The endoplasmic reticulum (ER) is the largest single intracellular organelle, which is present in all types of nerve cells. The ER is an interconnected, internally continuous system of tubules and cisterns, which extends from the nuclear envelope to axons and presynaptic terminals, as well as to dendrites and dendritic spines. Ca2+ release channels and Ca2+ pumps residing in the ER membrane provide for its excitability. Regulated ER Ca2+ release controls many neuronal functions, from plasmalemmal excitability to synaptic plasticity. Enzymatic cascades dependent on the Ca2+ concentration in the ER lumen integrate rapid Ca2+ signaling with long-lasting adaptive responses through modifications in protein synthesis and processing. Disruptions of ER Ca2+ homeostasis are critically involved in various forms of neuropathology.
I. ENDOPLASMIC RETICULUM AND CELLULAR PHYSIOLOGY: HISTORICAL REMARKS
In 1674 Antonius van Leeuwenhoek, after discovering the cross-striations of muscle fibers using one of his ingenious light microscopes, reflected “… who can tell, whether each of these filaments may not be enclosed in its proper membrane and contain within it an incredible number of still smaller filaments.” (236). One century later, Luigi Galvani discovered “animal electricity” (178). To explain his observations, Galvani suggested that nerve and muscle are capable of generating electricity by accumulating positive and negative charges on two opposite surfaces. Moreover, the electrical flow necessary to produce voltage changes was explained by Galvani in terms of specific fluid-filled pores between the internal and external surfaces, which were the harbingers of ion channels.
The 19th century witnessed the emergence of the cellular theory. Furthermore, it became apparent that the cell is a complex structure containing numerous organelles. Great histologists, such as Rudolf Virchow, Gustav Retzius, Santiago Ramon y Cajal, and Jan Purkinje, to name but a few, developed various techniques to look inside the cells, and it was Camillo Golgi who, using his “black (silver chromate) reaction” technique in 1898 discovered the “fine and elegant reticulum hidden in the cell body” of owl cerebellar Purkinje neurons (187; see also Refs. 37, 384), the structure known to us as the “Golgi complex.” The first detailed description of the endoplasmic reticulum was made by Golgi’s pupil, Emilio Veratti, who, by using black reaction technique on muscle fibers, was the first to discover a “true reticular apparatus constituted of filaments” (385, 668, 669; Fig. 1). This finding remained almost unnoticed for 50 years, and only when the endoplasmic reticulum (ER) was rediscovered by Keith Porter and George Palade, who used electron microscopy (481), were Veratti’s achievements generally recognized.
Insights into the physiological function of the ER and the appearance of the concept of intracellular Ca2+ stores resulted from studies of muscle contraction. Sidney Ringer’s experiments (535), which clearly identified Ca2+ as an ion mediating muscle contraction, did not address directly the question of the Ca2+ source, although Ringer noticed that contractions of skeletal muscle can be maintained in Ca2+-free media for hours and even days (536, 537), suggesting the existence of some barriers hampering the exchange of ions between the muscle interior and the external milieu. Further notions of the possible importance of intracellular Ca2+-containing systems were stimulated by experiments in which it was shown that movements of ameba are particularly sensitive to chelation of intracellular Ca2+. Injection of Ca2+-precipitating agents such as phosphates, sulfates, or alizarine sulfonate froze the movements of the amebas almost instantly (513, 532). The first direct evidence for the existence of a sizable intracellular Ca2+ storage compartment was obtained by Weise, who found that excessive physical exercise led to the appearance of Ca2+ in the ultrafiltrate of rat skeletal muscle (691). Later, an increased release of radioactive Ca2+ from stimulated frog sartorius muscle was demonstrated by Woodward (698), thus proving the existence of intracellularly stored calcium.
The idea of an intracellular Ca2+ store and the importance of regulated intracellular Ca2+ release for muscle contraction was formulated by Heilbrunn, who was also the first to test these concepts experimentally (210, 211). His theory was not readily accepted, and for many years K+ were considered to be the regulators of contraction and relaxation, a theory very much supported by Szent-Gyorgyi (619). Only after the free intracellular Ca2+ in skeletal muscle was measured directly by murexide (263) or aequorin (15, 533) could internal Ca2+ release be visualized and the intracellular Ca2+ store became the legitimate member of a Ca2+ signaling cascade. In the meantime, the concept of intracellular Ca2+ pumps associated with the endomembrane was developed by W. H. Hasselbach, Setsuro Ebashi, and Anne-Marie Weber (see Ref. 159), explaining how the ER could accumulate Ca2+. At the same time, the first indications of ER-residing regulated ion channels, which could provide a means of Ca2+ release, appeared (158).
At the beginning of the 1970s the crucial role of the ER (in its sarcoplasmic reticulum disguise) in regulation of contraction became generally accepted and the concepts of Ca2+-induced Ca2+ release and depolarization-induced Ca2+ release firmly established (see Refs. 140, 160 for review). At the same time, the receptor-activated, second messenger-induced intracellular Ca2+ release was discovered (81, 457), and a possible role for inositol phospholipids in signal transduction was suggested (410). Another decade, however, passed before the seminal experiments of Hanspeter Streb, Robin Irvine, Irene Schulz, and Michael Berridge (607) showed that this alternative pathway for intracellular Ca2+ release is mediated by the second messenger inositol 1,4,5-trisphosphate (InsP3). The intracellular Ca2+ release channels, residing within the endomembrane, were soon molecularly characterized, revealing two main families: the Ca2+-gated and InsP3-gated channels (505, 613). Thus the general concept of intracellular Ca2+ stores was established, and the stage was set for identification of their functional role in various cell types.
The ability of nerve cells to actively buffer cytosolic Ca2+ was initially associated with Ca2+ accumulation by mitochondria, the latter utilizing H+ transport to develop an electrochemical gradient favoring Ca2+ entry (334, 544). The functional importance of the neuronal ER as a dynamic Ca2+ pool was first appreciated after the demonstration that neuronal microsomal fractions can accumulate Ca2+ via an ATP-dependent mechanism (649). Almost simultaneously it was also discovered that various neuronal preparations, such as the squid axon (64), extruded axoplasm (21), and permeabilized synaptosomes (51, 52), are able to sequester Ca2+ even when mitochondrial uptake was obliterated (see also Ref. 416 for review). Finally, this nonmitochondrial Ca2+ uptake was unequivocally associated with the ER (212–214, 682), and the latter became firmly established as a dynamic intracellular Ca2+ store, deeply involved in neuronal signaling.
II. ORGANIZATION OF THE ENDOPLASMIC RETICULUM
A. The Neuronal ER as a Continuous Endomembrane Structure
The endoplasmic reticulum is arguably the largest single intracellular organelle, which appears as a three-dimensional network formed by an endomembrane and organized in a complex system of microtubules and cisternae. In neurons, the ER extends from the nucleus and the soma to the dendritic arborization and, through the axon, to presynaptic terminals. Classical histology subdivides the ER into smooth ER, rough ER, and the nuclear envelope (32). The continuity of the endomembrane that forms these three structures was initially deduced from microscopic observations (130, 131, 371, 682). Further insights into the continuity of the ER were gained by fluorescent microscopy on cells injected with the lipophilic probe 1,1′-dihexadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiIC16), which diffuses within strictly continuous lipid membranes. When injected into cells, DiIC16 labeled the intracellular membrane network endowed with ER markers (632, 633). Injection of DiIC16 into rat cerebellar Purkinje neurons in acutely isolated brain slices revealed a continuous membrane network extending throughout all parts of the cell (Fig. 2, A and B). The kinetics of dye spread was consistent with intramembrane diffusion; moreover, the ER labeling occurred even when slices were fixed 2 min after dye injection, thus excluding the possibility of active transport (634). Very similar ER labeling was observed in cultured hippocampal neurons (634). The concept of a continuous ER was further substantiated by experiments on nonneuronal cells. Initially, the continuity of the intra-ER aqueous space was directly demonstrated in pancreatic acinar cells, where soluble molecules, such as Ca2+ and fluorescent dyes, were shown to move throughout the ER lumen with remarkable ease (491, 509). Later on, it was found that much larger molecules of green fluorescence protein (GFP) targeted against the ER and expressed in Chinese hamster ovary (CHO) cells, were also able to freely diffuse within the ER lumen. Repeated local laser illumination of a spot 2 μm in diameter eventually bleached all GFP present in the cell, thus demonstrating the continuity of the ER lumen (115). Similarly, in RBL cells transfected with even larger ER-targeted elastase-GFP fusion protein (molecular weight 60 kDa), fluorescence recovery after photobleaching revealed rapid (diffusion coefficient ∼0.5 μm2/s) movements of this, rather hefty, protein within the entire ER lumen (610).
Although internally continuous, the ER exhibits an extraordinary heterogeneity in both its morphology and the functional properties. The ER can be represented as tubular networks as well as by flattened cisternae, the latter sometimes arranged in complex cisternal ER stacks or even crystalloid structures found in photoreceptors (682) and Purkinje neurons (625). In neuronal somata, the ER network is omnipresent (Fig. 2), and part of it forms the nuclear envelope. Portions of the ER in the soma and in the distal dendrites are molded into specialized structures known as subsurface cisternae, which come into close contact with the plasma membrane (620). These cisternae exist in several shapes and forms: type I is separated from the plasmalemma by a distance of 60–80 nm, whereas types II and III reside much closer, at a distance of ∼20 nm from the plasmalemma. These latter types of ER cisternae often follow the contour of the surface membrane (543). Sometimes the surface cisternae are organized in more complex structures, the cisternal organelles, which have been found in the proximal parts of axons (34, 267, 305, 548).
In axons, the ER mostly has a tubular structure and extends into the presynaptic terminals, where it often closely enwraps the mitochondria (48, 391, 693). The ER also spreads throughout the dendrites, terminating in the spine apparatus and thereby connects the latter with the entire ER lumen (604).
An important feature of the ER is its morphological plasticity. The ER network constantly remodels itself through multiple forms of tubular movements. These movements occur quite rapidly; new structures emerge within tens of seconds (57). Even large formations such as crystalline cisternal organelles are not static and may swiftly surface as part of a neuronal adaptive response, for example, against hypoxia (625).
The topographical heterogeneity of the ER is matched by its functional diversity. Protein synthesis within the rough ER occurs in specific compartments, with different mRNAs being specifically selected and targeted towards distinct regions of the reticulum (209, 525, 682). Sometimes mRNA is engaged in a long-distance travel to end up in the ER in dendrites, where the actual synthesis of a locally demanded protein (e.g., InsP3 receptors, Ref. 606) takes place (180, 566). Here, the ER closely interacts with another reticular structure, the Golgi apparatus, the outposts of which are present in dendrites and are involved in final trafficking of membrane proteins and proteins destined for secretion (237, 238). Although the precise mechanisms of the intra-ER mRNA sorting remain unclear, they appear to be an important part of proteins targeting cascades. The synthesis of phospholipids and cholesterol similarly appears to be spatially heterogeneous, with different steps of the process occurring in different ER compartments (32). The enormous functional heterogeneity of the ER is nonetheless integrated into a coherent system by the continuous lumen.
B. The ER as a Universal Signaling Organelle
The functions of the ER are many. First and foremost, the ER is the site of synthesis and maturation of proteins. Protein synthesis is accomplished in the rough ER, while posttranslational processing of proteins is governed by an extended family of ER-resident chaperones, which form complexes with newly synthesized proteins, assist their folding into the final tertiary structure and prevent them from aggregation. If the process of folding somehow fails, the chaperones remain assembled with misfolded proteins, thereby preventing them from proceeding through the ER exit sites and towards the Golgi complex. Whenever the concentration of unfolded proteins rises dangerously high, the ER develops a specialized reaction known as ER stress, as a result of which transcription-affecting signals are sent to the nucleus, which, in turn, adjusts gene expression according to environmental requirements. In addition to the synthesis of proteins, the ER is the primary site of formation of phospholipids, glucosylphosphatidylinositols, and leukotrienes. The ER may also serve as a graveyard for various unwanted molecules and toxins. Owing to its continuous lumen, the ER serves as a highway allowing haulage of transport RNAs, secretory products, numerous structural proteins and, as we shall see later, ions between different parts of polarized cells. The ER is also intimately involved in rapid cellular signaling, because it is a dynamic store of Ca2+ that regulates the cytosolic Ca2+ concentration and generates Ca2+ fluxes between the cytosol and the ER lumen in response to extracellular stimulation. Finally, the ER coordinates all these diverse processes with the physiological state of the cell. Therefore, the ER may be defined as a multifunctional organelle able to detect and integrate incoming signals and generate output signals in response to environmental changes (40, 57). For a comprehensive overview of ER functions, see References 65, 257, 364, 388, 392, 399–402, 479, 487, 508, 509, 591, 677.
The intimate mechanisms of ER integration remain largely undiscovered, yet a central role for Ca2+ is emerging. First, Ca2+ is a key input and output signal of the ER. Indeed, Ca2+ concentration increases in the cytosol affect its concentration in the ER, and in turn, the ER Ca2+ release and uptake influence the cytosolic Ca 2+concentration. Second, a number of intra-ER chaperones, such as calreticulin, calnexin, grp78/BiP, endoplasmin (or glucose-regulated protein, grp94), are Ca2+ binding proteins, and changes in free Ca2+ concentration in the lumen of the ER ([Ca2+]L) profoundly affect their functional activity (106, 409). Therefore, fluctuations in the ER Ca2+ content may provide the link between rapid signaling and long-lasting cellular adaptive responses.
C. The ER as an Excitable Organelle
The basic physiology of the ER calcium store has been extensively characterized in a variety of excitable and nonexcitable cells (for key reviews, see Refs. 38–42, 56, 75, 76, 147, 177, 308, 309, 381, 507, 508, 539, 670–672, 712). The ER acts as a dynamic Ca2+ store due to concerted activity of Ca2+ channels and transporters residing in the endomembrane, and intraluminal Ca2+-binding proteins, which serve as a high-capacity Ca2+ buffering system. Ca2+ efflux from the ER is executed by two families of Ca2+ channels, the Ca2+-gated Ca2+ channels, generally referred to as the ryanodine receptors (RyRs), and the InsP3-gated channels, commonly known as InsP3 receptors (InsP3Rs). Ca2+ accumulation into the ER lumen results from the activity of Ca2+ pumps of the sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA) family. The molecular physiology of ER channels and SERCA pumps has been comprehensively reviewed recently (RyRs, Refs. 59, 148, 161, 396, 545, 695; InsP3Rs, Refs. 44, 411, 412, 631; SERCA pumps, Refs. 121, 134, 587, 618, 699); hence, I shall limit myself to a brief discussion of their key features in the context of neuronal signaling.
1. Ca2+ release channels
The Ca2+ release channels, the RyRs and the InsP3Rs, are large tetrameric channel proteins that share a rather peculiar four-leaf clover-like structure when observed by electron microscopy. The RyRs were the first to receive experimental attention due to their vital role in excitation-contraction coupling in muscle. The RyR family comprises three major receptor subtypes, which were historically classified as “skeletal muscle,” “heart,” and “brain” types, currently referred to as RyR1, RyR2, and RyR3, respectively. The historic classification had a certain physiological sense, as the RyR1 type, dominant in skeletal muscle, possesses a rather unique activation mechanism; it can be activated solely by membrane depolarization by virtue of a direct link with the plasmalemmal Ca2+ channels through a huge cytosolic “foot” extension. All three RyRs can be activated by Ca2+ from the cytosolic side; the rank order of their sensitivity to cytosolic free Ca2+ concentration ([Ca2+]i) being RyR1 > RyR2 > RyR3 (93, 435). On the basis of studies on reconstituted channels in lipid bilayer membranes, the RyRs can be biophysically described as not perfectly selective Ca2+ channels (PCa/PK ∼6–7) with fast activation kinetics (τ of activation ∼0.5–1 ms) and a large conductance (100–500 pS, depending of the nature of ion carrier) (148). Yet, taking into account the imperfect selectivity toward monovalent cations and the virtual absence of selectivity between divalents, the physiological unitary Ca2+ current through RyRs can be as small as 0.5–0.9 pA at 2 mM intraluminal Ca2+ concentration (397).
All three types of RyRs have been detected in neurons; however, expression of the RyR2 isoform dominates. Interestingly, although the RyR3 was initially called the “brain” subtype, its total expression in the central nervous system does not exceed 2% of all RyRs (434). The RyR1 is heavily expressed in cerebellar Purkinje neurons, whereas RyR3 is found in hippocampal structures, the corpus striatum, and the diencephalon (170, 185, 434, 474, 574).
However, these are only general trends; in reality, many neurons coexpress either two or all three types of RyR isoforms. RyRs are found in high densities in neuronal somatas; in axons of hippocampal (574), cerebellar (475), and cortical (723) neurons; in cerebellar mossy fibers (475); and in presynaptic terminals (355). RyRs are also widely expressed in the dendritic tree; their expression is particularly high in dendritic spines of hippocampal CA1 neurons, where they represent the dominant Ca2+ release channel (574). Incidentally, RyRs were visualized (with fluorescent ryanodine) in the Golgi complex of sympathetic neurons, although the functional relevance of this discovery remains unclear (98).
The expression of RyRs changes considerably during development; for instance, throughout the embryonic stage, RyR1 mRNA levels are highest in the rostral cortical plate, whereas RyR3 mRNA was most prominent in the caudal cortical plate and hippocampus. Low levels of RyR2 mRNA were detected in the diencephalon and the brain stem. However, from postnatal day 7 onward, RyR2 mRNA became the major isoform in many brain regions, while RyR1 mRNA became prominent in the dentate gyrus and in the Purkinje cell layer. Postnatal downregulation in the caudal cerebral cortex restricted RyR3 mRNA expression to the hippocampus, particularly the CA1 region (427).
In vivo imaging of RyRs can be achieved using fluorescent BODIPY-ryanodine (e.g., Refs. 389, 603; Fig. 2C). Because ryanodine binding to the RyR is use dependent, the intensity of BODIPY-ryanodine staining may reflect the overall density of open channels. With the use of this technique, a redistribution of active RyRs from soma to dendrites was observed in developing hippocampal neurons (611); moreover, L-type Ca2+ channel-RyR coupling was identified at early developmental stages (611), reflecting a developmental switch between RyR1 and RyR2 expression in nerve cells (427).
Similar to RyRs, the InsP3 receptor family comprises three homotetrameric isoforms, InsP3R1, InsP3R2, and InsP3R3, although further diversity may arise from alternative splicing of InsP3R1 (up to 6 variants; Ref. 463) and heteromeric expression (399). The most important difference between isoforms lies in their different sensitivity to [Ca2+]i: the InsP3R1 has the “classical” bell-shaped dependence on [Ca2+]i with a maximal activation at ∼300–400 nM (45); the InsP3R2 and InsP3R3 are also stimulated by [Ca2+]i increase, but are not readily inhibited by larger [Ca2+]i rises (198).
In the central nervous system, the InsP3Rs were initially discovered in the cerebellum (172, 613), but thereafter all three isoforms were identified throughtout the brain, with particularly strong expression in Purkinje neurons and in somatas of CA1 cells (170, 171, 574). The InsP3R1 is the most abundant isoform in the brain, whereas InsP3R2 dominates in the spinal cord and is also highly expressed in glial cells. The InsP3R3 are expressed to a much lesser extent but are nonetheless clearly detectable in several brain regions, such as the cerebellar granule layer and the medulla (575). There is a prominent heterogeneity in the intracellular distribution of InsP3Rs. For example, InsP3R1 is found in dendrites, dendritic spines, cell bodies, axons, and axonal terminals of cerebellar Purkinje cells, but is mostly confined to somatic regions and proximal dendrites in other neurons (122, 551, 575). The InsP3R3 is very much concentrated in the neuropil and neuronal terminals (575).
