Redox Regulation of Protein Function via Cysteine S-Nitrosylation and Its Relevance to Neurodegenerative Diseases

Debilitating neurodegenerative diseases, such as Alzheimer's disease (AD) and Parkinson's disease (PD), can be attributed to neuronal cell damage in specific brain regions. An important hallmark of these diseases is increased oxidative and nitrosative stress that occurs via overproduction of highly reactive free radicals known as reactive oxygen species (ROS) and reactive nitrogen species (RNS). These molecules are normally removed by cellular antioxidant systems. Under physiological conditions, ROS/RNS are present at low levels, mediating several neurotrophic and neuroprotective signaling pathways. In contrast, under pathological conditions, there is a pronounced increase in ROS/RNS generation, impairing normal neurological function. Nitric oxide (NO) is one such molecule that functions as a signaling agent under physiological conditions but causes nitrosative stress under pathological conditions due to its enhanced production. As first reported by our group and colleagues, the toxic effects of NO can be in part attributed to thiol S-nitrosylation, a posttranslational modification of cysteine residues on specific proteins. Here, we review several reports appearing over the past decade showing that S-nitrosylation of an increasing number of proteins compromises important cellular functions, including mitochondrial dynamics, endoplasmic reticulum (ER) protein folding, and signal transduction, thereby promoting synaptic damage, cell death, and neurodegeneration.


Introduction
A delicate balance in redox state exists in cells, in large part because of production of ROS/RNS and the antioxidant systems that detoxify them. This homeostatic redox balance maintains a relatively low concentration of ROS/RNS. Under physiological conditions, ROS/RNS can activate specific signaling pathways required for diverse cellular functions, including cell growth and immune responses [1]. However, increased ROS/RNS production or decreased antioxidant capacity can lead to perturbation of the redox balance, causing oxidative/nitrosative stress [2] (Figure 1). We and others have demonstrated that sustained oxidative/nitrosative stress elicits counterattack mechanisms, including activation of transcriptional pathways that activate (i) endogenous antioxidant phase 2 enzymes (the Keap1/Nrf2 cascade) and (ii) chaperones for refolding misfolded proteins (heat-shock proteins of the Hsp90/HSF1 cascade). These transcription pathways can be activated directly by ROS/RNS or by electrophilic compounds generated in response to oxidation [3][4][5][6]. For example, upon reaction of an electrophile with Keap1, Nrf2 dissociates from the Keap1/Nrf2 complex in the cytoplasm and translocates into the nucleus to initiate transcription of phase 2 antioxidant genes [7][8][9]. HSF1 activates transcription of heat shock proteins to combat protein misfolding due to stress [10,11]. If oxidant counteraction mechanisms, including activation of the Keap1/Nrf2 and Hsp90/HSF1 pathways, fail to combat ROS/RNS-related stress, cell injury, and death ensues ( Figure 1). Synaptic loss and neuronal cell death due to excessive oxidative/nitrosative stress have been widely implicated in neurodegenerative disorders, including Alzheimer's disease (AD) and Parkinson's disease (PD).
ROS and RNS are highly reactive molecules or free radicals. For instance, free radical nitric oxide (NO) possesses an unpaired electron in its outer pi molecular orbital. Due to this nature, ROS and RNS can react somewhat indiscriminately with all classes of biological macromolecules (e.g., protein, lipid, DNA) and cause cellular damage ( Figure 1). In this paper, we will specifically address the effect of nitrosative stress triggered by NO species that react to form protein Snitrosothiols. It should be noted, however, that NO signaling can result in other types of posttranslational modifications, such as protein tyrosine nitration and S-glutathionylation, as well as reaction with heme, for example, to activate soluble guanylate cyclase to form cGMP [12].

