Mechanistic Role of Calpains in Postischemic Neurodegeneration

Physiologic Activity

The role of calpains in physiologic processes has been extensively reviewed (Goll et al, 2003) and encompasses functions including cell cycle regulation, differentiation, cell migration, adhesion, and signal transduction. A similarly wide range of physiologic functions has been attributed to calpain within the brain and neuronal culture systems. Calpain has been implicated in the differentiation of neuronal cell lines (Grynspan et al, 1997), neurite outgrowth in hippocampal neuronal culture (Song et al, 1994), as well as synaptic remodeling and long-term potentiation (Tomimatsu et al, 2002).

Through the use of transgenic mice and RNA interference (RNAi), there is an increasing literature concerning the isoform specificity of some physiologic functions of calpains. Perhaps the most convincing evidence for a functional difference between μ- and m-calpain is the fact that mice with a knockout of the calpain 1 catalytic subunit have been reported to have defects in platelet function (Azam et al, 2001) and to have normal long-term potentiation (Grammer et al, 2005), whereas knockout of calpain 2 (Dutt et al, 2006) or the common regulatory subunit (Zimmerman et al, 2000; Arthur et al, 2000) is embryonic lethal. Both calpain-3 and calpain-10 knockouts are viable, but the effects of these gene deletions on brain physiology and pathology have not been reported.

Studies using antisense oligonucleotides or RNAi have similarly shown isoform-specific physiologic effects. μ-Calpain has a demonstrated role in muscle cell differentiation (Moyen et al, 2004) and cell motility and migration (Satish et al, 2005; Wu et al, 2006). m-Calpain is involved in myoblast fusion (Balcerzak et al, 1995), membrane protrusion (Franco et al, 2004), and chromosome alignment during mitosis (Honda et al, 2004). Antisense oligonucleotides targeting the calpain 1 catalytic subunit reduced the effects of n -methyl-d-aspartic acid (NMDA) excitotoxicity in hippocampal slice culture (Bednarski et al, 1995). Similarly, RNAi for calpain 1, but not calpain 2, has been shown to be protective in dissociated neuronal culture (Cao et al, 2007). Unfortunately, the effects of translational suppression of calpains 3 and 10 on brain injury have not been reported, nor the effects of isoform-specific knockdown on in vivo injury.

Localization

The localization of calpain within the central nervous system has been repeatedly investigated. Immunohistochemical results have varied widely, with most studies of μ-calpain agreeing that it is expressed in neurons, and some noting glial expression as well. Specific brain regions identified include cerebellar Purkinje cells (Siman et al, 1985; Hamakubo et al, 1986), spinal cord nuclei (Siman et al, 1985; Hamakubo et al, 1986; Fukuda et al, 1990), cortex (Perlmutter et al, 1990), regions of diencephalon (Siman et al, 1985; Perlmutter et al, 1990), caudate nucleus (Siman et al, 1985; Perlmutter et al, 1990), and both granular and pyramidal neurons within the hippocampus (Siman et al, 1985; Hamakubo et al, 1986; Perlmutter et al, 1988; Fukuda et al, 1990). One immunohistochemical study specifically examining m-calpain noted that its expression was limited largely to glia (Hamakubo et al, 1986), whereas another observed m-calpain in hippocampal interneurons (Fukuda et al, 1990). Variable results such as these are common among calpain localization studies: in another example, Siman et al (1985) identified μ-calpain only in the hippocampal dentate gyrus, whereas Hamakubo et al (1986) and Fukuda et al (1990) reported that it is present in pyramidal cells throughout the hippocampus. Furthermore, Hamakubo et al (1986) observed only m-calpain in glial cells, whereas Perlmutter et al (1990) observed μ-calpain in glia. The major drawback of these studies, and undoubtedly the source of their variable results, is the fact that each used a different antibody. Although the isoform specificity of these antibodies was established in western blots, it is not clear as to what level of cross-reactivity each might have for other calpain isoforms when used in immuno-histochemical studies.

To avoid this problem, more recent studies have examined mRNA, rather than protein, expression. The most definitive of these used a combination of reverse transcriptase-polymerase chain reaction and in situ hybridization to examine the relative abundance and localization of mRNAs for calpains 1 and 2 and calpastatin (Li et al, 1996). Expression of calpain 2 mRNA was found to be 15-fold higher than calpain 1 in whole-brain homogenates. Calpastatin mRNA expression was three-fold higher than calpain 1. Calpain 1 and calpastatin message were observed to be diffused throughout the entire brain in both neurons and glia, whereas message for calpain 2 localized to distinct neuronal populations, including all cornu ammonis regions of hippocampus, cortical pyramidal neurons, and cerebellar Purkinje cells.