2. SERCA pumps
The SERCA pump belongs to the family of P-type Ca2+-ATPases and includes three gene products, SERCA1, SERCA2, and SERCA3; alternative splicing produces two distinct isoforms of SERCA2, namely, 2a and 2b. In central neurons, the SERCA2b is ubiquitously expressed, whereas SERCA2a and SERCA3 are found almost exclusively in cerebellar Purkinje neurons (18). The physiological role of SERCA3 remains quite mysterious, and genetic deletion of this Ca2+ pump does not result in any obvious phenotypic change, save minor deficits in relaxation of vascular and tracheal smooth muscles (351).
An important property of the SERCA2b pump that is relevant for ER Ca2+ homeostasis is the tight regulation of its activity by the free Ca2+ concentration in the ER lumen, mediated via the Ca+-binding proteins calreticulin (264) and ERp57 (341). Calreticulin senses the actual level of [Ca2+]L and, by directly interacting with SERCA2b, activates Ca2+ pumping whenever [Ca2+]L decreases. The ERp57 regulates Ca2+ uptake by modulating the redox state of SERCA2b thiols in a Ca2+-dependent manner.
The SERCA pumps have a distinct pharmacology; they are blocked irreversibly by thapsigargin (TG) in submicromolar concentrations (KD ∼20 nM) and reversibly by cyclopiazonic acid (CPA) at concentrations of ∼20–50 μM. Inhibition of the SERCA pumps results in a relatively slow emptying of releasable Ca2+ from the ER, with Ca2+ leaving the ER through poorly understood pathways (72). It must also be noted that TG is not perfectly specific for SERCAs; it also inhibits plasmalemmal voltage-gated Ca2+ channels of L/N-types in DRG neurons at concentrations of 0.2–2 μM (582). Likewise, micromolar concentrations of TG substantially decreased T and L Ca2+ currents in adrenal glomerulosa cells (546). The concentrations of TG that inhibit Ca2+ channels are comparable to those used in the majority of experiments in which it is intended to block the Ca2+ accumulation by the ER. A certain care, therefore, should be taken when interpreting data arising from such experiments.
3. ER Ca2+ buffers
ER Ca2+ buffering is accomplished by several Ca2+-binding proteins; in neurons, the most abundant is calreticulin, which accounts for nearly one-half of total ER Ca2+ binding. Calreticulin in the neuronal ER is endowed with 20–50 low-affinity Ca2+ binding sites (KD ∼1 mM) and therefore can bind a huge amount of Ca2+. As mentioned above, calreticulin acts not only as a Ca2+ buffer, but also as an important chaperone and regulator of SERCA pumps. Calsequestrin, the major ER Ca2+ buffer in skeletal myocytes, is also present in nerve cells, and like calreticulin may bind up to 50 Ca2+ with low affinity. Several other low-affinity Ca2+-binding proteins, such as endoplasmin (grp94), BiP/grp78, and proteins of the CREC family (reticulocalbin, calumenin, Cab55, etc.) also participate in intra-ER Ca2+ buffering (235, 399, 401, 409).
The most important property of the ER Ca2+ buffers is their low affinity to Ca2+, which allows the maintenance of high intra-ER free Ca2+ levels. There is a fundamental difference in Ca2+ handling in the cytosol and within the ER lumen. The cytoplasmic Ca2+ binding proteins are characterized by a very high (10–100 nM) affinity to Ca2+; therefore, they limit Ca2+ diffusion and favor localization of cytosolic Ca2+ signals. In contrast, the low Ca2+ affinity of the ER buffers allows Ca2+ diffusion throughout the lumen (423).
4. The ER as an excitable medium
The extensive complement of endomembrane Ca2+ channels and transporters, together with intra-ER Ca2+ storage proteins, gives rise to the excitable properties of the ER. Ca2+ accumulation in the ER lumen creates a large electrochemical driving force favoring Ca2+ movement from the ER lumen into the cytosol; hence, opening of Ca2+ channels in the ER membrane causes fast efflux of Ca2+ from the ER. In turn, the ER channels are sensitive to signals originating at the plasmalemma and so can be activated by both electrical and chemical stimulation of the cell. Furthermore, the ER channels are subject to positive autofeedback due to their sensitivity to cytosolic Ca2+, leading to a regenerative opening of RyRs and/or InsP3Rs beyond a certain threshold of [Ca2+]i. This regenerative process underlies propagating waves of channel activation along the ER membrane, very similar to the propagating wave of opening and closing of Na+ and K+ channels that generate plasmalemmal action potentials. Neuronal function is thus an interplay of two excitable membranes, the outer and the inner, which are coupled through a multitude of reciprocal signaling cascades (39). In contrast to plasmalemma, which is mostly specialized for integrating extracellular information, the ER uses Ca2+ to couple plasmalemmal and cytosolic events with intraluminal formation and transport of proteins, as well as with gene expression within the nucleus. The periodic fluctuations of the ER Ca2+ concentration are instrumental for this intracellular integration, because [Ca2+]L regulates the excitability of the ER membrane by controlling ER Ca2+ release channels and SERCA pumps, and governs intra-ER enzymatic cascades.
III. ENDOPLASMIC RETICULUM CALCIUM RELEASE IN NERVE CELLS
A. The Discovery of Ca2+ Release From the ER
Historically, the first observations of the release of Ca2+ from an intracellular pool were very much indebted to the previous discovery that caffeine can activate RyRs in muscle cells (688). The pioneering observation indicating that caffeine affects neuronal [Ca2+]i was made by K. Kuba and S. Nishi in 1976 (319). When monitoring membrane potential of neurons in isolated sympathetic ganglia of the bullfrog, Kuba and Nishi found that superfusion of the ganglia with an extracellular solution supplemented with 1–6 mM caffeine resulted in rhythmic hyperpolarizations. Further on, they found that the caffeine-induced effects on membrane potential (Vm) were mediated through a Ca2+-dependent K+ conductance, because intracellular injection of EDTA inhibited the action of caffeine. Several years later, Kenji Kuba made the logical suggestion that “rhythmic increases in the GK under the effect of caffeine are due to oscillations of the intracellular Ca2+ concentration and that there may be Ca storage sites in the bullfrog sympathetic ganglion cell which are comparable to the sarcoplasmic reticulum in the skeletal muscle fiber” (317). Using K+ permeability as a readout of [Ca2+]i, several groups confirmed these observations (219, 318, 431, 557). The advent of luminescent and fluorescent Ca2+ probes allowed direct monitoring of [Ca2+]i, which became instrumental for detailed investigations of ER Ca2+ release in nerve cells.
B. Ca2+-Induced Ca2+ Release
1. Caffeine-induced Ca2+ release
As already mentioned, the methylxantine caffeine was the first specific agent permitting identification of ER Ca2+ release, and it has remained the probe of choice (Table 1). Probably the first direct indication that another methylxantine, theophylline, triggers [Ca2+]i rises in vertebrate nerve cells was obtained in bullfrog sympathetic neurons injected with the bioluminescent Ca2+ probe arsenazo III (597). Yet, the first proper description of caffeine-induced Ca2+ release was published in 1984, when Ian Neering and Robert McBurney presented the results of their experiments on DRG neurons, which were isolated from newborn rats and injected with another luminescent Ca2+ probe, aequorin (452). When 10 mM caffeine was applied to DRG neurons, they responded by a large light transient, reflecting an elevation of [Ca2+]i (Fig. 3A). In fact, the caffeine-induced [Ca2+]i elevation was three to five times larger than the [Ca2+]i increase evoked by action potentials (the latter artificially prolonged by addition of 20 mM TEA). The caffeine-induced [Ca2+]i response was preserved in an extracellular solution containing 5 mM Co2+, suggesting that it was independent of plasmalemmal Ca2+ entry. Furthermore, Neering and McBurney (452) showed that caffeine depletes the Ca2+ store, which, however, can be rapidly replenished by stimulating the neuron with several action potentials (APs) (Fig. 3B). Further progress in studying caffeine-induced Ca2+ release was very much assisted by the introduction of fluorescent Ca2+ probes (653, 654) as well as by the widespread use of the patch-clamp technique (200). When combined, these techniques allowed simultaneous monitoring of membrane currents and [Ca2+]i dynamics. In 1988, caffeine-induced Ca2+ release was identified in fura 2-loaded sympathetic neurons of the frog (348, 349) and in sympathetic (636), sensory (637), and central (638) neurons of the rat. Within the next several years caffeine-induced Ca2+ release was discovered in all types of neurons studied regardless of the preparation; the phenomenon was found in freshly dissociated neurons, neurons in primary cultures, and neurons in situ in brain slices (see Table 1).
As a drug aimed at an intracellular target, caffeine is rather convenient as it freely diffuses through the plasma membrane, and its wash-in and wash-out kinetics are quite similar. The process of caffeine entering and exiting the cell can be directly monitored by exploring the ability of caffeine to quench the fluorescence of various Ca2+ probes in a wavelength-independent manner (471, 661, 662). This quenching is most prominent for indo 1, which allows real time recordings of intracellular caffeine concentration simultaneously with [Ca2+]i measurements.
The properties of caffeine-induced Ca2+ release are generally similar between different neurons (Table 1). Caffeine induces a [Ca2+]i rise without a requirement for extracellular calcium, and the [Ca2+]i elevation is not associated with plasmalemmal Ca2+ movements. Caffeine-evoked [Ca2+]i transients are sensitive to pharmacological modulators interacting with RyRs or with SERCA pumps; specifically, these [Ca2+]i responses are blocked by ryanodine, dantrolene, ruthenium red, and procaine and disappear after inhibition of ER Ca2+ uptake with TG or CPA (Table 1). Caffeine responses are found in all parts of neurons, from soma to dendrites and presynaptic terminals. Caffeine-induced Ca2+ release that is sensitive to TG and ryanodine has even been observed in individual spines of cultured hippocampal neurons, which are rich in ryanodine receptors (303).
Incubation of neurons isolated from bullfrog sympathetic ganglia with 10 mM caffeine frequently induces [Ca2+]i oscillations (e.g., Refs. 164, 166, 462), which are often quite diverse in different neurons, ranging from high-amplitude regular oscillatory activity to low-amplitude decaying [Ca2+]i fluctuations (111). This diversity of oscillatory patterns reflects complex mechanisms responsible for their generation, which involve ER (Ca2+-induced Ca2+ release and SERCA-dependent Ca2+ uptake, etc.) as well as non-ER (mitochondrial Ca2+ accumulation, plasmalemmal Ca2+ extrusion) cascades.
When using caffeine in nerve cell preparations, a certain caution is needed, because caffeine (and other methylxanthines) directly affects potassium channels (660) and effectively inhibits delayed rectifier and A-type potassium currents in vertebrate neurons (529). In certain types of neurons, caffeine may also activate plasmalemmal Ca2+ influx, for example, as demonstrated in acutely dissociated rabbit nodose ganglion neurons (229) and leech mechanosensitive neurons (565). The KD values for the activation of ER Ca2+ release and plasmalemmal Ca2+ influx by caffeine were close (1.5 and 0.6 mM, respectively); this needs to be taken into account when interpreting caffeine-induced [Ca2+]i responses as a sole consequence of ER Ca2+ release.
2. The “quantal” nature of caffeine-induced Ca2+ release
Caffeine releases Ca2+ from neuronal ER in a rather peculiar manner, which has been termed “quantal” release. When the cell is challenged with a low concentration of caffeine (∼2–5 mM), a transient Ca2+ response is triggered, and even if caffeine remains in the system, the [Ca2+]i recovers back toward the prestimulated level. Applications of caffeine in submaximal concentrations fail to empty the store completely; subsequent application of a higher caffeine concentration induces a further [Ca2+]i elevation. Interestingly, it appears that a submaximal caffeine concentration can empty the store “compartment” sensitive to this concentration: when 2 mM of caffeine is applied repetitively, the second application no longer produces a [Ca2+]i response (Fig. 4). This peculiar quantal release was found when studying [Ca2+]i dynamics in cultured chromaffin cells (92), in rat (661) and mouse (584) DRG neurons, and in myenteric neurons (286). A similar quantal release was also found when monitoring [Ca2+]L in chromaffin cells expressing ER-targeted aequorin (9; see Fig. 4A). These observations can be explained by assuming either that different neuronal Ca2+ release channels have different sensitivities to caffeine, or that the ER is composed of several compartments, which are discharged by different concentrations of caffeine (92). Further experiments showed, however, that the quantal caffeine-induced release is controlled by intraluminal Ca2+ concentration, and overcharging the stores with Ca2+ following massive depolarization-induced Ca2+ entry significantly enhances the sensitivity of ER Ca2+ release to caffeine (Fig. 4D, Refs. 300, 584). In the context of ER Ca2+ release, [Ca2+]L-dependent regulation of RyRs appears to be a general phenomenon, as the same sort of regulation is well established for another Ca2+ release channel, the InsP3R. Similar to RyRs, elevation of luminal Ca2+ increases the sensitivity of InsP3R to InsP3 (419, 465). Changes in “luminal” [Ca2+] in a planar lipid system directly influenced the purified InsP3R by affecting their inactivation (643).
3. Cytosolic Ca2+ regulates caffeine-induced Ca2+ release
The nature of interaction of caffeine with RyRs depends on the concentration of the former: direct electrophysiological investigations of isolated ryanodine receptor channels have demonstrated that low (<2 mM) concentrations of caffeine sensitize the channel to extraluminal (i.e., cytoplasmic) calcium, shifting RyR activation towards submicromolar Ca2+ concentrations. In the presence of low millimolar caffeine concentrations, the RyR channel open probability (Po) was significantly increased mostly due to a shortening of the lifetime of the closed state. At higher concentrations (>5 mM), caffeine opens RyRs even at picomolar [Ca2+]i and dramatically increases the open time of the channel (e.g., Refs. 221, 547, 593).
These biophysical properties of the caffeine-RyR interactions are also reflected by the [Ca2+]i dependence of caffeine action on nerve cells. The amplitude of the [Ca2+]i elevation induced by 10 mM caffeine increased sharply when the [Ca2+]i preceding the application of the drug was elevated by a conditioning KCl depolarization of rat sympathetic neurons (221). By changing the duration of the KCl application between 0.3 and 7 s, the conditioning [Ca2+]i rise could be graded, therefore making it possible to characterize the relation between [Ca2+]i and the amplitude of caffeine-induced Ca2+ release (Fig. 5A). This relation appeared bell-shaped: an increase of [Ca2+]i from the resting level of ∼50–70 to 100–300 nM markedly potentiated caffeine-induced [Ca2+]i response, but a further increase of [Ca2+]i to >400 nM led to an inhibition of the caffeine-evoked [Ca2+]i elevation. A similar effect of [Ca2+]i on the response to a submaximal (2 mM) caffeine challenge was observed in rat sensory neurons (Fig. 5B), where an increase in [Ca2+]i preceding caffeine application greatly potentiated Ca2+ release (584). Moreover, when precaffeine [Ca2+]i reached a level of ∼350 nM, 2 mM caffeine induced a full-blown Ca2+ release that almost completely emptied the ER.
An even more striking dependence of caffeine responses on precaffeine [Ca2+]i was observed in central neurons. In Purkinje cells studied in rat cerebellar slices, application of 20 mM caffeine to the resting cell (resting [Ca2+]i ∼20–40 nM) rarely resulted in Ca2+ release. Yet, when [Ca2+]i was slightly elevated by moderately depolarizing the cell (experiments were done under voltage-clamp) caffeine produced robust [Ca2+]i elevations (Fig. 6). The amplitude of the caffeine-induced [Ca2+]i response was linearly dependent on the preceding [Ca2+]i level (269). Similarly, in hippocampal neurons, an elevation of precaffeine [Ca2+]i from ∼40 to 284 nM led to a 3.5-fold increase of the caffeine-induced [Ca2+]i response (179).
4. Physiological CICR
A) THE ER AS A SOURCE OF AND SINK FOR CYTOSOLIC CA2+.
Although the ability of the ER to buffer [Ca2+]i and at the same time to release Ca2+ into the cytosol was recognized already by the mid 1980s (213, 214, 386), the concept of the ER Ca2+ store acting as a “Ca2+ sink and Ca2+ source” was for the first time clearly formulated by David Friel and Richard Tsien in 1992 (165). They backed up this concept with a rather thorough analysis of changes in depolarization-induced Ca2+ transients occurring upon various manipulations of the ER Ca2+ store. First, they convincingly demonstrated that ER Ca2+ release participates in the delivery of Ca2+ to the cytosol during cell depolarization, i.e., that CICR indeed acts as an amplifier of Ca2+ signals resulting from plasmalemmal Ca2+ entry. They employed a protocol, which later became a generally accepted test for CICR. This protocol compared depolarization-induced [Ca2+]i transients in control conditions and in conditions when 1) ER is recovering following treatment with a high (10–20 mM) concentration of caffeine, 2) RyRs are sensitized by a low (1 mM) concentration of caffeine, and 3) RyRs are activated by 1 μM ryanodine and the stores are permanently depleted. In perfect concordance with the idea of the ER having the dual role of [Ca2+]i buffer and Ca2+ source, these manipulations affected both the rising phase and the recovery of [Ca2+]i elevations in response to membrane depolarization with 30–50 mM KCl (Figs. 7 and 8). A complete depletion of the stores decreased the amplitude and rate of rise of KCl-induced [Ca2+]i transients, and accelerated their recovery by 1) removing the CICR contribution and 2) upregulating Ca2+ uptake into the depleted ER. Sensitization of RyRs by 1 mM caffeine increased the rate of rise and the amplitude of the depolarization-induced [Ca2+]i transient, and moderately accelerated the recovery of the [Ca2+]i elevation by 1) enhancing the CICR contribution and 2) augmenting Ca2+ uptake because increased CICR depleted the store to a higher extent compared with control. Finally, permanent emptying of the ER by 1 μM ryanodine decelerated the KCl-induced [Ca2+]i transient and slowed down the recovery by 1) removing the CICR contribution and 2) occluding the ER Ca2+ uptake, because the store remained permanently leaky due to constantly open RyRs.
The concept of “sink and source” quite adequately explains CICR behavior in nerve cells, and, most importantly, it emphasizes the role of the intra-ER Ca2+ content, which appears to be an essential regulator of ER function, controlling the mode (source or sink) of Ca2+ store operation. The relatively low ER Ca2+ content would therefore promote [Ca2+]i buffering as for example happens in hypoglossal motoneurons, where depletion of the store by caffeine prolonged the recovery, but did not alter the amplitude of AP-evoked [Ca2+]i transients (127).
B) EVIDENCE FOR PHYSIOLOGICAL CICR IN NEURONS.
With the use of the paradigm described above, the contribution of CICR to the [Ca2+]i elevation triggered by plasmalemmal Ca2+ entry has been demonstrated for many types of neurons, in both the peripheral and the central nervous system (Table 2). In addition to the protocols used by Friel and Tsien, several other experimental approaches were developed, e.g., CICR can be completely blocked by high concentrations of ryanodine, or the ER store can be depleted using the SERCA pump blockers TG or CPA. The changes observed in the depolarization-induced [Ca2+]i transients caused by these interventions are very much as expected from the sink and source theory: inhibition of CICR by ryanodine slows down the rate of rise and decreases the amplitude of KCl-induced [Ca2+]i transients, whereas inhibition of ER Ca2+ uptake has the same effect on the onset of [Ca2+]i signals in response to depolarization and significantly decelerates their recovery.
With the use of these methods, the contribution of CICR to depolarization-evoked [Ca2+]i signaling was discovered in many, but not in all, neurons (e.g., it was absent in cerebellar granule neurons, Refs. 250, 292). Furthermore, in experiments on brain slices with preserved synaptic inputs, a very significant CICR was recorded in response to synaptic stimulation in dendrites (Fig. 8B) and even spines of hippocampal neurons (6, 138). In these cases the trigger Ca2+ entered the cytosol via Ca2+-permeable NMDA receptors.