Nitric Oxide Production and Signaling
Cellular production of NO from l-arginine is catalyzed by a family of enzymes known as NO synthases (NOSs). The NOS family consists of endothelial NOS (eNOS), neuronal NOS (nNOS), and inducible NOS (iNOS) [13], and all three NOS subtypes are expressed in the mammalian brain. For instance, Ca 2+ -dependent nNOS catalyzes production of NO predominantly in neurons, whereas Ca 2+ -independent iNOS is primarily (but not exclusively) involved in NO production within microglia and astrocytes [14].
Many excitatory synapses contain N-methyl-d-aspartatetype glutamate receptor-(NMDAR-) operated channels. Activation of these channels results in Ca 2+ influx, triggering NO production by nNOS [15]. NO can undergo a number of reactions under normal physiological conditions. For example, NO reacts with soluble guanylate cyclase to produce cyclic GMP (cGMP) [16]. The second messenger cGMP then activates cyclic guanylate kinases (cGKs) [17,18]. Once activated, cGKs can phosphorylate various physiological substrates in neurons, thereby controlling various important processes, including synaptic transmission and synaptic plasticity [14,18] (Figure 2). An even more prominent physiological reaction of NO involves the posttranslational modification of S-nitrosylation or transfer of an NO group to a critical cysteine sulfhydryl to regulate protein function [19], in some sense analogous to phosphorylation of tyrosine, threonine, or serine residues. However, under pathological conditions, hyperactivation of NMDARs (often located at extrasynaptic or perisynaptic sites) causes massive Ca 2+ influx and overproduction of NO [19][20][21]. Nitrosative stress due to NO overproduction compromises cellular signaling via aberrant protein S-nitrosylation and tyrosine nitration, which can contribute to neuronal cell injury or death [12] ( Figure 2). In contrast, it should be noted that several of the proteins that are S-nitrosylated under physiological conditions, for example, the NMDAR itself and GOSPEL foster cell survival [22,23].

Protein S-Nitrosylation
As alluded to above, protein S-nitrosylation is a reversible posttranslational modification whereby an NO group is The nitrosative stress thus generated contributes to synaptic damage and neuronal loss, in part by fostering aberrant protein S-nitrosylation. Various cellular processes, including mitochondrial dynamics, protein folding, lipid metabolism, protein degradation, and signal transduction pathways can be perturbed by aberrant protein S-nitrosylation. Compromise in one or several of these cell processes can contribute to neurodegeneration. It should be noted, however, that a number of additional pathways, not related to NO, that are triggered by synaptic activity can also contribute to neuroprotection, while a number of pathways affected by extrasynaptic NMDAR activity appear to be involved in neuronal cell injury and death [99].
covalently attached to a cysteine thiol group (or more properly, a thiolate anion, -S − ) to form an S-nitroso derivative (R-SNO). Thus, we refer to S-nitrosylated proteins as SNOproteins. Importantly, not all the cysteines in a protein can be S-nitrosylated. Cysteines that are surrounded by a particular amino-acid motif are likely candidates for this modification. This "SNO motif " is a consensus grouping of amino acids that consists of nucleophilic residues (generally an acid and a base), which may result from protein tertiary or even quaternary structure [24]. A specific modification by Snitrosylation typically affects protein activity (either activating or inhibiting), thus mediating NO signaling pathways [23,25,26]. Proteins also can be denitrosylated (although transnitrosylation may also be involved) reportedly by redox-sensitive enzymes such as the thioredoxin (Trx) and S-nitroso-glutathione reductase systems, protein disulfide isomerase (PDI), and alcohol dehydrogenase (ADH) class III, which are now referred to as denitrosylases [27].

Implications of Protein S-Nitrosylation for Neurodegeneration
As discussed earlier, NO is produced in normal physiological conditions but does not induce nitrosative stress at low concentrations. However, under pathological neurodegenerative situations, NO production is highly increased, activating harmful signaling pathways, in large part due to aberrant protein S-nitrosylation. In this section, we will review several of the effects of protein S-nitrosylation in neurodegenerative diseases, including AD and PD.

Alzheimer's Disease.
AD is one of, if not the most, common forms of dementia, resulting in progressive decline of intellectual and social abilities that cause problems in dayto-day life. AD is a neurodegenerative disorder as there is a tremendous amount of cell injury and loss in various parts of the brain, including the hippocampus and neocortex [28]. One of the important observations in AD pathogenesis is that synaptic deficits precede cellular death and correlate well with decline in intellectual function [29,30]. Most AD cases (>95%) are sporadic, meaning that the majority of AD onset is not associated with obvious genetic mutations. Since AD mainly occurs in elderly people over 60 years old, age is one of the predisposing factors for the disease. One important hypothesis of aging and diseases of aging is the free radical theory, which states that an organism accumulates oxidative damage over time due to decreases in antioxidant systems and overproduction of free radicals [31,32]. Consistent with this theory, several studies have clearly shown that AD brains exhibit increased oxidative/nitrosative stress [33]. We have found that several proteins critical to neuronal survival are S-nitrosylated and, in some cases, further oxidized in AD, thereby disrupting the normal activity of the protein and contributing to disease pathogenesis, as described below.

