Cross talk between Ca²⁺ and redox signalling cascades in muscle and neurons through the combined activation of ryanodine receptors/Ca²⁺ release channels

Abstract

Calcium release mediated by the ryanodine receptors (RyR) Ca²⁺ release channels is required for muscle contraction and contributes to neuronal plasticity. In particular, Ca²⁺ activation of RyR-mediated Ca²⁺ release can amplify and propagate Ca²⁺ signals initially generated by Ca²⁺ entry into cells. Redox modulation of RyR function by a variety of non-physiological or endogenous redox molecules has been reported. The effects of RyR redox modification on Ca²⁺ release in skeletal muscle as well as the activation of signalling cascades and transcription factors in neurons will be reviewed here. Specifically, the different effects of S-nitrosylation or S-glutathionylation of RyR cysteines by endogenous redox-active agents on the properties of skeletal muscle RyRs will be discussed. Results will be presented indicating that these cysteine modifications change the activity of skeletal muscle RyRs, modify their behaviour towards both activators and inhibitors and affect their interactions with FKBP12 and calmodulin. In the hippocampus, sequential activation of ERK1/2 and CREB is a requisite for Ca²⁺-dependent gene expression associated with long-lasting synaptic plasticity. The effects of reactive oxygen/nitrogen species on RyR channels from neurons and RyR-mediated sequential activation of neuronal ERK1/2 and CREB produced by hydrogen peroxide and other stimuli will be discussed as well.

1. Introduction

The cytoplasmic free Ca²⁺ concentration ([Ca²⁺]_i), which cells at rest maintain at levels less than or equal to 100 nM, can increase rapidly and transiently in response to various types of stimuli (Berridge et al. 2000). Increases in [Ca²⁺]_i can initiate and modulate different cellular events; these include short-term responses, such as muscle contraction and secretion, as well as long-term responses including cell growth and proliferation and the expression of genes required for memory acquisition and learning (Berridge et al. 2003). Furthermore, an uncontrolled increase in [Ca²⁺]_i can lead to an alteration of intracellular Ca²⁺ homeostasis that may trigger apoptosis or cause irreversible cell injury and necrosis.

Two different Ca²⁺ channels mediate the fast release of Ca²⁺ from intracellular stores, the ryanodine receptors (RyR) and the inositol trisphosphate receptors. Both Ca²⁺ release channels can increase their activity in response to an increase in [Ca²⁺]_i. This response generates Ca²⁺-induced Ca²⁺ release (CICR), a universal cellular mechanism that allows amplification and propagation of the Ca²⁺ signals initially created by Ca²⁺ entry. In particular, ryanodine receptors (RyR)-mediated CICR is required for cardiac muscle contraction and has a central role in secretion and neuronal function, as discussed below.

There are three mammalian RyR isoforms (RyR1, RyR2 and RyR3), which are codified by three different genes localized in three different chromosomes in humans. These proteins were originally identified in skeletal muscle (RyR1), in heart muscle (RyR2) and in brain (RyR3), although it is now known that brain and other tissues express all three mammalian RyR isoforms (Coronado et al. 1994; Furuichi et al. 1994; Mori et al. 2000). Multiple cellular components—including ATP, H⁺, Ca²⁺ and Mg²⁺—and specific proteins, including kinases and phosphatases, as well as endogenous oxidative/nitrosative species, regulate RyR channels (Fill & Copello 2002). In particular, many studies have reported modifications of RyR cysteines by non-physiological redox compounds (for reviews, see Hamilton & Reid 2000; Pessah et al. 2002; Hidalgo et al. 2002). These studies have contributed to the current understanding of how these cysteine modifications change RyR function and affect Ca²⁺ release; they have also promoted active research into how redox compounds of physiological significance affect RyR activity. These endogenous redox components include the free radicals nitric oxide (NO) and superoxide anion, which through enzymatic or non-enzymatic chemical reactions can be readily converted into non-radical species of lower reactivity but longer half-life such as S-nitrosoglutathione (GSNO) or hydrogen peroxide (H₂O₂). All these endogenous redox components can modify RyR function (Hidalgo et al. 2004). Additionally, RyR channels are highly susceptible to modification by other endogenous redox agents, including glutathione (GSH), glutathione disulphide (GSSG), NADH, and by changes in the GSH/GSSG ratio (reviewed in Hidalgo et al. 2004).