It is worth pointing out that the functionality of the CICR pathway may critically depend on the experimental conditions. For instance, in Purkinje neurons, the parameters of ER Ca2+ release are greatly affected by cell culture. Photorelease of InsP3 or administration of caffeine induced only a tiny [Ca2+]i response in a small fraction of cultured Purkinje neurons, whereas the same manipulations resulted in large and frequent [Ca2+]i responses in freshly dissociated neurons (697) (however a caveat: the cultures were prepared from the embryonic cerebellum, E16-E18, whereas neurons were acutely isolated from mice at postnatal days 10–16). Conversely, no CICR in response to a single AP or trains of APs (tetanic stimulation) was observed in bullfrog sympathetic neurons, when studied in whole ganglia (462), but a prominent CICR in response to Ca2+ entry was invariably observed when the same cells were kept in culture conditions (240).
In certain cell types, physiological CICR exhibits a remarkable spatial heterogeneity. In leech Retzius neurons, for instance, depletion of the ER by 10 μM CPA reduced AP-induced [Ca2+]i transients in dendrites, but not in axons (33). In frog sympathetic ganglion neurons, either within the whole ganglia or in culture, a single AP was able to trigger several local CICR events concentrated in the ER-rich perinuclear region or around axon hillocks, as was revealed by fast fluo 4 confocal imaging (112).
At the subcellular level, activation of CICR is represented by local release events, resulting from spontaneous activation of RyRs and subsequent recruitment of RyR clusters. These elementary [Ca2+]i signals [observed in nerve growth factor-differentiated PC12 cells and in cultured hippocampal neurons (298, 398)] are analogs to “sparks” in muscle cell preparations. Neuronal [Ca2+]i sparks occur spontaneously, and their frequency increases upon addition of low concentrations of caffeine or following loading of the ER store by conditioning depolarization-induced Ca2+ entry. In contrast to other preparations, neuronal release sites appear to contain both InsP3Rs and RyRs, since metabotropic stimulation affected elementary Ca2+ release frequency at the same sites as caffeine (298). Spontaneous [Ca2+]i sparks were also found in presynaptic boutons of the lizard neuromuscular junction (398) and in cerebellar basket neurons (105, 355).
More accurate characterization of CICR in nerve cells requires a direct comparison between plasmalemmal Ca2+ entry and [Ca2+]i elevation. This necessitates simultaneous monitoring of Ca2+ currents (under voltage-clamp) and [Ca2+]i; the amount of Ca2+ entering the cell can be precisely calculated by integrating Ca2+ current (ICa) over time, and hence the Δ[Ca2+]i can be directly related to ∫ICa/Δt. Such an experiment was performed for the first time in rat sympathetic neurons (636), revealing a fairly linear relation between Ca2+ entry and Δ[Ca2+]i at short test pulses and its saturation at long test pulses (range of ICa durations varied between 10 and 640 ms). When the same experiment was repeated in the presence of 10 mM caffeine and 10 μM ryanodine, suppression of Δ[Ca2+]i was detected; however, this phenomenon was not very robust, leading the authors to conclude that CICR involvement in amplification of voltage-gated [Ca2+]i transients is rather modest, if it exists at all. Several years later, Shao-Yng Hua, Mitsuo Nohmi, and Kenji Kuba (240) added a degree of sophistication to this protocol by inventing a “unitary Ca2+ transient” (UT), which legitimized the Δ[Ca2+]i/∫ICa/Δt equation being in essence a measure of how many nanomolar Ca2+ will appear in the cytoplasm per nanoCoulombs of Ca2+-carried current entering the cell. If the transmembrane calcium current would be the only source for the [Ca2+]i increase, the UT should be constant irrespective of the duration or amplitude of ICa. An increase in UT thus reflects either the occurrence of CICR that delivers additional Ca2+, or saturation of cytoplasmic Ca2+ buffers and/or fast Ca2+ extrusion systems. Several groups, starting with Hua et al. have measured UT or else quantified the relation between Ca2+ entry and Δ[Ca2+]i in peripheral (240, 406, 585, 673) and central (68, 353) neurons, and all groups obtained virtually identical results (Table 2). When Ca2+ entry was graded by lengthening the depolarization step, the UT would eventually begin to increase, indicating a supralinear relation between Ca2+ entry and Δ[Ca2+]i (Fig. 9); the UT was similarly increased when Ca2+ entry was graded by increases in extracellular free Ca2+ concentration ([Ca2+]o) or by varying the depolarization amplitude (Figs. 9 and 10). Pharmacological inhibition of RyRs, or emptying the ER by preincubation with 20 mM caffeine, would depress UT and remove the supralinear relation between Δ[Ca2+]i and Ca2+ entry. Conversely, sensitization of RyRs by 1 mM caffeine results in an increase of UT (i.e., in increase of Δ[Ca2+]i in response to the same ICa; Fig. 10). These data strongly imply the existence of physiological CICR in nerve cells.
The final proof that CICR can be activated by physiological Ca2+ entry came from direct monitoring of [Ca2+]L in neural cells (the technique will be discussed in more details below). When performing these [Ca2+]L recordings in chromaffin cells (9) and in sensory neurons (603), it was demonstrated that cell depolarization with KCl in the former case, or activation of ICa under voltage-clamp conditions in the latter, resulted in a clear decrease in the [Ca2+]L, which reflected Ca2+ efflux from the ER lumen (Fig. 11). Furthermore, these [Ca2+]L transients can be enhanced by low caffeine concentrations and fully inhibited by ryanodine, thus unequivocally linking the [Ca2+]L decrease to the activation of RyRs.
C) NEURONAL CICR IS GRADED BY CA2+ ENTRY.
The activation of neuronal CICR proportionally follows the plasmalemmal Ca2+ entry, as was shown in experiments correlating ∫ICa/Δt and [Ca2+]i as described above. The “grading” of CICR by plasmalemmal Ca2+ influx was demonstrated even more directly in experiments where ICa, [Ca2+]i, and [Ca2+]L were simultaneously monitored (Fig. 12, Ref. 603). The relationship between ∫ICa/Δt and the magnitude of the transient [Ca2+]L decrease activated by Ca2+ entry was perfectly linear for ICa of increasing (from 50 ms to 3 s) durations. When the “unitary release potency” of ICa (defined as the amount of charge translocated necessary to decrease [Ca2+]L by 1 μM) was calculated, it remained the same for Ca2+ currents of different durations. This means that a small Ca2+ entry may generate only a moderate CICR, which may not significantly affect [Ca2+]i. This most likely explains the apparent threshold for CICR observed in unitary [Ca2+]i transient measurements (68, 240, 353, 406, 585).
The sensitivity of CICR to Ca2+ entry differs very significantly between different neuronal types. For example, in sensory neurons, the contribution of CICR (determined as the ryanodine-sensitive component of the [Ca2+]i transient) during physiological electrical activity became apparent only when >50 APs were generated in close succession (20-Hz frequency) (585). Likewise, in thalamocortical neurons, the contribution of CICR began to materialize only when bursts of APs exceeded 25 spikes with inter-AP interval of 7.5 ms (68). Conversely, in myenteric neurons, every single AP triggered a significant CICR, which, however, remained graded because each AP induced the same amount of Ca2+ release (227).
Nonetheless, in certain conditions, and in certain cells, neuronal CICR may acquire a regenerative character, when after reaching a threshold the release develops in an all-or-none fashion. A particularly strong CICR in response to a single AP was observed in nodose neurons, where Ca2+ entry produced by a single AP was amplified 5- to 10-fold by CICR, suggesting a regenerative process. In fact, the inhibition of CICR in these cells resulted in complete disappearance of [Ca2+]i elevations in response to less than eight APs (103). In sensory neurons regenerative CICR was also reported, although it required pharmacological sensitization of RyRs. In the presence of 5 mM caffeine, suprathreshold [Ca2+]i elevations produced by Ca2+ entry through plasmalemmal channels invariably triggered all-or-none CICR (663). Conversely, when the same protocol was applied to hippocampal neurons in slices, it did not reveal any regenerative CICR (446).
In fact, CICR is tightly regulated by the amount of trigger Ca2+ entering the cell and the rate of rise of [Ca2+]i in the close neighborhood of RyRs (222). In the presence of caffeine, the rise of [Ca2+] in the vicinity of RyRs is fast enough to trigger a regenerative response, which, by recruiting adjacent RyRs (already sensitized by caffeine), fully depletes the store. In the absence of caffeine, [Ca2+]i rises are slower and inactivation of RyRs prevents the occurrence of all-or-none CICR. One also has to remember the [Ca2+]i-sequestering ability of the ER, which balances the release, and sometimes may even overcome the latter. Indeed, as was elegantly shown by David Friel and colleagues (4, 234), the ER may switch between net Ca2+ release and net Ca2+ accumulation depending on the amount of Ca2+ entering the cell.
The mechanism that allows smoothly graded CICR in nerve cells is not entirely clear, although the very same phenomenon has been known for a long time from experiments on cardiomyocytes (149). Several theories (e.g., RyR adaptation, Ca2+-dependent inactivation of RyRs, [Ca2+]L depletion, [Ca2+]L regulation of RyRs, etc.) have been proposed to explain the inhibition of autocatalytic CICR (149, 326, 595); most likely a combination of many if not all of these factors is responsible for Ca2+ entry-dependent grading of Ca2+ release.
5. Depolarization-induced CICR in nerve cells
Although many neurons express the “skeletal” RyR1, identification of direct depolarization-induced Ca2+ release, which does not involve plasmalemmal Ca2+ entry, has long been missing. Some recently published data indicate that direct coupling between the plasmalemmal and ER Ca2+ channels in neurons may indeed exist. At the molecular level, for instance, direct coupling between the Ca2+ channels Cav1.2 or Cav1.3 and RyR1 was identified in solubilized rat brain membrane preparations (433). Some indirect evidence favoring the expression of depolarization-induced CICR in axons was presented by Ouardouz et al. (473), who demonstrated that inhibition of the voltage sensor of Ca2+ channel provides the same degree of protection against ischemic axonal damage as inhibition of RyRs. These authors also observed coimmunoprecipitation (suggestive of a direct physical link) of RyR1 and Cav1.2 and of RyR2 and Cav1.3, both in the whole brain and in the dorsal spinal cord column (such combinations are quite different from skeletal muscle where RyR1 is linked to Cav1.1). Finally, elementary Ca2+ release events, which occurred in the absence of Ca2+ entry but were potentiated by depolarization, were identified in isolated nerve terminals of magnocellular hypothalamic neurons (116). The localized [Ca2+]i transients that resulted from these elementary Ca2+ releases were named Ca2+ scintillas (a spark or a glimmer in latin). They were not affected by the removal of extracellular Ca2+, they were sensitive to ryanodine, and their frequency increased upon depolarization.
6. Endogenous CICR modulators
A) CYCLIC ADP RIBOSE.
The effects of cyclic ADP ribose (cADPR) on Ca2+ homeostasis were initially discovered in sea urchin eggs microsomes where this pyridine nucleotide applied in low micromolar concentrations was found to trigger Ca2+ release (99, 333). The primary target of cADPR is the ryanodine receptor type 2, and cADPR triggers opening of RyR2 in various nonexcitable cells (176, 177). In neuron-related cells or in neuron-derived cell lines, the injection of cADPR was reported to induce Ca2+ release. Such direct cADPR-triggered CICR was observed in permeabilized bovine chromaffin cells (430), and PC12 pheochromocytoma cells (100), as well as in neuroblastoma × fibroblast hybrid cells injected with cADPR (251). An increase in [Ca2+]i due to Ca2+ release was also observed in buccal Aplysia neurons injected with cADPR (432); moreover, cADPR also potentiated the release of ACh from cholinergic synapses belonging to these neurons. Apart from affecting the ER, cADPR may exert several other effects not connected with RyRs at all, e.g., cADPR was reported to interact with M-type potassium channels by affecting the ACh-induced inhibition of IK(M) (61). It has to be noted also that cADPR competes with ATP for the same binding site on the RyR2 (592).
In mammalian neurons, cADPR usually does not induce Ca2+ release by itself, yet it potentiates CICR in response to plasmalemmal Ca2+ entry. In the majority of neurophysiological experiments, administration of cADPR into the nerve cells was not accompanied by acute changes in [Ca2+]i. Injection of cADPR via patch pipette into NG108–15 neuroblastoma cells did not affect resting [Ca2+]i, but greatly potentiated depolarization-induced [Ca2+]i elevation, the effect being, at least in part, blocked by ryanodine. Apart from facilitating CICR, in these experiments cADPR also potentiated Ca2+ entry via L-type channels (204). In another set of experiments on the same neuroblastoma NG108–15, the inclusion of 10 μM cADPR into the pipette solution potentiated [Ca2+]i transients evoked by Ca2+ currents and resulted in a strong nonlinearity between Ca2+ entry and [Ca2+]i transient amplitude (thus indicating potentiated CICR, Fig. 13A, Ref. 137). Similarly, intracellular administration of cADPR facilitated depolarization-induced [Ca2+]i transients in CA1 hippocampal pyramidal cells (612) and in thalamocortical neurons (68).
In bullfrog sympathetic neurons, the inclusion of cADPR into the intracellular solution enhances the amplitude of depolarization- and caffeine-induced [Ca2+]i elevations (241). Similar facilitation of CICR by cADPR was found in DRG neurons studied under whole cell voltage-clamp with simultaneous monitoring of [Ca2+]i using fura 2. The inclusion of cADPR into the dialyzing solution significantly enhanced the unitary Ca2+ transient, thus reflecting increased nonlinearity between Ca2+ entry through voltage-gated Ca2+ channels (VGCCs) and the amplitude of the resulting [Ca2+]i elevation (Fig. 13B, Ref. 580). The opposite results, however, were obtained by Usachev and Thayer (663), who did not observe any potentiation of CICR following intracellular dialysis of rat DRG neurons with 1–10 μM cADPR.
To act as a second messenger, production of cADPR must be controlled by relevant neurotransmitter-activated signaling cascades. Indeed, the activity of ADP-ribosyl cyclase may be regulated by metabotropic receptors, such as muscarinic cholinoreceptors (MChRs) or even glutamate receptors. Coupling between MChR and ADP-ribosyl cyclase was identified in the NG108–15 neuroblastoma, indicating that ACh may regulate cADPR production (225). The neurotransmitter-regulated cADPR-dependent Ca2+ release pathway was also implicated in glutamate-induced [Ca2+]i signaling in midbrain dopamine neurons: local activation of mGluRs triggered propagating Ca2+ waves that were mediated by both InsP3Rs and RyRs, and were blocked by concerted action of heparin and the cADPR antagonist 8-NH2-cADPR. Quite naturally, the idea of mGluRs controlling both phospholipase C (PLC) and ADP-ribosyl cyclase was born (429), although more direct data supporting such a suggestion would be very welcome.
In addition, the activity of ADP ribosyl cyclase may be controlled by the nitric oxide (NO)/cGMP/cGMP-dependent protein kinase cascade (224); NO and cGMP do enhance the formation of cADPR in homogenates of sea urchin eggs (176). Photorelease of cGMP from a caged precursor delivered into cultured DRG neurons via intracellular dialysis triggered a delayed Ca2+-dependent inward current in ∼50% of all neurons tested (514). This was interpreted as an indication of cADPR-induced Ca2+ release resulting from the activation of ADP ribosyl cyclase caused by cGMP-dependent phosphorylation. In support of this suggestion, the inclusion of the cADPR antagonist 8-NH2-cADPR into the pipette solution inhibited the effects of cGMP. Finally, there is some evidence that cADPR may be involved in assisting CICR necessary for LTP generation in Schaffer collaterals-CA1 neuron synapses in rat hippocampus (531).
All in all, cADPR remains a strong candidate as an endogenous modulator of CICR. In particular, neurotransmitter-regulated synthesis of cADPR in synaptic regions may be very important for local modulation of [Ca2+]i signals and hence relevant for synaptic plasticity.
B) NICOTINIC ACID ADENINE DINUCLEOTIDE PHOSPHATE.
Another nucleotide involved in the regulation of ER Ca2+ release in many types of nonexcitable cells is nicotinic acid adenine dinucleotide phosphate (NAADP). This molecule is believed to trigger Ca2+ release through specific, but so far unidentified, channels, the NAADPRs (75, 77, 96, 177, 374), although NAADP may also directly activate RyRs (182). In the neural tissue, the NAADP-induced Ca2+ release was described in brain microsomes (20) and in neurons from buccal ganglion of Aplysia californica (86). The existence and importance of NAADP-dependent signaling cascade in vertebrate neurons remains unknown.
Calexcitin (CE), a Ca2+ and GTP-binding protein, is a recently discovered endogenous modulator of neuronal RyRs and CICR. This protein was initially isolated from neurons of conditionally trained Hermissenda (451). CE binds Ca2+ by virtue of an EF Ca2+-binding domain and undergoes considerable and reversible conformational changes when [Ca2+] fluctuates between 0.1 and 10 μM (14). The physiological effects of CE are exerted through modulation of K+ channels and RyRs. CE binds to RyRs in a Ca2+-dependent manner (7, 453) and facilitates Ca2+ release; injection of CE into hippocampal neurons resulted in a slow-developing [Ca2+]i elevation, which originated from the ER (454). CE also stimulated Ca2+ release from squid optic lobe microsomes; this release was blocked by dantrolene and an anti-RyR antibody (454). CE is known to be phosphorylated upon memory consolidation, which raises the interesting possibility that RyRs are involved in this process.
D) VANILLOID RECEPTORS.
The effects of capsaicin on nerve cells are executed through the TRPV1/VR1 vanilloid receptor, which belongs to a broad family of TRP channels (35, 49). In cultured DRG neurons, capsaicin increased [Ca2+]i in a dose-dependent manner with ED50 ∼13.5 μM. This [Ca2+]i elevation was mimicked by the capsaicin analog resiniferatoxin (100 nM) and blocked by the VR1 receptor antagonist capsazepine (10 mM). Capsaicin-induced [Ca2+]i signaling did not require extracellular Ca2+ and disappeared after the ER was depleted by incubating cells with a mixture of 10 mM caffeine and 12 μM ryanodine (144). The [Ca2+]i transients evoked by capsaicin were also inhibited by 10 μM dantrolene or by 10 μM ruthenium red (144), as well as by 2 μM TG (271). Taken together, these data indicate that capsaicin stimulates CICR in sensory neurons.
A rather unusual suggestion of direct involvement of TRPV1/VR1 in Ca2+ release from the ER was made recently by Liu et al. (352), who confirmed that capsaicin, when applied in Ca2+-free media, produces [Ca2+]i elevation in small DRG neurons. In contrast to previous findings, however, this [Ca2+]i signaling pathway was not affected by TG but was sensitive to the TRPV1/VR1 receptor antagonist capsazepine. Morphological examination revealed a substantial overlap between polyclonal antibody (TRITC labeled) TRPV1/VR1 staining of small DRG neurons and staining of the same cells with ER-tracker or ER marker DiOC5 (352). This overlap in staining was considered to indicate specific localization of TRPV1/VR1 in the ER membrane. These data prompted the intriguing speculation that TRPV1/VR1s may act as ER-resident Ca2+ release channel. These authors also proposed that the vanilloid receptors act as store-operated channels (but this is disputed by Karai and co-workers, Ref. 271).
“It is a truth universally acknowledged that all intracellular reactions are sensitive to pH” wrote Roger Thomas (641),1 and RyR function as well as CICR in various cell types are sensitive to the concentration of protons (e.g., Refs. 128, 395). In neurons, acidosis inhibits CICR and increases ER Ca2+ content (641), whereas alkaline shifts stimulate the release of Ca2+ and/or inhibit reuptake of Ca2+ into the ER (696).
It is generally accepted that the action of extracellular glutamate on ER Ca2+ release is accomplished via activation of group I metabotropic glutamate receptors (mGluRs) coupled, through G protein pathways, to PLC and InsP3 metabolism. Data accumulated during the past decade, however, indicate the existence of an alternative route that involves direct interactions of mGluRs and RyRs (145). Initial evidence in support of this pathway was obtained in cell-attached patch-clamp experiments on cultured cerebellar granule cells (90). This study demonstrated mGluR-dependent activation of L-type VGCCs, which could not be mimicked by intracellular dialysis with InsP3, was not blocked by heparin, but was completely obliterated by 1 μM ryanodine and simulated by caffeine. Subsequently, it was found that ryanodine effectively (∼90%) inhibited 1-aminocyclopentane-trans-1,3-dicarboxylic acid (trans-ACPD)-induced [Ca2+]i elevation, while it was much less potent against [Ca2+]i signals triggered by the other metabotropic agonists carbachol and histamine (119). These data led to an interesting idea about direct functional coupling between mGluRs and RyRs, thought to be mediated by Homer proteins (145). This assembly may be even more complex and may involve voltage-gated Ca2+ channels and Ca2+-dependent potassium channels. Although extremely stimulating, this hypothesis still requires additional experimentation.