Less well studied is the subcellular localization of various components of the calpain system. Early immunoelectron microscopy studies were able to identify μ-calpain throughout neuronal and glial cytoplasm, typically associated with cytoskeletal elements. In neurons, staining extended into axons and dendrites, and, more specifically, was associated with dendritic spines and postsynaptic densities (Perlmutter et al, 1988). As a result of this study and of rough localization claims in earlier immunohistochemistry studies, calpain has typically been considered a ‘cytoplasmic’ protein. More recent immunoelectron microscopic studies have also placed calpain in the cytoplasm, but associated with the membranes of the endoplasmic reticulum (ER) and Golgi apparatus (Hood et al, 2003). Calpains' involvement in nuclear processes such as mitosis (Honda et al, 2004), however, suggests that calpain at least possesses the ability to translocate to the nucleus. In fact, fluorescein-tagged μ-calpain has been shown to be selectively transported to the nucleus under conditions known to allow energy-dependent nuclear transport of proteins (Mellgren and Lu, 1994), both μ- and m-calpain have been identified in the nuclear fraction of rabbit hippocampal homogenates (Ostwald et al, 1994), and m-calpain has been observed in cerebellar granule cell nuclei by immunohistochemistry and subcellular fractionation (Tremper-Wells and Vallano, 2005). Additionally, a recent study showed the presence of μ-calpain within isolated mitochondria (Garcia et al, 2005). Further studies have localized μ-calpain specifically to the mitochondrial intermembrane space (Cao et al, 2007). Calpain 10 has also been localized to mitochondria (Arrington et al, 2006), whereas calpain 3 is present in the cytoplasm and nucleus (Marcilhac et al, 2006). The subcellular localization of calpain 5 within the central nervous system has not been reported.

In addition to the localization of calpain under baseline conditions, there have been numerous studies of calpain translocation throughout the cell. Activation of calpain is associated with translocation to the plasma membrane in neutrophils (Pontremoli et al, 1989) and platelets (Ariyoshi et al, 1993). The importance of calpain translocation in the brain is not clear, and there is some evidence that calpain may be present at the membrane before any excitotoxic insult, and therefore no translocation is necessary before calpain activation (Hewitt et al, 1998). Given the similar substrate profiles of the two major calpain isoforms, it is likely that localization and translocation within the cell play a key role in differentiating their physiologic roles, and may impact their function in pathology as well.

Plasma Membrane and Synapse

A major component of the postischemic injury cascade in the brain is the membrane depolarization that occurs after ATP depletion (for review, see Neumar, 2000). This membrane depolarization results in the opening of voltage-gated sodium and calcium channels and in the release of excitatory neurotransmitters, such as glutamate. Transmitter release further alters ion homeostasis as it results in activation of calcium-permeable NMDA receptors and sodium-permeable AMPA receptors. Resulting loss of the normal Na gradient then leads to reversal of the Na⁺/Ca²⁺ exchanger. Although these events are likely sufficient to activate calpain, the initial change in ion balance is reversible within minutes after a mild or moderate ischemic insult. However, there is a secondary disruption of ion homeostasis that occurs in many neurons between 4 and 48 h after reperfusion. This further disruption of calcium regulatory processes is necessary to produce the delayed calpain activation that potentially contributes to delayed postischemic neurodegeneration. Calpain itself may contribute to the sustained or secondary calcium overload by cleaving and altering the function of many of these proteins during the recovery period.

Plasma Membrane: The most obvious effects of calpain activity on calcium regulation come through its cleavage of l-type calcium channels, the Na⁺/Ca²⁺ exchanger, and the plasma membrane Ca²⁺ ATPase (Figure 2A). Calpain-mediated truncation of the l-type calcium channel occurs both in vitro (De Jongh et al, 1994) and in hippocampal neurons exposed to NMDA (Hell et al, 1996). This truncation removes the major site of channel phosphorylation and results in a persistent increase in Ca²⁺ influx (Wei et al, 1994). A second potential contributor to calpain-mediated calcium dysregulation at the plasma membrane is the Na⁺/Ca²⁺ exchanger, which is a major extruder of intracellular calcium. After in vitro excitotoxicity and in vivo brain ischemia, the Na⁺/Ca²⁺ exchanger undergoes a cleavage that is prevented by calpain inhibitors or calpastatin overexpression (Bano et al, 2005). The cleaved exchanger is unable to expel calcium.