RyR channels have been proposed to act as intracellular redox sensors (Eu et al. 2000; Hidalgo et al. 2005). The redox sensitivity of skeletal RyR1 is due to the presence of highly reactive cysteine residues in the RyR1 molecule, which have a pK value of 7.4 (Liu et al. 1994; Pessah et al. 2002) and allow RyR1 redox modification at physiological pH. Additionally, RyR1 exhibit a well-defined redox potential (Feng et al. 2000; Xia et al. 2000); through this property they may be susceptible to changes in cellular redox potential. Oxidative/nitrosative modifications that affect RyR function are likely to modify RyR-mediated Ca²⁺ release from intracellular stores. Hence, changes in cellular redox state or generation of reactive oxygen species (ROS) or reactive nitrogen species (RNS) may result in inhibition or activation of Ca²⁺ signalling cascades. A review of recent data, showing that redox modifications of the RyR channels present in skeletal muscle cells or neurons exert a central role in regulating single channel properties and CICR from vesicles and cells, will be presented next. A discussion of the effects of redox modifications on the function of cardiac RyR2 channels has been presented elsewhere (Hidalgo et al. 2004).

2. Skeletal muscle

(a) Functional effects of modifications of skeletal muscle RyR1 by ROS/RNS

Physiological activation of RyR1-mediated Ca²⁺ release in response to muscle depolarization is a very fast process (ms range) which does not require Ca²⁺ entry into the muscle cells. RyR1 channels open following membrane depolarization, presumably by direct coupling with plasma membrane voltage sensors (Rios & Pizarro 1991). Each of the four homologous 565 kDa RyR1 protein subunits contains approximately 100 cysteine residues (Takeshima et al. 1989). In native RyR1, approximately 50 of these residues appear to be in the reduced state; of these, approximately 10–12 are highly susceptible to oxidation/modification by exogenous sulphydryl (SH) reagents (Sun et al. 2001a). Endogenous redox-active molecules, including molecular oxygen (Eu et al. 2000) and GSSG (Zable et al. 1997; Sun et al. 2001a,b; Aracena et al. 2003), enhance RyR activity. Likewise, H₂O₂ markedly activates single RyR2 channels (Boraso & Williams 1994) or RyR1 channels maintained under redox control (provided by controlled cytoplasmic and luminal GSH/GSSG ratios; Oba et al. 2002a). Furthermore, superoxide anion generation by a NADH oxidase activity that is present in heavy sarcoplasmic reticulum (SR) vesicles and which copurifies with RyR1 channels is presumably responsible for the activation of [³H]-ryanodine binding (a reflection of increased RyR1 activity) produced by 1 mM NADH (Xia et al. 2003).

In certain conditions, NO or NO donors enhance RyR1 activity, while in others they exert an inhibitory effect (Stoyanovsky et al. 1997; Suko et al. 1999; Sun et al. 2001a,b, 2003; Eu et al. 2003). In particular, submicromolar NO concentrations activate RyR1 by S-nitrosylation of a single cysteine residue (Cys-3635); this reaction occurs only at low (physiological) pO₂ but not at ambient pO₂ (Sun et al. 2003). Likewise, in situ generation of NO stimulates whole muscle contractility at low pO₂, but inhibits contractility at higher pO₂ (Eu et al. 2003). Contrary to NO, GSNO stimulates RyR1 channel activity independently of pO₂ through S-nitrosylation of a few critical SH residues other than Cys-3635 (Sun et al. 2003). We have shown that S-nitrosylation of RyR1 enhances its activation by Ca²⁺, but does not affect the strong inhibitory effect of Mg²⁺ on CICR (Aracena et al. 2003). Both GSNO and the NO donor NOR-3 produce a significant increase in the S-nitrosylation of RyR1 over the endogenous levels and affect the interactions of RyR1 with FKBP12 and calmodulin, as detailed below. In addition, GSNO or the combination of millimolar GHS plus micromolar H₂O₂ enhance significantly the endogenous S-glutathionylation of RyR1 (Aracena et al. 2005), a modification that decreases specifically RyR1 inhibition by Mg²⁺ without affecting its activation by Ca²⁺ (Aracena et al. 2003). These findings suggest that RyR1 possesses cysteine residues with different susceptibilities to S-glutathionylation or S-nitrosylation, and that these cysteine modifications cause independent effects on RyR1 channel activation by Ca²⁺ or inhibition by millimolar Mg²⁺ or Ca²⁺ (figure 1). The implications of these findings for redox enhancement of CICR in skeletal muscle, as well the possible sources and the activation mechanisms of ROS/RNS generation in living skeletal muscle fibres are discussed in §2b.

