- Top of page
- Materials and methods
Calcineurin is a serine/threonine phosphatase involved in a wide range of cellular responses to calcium mobilizing signals. Previous evidence supports the notion of the existence of a redox regulation of this enzyme, which might be relevant for neurodegenerative processes, where an imbalance between generation and removal of reactive oxygen species could occur. In a recent work, we have observed that calcineurin activity is depressed in two models for familial amyotrophic lateral sclerosis (FALS) associated with mutations of the antioxidant enzyme Cu,Zn superoxide dismutase (SOD1), namely in neuroblastoma cells expressing either SOD1 mutant G93A or mutant H46R and in brain areas from G93A transgenic mice.
In this work we report that while wild-type SOD1 has a protective effect, calcineurin is oxidatively inactivated by mutant SOD1s in vitro; this inactivation is mediated by reactive oxygen species and can be reverted by addition of reducing agents. Furthermore, we show that calcineurin is sensitive to oxidation only when it is in an ‘open’, calcium-activated conformation, and that G93A-SOD1 must have its redox-active copper site available to substrates in order to exert its pro-oxidant properties on calcineurin. These findings demonstrate that both wild-type and mutant SOD1s can interfere directly with calcineurin activity and further support the possibility of a relevant role for calcineurin-regulated biochemical pathways in the pathogenesis of FALS.
Mutations in the gene coding for the enzyme Cu,Zn superoxide dismutase (SOD1) have been demonstrated in patients of familial amyotrophic lateral sclerosis (FALS; Rosen et al. 1993). The mechanism(s) by which FALS-SOD1 mutants exert their toxic properties in the pathogenesis of this disease are still very controversial. One popular hypothesis is that FALS-SOD1 mutations cause the appearance of a pro-oxidant, pro-apoptotic function in a typically antioxidant enzyme (Rabizadeh et al. 1995; Wiedau-Pazos et al. 1996; Yim et al. 1996); this function might be mediated by an altered copper chemistry (Corson et al. 1998; Estévez et al. 1999; Gabbianelli et al. 1999) consequent to loose binding of zinc to the active site. In turn, this may cause the appearance of an increased catalysis of tyrosine nitration by SOD1 (Estévez et al. 1999).
During the last few years an impressive set of data have been collected in order to understand cellular alterations induced by mutation of this key enzyme (Cleveland 1999; Liu et al. 1999 Julien 2001). However, the targets of FALS-SOD1s toxic activity have eluded identification so far, and discussion is still open on which is the molecular species produced by that enzyme activity.
We have recently obtained evidence that expression of mutant FALS-SOD1s impairs the activity of calcineurin (CaN), a key enzyme in the signal transduction cascade, both in transfected human neuroblastoma cell lines and in the motor cortex of brain from FALS-transgenic mice (Ferri et al. 2000).
Calcineurin, also known as protein phosphatase 2B, is a heterodimeric enzyme composed of a catalytic A-subunit and a myristoylated regulatory B-subunit (Rusnak and Mertz 2000 and refs. therein). Besides the catalytic domain, where one zinc and one iron atom are bound, the A-subunit contains the auto-inhibitory domain masking the metal ions at the active site. The activation of CaN by Ca2+/calmodulin probably results from the displacement of this domain, causing the exposure of the catalytic center. In this ‘open’ conformation, CaN is sensitive to inactivation by small (undefined) molecules (Wang et al. 1996). Calcineurin is enriched in neural tissue and comprises over 1% of the total protein in the nervous system. Its role in regulating neuronal excitability by controlling the activity of ion channels, the release of neurotransmitter and hormone, synaptic plasticity and gene transcription is well documented (Morioka et al. 1999; Rusnak and Mertz 2000). A role for CaN in cell cycle regulation and in Ca-dependent apoptosis, mediated by interaction of CaN with Bcl-2 and by dephosphorylation of BAD, has also been proposed (Wang et al. 1999). Furthermore, CaN is responsible for the hyperphosphorylation of tau in Alzheimer's disease (Gong et al. 1994; Garver et al. 1999).
Recent work has reported evidence on the redox modulation of this enzyme in vitro, suggesting that CaN is inactivated by oxidants such as hydrogen peroxide, superoxide and glutathione disulfide and protected – or even increased by some common antioxidants (e.g. ascorbate, GSH, N-acetyl-l-cysteine) but not by others (e.g. caffeic acid, butylated hydroxytoluene and dimethyl thiourea) (Carballo et al. 1999; Bogumil et al. 2000; Sommer et al. 2000). Those data, together with our evidence for the inactivation of CaN by FALS-SOD1s in vivo (Ferri et al. 2000), has prompted us to further investigate on the possibility that mutant SOD1s act directly on CaN through an oxidative mechanism. By challenging recombinant CaN and mutant SOD1s in vitro, we have obtained evidence that mutant SOD1s can exert their toxic properties directly on calcineurin activity, in line with the hypothesis that calcineurin-regulated biochemical pathways might play a relevant role in the pathogenesis of SOD1-linked FALS.