OPEN IN VIEWER

Figure 2

Subcellular localization of calpain pathology in postischemic neurons. Postischemic calpain activity occurs in a wide range of subcellular regions including the synapse and plasma membrane (A), endoplasmic reticulum and lysosomes (B), and mitochondria and nucleus (C). The substrate profile within each location suggests multiple distinct mechanisms by which calpain activity can contribute to postischemic neurodegeneration. Calpain-mediated proteolysis of calcium regulatory proteins at the synapse, plasma membrane, and endoplasmic reticulum could work synergistically in a feed-forward pathway of disrupted calcium homeostasis. Calpain-mediated proteolysis of signaling proteins in the cytosol, mitochondria, and nucleus potentially modulates caspase-dependent and caspase-independent cell death. Finally, calpain-mediated proteolysis of cytoskeletal and membrane proteins could result in structural failure of the plasma membrane and lysosomal membranes. AMPA, alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; NMDA, n -methyl-d-aspartic acid receptor; mGluR1α, metabotropic glutamate receptor 1α; CamKIIα, calmodulin-dependent protein kinase IIα; PKC, protein kinase C; nNOS, neuronal nitric oxide synthase; NO, nitric oxide; MAP2, microtubule-associated protein 2; PMCA, plasma membrane calcium ATPase; NCX, sodium—calcium exchanger; RYR, ryanodine receptor; IP₃R, inositol 3-phosphate receptor; SERCA, sarcoplasmic/endoplasmic reticulum calcium ATPase; AIF, apoptosis-inducing factor; CytC, cytochrome c; APAF-1, apoptosis-activating factor 1; β-cat, β-catenin; GSK-3, glycogen synthase kinase 3; CDK5, cyclin-dependent kinase 5; CRMP, collapsin response mediator protein; PARP, poly(ADP-ribose) polymerase; CamKIV, calmodulin-dependent protein kinase IV.

The plasma membrane calcium ATPase (PMCA) has long been known to be a calpain substrate in erythrocytes, where truncation results in an active form of the protein that is no longer regulated by calmodulin (James et al, 1989). However, at least one study has shown that calpain proteolysis of erythrocyte PMCA in situ actually causes a decrease in function (Salamino et al, 1994b). More recently, glutamate excitotoxicity has been shown to result in internalization of PMCA in hippocampal neurons, an effect that was blocked by calpain inhibition (Pottorf et al, 2006). This result is consistent with studies showing loss of PMCA activity (Oguro et al, 1995) and expression (Lehotský et al, 1999) in postischemic gerbil hippocampus. The result of calpain-mediated PMCA truncation therefore appears to parallel the effects seen for calpain cleavage of l-type channels and the Na⁺/Ca²⁺ exchanger; cleavage of all three proteins reduces the ability of the cell to maintain a low internal calcium concentration and therefore likely contributes to a feed-forward pathway of cytosolic calcium overload.

Synapse: The manner in which calpain activity may alter calcium regulation through cleavage of neurotransmitter receptors is less straightforward (Figure 2A). Classic NMDA-type glutamate receptors are heteromeric proteins composed of two NR1 subunits and two NR2 subunits, selected from subtypes NR2A, NR2B, NR2C, and NR2D. Of these, NR2A, -2B, and -2C are known to undergo C-terminal proteolysis by calpain in vitro (Guttmann et al, 2002), a truncation that does not appear to prevent any aspects of receptor function. However, only the NR2B subunit has been (Xu et al, 2007) shown to be cleaved by calpain in mature neurons (Dong et al, 2006). This cleavage occurs both in vivo after transient bilateral carotid artery occlusion and in vitro after excitotoxic insults in hippocampal neurons (Simpkins et al, 2003). The cleaved receptor remains on the cell surface, and, although it has not been shown to be active in neurons, it is predicted to be so based on the cleavage sites. Separate published reports have shown either a reduction in NMDA-mediated transmission (Wu et al, 2005) or essentially no change in electrophysiologic properties (Guttmann et al, 2001) after calpain proteolysis. What should be altered is the ability of the truncated C-terminus to interact with intracellular signaling molecules. The downstream effects of this modification on signaling and calcium regulation are unknown.

Activation of NMDA receptors is also associated with calpain-mediated cleavage of a member of the metabotropic family of glutamate receptors. The C-terminus of the mGluR1α G-protein-coupled receptor is cleaved by calpain after NMDA exposure (Xu et al, 2007). The truncated receptor retains its ability to signal for the release of calcium from intracellular stores, but is unable to activate the neuroprotective PI₃ kinase—Akt signaling pathway. The calpain-cleaved mGluR1α is therefore able to contribute to cytosolic calcium overload without any balancing neuroprotective effect.

