Neurotoxicity Linked to Dysfunctional Metal Ion Homeostasis and Xenobiotic Metal Exposure: Redox Signaling and Oxidative Stress

Copper

Cu is present in biological systems as Cu⁺ (cuprous ion) and Cu²⁺ (cupric ion) (Supplementary Table S1). Cu is a redox-active metal and a cofactor of many enzymes involved in cellular respiration, radical detoxification, as well as biosynthesis of neurotransmitters, neuropeptides, and hormones. For example, Cu is required as cofactor of several important enzymes in the brain, such as peptidylglycine monooxygenase (PHM), dopamine β-monooxygenase (DBM), tyrosinase (TYR), and cytochrome C oxidase (COX). Cu can activate O₂ for reduction and although its high reactivity with O₂ is essential for life, if uncontrolled it can promote oxidative stress and cellular damage. Cu⁺ can react with H₂O₂ to produce highly reactive ^•OH. Cu also induces microglial activation and mitochondrial ROS formation (137).

The control of Cu homeostasis in the brain requires a close interrelationship between the BBB, neurons, and astrocytes (300) (Fig. 1). Astrocytes regulate the properties of the BBB, which is the entry point for Cu into the brain from the blood stream, where Cu is bound to albumin or ceruloplasmin (Cp) (58) (Fig. 1a). At the same time, neurons require Cu as a cofactor and neuromodulator, while astrocytes are key players in synaptic transmission and Cu homeostasis (58). The Cu trafficking machineries of the BBB endothelial cells, neurons, and astrocytes resemble those of other extensively studied mammalian cells (Fig. 1a–c). Extracellular Cu is primarily transported into cells as Cu⁺ via the Cu transporter 1 (CTR1) (58). Cu²⁺ reduction to Cu⁺, and Cu uptake from Cp via CTR1, has been proposed to involve a reduction step but no Cu²⁺ reductase has been identified (281). CTR1-independent mechanisms have also been proposed. The divalent metal transporter 1 (DMT1) seems to play a compensatory role for Cu uptake under certain conditions such as in the absence of CTR1 or under low Fe conditions (147, 231). Interestingly, DMT1 loss promotes brain Cu accumulation and oxidative stress (122). Other potential candidates recently proposed to mediate Cu uptake are the Zrt (Zn-regulated transporter)- or Irt (Fe-regulated transporter)-like protein 4 (ZIP4) (11, 29).

OPEN IN VIEWER

FIG. 1.

Cu homeostasis in the brain and its redox control. Regulation of Cu homeostasis is a complex process requiring intercommunication between endothelial cells of the BBB, astrocytes, and neurons. These three types of cells have the same protein machinery as other mammalian cells: CTR1 and DMT1 for Cu uptake; Atox1, CCS1, and GSH for Cu delivery to ATP7(A/B), SOD3, and MT, respectively; COX17 and COX19 for Cu delivery to mitochondria, where SCO1, SCO2, and COX11 load Cu into COX; while ATP7 (A/B) participates in Cu export through the secretory pathway (a–c). (a) Cp- and albumin- bound Cu enters into the brain by crossing endothelial cells of BBB, which are in tight communication with astrocytes. (b) Astrocytes serve as an interface to deliver Cu to (c) neurons. (d) Recently, a redox control of Cu efflux was elucidated in neurons during differentiation: the redox state of Atox1 controls Cu delivery to the secretory pathway. ATP7A, ATPase copper transporting alpha; ATP7B, ATPase copper transporting beta; Atox1, antioxidant protein 1; BBB, blood-brain barrier; CCS1, copper chaperone for superoxide dismutase 1; COX, cytochrome C oxidase; Cp, ceruloplasmin; CTR1, copper transporter 1; Cu, copper; DMT1, divalent metal transporter 1; Glu, glutamate; GSSG, glutathione disulfide; GSH, glutathione; MT, metallothionein; SCO, cytochrome c oxidase assembly protein; SOD, superoxide dismutase.

