Roles for the Endoplasmic Reticulum in Regulation of Neuronal Calcium Homeostasis

Abstract

By influencing Ca²⁺ homeostasis in spatially and architecturally distinct neuronal compartments, the endoplasmic reticulum (ER) illustrates the notion that form and function are intimately related. The contribution of ER to neuronal Ca²⁺ homeostasis is attributed to the organelle being the largest reservoir of intracellular Ca²⁺ and having a high density of Ca²⁺ channels and transporters. As such, ER Ca²⁺ has incontrovertible roles in the regulation of axodendritic growth and morphology, synaptic vesicle release, and neural activity dependent gene expression, synaptic plasticity, and mitochondrial bioenergetics. Not surprisingly, many neurological diseases arise from ER Ca²⁺ dyshomeostasis, either directly due to alterations in ER resident proteins, or indirectly via processes that are coupled to the regulators of ER Ca²⁺ dynamics. In this review, we describe the mechanisms involved in the establishment of ER Ca²⁺ homeostasis in neurons. We elaborate upon how changes in the spatiotemporal dynamics of Ca²⁺ exchange between the ER and other organelles sculpt neuronal function and provide examples that demonstrate the involvement of ER Ca²⁺ dyshomeostasis in a range of neurological and neurodegenerative diseases.

1. Overview

For well over a century, perhaps starting with the remarkable drawings of Ramón y Cajal [1], neurons have been appreciated for their exquisite structural intricacy. Mirroring the complexity of neuronal morphology, architecture of their organelles varies significantly from one subcellular location to the next. For instance, the endoplasmic reticulum (ER), whose surface area vastly exceeds that of the plasma membrane [2], is comprised of tightly-packed interconnected sheets in cell bodies and dendrites, whereas the same organelle in axons takes on a characteristically tubular shape [3,4,5,6,7]. Recent studies have suggested that the morphological features of the ER are not ‘accidental’ consequences of neuronal arborization. Rather, the observed structural heterogeneity is inherent to the functional diversity and pleiotropy of these organelles in different neuronal compartments. In the following sections, we describe our current understanding of the mechanisms involved in the regulation of ER Ca²⁺ homeostasis through the lens of neuronal architecture and elaborate upon how changes in ER Ca²⁺ dynamics across space and time sculpt various aspects of neuronal function.

2. Neuronal Polarity Imposes Structural Constraints on the ER

Most neurons exhibit stereotypical compartmentalization that arises from partitioning into functional modules called dendrites, cell bodies, and axons. Whereas cell bodies house the nucleus and serve as de facto guides of overall cellular function, dendrites and axons are responsible for the transmission of electrical and chemical signals across neuronal networks. Dendrites, the signal receiving compartments, are comprised of tiny protrusions called spines, each of which contains a specialized ER structure called the spine apparatus [5,6]. The spine apparatus has roles in the regulation of spine morphogenesis, Ca²⁺ dynamics, and synaptic plasticity [5,8,9]. Structurally, the spine apparatus is made up of interconnected stacks of ER, and typically occupies <5% of the spine volume [10]. Despite the relatively small fraction of the total spine volume, the complexity of ER structure, in terms of branching and local arborization, increases dramatically as the ER traverses into spines from the ‘bulk’ dendrite [11]. This increase in structural complexity serves to limit the movement of proteins embedded in the ER, thus facilitating the translocation of proteins from the spine apparatus to local plasma membrane [11,12]. These findings underscore the validity of the notion that the structural heterogeneity of the ER is not simply a passive outcome of neuronal arborization, but rather represents an example of ‘function arising from form’.

The morphology of ER in neuronal cell bodies is similar to that observed in spines, and resembles stacked sheets connected by helicoidal structures interspersed between the sheets [4]. ER sheets in neuronal soma and dendrites exhibit continuity and dynamic exchange of membranes, which is necessary for dendrites to reliably maintain Ca²⁺ homeostasis for extended durations [13,14]. The stacked-sheet architecture also permits an exceptionally high density of ER membrane in cell bodies, which provides an expansive surface for ribosome attachment [4]. Attachment of ribosomes to ER sheets is necessary for protein translation, and morphologically distinguishes rough ER (RER) from smooth ER (SER). Somatodendritic compartments contain both SER and RER, which exhibit dynamic transitions between each other [15].

In contrast to the sheet-like structures in cell bodies and dendrites, axonal ER is comprised of thin (~20–30 nm in diameter), elongated, and smooth tubules [3,16]. RER constitutes a very small fraction of axonal ER [17]. The relative paucity of RER in axons suggests a diminished contribution of ER-associated ribosomes in axonal protein translation. ER tubules are sculpted from sheets by the complex interplay between many proteins including atlastin, the reticulon complex, and members of the receptor expression enhancing proteins (REEPs) [7,18,19,20,21]. Once tubulated, growth of nascent ER along microtubules allows the organelle to populate the entire length of the axon [22]. The axonal ER and microtubules exhibit multiple levels of crosstalk, and function together to mediate axonal elongation and neuronal polarity [20,22,23]. In addition to microtubules, neurofilaments and atypical myosins (e.g., myosin Va) also regulate the shape and distribution of axonal ER, which in return guides axonal growth [24,25].