Calpain also cleaves the GluR1, GluR2, and GluR3 subunits of AMPA-type ionotropic glutamate receptors (Bi et al, 2000). Again, calpain proteolysis produces a C-terminal truncation of the receptor, an effect that is seen in vitro and after NMDA excitotoxicity in hippocampal slice cultures (Bi et al, 1998). The effects of calpain cleavage are clearer for AMPA receptors than for NMDA receptors, with studies suggesting that calpain activation leads to both a loss of total GluR1–3 protein (Jourdi et al, 2005) and their dissociation from postsynaptic densities (Lu et al, 2000b). The overall conclusion from these studies appears to be that calpain cleavage reduces AMPA-mediated currents and increases turnover of AMPA receptors. This is particularly interesting in light of studies showing a decrease in expression of mRNA for the calcium-impermeable GluR2 subunit after ischemia, suggesting that replacement of AMPA receptors after ischemia would be biased toward calcium-permeable isoforms (Pellegrini-Giampietro et al, 1997). This concept is supported by the fact that overexpression of a Ca²⁺-impermeable form of the receptor rescues neurons from ischemic damage (Liu et al, 2004). Conversely, it has recently been shown that the calpain-mediated downregulation of AMPA receptors acts primarily on the GluR1 subunit, leaving GluR2 at the cell surface both in cortical neuron culture and in rat brain (Yuen et al, 2007). It is not clear what effect this sparing of GluR2 subunits might have on potential changes in Ca²⁺ permeability associated with AMPA receptor turnover.

In addition to direct effects on neurotransmitter receptors, postischemic calpain activation may alter synapse structure and function through cleavage of postsynaptic cytoskeletal elements (summarized in Table 1 and Figure 2A). Postsynaptic density protein 95 ordinarily binds numerous receptors and protects NR2A from being cleaved by calpain (Dong et al, 2004). However, postsynaptic density protein 95 itself is cleaved by calpain in calcium-treated rat brain sections and NMDA-exposed hippocampal slice culture (Lu et al, 2000a), which may modify the anchoring of NMDA receptors (Vinade et al, 2001). Calpain also cleaves proteins associated with the stabilization of AMPA receptors, such as synapse associated protein-97 (SAP-97) and glutamate receptor-activating protein (Jourdi et al, 2005), which may have an additive effect with direct calpain cleavage of AMPA subunits on receptor turnover.

The structural changes that occur in synapses after an ischemic insult extend beyond tethering of receptors. Rapid changes in dendritic spine density have been observed within hours of in vitro excitotoxicity, where calpain appears to be involved in synaptic remodeling (Faddis et al, 1997). These changes in neuronal structure likely involve cleavage of cytoskeletal calpain substrates such as αII-spectrin and microtubule-associated protein II (MAP2).

Spectrin is the major cytoskeletal protein component of the cell membrane (Czogalla and Sikorski, 2005). Intact spectrin is a 280 kDa protein, calpain cleavage of which produces unique 150 and 145 kDa fragments. Antibodies specific to these fragments provide a means to identify calpain activity, both through Western blotting of tissue lysates and through immunohistochemistry. Studies in postischemic gerbil hippocampus have shown an initial rapid rise in calpain-cleaved spectrin that occurs within 15 mins of reperfusion and which is then followed by a secondary accumulation of spectrin breakdown product between 4 and 24 h after injury (Saido et al, 1993; Roberts-Lewis et al, 1994). Transient forebrain ischemia in rats produces a similar pattern, with a peak in spectrin breakdown products observed at 1 h after injury that subsides only to be followed by a second peak beginning between 24 and 48 h in those cornu ammonis 1 hippocampal neurons that subsequently degenerate (Neumar et al, 2001). Spectrin proteolysis is, in fact, a hallmark of postischemic brain injury to the degree that the appearance of spectrin breakdown products in cerebrospinal fluid can be used as a biomarker of ischemia (Siman et al, 2004). Given the interactions between spectrin and many other membrane proteins such as ankyrin, actin, calmodulin, and microtubules (Bennett, 1990), it is likely that postischemic spectrin cleavage by calpain has substantial effects on membrane structure and function.

Microtubule-associated protein II serves to stabilize microtubules and regulate the microtubule network and transport system in neuronal dendrites (Dehmelt and Halpain, 2005). Microtubule-associated protein II is susceptible to calpain-mediated cleavage, as in vivo excitotoxicity leads to calpain-mediated loss of MAP2 immunoreactivity (Siman and Noszek, 1988). Similar degradation of MAP2 occurs after oxygen—glucose deprivation in hippocampal slice culture (Buddle et al, 2003) and in in vivo focal (Pettigrew et al, 1996) and global ischemia (Ziemka-Nałecz et al, 2003). This degradation of MAP2 is preceded by calpain activation, and is prevented by calpain inhibition (Inuzuka et al, 1990). Taken together, evidence for calpain-mediated cleavage of spectrin and MAP2 strongly suggests an important role of calpain in postischemic degradation of the neuronal cytoskeleton and alterations in synaptic structure.

