondria can be deleterious when the uptake of Ca2+ is abnormally high, resulting in activation of the MTP (mitochondrial permeability transition pore) and the release of proteins such as cytochrome c to trigger apoptosis.
The second messenger function of Ca2+ is carried out by a wide range of Ca2+ sensors that detect changes in Ca2+ concentration and then activate different downstream responses. Most of these sensors fall into two main families characterized by having either EF-hand or C2 Ca2+-binding domains.
Ca2+-sensitive cellular processes
Many cellular processes are regulated by Ca2+ stimulating downstream effector systems either directly or indirectly. For example, the C2 domain synaptotagmins can trigger exocytosis directly. Similarly, troponin C acts directly to stimulate the interaction between actin and myosin to control contraction in muscle cells. The Ca2+-sensitive K+, Cl− and Na+ channels that regulate membrane excitability and fluid secretion are also activated directly by Ca2+ or by the Ca2+ sensor calmodulin. Alternatively, the Ca2+ sensors act indirectly by recruiting a range of intermediary effectors such as Ca2+-sensitive enzymes [e.g. CaMKs (Ca2+/calmodulin-dependent protein kinases), CaN, MLCK (myosin light chain kinase) and phosphorylase kinase]. Many other signalling pathways can be recruited by the Ca2+ signalling pathway through complex signalling cross-talk mechanisms. For example, Ca2+ activates the cAMP, MAPK (mitogen-activated protein kinase) and PI3K (phosphoinositide 3-kinase) signalling pathways to control changes in synaptic strength during learning and memory. The ability of Ca2+ to regulate the dynamics of actin polymerization/depolymerization controls the cytoskeletal remodelling necessary for processes such chemotaxis, formation of pseudopodia during phagocytosis, cytokinesis during cell division, formation of osteoclast podosomes during bone resorption and spine growth during learning and memory. The ability of Ca2+ to regulate longer-term changes in cells such as protein synthesis, gene transcription and chromatin remodelling has a major impact on cell remodelling and cell proliferation.
Phenotypic expression and stability of Ca2+ signalling mechanisms
During the final phase of development, specific cell types select out and express those components of the toolkit that are necessary to control their particular cellular functions. For example, cardiac ventricular cells select out the CaV1.2 channel that provides the Ca2+ sparklets that are amplified by type 2 RyRs to provide the sparks that activate contraction. In contrast, the antigen-dependent Ca2+ signals in T-cells depend on InsP3-induced Ca2+ release from the ER. As the latter empties, STIM1 (stromal-interaction molecule 1) activates Orai1 channels in the plasma membrane to induce the entry of external Ca2+ to provide the much longer-lasting Ca2+ signals that drive proliferation . The expression of such cell-type-specific signalling phenotypes are generated and maintained by differential gene transcription. However, such signalosomes are not set in stone, but can be remodelled during both normal and pathological conditions. Indeed, abnormal remodelling of signalosomes is a major cause of disease.
Cells can regulate the transcription of their toolkit components in order to maintain the phenotypic stability of their individual signalling systems. It seems that cells operate a quality assessment system whereby the properties of the output signals are constantly monitored and any deviations are fed back to the transcriptional system to make the necessary adjustments. Such autoregulatory mechanisms are particularly evident for the Ca2+ signalling system and may form the basis of the many examples of compensation that have been observed when specific signalling components are either deleted or overexpressed. For example, overexpression of the L-type channel in cardiac cells, which increases the amount of Ca2+ entering the cell, is counteracted by an up-regulation of the Na+/Ca2+ exchanger to increase Ca2+ extrusion .
A process of Ca2+-induced transcription of Ca2+ toolkit components may be responsible for maintaining the stability of Ca2+ signalling systems (Figure 2). The genes that encode many components of the Ca2+ signalling toolkit are regulated by Ca2+-dependent transcription factors such as NFAT and CREB (cAMP-response-element-binding protein). For example, NFAT can activate the transcription of genes such as endothelin (a potent activator of receptors that generate Ca2+ signals), the TRP (transient receptor potential) family member TRPC3 (canonical TRP channel 3), InsP3R, NFAT2, DSCR1 (Down's syndrome critical region 1), which encodes a protein that inhibits the activity of CaN, DYRK1A (dual-specificity tyrosine-phosphorylated and regulated kinase 1A), SERCA, the large conductance K+ channels (BK) and its KCaβ1 subunit (Figure 2). CREB activates the expression of the anti-apoptotic factor Bcl-2, which can inhibit Ca2+ release by the InsP3Rs (see also Figure 4).