Further evidence for glutamate-mediated regulation of CICR was obtained on neurons from the avian nucleus magnocellularis, in which activation of mGluRs inhibited both CICR and InsP3-induced Ca2+ release (IICR). It is hypothesized that this inhibition provides a mechanism by which glutamate protects highly active neurons from calcium overload (275).
Neurotrophins, which include nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophins 3/4/5 (NT-3, -4, and -5), operate through an extended family of TrkA, -B, and -C receptors, tumor necrosis factor (TNF) receptors, and p75 and Fas receptors (89). Among these receptors TrkA and p75 are abundantly expressed in the brain, and their activation is important for neuronal physiological activity; for example, BDNF-dependent neuronal excitation, synaptic plasticity, and release of neurotransmitters have been reported (265, 340, 359, 685, 701). Several of these effects involve intracellular Ca2+ release mediated by RyRs. For instance, NGF induces release of glutamate from cultured cerebellar neurons through activation of p75 receptors and ryanodine-sensitive Ca2+ mobilization (464). Similar activation of glutamate release, sensitive to dantrolene and to store depletion by caffeine, was observed in cortical synaptosomes; in this case, NGF acted through TrkA receptors, and glutamate release was sensitive to heparin (527).
Another neurotrophin, NT-3, significantly potentiated the spontaneous neurotransmitter release in the neuromuscular junction; this action required both Ca2+ release via RyR/InsP3Rs and the activation of Ca2+/calmodulin-dependent kinase II (208). In pyramidal neurons from the visual cortex, NT-4 potentiated depolarization-induced [Ca2+]i elevations partly though enhancement of CICR (278).
Alternatively, neurotrophins may regulate CICR by affecting the expression of RyRs. For instance, NGF treatment of cultured adrenal chromaffin cells significantly elevated the expression of RyR2 and dramatically (>1,000% over 10 days) increased the amplitude of caffeine-induced [Ca2+]i responses (260). In cerebellar granule cells, the addition of 10 ng/ml BDNF from the first day in culture augmented the contribution of CICR to depolarization-induced [Ca2+]i transients by a yet unknown mechanism (408).
Treatment of cultured rat DRG neurons with TNF-α induced a Ca2+-dependent current in ∼70% of cells and triggered [Ca2+]i increase in ∼35% of cells tested; both the current and [Ca2+]i elevation were inhibited by TG (10–50 μM) and ryanodine (10 μM). The action of TNF-α was mimicked by the uncaging of sphingosine or by addition of 0.4 U/ml sphingomyelinase (515). In parallel, two subtypes of TNF-α receptors, TNFR1 and TNFR2, were identified in DRG neurons by immunohistochemical and RT-PCR techniques. Whether the activation of TNFRs represents a novel pathway for Ca2+ release mediated by sphingosine-1-phosphate through separate ER channels or through modulation of RyRs remains rather unclear (712).
NO, which acts as a rapidly diffusible gaseous messenger in many tissues including the brain, seems to be another endogenous regulator of RyRs (143). NO-induced activation of RyR-mediated Ca2+ release was observed in hippocampal CA1 neurons; this Ca2+ release was important for both CREB phosphorylation and LTP induction (360). NO-induced activation of Ca2+ release from a ryanodine-sensitive store was also implicated in hypoxic [Ca2+]i rises in brain slices (394). Similarly, NO-mediated ER Ca2+ release via RyRs is considered to be an important pathological step in interleukin (IL)-1β-induced pyrogenic effects (485).
C. IICR in Nerve Cells
1. Direct evidence for neuronal IICR
The initial suggestion of InsP3-induced release of Ca2+ in neurons arose from electrophysiological studies, which demonstrated that injection of InsP3 into various neurons or neuronlike cells [e.g., neuroblastoma × glioma cells (223), rat hippocampal neurons (387), rat dorsal raphe neurons (162), etc.] resulted in activation of Ca2+-dependent K+ currents. This suggestion was immediately confirmed by measuring [Ca2+]i (Fig. 14A, Ref. 151), which demonstrated directly that intracellular administration of InsP3 induces a [Ca2+]i elevation due to Ca2+ release from the intracellular compartments (Table 3).
Similar experiments, in which InsP3 was administered intracellularly by photorelease from a caged precursor or by InsP3 application to permeabilized cells, revealed relatively poor sensitivity of neuronal IICR to the agonist. In Purkinje neurons in situ, for example, even 80 μM InsP3 did not result in an obvious saturation of the [Ca2+]i response (Fig. 14B; Ref. 284). Likewise, the EC50 for InsP3-mediated decrease in [Ca2+]L determined in permeabilized embryonic Purkinje neurons loaded with Mag-fura 2 was 25.8 μM (169). In single permeabilized DRG neurons, maximal depletion of the ER was observed at InsP3 concentrations of ∼20 μM (602). These values are many times greater that effective InsP3 concentrations determined for purified cerebellar InsP3Rs (EC50 ∼1 μM; Refs. 45, 362) and InsP3R1 heterologously expressed in artificial systems (EC50 ∼0.2–2 μM; Refs. 133, 420). Similarly, in many nonexcitable cells, such as astrocytes, hepatocytes, exocrine cells, and vascular epithelium, the maximal IICR was observed at InsP3 concentrations not exceeding 5–10 μM (285); however, very high doses of InsP3 (up to 100 μM) were required to trigger maximal response in pancreatic acinar cells (681).
The reasons for low sensitivity of IICR in nerve cells are not quite clear. The effective InsP3 concentrations in the cytosol can be substantially reduced by an extensive InsP3 buffering, or else InsP3 can be quickly degraded by enzymatic systems. These two possibilities, however, are questioned by the experiments on permeabilized cells, where cytoplasm is essentially washed out together with the InsP3 buffers and InsP3-degrading enzymes. Alternatively, the InsP3 binding to the receptor may be inhibited by endogenous inositol trisphosphate receptor inhibitor (IRI), recently discovered in the cerebellum (686). The low sensitivity of neuronal IICR could be important for the maintenance of synaptic specificity by localization of InsP3-induced signaling.
2. Metabotropic receptors and neuronal IICR
Physiological activation of IICR requires stimulation of plasmalemmal receptors coupled to PLC, which in turn hydrolyzes phosphatidylinositol 4,5-bisphosphate and gives birth to InsP3. These plasmalemmal receptors, which usually share a common seven-transmembrane domains structure, are coupled to G proteins and are generally known as metabotropic receptors. Neurons express an impressive complement of metabotropic receptors, many of which are coupled to the InsP3 signaling cascade. Numerous observations have shown that activation of many different types of neuronal metabotropic receptors trigger IICR (Table 4).
The metabotropic receptors are activated upon synaptic transmission (Table 5, Fig. 15) and initiate IICR responsible for several distinct forms of [Ca2+]i responses, manifested either as propagating Ca2+ waves, discussed below, or in a form of highly localized [Ca2+]i elevations. The local [Ca2+]i microdomains are particularly important in cerebellum, where synaptic stimulation of mGluRs results in Ca2+ release strictly confined to postsynaptic spines or defined dendritic areas of the Purkinje neurons (150, 624). This localization of [Ca2+]i signaling seems to be an intrinsic property of the IICR mechanism, because focal photorelease of InsP3 triggered [Ca2+]i elevation which did not spread far away from the site of InsP3 appearance (150). This localization may reflect specific properties of the Purkinje neurons such as low sensitivity of IICR to InsP3; high level of cytosolic Ca2+ buffering; large densities of InsP3 receptors/buffers, which bind InsP3; speed of InsP3 degeneration, or the effects of IRI discussed in the previous section. This local IICR modulates the neighboring synapses by inducing long-term depression (LTD); the localized [Ca2+]i signaling can be important for spatial segregation of synaptic plasticity events.
A further degree of complexity is added by the fact that neuronal InsP3Rs can be directly linked to plasmalemmal mGluRs via Homer proteins (656). This physical link may have important functional consequences, because the expression of a modified form of Homer protein, lacking the ability to cross-link with InsP3R, reduces amplitude and speed of [Ca2+]i transients mediated by activation of mGluRs (656).
3. CICR-assisted IICR
An important property of InsP3 receptors and IICR is their regulation by [Ca2+]i. An increase in the latter facilitates IICR and may trigger Ca2+ release through InsP3Rs in the presence of a constant level of the second messenger. The synergism between [Ca2+]i and InsP3 may act as a coincidence detector allowing the system to identify the simultaneous activation of both signaling cascades; furthermore, pairing of InsP3 and [Ca2+]i elevations may be important in converting local signaling events into propagating waves of ER membrane excitation.
Indeed, IICR was found responsible for generation of propagating Ca2+ waves triggered by synaptic excitation of hippocampal pyramidal neurons. These Ca2+ waves were found almost exclusively in the thick apical shafts of CA1 neurons (445). The generation of the Ca2+ wave has a threshold nature, which most likely reflects InsP3 accumulation (717) and amplification of the release by CICR through InsP3Rs. This amplification can be seen directly. Pairing of mGluR stimulation (either by synaptic stimulation or by application of a metabotropic receptors agonist) with plasmalemmal Ca2+ entry, activated by back-propagating action potentials, greatly enhanced [Ca2+]i signals (Fig. 16, Refs. 444, 446). In fact, the degree of mutual potentiation was quite remarkable: a backpropagated AP by itself resulted in a fast low-amplitude [Ca2+]i increase, and the application of mGluR antagonists alone did almost nothing to [Ca2+]i, yet when an AP was generated in the presence of continuous mGluR stimulation, a large (up to 40 μM) [Ca2+]i increase was produced (446). Similarly, [Ca2+]i waves could be induced in response to electrical stimulation of mossy fibers, which form synaptic input to CA3 hippocampal neurons. These waves were inhibited by 1 μM heparin and were potentiated by pairing mossy fiber stimulation with AP-driven Ca2+ influx through VGCC (270). Comparable synaptically activated propagating [Ca2+]i waves resulting from IICR were also observed in neocortical pyramidal neurons from layers II/III and V. These [Ca2+]i waves were dependent on stimulation of metabotropic glutamate receptors and were blocked not only by intracellularly applied heparin, but also by 5–20 μM ryanodine, but not by ruthenium red (331). These results may indicate either an amplifying role of RyR-mediated CICR or depletion of a single store shared by RyRs and InsP3Rs.
Activation of muscarinic cholinoreceptors in hippocampal CA1 neurons by electrical stimulation of cholinergic synaptic inputs induced [Ca2+]i waves propagating from dendrites towards soma. These [Ca2+]i waves were potentiated by AP-induced Ca2+ entry, disappeared after inhibition of ER Ca2+ accumulation with TG (100 nM) or CPA (30 μM), and were blocked by intracellular administration of heparin (500 μg/ml), demonstrating the involvement of InsP3Rs (518). The CICR from InsP3Rs was also described in layer II/III neocortical neurons in which AP-induced [Ca2+]i transients were remarkably augmented upon stimulation of MChRs; the latter apparently provided an increase in InsP3, which sensitized InsP3Rs to Ca2+ entering through the plasmalemma (704). Most interestingly, this CICR via InsP3Rs was absent in very young (<P7) rats, although in these young neurons direct application of InsP3 resulted in IICR, and cell depolarization produced [Ca2+]i elevation of an appreciable amplitude (705). This developmental difference prompted the suggestion that during development, the plasmalemmal Ca2+ channels and InsP3Rs become closer spatially, allowing modulation of InsP3Rs by Ca2+ influx.
4. Endogenous IICR modulators
A) PROTEIN KINASES AND CALCINEURIN.
Endogenous regulation of InsP3Rs is mainly achieved through phosphorylation/dephosphoryaltion of the receptor molecule. Phosphorylation of the InsP3R, which results in enhanced channel activity, is catalyzed by protein kinases A, C, and G (412, 413) and by tyrosine kinase (258). The latter is also coupled to InsP3 metabolism and therefore may affect the IICR via several routes. Dephosphorylation of InsP3R and hence reduction of the channel activity is catalyzed by the Ca2+/calmodulin-dependent phosphatase calcineurin (73), which also regulates the expression of InsP3R at the transcriptional level, because pharmacological blockade of calcineurin abolished the expression of InsP3Rs in cerebellar granule neurons (181).
B) IMMUNOPHILIN FK506 BINDING PROTEIN 12.
Immunophilin FK506 binding protein 12 (FKBP12), which is abundantly expressed in brain tissue, was shown to be associated with InsP3R1. This complex is disrupted upon the addition of the immunodepressants FK506 or rapamycin. Dissociation of InsP3R1-FKBP12 causes an increase of Ca2+ flux through the channel, and this can be reversed by adding FKBP12 (74). It seems that the latter serves as a physiological regulator of the conductive properties of InsP3Rs. In biophysical experiments on purified InsP3R1s in planar lipid bilayers, the addition of exogenous recombinant FKBP12 substantially increased the open time of the channels (113).
The InsP3R type 1, but not type 3, possesses a high-affinity binding site for calmodulin, occupation of which inhibits the channel. Inhibition of calmodulin by calmodulin inhibitory peptide sensitizes IICR in Xenopus oocytes (376).
The InsP3Rs are also regulated by the cytoplasmic Ca2+-binding proteins caldendrins, which, depending on their structure, may either activate or inhibit IICR. A splice variant of caldendrin was found to bind specifically to all three InsP3R subtypes. This binding activated InsP3Rs, suggesting the possibility of IICR initiation even in the absence of the natural agonist (707). At the same time a long-splice variant of caldendrin, when expressed in PC12 or HeLa cells, inhibited IICR triggered by activation of histamine and purinoreceptors (206).
D. Do InsP3Rs and RyRs Share a Common Ca2+ Pool?
Although morphological and functional evidence (see sect. ii) clearly defined the ER as a single continuous space, the analysis of Ca2+ release divided the field into supporters of a single functional Ca2+ pool available for both CICR and IICR, and supporters of at least two separate pools, selectively sensitive to either InsP3 or Ca2+/caffeine (see Table 6). The experimental strategy employed by both camps is similar: the cells are treated with maximal concentrations of CICR agonists and thereafter the ability of IICR-stimulated agents to produce Ca2+ release is tested, and vice versa. Provided the cell has a single Ca2+ pool, each agonist should be able to deplete it completely, thus rendering subsequent stimulations with other agents ineffective. Alternatively, if the cell expresses more than one functional pool, each agonist would discharge a compartment with appropriate complement of receptors. The story is old and at the moment opposing factions are far from a consensus. As far as neurons are concerned, it looks as if more data are in favor of a single continuous ER Ca2+ store. Furthermore, some studies supporting separate Ca2+ pools mention peculiar unidirectional relations between RyR-bearing and InsP3Rs bearing stores: either ryanodine occluded IICR but not vice versa (541), or else stimulation of IICR depletes caffeine-sensitive pool but stimulation of RyRs does not affect the InsP3-sensitive one (528). Moreover, in most cases, the separate pools were observed in cell lines or in embryonic cultures, whereas the existence of a single Ca2+ store is supported by experiments on primary cultures and brain slices, the latter preparations being rather more physiological (Fig. 17A). When directly investigating intra-ER Ca2+ dynamics in sensory neurons, a very clear and full overlap between InsP3/ryanodine/caffeine/Ca2+ and TG/CPA pools was observed (Fig. 17B, Ref. 602).
E. Luminal Ca2+ as a Regulator of Ca2+ Release
The level of free Ca2+ concentration within the ER lumen is critically important for regulation of Ca2+ release as it creates the driving force for Ca2+ exit and regulates the functional availability of Ca2+ release channels. That [Ca2+]L in nerve cells is a labile parameter became obvious from the very beginning of experimental investigations of the ER Ca2+ store. The first attempts to induce Ca2+ release in single central neurons, either by caffeine or by metabotropic agonists, demonstrated that application of stimulating agents to resting cells triggered very tiny [Ca2+]i increases in a small proportion of neurons. However, when caffeine or a metabotropic agonist was applied shortly after the cell was depolarized, and thus had experienced a large Ca2+ influx into the cytosol, a considerable [Ca2+]i elevation due to Ca2+ release was apparent in a large proportion of neurons (249, 250, 375, 437, 526, 581). A small depolarization (i.e., with ∼10 mM KCl or with low concentrations of NMDA) which led to a steady-state [Ca2+]i increase had the same effect (249).
The stimulating effect of the conditioning depolarization was transient because the amplitude of Ca2+ release decreased progressively when increasing the time gap between the depolarization and the application of a Ca2+-release stimulator (Fig. 18). Quite logically, this peculiar behavior was explained in terms of a low releasable Ca2+ content of the ER in resting central neurons. The empty Ca2+ store favors Ca2+ accumulation and therefore Ca2+ entering the cell following depolarization is taken up by the ER, thus increasing the Ca2+ content within the store and permitting Ca2+ release. An extreme situation, in which the store is empty or almost empty at rest, is usually observed in cultured cells. Application of caffeine to resting neurons in acute hippocampal slices (179) or in acutely isolated thalamocortical neurons (68) routinely evokes [Ca2+]i responses; similarly stimulation of IICR triggered [Ca2+]i elevation in Purkinje neurons (624) and in pyramidal cortical neurons (609) in situ. Nonetheless, the conditioning depolarization dramatically (by 500%) increased the amplitude of a subsequent caffeine-induced [Ca2+]i transient in central neurons studied in acute slices (179); and similarly to what was found in cultured neurons this potentiation was short-lived (Fig. 18B). In peripheral neurons, the effect of conditioning depolarization also existed, although it was much less dramatic; conditioning depolarization increased the amplitude of subsequent Ca2+ release by 10–30% (583, 661). The nature of this rather prominent difference between central and peripheral neurons is unclear; similarly unclear are the mechanisms responsible for the loss of surplus Ca2+ in the ER after the store has been charged.
The ability of the ER of central neurons to rapidly accumulate huge amounts of Ca2+ was confirmed in direct recordings of total ER Ca2+ content using energy dispersive X-ray microanalysis. In these experiments, hippocampal slices were rapidly frozen at different time intervals after tetanic afferent stimulation. It was observed that 30 s after stimulation, the total ER Ca2+ content had increased up to 20 times and stayed elevated for 15–30 min (519, 520). The authors concluded that the ER acts as a major Ca2+ buffering system in dendrites upon synaptic stimulation. However, this may vary between different regions and may depend on heterogeneous distribution of SERCA pumps (365). Importantly, the X-ray microanalysis also showed a remarkable heterogeneity of dendritic ER with respect to Ca2+ accumulation; only ∼40% of the reticulum demonstrated an increase in total Ca2+ content. These data once more raise questions about the functional continuity of the ER, although a heterogeneous increase in total Ca2+ may reflect specific concentration of Ca2+ binding proteins in particular ER regions.
The transient Ca2+ loading of the Ca2+ store in central neurons certainly has very important implications. Indeed, a transient increase in [Ca2+]L and the consequent enhancement of Ca2+ release would provide the ER with a coincidence detection mechanism and with (at least short-term) memory. Further understanding of the mechanisms controlling changes in the ER Ca2+ concentration ultimately requires direct monitoring of the free Ca2+ dynamics in the ER lumen.