The final role of calpain at the plasma membrane and in the synapse is to proteolytically modify a number of signaling molecules, triggering downstream effects throughout the cell. Many of these molecules are regulated by calmodulin, and a common effect of calpain cleavage is to remove this regulation. Excitotoxicity in hippocampal neurons (Wu et al, 2004) and focal brain ischemia in mice (Shioda et al, 2006) induces calpain cleavage of the calmodulin-dependent phosphatase calcineurin, producing a constitutively active form. Calpain also produces an active form of type IIα calmodulin-dependent protein kinase and degrades neuronal nitric oxide synthase after both in vitro and in vivo administration of a variety of neurotoxins (Hajimohammadreza et al, 1997). The presynaptic growth-associated protein 43, which is involved in the regulation of neurotransmitter release, is cleaved by calpain, decreasing calmodulin binding (Zakharov et al, 2005). Although not regulated by calmodulin, protein kinase C is cleaved by calpain, again producing a catalytically active fragment (Kishimoto et al, 1989). The precise downstream effects of cleavage of these signaling molecules on postischemic neurodegeneration are not known.

Endoplasmic Reticulum

Although the influx of extracellular calcium is certainly an important component of postischemic injury, it is important to also consider ways in which intracellular calcium stores can contribute to alterations in ion homeostasis. As noted above, calpain has been shown to be associated with the membrane of the ER (Hood et al, 2003, 2004). It is therefore only natural to assume that it is able to cleave ER proteins, and, in fact, several calpain substrates have been identified in the membrane of the ER (Figure 2B).

The ryanodine receptor (RYR) and inositol triphosphate receptor (IP₃R) are the two Ca²⁺ channels found in the membrane of the ER. The primary physiologic mechanism of RYR channel activation is thought to be increased cytosolic Ca²⁺ (Ca²⁺-induced Ca²⁺ release). Calpain-mediated proteolysis of cardiac RYR2 has been studied in detail. Digestion of cardiac junctional sarcoplasmic reticulum vesicles by m-calpain causes cleavage of the 400 kDa RYR2 isoform into 350 and 315 kDa intermediates followed by generation of a 150 kDa limit protein (Rardon et al, 1990). Electron microscopic analysis of sarcoplasmic reticulum vesicles after m-calpain proteolysis revealed a ‘shaved-off’ appearance of the cytoplasmic extension of the proteins. The functional effect of calpain-mediated proteolysis of cardiac RYR2 has been studied with the protein reconstituted in lipid bilayers. In these studies, calpain-mediated cleavage of RYR2 renders the channel unable to close in the presence of elevated cytosolic Ca²⁺ levels. Specifically, addition of calpain to the cis (cytosolic) chamber containing 1 μmol/L Ca²⁺ causes the RYR2 percentage open time to increase from 36% to >90% with no change in unitary channel conductance (Rardon et al, 1990). Ryanodine receptor 2 channels in a ‘locked open’ state are a potential mechanism for permanent ER Ca²⁺ depletion and loss of the ER's ability to buffer elevations in cytosolic Ca²⁺.

Published work evaluating the effect of brain ischemia on RYR is limited to studies showing decreased ³H-ryanodine binding in situ after permanent unilateral carotid artery occlusion in gerbils (Nozaki et al, 1997, 1999). Interpretation of these studies is limited because ryanodine binding is not significantly altered by calpain cleavage (Rardon et al, 1990). However, postischemic administration of the RYR channel antagonist dantrolene has been reported to be neuroprotective both in vitro (Wei and Perry, 1996) and in vivo (Nakayama et al, 2002), suggesting that increased Ca²⁺ efflux via RYR could contribute to sustained or secondary postischemic calpain activation and eventual neuronal death.

The IP₃R is the second Ca²⁺ channel in the ER. Inositol triphosphate receptor channel opening is stimulated by binding of IP₃, a second messenger generated when phospholipase C converts phospha tidylinositol bisphosphate (PIP₂) to IP₃ and diacylglycerol. There is also evidence that IP₃R Ca²⁺ release is enhanced by increased intracellular Ca²⁺, similar to RYR. The Ca²⁺-dependent channel opening probability displays a bell-shaped curve with maximum probability of opening at 0.2 μmol/L free Ca²⁺, which is significantly lower than the RYR (Bezprozvanny et al, 1991). Fifty percent of brain ER Ca²⁺ can be released by IP₃ (Ferris and Snyder, 1992), making the IP₃R a major potential player in postischemic calcium overload.