3. ER-Resident Ca²⁺ Channels Are Required for the Regulation of Axodendritic Growth and Morphology

As mentioned at the end of the previous section, the relationship between the structures of neurons and the underlying ER is inherently bidirectional. While neuronal architecture imposes specific constraints on ER morphology, ER Ca²⁺ release and the channels involved therein influence the structure of neurons. Analyses of invertebrate synapses have revealed several aspects of the interplay between Ca²⁺ and neuronal structure and function. Since the sites of neurotransmitter release are located within clearly identifiable presynaptic boutons at Drosophila neuromuscular junctions (NMJ) [26,27,28], determinants of bouton morphology and number at these synapses have been matters of intense study. The stereotypical arborization of axon termini at Drosophila NMJs is strongly influenced by the stability of presynaptic microtubules, which depends on the microtubule binding protein 1b (MAP1b) homolog, Futsch (Figure 1) [29,30,31]. Futsch–microtubule association promotes microtubule stability, whereas diminished interaction with Futsch results in microtubule depolymerization and stereotypical alterations in NMJ morphology—larger but fewer boutons [23,29,30,31]. As is the case for many microtubule binding proteins, Futsch–microtubule association is inversely correlated with Futsch phosphorylation [32]. Ca²⁺ release via the ER-resident transient receptor potential vanilloid (TRPV) channel, Inactive (Iav) [33,34], leads to activation of the Ca²⁺-dependent phosphatase, calcineurin (Figure 1) [23]. Once activated, calcineurin mediates Futsch dephosphorylation, increased microtubule stability, and normal bouton morphology [23]. Decreased abundance of Iav and Drosophila homologs of the ER channels, inositol triphosphate receptors (IP₃Rs), and ryanodine receptors (RyRs), results in diminished calcineurin activity, loss of presynaptic microtubules, and attendant alterations in bouton number and morphology [23]. The observation that activities of Iav, IP₃R, and RyR are required for regulation of presynaptic morphology points to functional interactions between these channels in the regulation of ER Ca²⁺ release and synapse development.

ER Ca²⁺ and microtubules are also involved in axon guidance [23]. Ca²⁺ release via RyRs, and the subsequent activation of store-operated channels mediate the effects of netrin-1 on growth cone guidance [35,36,37,38]. IP₃Rs also influence netrin-1-dependent turning by regulating the rate of microtubule invasion into growth cones [37]. Irrespective of the mechanism of store depletion, the luminal [Ca²⁺] sensor, stromal interacting molecule 1 (STIM1), couples store release with microtubule polymerization via interactions with the microtubule plus-end binding protein, EB1 [36,39,40,41,42]. Furthermore, membranes necessary for neurite extension are derived from ER Ca²⁺-dependent asymmetric secretion of vesicles towards the growing end of axons [36,39,40,41,42]. The movement and fusion of these vesicles with the axonal membrane depend on myosin Va, and are facilitated by activation of IP₃Rs or RyRs [24,25]. Myosin Va is also necessary for maintaining the appropriate distribution of ER in axonal periphery, without which proper axonal growth would be impeded [24,25].

STIM and ER [Ca²⁺] also regulate the growth and morphology of dendritic spines. Neuronal activity-dependent modification of the spine structure during long-term potentiation involves ER Ca²⁺ store depletion, STIM1-mediated induction of store-operated Ca²⁺ entry, and gene expression driven by the transcription factor NFATc3 [43,44]. In hippocampal neurons, directional movement of microtubules into spines, an intermediary in the regulation of spine morphology by ER Ca²⁺ release, depends on interactions between STIM2 and the microtubule plus-end binding protein EB3 [45]. As is the case in axons, release of ER Ca²⁺ within synaptic spines is predicated on the presence of ER. Myosin Va-dependent movement of ER along actin fibers drags the organelle into growing spines [46,47], thus facilitating spine development via pathways involving ER Ca²⁺ release. Together, these findings suggest that actomyosins and microtubules influence temporally distinguishable events that sculpt neuronal morphology. Myosin Va appears to set the stage by ensuring the availability of ER in the relevant compartments such that subsequent release of ER Ca²⁺ facilitates local remodeling in a microtubule-dependent manner.

4. ER Ca²⁺ and the Regulation of Neurotransmission

Axons contain the domains responsible for the release of neurotransmitter-laden synaptic vesicles (SVs). The fundamentals of SV release, which are reasonably well-understood, can be succinctly summarized as a sequence of events starting with the arrival of action potentials to presynaptic termini, opening of voltage-gated Ca²⁺ channels (VGCCs), Ca²⁺ influx, and Ca²⁺/SNARE-dependent SV exocytosis (Figure 2A) [48,49]. According to this model, instructive signals that elevate presynaptic [Ca²⁺] to a range necessary for SV release are plasma membrane delimited, i.e., channels responsible for both action potential propagation and Ca²⁺ entry are located on the cell surface. Nevertheless, many studies have demonstrated the involvement of ER Ca²⁺ channels in sculpting the temporal dynamics and extent of SV release. The probability of SV release, defined as the likelihood that a primed SV will be released upon arrival of the action potential, is influenced by presynaptic resting [Ca²⁺] (Figure 2B) [23,49,50,51,52]. Given the 4^th power relationship between [Ca²⁺] and quantal content (referred to as the nonlinear relationship between Ca²⁺ and vesicle release rates [53]), small changes in resting [Ca²⁺] exert considerable influence on neurotransmitter release (Figure 2B, inset) [23,49,50,51]. At the calyx of Held, for instance, even a 50 nM increase in resting [Ca²⁺] is sufficient to elicit an appreciable increase in SV release probability [51]. Many different processes, including Ca²⁺ entry and extrusion mechanisms, cooperate to determine presynaptic resting [Ca²⁺] [50]. Roles for ER Ca²⁺ and ER-resident channels in crafting resting [Ca²⁺] and SV release probability have been widely acknowledged (Figure 2A) [23,54,55,56,57,58,59,60,61,62]. At the hippocampal Schaffer-collateral pathway, deletion of presenilin in presynaptic CA3 neurons perturbs short-term plasticity via hyperactivation of RyRs [54,59]. Elevations in RyR-mediated ER Ca²⁺ release promote both spontaneous SV release and the probability of evoked events [55,63,64]. Furthermore, either the induction of passive leak of ER Ca²⁺ following inhibition of the Ca²⁺ ATPase, SERCA, or activation of IP₃R mediated ER Ca²⁺ release are sufficient for presynaptic facilitation [54,65,66]. These data point to a role for ER Ca²⁺ release per se rather than an obligatory role for any specific channel type. Needless to say, the influence of ER Ca²⁺ release on overall network dynamics depends on the identity of the neurons experiencing the Ca²⁺ release events, as exemplified by the findings that increased RyR activity or inhibition of SERCA elevates the probability of GABA release in Purkinje neurons, which serves to dampen network activity in the cerebellum [56,57].