Several experimental approaches permitting real time measurements of [Ca2+]L have been developed during the last two decades (Table 7). These techniques employ either ER-targeted Ca2+-sensing luminescent or fluorescent proteins (10, 120) or conventional fluorescent Ca2+ probes with a sufficiently low Ca2+ affinity (232, 492, 601). The [Ca2+]L measurements in neurons began only recently (Table 7; Fig. 19), and our knowledge about ER Ca2+ dynamics in nerve cells remains very incomplete. The quantitative [Ca2+]L measurements performed in PC12 cells and isolated sensory neurons showed a high resting [Ca2+]L ranging between 100 and 800 μM and demonstrated that activation of Ca2+ release channels by Ca2+, caffeine, ryanodine, and InsP3 results in a rapid decrease of [Ca2+]L reflecting Ca2+ release. Direct [Ca2+]L measurements in physiologically intact central neurons have not yet been performed.
F. Refilling the Ca2+ Store and Store-Operated Ca2+ Entry
Control of [Ca2+]L and refilling of the Ca2+ store after depletion is accomplished by SERCA pumps. Numerous experiments (see Tables 1–5) have demonstrated that inhibition of the latter prevents Ca2+ accumulation and results in progressive depletion of the Ca2+ store due to unopposed leakage of Ca2+ from the ER lumen. Occasional reports mention the possible existence of a TG-resistant Ca2+ store in nerve cells, and a TG-resistant Ca2+ transport was found in brain microsomes and in certain cell lines (687). The relevance of this type of Ca2+ pool for neuronal physiology remains unclear.
Intraluminal Ca2+ recordings revealed an important role of the [Ca2+]L level in the regulation of the velocity of ER Ca2+ uptake. Decreases in [Ca2+]L significantly (5–7 times) increased the speed of Ca2+ uptake; replenishment of the ER slowed the rate of Ca2+ uptake (603). Most importantly, at least in sensory neurons, the influence of an increased [Ca2+]i on the Ca2+ uptake velocity seems to be negligible (Fig. 20). To quantify Ca2+ uptake velocity, Ca2+ release was stimulated by brief application of caffeine, and the recovery kinetics of [Ca2+]L were monitored during the washout period. After [Ca2+]L reached the prestimulation level, the SERCA pumps were inhibited by TG, which resulted in a relatively slow decrease in [Ca2+]L due to Ca2+ leakage. The rate of leakage declined in proportion to the decrease in [Ca2+]L, presumably reflecting a falling electrochemical driving force for Ca2+. It turned out that neither the rate of Ca2+ uptake nor the velocity of Ca2+ leakage was affected when a BAPTA/Ca2+ (10 mM/2 mM) buffer was washed into the cytosol (Fig. 20, A and B). In the latter condition [Ca2+]i is clamped, yet plenty of Ca2+ is available for ER accumulation. Therefore, in peripheral neurons store refilling does not require an increase in [Ca2+]i, albeit it requires Ca2+ availability in the cytosol. This seems to be a general mechanism for many cell types, and the very similar dependence of ER Ca2+ uptake on [Ca2+]L and independence on [Ca2+]i rises, was shown for pancreatic acinar cells (425; see Fig. 20C) and BHK-21 fibroblasts (231).
In frog sympathetic ganglion neurons, a special mechanism which greatly assists the refilling of the stores after their depletion, the release-activated calcium transport (RACT) was also suggested (110). The nature of this pathway remains enigmatic, and it is possible that it simply reflects greater activation of SERCA pumping following severe depletion of the ER (although according to the initial description RACT appeared to be insensitive to TG).
All in all, the regulation of [Ca2+]L in peripheral neurons seems to be quite straightforward; these stores are full in the resting conditions, and an efficient [Ca2+]L dependent regulation of SERCA pumping keeps [Ca2+]L under tight control. This control is assisted by a [Ca2+]L-dependent leakage such that every increase in [Ca2+]L immediately increases the Ca2+ leak and decreases SERCA activity. Store depletion leads to the opposite effect, i.e., a decrease in Ca2+ leak and rapid upregulation of SERCA pumping velocity. Intuitively, the mechanism controlling [Ca2+]L in central neurons must be very different as it should permit a high flexibility of the [Ca2+]L regulation to tolerate the relatively low resting ER Ca2+ content. Nevertheless, this low resting [Ca2+]L seems to prime SERCAs for rapid accumulation of the bulk of the Ca2+ entering the cytosol upon depolarization or synaptic stimulation. This also assumes the high sensitivity of the Ca2+ uptake mechanism to increases in [Ca2+]i, which represents another clear difference between central and peripheral nerve cells.
The problem of store refilling is also inseparable from the question of neuronal expression of the store-operated Ca2+ entry pathway. This pathway, discovered by Jim Putney (522), is a general feature of nonexcitable cells, which lack other means of Ca2+ entry. A direct relation between ER depletion and SOC activation is firmly established (458, 488, 490, 523). In nerve cells, the replenishment of the Ca2+ store after depletion by either several consecutive applications of caffeine/metabotropic agonists or prolonged incubation of cells with Ca2+ release stimulating agents, requires plasmalemmal Ca2+ influx (179, 661). Does this Ca2+ influx employ a classical store-operated mechanism? Or, does the Ca2+ enter the cell via numerous Ca2+-permeable channels present in the neuronal plasmalemma? This remains a debatable matter. From the general logic of neuronal physiology, the SOC pathway can hardly be justified. Indeed, neurons possess plenty of Ca2+ channels, and moreover, these channels are repeatedly activated, as neurons rarely enjoy complete idleness. Furthermore, activation of SOC, according to several reports, has a threshold nature, i.e., it requires a very substantial depletion of the store (202, 489, 490). This is very difficult to reconcile with the observations of empty stores in central neurons; if SOC exists it surely would increase [Ca2+]i up to the point when the stores become replenished.
Nonetheless, several groups have suggested that neurons possess a store-operated Ca2+ influx pathway very similar to that of nonexcitable cells (Table 8, Ref. 524), yet the data gathered are not entirely convincing. Quite often SOC expression is judged upon pharmacological sensitivity of [Ca2+]i transients produced by Ca2+ readmission after store depletion in Ca2+-free media. Because the pharmacology of store-operated channels is not very well defined, this interpretation is open to criticism. The existence of SOC in central neurons, for example, was postulated by Baba et al. (17) based on [Ca2+]i imaging of primary cultured pyramidal and granule neurons isolated from Ammon’s horn and dentate gyrus, respectively. After these neurons were incubated with 1 μM TG for 5 min in Ca2+-free solution, readmission of external Ca2+ triggered [Ca2+]i elevation, which was inhibited by 100 μM La2+, 3 μM SKF96365, or 30 μM 2-aminoethoxydiphenyl borate (2-APB). No attempts, however, were made to activate SOC by more physiological depletion of the ER store such as administration of caffeine or stimulation of metabotropic receptors, yet a conclusion about SOC activation by synaptic transmission (based on partial and rather slight inhibition of [Ca2+]i transients evoked by tetanic field stimulation of cultured neurons by 2-APB or SKF96365) was announced. As 2-APB and SKF96365 also attenuated LTP recorded in hippocampal slices, another fundamental conclusion of SOC involvement in synaptic plasticity was made by the same authors (17). Interestingly, La2+, 2-APB, and SKF96365 also inhibited the plateau phase of NMDA-induced [Ca2+]i elevation, which may reflect their poor selectivity, rather than activation of SOC upon stimulation of ionotropic glutamate receptors.
The protocols of inducing SOC in neurons vary, e.g., cultured cortical neurons were incubated in a Ca2+-free, 50 μM EGTA and 2 μM CPA containing solution for 30 min, and [Ca2+]i elevation developing after external Ca2+ restoration was regarded as a SOC-related phenomenon (709), yet the possibility that prolonged treatment with a Ca2+-free solution may affect neuronal status and viability was not explored. Sometimes the existence of SOC, as well as its physiological significance, is deduced from indirect data such as an increase in miniature excitatory postsynaptic potential frequencies in Ca2+-containing solutions after treatment with CPA (139). These observations favoring neuronal SOC should be balanced by very many experiments, that employed TG/CPA for inhibition of Ca2+ release in nerve cells. In most of these cases, no significant changes in resting [Ca2+]i, indicative of SOC were observed (e.g., Refs. 179, 303, 418, 582, 583, 629, 655 to name but a few). All in all, the evidence for neuronal SOC derived from pharmacological inhibition of post-TG/CPA-induced [Ca2+]i elevation remains, in my view, unconvincing, and conjectures about the role of SOC in synaptic plasticity, though very stimulating, currently lack real experimental basis.
Perhaps the most convincing case for SOC expression in neurons was presented by Usachev and Thayer, who investigated the process of store replenishment in rat cultured DRG neurons. They demonstrated that Ca2+ entry, essential for the refilling of ER stores after they have been discharged with caffeine, can be blocked by 2 mM Ni2+, and the refilling of the store is potentiated when the cells are hyperpolarized from −55 mV to −80 mV. Furthermore, Usachev and Thayer (664) substantiated their claim for SOC in nerve cells by demonstrating that ER store depletion potentiates Mn2+ entry into neurons. Still, these data may be explained by a very different mechanism, i.e., Ca2+ entry through T-type Ca2+ channels, which have a “window” current region around −75 to −80 mV, where they may provide for unstimulated Ca2+ influx. Calcium entry through conventional Ca2+ channels following ER depletion was, for instance, observed in cultured guinea pig myenteric neurons, in which the treatment with TG (1 μM) induced long-lasting irregular [Ca2+]i oscillations arising solely from Ca2+ influx, yet these oscillations were completely blocked by ω-conotoxin MVIIA, suggesting that these neurons may use N-type Ca2+ channels for store refilling (665). In the same neurons, application of ATP triggered a biphasic (peak − plateau) [Ca2+]i response; the plateau required extracellular Ca2+ but was blocked by 10 μM nifedipine (287). Likewise, the plasmalemmal voltage-gated Ca2+ channels were responsible for ER store refilling in cultured cerebellar granule cells (249) as well as in cultured motoneurons from larval lampreys (280).
G. Pharmacology of Ca2+ Release
1. Ryanodine receptors/CICR
Mechanisms of pharmacological modulation of RyRs have received much attention, and the properties of drugs that selectively modulate RyRs have been summarized in numerous reviews (e.g., Refs. 135, 216, 217, 259, 308, 310, 483, 614, 703, 720). Therefore, I shall limit myself to a brief discussion of some of these tools, which are instrumental for studying CICR in nerve cells.
Ryanodine was probably the first and the foremost pharmacological agent used for probing for Ca2+ stores and Ca2+ release. Its action on skeletal and cardiac muscle had been known for the last 35 years (259), and its ability to specifically bind to Ca2+-gated Ca2+ release channels led to naming the latter “ryanodine receptor” (505) a little confusing (as no ryanodine was ever in attendance in living tissues, save in roots and stem of the plant Ryania speciosa, which is the natural source of ryanodine), but generally accepted. Ryanodine exerts multiple actions on the RyRs, depending on its concentration. At very low concentrations (5–40 nM), ryanodine was reported to increase the frequency of RyR channel opening, without affecting its conductance (67). At low micromolar (1 to ∼5 μM) concentrations, ryanodine promotes RyR channel opening in a subconducting (∼40–60% compared with a normal) state, which is also characterized by a dramatic (∼20 times) increase in the channel open time. At higher concentrations (50 to >100 μM) ryanodine completely blocks the channel (67, 440). These multiple actions most likely reflect the existence of high- and low-affinity binding sites for ryanodine within the RyR. Notably, ryanodine binding to the low-affinity site is almost irreversible. An important feature of ryanodine is its “use dependence” (440), i.e., preferential binding of the drug to the open channel. In practical terms, this became apparent as a facilitation of ryanodine action by repetitive activation of RyRs (either by caffeine or by [Ca2+]i increase), a phenomenon that is often observed in nerve cells (29, 92, 269, 308, 636, 661). Ryanodine is still the most specific pharmacological tool used for probing CICR in neurons (see Tables 1 and 2).
B) RUTHENIUM RED.
A polycationic dye, ruthenium red, is widely used as CICR blocker in nerve cells (see Tables 1 and 2). Its effective concentrations for CICR inhibition in muscle preparations range between 1 and 20 μM (720). At the single-channel level, ruthenium red effectively reduces Po by prolonging the closed state of the RyR. Unfortunately, as an inhibitor ruthenium red is rather nonspecific as it also inhibits VGCC (97) epithelial Ca2+ channels ECaC1 (459), TRPV channels (194), GABA release from hippocampal synaptic terminals (568), and mitochondrial calcium uptake (534). Moreover, when applied to permeabilized chicken atria, ruthenium red prevented contractile responses induced by InsP3 and potentiated responses to caffeine (676), which makes its employment as a specific CICR inhibitor even more dubious.
Dantrolene was originally synthesized in 1967 (598) as a potent muscle relaxant, and it is currently the only available drug for the specific treatment of malignant hyperthermia, a life-threatening complication of general anesthesia (313). Incidentally, dantrolene may also help against ecstasy intoxication (28). The action of dantrolene on the ER consists of reducing the Po of RyRs, thereby inhibiting CICR; the effective concentrations of the drug lie between 10 and 90 μM (469). Dantrolene is widely used as a CICR inhibitor in neurons (Tables 1 and 2), although there are some indications that dantrolene is effective only against RyR1 and does not inhibit RyR2 (167).
D) GENERAL ANESTHETICS.
Halothane and enflurane enhance Ca2+ release in cardiac preparation at low micromolar (i.e., in subtherapeutic) concentrations and increase Po of RyR channels (720). In the neural cholinergic cell line SN56, halothane triggered Ca2+ release from the ryanodine-, dantrolene- and CPA-sensitive calcium store (189). Considering that volatile anesthetics affect synaptic transmission (94, 95, 293) and neuronal excitability (414), the possibility that they act by modulating CICR should not be forgotten. In addition, however, halothane also inhibits plasmalemmal Ca2+-ATPase (157, 683), so the resulting effects on intracellular Ca2+ dynamics and Ca2+-dependent processes may be rather complex and difficult to interpret.
E) LOCAL ANESTHETICS.
Most of the local anesthetics, such as procaine, tetracaine, lidocaine, prilocaine, the quaternary amines QX 572 and QX 314, and benzocaine, inhibit caffeine-induced Ca2+ release and CICR in millimolar concentrations (720), and this action was also observed in nerve cells (308, 661).
Nifedipine, a well-known voltage-gated channel antagonist, was reported to increase the frequency of miniature end-plate potentials in synaptic terminals studied in the muscular junction of neonatal (up to 3 wk) rats. Effects of nifedipine persisted in Ca2+-free media and were inhibited by TG (2 μM) and ryanodine (10 μM), which allowed the authors to suggest possible stimulatory action of the drug on RyRs (512).
Heparin, which is customarily used as a selective inhibitor of IICR, does affect RyRs at the very same concentrations that are effective against InsP3Rs. The effects of heparin on RyRs were initially suggested when it was found that heparin stimulates Ca2+ release from skeletal muscle SR vesicles (404, 538). Further experiments on purified RyR channels demonstrated that heparin increases channel openings with EC50 values of ∼0.23 μg/ml (compared with 1–4 mg/ml concentrations usually employed for IICR inhibition) in a Ca2+-dependent manner (46, 135).
H) OTHER PHARMACOLOGICAL CICR REGULATORS.
The interaction with RyRs and modulation of CICR have been reported for many other pharmacological agents, which generally are not considered to possess such ability. For example, cardiac glycosides and, in particular, digoxin in therapeutic concentrations increase the rate of CICR in cardiac preparations through sensitizing RyRs to [Ca2+]i and increasing channel Po (390). Suramin, widely used as a purinoreceptor blocker, induces CICR and increases Po of RyRs (594). Another agent, believed to activate CICR and the RyR3 channel, in particular, is chloro-m-cresol, which triggers Ca2+ release in skeletal muscle (719), and in glial cells (383). So far, however, no systematic attempts to use chloro-m-cresol in nerve cells were made. Another class of drugs often employed in experiments with cultured neurons, the aminoglycoside antibiotics neomycin and gentamycin, are potent inhibitors of CICR with effective concentrations in a range between 50 nM and 10 μM (720).
2. Pharmacology of IICR
IICR pharmacology is less well defined than the pharmacology of CICR. From the very beginning of InsP3R studies it became apparent that a number of quite different pharmacological agents, including heparin, p-chloromercuribenzoic acid, cinnarisine, flunarisine, local anesthetics, such as tetracaine and lidocaine, and potassium channel blockers such as TEA, can effectively inhibit InsP3-iduced Ca2+ release from microsomal ER preparations (see, e.g., Refs. 482, 484). Further complexity was added when it became clear that caffeine, which was believed to selectively interact with RyRs, inhibits InsP3Rs (135, 680), and heparin, the most widely used InsP3R blocker, activates RyRs (46, 135). As if this was not enough, caffeine also effectively inhibits InsP3 production (645). All in all, the ideal inhibitor of IICR, which would penetrate biological membranes and demonstrate high selectivity against InsP3R, remains in the realm of dreams.
Heparin was shown to bind to InsP3Rs (613) and block the activity of InsP3-gated channels incorporated into lipid bilayers (43). Heparin was also shown to inhibit IICR in many types of nonexcitable and excitable cells, and it is constantly used as a pharmacological probe for neuronal IICR (see Tables 3–5). A big disadvantage of heparin is its membrane impermeability; hence, it must be delivered into the cell interior either by injection or via intracellular dialysis. Although heparin activates RyRs in artificial membranes, I failed to find a single report that would describe heparin-induced Ca2+ release in nerve cells. If that would be the case, however, heparin may still inhibit IICR by depleting ER via activated RyRs, although the interpretation of such data can be difficult.
The inhibitory action of caffeine on IICR was first discovered by Ole Petersen and co-authors (680) who studied agonist-mediated Ca2+ signaling in pancreatic acinar cells. Later on, the inhibition of IICR by millimolar concentrations of caffeine was found in Xenopus oocytes (493), cerebellar microsomes (66), and permeabilized smooth muscle cells (228). Direct electrophysiological investigations of purified InsP3Rs supported these findings by showing inhibition of unitary InsP3R current. The KD of caffeine action on InsP3R currents was ∼1.6 mM, and at 10 mM caffeine, a complete block of InsP3R currents was attained (228).
The ability of 2-APB to inhibit IICR was initially found in cerebellar microsomes stimulated with 100 μM InsP3; 2-APB blocked Ca2+ release with an IC50 value of ∼40 μM. Incubation with 2-APB also inhibited agonist-stimulated IICR in human platelets and neutrophils (373) and in HeLa cells (503). However, further investigations demonstrated that 2-APB lacks selectivity as it affects not only IICR but also Ca2+ pumps and store-operated Ca2+ entry in nonexcitable cells (55, 503).
Some hope for a selective, membrane-permeable InsP3R antagonist appeared in 1997 when Isaac Pessah’s group purified and characterized a group of macrocyclic bis-1-oxaquinolizidines isolated from the Australian sponge Xestospongia species (174). Several of these agents, dubbed Xestospongines (Xes), namely, XeA, C, and D, were reported to be powerful membrane-permeable blockers of IICR when tested on ER vesicles isolated from the cerebellum; of these, XeC was the most potent inhibitor with a KD of ∼360 nM. Further experiments have demonstrated that XeC blocks [Ca2+]i transients mediated by metabotropic pathways in the PC12 cell line and in primary rat cortical astrocytes. Incubation of cells with 5–20 μM XeC effectively abolished [Ca2+]i responses to bradykinin and carbachol (174). Yet more detailed investigations performed later somewhat dampened the optimism for the action of xestospongines as selective membrane-permeable blockers of InsP3Rs because it emerged that XeC acts as an effective blocker of SERCA pumps and thus empties the ER of Ca2+ (84, 123, 600). It is noteworthy that an indication that XeC interacts with SERCA pumps was already present in the initial paper by Gafni et al. (174); Figure 4 of this study clearly shows that incubation of PC12 cells with XeC decreases the amplitude of consequent TG-induced [Ca2+]i rise (which was interpreted by authors as an inhibition of leak pathway, yet it may reflect depletion of the ER as a result of SERCA blockade).