Calpain-mediated proteolysis of IP₃R1 from the rat cerebellum or striatum results in loss of intact protein (260 kDa) and generation of 200, 130, and 95 kDa (predominant) fragments detected by antibody raised against a peptide corresponding to the C-terminal domain (Igwe and Filla, 1997). This proteolysis of the IP₃R resulted in decreased IP₃ binding, suggesting that site-specific cleavage decreases the affinity of the remaining protein species for the IP₃ ligand. A possible calpain-mediated reduction in ligand binding is consistent with studies of IP₃R in postischemic brain. Reduced ³H-IP₃ binding has been observed in cornu ammonis 1 (CA1) hippocampus after unilateral carotid occlusion in gerbils, an effect that was not associated with any decrease in IP₃R immunoreactivity (Nagata et al, 1994). A more recent study of global ischemia in the rat produced similar results—a reduction in ³H-IP₃ binding in CA1, but no change in receptor expression (Dahl et al, 2000). The loss of IP₃ binding has also been shown functionally in postischemic brain tissue. Flash-frozen hippocampal slices taken from gerbils after unilateral carotid artery occlusion showed normal uptake of ⁴⁵Ca²⁺, but failed to release that calcium when stimulated with IP₃. Although this would seem to indicate that calpain cleavage of the receptor prevents evoked calcium release, it is possible that the proteolysis simply removes the ligand regulation of the channel. It is easy to imagine a model in which calpain-mediated loss of the ligand-binding domain of the IP₃R leads to baseline calcium release from the ER, contributing to postischemic calcium overload.

In addition to its effects on the IP₃R, calpain may also alter ER calcium release by affecting processing of IP₃ itself. A major metabolic pathway that serves to limit the half-life of IP₃ is its phosphorylation to form IP₄ by a family of IP₃ kinases (Irvine et al, 1986). The B isoform of the IP₃ kinase (IP₃-KB) is cleaved by calpain (Pattni et al, 2003). This cleavage separates the catalytic and membrane-anchoring domains of the molecule, releasing IP₃-KB from the membrane of the ER. This could then prevent metabolism of IP₃, allowing it to act longer on the IP₃R, thereby potentiating calcium efflux from the ER.

The sarcoplasmic/ER Ca²⁺ ATPase (SERCA) is responsible for energy-dependent ER Ca²⁺ sequestration. Sarcoplasmic/ER Ca²⁺ ATPase has been shown to be susceptible to calpain proteolysis in vitro, but only when the ATPase is in its oxidized form. Although calpain-mediated degradation of SERCA has been repeatedly shown after ischemia—reperfusion injury in the heart (French et al, 2006), little data exist for the effects of ischemia on brain SERCA. One study has measured SERCA ATPase activity in brain microsomes after decapitation ischemia (Parsons et al, 1999). In this model, SERCA-mediated ⁴⁵Ca²⁺ uptake was decreased after as little as a 10-min ischemia, but SERCA-specific ATP hydrolysis was unaffected. This apparent uncoupling of SERCA ATPase activity and Ca²⁺ uptake could be explained by modification of the SERCA molecule itself or increased ER membrane Ca²⁺ conductance via RYR and/or IP₃R channels. One would not expect calpain-mediated proteolysis of SERCA in a model of decapitation ischemia because unoxidized SERCA is not susceptible to calpain-mediated proteolysis. Sarcoplasmic/ER Ca²⁺ ATPase function has not yet been examined in the reperfused brain where conditions favor SERCA oxidation. Still, loss of SERCA function because of calpain-mediated proteolysis, like calpain alteration of RYR and IP₃R systems, is a potential mechanism of Ca²⁺ overload in postischemic neurons.

Mitochondria

Mitochondrial dysfunction has a well-established role in ischemic brain injury. Like cytosolic calcium overload, mitochondrial calcium overload is known to occur both immediately after brain ischemia and again at later time points (Zaidan and Sims, 1994). The initial postischemic calcium overload is accompanied by a burst of free radical production (Hall et al, 1993), an effect that can be inhibited with mitochondrial complex I blockers (Piantadosi and Zhang, 1996). Mitochondrial permeability transition has been implicated in models of both focal (Shiga et al, 1992) and global (Uchino et al, 1995) brain ischemia by the neuroprotective effects of cyclosporine A, a mitochondrial permeability transition inhibitor. Although such mitochondrial dysfunction is classically associated with apoptotic cell death and activation of the caspase family of proteases, a number of studies have identified a role of calpain in cell death associated with mitochondrial dysfunction. Calpain inhibitors have been shown to preserve mitochondrial function in cardiac tissue after ischemia—reperfusion injury (Neuhof et al, 2003; Trumbeckaite et al, 2003). This effect is likely independent of any inhibition of caspases, as the specific inhibitors used have been shown not to inhibit caspase-3 (Lubisch et al, 2003).