S-Nitrosylation of Protein Disulfide Isomerase (PDI).
Once polypeptides are synthesized on ribosomes, those destined to be secreted are translocated to the endoplasmic reticulum (ER) for proper folding and disulfide bond formation. PDI is one of the enzymes that catalyze correct disulfide bond formation through a series of thiol-disulfide exchange reactions [34,35]. In the absence of proper disulfide bond formation, proteins misfold and aggregate in the ER, resulting in ER stress [36]. If ER stress persists, it can result in cell death [37]. Several lines of research have implicated a role for ER stress in AD pathophysiology [38]. Our laboratory discovered that PDI is S-nitrosylated in human AD brain compared to control brain. S-nitrosylation of PDI facilitates further oxidation of cysteine residues to sulfenic (-SOH), sulfinic (-SO 2 H), and sulfonic (-SO 3 H) acid PDI derivatives. These redox modifications compromise PDI chaperone/protein folding function, leading to protein misfolding and ER stress [26]. These results highlighted the role of nitrosative stress and SNO-PDI in neuronal cell injury and death in AD.

S-Nitrosylation of Dynamin Related Protein 1 (Drp1).
Neurons, and particularly their synaptic connections, require a tremendous amount of energy due to their high metabolic activity. Mitochondria, being the powerhouses of the cell, generate the vast majority of this energy. Recent reports suggest that to meet energy demand in an efficient manner, mitochondrial dynamics, consisting of fission and fusion events to generate new mitochondria, have to be carefully regulated [39]. Perturbation of mitochondrial dynamics can have deleterious effects on neuronal function and survival [40,41]. Studies from our laboratory have shown that aberrant S-nitrosylation of Drp1 (a protein required for mitochondrial fission) hyperactivates Drp1, and, in turn, causes a dramatic increase in mitochondrial fission. We demonstrated that the altered mitochondrial dynamics due to S-nitrosylated Drp1 (SNO-Drp1) contributes to synaptic loss in neurons and subsequent neuronal cell death. In addition, SNO-Drp1 levels are significantly increased in postmortem sporadic human AD patient brains compared to controls [42]. Hence, this study clearly implicated the pathophysiological role of SNO-Drp1 in AD pathophysiology.

Transnitrosylation of Cdk5 to Drp1
. Molecular signaling pathways are central to cellular physiology and function. Several signaling molecules have been implicated in AD pathophysiology [43]. One such molecule is cyclindependent kinase 5 (Cdk5), the activity of which has been shown to be altered in AD [44]. Cdk5 in neurons does not function as a cell cycle regulator, yet it exerts control over various aspects of neuronal function including cell survival, neuronal migration, dendritic spine density, and synaptic plasticity [45][46][47]. In a recently published article from our laboratory, Qu et al. showed that Cdk5, in addition to being a kinase, is also a nitrosylase, capable of Snitrosylating other targets involved in both AD and PD. Initially, we found that Cdk5 itself could be S-nitrosylated in an Aβ-and NMDAR-dependant manner in neurons due to generation of NO by these insults [25]. Furthermore, we showed that S-nitrosylation of Cdk5 results in activation and contributes to Aβ-induced dendritic spine loss, representing a decrease in synapses, the only pathological correlate to clinical dementia in AD. Moreover, SNO-Cdk5 levels are significantly increased in postmortem sporadic AD patient brains compared to age matched control brains. Importantly, SNO-Cdk5 then appears to contribute to synaptic failure by acting as an endogenous nitrosylase for Drp1, transferring the NO group from Cdk5 to Drp1 to form SNO-Drp1. This study revealed a role for protein S-nitrosylation of Cdk5 in aberrant cell signaling and links this nitrosylase activity to neuronal damage in AD [25].