Although several studies have shown that Ca²⁺-induced Ca²⁺ release (CICR, mediated by either RyR or IP₃R) is necessary for ER stores to adequately influence presynaptic plasticity and SV release, the notion of CICR influencing SV release has incompletely understood nuances [54,55,61,62]. For instance, other studies argue in favor of a model that residual Ca²⁺ from VGCC opening, and not CICR, is the major regulator of resting [Ca²⁺] and presynaptic plasticity in cerebellar and hippocampal slices [67]. A potential explanation for these conflicting viewpoints could be gleaned from studies focused on the role of ER Ca²⁺ release in presynaptic termini of Drosophila NMJs. Loss-of-function mutations in the gene encoding the TRPV channel, Iav, result in lower presynaptic resting [Ca²⁺] and decreased SV release probability [23]. Interestingly, these phenotypes were only evident when the extracellular [Ca²⁺] was lowered to <0.5 mM from the usual 2 mM [23]. These phenotypes, which have been recapitulated in other mutants with diminished ER Ca²⁺ release at the Drosophila larval NMJ [20,68], show that the contribution of ER channels on resting [Ca²⁺] can be overridden by Ca²⁺ influx. We envision that only during periods of diminished presynaptic Ca²⁺ entry would the contribution of ER Ca²⁺ stores to SV release become apparent. For example, during high frequency stimulation when massive Ca²⁺ influx into postsynaptic neurons results in dramatic reduction of [Ca²⁺] in the synaptic cleft [69,70,71,72], ER Ca²⁺ release could have a significantly larger impact on determining resting [Ca²⁺], and thus SV exocytosis.

5. ER Ca²⁺ Release Regulates Neuronal Gene Expression during Development, Synaptic Plasticity, and Cell Death

While axons and dendrites participate in intercellular communication, cell bodies contain the nuclei, and thus serve as the seat of neuronal gene expression. Ca²⁺ is a key intermediate in the functional relationship between neuronal activity and gene expression. Alterations in cytosolic and nuclear [Ca²⁺] exert differential effects on neuronal gene expression, the dynamics of which are synchronized with the spatiotemporal patterns of the Ca²⁺ transients. Depletion of ER Ca²⁺ stores by application of the SERCA inhibitor, thapsigargin, in cortical, hippocampal, and other neurons leads to the dramatic and rapid increase in the transcription of the immediate early genes encoding c-Fos and Homer-1a [73,74]. Stimulation of ER Ca²⁺ release in hypothalamic neurons following application of serotonin also results in c-Fos induction via a process requiring ER Ca²⁺ release through IP₃Rs [75,76]. The relationship between c-Fos induction and ER Ca²⁺ release reinforces the role for the latter in synaptic plasticity. Furthermore, CREB-dependent gene expression, which is also critically important for synaptic plasticity, is influenced by different sources of neuronal Ca²⁺ elevation [77,78,79]. Indeed, c-Fos expression is itself dependent on CREB, potentially via IP₃R-dependent nuclear translocation of a CREB coactivator called TORC1 (not to be confused with the kinase complex containing mTOR) [76,80]. IP₃ production and ER Ca²⁺ release via IP₃Rs during synaptic plasticity is often triggered by stimulation of metabotropic glutamate receptors [81]. It is notable that in the context of synaptic plasticity, elevation of nuclear [Ca²⁺] arising from the passive diffusion from the cytosol through the nuclear pore or due to activation of IP₃Rs embedded in the nuclear envelope promotes CREB-dependent transcription and the changes in synaptic strength and dendritic arborization that characterize plasticity [82,83,84,85,86,87,88,89].

The canonical mode of CREB activation in response to cytosolic or nuclear Ca²⁺ elevation involves CREB phosphorylation by Ca²⁺/calmodulin-responsive enzymes, CaMKII and CaMKIV (Figure 3) [79,90,91]. The elevation of [Ca²⁺] that precedes CaMK- and CREB-dependent synaptic plasticity is triggered by synaptic transmission occurring at dendritic spines, which begs the question: how are signals transmitted from dendrites to the nucleus? One explanation for this long-range spatial communication is that dendritic Ca²⁺ elevations result in CICR (RyR-dependent) and the propagation of ER Ca²⁺ waves (likely IP₃R-dependent), ultimately resulting the elevation of nuclear Ca²⁺ and CREB-dependent gene expression (Figure 3) [83,92].

In T lymphocytes, NFAT-dependent gene expression is predicated upon the movement of the transcription factor from the cytosol to nucleus [93,94,95,96]. As phosphorylation inhibits nucleocytoplasmic transport of NFAT, calcineurin activated by ER Ca²⁺ release and store-operated Ca²⁺ entry is needed for NFAT dephosphorylation and attendant gene expression [93,94,95,96]. Similarly, cytosolic Ca²⁺ elevations in neurons lead to NFAT-dependent gene transcription in a calcineurin-dependent manner [97,98]. In addition to Ca²⁺ entry accompanying neuronal activity, IP₃R-mediated ER Ca²⁺ release downstream of BDNF application or alterations in mitochondrial Ca²⁺ uptake drive nuclear translocation of NFAT in neurons [97,98,99]. Whereas calcineurin/NFAT participate in neurodevelopment and synaptic plasticity in some contexts [43,100,101], activation of this transcriptional axis by elevated Ca²⁺ release in mature neurons usually promotes cell death [102,103]. Furthermore, alterations in ER Ca²⁺ homeostasis is also thought to be a determinant of gene expression changes observed in aged neurons, and could involve calcineurin and NFAT [104]. Interestingly, expression of the genes encoding IP₃R1 and IP₃R2 are under the control of calcineurin and NFAT [105,106,107,108], which prompts us to speculate that elevated channel activity could indirectly promote its abundance, and thus compromise neuronal viability during aging and disease.