- Top of page
- Materials and methods
CaN phosphatase activity is modulated by oxidative mechanisms: previous work has indicated that CaN is inactivated by oxidants such as hydrogen peroxide, superoxide and glutathione disulfide and protected by some antioxidants (Carballo et al. 1999; Bogumil et al. 2000; Ferri et al. 2000; Sommer et al. 2000) as well as by wtSOD1 (Wang et al. 1996; Ferri et al. 2000). Those data, together with our previous evidence for the inactivation of CaN by FALS-SOD1s in human neuroblastoma cell lines and in brain cortex of FALS transgenic mice (Ferri et al. 2000), has prompted us to ask whether mutant SOD1s act directly on CaN through an oxidative mechanism. To this aim, we have used human recombinant CaN purified according to a described procedure (Mondragon et al. 1997). This enzyme exists as a mixture of reduced and oxidized forms, which is sensitive to the presence of molecular oxygen in the assay, but can be re-activated under either anaerobic (Fig. 2) or reducing (Fig. 4) conditions. Inactivation by molecular oxygen most probably is due to the presence of traces of redox-active metal ions in the assay and both superoxide and hydrogen peroxide are able to inactivate CaN, as suggested by the fact that both wtSOD1 and catalase are able to shift the ratio reduced (active) to oxidized (inactive) enzyme toward the active form (Fig. 2). Such free redox-active metal ions would probably not exist in vivo, as it has been estimated for copper (Rae et al. 1999). However, the fact that CaN inactivation is observed in vivo in systems where FALS-SOD1s are expressed suggests that copper ions loosely (or incorrectly) bound to mutant SOD1s might serve as a source of ROS. Indeed, increase of ROS production and of oxidised species has been observed both in ALS patients and in model systems (Ferrante et al. 1997; Bogdanov et al. 1998; Liu et al. 1999; Lee et al. 2001).
The reversibility of CaN inactivation implies the existence of one or more intramolecular targets sensitive to oxidation. These might be represented either by the bi-metal center or by free Cys residues at the active site, as unmasking of the active site by displacement of the autoinhibitory domain is needed to inactivate CaN under all conditions tested. In particular, incubation in a calcium-free medium prevents partial inactivation of CaN upon exposure to air (Fig. 3).
Previous work has suggested that oxidation of iron at the Fe/Zn active site could be responsible for the inactivation of CaN (Wang et al. 1996). However, the state of oxidation of iron at the active site was extensively examined in a study by Yu and co-workers (Yu et al. 1997) and evidence was collected indicating that the functional state of purified bovine CaN is Fe(III)/Zn(II) or Fe(III)/Fe(II). In another recent work using ESR spectroscopy of H2O2-treated CaN, the authors have been able to rule out that Fe2+ is present in the active enzyme and that the dinuclear metal center is the target for the oxidative inactivation of CaN (Bogumil et al. 2000). Furthermore, we have observed (Ferri et al. 2000) that a strong iron reductant like dithionite causes a 50% decrease in CaN activity in cell extracts, supporting the view that the oxidised state of the metal at the active site is crucial for this enzyme.
We have now obtained evidence that the activity of recombinant CaN can be increased by addition of reducing agents (Fig. 4). The fact that thiol-reducing agents such as dithiotreitol and reduced glutathione are able to reactivate CaN is strongly suggestive of a critical role for Cys residues in this enzyme, in line with the work of Bogumil et al. (2000). Human CaN has 12 Cys residues in the A subunit and 3 in the B subunit. Most of those Cys residues appear to be in their reduced form in purified CaN and it has been reported that oxidation of one of those Cys residue is critical for the activation of this enzyme by Ni2+ (King 1986). Cysteines are among those compounds that ‘autoxidize’ upon exposure to air, with a rate depending on the amount of contaminating transition metal ions in the reaction mixture, giving rise to thiyl (RS·) and sulfenyl (RSO2·) or thiyl peroxyl (RSO2.) radicals (Halliwell and Gutterridge 1990; Halliwell 1992). This might explain the fact that purified CaN is present as a mixture of ‘reduced and oxidized’, i.e. ‘active and inactive’ enzyme, a condition that might be relevant to CaN modulation in the cell. In turn, this modulation might be very relevant to processes of aging and neurodegeneration, where oxidative damage mediated by the imbalance between generation of ROS and antioxidant defence is involved (Olanow 1993; Wallace and Melov 1998). Furthermore, an age-related decline in activation of calcineurin was described in rat T cells by Pahlavani and Vargas (1999).