Expression of the Ca2+ signalling toolkit components by Ca2+-sensitive transcription factors such as CREB and NFAT
The DSCR1 and DYRK1A genes are located within the critical region of human chromosome 21 where trisomy occurs in patients with Down's syndrome. The increased expression of these two genes will conspire to reduce the activation of NFAT: DSCR1 inhibits CaN and DYRK1A enhances the phosphorylation and nuclear export of NFAT. Some of the symptoms of Down's syndrome may be explained by such a reduction in the activity of NFAT, since this transcription factor has been implicated in the development of both the brain and facial features. Down's syndrome may thus be an example of how subtle alterations in Ca2+ signalling pathways can have devastating pathological consequences as described in more detail in the following section.
Ca2+ signalling remodelling and disease
Many of the major diseases in humans may arise through the remodelling of Ca2+ signalling pathways  to generate inappropriate Ca2+ responses that are set either too high or too low. What is remarkable about this remodelling is that alterations in the activity of the InsP3/Ca2+ signalling pathway seem to be responsible for some of the major diseases in humans.
P3/Ca2+ signalling mechanisms and atrial arrhythmias
The InsP3/Ca2+ signalling pathway has a major role in regulating the contraction of atrial cells [12–15]. These atrial cells express the type 2 InsP3Rs that are localized at the junctional zone where Ca2+ entry triggers the release of stored Ca2+ to stimulate contraction . Under normal conditions, these InsP3Rs play little role in the activation process, but if these atrial cells are stimulated in the presence of cardiac hormones, such as ET-1 (endothelin-1) or α-adrenergic agents that generate InsP3, there is an increase in the force of contraction and the regular beats can become chaotic just like the atrial arrhythmias seen in the diseased heart. Since these cardiac hormones are elevated during heart disease, it is likely that abnormal increases in InsP3 may explain the development of the arrhythmias that are a major cause of sudden death syndrome. Increases in InsP3 are also responsible for the arrhythmogenic action of ET-1 in ventricular cardiac myocytes .
These InsP3-induced cardiac arrhythmias can be prevented by adding the channel inhibitor 2-APB (2-aminoethoxydiphenylborate), which is known to act by blocking Ca2+ release from the InsP3R . The fact that 2-APB can completely restore the normal Ca2+ transients suggests that the InsP3R may be a very significant target for the development of drugs to prevent the cardiac arrhythmias that are one of the main causes of sudden heart death.
P3/Ca2+ signalling mechanisms and cardiac hypertrophy
There is increasing evidence that cardiac hypertrophy and CHF (congestive heart failure) may arise from an inappropriate remodelling of the cardiac Ca2+ signalling system. Heart disease is characterized by a decrease in the ability of the heart to pump blood around the body. This dysfunction is a major cause of human morbidity and mortality. CHF can be induced by multiple factors, many of which appear to act by increasing the workload of the heart. For example, blood pressure increases during hypertension is a major cause of heart disease. The onset of CHF can also be induced by endocrine disorders that increase the circulation of hormones such as catecholamines, ET-1 and angiotensin II. The last two hormones operate through the InsP3/Ca2+ signalling system, which seems to play an important role in the induction of hypertrophy.
There is now convincing evidence that Ca2+ itself may be the major signal for inducing the onset of cardiac hypertrophy that then leads inexorably to CHF. Under normal conditions, however, the heart receives continuous pulses of Ca2+ to drive contraction without initiating a change in transcription. So what is it about the hypertrophic Ca2+ signals that initiate the remodelling of cardiac gene transcription? An early explanation of this conundrum was that hypertrophy might be driven by subtle changes in the spatial or temporal properties of the individual Ca2+ transients [2,17]. Just how these changes were induced was not clear until it was revealed that a localized activation of the InsP3/Ca2+ signalling system in the perinuclear region was capable of activating transcriptional processes  and suggested that a distinct nuclear Ca2+ signal comes into play to drive hypertrophy .
The changes in the spatiotemporal properties of each transient that were proposed to be responsible for driving hypertrophy [2,17] may result from these InsP3-dependent nuclear Ca2+ signals [20,21]. Under normal conditions, contraction is activated by the RyRs releasing a Ca2+ signal that is largely confined to the cytoplasm with little impact on events in the nucleus. If the ambient level of InsP3 is elevated due to hypertrophic stimuli, then each of these transients will be amplified near the nucleus by the perinuclear InsP3Rs. The stimulation of hypertrophy may be enhanced by NCS-1 (neuronal calcium sensor 1), which interacts with InsP3Rs to increase the InsP3-dependent signals . The basic idea is that these perinuclear InsP3Rs function as coincident detectors that respond to both the ambient elevation of InsP3 and the brief elevation of Ca2+ that occurs during each transient to create a nuclear Ca2+ microdomain responsible for driving the nuclear transcriptional processes that initiate hypertrophy and subsequent heart disease [20,21]. There is now considerable experimental evidence to show that InsP3Rs can function to enhance nuclear Ca2+ transients and to control hypertrophy [23–26]. These InsP3-induced elevations in nuclear Ca2+ signalling bring about the subtle alteration in the normal processes of cardiac excitation–contraction coupling to initiate the processes of hypertrophy that then leads inexorably to CHF.