6. Relationship of ER Ca²⁺ Release with Neuronal Bioenergetics and Excitability

The healthy brain is a veritable powerhouse, a fact that is best illustrated by the involvement of this organ in the production and consumption of ~20% of an individual’s energy despite contributing only ~2% to the total body mass [109]. Within the brain, most of the bioenergetic activity occurs in neurons, which utilize ATP generated from mitochondrial oxidative phosphorylation (OXPHOS) for a variety of key functions. The reducing equivalents that power OXPHOS are produced by the tricarboxylic acid (TCA) cycle in the mitochondrial matrix [110]. While it is widely acknowledged that the TCA cycle requires metabolites such as pyruvate and α-ketoglutarate [111,112,113], a less appreciated fact is that >20 μM [Ca²⁺] is needed within the matrix for activation of the TCA dehydrogenases—pyruvate dehydrogenase, α-ketoglutarate dehydrogenase, and isocitrate dehydrogenase (Figure 4) [113]. This requirement necessitates an acute elevation of matrix [Ca²⁺] from the resting levels of ~500 nM, which is made challenging by the relatively low Ca²⁺ affinity (20–30 μM) of the mitochondrial Ca²⁺ uniporter (MCU) [114]. Within the presynaptic compartment, routine machinations of neurotransmission elevate cytosolic [Ca²⁺] to levels high enough for mitochondrial uptake and the attendant activation of OXPHOS [114]. At these sites, locally produced ATP is immediately made available to plasma membrane Ca²⁺ ATPases (PMCA), which use this source of energy to actively extrude intrasynaptic Ca²⁺ (Figure 5) [115,116]. The situation is substantially more complex in cell bodies where the threat of Ca²⁺ cytotoxicity precludes sustained [Ca²⁺] elevation in the bulk cytosol. To overcome these restrictive conditions, a fraction of ER is physically associated with mitochondria allowing the two organelles to form specialized structures called mitochondrial associated membranes (MAMs) [117,118]. Amongst their myriad functions, MAMs are characterized by the presence of IP₃Rs, and thus serve as conduits for interorganellar Ca²⁺ transfer [117,119,120,121,122,123,124]. ER Ca²⁺ release by IP₃Rs in MAMs enables sufficient elevation of perimitochondrial [Ca²⁺] and MCU-mediated Ca²⁺ uptake [119]. IP₃Rs are maintained in MAMs by physical tethering to the outer mitochondrial membrane-resident protein, porin1, with GRP75 serving as an intermediary link (Figure 4) [119]. These interactions allow IP₃Rs to sculpt both cytosolic and mitochondrial [Ca²⁺], with the latter playing a major role in the regulation of cellular bioenergetics. It is important to note that despite the definitive involvement of MAMs in interorganellar Ca²⁺ transfer, cytosolic [Ca²⁺] elevations that exceed the MCU threshold would increase mitochondrial [Ca²⁺] in a manner that is agnostic to the source of Ca²⁺. For instance, cytosolic [Ca²⁺] elevations subsequent to RyR activation also result in mitochondrial Ca²⁺ uptake, albeit in a MAM-independent manner [125,126]. Concordantly, pharmacological inhibition of RyRs are beneficial in neurological diseases associated with Ca²⁺-dependent mitochondrial dysfunction [127].

The extent of energy production in neurons is intimately related to neuronal excitability. Almost all the processes involved in the regulation of neuronal membrane potential, excitability, and propagation of action potentials exhibit steep dependences on [ATP]. Thus, the interplay between IP₃R activation and mitochondrial ATP production ensures a critical requirement for IP₃Rs in the dynamic regulation of neuronal membrane potential and excitability. This notion is exemplified by the Na⁺/K⁺ ATPase, which is needed for the maintenance of the electrical polarity across the neuronal membrane, and for this purpose the pump consumes ~75%–90% of the ATP produced in the cell [128]. Computational studies modeling the relationships between OXPHOS and neuronal excitability point to the involvement of the Na⁺/K⁺ ATPase in the susceptibility of neurons to ATP deprivation [129]. In this schema, even a 15% deficit in [ATP] is expected to result in fulminant loss of neuronal membrane potential due to the progressive decline in Na⁺/K⁺ ATPase activity [129]. Thus, it stands to reason that alterations in ER Ca²⁺ release are reflected in the extent of ATP production and membrane potential. In addition, dynamic changes in neuronal excitability are also subject to regulation by ER Ca²⁺ channels. For example, in cortical neurons previously treated with caged IP₃, action potentials were consistently stronger [130]. Along these lines, the Ca²⁺ store-responsive protein, STIM1, has been shown to play a role in ATP production and the resulting regulation of Ca²⁺ extrusion and the intrinsic firing properties of Purkinje neurons [131]. On the other hand, Ca²⁺ release via either RyRs or IP₃Rs can activate Ca²⁺-responsive K⁺ channels (e.g., BK channels), which hyperpolarize neurons and repress excitability [132]. Together, these findings implicate several interlocking functions of ER Ca²⁺ channels in the regulation of neuronal excitability.

7. Mechanisms Involved in the Activation of ER Ca²⁺ Release in Neurons

7.1. Role for CICR and Ca²⁺ Pumps in ER Ca²⁺ Homeostasis

Since IP₃R-dependent cytosolic Ca²⁺ elevation is usually sufficient for stimulating RyRs, IP₃Rs and RyRs are thought to cooperate in the initiation and propagation of CICR [133,134,135,136]. Both types of channels exhibit biphasic dependence on cytosolic [Ca²⁺] owing to the presence of high Ca²⁺-affinity activating sites and low Ca²⁺-affinity inhibitory sites on the proteins [137,138,139,140]. When cytosolic [Ca²⁺] is below a critical threshold, Ca²⁺ ions activate the channels leading to store release. Upon exceeding the threshold, Ca²⁺ ions start to inhibit the channels, which lead to attenuation of Ca²⁺ release and restoration of cytosolic [Ca²⁺] to resting levels. In addition to IP₃Rs and RyRs, other purported triggers for CICR include Ca²⁺ influx via TRP channels and VGCCs, and alterations in mitochondrial Ca²⁺ uptake [141,142,143,144]. Given the inherently self-sustaining nature of CICR, functional crosstalk between RyRs and other modulators of cellular Ca²⁺ homeostasis permits the amplification of even small or transient elevations in cytosolic [Ca²⁺].