The adenophostines A and B were isolated from metabolites of fungi Penicillium brevicompactum (622) and found to be extremely potent (10–100 times more potent than InsP3) agonists of InsP3Rs (694). They were tested in turtle olfactory sensory neurons (272) and in single InsP3Rs isolated from fish olfactory cilia (70). In both cases adenophostines acted as powerful agonists of InsP3Rs inducing channel opening (70) and promoted IICR as judged by the appearance of InsP3/[Ca2+]i-dependent currents.
IV. ENDOPLASMIC RETICULUM CALCIUM SIGNALING AND REGULATION OF NEURONAL FUNCTIONS
A. Neuronal Excitability
Calcium release from the ER regulates neuronal excitability through several types of plasmalemmal Ca2+-activated channels, responsible for K+, Cl−, and nonselective currents (see Tables 9 and 10). In addition, Ca2+ release may (at least theoretically) suppress VGCC by increased Ca2+-dependent inactivation. The best-characterized effect of CICR is activation of K(Ca) channels, which control postspike afterhyperpolarization (AHP). The AHP comprises three distinct components (554), the fast AHP (mediated by BK channels), the AHP (mediated by SK channels), and the (ultra)slow AHP (mechanism not yet identified). Activation of the fast AHP is mediated solely by plasmalemmal Ca2+ entry, while both AHP and slow AHP are sensitive to CICR inhibitors. The contribution of CICR to the AHP is quite substantial: up to 50% of the AHP current is activated by Ca2+ released from the ER. The mechanism of CICR-mediated regulation of the (ultra)slow AHP remains entirely unknown. In vagal afferent neurons, Ca2+ release activates IAHP only when CICR is triggered by Ca2+ entry through N-type VGCC, which may reflect a very special colocalization of N-type Ca2+ channels, RyRs, and Ca2+-dependent K+ channels (107). A similar functional colocalization of RyRs and voltage-gated N-type Ca2+ channels was also demonstrated in bullfrog sympathetic neurons, where Ca2+ entry through N-type VGCC triggers regenerative CICR near the plasma membrane (2). In these cells, RyRs also seem to be functionally coupled to BK channels, thus providing a feedback between ER Ca2+ release, membrane excitability, and AP frequency. Finally, the same functional interactions between N-type Ca2+ channels and RyRs were confirmed in PC12 cells, where CICR is triggered by Ca2+ influx through N-type but not through L-type Ca2+ channels (657).
CICR also regulates membrane excitability through Ca2+-dependent Cl− channels underlying depolarizing afterpotentials (DAPs) found in the magnocellular neurons of the supraoptic nucleus, in sympathetic neurons, and in vagal nodose neurons subjected to vagotomy (328, 342, 370).
B. Neurotransmitter Release
The crucial importance of Ca2+ for neurotransmission was initially demonstrated by F. S. Locke for the neuromuscular junction (356), and the concept of Ca2+-dependent neurotransmitter release was coined by Bernard Katz and colleagues (118, 146). According to this theory, regulated exocytosis, of which neurotransmitter release represents a particular case, requires activation of numerous Ca2+-regulated transduction cascades that eventually result in the fusion of vesicles with the plasmalemma and release of vesicle contents into the extracellular space (in the case of neurotransmission, the synaptic cleft). Activation of exocytotic cascades requires a large (up to 100 μM) local [Ca2+]i concentration, and the role of plasmalemmal Ca2+ entry in neurotransmitter release is well established (36). The possible role of Ca2+ originating from the ER store is much less understood. Certainly, exocytosis in several types of nonexcitable and excitable cells almost exclusively depends on ER Ca2+ release [good examples are pancreatic acinar cells (506) and pituitary gonadotrophs (652)]. In a nerve-related cell preparation, neuroblastoma cocultured with rat skeletal muscle cells, bradykinin-induced IICR elevated [Ca2+]i and triggered acetylcholine release (468). However, whether ER Ca2+ release by itself can trigger exocytosis in presynaptic terminals in vivo remains an unanswered, albeit challenging, question.
Numerous morphological studies confirmed that functional ER is present in presynaptic terminals [e.g., Refs. 50, 82, 203, 391; see also an excellent review by Ron Bouchard et al. (59)], and moreover, ER structures are located in close proximity (as near as 40–400 nm) to the presynaptic active zone. As demonstrated already in 1986 by Westrum and Grey (693), the ER within the presynaptic terminal appears to be in close association with microtubules and synaptic vesicles and, moreover, ER portions impinge directly on the presynaptic membrane forming “tubular-fibrillar” extensions. Such close apposition of the ER and the presynaptic membrane may perfectly well allow Ca2+ released from the store to participate in creation of [Ca2+]i microdomains relevant for exocytosis. Yet the role of ER Ca2+ release in regulation of neurotransmitter release remains controversial, with arguments supporting (59) or denying (721) its importance.
Indeed, there are not many indications that ER Ca2+ release in the absence of plasmalemmal Ca2+ entry can induce massive exocytosis in synaptic terminals, although numerous data (see Tables 9 and 10) demonstrate that both CICR and IICR may modulate/amplify neurotransmitter release from presynaptic compartments.
Probably the first to contemplate the importance of presynaptic Ca2+ release in the regulation of neurotransmission were Erulkar and Rahamimoff (141), who arrived at this idea based on analysis of tetanic stimulation-induced changes in the frequency of the miniature end-plate potentials at varying Ca2+ electrochemical gradients. Further experimentation, aimed at investigating the effects of pharmacological manipulations with ER Ca2+ handling on either spontaneous or evoked postsynaptic currents, demonstrated that both Ca2+ release from and Ca2+ uptake into the ER may significantly modulate exocytosis of neurotransmitter in synaptic terminals. It appeared that spontaneous RyR-mediated presynaptic [Ca2+]i transients may trigger neurotransmitter release in resting conditions. This conclusion was based on experiments demonstrating that the frequency or amplitude of miniature postsynaptic currents (mIPSCs as well as mEPSCs) is sensitive to inhibition of RyRs (by ryanodine) or SERCA pumps (by TG and CPA). This phenomenon was observed in cerebellar neurons (26, 355) as well as in hippocampal pyramidal (139, 563) and cortical (589) neurons. In parallel, Ca2+ imaging of presynaptic terminals indeed revealed the spontaneous [Ca2+]i release events (139, 355). Moreover, in the presynaptic terminals of basket cells, the amplitude of [Ca2+]i signals resulting from spontaneous Ca2+ release was comparable with [Ca2+]i rises produced by single AP (105).
At the same time, CICR triggered by depolarization-induced Ca2+ entry was detected in single presynaptic terminals in intact bullfrog sympathetic ganglia, where it accounted for ∼46% of the total [Ca2+]i elevation induced by electrical stimulation of the nerve (502) and in the presynaptic mossy fiber terminals, where it contributed as much as ∼50% of the [Ca2+]i elevation induced by a train of action potentials (344). In hair cells (which are presynaptic to primary afferent neurons), direct activation of RyR-mediated Ca2+ release by caffeine induced a [Ca2+]i elevation and an increase in membrane capacitance that may indicate activation of exocytosis (337). Finally, pharmacological inhibition of ER reduced not only spontaneous but also evoked GABA-mediated IPSCs in basket cell-Purkinje neuron synapses in the cerebellum (175), whereas potentiation of CICR by intracellular addition of cADPR augmented ACh release from terminals of Aplysia buccal neurons (432). Likewise, CICR activated by Ca2+ entry via VGCC enhanced neurotransmitter release in the frog neuromuscular junction (450). In this preparation, a close apposition of RyRs and plasmalemmal Ca2+ channels was postulated, and it was also suggested that Ca2+ entry primes and activates CICR, which in turn contributes to neurotransmitter release (449).
The intimate mechanism of CICR-dependent potentiation of evoked neurotransmitter release may, however, be more complicated than simple amplification of Ca2+ entry and therefore an increase in [Ca2+]i domains near the active zone. The simplest hypothesis argues that CICR events trigger neurotransmitter release and moreover favor the multiquantal release from the same active zone (105, 355). However, analysis of ryanodine effects on the amplitude and frequency of evoked IPSCs in basket cell-Purkinje neuron synapse, led Galante and Marty (175) to suggest that Ca2+ released by CICR arrives at the active zone too late to affect exocytosis, but it does sensitize the exocytotic machinery to a subsequent depolarization.
Sometimes the effective modulation of neurotransmitter release requires synergism between IICR and CICR. When recording simultaneously from presynaptic axons and postsynaptic motoneurons in a Lamprey spinal cord preparation, Cochilla and Alford (102) observed potentiation of synaptic transmission by Ca2+ release from the ER; both IICR and CICR were necessary as this potentiation could be blocked by pharmacological inhibition of mGluRs as well as by 20 μM ryanodine.
Incidentally, the CICR responsible for spontaneous neurotransmitter release may also be activated by Ca2+ entry through presynaptic Ca2+-permeable ionotropic receptors. The frequency and amplitude of spontaneous mEPSCs in CA3 hippocampal neurons were greatly enhanced following activation of presynaptic nicotinic cholinoreceptors (NChRs). Subsequent analysis revealed that CICR triggered in the presynaptic terminal by Ca2+ entry through Ca2+-permeable NChRs was instrumental in the action of nicotine on mEPSCs. Pharmacological inhibition of presynaptic CICR completely abolished NChR-dependent modulation of mEPSCs (573). Most importantly, CICR in the presynaptic terminal remodeled neurotransmitter release quite substantially, by a shift towards a multiquantal mode, or else in synchronizing release across several active zones, as judged by a dramatic (∼3 times) increase in mean mEPSC amplitude and appearance of a relatively high number of extralarge mEPSCs reaching ∼200 pA (as compared with a mean mEPSC amplitude of 20 pA in control conditions) upon activation of NChRs. This remodeling has important functional consequences as nicotine-induced facilitation of neurotransmitter release was able to trigger a burst of action potentials in the postsynaptic cell, thus indicating the existence of synaptic transmission without AP-induced depolarization of the presynaptic membrane (573). Local CICR in response to Ca2+ entry through NChRs was also found in nerve terminals in vas deferens; this CICR facilitated neurotransmitter release, which resulted in a 70% increase in the amplitude of excitatory junction potentials (63).
Very much in line with the concept of Ca2+ store/sink duality, the presynaptic ER may not only amplify, but also limit neurotransmitter release, by buffering Ca2+ in the terminal, thus reducing the lifetime of elevated [Ca2+]i microdomains. This, for example, happens in the frog neuromuscular junction, where inhibition of SERCAs facilitates synaptic transmission (83). In the rat neuromuscular junction, CICR also limits the transmission, albeit in a completely different way, by limiting Ca2+ entry through VGCC as a result of increased Ca2+-induced inactivation of the latter (567).
However comprehensively the data discussed above favor the role of presynaptic Ca2+ release in the regulation of neurotransmission, it may not be regarded as an ubiquitous mechanism always present in all synapses. For example, Carter et al. (80) while investigating transmission in the hippocampus (associational-commissural synapses and mossy fiber synapses in CA1 and CA3 neurons) and cerebellum (parallel fiber-Purkinje neuron synapse) did not observe any involvement of CICR in the regulation of neurotransmitter release. Moreover, they failed to detect either caffeine-induced Ca2+ release or CICR in cerebellar parallel fibers. Likewise, a study of giant EPSCs [which were quite similar to extralarge ACh-mediated mEPSCs observed by Sharma and Vijayaraghavan (573)] in the mossy fiber-CA3 synapse did not show dramatic effects of either TG/CPA or ryanodine/caffeine on occurrence of giant EPSPs (215). Furthermore, the inhibition of Ca2+ release affected the parameters of normal EPSPs rather slightly, if at all (e.g., 10 μM ryanodine changed EPSC frequency to 126 ± 42% of the control; Ref. 215). No significant CICR (as probed by 3 mM caffeine, 10μM ryanodine, or 1 μM TG) was found during [Ca2+]i recordings from the single synaptic boutons of rat sympathetic preganglionic terminals (347) and from sympathetic varicosities (as probed by 10 μM ryanodine; Ref. 62). CICR was also not found in presynaptic terminals of goldfish retinal bipolar cells (296). Or was it? The same group published results supporting the importance of intracellular Ca2+ release for amplification of intraterminal [Ca2+]i transients and for facilitation of transmitter release from the same goldfish retinal bipolar cells (295); these two contradictory papers are separated by exactly 6 wk and were published in the same journal. These discrepancies most likely reflect the complexity of the system responsible for synaptic transmission. In particular, they may arise from differential expression of ER in various terminals, from differences in ER functional conditions (e.g., the degree of Ca2+ loading), and, naturally, from the experimental difficulties, as all these measurements are technically quite demanding and are performed at the very edge of the sensitivity of existing equipment. One also has to consider that incubation of whole brain slices with rather powerful pharmacological tools such as TG or ryanodine may trigger obscure indirect effects, thus complicating data interpretation.
C. Synaptic Plasticity
The living brain constantly remodels and modifies its cellular networks. Throughout life, synapses weaken and strengthen or else form and die. These processes underlie the adaptation of the brain to the constantly changing environment and represent what we know as learning and memory. Modification of synaptic transmission efficacy is an important part of this adaptation, which is manifested in short- and long-term synaptic plasticity. The electrophysiological correlates of long-term synaptic plasticity are long-term potentiation (LTP) or depression (LTD) of synaptic potentials in response to either high-frequency stimulation of single synaptic inputs or to simultaneous activation of several distinct synaptic inputs.
LTP, discovered by Tim Bliss and Terje Lomo in hippocampal neurons (53), was also the first form of synaptic plasticity, which highlighted the importance of cytoplasmic Ca2+ signals, as intracellular injection of the Ca2+ chelator EGTA effectively obliterated the potentiation of synaptic strength (363). Very soon thereafter, the importance of the cytoplasmic Ca2+ rise was also appreciated for cerebellar LTD (136, 301). More recent experiments also clearly demonstrate that the ER Ca2+ store plays an important role in the generation of [Ca2+]i elevations relevant to the induction of synaptic plasticity (154, 542, 552).
1. Short-term synaptic plasticity
Short-term synaptic plasticity is a form of presynaptic strengthening of neurotransmitter release, and as such, it ultimately depends on presynaptic Ca2+ signaling as discussed above. The involvement of CICR in a form of short-term plasticity, known as paired-pulse facilitation (PPF), was recently proposed for CA3 hippocampal neurons studied in organotypic slices (139). In contrast, a detailed investigation of PPF in excitatory synapses in the hippocampus and cerebellum failed to reveal any contribution of ER Ca2+ release to this phenomenon (80).
2. Long-term synaptic plasticity
Historically, LTP and LTD were mostly investigated in the hippocampus and cerebellum, respectively. Incidentally, as far as ER Ca2+ release is concerned, the mechanisms of hippocampal and cerebellar plasticity clearly differ in that cerebellar LTD involves IICR, whereas CICR is important for hippocampal LTP.
A) SYNAPTIC PLASTICITY IN THE HIPPOCAMPUS: A ROLE FOR CICR.
Much of our understanding of hippocampal synaptic plasticity comes from investigations of synapses formed by CA3 neurons through Schaffer collaterals on CA1 pyramidal cells. A peculiarity of this synapse is that it possesses both LTP and LTD depending on the stimulation protocol. LTP is usually induced by high frequency (∼100 Hz; 100 stimuli or so) stimulation, whereas LTD requires prolonged low-frequency stimulation of Schaffer collaterals (many 100s of stimuli at 1–2 Hz) (154). Both LTP and LTD critically require a [Ca2+]i increase. The precise mechanisms of differentiating between the different types of synaptic input in favor of potentiation versus depression are unclear, yet the signals arising from the ER may hold the key to further understanding.
The first and foremost task in the elucidation of the role of the ER in synaptic plasticity is to show that synaptic stimulation can result in Ca2+ release in dendritic spines and proximal dendrites, i.e., where exactly do the signaling events responsible for LTP/LTD occur. Hippocampal neurons have a rather peculiar distribution of the ER and Ca2+ release channels. First, in vivo the ER is not present in every spine. Only ∼25–50% of the small spines contain ER. In contrast, the latter is present in ∼80–90% of the big mushroom-like spines (604). Second, hippocampal pyramidal neurons have a peculiar distribution of Ca2+ release channels (617). Only RyRs (particularly RyR3) are present in the spines, while both RyRs and InsP3Rs reside in the dendritic ER [although prominent species-dependent differences exist, i.e., RyR expression in the rabbit hippocampus appears to be almost negligible compared with the rat (558)].
Such a distribution of Ca2+ release channels suggests strong CICR in those CA1 neuron spines, which possess a functional ER. Because the distribution/localization of the latter may change depending on the experimental conditions (i.e., the animal’s age), conflicting results could be anticipated safely, and, sure enough, the controversy did not wait long to emerge. To begin with, a robust caffeine-induced Ca2+ release, sensitive to TG and ryanodine, and CICR in response to glutamate application, were observed in spines from cultured hippocampal neurons (303). This observation was confirmed by [Ca2+]i imaging on hippocampal neurons in organotypic slices (138), which revealed that CICR is responsible for by far the largest (∼80–90%) part of the [Ca2+]i elevation triggered by stimulation of NMDA receptors (Fig. 21A). Yet, when confocal Ca2+ imaging was applied to hippocampal neurons in acute slices, no signs of significant CICR following NMDA-mediated Ca2+ influx were detected (312), and the NMDA-induced Ca2+ flux was found to be almost solely responsible for the [Ca2+]i elevation (Fig. 21B). All in all, the question about functional CICR in spines remains unresolved, which most likely reflects the complexity and heterogeneity of the preparations. Yet, physiological data support the notion of at least some involvement of CICR in LTP in SC-CA1 synapses, since emptying the ER with TG (1 μM) or CPA (1 μM), or blocking CICR with 10 μM ryanodine, inhibited the induction but did not affect the maintenance of LTD in acute hippocampal slices (530).
The next step in investigating the role of ER Ca2+ release in LTP quite naturally involved genetic manipulation of Ca2+ release channels. The obvious candidate to check was RyR3, because of its expression in CA1 neurons. In any case deletions of RyR1 and RyR2 are not compatible with life. Genetic deletion of RyR3 led to facilitated LTP and a complete disappearance of LTD in CA1. Furthermore, RyR3-deficient mice demonstrated improved spatial learning on a Morris water maze task (173). LTP in the RyR3 knockout mouse was, however, principally different from the wild type; it could not be blocked by NMDA receptor inhibitors and was partially dependent on Ca2+ influx through VGCC, and on metabotropic glutamate receptors. Another RyR3 knockout mouse, also generated in Japan, displayed a smaller magnitude of LTP and diminished synaptic AMPA responses, although this was not due to downregulation of AMPAR expression (578). Finally, a RyR3 knockout mouse generated in Italy showed no changes in LTP induction, although depotentiation of LTP was impaired. In the water maze, the Italian mice performed rather poorer than controls, and “in the open-field, RyR3−/− mice displayed a normal exploration and habituation, but had an increased speed of locomotion and a mild tendency to circular running” (23). To add to the degree of complexity, the genetic deletion of InsP3R1 led to an increased magnitude of CA1 hippocampal LTP. Homosynaptic LTD remained unchanged, whereas heterosynaptic LTD of nonassociated pathway was blocked (168, 441). In summary, the role of ER Ca2+ release in synaptic plasticity in CA1 hippocampal neurons remains obscure.
Some indications of the importance of CICR in generation of LTP/LTD are coming from other neurons (CA3 hippocampal cells, cortical neurons), although they often involve both pre- and postsynaptic mechanisms (see Table 9). IICR seems to play a more important role in CA3 neurons, as deletion of InsP3Rs facilitated LTP and completely inhibited LTD in mossy fiber-CA3 neuron synapses (254).
B) SYNAPTIC PLASTICITY IN THE CEREBELLUM: A ROLE FOR IICR.
The role for IICR in LTD generation in the cerebellum is rather well established. The LTD of parallel fibers, which form their synapses on dendrites of Purkinje neurons, occurs as a result of simultaneous activation of parallel and climbing fibers. Stimulation of climbing fibers can be fully substituted by depolarization of the neuronal soma and [Ca2+]i elevation (301). Whatever protocol of LTD induction is used, [Ca2+]i elevation in the PN is necessary for LTD development. In fact, stimulation of climbing fibers or somatic depolarization may be replaced by an artificial increase in [Ca2+]i due to photorelease of caged Ca2+ (338) (incidentally, stimulation of parallel fibers can be fully substituted by NO generation, Ref. 338).