S-Nitrosylation of ApoE. Apolipoprotein E (ApoE)
represents a major risk factor locus for late onset Alzheimer's disease [48]. The different isoforms of ApoE vary at their cysteine residues, which are potential sites for S-nitrosylation, as our group had observed a number of years ago. A recent study showed that all ApoE isoforms can bind nNOS and that ApoE2 and ApoE3 can be found in the S-nitrosylated state in human hippocampal lysates [49]. S-Nitrosylation of ApoE isoforms has been suggested to cause loss of binding to low density lipoprotein (LDL) receptors. Thus, S-nitrosylation of ApoE may affect lipid metabolism, which is postulated to affect the progression of AD.
The above-mentioned studies highlight some of the roles of S-nitrosylated proteins and how they can alter diverse cellular functions, including mitochondrial dynamics and synapse loss, ER protein folding, signal transduction pathways, and lipid metabolism, thereby affecting the progression of AD (Figure 2). In addition to these pathways, we suspect that there are many more pathways altered by protein Snitrosylation in AD pathophysiology.

Parkinson's Disease.
PD is second only to AD in the prevalence of neurodegenerative disorders. It affects approximately 1% of people over 65 years of age [50] and is characterized by motor sequencing impairment and often has a component of dementia. Although there are some symptomatic treatments for patients suffering from PD, currently there is no successful therapy to prevent progression or restore function. The histopathological aspects of PD include the loss of dopaminergic neurons, primarily in the substantia nigra pars compacta, often with the simultaneous presence of intracellular inclusions called Lewy bodies. Lewy bodies are mainly distributed in the substantia nigra, neocortex, basal forebrain nuclei, and hippocampus [51,52]. Despite the fact that some familial cases have been identified, more than 95% of PD cases are reported as sporadic, some of which appear to be correlated with exposures to agricultural pesticides, herbicides, fungicides, heavy metals, or neurotoxins [53,54], although this epidemiological information has remained International Journal of Cell Biology 5 contentious in some circles. We and others have shown that several of these environmental factors induce the generation of potentially toxic ROS/RNS species within neuronal cells [55]. Interestingly, dopaminergic neurons are especially vulnerable to oxidative/nitrosative stress, perhaps partly because of the oxidizing nature of dopamine. These observations have raised the hypothesis that in sporadic PD cases, oxidative or nitrosative stress contributes to PD pathogenesis via altering the function of PD-associated proteins. In several cases, the same gene product that is encoded in hereditary cases of PD may be affected by environmental factors to mimic the more rare genetic form or increase the susceptibility or severity of the hereditary phenotype, as highlighted below.

S-Nitrosylation of Parkin.
As an example, mutations in the parkin gene are known to cause many cases of the autosomal-recessive juvenile Parkinsonism and some rare cases of adult-onset PD [56][57][58][59]. The parkin gene encodes an ubiquitin E3 ligase that targets many proteins for proteasomal degradation and also has a neuroprotective role in PD-related apoptotic events [56,60]. Mutations in the parkin gene result in the disturbances in parkinmediated protein ubiquitination [61][62][63], which leads to the accumulation of potentially neurotoxic protein aggregates of parkin substrates with consequent dysfunction of the ubiquitin-proteasome system degradative pathway [61,64,65]. Interestingly, recent reports suggest that, independently from its ubiquitin-ligase role, parkin also functions as a transcriptional repressor of p53 to protect dopaminergic neurons from PD-related stress [66,67].
In addition to these rare mutations, several environmental toxins that trigger oxidative/nitrosative stress are believed to affect the enzymatic activity of parkin protein.
For instance, certain pesticides, herbicides, and fungicides that generate ROS and RNS and have been linked epidemiologically to PD can cause alterations in parkin solubility, inducing its aggregation and compromising its protective function. Parkin has multiple cysteine residues in its RING domain and elsewhere [66,68], which can react with NO to form SNO-parkin. This S-nitrosylation reaction compromises parkin's neuroprotective function. Our group reported that S-nitrosylation of parkin initially increases E3 ligase activity, but with additional time this activity is inhibited. This dysfunctional E3 ligase activity is associated with abnormal protein aggregation resembling Lewy bodies, thus contributing to the parkinsonian phenotype [68]. Moreover, S-nitrosylation of parkin has also been found by our group and others in a mouse MPTP model of PD and in brains of human patients with Lewy body disease (LBD) and PD [68,69]. These findings support our notion that posttranslational changes to PD-related proteins via S-nitrosylation or other oxidation reactions may well contribute to the etiology of sporadic PD.