Neurons have a highly developed Ca2+ signalling system  that is responsible for regulating a large number of neural functions such as the control of brain rhythms, information processing and the changes in synaptic plasticity that underpin learning and memory. Remodelling of the Ca2+ signalling pathway has been implicated in the development of some of the major neural diseases such as AD, BD and schizophrenia.
The Ca2+ hypothesis of AD proposes that dysregulation of Ca2+ signalling is responsible for AD [28,29]. The basis of the Ca2+ hypothesis is that abnormal amyloid metabolism induces a change in the Ca2+ signalling homoeostasis that then initiates both the progressive decline in memory and the increase in neuronal cell apoptosis. Any explanation of AD, particularly sporadic AD, has to account for the slow progression of the disease and for the fact that the changes in synaptic physiology and onset of memory loss often precedes any evidence for the massive cell loss that characterizes the later stages of AD. In animal models of AD, the resting level of Ca2+ is elevated from approximately 100 nM to 300–500 nM and this might be responsible for the cognitive decline . A number of mechanisms have been proposed to account for this elevation in resting Ca2+, including an increase in the activity of InsP3Rs . In a mouse model, some of the pathogenic symptoms of AD, such as neuronal cell death, were alleviated by reducing the activity of the InsP3Rs . The challenge is to understand how this early dysregulation of Ca2+ signalling brings about this early loss of memory.
Our current understanding of the role of Ca2+ in learning and memory was used to extend the Ca2+ hypothesis of AD by suggesting a mechanism to explain how the early cognitive decline in AD may depend on a persistent elevation in the resting level of Ca2+ [33,34]. Fluctuations in the level of Ca2+ regulate the changes in synaptic plasticity responsible for learning and memory. Memory formation depends on brief high concentration spikes of Ca2+ that activate the process of LTP (long-term potentiation). Such recently acquired memories, which are located in the working memory store mainly in the hippocampus, are then uploaded and consolidated in more permanent memory stores in the cortex during certain phases of sleep. In contrast, smaller elevations in Ca2+ activate a process of LTD (long-term depression) that can erase the information in the working memory stores. In other words, Ca2+ has two diametrically opposed actions: it can both form and erase memories. In the case of AD, it is argued that the permanent elevation in the resting level of Ca2+  that occurs during early-onset AD may act to quickly erase memory traces before they can be consolidated. This hypothesis thus focuses attention on how the amyloid-dependent remodelling of Ca2+ signalling disrupts the mechanisms responsible for learning and memory.
Schizophrenia, which is a severe psychiatric condition that affects approximately 1% of the human population, is characterized by both positive (hallucinations and paranoia) and negative (poor attention, decline in social interactions and lack of motivation) symptoms . Some of these symptoms may be linked to subtle changes in the brain rhythms responsible for driving processes such as perception, consciousness and memory. One of the most consistent features of schizophrenia is a decline in working memory, which depends on storing bits of information for short periods in order to control a particular train of thought or a behavioural response. Such working memory, which is what is lost during early-onset AD as described above, depends on pyramidal neurons in different regions of the brain firing in a sustained and synchronous manner in the γ frequency range of approximately 40 Hz. Such γ rhythm generation and synchronization is impaired in schizophrenia [36–38] and has focused attention on the neurons responsible for generating this oscillatory activity. The essential components of the network oscillator are the inhibitory GABAergic interneurons and the excitatory pyramidal glutamatergic neurons . As part of the oscillatory cycle, the glutamatergic neurons release glutamate to excite the inhibitory interneurons to release GABA (γ-aminobutyric acid), which then feeds back to inhibit the pyramidal neurons to generate the γ rhythms.
The abnormal γ rhythms in schizophrenia seem to arise through a defect in the ability of the inhibitory interneuron to respond to glutamate . In particular, there is a decrease in the activity of the NMDAR (N-methyl-D-aspartate) receptor). This glutamate hypothesis  is supported by the observation that schizophrenic symptoms in healthy adults can be induced by administering NMDAR antagonists such as ketamine and PCP (phencyclidine). One of the functions of the NMDARs is to maintain the GABAergic phenotype of the inhibitory interneurons by generating Ca2+ signals that act through CaMKIV in the nucleus to phosphorylate the transcription factor CREB to control the expression of a number of signalling components that define the GABAergic phenotype (Figure 3). Hypofunction of the NMDARs results in a reduction in the expression of GAD67 (glutamic acid decarboxylase 67) that synthesizes the inhibitory transmitter GABA. This GABA is packaged into vesicles that are transported to the presynaptic terminals where they are released as part of the network oscillatory mechanism that generates the γ rhythms. A reduction in the level of GAD67 and the resulting decreased availability of GABA is one of the most characteristic features of schizophrenia. The transport of GABA-containing vesicles down the axon to the presynaptic ending is mediated by the dynein motor travelling down microtubules (Figure 3). One of the genes mutated in schizophrenia is DISC1 (disrupted in schizophrenia 1), which functions as a dynein adaptor. The mutations in DISC1 will thus disrupt the transport of GABA vesicles.