Intracellular Cu distribution depends on the relative concentration and metal affinity of chaperones or chelators (18) (Fig. 1b, c). The antioxidant protein 1 (Atox1), the Cu chaperone for superoxide dismutase 1 (CCS1), and GSH have been proposed to take Cu⁺ from CTR1 (197). Chaperones not only bind Cu but they also deliver it to specific targets. CCS1 transfers Cu to SOD1, where its reactivity with O₂ is required for SOD1 maturation via the formation of a disulfide bridge (17). Copper chaperones COX19 and COX17 deliver Cu to the COX assembly proteins (SCO1 and SCO2) and COX11. Finally, Atox1 transports Cu to the ATPase copper transporting alpha (ATP7A) and beta (ATP7B) in the secretory pathway, where cuproenzymes such as SOD3 are metalized (Fig. 1b, c). Upon an excess in cytosolic Cu levels, vesicles in the secretory pathway are loaded with Cu and trafficked to the plasma membrane, where Cu is released into the extracellular space (196, 197).

Strikingly, although GSH has a lower affinity for Cu, compared to Atox1 and CCS1, the rate of Cu entry into the cell via CTR1 is affected by GSH, but not by Atox1 or CCS1 depletion, likely due to the higher concentration of GSH (208). Thus, GSH is the most important cytosolic first acceptor of Cu from CTR1, providing a tight link between cellular Cu uptake and cellular redox homeostasis. GSH in turn is known to transfer Cu ions to metallothioneins (MTs), small Cys-rich proteins that play a major role as scavengers for metal ions (93) (Fig. 1b, c). Three distinct isoforms of MTs are expressed in the human brain. MT-I, MT-II, and MT-III are found in astrocytes, while MT-III is the main isoform expressed in neurons. MTs can be secreted, playing a crucial role in modulating Cu homeostasis and protecting the cell from oxidative damage (299).

Chaperones use Cys residues to coordinate Cu⁺ in their reduced state. Thus, Cys oxidation affects Cu dynamics. Atox1 coordinates Cu using a CysXXCys motif that can form a disulfide bond, which can be reduced directly by GSH or by Grx1 in a GSH-dependent manner (Fig. 1d) (41, 127). During neuronal differentiation, the GSH/GSSG ratio increases promoting a more reductive environment, which in turn reduces Cys residues at the Cu binding site of Atox1. These events enable Cu transport from Atox1 to ATP7A/B and enhance Cu availability to load the active sites of newly synthetized cuproenzymes (128) (Fig. 1d). After neuronal differentiation, both Cu and MT-III levels increase (249).

The recent development of fluorescent sensors has revealed new important roles of intracellular Cu in neuronal activity (78); for example, in the spine neck of hippocampal neurons, Cu is essential for the control of the dendritic actin cytoskeleton (269). Cu export by ATP7A has been reported to be triggered by the activation of glutamate (Glu)/N-methyl-d-aspartate receptors (NMDARs) (301) (Fig. 2a). At the synapse, Cu can modulate many neurotransmitter receptors (66). For example, Cu inhibits NMDAR activity by Cys nitros(yl)ation, a neuroprotective mechanism associated with neuronal plasticity that requires the participation of the cellular prion protein (PrP^C) (110) (Fig. 2b). Accordingly, selective depletion of ATP7A in neurons and glia increases the susceptibility to NMDA seizures (132).

OPEN IN VIEWER

FIG. 2.

Cu at synapse and redox signaling. (a) After activation of NMDAR, Cu is released at the synapse through the secretory pathway. Several Cu binding proteins are present in the synaptic cleft, including PrP^C, APP, Aβ peptides, and NKB. (b) The Cu-PrP^C interaction is particularly important for the modulation of NMDAR activity via S-nitros(yl)ation. In turn, NO production is dependent on the activation of NMDAR and its interaction with nNOS through the PSD-95. (c) While NKB can assist Cu uptake into astrocytes, Cu distribution among Cu-binding proteins at the synapse likely depends on their relative concentration and affinities for Cu. (d) On the contrary, α-synuclein is an important presynaptic Cu-binding protein found in neurons; it is an IDP that can interact with membranes when it acquires α-helix structure at N-terminal region. Interestingly, Cu⁺ binding to the N-Terminal of α-synuclein enhances its α-helical conformation. Aβ, amyloid beta; APP, amyloid precursor protein; Cu⁺, cuprous ion; IDP, intrinsically disordered protein; NKB, neurokinin B; NMDAR, glutamate/N-methyl-d-aspartate receptor; nNOS, neural nitric oxide synthetase; PrP^C, cellular prion protein; PSD-95, postsynaptic density protein 95.