Figure 1.

Calcium dependence of the activity of single RyR1 channels. The theoretical curves were obtained as described (Marengo et al. 1998), except that the maximal value of open probability was taken as 1.0 and both the activation and inactivation functions were considered non-cooperative. The solid black curve represents the response to cytoplasmic (cis) [Ca²⁺] of native RyR channels, which is obtained as the combination of two independent functions (dotted lines), an activation function with K_a=5 μM and an inactivation function with K_i=41 μM. The activation function depicts the channel activation produced by Ca²⁺ binding to the high-affinity site(s), while the inactivation function describes binding of Ca²⁺ or Mg²⁺ to the low-affinity divalent cation inhibitory site(s). Incubation of RyR1 channels with reducing agents increases K_a from 5 to 50 μM and decreases K_i from 41 to 2.2 μM (grey dashed lines), yielding the bell-shaped function represented by the solid light grey line. In contrast, oxidation/alkylation of free sulphydryl residues present in native RyR1 channels decreases K_a from 5 to 0.6 μM and increases K_i from 41 μM to 5 mM (grey dotted lines); these values yield the bell-shaped function represented by the solid dark grey line. RyR1 S-nitrosylation produces only a decrease in K_a, whereas RyR1 S-glutathionylation produces only an increase in K_i.

In mammalian skeletal muscle, RyR1 channels interact with several different accessory proteins, which contribute to tightly regulate their activity. These proteins include the skeletal L-type Ca²⁺channel dihydropyridine receptor, which acts as voltage sensor during excitation–contraction coupling, triadin, calmodulin, FKBP12 (a 12 kDa FK506 binding protein) and the SR luminal protein calsequestrin (MacKrill 1999; Fill & Copello 2002). Calmodulin in its apocalmodulin (Ca²⁺-free) form enhances RyR1 channel activity, whereas when present as Ca²⁺-calmodulin it behaves as a channel inhibitor (Tripathy et al. 1995; Rodney et al. 2000, 2001). At physiological tissue pO₂ (approx. 10 mmHg), submicromolar NO activates RyR1 by S-nitrosylation of a single cysteine residue (Cys-3635) (Sun et al. 2003) without affecting Ca²⁺-calmodulin binding to RyR1 (Eu et al. 2000). The specific S-nitrosylation of Cys-3635, which resides within a calmodulin binding domain, by NO or the NO donor NOC-12 does not take place at ambient pO₂; S-nitrosylation of Cys-3635 activates RyR1 and reverses its inhibition by Ca²⁺-calmodulin (Sun et al. 2003). In contrast, GSNO activates RyR1 by oxidation and S-nitrosylation of thiols other than Cys-3635, and calmodulin is not involved (Sun et al. 2003). In triad vesicles isolated from skeletal muscle, we have found that S-nitrosylation or S-glutathionylation of triad proteins, including RyR1, increase the K_d of [³⁵S]-Ca²⁺-calmodulin and of [³⁵S]-apocalmodulin binding to triads without affecting maximal binding capacity (B_max). Moreover, the combined S-glutathionylation and S-nitrosylation induced by GSNO increases by fourfold the K_d of [³⁵S]-Ca²⁺-calmodulin binding to triads and abolishes the binding of [³⁵S]-apocalmodulin (Aracena et al. 2005).