Functional modulation of GABAergic interneurons in schizophrenia
Some of the phenotypic changes that occur in the interneurons seem to be compensatory responses caused by the primary defect in the expression of GAD67. This may explain the decline in the level of PV, which modulates the Ca2+-dependent release of GABA. This release of GABA is controlled by a brief pulse of Ca2+ resulting from the entry of Ca2+ through CaV2.1 P/Q-type channels, which are located close to the exocytotic vesicles in order to increase the speed of the inhibitory response. This pulse of signal Ca2+ is rapidly buffered by PV to reduce the process of facilitation that depends on the build-up of Ca2+ during synaptic activity. In schizophrenia, PV levels decline, and this will enhance facilitation and may be a compensatory response to the reduction in GABA levels caused by the decrease in GAD67. Another compensatory mechanism depends on a decrease in the expression of GAT1 (GABA membrane transporter 1), which is located in the presynaptic ending where it pumps GABA back into the interneuron. A reduction in this removal mechanism will help to enhance the activity of the reduced amount of GABA being released in schizophrenia.
In summary, hypofunction of the NMDARs and the resulting reduction of Ca2+ signalling sets in train a complex series of transcriptional and compensatory events that remodel the GABAergic phenotype. The resulting reduction in the release of the inhibitory neurotransmitter GABA may explain the γ rhythm changes that occur in schizophrenia. Identifying the reason for the decrease in NMDAR function may thus provide a clue as to the biochemical basis for schizophrenia. The following working hypothesis attempts to integrate much of the genetic [e.g. NRG1 (neuregulin-1), DISC1 and VIPR2 (vasoactive intestinal peptide receptor 2)] and biochemical susceptibility factors [BDNF (brain-derived neurotrophic factor), DAOA (D-amino acid oxidase activator) and GSH (glutathione)] into a unifying concept of schizophrenia that describes the multiple mechanisms responsible for the decline in NMDAR activity and the Ca2+-dependent remodelling of the GABAergic phenotype.
NMDAR expression in schizophrenia
One of the prominent mutations that has been linked to schizophrenia is the NRG1 gene that encodes neuregulin-1 that acts through the ErbB3 signalling pathway [35,37] and contributes to the function of GABAergic neurons in two ways. First, it contributes to the GABAergic neuronal phenotype by controlling the expression of various signalling components that are related to NMDAR function (Figure 3). It regulates expression of the NMDAR itself, the scaffolding protein PSD95 (postsynaptic density 95) that links the NMDAR to nNOS (neuronal nitric oxide synthase) and the nAChR (nicotinic acetylcholine receptor), which regulates the release of glutamate from the presynaptic endings. A decrease in the activity of NRG1 may thus contribute to the remodelling of the GABAergic phenotype of the inhibitory interneurons by decreasing the activity of the NMDARs. Secondly, the NRG1/ErbB3 signalling pathway may exert a direct control by inhibiting the Src kinase that is a regulator of the NMDAR .
BDNF and schizophrenia
The expression of BDNF and its receptor TrkB (tropomyosin receptor kinase B) are reduced in schizophrenia and this may have a major impact on the function of the inhibitory interneurons particularly during early development. The signalling pathways induced by TrkBs control the expression of a number of the components that could account for the phenotypic alterations that disrupt the function of the inhibitory interneurons particularly during early development. TrkBs are typical PTKRs that are coupled to a number of signalling pathways (Figure 3). One of these is the MAPK signalling pathway that uses ERK1/2 (extracellular-signal-regulated kinase 1/2) to activate the transcription factor CREB, which is the same transcription factor that is activated by Ca2+ to control the expression of GAD67 as described above.
The TrkBs also activate the PI3K signalling pathway, which can control gene transcription by regulating the activity of GSK3β (glycogen synthase kinase 3β) to activate the transcription factor β-catenin and to inhibit CREB. This possibility is strengthened by the observation that the activity of GSK3β is also regulated by DISC1, which is encoded by one of the prominent genes mutated in schizophrenia . The mutated form of DISC1 is incapable of inhibiting GSK3β, resulting in a more pronounced inhibition of CREB that will contribute to a decline in the GABAergic phenotype and inhibitory interneuron function.
NMDAR hypofunction and allosteric modulators
The NMDAR channel consists of NR1 and NR2A subunits (Figure 3). The NR2A subunit has the glutamate-binding site, whereas the NR1 subunit responds to serine and glycine, which function as allosteric modulators to facilitate channel opening. A reduction in the level of serine has been reported in the blood and CSF (cerebrospinal fluid) of individuals with schizophrenia. Such reductions may be caused by mutations in the DAO (D-amino acid oxidase) that metabolizes D-serine and in the DAOA, which functions to enhance DAO activity .