Wild type-SOD1 and FALS-SOD1s exert opposite effects on CaN: while the wild-type enzyme has a protective effect on exposure to air, the FALS mutants increase oxidative damage in the same conditions (Fig. 2). This fact cannot be ascribed simply to the lower superoxide dismutase activity of FALS-SOD1s respect to the wild-type enzyme (about 50% for G93A-SOD1 and 5–10% for H46R-SOD1), consequent to their reduced content of catalytic copper. In fact, it is well- known that SOD1 is a very efficient enzyme (K = 2 × 109 m−1 s−1 between pH 5 and pH 10) and all SOD1s were present in a 10-fold molar excess with respect to CaN in the assay. Furthermore, wtSOD1 proved unable to revert the oxidative effect of G93A-SOD1 (Fig. 2). Therefore, that FALS-SOD1s inactivate CaN is not due to loss of superoxide dismutase activity, but can only be explained in terms of the widely accepted concept of ‘gain of funtion’ of the mutant enzyme.
Both protection and inhibition of CaN are mediated by availability of SOD1s' active site to substrates, as we have observed that inhibition with sodium azide, a drug acting on the active site-bound copper, abolishes both effects (Fig. 3). Copper chelator TEPA has a protective effect on CaN activity in the presence of G93A-SOD1, a fact that might be interpreted assuming that cellular toxicity of FALS-SOD1 mutants arises by alterations in their metal binding properties (Corson et al. 1998; Gabbianelli et al. 1999; Liu et al. 2000; and refs therein). Furthermore, H46R-SOD1 has by itself a much lower copper content than G93A-SOD1, and has a lower effect in inactivating CaN (Fig. 2). This might be an explanation of why H46R-SOD1 is associated with a late-onset, mild form of FALS (Ogasawara et al. 1993). In line with these observations, it must be recalled that inhibition of activity by chelation of copper is effective in reducing FALS-SOD1 toxic activity in several experimental systems (Rabizadeh et al. 1995; Ghadge et al. 1997; Hottinger et al. 1997; Gabbianelli et al. 1999; Ciriolo et al. 2000, 2001). It is widely accepted that free copper ions do not exist inside the cell: in a recent elegant study, it has been estimated their concentration as less than one Cu atom/cell (Rae et al. 1999). However, when one of the major intracellular copper-binding protein partially looses its ability to buffer that redox-active metal, as in the case of FALS-SOD1s, aberrant production of ROS may occur, even in the presence of other antioxidants. In fact, a partially correct, more solvent-accessible cooper site in SOD1 might still have full superoxide dismutase activity, but be more prone to side reactions with H2O2. This would result in a spurious peroxidase activitiy, which is quenchable by copper chelators. In this respect, several observations in vitro and in vivo have suggested that FALS-typical mutant SOD1s have acquired a copper-mediated pro-oxidant function. Evidence for increased ·OH formation upon reaction with H2O2 from FALS-associated SOD1 mutants (A4V and G93A) was obtained independently by Valentine and co-investigators (Wiedau-Pazos et al. 1996) and by the group of Stadtman (Yim et al. 1996; Yim et al. 1997). Such peroxidase activity has a Michaelis constant (Km) value for H2O2 lower for mutants G93A and A4V than for the wild-type enzyme, while the catalytic constant (Kcat) resulted identical for all three enzymes. The relevance of such oxidative activity in vivo has been supported by a study of Liu et al. (1999) in transgenic FALS mice, and by us (Ferri et al. 2000).
As shown in Figs 5 and 6, reactive oxygen species inactivate CaN and the effect is prevented only in part by DTT and by catalase when the experiment is carried out on an already partially oxidized protein (basal conditions). When the same experiment was carried out in the presence of wtSOD1, both X/XO and H2O2 caused depression of CaN activity, as wtSOD1 may also have some peroxidase activity in the absence of H2O2-inactivating enzymes (Hodgson and Fridovich 1975). In fact, in this case the recovery by DTT and catalase was more pronounced than in basal conditions. In the presence of G93A-SOD1, X/XO caused nearly complete inactivation of CaN, while H2O2 had no effect on an already oxidized CaN. These data can be interpreted recalling that G93A-SOD1 possesses significant superoxide dismutase activity and that the endogenously produced H2O2 is more prone to peroxidative reaction in the presence of the mutant enzyme (Wiedau-Pazos et al. 1996; Yim et al. 1996; Yim et al. 1997) than with the wild-type enzyme. This reaction generates ‘copper-bound hydroxyl radical’ (SOD-Cu2+-OH·) that oxidizes several anionic ligands including formate, azide, and nitrite anions (Hodgson and Fridovich 1975). The oxidizing potential of this species is similar to that of the ‘free’ hydroxyl radical, which is considerably higher than those associated with conventional peroxidases. SOD-Cu2+-OH· is likely to be the molecular species actually damaging CaN.
On the whole, our data further support the notion that copper-dependent peroxidation is the real novel function acquired by mutant SOD1s typical of familial amyotrophic lateral sclerosis and indicate that calcineurin is a direct target of such a function.