FKBP12 is another RyR1 regulatory protein that may specifically coordinate RyR1 channel gating (Gaburjakova et al. 2001). Removal of this protein with rapamycin or FK506 leads to RyR1 activation and promotes the emergence of sub-conductance states in single RyR1 channels incorporated in lipid bilayers (Ma et al. 1995; Kaftan et al. 1996). In addition, tissue-specific FKBP12 knockout mice exhibit altered orthograde and retrograde signalling between the voltage sensor and RyR1, leading to altered L-type currents (Tang et al. 2004). We have found that [³⁵S]-FKBP12 binds to RyR1 present in triads or heavy SR vesicles with similar affinity (K_d=13.1 nM) (Aracena et al. in press). These results indicate that RyR1–voltage sensor interactions do not affect the high-affinity binding of FKBP12 to RyR1. RyR1 S-nitrosylation, but not S-glutathionylation, specifically increases fivefold the K_d value of FKBP12 binding to RyR1 without affecting B_max (Aracena et al. 2005).

Since both FKBP12 and Ca²⁺-calmodulin inhibit RyR1 activity, decreased binding of these proteins to RyR1 may contribute to the activation of CICR release kinetics induced by S-glutathionylation and S-nitrosylation (Aracena et al. 2003). Furthermore, in vivo conditions that promote the combined S-nitrosylation and S-glutathionylation of RyR1 are bound to alter both FKBP12 and calmodulin binding to RyR1, and thus may alter excitation–contraction coupling in skeletal muscle fibres.

(b) Relevance of RyR1 modifications induced by ROS/RNS to skeletal muscle function

The results discussed so far indicate that channel S-nitrosylation enhances Ca²⁺ release by increasing the affinity of Ca²⁺ binding to RyR1 (Aracena et al. 2003). Mammalian skeletal muscle contains endogenous sources of NO, and it has been reported that endogenous NO generated by nNOS induces a significant enhancement of whole-muscle contractility at low (physiological) pO₂ but not at ambient pO₂ (Eu et al. 2003). Moreover, cysteine RyR1 modifications that enhance channel activation by Ca²⁺ and ATP (Marengo et al. 1998; Oba et al. 2002b; Aracena et al. 2003; Bull et al. 2003), and especially those that reduce the powerful inhibitory effect of Mg²⁺ on RyR1-mediated Ca²⁺ release (Donoso et al. 2000; Aracena et al. 2003) may stimulate CICR in living skeletal muscle fibres. In vitro, submillimolar [Mg²⁺] strongly inhibits RyR1 channels (Laver et al. 1997) and RyR1-mediated Ca²⁺ release (Meissner et al. 1986; Moutin & Dupont 1988; Donoso et al. 2000; Aracena et al. 2003). Mg²⁺ inhibition may be a contributing factor to the reported lack of sparks due to spontaneous CICR in skeletal muscle fibres (Shirokova et al. 1998).

(c) RyR1 S-glutathionylation

Endogenous generation of ROS or GSNO (via NO generation) may enhance RyR S-glutathionylation, and thus may overcome the strong inhibitory effect of Mg²⁺ on CICR. Skeletal muscle fibres generate ROS following physiological contraction (Bejma & Li 1999; Reid & Durham 2002) as well as during heat stress conditions (Stofan et al. 2000). Skeletal muscle homogenates contain a constitutively active non-phagocytic NAD(P)H oxidase complex that generates superoxide anion (Javesghani et al. 2002). Enzymatic or spontaneous dismutation of superoxide anion generates hydrogen peroxide, which markedly activates RyR1 in vitro (Favero et al. 1995; Oba et al. 2002a) but does not seem to play a significant role in the post fatigue recovery process (Oba et al. 2002a). We have found the membrane bound components of NAD(P)H oxidase (gp91^phox and p22^phox) in skeletal muscle transverse tubules, which are absent from heavy SR vesicles (figure 2). Activation of this transverse tubular NAD(P)H oxidase, which generates superoxide anion and hydrogen peroxide, leads to S-glutathionylation and stimulation of the neighbouring RyR1 in isolated triad vesicles (Hidalgo et al. in preparation). It has been reported that exogenously added hydrogen peroxide stimulates caffeine-induced contractions in skinned skeletal muscle fibres; in contrast, addition of hydrogen peroxide does not affect action potential-generated contractions (Posterino et al. 2003). These results support activation by hydrogen peroxide of RyR1-mediated CICR but not of depolarization-induced Ca²⁺ release. Furthermore, in skinned skeletal muscle fibres hydrogen peroxide activates contraction without producing an increase in [Ca²⁺]_i (Andrade et al. 1998) or even after causing [Ca²⁺]_i to decrease (Andrade et al. 2001). Hence, a definite demonstration of the relevance of RyR1 modifications by hydrogen peroxide in the context of physiological gating of the channel during excitation–contraction coupling is still missing (Lamb & Posterino 2003). It is not known, however, if reducing or oxidizing agents externally added to skeletal muscle cells effectively cause RyR1 redox modifications. This issue is important since RyR1 present endogenous levels of S-glutathionylation when isolated (Aracena et al. 2005), an indication that RyR1s effectively function in vivo with some S-glutathionylated cysteine residues. Although the cellular sources of this RyR1 redox modification are not established, the transverse tubule NAD(P)H oxidase may contribute to this modification. This enzyme may provide direct redox control of RyR1 function, so that increased muscle activity may result in enhanced RyR1 S-glutathionylation and Ca²⁺ release. We have preliminary data indicating that depolarization of muscle cells in culture activates this transverse tubule NAD(P)H oxidase, which in turn stimulates RyR1 activity (Espinosa et al. in preparation; P. Aracena, C. Gilman, S. L. Hamilton & C. Hidalgo, unpublished observations). It remains to be investigated, however, whether RyR1 S-glutathionylation increases following physiological activation of skeletal muscle fibres, and whether this specific RyR1 redox modification affects the physiological excitation–contraction coupling process.