NMDAR hypofunction and redox signalling in schizophrenia
Hypofunction of the NMDAR has been linked to a number of mechanisms, many of which are related to the way in which this receptor is modulated. The fast spiking inhibitory interneurons in the cortex seem to be particularly sensitive to the redox state of the brain. An increase in the oxidation state of these inhibitory neurons has been linked to the onset of schizophrenia [44,45]. Such dysregulation of the redox signalling pathway may provide an explanation for the developmental origins of schizophrenia because there appears to be a link between maternal viral infection during gestation and the incidence of this disease in the offspring. In rat models that reproduce this phenomenon, the developmental defects induced by such inflammatory responses are caused by activation of redox signalling.
During viral infections, there is an increase in IL-6 (interleukin 6), which has a prominent role in activating the redox signalling pathway . IL-6 acts through the JAK (Janus kinase)/STAT (signal transducer and activator of transcription) signalling pathway to increase the expression of Nox2, which is one of the NADPH oxidases responsible for generating superoxide (O2•−) (Figure 3). Both O2•− and hydrogen peroxide (H2O2), which is formed from O2•− by SOD (superoxide dismutase), function as second messengers during redox signalling. These ROS (O2•− and H2O2) can also react with nitric oxide (NO) to form peroxynitrite (ONOO−), which is very much more reactive than the two parent molecules. The NMDAR has an important role in generating NO because it is physically linked to the nNOS in the postsynaptic density through the scaffolding protein PSD95. nNOS is thus ideally positioned to respond to the pulses of Ca2+, which enter through the NMDAR, to generate NO that then interacts rapidly with O2•− to form ONOO−. The open probability of the NMDAR is reduced when the NR2A subunit is nitrosylated at Cys399 by ONOO−. This reduction in the ability of the NMDAR to generate a Ca2+ signal may have a direct bearing on the cognitive defects in schizophrenia. In addition, the reduction in Ca2+ signalling can explain the phenotypic remodelling of the GABAergic phenotype as described above.
The role of enhanced oxidation in driving these phenotypic alterations of the inhibitory interneurons is also consistent with studies on the changes in the antioxidant mechanisms that have been described in neurodegenerative disorders  including schizophrenia. The denitrosylation reaction, which functions to reverse the nitrosylation reaction that reduces the activity of the NMDAR, is carried out by two main antioxidants: GSH and Trx (thioredoxin) (Figure 3). In schizophrenia, gene polymorphisms have been described in the enzymes responsible for the synthesis of GSH such as the GCLC [GCL (glutamate cysteine ligase) catalytic subunit] and GCLM (GCL modifier subunit) that combine to form the GCL responsible for one of the steps of GSH synthesis . Such defects in GSH synthesis may explain the decreased GSH levels found in schizophrenia. The NMDAR plays an important role in regulating the redox state by reducing the expression of Txnip (thioredoxin-interacting protein), which binds to Trx and inhibits its participation in the denitrosylation reaction. Such a decline in the levels of GSH and Trx may thus contribute to the decline in NMDAR function that reduces the activity of the fast-spiking interneurons in schizophrenia.
Modulation of the tonic excitatory drive of GABAergic interneurons and schizophrenia
The neuronal oscillator that generates the γ rhythm depends on a tonic excitatory drive that functions to reduce the membrane potential of the participating neurons to a level that enables them to oscillate. This tonic drive seems to be located on both the inhibitory GABAergic and excitatory glutamatergic neurons. For the sake of argument, I have used the GABAergic neurons as an example to explore the potential significance of this pacemaker mechanism that mediates the action of the ascending arousal system that controls the awake/sleep cycle. This arousal system, which is inactive during sleep, is rapidly switched on during wakefulness to maintain the brain rhythms that characterizes consciousness. Not much is known about how the arousal systems functions to generate the tonic excitatory drive, but some potential signalling pathways can be inferred on the basis of the receptors known to mediate the modulatory function of transmitters such as ACh (acetylcholine), dopamine, VIP (vasoactive intestinal peptide) and oxytocin (Figure 3). These metabotropic actions seem to be mediated through either the cAMP or InsP3/Ca2+ signalling pathways.