Genetic manipulation of components of the mGluR-IICR signaling cascade evidently demonstrated the crucial importance of this pathway for cerebellar LTD. Deletion of mGluR1 resulted in the appearance of a clear cerebellar phenotype, with ataxia, motor discoordination, and impaired LTD (1, 104). Both LTD and locomotive deficits can be restored by expression of mGluR1 under the control of a Purkinje neuron-specific promotor (243).
Almost simultaneously, the involvement of IICR in cerebellar LTD was suggested by physiological experimentation. Direct activation of IICR by photorelease of caged InsP3 in Purkinje neurons produced LTD when coincident with somatic depolarization, thus fully substituting for the stimulation of parallel fibers (282). Furthermore, as shown by Finch and Augustine (150), local InsP3 uncaging and subsequent IICR were sufficient to induce local LTD in synapses close to the uncaging foci.
It has to be noted, though, that IICR appeared not to contribute to LTD in cultured or freshly isolated Purkinje neurons, as no difference was found in [Ca2+]i transients induced by depolarization and combination of depolarization and glutamate application. Moreover, LTD was insensitive to XeC, and diacylglycerol was able to substitute for mGluR activation in the induction of LTD (448). Yet, these results most likely reflect certain changes in signaling pathways introduced by tissue culture conditions or the isolation procedure.
Further evidence favoring the key role of IICR in the generation of cerebellar LTD was obtained from mice with genetically deleted InsP3R1s, in which LTD completely vanished (Fig. 22A) (247). Interestingly, InsP3R1−/− animals never survived beyond the 23rd day. The same authors also reported that a similar disappearance of LTD can also be achieved by perfusing neurons with specific antibodies raised against InsP3R1. Not only the expression but also the localization of InsP3R1s matters, as was demonstrated on animals mutated for myosin Va. This mutation causes the ER not to enter PN spines, which therefore effectively eliminates IICR specifically from this compartment. The LTD in mutant mice and rats was completely lost (Fig. 22B); nonetheless, it could be rescued by Ca2+ uncaging within the spine (421).
C) SYNAPTIC MODIFICATION.
The most amazing results were obtained when studying GABAergic synapses formed by basket interneurons on CA1 hippocampal pyramidal cells. When the CICR mechanism was primed by intracellular cADPR, postsynaptic depolarization triggered a large [Ca2+]i elevation that changed GABA-mediated IPSPs into excitatory postsynaptic potentials (612). The actual mechanism underlying such remodeling is not precisely understood, although it is sensitive to carbonic anhydrase inhibition. This may indicate certain changes in the GABA-operated Cl− channel, resulting in the appearance of a depolarizing HCO3− efflux.
D) MEMORY PROCESSES.
Not much is known about the role of Ca2+ release in consolidative memory processes. There are some indications that synchronous oscillations of membrane potential in the neocortex, which were suggested to be important for cognitive brain function (218), are governed by CICR triggered by Ca2+ entering through NMDA receptors (710). A certain importance of ER Ca2+ release may be also deduced from the observation that a significant increase in RyR2 expression, at both the mRNA and protein level, was found in the hippocampus of rats subjected to intensive water maze training (716).
D. Gene Expression, ER Ca2+ Waves, and ER Ca2+ Tunneling
The importance of cytoplasmic and nuclear Ca2+ signals in the regulation of gene expression is well established, and a variety of Ca2+-dependent cascades controlling transcription have been characterized (19, 91, 152, 183, 184, 692), which shall not be discussed here. However, the problem of how Ca2+ reaches the nucleus during physiological activity of the geometrically complex and highly polarized nerve cell remains unsolved. It is well established that cytosolic Ca2+ buffering hampers Ca2+ diffusion from often distant entry sites towards the cell interior. The ER can be instrumental in conveying Ca2+ over long distances and therefore providing nuclear systems with appropriate Ca2+ signals. Intracellular propagation of Ca2+ signals can be achieved by virtue of endomembrane excitability. Either RyR-mediated CICR or CICR-assisted Ca2+ release from InsP3Rs or their combination may create propagating Ca2+ waves transforming near-plasmalemmal [Ca2+]i elevations into nuclear Ca2+ rises. Indeed, such propagating [Ca2+]i waves have been observed experimentally. For instance, CICR maintains spreading Ca2+ waves between the plasmalemma and the nucleus of the frog sympathetic ganglion neurons. Moreover, the speed of this wave remains constant all the way to the nucleus, suggestive of an active propagation, whereas at the moment the Ca2+ signal reaches the nucleus, its spread becomes diffusion governed (112, 389). Similarly, InsP3Rs-dependent propagating Ca2+ waves have been described in a variety of central neurons (see Ref. 25 for review). Most interestingly, the Ca2+ wave can become amplified when approaching the center of the cell. For example, pairing of muscarinic stimulation of CA1 hippocampal neurons with trains of action potentials resulted in a focal Ca2+ elevation, which triggered a propagating Ca2+ wave, and the [Ca2+]i increase resulting from this wave was largest in the soma and the nuclear region (518). This amplification critically depended on ER Ca2+ release, as emptying the ER by caffeine completely prevented wave propagation.
Yet, the ER may offer an alternative route for long-range Ca2+ propagation, which relies on the much faster Ca2+ diffusion within the ER lumen compared with the cytoplasm. This pathway, known as the intraluminal Ca2+ tunnel, has been described in pancreatic acinar cells, where Ca2+ was shown to rapidly equilibrate throughout the whole of the ER (491), and moreover, intra-ER Ca2+ diffusion can effectively replenish the depleted store in conditions of highly localized Ca2+ entry (424). As discussed above, synaptic stimulation can result in a very substantial accumulation of Ca2+ in the dendritic ER. This local increase of luminal Ca2+ may, by Ca2+ tunneling, rapidly reach the nucleus even from remote cellular processes. Because the ER is connected with the nuclear envelope (or rather the nuclear envelope is part of the ER), Ca2+ can be released directly into the nucleus via Ca2+ release channels residing in the envelope membrane (182, 642). Furthermore, sustaining Ca2+ waves in the neuronal soma would also require Ca2+ transport through the ER lumen. As mentioned above, the ER in central neurons may not be full in resting conditions, and therefore, initial loading of the store is necessary for the maintenance of a propagating Ca2+ wave. The existence of neuronal ER Ca2+ tunneling has not been demonstrated directly, although recent experiments performed by Hilmar Bading and co-workers (201) may provide some evidence for such a pathway. They investigated CREB-mediated gene expression in cultured hippocampal neurons treated with bicuculline. This treatment facilitates spontaneous firing in peripheral dendrites associated with substantial peripheral Ca2+ influx and [Ca2+]i elevation in dendrites, which then had to be conveyed to the nucleus to activate CREB. Bading and co-workers (201) found that inhibition of ER Ca2+ accumulation by TG completely prevented the signal from reaching the nucleus. In the described experimental conditions, Ca2+ entry into the ER must occur in dendrites, precisely where [Ca2+]i is elevated, and therefore, the possibility of Ca2+ diffusing through the ER lumen to the nucleus appears probable. Clearly many more experiments are needed to test for ER Ca2+ tunneling in nerve cells.
E. Neuronal Growth and Morphological Plasticity
Neurite outgrowth is governed by [Ca2+]i fluctuations within the growth cone, where both the level of [Ca2+]i and patterns of [Ca2+]i fluctuations regulate the velocity of cone extension (273). For example, complex [Ca2+]i transients are produced in growth cones during their migration in the embryonic spinal cord. The larger the frequency and magnitude of these [Ca2+]i spikes, the slower the outgrowth. Inhibition of [Ca2+]i elevation accelerates the rate of neurite extension, whereas artificial increase of [Ca2+]i by Ca2+ photorelease inhibits axonal extension (190).
The [Ca2+]i signals relevant for neurite outgrowth are produced by both Ca2+ entry and Ca2+ release, and the particular importance of CICR was appreciated quite early (297, 330). Recent analysis of growth cone movements of Xenopus neurons treated with netrin-1, an established regulator of neurite outgrowth, has further substantiated the importance of CICR for growth cone extension (233). Focal administration of netrin-1 to the neuronal culture created a local concentration gradient, which, in turn, forced the neurites to turn towards the source of netrin-1. This turning movement was modified by pharmacological inhibition of either plasmalemmal Ca2+ entry (by 50 μM Cd2+) or Ca2+ release from the ER (by 10 μM BHQ or 1 μM TG). When the cultures were treated with 100 μM ryanodine, however, the administration of netrin-1 produced repulsion of the cone. Most interestingly, applications of low concentrations of ryanodine (∼100 nM around the growth cone) caused a [Ca2+]i elevation within the cone and turning of the latter towards the source of the drug (233). Thus, by regulating the pattern of local [Ca2+]i signaling in the growth cone, CICR determines the direction and rate of axon extension/repulsion. A somewhat similar role of CICR was observed in the embryonic chick retina, where local CICR stabilized dendrite formation during development. Inhibition of this local Ca2+ release induced rapid retraction of dendrites (358).
In contrast, in embryonic chick DRG neurons, the regulation of growth cone motility seems to be regulated by InsP3Rs. Growth cones of these neurons are particularly rich in InsP3R1. Depletion of the ER store with TG (10 μm) arrested the extension of the cones, and inhibition of IICR with intracellularly injected heparin induced neurite retraction (626).
ER Ca2+ release controls not only the extension of growth cones, but also the extension of dendritic spines. Caffeine-activated release of Ca2+ from the ER triggers rapid elongation of spines in cultured embryonic hippocampal neurons. Spine elongation was already obvious 30 min after a short application of 5–10 mM caffeine, and after 90 min, the spine length had increased by 0.35 μm (∼33%; Ref. 304). Short puffs of caffeine even triggered a de novo formation of spines, although this was not very frequent (new spines appeared in 4 of 22 preparations). The effects of caffeine were a specific consequence of ER Ca2+ release, as 10 μM ryanodine obliterated them completely (304).
In summary, ER Ca2+ release acts as a powerful regulator of the motility of growth cones and dendrites, most likely through interaction with enzymes controlling polymerization/depolymerization of the cytoskeleton.
F. Neurohormone Release
An interesting hypothesis postulating that the ER Ca2+ store may regulate dendritic (i.e., extraterminal) release of oxytocin was developed as a result of in vivo experiments, which measured oxytocin release in the supraoptic nucleus region of anesthetized rats in response to local dialysis of ER-specific agents. TG, but not caffeine or ryanodine, induced release of the neurohormone and greatly potentiated depolarization-induced oxytocin secretion. Interestingly, this potentiation was not observed shortly after (5 min) application of TG, but a large potentiation was detected 30–90 min later (361). The authors explained their results in terms of TG-induced sustained Ca2+ release, which, in fact, does not seem a plausible explanation. First, [Ca2+]i rises caused by SERCA inhibition are short-lived, and second, depletion of the ER store renders additional Ca2+ release from this compartment impossible. If anything, post-TG potentiation of depolarization-induced oxytocin release may be explained by larger [Ca2+]i rises in the presence of TG due to exclusion of ER Ca2+ buffering, or else due to activation of an alternative Ca2+ entry pathway.
IICR triggered by stimulation of P2Y purinoreceptors stimulated the release of a pain-related neuromodulator, calcitonin gene-related peptide (CGRP), from small (presumed nociceptive) cultured DRG neurons (559).
Finally, the release of BDNF from electrically active neurons required activation of Ca2+ influx through N-type Ca2+ channels and CICR, as BDNF secretion was inhibited by ω-conotoxin GVIA, TG, and dantrolene (22).
G. Circadian Rhythms
Ca2+ release via ryanodine receptors is intimately involved in the control of circadian rhythms in the suprachiasmatic nucleus (SCN). Transfection of organotypic SCN slices with cytosol- and nucleus-targeted chameleon YC2.1 (422), by employing gene gun technology, allowed long-term monitoring of intracellular Ca2+ in neurons that maintain circadian rhythms (244). These long-lasting recordings revealed a circadian rhythm for cytosolic but not nuclear [Ca2+], and the former was fully synchronized with the electrical circadian rhythm measured by a multiple electrode array. Treatment with ryanodine (100 μM) depressed both the [Ca2+]i and electrical rhythms, although TG (1 μM) was ineffective (244). Hence, the precise role played by the ER store remains to be investigated in much greater detail.
Most interestingly, the expression of RyR2 in SCN neurons is also governed by the circadian rhythm, with the maximal level in the second half of the day phase. This rhythm of RyR2 expression was preserved even when the animals were constantly kept in dim light, suggesting that it has a fundamental importance in setting the circadian clock (125). A role for InsP3Rs in regulation of circadian rhythms has also been suggested, although the only evidence for this is the inhibitory action of 2-APB (199), which may not be associated with suppression of IICR at all due to the lack of specificity.
V. ENDOPLASMIC RETICULUM CALCIUM HOMEOSTASIS AND NEURONAL PATHOPHYSIOLOGY
A. ER Stress Response and Neurodegeneration
Since ER is responsible for transcription, posttranslational protein modification, and selective transport of proteins to different destinies, it has developed a highly sophisticated system for controlling the quality of the final protein products. Any imbalance of protein handling triggers a specific reaction generally defined as the ER stress response. This response is triggered either by accumulation of unfolded proteins (the “unfolded protein response,” UPR) or by overexpression of particular proteins (“the ER overload response,” EOR). The ER stress results in activation of several transcription factors (most notably IRE1 and its receptor PERK) that upregulate the expression of chaperones and reduce ER protein load by inhibiting overall protein synthesis. In addition, the ER triggers the activation of NF-κB and the transcription factor C/ERP-homologous protein, which in turn control synthesis of proinflammatory and hemopoetic proteins and act as inhibitors of growth as well as promoters of apoptosis (see Refs. 12, 155, 364, 477, 478, 500). If the ER stress persists, it may result in the release of necrotic factors, which trigger cell death. As already discussed above, the activity of many intra-ER enzymatic cascades ultimately depends on the intraluminal Ca2+ concentration, and a strong link between ER Ca2+ homeostasis, ER stress, and neurodegeneration therefore has been postulated (495–498). This hypothesis is supported by several lines of evidence, which show that 1) depletion of ER stores induces upregulation of ER stress markers, 2) depletion of ER stores causes neuronal cell death, 3) exposure of neuronal cells to various insults upregulates ER stress markers, and 4) inhibition of ER Ca2+ release protects against neurotoxicity. Several examples of such data are discussed below.
The depletion of ER Ca2+ stores in primary cultured rat embryonic cortical neurons by incubation with TG (1 μM), CPA (30 μM), or caffeine (10 mM) reduced protein synthesis to 10–70% of the control and caused complete disaggregation of polysomes (129, 499). A significant decrease in protein synthesis was also observed in cultured embryonic rat cortical neurons treated with 100 nM ryanodine, a concentration which depletes the ER store by opening RyRs (5). At the same time, ER depletion following 30-min incubation with 1 μM TG led to an almost 200-fold increase in the expression of ER stress markers grp78, grp94, gadd34, and gadd153 and in a 6-fold increase of SERCA2b and BCl-2 (403). Therefore, severe disruption of ER Ca2+ homeostasis triggers ER stress and has a clear neurotoxic effect.
ER stress is also induced by exposure of nerve cells to various neurotoxic conditions. For example, application of excitotoxic concentrations of glutamate as well as oxidative insults elevated the level of grp78 in rat cultured hippocampal neurons (713). This increase in grp78 expression was neuroprotective, as treatment of cultures with grp78 antisense substantially increased neuronal cell death. Hypoxia/reperfusion increased the expression of another chaperon, grp94, in neuroblastoma cells. The increased level of grp94 was also cytoprotective, as treatment of neuroblastoma cells with grp94 antisense exacerbated cell death (24). Similarly, CA1 hippocampal neurons overexpressing grp94 were resistant to ischemic damage in gerbils subjected to transient carotid artery occlusion (24).
Finally, disturbances of ER Ca2+ handling cause neuronal death. Treatment of cultured cerebellar granule neurons with the neurotoxin mastoparan triggered massive IICR, which in turn initiated apoptosis and cell death. Inhibition of PLC (and hence InsP3 production) had a neuroprotective effect (346). Depletion of Ca2+ stores with TG (300 nM), CPA (300 μM), and BHQ (300 μM) resulted in 50–60% neuroblastoma cell death within 48 h and in an almost complete demise of the culture within 72 h (455). Interestingly, loading the cells with BAPTA-AM did not protect against TG toxicity, although it prevented TG-induced [Ca2+]i increases, thus demonstrating that the ER depletion, rather than [Ca2+]i disturbances, were responsible for neurotoxicity (455). Experiments on the same cell line demonstrated that toxicity induced by oxygen-glucose deprivation is also mediated through Ca2+ depletion of the ER (684). In Caenorhabditis elegans, necrotic neuronal death can be prevented by loss of function mutations in Ca2+ release channels; at the same time, treatment with TG invariably exacerbated necrosis (e.g., Ref. 702).
The ER stress response may regulate neuronal death through another pathway, which involves caspase-12. This enzyme resides within the ER lumen and is activated upon various insults initiating ER stress (577). Most notably, ER stress triggered by disruption of ER Ca2+ homeostasis also activated caspase-12 cascade (442). This pathway is endowed with a degree of specificity as cortical neurons isolated from caspase-12-deficient mice were resistant to apoptosis induced by amyloid-β proteins, but not to apoptosis induced by staurosporine or trophic factor deprivation (442). Caspase-induced cell death in embryonic hippocampal neurons similarly depended on ER Ca2+ signaling and required the activation of ryanodine receptors (239).
Overall, it is quite clear that nerve cells do develop ER stress upon severe disruption of ER Ca2+ homeostasis. Yet, the maneuvers employed for damaging ER Ca2+ handling, such as the irreversible block of SERCA pumps by TG, are rather brutal and are very far from the conditions that neurons may encounter even in a diseased state. Clearly more experimentation is needed to precisely characterize the link between [Ca2+]L and ER stress. Another very important point lies in the determination of harmful levels of [Ca2+]L. Indeed, it is evident that dramatic falls in intra-ER free Ca2+ are detrimental for the cell; however, increased [Ca2+]L can also be toxic (147, 511). It seems that a certain normal “window” of [Ca2+]L exists, and when ER Ca2+ rises or falls beyond a particular limit, grave consequences follow. Lastly, changes in ER Ca2+ concentration may cause severe fragmentation of the ER (610, 634), which may also have pathological consequences.
B. ER Stores and Ischemia
As a rule, glucose/oxygen deprivation triggers a substantial elevation of [Ca2+]i in nerve cells (521, 630, 708), part of which is due to ER Ca2+ release (569, 644). In striatal slices exposed to hypoxic conditions, Ca2+ release from the ER (partially mediated by NO) was responsible for a large part of the 45Ca2+ liberation (394). When cultured hippocampal neurons were subjected to an anoxic environment, mimicked by incubation with NaCN, a large [Ca2+]i increase was induced. This [Ca2+]i elevation was blocked by dantrolene and ruthenium red, therefore implying an important role of CICR (132) The NaCN-induced [Ca2+]i increase considerably exceeded the [Ca2+]i elevation in response to glutamate, although the action of the latter was grossly amplified in conditions of metabolic stress.