S-Nitrosylation of Peroxiredoxin.
Our group and colleagues have also demonstrated that S-nitrosylation of another protein, peroxiredoxin 2 (Prx2), may be related to PD pathogenesis. Prxs are a highly abundant family of antioxidant enzymes that reduce intracellular peroxides by redox reactions [70][71][72]. Among the Prx enzymes, Prx2 is the most abundant in the mammalian brain and neurons. The active site cysteine residues in Prx2 reduce peroxides to H 2 O, thus forming a sulfenic acid (-SOH) derivative of Prx2. Subsequently, the oxidized Prx2 cysteine(s) can either form an intermolecular disulfide bond (-S-S-) with another Prx2 molecule, undergo reduction/regeneration back to free sulfhydryl (-SH) by thioredoxin (Trx), or be further oxidized (termed hyperoxidation) to produce a sulfinic (-SO 2 H) or sulfonic (-SO 3 H) acid derivative.
In several neurodegenerative diseases linked to oxidative/nitrosative stress, Prx2 levels are increased [73,74], which may represent an attempt of the cell to counteract oxidative/nitrosative insult during neurodegeneration. We and others recently reported that Prx2 activity can be regulated in vitro and in cell-based systems by NO through Snitrosylation of redox-active cysteine residues, which would prevent the reaction of this protein with peroxides, thus preventing the neuroprotective action of Prx2 [75,76]. In human samples of PD brains and cell-based models of PD, S-nitrosylation of Prx2 has been found to be increased compared to control samples [75]. Since SNO-Prx2 cannot react with peroxide because the active cysteines are already nitrosylated, the normal redox cycle to detoxify ROS is disrupted, inducing oxidative stress that can contribute to neuronal cell death.

S-Nitrosylation of XIAP.
The protein X-linked inhibitor of apoptosis (XIAP) has also been found to be Snitrosylated in several neurodegenerative disorders, including Alzheimer's, Parkinson's, and Huntington's diseases (HD), by our laboratory and others. Inhibitors of apoptosis (IAPs) are a family of proteins that regulate cell survival through binding to caspases to repress their catalytic activity [77,78]. XIAP is the most commonly expressed and the most potent endogenous caspase inhibitor among the IAPs. XIAP has three copies of the baculovirus IAP repeat (BIR) domain and one RING domain at the C terminal. Biochemical and structural studies demonstrated that BIR domains confer the anticaspase activity [79], whereas the RING domain can act as an E3 ubiquitin ligase in the proteasome system [80][81][82][83][84].
Recent studies show a significant increase of S-nitrosylated XIAP in both cell-based and animal models of PD as well as in human brain samples from PD, AD, and HD patients [85,86]. Our detailed experiments have identified that the RING domains of XIAP can react with NO by S-nitrosylation [85], although very high, nonphysiological concentrations of NO can also induce S-nitrosylation of the BIR domain [86]. S-Nitrosylation of XIAP at the RING domain inhibits its E3 ligase and antiapoptotic activity. Furthermore, we demonstrated recently that S-nitrosylated caspases can transfer their NO group to XIAP in a process called transnitrosylation. This reaction inhibits XIAP ubiquitin E3 ligase activity on caspases, thus effectively enhancing 6 International Journal of Cell Biology caspase activity and thus promoting proapoptotic signaling [85].

S-Nitrosylation of GAPDH.
Solomon Snyder's group has shown that the important metabolic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) can be Snitrosylated to form SNO-GAPDH. SNO-GAPDH manifests a loss of enzymatic activity [87]. More importantly, Snitrosylated GAPDH efficiently binds Siah1 protein and then translocates to the nucleus. In the nucleus, this protein complex activates ubiquitination and thus degradation of several nuclear proteins, including nuclear receptor corepressor (N-COR); this process contributes to cell death [87]. These studies suggest that S-nitrosylation of GAPDH, on one hand, compromises its metabolic enzymatic activity, but, on the other hand, in conjunction with Siah1, forms an important signaling complex to promote cell death and neurodegeneration.