Dopamine acts through the D2 receptor to inhibit the cAMP signalling pathway that may alter the input resistance by acting through the HCN1 (type 1 hyperpolarizing-activated cyclic nucleotide-gated) channels that provides an inward Na+ current (Figure 3). Since the activity of HCN1 is regulated by cAMP, it could explain how alterations in the activity of the dopamine D2 receptor and VIPR2 may contribute to schizophrenia. Schizophrenic-like symptoms occur in response to drugs such as the amphetamines that release dopamine, whereas drugs that inhibit the D2 receptor are antipsychotic. These pharmacological actions would thus predict that schizophrenic symptoms may arise through excessive activation of the D2 receptor, resulting in a reduction in the level of cAMP and a decrease in the tonic drive. Conversely, duplications of VIPR2, which confers significant risk of schizophrenia , would have the opposite effect, since the action of VIP on VIPR2 enhances the cAMP signalling pathway to increase the tonic drive. It would seem that changes in the cAMP signalling pathway can either enhance or reduce the tonic excitatory drive and this would have repercussions on the generation of the γ rhythms resulting in schizophrenia. Such a possibility is consistent with the fact that alterations in the activity of PDE4B (phosphodiesterase 4B), which hydrolyses cAMP (Figure 3), have been identified in schizophrenia. SNPs (single nucleotide polymorphisms) associated with the gene that codes for PDE4B have been described in schizophrenia. In addition, DISC1, which is mutated in schizophrenia, has also been shown to interact with PDE4B to alter the metabolism of cAMP.
The cholinergic system, which projects diffusely throughout the cortex, controls processes such as sensory perception, cognition and consciousness by activating this tonic excitatory drive . ACh functions as an arousal-promoting neuromodulator that can switch the brain from a sleep to an awake mode of firing that is characterized by the appearance of γ rhythms. This ability of ACh to enhance neuronal excitability is not fully understood, but seems to depend on two separate mechanisms. First, ACh can facilitate the release of glutamate at presynaptic endings by acting through ionotropic nAChRs . In patients with schizophrenia, the expression of nAChRs, which is regulated by the NRG1/ErbB pathway described above, is reduced and this could contribute to hypofunction of glutamatergic signalling in the interneurons. Secondly, ACh also acts through muscarinic M1 receptors that act by increasing the formation of InsP3 to release Ca2+ from the internal stores (Figure 3). The InsP3/Ca2+ signalling pathway is also induced by activation of mGluR1s (type 1 metabotropic glutamatergic receptors). The increase in Ca2+ may then act to enhance membrane depolarization by stimulating an inward Ca2+ current (ICat) . A decrease in oxytocin, which acts through the InsP3/Ca2+ signalling pathway, has also been implicated in schizophrenia and other mental disorders such as autism.
The Ca2+ signals generated by these different metabotropic receptors are influenced by various intracellular feedback mechanisms. For example, NMDAR hypofunction will result in an increase in the sensitivity of the InsP3R as a result of the reduction in the CREB-dependent expression of Bcl-2, which normally acts to suppress InsP3R activity. In addition, Ca2+ can activate NCS-1 that desensitizes the dopamine D2 receptor, thus increasing cAMP levels to enhance membrane excitability . Such a possibility is of interest because NCS-1 levels are increased in the prefrontal cortex in both schizophrenia and BD . NCS-1 may also act by enhancing Ca2+ release by the InsP3Rs  and this would also increase membrane excitability.
A puzzling aspect of the actions of these metabotropic receptors is that they seem to have contradictory effects on the tonic excitatory drive. However, this might be reconciled by the fact that γ rhythms may be increased or reduced in different schizophrenia syndromes . It is clear that much more needs to be learnt about the signalling mechanisms responsible for regulating this excitatory drive in order to provide further insights into the nature of consciousness and how dysregulation of this pathway may contribute to schizophrenia, and this will also be relevant to BD as described below.
Over 2 million people in the U.K. are diagnosed as suffering from depression each year. Almost half of these have the more severe disorder of manic–depressive illness, also known as BD, which is characterized by extreme mood swings between mania and depression. The underlying causes of BD are still somewhat of a mystery. Early attempts to understand BD focused on possible defects in neurotransmitters as a cause of the neuronal changes responsible for these profound alterations in behaviour. The changes in mood were thought to be caused by alterations in the activity of various neurotransmitters such as 5-HT (5-hydroxytryptamine) (serotonin), noradrenaline (norepinephrine), dopamine and ACh. The properties of the receptors that respond to these neurotransmitters drew attention to the possibility that mood might be controlled by modulating neural activity through downstream signalling pathways. In the case of ACh, inhibition of acetylcholinesterase that hydrolyses ACh causes depression, whereas inhibition of muscarinic M1 receptors by scopolamine has the opposite effect by inducing symptoms of mania. Such M1 receptors stimulate phosphoinositide hydrolysis, which induces the InsP3/Ca2+ and DAG (diacylglycerol)/PKC (protein kinase C) signalling pathways that may play a role in BD by enhancing the tonic excitatory drive responsible for maintaining brain rhythms (Figure 4). Dopamine, which acts through the cAMP signalling pathway, also has a marked effect on mood in that drugs such as haloperidol that decrease dopaminergic transmission are antimanic. It seems that low cAMP levels promote mania, whereas elevated levels induce depression. Despite uncertainties as to exactly how these signalling pathways operate, there is sufficient evidence to suggest that BD may arise through remodelling of some of the key intracellular signalling pathways that function to modulate neuronal excitability.