ER Ca²⁺ release is also coupled to the mechanisms of Ca²⁺ removal. Whereas extrusion of Ca²⁺ by PMCA serves to lower cytosolic [Ca²⁺] and limit CICR, Ca²⁺ uptake into the ER promotes subsequent release [60,144]. Given the 10⁴-fold [Ca²⁺] gradient between the ER lumen and the cytosol, the overall Ca²⁺ content of the ER serves as a powerful driving force for Ca²⁺ release [60]. Neuronal activity results in SERCA-dependent sequestration of Ca²⁺ into the ER stores, which when released, promotes store-operated Ca²⁺ entry and modulates SV release probability [60]. These findings demonstrate the functional relationships between plasma membrane-resident channels, Ca²⁺ pumps, and ER channels in the regulation of neuronal Ca²⁺ homeostasis.

7.2. Physiological Agonists of IP₃Rs and RyRs

IP₃R1 and RyR1/RyR3 are the respective channel isoforms that are most enriched in the nervous system [145,146]. Despite exhibiting qualitatively similar relationships with Ca²⁺, the two families of channels differ markedly in biophysical parameters such as ionic conductance, adaptation and inactivation mechanisms, and kinetics [136]. IP₃Rs are activated by the soluble second messenger, IP₃, which lowers the apparent Ca²⁺ responsiveness of the channels towards resting cytosolic [Ca²⁺] [147,148]. Phospholipase C (PLC) enzymes, predominantly comprised of PLCβ, PLCγ, and PLCδ subtypes, hydrolyze the phosphoinositide, PI(4,5)P₂ (henceforth referred to as PIP₂), to generate diacylglycerol (DAG) and IP₃ (Figure 5) [147,149,150]. Although all three subtypes of PLC hydrolyze PIP₂, they exhibit distinct activation and inactivation kinetics such that the shapes and dynamics of the attendant ER Ca²⁺ release transients are a function of the subtype being stimulated. Whereas PLCβ isoforms are activated by Gq-coupled receptors, PLCγ isoforms are coupled to receptor tyrosine kinase signaling pathways, and PLCδ isoforms are regulated by phospholipids and Ca²⁺ [149,150,151,152,153]. The enzyme subtypes also have distinct functions in the nervous system. For instance, PLCβ activation is involved in regulation of neuronal tone and excitability, whereas PLCγ activation has a bigger influence on neurodevelopment and synaptic plasticity [38,150,154,155,156,157,158,159]. It is important to note, however, that only a subset of PLC functions can be attributed to subsequent IP₃R activation since PLC-mediated PIP₂ depletion can influence neuronal Ca²⁺ homeostasis via a number of different pathways [34]. IP₃ kinases, IP₃K1 and IP₃K2, convert IP₃ to the higher inositol phosphate, IP₄, which is unable to bind IP₃Rs [160]. Thus, the combined activities of IP₃K1 and IP₃K2 serve to restrict channel activity.

RyR activity is modulated by ryanodine and Ca²⁺, and promoted by the second messenger, cyclic ADP ribose (cADPR), synthesized from NAD⁺ by ADP-ribosyl cyclases such as CD38 and CD157 (Figure 5) [161,162,163,164,165]. Depending upon the context and cell type, RyR activation by cADPR could also involve accessory proteins such as calmodulin [166,167,168]. Not surprisingly, many functions of RyRs in neurons are subject to regulation by ADP-ribosyl cyclases and cADPR [148]. For instance, cADPR-mediated release of ER Ca²⁺ affects the excitability of dorsal root ganglion (DRG) neurons via Ca²⁺-activated currents [169,170]. Activation of RyRs by cADPR is also involved in the regulation of neuronal firing frequency, SV release, and synaptic plasticity in both vertebrate and invertebrate neurons [171,172,173,174].

7.3. Mechanisms Involved in the Activation of ER-Resident TRP Channels

TRPVs, TRPM1, TRPM8, and TRPP2 subtypes of the TRP superfamily of cation channels have also been shown to localize and/or function in ER Ca²⁺ release [175,176,177,178,179]. TRPP2 channels have been suggested to function as Ca²⁺-activated release channels, although this finding needs further verification [177,178]. While TRPM8 activation in the ER has been suggested to require phosphoinositides [176], the activation mechanisms for ER-resident TRPM1 remain unknown. As mentioned previously, the Drosophila TRPV channel, Iav, is localized to axonal and presynaptic ER in motor neurons [23,180]. Iav-mediated ER Ca²⁺ release is required for the maintenance of presynaptic resting [Ca²⁺] and SV release probability [23]. The localization and function of a Drosophila TRPV channel in the ER is reminiscent of a fraction of mammalian TRPV1 being localized to the ER, and in close proximity to IP₃Rs [181,182,183,184,185,186,187,188,189]. Concordantly, either the overexpression of TRPV1 or application of the channel agonist, capsaicin, leads to ER Ca²⁺ release [23]. Involvement of TRPV1 in ER Ca²⁺ release also explains why ectopic expression of TRPV1, but not TRPV4, in Drosophila motor neurons rescued the defects associated with ER Ca²⁺ release in hypomorphic alleles of the gene encoding Iav [183,184,185,186,187,188,189]. The exact mechanisms involved in the stimulation of TRPV1 or Iav in the ER remain unknown. Given the activation of TRPV1 by heat and endocannabinoids, it is possible that these modalities also activate the channel in the ER. Iav activity is regulated by mechanical stretch [181,190], although the involvement of stretch in Iav-mediated ER Ca²⁺ release has not been examined. We speculate that the role of Iav and TRPV1 in the regulation of resting [Ca²⁺] could reflect a potential mode of activation. Both Iav and TRPV1 contain Ca²⁺/calmodulin-binding motifs in their cytosolic domains. In case of TRPV1, occupancy of these sites by Ca²⁺/calmodulin leads to channel desensitization [33]. Notably, deletion of the calmodulin-binding site on TRPV1 potentiates capsaicin-induced ER Ca²⁺ release, which points to a role for Ca²⁺/calmodulin in channel inactivation at the ER [191,192]. It is possible that a similar mechanism suppresses Iav at resting or higher cytosolic [Ca²⁺]. If so, a drop in cytosolic [Ca²⁺] would disinhibit Iav, leading to ER Ca²⁺ release. We predict that this axis serves as a feedback mechanism for the maintenance of resting [Ca²⁺] within a narrow physiologically important level at presynaptic termini.