S-Nitrosylation of PDI.
As discussed above, protein misfolding and ER stress can be precipitated by Snitrosylation of PDI, thereby potentially contributing to neuronal injury in AD. Our laboratory has also shown this scenario to be true in PD models. For example, when SH-SY5Y dopaminergic cells were treated with rotenone, a pesticide implicated in the pathogenesis of PD, we observed an increase in SNO-PDI levels concomitant with a decrease in PDI chaperone activity. Additionally, we found dramatically increased levels of SNO-PDI in human postmortem PD brains compared to controls [26]. Since ER stress due to protein misfolding is thought to contribute to the neurodegenerative process in PD [88], our finding of a substantial degree of SNO-PDI in PD brains has both pathogenic and therapeutic implications.

S-Nitrosylation of DJ-1.
Deletions or point mutations in the protein DJ-1 (PARK7) have been shown to be responsible for an early-onset, autosomal-recessive form of PD [89]. Interestingly, DJ-1-mediated signaling pathways have also been implicated in the much more common sporadic form of PD. It has been postulated that the increase in DJ-1 expression observed in cells undergoing nitrosative stress induced by the herbicide, paraquat, represents an attempt to protect the cells [90]. Consistent with this notion, DJ-1 knockdown makes neuronal-like cells more susceptible to peroxide-, MPP + -, and 6-hydroxydopamine-induced cell death [91,92]. Additionally, DJ-1 deficient flies [93][94][95][96] or mice [97] are more vulnerable to environmental neurotoxins associated with dopaminergic degeneration. Sequence analysis of protein DJ-1 reveals three potentially redox-active cysteine residues, two of which (Cys46 and Cys53) appear to be susceptible to S-nitrosylation in vitro and in cell-based systems [98]. Our group has also observed S-nitrosylation of a critical redox-active cysteine in the crystal structure of DJ-1. These findings and others suggest that posttranslational modifications of DJ-1, including protein S-nitrosylation, can disrupt the antioxidant action of DJ-1 in dopaminergic neurons, rendering them more susceptible to damage in sporadic PD. However, the elucidation of additional effects of SNO-DJ-1 in PD will require further investigation.
In summary, a number of studies suggest that nitrosative stress contributes to PD pathogenesis by altering neuroprotective proteins such as parkin, Prx2, PDI, GAPDH, and XIAP ( Figure 2). These findings indicate that aberrant S-nitrosylation reactions may play an important role in this neurodegenerative disorder, providing additional insight into nitrosative mechanisms of PD pathogenesis as well as potential novel targets for the treatment of PD.

Conclusions
Nitric oxide signaling can be both beneficial and harmful to the nervous system depending on (i) the concentration of NO and (ii) the cell signaling pathways affected by various levels of NO. Physiological levels of NO activate both cGMP-cGKI and S-nitrosylation pathways responsible for various physiological processes, including those affecting synaptic transmission and plasticity. In contrast, high levels of NO compromise cellular functions by a variety of posttranslational modifications including aberrant Snitrosylation reactions that would not normally occur in the presence of physiological levels of NO. In this paper, we have discussed data accumulated over the past several years that highlight the importance of protein S-nitrosylation in perturbing vital cell functions, including mitochondrial dynamics, protein folding, ubiquitination, synaptic transmission, and signal transduction pathways. Alteration of one or several of these events contributes to neuronal cell death and the development of neurodegenerative disorders ( Figure 2). Although we have discussed the role of Snitrosylation of several proteins here, including Drp1, PDI, GAPDH, ApoE, parkin, XIAP, Prx2, and DJ-1 in AD and PD, this list is by no means complete. Proteome-wide studies have already found hundreds, if not thousands, of proteins that are S-nitrosylated [99]. Future studies will unravel the role of S-nitrosylation of additional proteins in various cellular cascades and its implications for the pathogenesis and treatment of neurodegenerative disorders.