Similarly, energy deprivation (induced by perfusion with an oxygen/glucose free solution) produced a large [Ca2+]i elevation in hippocampal neurons in slices ([Ca2+]i increased up to 28 μM), and this increase was invariably followed by cell edema and death. Similar massive (up to 30 μM) increases in [Ca2+]i following an interruption in blood flow were observed in hippocampal neurons impaled with ion-sensitive electrodes (588). Depletion of the ER stores by preincubation of hippocampal slices from young Wistar rats with TG (1 μM, 30 min) dramatically (∼5 times) reduced the [Ca2+]i elevation triggered by energy deprivation (192). Promotion of ER Ca2+ release (by intracellular injection of InsP3 into gebril hippocampal neurons) greatly facilitated the irreversible depolarization induced by ischemia (288). Dantrolene significantly reduced neuronal death in the CA1 area of the hippocampus in adult rats, subjected to global cerebral ischemia, but this effect could be partly attributed to reduced release of glutamate (447). Interestingly, however, the glutamate release was diminished to an equal extent by dantrolene as well as by DMSO (used as a solvent for dantrolene); DMSO alone also protected neurons against ischemia, yet dantrolene was twice as potent. Inhibition of ER Ca2+ release also prevented the liberation of neurotoxic free fatty acids from the ischemic cortex (510).
Ca2+ release from the ER also appears instrumental in ischemic damage of white matter. Destruction of axons upon ischemic stress is mediated through an increase in cytosolic Ca2+. Experiments on rat dorsal columns showed that this [Ca2+]i elevation persisted after removal of Ca2+ from the perfusate; similarly, ischemic injury of axons persisted in Ca2+-free conditions. At the same time, the axonal [Ca2+]i increases were sensitive to ER depletion by TG or by inhibition of CICR by ryanodine (473). Most intriguingly, the CICR in axons may be of the “skeletal muscle” variety, i.e., involving direct coupling between plasmalemmal Ca2+ channels and RyRs. This suggestion is based on the fact that blockage of the Ca2+ channel voltage sensor (by dihydropyridines), but not of the Ca2+ channel pore (by Cd2+), was hugely protective against ischemia. At the same time, simultaneous inhibition of RyRs (by ryanodine) and of Ca2+ channels by nimodipine was not additive, thus favoring interdependence of two molecules. In concordance with this suggestion, replacement of extracellular Na+ with impermeable cations, limiting depolarization, protected axons against ischemia, whereas substitution of Na+ with Li+ (which is carried through Na+ channels and hence allows depolarization) was not effective (473). Together, these findings provide very important information on mechanisms of ischemic damage of white matter, which seem to be different from those in gray matter. White matter damage is accomplished by intracellular Ca2+ release, whereas injury of gray matter very much depends on plasmalemmal Ca2+ entry.
The link between ischemia-induced changes in ER Ca2+ homeostasis and intra-ER chaperone activity has also been described recently, when the ischemia-induced ER chaperone orp150 (oxygen-regulated protein of 150 kDa) was isolated and characterized. Orp150 is strongly upregulated in astrocytes following ischemic stress, which may explain their exceptional resistance to ischemic conditions. Overexpression of orp150 in cultured rat cortical neurons protected them against ischemia, and vice versa, inhibition of orp150 synthesis in cultured astrocytes increased their vulnerability to hypoxic shock (627). Transgenic mice expressing orp150 under control of a platelet-derived growth factor B-chin promoter displayed high levels of orp150 in the cortex, hippocampus, and cerebellum and were much more resistant to global cerebral ischemia compared with wild-type controls (627). Interestingly, in cultured neurons with an increased level of orp150, glutamate-induced [Ca2+]i loads were substantially reduced, implying that this chaperone may also stabilize cellular Ca2+ homeostasis, most likely by limiting Ca2+ release from the ER, which accompanies excessive glutamate stimulation (294).
C. ER Stores and Human Immunodeficiency Virus
The human immunodeficiency virus (HIV) envelope glycoprotein GP120 was shown to induce large [Ca2+]i increases in cultured hippocampal neurons and promote their degeneration. Both the [Ca2+]i elevation and the neurodegeneration can be prevented by the CICR blocker dantrolene, or by ER depletion produced by preincubation of cells with caffeine, carbachol, or TG (393).
D. ER Stores and Diabetes
The involvement of the ER Ca2+ store in diabetic peripheral neuropathies can be suggested on the basis of a rather limited number of observations demonstrating a reduction in the size of caffeine-induced Ca2+ release in peripheral neurons isolated from animals with experimental (usually streptozotocin induced) diabetes type I. Such a decrease was found in diabetic DRG neurons (242, 306, 316), and the reduced [Ca2+]i responses to caffeine were especially prominent in cells isolated from lumbar ganglia (242), which are the first to be affected by diabetic neuropathies. A very substantial decrease in caffeine-induced [Ca2+]i transients (from 270 nM in control animals to 30 nM in diabetic ones) was also observed in dorsal horn neurons from streptozotocin-induced diabetic rats (316, 679). In in vivo experiments, ryanodine and TG significantly modified the formalin-induced nociceptive response in streptozotocin-diabetic mice (266) providing further evidence that the ER Ca2+ store is involved in the pathogenesis of diabetic neuropathies.
E. ER Stores and Gangliosidoses (Gaucher’s and Sandhoff’s Disease)
Mutations of the human gene encoding glucocerebrosidase result in a reduction of activity of this enzyme and accumulation of glycosylceramide, which, in turn, promotes neurodegeneration in the central nervous system. Clinically this type of neurodegenerative disorder is known as Gaucher’s disease (27). When the activity of glucocerebrosidase in cultured hippocampal neurons was blocked by incubation with conduritol-B-epoxide, accumulation of glycosylceramide occurred, thus mimicking the pathological conditions associated with Gaucher’s disease. This condition was accompanied by a significant increase in ER tubular elements and by a great increase in the release of Ca2+ from the ER in response to both metabotropic stimulation and caffeine (302). Furthermore, hippocampal neurons with inhibited glucocerebrosidase were significantly more vulnerable to excitotoxic shock (exposure to high glutamate concentrations), and this increased vulnerability was blocked by ryanodine (302). Therefore, remodeling of the ER Ca2+ homeostasis which results in an increase of ER Ca2+ content may be regarded as a key element of neurodegeneration associated with Gaucher’s disease.
Another type of gangliosidoses, when gangliosides accumulate in cells due to a defect in lysosomal proteins catalyzing their degradation, the GM2 gangliosidoses, or Sandhoff’s disease, also affects the ER, as the rate of SERCA-driven Ca2+ uptake was found to be considerably lower in the Sandhoff mouse model (501).
F. ER Stores and Alzheimer’s Disease
Alzheimer’s disease (AD) is a form of senile dementia characterized by the advent of plaques in the brain matter (54, 153) and appearance of fibrillous structures within the neuronal cytoplasm (11). It receives much attention because of its social impact affecting the progressively ageing mankind. The importance of cellular Ca2+ homeostasis in the pathogenesis of AD was perceived at the beginning of 1980s, when Zaven Khachaturian promulgated his “Ca2+ hypothesis of AD and neuronal ageing” (281). This hypothesis postulated that long-lasting subtle changes in [Ca2+]i homeostasis would eventually result in the neurodegeneration so characteristic of AD.
Investigations in the last decade confirmed this early reflection by clearly demonstrating that disruption of cellular Ca2+ homeostasis is a prominent feature of AD pathophysiology. Furthermore, these investigations identified ER Ca2+ dysregulation as a key step in the development of AD.
As already mentioned, the histopathology of AD is represented by plaques formed by aggregates of amyloid-β peptide (Aβ) and by the appearance of intracellular filamentous structures of a macrotubule-associated protein tau (378). This reflects several complex alterations of enzymatic systems controlling the synthesis and processing of Aβ. The familial form of AD (FAD) is caused by mutations of genes encoding the Aβ precursor protein (APP, chromosome 21) and presenilins 1 (PS1, chromosome 14) and 2 (PS2, chromosome 1). Mutations in PS1 are responsible for the majority of FAD cases. The central question for the pathophysiology of AD is related to events proximal to the malformation of Aβ, as they could represent potential therapeutic targets.
Disrupted Ca2+ homeostasis has been identified as a feature of AD. Increased levels of free and bound Ca2+ have been detected in neurons containing neurofibrillary tangles (439). Likewise, neurons possessing tangles showed increased activity of various Ca2+-dependent proteases (461). Early studies on fibroblasts isolated from AD patients also demonstrated impairment of Ca2+ mobilization in response to metabotropic stimulation (186, 252). Direct administration of Aβ to nerve cells caused elevation in resting [Ca2+]i and greatly diminished the ability of cells to cope with Ca2+ loads (207, 367, 379; see also Ref. 377 for review). Most importantly however, cellular Ca2+ dishomeostasis occurs at the initial stages of AD (142), and studies on transgenic mice with APP or PS mutations found that impaired Ca2+ handling occurs long before any histopathological changes (195, 323). The mechanisms of AD-related disruption of [Ca2+]i homeostasis can be multifacetous and involve plasmalemmal systems as well as intracellular organelles. The effects of Aβ and its fragments, for instance, are reported to influence plasmalemmal Ca2+ transport by forming Ca2+-permeable pores (47, 279), although acute application of Aβ additionally augments intracellular Ca2+ release (207, 299). Several recently conducted studies suggest that disrupted ER Ca2+ homeostasis may be regarded as a specific feature of the AD-related neuronal Ca2+ dysregulation (see Table 11 and Refs. 323, 378, 380, 608, 709).
Among molecules involved in the pathogenesis of AD, presenilins are particularly relevant for disruption of ER function. Presenilins are integral membrane proteins that contain eight transmembrane domains and serve as part of a multiprotein protease complex, the γ-secretase, which is responsible for the intramembranous cleavage of the APP and the Notch receptors (124). Most importantly, presenilins are involved in integrative neuronal functions, as conditioning knockout of both PS1 and PS2 in the forebrain of adult mice causes a remarkable impairment of memory, LTP, NMDA responses, and CREB/CBP-mediated gene expression and promotes neurodegeneration (562).
Presenilins are highly expressed in neurons and reside mostly in the ER membrane. Not much is known about the involvement of presenilins in ER Ca2+ handling, although it has been shown that PS1 can bind to RyRs (88) and interact with sorcin (476), calsenilin (69), and calpain (372), i.e., proteins known to regulate Ca2+ release channels. Nonetheless, every alteration of PS, be it in the form of its genetic deletion or of expression of mutant (FAD associated) PS, results in impairment of ER function. As a rule, the latter is manifested as an increased amplitude of the Ca2+ response to metabotropic agonists, caffeine, or the SERCA inhibitors TG and CPA (see Table 11). This increase in [Ca2+]i transients is generally regarded as a consequence of an elevated [Ca2+]L. The changes in ER Ca2+ content are not confined to neurons, but are also observed in various peripheral cells such as, e.g., fibroblasts (336).
Manipulations of presenilin expression affect several Ca2+-dependent functions as well as neuronal vulnerability to various types of insults. In transgenic mice expressing mutant PS1, neurons in hippocampal slices show an increased sensitivity to kainate excitotoxicity and demonstrate facilitated LTP induction (564). Both effects are attributable to an increased Ca2+ release from the ER. Similarly, expression of mutant PS1 increased the size of infarction following middle cerebral artery occlusion and reperfusion in vivo and increased cell death in cortical neurons in vitro, when exposed to hypoxia and glucose deprivation (382). Likewise, cultured hippocampal neurons from the same animals displayed increased vulnerability to glutamate excitotoxic shock (195). Inhibition of ER stores by dantrolene protected PS1 mutant-bearing neurons (382). Higher susceptibility of PS1 mutant neurons to various insults correlated with failures of the Ca2+ homeostatic machinery. Recordings of [Ca2+]i in PS1 mutant neurons revealed larger [Ca2+]i elevations induced by hypoxia, glutamate, and intracellular delivery of InsP3 (195, 382, 608). Modulation of PS functions by overexpressing calsenilin, a cytoplasmic protein interacting with PS1 and PS2, enhanced apoptosis and increased the amount of TG-released Ca2+, once more pointing to a possible increase in [Ca2+]L (345).
The PS-mediated alterations of ER function were accompanied by biochemical changes characteristic of the ER stress response. Specifically, expression of mutant PS1 in hippocampal neurons was associated with an abnormal activation of ER resident caspase-12 (87), which, as has been mentioned previously, mediates ER stress-related apoptosis. Expression of an aberrant spliced form of PS2 (PS2V) altered UPR and increased production of Aβ40 and Aβ42 (561). Furthermore, neurons overexpressing the mutant (but not the wild-type) form of PS1 were more susceptible to spontaneous apoptosis and apoptosis induced by ER stress, through decreased UPR (635).
An extremely important achievement in studying the cellular mechanisms of AD occurred recently after several genetic AD models were created. Conceptually these models aimed at the insertion of various forms of APP, which replicated Aβ deposition and plaque formation, or inclusion of mutant PS genes, which produced numerous AD-specific neuropathological changes. The recently discovered tau gene mutations (191), associated with frontotemporal dementia, have provided an additional tool for genetic modeling. A big step forward was made by Frank LaFerla, Mark Mattson, and co-workers who succeeded in creating a triple transgenic mouse bearing presenilin 1 (PS1m146V), amyloid precursor protein (APPSwe), and tau (tauP301L) transgenes (466). These animals possessed several hallmarks of AD such as progressively developing plaques and tangles as well as deficiencies in LTP, which preceded the appearance of the histological markers of AD. The functional changes were clearly age dependent, as both the amplitude of field-stimulation evoked EPSPs and LTP were significantly impaired in 6-mo-old triple transgenic animals, whereas at 1 mo there was no difference between the transgenic mice and wild-type controls (466). However, a detailed analysis of the ER signaling changes associated with this model has yet to be performed.
Although there is no doubt about the AD-related impairment of ER Ca2+ homeostasis, the important question of whether ER changes precede Aβ formation or conversely Aβ malsynthesis causes ER dysfunction remains open.
G. ER Stores and Ageing
The process of physiological ageing (646), which is not associated with any prominent neuronal loss, yet is accompanied by cognitive decline, involves changes in neuronal Ca2+ homeostasis, which comprise alterations of Ca2+ influx, mitochondrial and cytoplasmic Ca2+ buffering, and Ca2+ extrusion (108, 156, 289, 307, 329, 516, 639, 640, 647, 675, 700). The amplitudes of caffeine-induced [Ca2+]i responses are moderately decreased in aged DRG, neocortical, and cerebellar granule cells (290, 291, 674) and in acutely dissociated rat basal forebrain neurons (436). This may reflect the decreased [Ca2+]L, and indeed, the studies mentioned above found a decreased Ca2+ sequestering capacity of the ER in aged nerve cells. In contrast, old cultured hippocampal neurons (101) demonstrated prolonged CICR in response to glutamate exposure, although the size of the caffeine-releasable pool was similar in young and old cultures. Experiments in hippocampal slices isolated from old (22–24 mo) rats found that increased ER Ca2+ release affects LTP induction in the aged hippocampus (321). All in all, however, the age-dependent changes in ER Ca2+ homeostasis are rather mild when compared with the ER alterations in AD. Moreover, no obvious increase in the ER Ca2+ content has been discovered in old neurons, which may reflect a fundamental difference between the physiological processes of brain ageing and age-dependent neurodegeneration.
H. Spinocerebellar Ataxia Type 1
Spinocerebellar ataxia type 1 (SCA1) is an autosomal dominant neurological disorder characterized by loss of Purkinje neurons and neurons in pontine nuclei, which is clinically manifested as a malfunction of the brain stem and ataxia. SCA1 is caused by the expansion of a CAG trinucleotide repeat, which results in an expanded polyglutamine tract in its gene product, ataxin-1 (248). In SCA1 transgenic mice, which carry the expanded allele, a significant reduction in the expression of the InsP3R1 and the SERCA2 pump was observed, although quantitatively parameters of [Ca2+]i signaling and mGluR-mediated Ca2+ release did not differ much from wild-type controls (248).
I. ER and Neuronal Trauma
A direct link between neurotrauma and ER Ca2+ homeostasis has been demonstrated for cultured cortical neurons subjected to mechanical deformation. It appeared that such a crude traumatization of neurons resulted in a remarkable decrease in the proportion of cells able to produce ER-associated Ca2+ release. The percentage of cells responding to caffeine fell from 70% in control to 30%, and the percentage of neurons producing IICR from ∼90% in normal cells to ∼20% already 15 min after the trauma (690). Three hours later the ER Ca2+ release was normalized, although an increase in SOC was observed (689). SOC was determined as SKF96365-sensitive component of the TG-induced [Ca2+]i elevation.
J. ER and Parkinson’s Disease
A possible role for the ER in the pathogenesis of Parkinson’s disease (PD) has been suggested based on investigations of a cellular model of PD, namely, exposure of cells to 6-hydroxydopamine, an agent which causes a type of neurodegeneration similar to that observed in PD (659). When PC12 cells were exposed to 100 μM 6-hydroxydopamine for 8 h, they responded by an upregulation of the expression of numerous genes indicative of UPR (550). Similarly, the ER stress response was evoked by rotenone and 1-methyl-4-phenyl-pyridinium, other agents used for cellular modeling of PD. Finally, the recently identified PD-related G protein-coupled receptor protein, termed the Pael receptor, was found to accumulate in the ER and cause ER-related cell death (623). Whether these biochemical changes become translated into impairments of ER Ca2+ homeostasis remains totally unexplored.
K. ER and Epileptiform Activity
There is some evidence indicating that incubation of hippocampal slices with 1–5 μM TG or with 30–100 μM dantrolene suppresses epileptiform activity in CA3 hippocampal area (549). Similarly, TG is shown to inhibit bicuculline-induced epileptiform activity in hippocampal neuronal networks in culture (599). In a pilocarpine model of epilepsy in hippocampal neuronal cultures, a reduction of TG-sensitive Ca2+ uptake in “epileptic” neurons together with augmented CICR and altered IICR were described (480). Whether these findings reflect any involvement of the ER Ca2+ store in the pathogenesis of epilepsy remains unknown.
L. ER and Huntington’s Disease
Huntington’s disease is a fatal autosomal dominant neurodegenerative disorder, which is characterized by a triad of motor (chorea), cognitive (progressive loss of mental abilities), and psychiatric/behavioral disturbances. This neuropathology is caused by mutations (polyglutamine expansion) of the protein huntingtin (85). The mutant huntingtin sensitizes the InsP3R1 to InsP3, and moreover, transfection of medium spiny striatal neurons with this mutant protein facilitates [Ca2+]i responses to mGluR1/5 activation (628).
VI. CONCLUDING REMARKS
The ER is deeply engaged in neuronal integration. This integration involves coordinated activities of numerous molecular cascades that provide the ER with incoming signals, intra-ER information processing, and generation of output signals which control rapid neuronal signaling and long-lasting adaptive responses (Fig. 23). The movements of Ca2+ across the ER membrane and within the ER lumen are responsible for many of these duties. Regulated Ca2+ release from the ER is implicated in rapid neuronal responses to synaptic inputs and action potentials and is important for synaptic plasticity. Fluctuations of the intra-ER Ca2+ concentration couple these rapid signaling events to protein synthesis and modification. Finally, disruptions of ER Ca2+ homeostasis may be intimately involved in numerous neuropathological processes.
Our understanding of the ER function in nerve cells is clearly incomplete, and future investigations will undoubtedly result in many exciting discoveries in this field.
I am very grateful to Dr. Denis Burdakov for invaluable help with the manuscript editing. I also thank Prof. Marina Bentivoglio, Dr. Antony Galione, Dr. Olga Garaschuk, Dr. Paolo Mazzarello, Prof. David Friel, Prof. Masanobu Kano, Prof. Arthur Konnerth, Dr. Yuri Kovalchuk, Dr. David Ogden, Prof. Alexei Tepikin, and Prof. William Ross for providing me with originals of figures. I thank the reviewers for their meticulous revision of the manuscript and for very constructive comments.
The author’s research was supported by The Wellcome Trust, The Biotechnology and Biological Science Research Council UK, Royal Society, North American Treaty Organization, and International Association of European Community.
Address for reprint requests and other correspondence: A. Verkhratsky, Univ. of Manchester, Faculty of Life Sciences, 1.124 Stopford Building, Oxford Road, Manchester M13 9PT, UK (E-mail:)
↵1 Assistance from Jane Austin also should be acknowledged: “It is a truth universally acknowledged that a single man in possession of a good fortune must be in want of a wife” (Pride and Prejudice, p. 1).
- Copyright © 2005 the American Physiological Society