8. Pathologies Associated with Neuronal ER Ca²⁺ Dyshomeostasis

Perturbed Ca²⁺ homeostasis occurs in many neurological diseases. Being the largest intracellular reservoir of Ca²⁺, ER Ca²⁺ dyshomeostasis is of particular relevance to these diseases. In this section, we present an overview of the relationships between ER Ca²⁺ and pathology in a sampling of neurological diseases. We hope to highlight the tenet that qualitatively distinct alterations in ER Ca²⁺ signaling (either too much or too little release or uptake) can result in pathology. In fact, ostensibly similar changes in ER Ca²⁺ homeostasis can induce diverse spatiotemporal outcomes, indicating that changes in ER Ca²⁺ dynamics accompany neuronal processes perturbed in disease.

8.1. Autism Spectrum Disorder (ASD)

ASD is a constellation of neurodevelopmental disorders characterized by limited interpersonal communication and social skills. Although the molecular etiology of ASD is highly complex, evidence points to the involvement of ER Ca²⁺ alterations in its pathophysiology [193]. In cells derived from patients suffering from fragile X syndrome and tuberous sclerosis, which are monogenic syndromes that present with ASD, IP₃-mediated Ca²⁺ signaling is compromised [194]. This paradigm has also been extended to sporadic forms of ASD that are characterized by diminished ER Ca²⁺ release in response to stimulation of PLCβ-coupled receptors [195]. The purported involvement of IP₃Rs in ASD likely stems from the role of these channels in the regulation of neuronal excitability and the excitation–inhibition balance of neuronal circuits, both of which are necessary for the regulation of social behaviors [193].

Mutations in Ryr1 accompany enhanced dendritic arborization and impaired social behavior in mice [196]. Behavioral analyses of Ryr3-deficient mice also suggested abnormal social interactions [197]. The relationship between RyRs and social behavior is not restricted to mice since a gene duplication has pointed to RYR2 as a target gene in a Lebanese cohort of ASD patients [198]. Additional insights into ER Ca²⁺ dyshomeostasis in ASD may be gleaned from studies of fragile X, in which loss of fragile X mental retardation protein (FMRP) results in the broadening of action potential waveforms, and elevated neurotransmitter release [199]. The involvement of Ca²⁺-activated BK-type K⁺ channels in this process raises the possibility that the demonstrated relationship between BK channels and ER Ca²⁺ release could contribute to autistic behaviors [132,199].

8.2. Lysosomal Storage Diseases (LSDs)

LSDs are inborn errors of metabolism that are often associated with severe neurodevelopmental defects [200]. Although primarily associated with endolysosomal dysfunction, sustained changes in ER Ca²⁺ dynamics have been reported in several LSDs [201,202,203,204,205,206,207]. The relationship between lysosomes and ER in the context of Ca²⁺ homeostasis agrees with the findings that endolysosomal Ca²⁺ can be traced to the ER. Ca²⁺ released by activated IP₃Rs is loaded into lysosomes, which in turn, mediates various aspects of endolysosomal function [208,209]. Although the mechanisms of endolysosomal Ca²⁺ uptake remain incompletely understood, involvement of vesicular K⁺ channels or Ca²⁺/H⁺ exchangers have been suggested [210,211,212]. Reminiscent of MAMs that from between ER and mitochondria, organellar contact sites between the ER and endolysosomes form in a VPS13C-dependent manner, and permit localized [Ca²⁺] elevations needed for interorganellar ion transfer [208,213,214].

The aforementioned relationships between ER and lysosomal Ca²⁺ also explain some of the pathological features of LSDs. For instance, in primary neuronal cultures generated from murine models of Gaucher’s disease and patient-derived tissues, accumulation of glucosylceramide potentiated RyRs resulting in excessive Ca²⁺ release [201,202]. Concordantly, increased expression of SERCA2b or inhibition of RyR, two strategies that increase ER luminal Ca²⁺, improved proteostasis and the stability of glucocerebrosidase, the lysosomal enzyme mutated in Gaucher’s disease [203]. These findings suggest that elevated ER Ca²⁺ release is a pathological event in Gaucher’s disease, likely triggering the unfolded protein response due to sustained ER Ca²⁺ store depletion [204]. RyR antagonists were also found to be beneficial in restoring lipid homeostasis in cells isolated from patients suffering from Niemann-Pick type C [205]. Similarly, in murine models of GM1-gangliosidosis, the accumulation of GM1-ganglioside within MAMs results in the activation of IP₃R1 and attendant ER Ca²⁺ depletion [206], whereas in Sandhoff’s disease (caused by deletion of HexB), GM2-gangliosides inhibit SERCA, leading to the passive depletion of ER Ca²⁺ stores [207]. Together, these studies prompt the speculation that suppression of ER Ca²⁺ release mechanisms could restore proteostasis, and thereby prove beneficial in LSDs.

8.3. Neuropsychiatric Diseases

In accordance with the involvement of ER Ca²⁺ dynamics in neuronal development, synaptic plasticity, and excitability, Ca²⁺ dyshomeostasis has been described in neuropsychiatric diseases [215]. Darier’s disease, which is caused by mutations in the SERCA2 encoding ATP2A2 gene, demonstrates a causal link between ER Ca²⁺ and neuropsychiatric disorders [215]. Amongst the spectrum of pathophysiological features in Darier’s disease is the prevalence of schizophrenia, bipolar disorder, and major depression [215,216,217,218,219,220]. Interestingly, Darier’s patients exhibit distinct neuropsychiatric outcomes depending on the locations of the mutations in SERCA2 [215]. Consistent with the appearance of psychiatric symptoms in Darier’s patients, it is notable that ER stress, an established outcome of sustained Ca²⁺ store depletion, is associated with the neuroplastic changes observed in major depression [221]. This notion informs the idea that attenuation of ER Ca²⁺ release would counteract, and thus potentially reverse, neuropsychiatric outcomes. Indeed, knockdown of RyR genes in mouse brains elicited antidepressant-like effects [222]. Moreover, the prototypical mood stabilizer, lithium, has been suggested to exert some of its therapeutic effects by depleting free inositol, and the consequent attenuation of phosphoinositide signaling and IP₃ production [223,224,225]. Diminished phosphoinositide signaling following lithium administration is thought to decrease IP₃ production and thus, restrict IP₃R-mediated ER Ca²⁺ release.