Figure 2.

Western blot analysis to test the presence of NAD(P)H subunits in triads, transverse tubules (TT) and junctional sarcoplasmic reticulum (heavy SR) vesicles devoid of transverse tubules. Isolated triad vesicles contain 10–15% TT still attached to junctional SR membranes. All membrane fractions were isolated from mammalian skeletal muscle as described (Barrientos & Hidalgo 2002). Mouse monoclonal anti-gp91^phox, anti-p22^phox, anti-p47^phox and anti-p67^phox antibodies were a kind gift from Dr Mark T. Quinn (Veterinary Molecular Biology, Montana State University, Bozeman, MT).

(d) RyR1 S-nitrosylation

There is experimental evidence indicating that NO-induced RyR1 cysteine S-nitrosylation affects muscle function. RyR1 channel S-nitrosylation caused by endogenous NO generation is likely responsible for the reported increase in skeletal muscle contractility observed at physiological pO₂ (Eu et al. 2003). In single skeletal muscle fibres from mouse, challenged by short membrane depolarizations under voltage-clamp conditions, NO donors induce a progressive increase in resting [Ca²⁺]_i (Pouvreau et al. 2004). High NO levels may produce continuous Ca²⁺ leak from the SR, possibly mediated by redox-modified RyR1s that do not close once they are activated by membrane depolarization. This SR Ca²⁺ leak and the resulting increase in resting [Ca²⁺]_i may be important in mediating the effects of excess NO on voltage-activated Ca²⁺ release.

3. Neurons

Several neuronal functions, including synaptic plasticity and gene expression, require transient elevation of [Ca²⁺]_i (Berridge 1998; Berridge et al. 2000). This increase is initially produced by Ca²⁺ influx through plasma membrane voltage- or receptor-activated Ca²⁺ channels following neuronal stimulation. In addition, Ca²⁺ entry can stimulate Ca²⁺ release from intracellular stores via CICR or depolarization-induced Ca²⁺ release. A variety of experimental findings indicate that Ca²⁺ release from stores elicits diverse neuronal responses (Simpson et al. 1995; Chameau et al. 2001; Meldolesi 2001; Verkhratsky & Petersen 2002; Bouchard et al. 2003; Ouardouz et al. 2003; Pape et al. 2004; Gafni et al. 2004). Through CICR neurons can amplify and propagate the initial Ca²⁺ entry signal, leading to activation of most calcium-regulated transcription factors (Berridge 1998). In particular, RyR-mediated Ca²⁺ release may have a role in synaptic plasticity and gene expression in neurons (Berridge 1998; Futatsugi et al. 1999; Berridge et al. 2000; Carafoli 2002; Carrasco et al. 2004; Verkhartsky 2004). RyR-mediated Ca²⁺ release may also be involved in causing neurodegenerative processes, as discussed below.