Intracellular signalling pathways and BD
One of the problems with trying to identify the nature of the primary signalling defects in BD is the long latency between drug administration and the change in behaviour. Even though many mood-stabilizing drugs reach their targets very quickly, it can take 2–3 weeks before a change in behaviour occurs. This long lag probably depends on the time it takes to remodel the aberrant neural signalling pathways to restore the neural circuitry responsible for regulating mood. There is considerable evidence to link the dysfunctional signalling mechanisms to a decline in the process of neurogenesis responsible for the differentiation and survival of new neurons necessary to maintain the neural circuitry in the adult brain . This neurogenesis hypothesis considers that BD is caused by deterioration in the neuronal circuitry in specific regions of the brain caused by a loss of neurons. For example, the volume of the hippocampus in patients suffering from this disease is much reduced. It is known that the onset of this disease is often triggered by a period of stress that is also known to reduce the volume of the hippocampus. Some of the symptoms of the disease may arise from a decrease in hippocampal circuitry because this region provides inputs to the prefrontal cortex, cingulate cortex and amygdala, which contribute to the altered mood and emotions of depression. A defective hippocampus is also consistent with the fact that depressed patients display cognitive defects. Much interest is now focused on the idea that BD may be caused by a decline in the action of neurotrophins, such as BDNF, which regulate this process of neurogenesis .
The relationship between cell signalling and neurogenesis is beginning to emerge from studies on the mode of action of mood-stabilizing drugs, such as lithium (Li+) and valproate, that act by modulating downstream intracellular signalling pathways (Figure 4). For example, Li+ is a potent inhibitor of enzymes such as GSK3β and IMPase (inositol monophosphatase) that function in different but related signalling pathways. The effectiveness of Li+ in controlling BD may thus depend on its ability to inhibit a number of signalling pathways that are intimately related to each other through a number of feedback mechanisms and they also play a role in regulating both neurogenesis and the tonic excitatory drive.
GSK3β inhibition and BD
One of the primary intracellular targets of Li+ is GSK3β, which regulates the activity of transcription factors such as CREB and β-catenin responsible for the expression of signalling components such as BDNF and Bcl-2, which control neurogenesis and survival. Restoration of normal hippocampal volumes by Li+ depends on its ability to increase neurogenesis by inhibiting GSK3β, which is a potent inhibitor of CREB . It could therefore be argued that one of the causes of BD may be an overactive GSK3β that suppresses the expression of both BDNF and Bcl-2. Patients with depression have low serum levels of BDNF. Bcl-2 gene SNPs, which resulted in elevated basal Ca2+ levels and enhanced InsP3-mediated cytosolic Ca2+ release, have been associated with a risk of developing BD . The low level of Bcl-2 will have two important consequences (Figure 4). First, a decrease in the level of this anti-apoptotic factor will enhance apoptosis. Secondly, a decrease in Bcl-2 will reduce its inhibitory effect on InsP3-induced Ca2+ release that will contribute to the increase in both resting and activated levels of Ca2+, which are a characteristic feature of BD . These elevated levels of Ca2+ will enhance apoptosis, and the decrease in neuronal survival will then contribute to the decline in the neurogenesis necessary to maintain the neural circuitry responsible for regulating behaviour. An important aspect of the neurogenesis hypothesis is that it can explain the long time lag for antidepressants to work. Because the process of neurogenesis is relatively slow, the recovery of hippocampal volume following antidepressant treatment takes several weeks to occur, exactly in line with the time it takes to see any behavioural improvements.
Inositol-depletion hypothesis of BD
Another important Li+ target is the IMPase that hydrolyses inositol monophosphates (Ins1P, Ins3P and Ins4P) to free inositol. By inhibiting the formation of inositol, Li+ chokes off the supply of the free inositol required to resynthesize the PtdIns necessary to provide the PtdIns(4,5)P2 required for the phosphoinositide signalling pathway  (Figure 4). This inositol-depletion hypothesis is consistent with the action of valproate, which is another potent mood-stabilizing drug, that also has the potential to deplete internal inositol by inhibiting the inositol synthase responsible for the de novo synthesis of inositol from glucose 6-phosphate (Figure 4). The central feature of the inositol-depletion hypothesis is that Li+ and valproate act to inhibit the supply of inositol required to maintain the inositol lipid signalling pathway. Cells in the periphery have access to dietary inositol that circulates in the plasma and is taken up by SMIT1 (sodium-dependent myo-inositol co-transporter-1). Neurons closeted behind the blood–brain barrier, which is relatively impermeable to inositol, have reduced access to this dietary inositol. Since they rely on inositol obtained through recycling and de novo synthesis, neurons are uniquely sensitive to Li+ and valproate. Conversely, cells in the periphery are protected against inositol depletion induced by these two drugs by obtaining inositol from the plasma.