Apoptosis Regulatory Proteins: One way in which calpain may contribute to mitochondrial dysfunction is through cleavage of a number of proteins that regulate apoptosis (Figure 2C). This includes the Bcl2 family of proteins, several of which are calpain substrates. Calpain cleavage of Bax produces an 18 kDa fragment that leads to cytochrome c release in dopaminergic neurons exposed to mitochondrial toxins (Choi et al, 2001). Similar results have been observed for Bid, another proapoptotic Bcl2 family protein. Calpain-mediated cleavage and activation of Bid have been observed in rabbit heart after ischemia reperfusion injury (Chen et al, 2001) and in mouse brain after kainate excitotoxicity (Takano et al, 2005). Bid activation in this model was enhanced in mice deficient for the endogenous calpain inhibitor calpastatin. There is also evidence that calpain cleaves the loop region of the antiapoptotic protein Bcl-xL, converting it to a proapoptotic molecule (Nakagawa and Yuan, 2000). Calpain cleavage of these molecules leads to mitochondrial permeability transition, and release of death trigger molecules such as AIF and cytochrome c. Calpain is also thought to regulate apoptosis through degradation of the transcription factor p53 (Pariat et al, 1997), a finding that is supported by the fact that inhibition of calpains increases activated p53 and enhances p53-dependent apoptosis in tumor cell lines (Atencio et al, 2000).

Caspase-Dependent Programmed Cell Death: Calpain may also play an important role in apoptotic cell death through direct and indirect interactions with members of the caspase family of proteases (Figure 2C). Apoptosis protease-activating factor-1, which combines with cytochrome c to bind and activate caspase-9, is cleaved by calpain. The resulting loss of apoptosis protease-activating factor-1 was correlated with reduced activity of caspase-3-like proteases (Reimertz et al, 2001). Direct calpain truncation of caspase-9 also results in a loss of ability to activate caspase-3 (Chua et al, 2000). Conversely, calpain cleavage converts procaspase-7 to the active form, independent of the activity of other caspases (Ruiz-Vela et al, 1999). Conflicting results have been reported for caspase-3, calpain cleavage of which has been shown in separate studies to be inhibitory (McGinnis et al, 1999) or activating (Blomgren et al, 2001). Crosstalk between the calpain and caspase systems also occurs via calpastatin. Caspases have been shown to cleave calpastatin and reduce its inhibitory ability (Wang et al, 1998), setting up a complex interaction between the calpain and calpastatin systems. Although overexpression of calpastatin initially enhances caspase-3 activity, likely by preventing calpain-mediated degradation of caspase-9 and caspase-3, calpastatin itself is degraded by caspase-3, resulting in a secondary increase in calpain activity (Neumar et al, 2003).

Caspase-Independent Programmed Cell Death: In addition to promoting formation of the mitochondrial permeability transition pore, calpain may play a more direct role in the release of AIF. Calpain has recently been shown to cleave AIF, and to induce its release from mitochondria (Polster et al, 2005; Cao et al, 2007). This cleavage can occur both with exposure to exogenous calpain and when the only calpain present is that already within the mitochondria. Release of AIF from the mitochondria has also been observed in concert with activation of caspase-12, another calpain substrate (Nakagawa and Yuan, 2000), in models of ER stress-induced apoptosis in retinal ganglion cells (Sanges et al, 2006). Release of AIF in response to excitotoxic stimuli is suppressed by adenovirus-mediated expression of calpastatin (Volbracht et al, 2005; Cao et al, 2007), in transgenic mice overexpressing calpastatin (Takano et al, 2005), and after RNAi-mediated knockdown of μ-calpain in neurons (Cao et al, 2007). Therefore, although calpain appears to downregulate forms of caspase-mediated cell death, it may be simultaneously working to promote caspase-independent programmed cell death via an AIF-mediated mechanism.

Lysosomes

One proposed mechanism by which calpain activation may bring about necrotic destruction of the cell is through release of the hydrolytic enzymes typically sequestered within lysosomes, primarily the cathepsins (Figure 2B). This process, described by Tetsumori Yamashima as the ‘calpain—cathepsin cascade’, consists of calpain-mediated disruption of lysosomal membranes followed by diffusion of cathepsin B throughout the cytoplasm and nucleus. The evidence supporting this hypothesis in models of brain ischemia has recently been reviewed (Yamashima, 2004). Independent support for the interaction between calpains and cathepsins comes from work in C. elegans where loss of function in the proteases CLP-1 and TRA-3 (equivalent to calpains) as well as ASP-3 and ASP-4 (equivalent to cathepsins) is neuroprotective (Syntichaki et al, 2002). Combined loss of both calpain and cathepsin equivalent proteases did not have an additive neuroprotective effect, suggesting that the two protease families may induce neurodegeneration via a common sequential pathway.