Cu trafficking at the synapse is complex and involves several Cu binding proteins, such as the membrane-bound PrP^C, the amyloid precursor protein (APP), and the amyloid beta (Aβ) peptide from neurons, or the neurokinin B (NKB) peptide from astrocytes (Fig. 2). PrP^C has several Cu binding sites and it might be involved in Cu sensing and transport into neurons (Fig. 3a). During synaptic transmission, the extracellular Cu concentration may reach ∼100 μM. Cu binding to PrP^C induces its endocytosis, possibly contributing to Cu delivering into the cytosol (263). APP has been reported to regulate Cu efflux, as the APP knockout mice display higher levels of Cu in the brain and in neurons, while APP overexpression leads to decreased intracellular Cu concentrations (28, 330, 375). In contrast, Aβ peptides produced from the proteolytic cleavage of APP have been proposed to act as Cu scavengers (264), particularly those cleaved to yield 4–40 peptides that have higher affinity for Cu and make up to 50% of the Aβ in plaques (374). NKB has been suggested to compete for Cu from PrP^C and transport it into astrocytes by endocytosis (Fig. 2c) (308). PrP^C and Aβ contain intrinsically disordered regions with Cu binding sites that have the capacity to adopt different Cu coordination modes, some of which have been proposed to activate O₂ and produce ROS (Fig. 3a) (184, 350).

OPEN IN VIEWER

FIG. 3.

Structural features and redox properties of relevant metal-protein complexes in neurodegenerative diseases. (a) Cu can interact with PrP^C, Aβ, and α-synuclein. PrP^C has three coordination sites: histidine (His) His96, His111, and an octarepeat. Cu²⁺ coordination to His96 and His111 yields very similar complexes different to those generated upon Cu⁺ binding to His111. The octarepeat region coordinates up to four Cu²⁺ ions depending on the Cu²⁺/PrP^C stoichiometric ratio. At low Cu:protein ratios, four His residues coordinate Cu²⁺, while at high ratios each His can bind one Cu²⁺ ion. The different coordination chemistry of these Cu-PrP^C complexes is reflected in their redox properties. Similarly, Cu²⁺ coordination to Aβ (1 –40/42) yields two different coordination modes that display different redox properties. Recently, coordination to Aβ (4 –40/42) has also been described; this site displays higher affinity for Cu²⁺ than Aβ (1 –40/42), and it yields a redox silent complex. On the contrary, nonacetylated α-synuclein has two high-affinity binding sites for Cu²⁺: site one involving residues at the N-terminal, and site two at His50. Acetylation of α-synuclein, as described to occur in vivo, abolishes Cu²⁺ binding at site 1, although both forms of α-synuclein can bind Cu⁺. (b) Fe²⁺ can interact with Aβ, but the redox properties of Fe-Aβ complexes remain unclear. Interestingly, recent studies propose that heme can bind to Aβ, yielding a heme (Fe)-Aβ complex with peroxidase activity. (c) Finally, Zn²⁺ can bind to the octarepeat region of PrP^C and with Aβ. Cu²⁺, cupric ion; Fe, iron; Fe²⁺, ferrous; Zn, zinc.

A dysfunction in Cu homeostasis is reported to alter neuronal function and lead to disease progression, including neurodegeneration (Supplementary Table S1). Menkes disease and Wilson disease are caused by mutations or partial deletions in ATP7A and ATP7B, respectively. These Cu transporters have different patterns of expression in the CNS, explaining the distinct pathological features of each disease. ATP7B is found in the visual cortex, anterior cingulate cortex, caudate, putamen, substantia nigra (SN), and cerebellum. ATP7A is detected in astrocytes and neurons from the hippocampus and cerebellum, the BBB and choroid plexus, and during neural development (338). Wilson patients display parkinsonism, underscoring the importance of Cu homeostasis in the motor controlling systems. On the contrary, the ubiquitous expression of ATP7A has challenged mechanistic investigations in Menkes disease. However, a recent study shows that depleting ATP7A in neurons and glia does not lead to neurodegeneration, but to an increased susceptibility to NMDA seizures (132), underscoring its neuroprotective role as described above (Fig. 2a,b). Although neurodegeneration is clearly linked to alterations in Cu homeostasis in Menkes and Wilson diseases, the mechanisms are still unknown.