The inhibition of IMPase by Li+ occurs through an uncompetitive mechanism that has an unusual consequence with regard to its therapeutic action in that Li+ will have little action when signalling pathways are operating normally, but will become increasingly efficacious the more abnormal signalling becomes. In this way, Li+ is a perfect homoeostatic drug in that it has no effect if the signalling system is operating normally, but begins to exert its therapeutic action only when signalling becomes excessive. In effect, the efficacy of Li+ is tailored to the severity of the disease state.
A corollary of the inositol-depletion hypothesis is that BD may depend on a remodelling of the phosphoinositide signalling pathway resulting in an increase in both the InsP3/Ca2+ and DAG/PKC information transfer pathways that can have a number of consequences. A significant risk of BD is conferred by SNPs in the DGKH (diacylglycerol kinase) gene that metabolizes DAG  (Figure 4). An increase in the activity of the DAG/PKC pathway may influence mood through its established action on membrane excitability. The accompanying increase in Ca2+ signalling is consistent with a large number of observations indicating that the resting and activated levels of Ca2+ are elevated in BD . NCS-1, which is known to be elevated in the prefrontal cortex in both schizophrenia and BD , is known to enhance the activity of the InsP3Rs  and this would contribute to an increase in the intracellular level of Ca2+. Such a conclusion is also in line with the observation that depression occurs when the activity of M1 receptors is enhanced following treatment with organophosphates (Figure 4). As described above, such elevations of Ca2+ can activate apoptosis, which will contribute to a decrease in neurogensis. Such an inhibitory effect on neurogenesis may be facilitated by Ca2+ stimulating Pyk2 (proline-rich tyrosine kinase 2) to phosphorylate GSK3β, thus enhancing its inhibitory action on CREB . The membrane excitability of the neural circuits that control mood may also be distorted by abnormal elevations in Ca2+ through two mechanisms. First, it can have marked effects on the tonic excitatory drive that controls the membrane excitability of the neural circuits, resulting in changes in mood. Secondly, Ca2+ can act through NCS-1 to desensitize the dopamine D2 receptor  and this will also have an effect on membrane excitability through the mechanisms described above (Figure 4).
One of the puzzling aspects of the mode of action of Li+ is how it can act on both mania and depression. A possible solution to this conundrum is that these two extremes of the bipolar spectrum may depend on the nature or location of the neurons that have excessive tonic drive resulting from the enhanced InsP3/Ca2+ and DAG/PKC information transfer pathways. Mania may result from excessive activation of the excitatory neurons, whereas depression may arise from hyperstimulation of the inhibitory neurons. Through its unique uncompetitive mode of action, Li+ seeks out and dampens down this hyperactivity responsible for either mania or depression.
Each cell type has a unique signalling phenotype capable of delivering Ca2+ signals with the spatial and temporal properties necessary to regulate its particular function. Phenotypic stability seems to be maintained by Ca2+-dependent feedback mechanisms that adjust the expression levels of individual toolkit components that contribute to these cell-specific signalling systems. Signalling through InsP3/Ca2+ is an integral part of the control mechanisms in many different cell types. In non-excitable cells, it often plays a direct role in cell activation. In excitable cells, fast-acting voltage-sensitive Ca2+ entry mechanism provide the primary Ca2+ signal for cell activation, whereas the InsP3/Ca2+ pathway has a more modulatory role to adjust the intensity of the primary Ca2+ signal. Many of the major diseases in humans seem to result from subtle alterations in the function of this ubiquitous InsP3/Ca2+ signalling pathway to modulate the activity of excitable cells such as cardiac cells (heart disease) and neurons (AD, BD and schizophrenia).
Biochemical Society Award Delivered at the Biochemical Society Centenary Event held at the Royal Society, London, on 16 December 2011 Sir Michael Berridge
brain-derived neurotrophic factor
Ca2+/calmodulin-dependent protein kinase
congestive heart failure
D-amino acid oxidase
D-amino acid oxidase activator
disrupted in schizophrenia 1
Down's syndrome critical region 1
dual-specificity tyrosine-phosphorylated and regulated kinase 1A
glutamic acid decarboxylase 67
glutamate cysteine ligase
glycogen synthase kinase 3β
type 1 hyperpolarizing-activated cyclic nucleotide-gated
mitogen-activated protein kinase
nicotinic acetylcholine receptor
neuronal calcium sensor 1
nuclear factor of activated T-cells
neuronal nitric oxide synthase
protein kinase C
postsynaptic density 95
protein tyrosine kinase-linked receptor
reactive oxygen species
sarcoplasmic/endoplasmic reticulum Ca2+-ATPase
single nucleotide polymorphism
tropomyosin receptor kinase B
transient receptor potential
vasoactive intestinal peptide
VIP receptor 2