Of the three RyR isoforms expressed in different brain regions, RyR2 is the most abundant (Furuichi et al. 1994; Giannini et al. 1995; Mori et al. 2000). Despite the emerging importance of RyR-mediated Ca²⁺ release for brain function, there is limited information on how these different RyR isoforms contribute to the generation and/or regulation of the Ca²⁺ signals that underlie diverse Ca²⁺ dependent neuronal functions. It is known, however, that membrane depolarization activates RyRs in several neuronal types (Chavis et al. 1996; Mouton et al. 2001; De Crescenzo et al. 2004), Yet, the properties and regulation of RyR channels from brain have been less studied than those of their skeletal or cardiac counterparts.

(a) Redox modulation of brain RyR

CICR relies on the fact that RyR are activated by an increase in cytoplasmic Ca²⁺ concentration, from the submicromolar to the micromolar range; in cells this activation occurs in the presence of ATP and Mg²⁺. We have studied at the single-channel level the activation induced by Ca²⁺ and ATP of native and oxidized RyR channels from rat brain endoplasmic reticulum. Native RyR channels from rat brain cortex incorporated in planar lipid bilayers display three different Ca²⁺ dependencies (Marengo et al. 1996), which can be interchanged through modifications of RyR redox state (Marengo et al. 1998; Hidalgo et al. 2004). In brief, highly reduced channels (a condition in which most RyR1 cysteine residues have free SH groups) respond poorly to Ca²⁺ activation, whereas increasing cysteine oxidation/alkylation increases the channel response to micromolar [Ca²⁺] and decreases the inhibitory effect of higher [Ca²⁺]. Noteworthy, reducing agents reverse all these changes whereas ATP differentially activates RyR channels depending on their Ca²⁺ response, so that more oxidized channels from brain require less ATP to attain maximal activation (Bull et al. 2003). We have found that modification of brain RyR cysteine residues with thimerosal decreases the inhibitory effect of Mg²⁺ on RyR activity whereas reducing agents have the opposite effect (R. Bull et al. in preparation). Changes in the redox state of RyR channels from brain affect as well RyR modulation by glucosylceramide (Lloyd-Evans et al. 2003), a type of glycosphingolipid that may be implicated in the cell death observed in Alzheimer's disease (Cutler et al. 2004) and HIV dementia (Haughey et al. 2004). Interestingly, the finding that glucosylceramide augments agonist-stimulated Ca²⁺ release via RyR channels (Lloyd-Evans et al. 2003) may be relevant to understand the development of pathological conditions that entail a faulty regulation of Ca²⁺ homeostasis in pathological conditions.

(b) Participation of RyR-mediated Ca²⁺ release in long-term potentiation induction and CREB activation in the hippocampus

Activation of the nuclear transcription factor CREB (cAMP/calcium response element binding protein) is crucial for several neuronal functions, including activity-dependent synaptic plasticity (Silva et al. 1998; Lonze & Ginty 2002). Activation of CREB and the ensuing CRE-mediated transcription are required for induction of long-term potentiation (LTP), a reversible increase in synaptic transmission inducible in several brain regions in response to tetanic stimulation of afferent fibres. LTP is considered a cellular model of synaptic plasticity that may underlie higher brain functions such as learning and memory (Lynch 2004). A number of genes activated by neuronal electrical activity, among them c-fos and BDNF, have functional CRE sequences in their promoters (Sheng et al. 1990; Tao et al. 1998); in the last few years these findings have prompted many studies on the cellular factors that regulate CREB activation.

Participation of Ca²⁺ release from intracellular stores in LTP induction has been described (Auerbach & Segal 1994; Wang et al. 1996; Reyes-Harde et al. 1999; Lu & Hawkins 2002; Lauri et al. 2003; Matias et al. 2003; Lynch 2004). LTP induction in most hippocampal synapses requires a rise in intracellular postsynaptic [Ca²⁺] mediated by N-methyl-d-aspartate (NMDA) receptors, with contributions from L-type channel activation and intracellular Ca²⁺ stores (Matias et al. 2003; Lynch 2004). In the CA1 hippocampal region, RyR blockade significantly reduces tetanic-, NO- and cGMP-induced LTP, and decreases P-CREB immunofluorescence in postsynaptic neurons of hippocampal slices (Lu & Hawkins 2002). These results point to a well-defined postsynaptic location of the RyR channels involved in hippocampal LTP induction. A presynaptic effect of NO resulting in induction of long-term depression (LTD) as well as LTP in the same hippocampal region has also been described (Reyes-Harde et al. 1999). A contribution of RyR, in this case presynaptic, has been suggested by the marked reduction of LTD produced by ryanodine when used at inhibitory concentrations. Interestingly, a switch from LTD to LTP was induced at the hippocampal dentate gyrus by a low stimulatory ryanodine concentration, suggesting that RyR-mediated Ca²⁺ release plays a significant role in LTP induction (Wang et al. 1996).