8.4. Peripheral Neuropathies

Charcot Marie Tooth disease (CMT) constitutes a spectrum of peripheral neuropathies associated with the degeneration of the long neuronal processes of motor and sensory neurons. Mutations in many genes are associated with CMT, and at least a subset of these lead to disease-causing defects in ER Ca²⁺ homeostasis and MAM function. A quintessential example illustrating the involvement of MAMs in this disease is CMT2A, which is caused by mutations in the gene encoding a mitochondrial protein, mitofusin-2 [226]. Further supporting a role for mitochondrial proteins in the pathophysiology of CMT, mutations or loss of the gene encoding GDAP1 lead to CMT2K and CMT4A [227]. However, in contrast to CMT2A, GDAP1 mutations have been shown to result in decreased store-operated Ca²⁺ entry [227,228,229]. Forms of CMT attributed to mutations in the gene for peripheral myelin protein 22 (PMP22) also result in diminished store-operated Ca²⁺ entry, but in Schwann cells [230]. These data point to the complexities of ER-driven Ca²⁺ entry mechanisms in forms of CMT.

Although the exact role of mitofusin-2 in the regulation of Ca²⁺ homeostasis remains poorly understood, deletion or CMT2A-associated mutations in mitofusin-2 have been reported to prevent elevations in mitochondrial [Ca²⁺] in response to IP₃R activation [226]. In contrast, another study has reported increased interorganellar Ca²⁺ transfer upon the knockdown of mitofusin-2 [231]. The conflicting findings regarding the role of mitofusin-2 likely reflect hitherto undescribed roles for additional factors that modulate ER–mitochondrial Ca²⁺ transfer. In addition, mitofusin-2 could be regulating attachment of a variety of different organelles, such as the overall effects of mitofusin-2 knockdown, which is highly context-dependent. Indeed, the association between the ER and the plasma membrane necessary for STIM-dependent activation of store-operated Ca²⁺ entry has also been proposed to involve mitofusin-2 [232].

8.5. Neurodegenerative Diseases Directly Attributed to ER Ca²⁺ Release Channels

Neurodegeneration refers to the progressive loss of neuronal function, often resulting from premature neuronal demise. In general, neurodegenerative diseases represent neurological diseases with the clearest and best-documented involvement of ER Ca²⁺ dyshomeostasis. This notion is exemplified by the findings that mutations in the gene encoding IP₃R1, ITPR1, result in forms of spinocerebellar ataxia (SCA15 and SCA29) and Gillespie syndrome [233,234,235,236,237,238,239,240,241,242]. SCA15 is an adult-onset disease that presents with autosomal dominant cerebellar ataxia and attendant gait impairment due to heterozygous deletions spanning the ITPR1 locus [233,234,235,236]. Mice that are heterozygous for a spontaneously arising Itpr1 loss-of-function allele also exhibit severe cerebellar degeneration leading to abnormal locomotion starting around postnatal day 14 [234]. The exceptionally high abundance of IP₃R1 in the cerebellar Purkinje neurons [234,243] could explain the haploinsufficiency observed upon deletion of one of the two copies of the gene encoding the channel. Constitutive homozygous deletion of Iptr1 in mice leads to significant embryonic lethality [243]. Of the few animals that are born, all exhibit severe ataxia, tonic-clonic seizures, and die by the weaning period [243].

SCA29 is an infantile-onset form of the disease, and is associated with cerebellar atrophy, hypotonia from infancy, non-progressive ataxia, and related psychomotor deficiencies [238]. Patients with SCA29 are generally heterozygous for missense mutations in ITPR1 that occur in exons encoding the IP₃-biding domain of the protein, and thus, diminish or abolish channel function [237,238]. While most cases of SCA stem from decreased IP₃R1 activity as the consequence of haploinsufficiency, one missense variant (IP₃R1^R36C) results in elevated IP₃-binding affinity and sustained channel activation [244]. Interestingly, these patients also present with typical features of SCA, including gait abnormalities and delayed motor development [244]. These data point to the importance of ITPR1 dosage and maintenance of channel activity within a ‘Goldilocks’ zone for Purkinje neuron viability.

Gillespie syndrome, which is a rare congenital disorder characterized by hypotonia, ataxia due to progressive cerebellar atrophy, and intellectual disability, is also caused by autosomal recessive or dominant missense mutations in ITPR1 [240,241,242]. ITPR1 mutations in Gillespie syndrome are varied, and often involve partial expression of the wild-type gene accompanied by truncated variants that exert dominant-negative effects [240,241,242]. Although the cerebellar and motor outcomes in Gillespie syndrome mimic those in SCA, an intriguing feature that distinguishes Gillespie syndrome is the appearance of partial aniridia [240,241,242,245]. The mechanisms underlying the specificity of aniridia in Gillespie syndrome remain unknown, and mice with ITPR1 mutations do not exhibit aniridia [242].