Nucleus

Calpain directly cleaves nuclear proteins as well as causes the translocation of cytosolic proteins to the nucleus after proteolysis (Figure 2C). Within the nucleus of primary cerebellar granule neurons, calpain cleaves Ca²⁺/calmodulin-dependent protein kinase type IV (CaMKIV), leading to a reduction in phosphorylation of CaMKIV targets (Tremper-Wells and Vallano, 2005). The role of calpain-mediated CaMKIV downregulation in neurodegeneration has not been investigated. Calpain also cleaves the nuclear protein poly(ADP-ribose) polymerase-1 (PARP). Poly(ADP-ribose) polymerase-1 is normally a DNA repair enzyme that becomes overactivated after numerous toxic insults, leading instead to NAD and ATP depletion and eventual cell death (Pellicciari et al, 2004). Evidence for a role in ischemic brain injury includes findings that PARP becomes activated after excitotoxicity in cerebellar granule neurons (Cosi et al, 1994) and that PARP gene expression increases after global brain ischemia or kainate injection in gerbil and rat brains (Liu et al, 2000). Knockout of PARP reduces infarct size in a mouse model of focal cerebral ischemia (Eliasson et al, 1997). There is also evidence that PARP plays a key role in AIF-mediated programmed cell death (Yu et al, 2006). Although PARP is a known calpain substrate (McGinnis et al, 1999), it is again not clear what role, if any, calpain-mediated PARP cleavage plays in postischemic neurodegeneration.

Calpain also cleaves a number of cytosolic molecules that subsequently translocate to the nucleus. One such molecule is β-catenin, which undergoes calpain-mediated truncation in neurons exposed to glutamate (Abe and Takeichi, 2007). The truncated molecule then translocates to the nucleus where it induces gene transcription. The precise role of this gene regulation in postischemic neuronal death has not been examined. A related molecule in the Wnt-signaling pathway, glycogen synthase kinase 3, is also cleaved by calpain. Like many other calpain substrates, cleaved glycogen synthase kinase remains functional, and, in fact, displays an increase in kinase activity (Goñi-Oliver et al, 2007). Interestingly, glycogen synthase kinase 3 phosphorylation of β-catenin should prevent its translocation to the nucleus, further complicating the role calpain cleavage of these molecules may have in postischemic pathology.

More substantial evidence for a causal role in calpain-mediated cell death exists for cyclin-dependent kinase 5 (CDK5). Cyclin-dependent kinase 5 is unique in the CDK family in that it is primarily active in the central nervous system, where it is involved in regulating neuronal migration and synaptic function (Dhavan and Tsai, 2001). Activity of CDK5 kinase is regulated by the binding of one of its activator proteins, p35 or p39, which are cleaved by calpain to form to p25 or p29, respectively (Kusakawa et al, 2000; Patzke and Tsai, 2002). Accumulation of p25 has been shown after excitotoxicity, hypoxia, calcium overload, and ER stress in primary neuronal culture (Lee et al, 2000; Saito et al, 2007), as well as after middle cerebral artery occlusion in vivo (Nath et al, 2000). Similar results have been reported for p29 (Patzke and Tsai, 2002). Calpain-mediated truncation of p35 to p25 produces a more stable form of the protein, which is still able to activate CDK5. The CDK5–p25 complex translocates from the cytoplasm to the nucleus, where it appears to play an essential role in neuronal cell death (O'Hare et al, 2005; Saito et al, 2007).

The collapsin response mediator proteins (CRMPs) are another family of proteins with important roles in normal neuronal development that undergo calpain cleavage and nuclear translocation in disease states. Both CRMP-2 (Zhang et al, 2007) and CRMP-3 (Hou et al, 2006) are cleaved by calpain in vitro after excitotoxic injury and in vivo after traumatic or ischemic brain injury, respectively. Calpain cleavage of both CRMP-2 and CRMP-3 leads to their translocation from the cytosol to the nucleus. Furthermore, overexpression of the calpain-cleaved form of CRMP-3 was sufficient to induce neuronal death, whereas knockdown of endogenous CRMP-3 was neuroprotective (Hou et al, 2006), suggesting a causal role of nuclear CRMP in neurodegeneration.