Presynaptic RyR also contribute to LTP generation in mossy hippocampal fibres (Lauri et al. 2003). This form of LTP is independent of NMDA receptors; instead, Ca²⁺ entry is mediated by presynaptic kainate glutamate receptors, which initiate a calcium-signalling cascade involving RyR-mediated Ca²⁺ release that is abolished by inhibitory ryanodine concentrations. In addition, caffeine (10 mM) can evoke LTP in CA1 hippocampal neurons, which is different to the classical LTP since it does not require postsynaptic receptor activation or Ca²⁺ increase in this area (Martin & Buño 2003). This new form of LTP is generated presynaptically, through the interaction of caffeine with presynaptic purinergic receptors and RyR channels. The resulting increased Ca²⁺ signals enhance neurotransmitter release.

In the hippocampus, Ca²⁺-dependent activation of the Ras/ERK (extracellular signal-regulated kinase) pathway is essential for long-term CREB phosphorylation and CREB-dependent transcription of genes involved in synaptic plasticity (Hardingham & Bading 1999; Impey & Goodman 2001; Thiels & Klann 2001; Wu et al. 2001; Adams & Sweatt 2002; Dolmetsch 2003; Lynch 2004). Furthermore, functionally active neurons display increased metabolic activity plus increased oxygen consumption and generation of ROS/RNS (Davis et al. 1998; Yermolaieva et al. 2000). Several studies indicate that ROS/RNS may regulate synaptic plasticity. Activity-dependent nitric oxide generation has been associated with synaptic plasticity and Ca²⁺ signalling (Peunova & Enikolopov 1993; Yermolaieva et al. 2000), including Ca²⁺-dependent activation of CREB-phosphorylation in the hippocampus (Lu & Hawkins 2002). Activation of NMDA receptors generates superoxide anion in rat hippocampal pyramidal neurons in culture and in brain hippocampal slices (Bindokas et al. 1996), while cell-permeable scavengers of superoxide anion block LTP induction in area CA1 of the hippocampus (Klann 1998). Noteworthy, transgenic mice that overexpress extracellular superoxide dismutase (a superoxide anion scavenger) exhibit impaired LTP in hippocampal area CA1, despite normal LTP in area CA3, and also show defects in memory acquisition (Thiels et al. 2000). These findings strongly suggest that superoxide anion (or its derivative, hydrogen peroxide) is a required signalling molecule for LTP induction and normal neuronal function. Yet, dissimilar results on the effects of hydrogen peroxide on hippocampal LTP induction have been reported (Kamsler & Segal 2004). Initial augmentation, and subsequent long-lasting depression, of population spikes and excitatory postsynaptic potentials by hydrogen peroxide (0.3–2.9 mM) has been described in the CA1 region of rat hippocampal slices (Katsuki et al. 1997). In contrast, low concentrations of hydrogen peroxide (1 μM) cause a twofold increase in tetanic LTP and enhance NMDA-independent LTP in the hippocampus compared to controls (Kamsler & Segal 2003).