8.6. Age-Related Neurodegenerative Diseases

Age-related neurodegenerative diseases (Alzheimer’s disease (AD), Parkinson’s disease (PD), Huntington’s disease (HD), and amyotrophic lateral sclerosis (ALS)) exhibit ER Ca²⁺ dyshomeostasis. A “Ca²⁺ hypothesis” has been advanced to bridge the mechanistic gap between amyloid accumulation and cognitive decline in AD [246]. This model is built on the understanding that AD neurons exhibit elevated cytosolic Ca²⁺ levels. For instance, baseline cytosolic [Ca²⁺] in neurons from the 3xTg-AD mouse model is double that in control cells [247]. Mutated variants of presenilin, which are a familial cause of AD, potentiate the activity of IP₃Rs via direct regulation of channel gating or other indirect mechanisms (Figure 6) [65,248,249,250]. A number of other studies have also reported increases in the abundance and activity of RyRs in AD mouse models, which has raised the possibility of treating AD with the RyR antagonist, dantrolene [59,251,252]. Adding to this complexity, presenilins also regulate the activity of SERCA, ER–mitochondrial transfer of Ca²⁺ and phospholipids, mitochondrial ATP production, and overall cellular Ca²⁺ homeostasis (Figure 6) [253,254,255]. In contrast, amyloid β oligomers have been shown to increase the transfer of Ca²⁺ from the ER to mitochondria, leading to mitochondrial Ca²⁺ overload, sustained store depletion, and toxic activation of store-operated Ca²⁺ entry [256]. As mentioned in a previous section, ER Ca²⁺ release and store-operated Ca²⁺ entry (Figure 6) can lead to calcineurin-dependent nuclear translocation of NFAT, and potentiate cell death in mature neurons [98,102,103]. Pointing to the involvement of this axis in AD, inhibition of the calcineurin/NFAT pathway alleviates amyloid β-induced neurodegeneration in a murine model of the disease [257]. An additional consequence of ER Ca²⁺ store depletion is ER stress and activation of the unfolded protein response (UPR), which also contributes to AD pathology [258,259,260]. Despite these studies pointing to a pathological role for excessive ER Ca²⁺ release in AD, further studies are needed to reconcile the conflicting reports suggesting either absent or opposite changes in ER Ca²⁺ homeostasis in AD [261,262,263,264].

One view of the pathological mechanisms underlying PD, a disease associated with the loss of dopaminergic neurons of the substansia nigra, revolves around the concept of dysregulated MAMs and perturbations in ER–mitochondria Ca²⁺ signaling [265]. In induced pluripotent stem cell (iPSC)-derived dopaminergic neurons from familial PD patients with triplication of the α-synuclein gene, interactions between the ER and mitochondrial counterparts of MAMs VAPB and PTPIP51, respectively, are disrupted [266]. The resulting cessation of interorganellar Ca²⁺ transfer decouples IP₃R activation from mitochondrial Ca²⁺ elevation, and thus, limits mitochondrial ATP production. Deprivation of ATP in the dopaminergic neurons of the substansia nigra may be especially consequential, as these cells perform a particularly energy demanding pacemaker function that requires tight homeostatic control over ionic gradients [265]. As is the case in AD, calcineurin and NFAT mediate α-synuclein-induced loss of dopaminergic neurons [267,268]. Since ER Ca²⁺ release and store-operated entry activate calcineurin/NFAT-dependent neurotoxicity, these findings are consistent with augmented ER Ca²⁺ release as being a causal insult in PD.

HD is a neurodegenerative disease caused by CAG repeat extensions in the gene encoding Huntingtin, and is characterized by extensive loss of neurons in the striatum resulting in characteristic chorea and progressive dementia [269]. IP₃R1–GRP78 coupling and enhanced transfer of Ca²⁺ between the ER and mitochondria have been suggested to underlie ER stress and degeneration in a mouse model of HD [270]. In agreement with these reports, expression of mutant Huntingtin (i.e., with polyglutamine repeat expansions), but not wild type Huntingtin, in murine medial spiny neurons enhanced IP₃-mediated Ca²⁺ release [271]. Genetic studies in Drosophila further support this interaction given that knockdown of IP₃R suppressed neurodegeneration in a fly HD model [272]. As expected from elevated ER Ca²⁺ release, store-operated Ca²⁺ entry is enhanced in striatal neurons of a mouse model of HD, which underlies the synaptic loss observed in those animals [273]. A unique feature of HD in the context of ER Ca²⁺ is that mutant Huntingtin exists in a complex with IP₃R1 via interaction with the C-terminus of the channel [274]. Thus, expression of a C-terminal fragment of IP₃R1 attenuates the interaction between mutant Huntingtin and IP₃R1 leading to the mitigation of neuronal loss in a mouse model of the disease [274].

Various lines of evidence also point to ER Ca²⁺ dyshomeostasis stemming from perturbations in ER Ca²⁺ release and MAM assembly as being important pathological mechanisms in ALS, a lethal neurodegenerative disease characterized by motor neuron degeneration [275]. A familial form of ALS (ALS8) is caused by a mutations in the MAM resident protein, VAPB (for example, VAPB^P56S), which participates in ER-mitochondrial tethering [275,276,277]. Although ALS8 is a rare form of ALS, decreased VAPB protein abundance has been observed in spinal cord motor neurons from patients with sporadic ALS [278]. In other forms of familial ALS, such as those caused by mutated variants of FUS and TDP-43, VAPB–PTPIP51 interactions are comprised with attendant decreases in mitochondrial [Ca²⁺] and ATP production (Figure 7) [279,280]. In accordance with the relationship between TDP-43 and ER Ca²⁺ homeostasis, a worm model of ALS exhibits ER Ca²⁺ dyshomeostasis that led to decline of neuronal function [281]. ALS-causing mutations in the gene encoding Sigma receptor-1, and C9orf72 hexanucleotide expansions (a leading cause of both sporadic and familial forms of ALS in populations with European ancestry) are also associated with ER Ca²⁺ dyshomeostasis due to combinations of elevated IP₃R activity and diminished uptake into stores [282,283].

9. Closing Remarks

In this review, we have described the involvement of ER Ca²⁺ homeostasis in neuronal functions ranging from development and plasticity to gene expression and cell survival. We also provided examples of neurological and neurodegenerative diseases that arise when the processes regulating the movement of Ca²⁺ in and out of the ER go awry. It is important to note that despite being one of the most extensively studied organelles, much still remains to be known about ER Ca²⁺ and its relationship to neuronal function and disease. In particular, we anticipate a proliferation of studies focused on the domains that the ER forms with other organelles such as the mitochondria, endolysosomes, and the plasma membrane. These studies should go a long way to illustrate that in the ultimate analysis, intracellular organelles function in an ecosystem that requires invaluable contributions from all the players. By remaining open to these possibilities, we can hope for novel therapies for a variety of debilitating neurological diseases that arise from miscommunication between neuronal organelles.