The preceding discussion of the literature indicates that LTP induction in hippocampal neurons entails a combined increase of [Ca²⁺]_i, ROS and RNS. All these factors can potentially contribute to enhance RyR-mediated Ca²⁺ release; the resulting amplified Ca²⁺ signals may in turn enhance the long-lasting Ca²⁺-dependent ERK/CREB phosphorylation cascade required for LTP induction. Hence, it is of interest to investigate whether ROS/RNS stimulate the Ca²⁺-dependent Ras/ERK pathway through stimulation of RyR-mediated CICR. To this aim, we have studied the effects of hydrogen peroxide and membrane depolarization on ERK/CREB phosphorylation in neurons with functional or inhibited RyR. We exposed N2a and hippocampal cells in primary culture cells to hydrogen peroxide (Carrasco et al. 2004; Kemmerling et al. submitted). In both cell types, exogenous hydrogen peroxide activates CREB and ERKs and promotes RyR-dependent Ca²⁺ release; these effects are independent of extracellular Ca²⁺ and are inhibited significantly by ryanodine concentrations that block RyR-mediated Ca²⁺ release. Furthermore, thapsigargin antagonizes the increase in P-CREB levels induced by hydrogen peroxide, which also promotes RyR s-glutathionylation (Kemmerling et al. submitted). These results suggest that oxidative stimulation of RyR-mediated Ca²⁺ release is an intermediate step in the activation of the ERK/CREB phosphorylation cascade required for synaptic plasticity (figure 3). We also found that depolarization induced by high K⁺ concentration stimulates CREB phosphorylation in the neuroblastoma cell line N2a (Carrasco et al. 2004). Inhibitory concentrations of ryanodine partially reduced depolarization-induced CREB phosphorylation, suggesting that RyR-mediated Ca²⁺ release contributes to the observed stimulation.

Figure 3.

Proposed model to illustrate the potential role of neuronal RyR as coincidence detectors of a simultaneous increase in [Ca²⁺] and ROS/RNS. As discussed in the text, stimulation of postsynaptic NMDA receptors (NMDAR) in hippocampal neurons elicits Ca²⁺ entry and promotes ROS generation; the resulting localized [Ca²⁺] increase stimulates the Ca²⁺-dependent neuronal NO synthase that is attached to the postsynaptic membrane, resulting in the generation of NO and possibly of other NO-derived RNS. The model proposes that the concomitant local increases in Ca²⁺ and ROS/RNS activate jointly RyR-mediated Ca²⁺ release from the endoplasmic reticulum (ER), generating larger Ca²⁺ signals. The resulting RyR-mediated Ca²⁺ signals would stimulate Ca²⁺-dependent signalling cascades, such as the Ras MAP kinase pathway, producing increased ERK1/2 phosphorylation. Translocation of phospho-ERK1/2 (P-ERK1/2) to the nucleus (N) would stimulate the sustained phosphorylation of the transcription factor CREB, which is a requisite for transcription of genes involved in synaptic plasticity. According to this model, cross talk between Ca²⁺ and redox signals through joint stimulation of RyR-mediated Ca²⁺ release may play a key role in hippocampal synaptic plasticity. Furthermore, conditions that promote oxidative/nitrosative stress may result in excessive stimulation of RyR-mediated Ca²⁺ release that if unchecked can provoke pathological responses and neuronal death.

4. Conclusions

Through either depolarization-induced Ca²⁺ release or via CICR, RyR channels have a pivotal role in the generation or amplification of Ca²⁺ signals that are required for muscle contraction and neuronal plasticity. Endogenous activation of RyR channels by cellular ROS/RNS may represent a physiological mechanism of cross talk between Ca²⁺ and redox signalling pathways. Thus, cells may use redox-modulated RyR-mediated Ca²⁺ release as an additional mechanism to either amplify or inhibit Ca²⁺ signals as needed for a specific response. In neurons, generation of ROS/RNS such as hydrogen peroxide or NO, which may act as diffusible signal molecules in synaptic plasticity, may modify cellular processes that depend on RyR-mediated Ca²⁺ release from the ER, including LTP and LTD, and presumably learning and memory. Oxidative stress and alterations in Ca²⁺ homeostasis may contribute to neuronal apoptosis and excitotoxicity, which may underlie the pathogenesis of several neurodegenerative disorders (Mattson 2000; Mattson & Chan 2003). In particular, RyR channels may be involved in the pathophysiology of neurodegeneration in Alzheimer's disease (Kelliher et al. 1999; Chan et al. 2000; Mungarro-Menchaca et al. 2002). As discussed here, oxidative stress is likely to enhance RyR-mediated CICR in neurons. Thus, redox modification of RyR channels may have physiological and pathological consequences for neuronal function.