Neurodegenerative disorders such as amyotrophic lateral sclerosis (ALS), Huntington's disease (HD), Parkinson's disease (PD), and Alzheimer's disease (AD) have been associated with alterations in Cu homeostasis, but a link to specific genetic alterations in Cu transport or handling is missing (Supplementary Table S1). These neurodegenerative diseases are associated with the formation of amyloid aggregates composed of proteins that are either a Cu-dependent antioxidant enzyme, such as SOD1 in ALS, or Cu-binding proteins, such as Aβ and α-synuclein in AD and PD, respectively.

ALS is characterized by the degeneration of motor neurons, and mouse models show increased intracellular Cu levels and the formation of protein aggregates composed of SOD1 and Cu transport proteins such as Ctr1, CCS, Atox1, and Cox17 (344, 346). SOD1 aggregation has been associated with an alteration in protein stability, which is impacted by metallation and Cu-dependent dimerization. Although SOD1 plays an important role in clearing O₂ ^•−, studies have demonstrated that the mechanism is likely other than the alteration in the antioxidant capacity of SOD1 (312). Increased intracellular Cu levels have been reported in ALS models (346). Consistently, Cu chelators or MT-I overexpression extend the life span and slow disease progression of the SOD1 (G93A) ALS mouse model (345).

HD is an autosomal dominant genetic disorder caused by polyglutamine (polyQ) repeat expansions near the N-terminus of the huntingtin (Htt) protein. HD is characterized by movement dysfunction, psychiatric and cognitive alterations linked to the degeneration of striatal spiny neurons. Cu accumulation in the striatum of HD transgenic mice has been reported (102). Mutant Htt forms toxic aggregates, but the mechanisms of toxicity are still unclear. A putative Cu binding site in Htt involving Met8 and His82 has been identified, and this interaction promotes polyQ aggregate formation (102, 381), which is reduced by MT-III overexpression (123).

PD is characterized by the degeneration of dopaminergic neurons in the SN, which are rich in Cu. PD patients show decreased Cu content in the SN (70) without changes in Cu levels in serum, plasma, and cerebrospinal fluid (CSF) (203). Accordingly, decreased levels of CTR1 (70) have been reported in the SN, while MT-I and MT-II levels in active astrocytes are increased, reflecting a glial response to the loss of Cu homeostasis in PD (226). In contrast, occupational exposure to Cu has been linked to an increased risk to develop PD (115).

The accumulation of intracellular protein inclusions (Lewy bodies), where α-synuclein is the main protein component, is another hallmark of PD (362). α-Synuclein is a small intrinsically disordered protein (IDP) enriched in presynaptic terminals and nucleus that can interact with cytoskeleton components and lipid membranes (170). IDPs are proteins that can adopt different conformations, and thus respond to changes in their biochemical environment. This property allows them to engage in interactions with multiple protein targets (23). Cu²⁺ and Cu⁺ ions are capable of binding to α-synuclein at three different sites (20, 33, 228, 361) (Fig. 3a). Interestingly, the H50Q SNCA/α-synuclein mutation linked to hereditary PD abolishes one Cu binding site altering Cu-induced α-synuclein aggregation (361). Moreover, Cu binding to the high-affinity Cu-binding site at the N-terminal region of α-synuclein accelerates its amyloid aggregation in vitro (20, 33); although this effect is abolished in acetylated α-synuclein (234), which is found in Lewy bodies (10).

While several studies have suggested that Cu-induced aggregation of α-synuclein is directly linked to its neurotoxicity, recent studies suggest a lack of correlation between protein aggregation and cytotoxicity (361); in fact, it has been demonstrated that α-synuclein potentiates the toxicity of Cu in dopaminergic cells in the absence of enhanced accumulation of protein aggregates (8). Clearly, further investigations are needed to completely understand the structural impact of Cu-α-synuclein interactions and their role in PD. On the contrary, acetylation and Cu⁺ binding to α-synuclein are two synergistic events that turn the intrinsically disordered N-terminal region into an α-helix conformation (228) (Fig. 2d), which displays higher affinity for membranes (75). These observations suggest a potential link between Cu⁺-α-synuclein interactions and the proposed function of α-synuclein in vesicle trafficking; a link that might be perturbed in PD.

Different redox modifications of α-synuclein are found in Lewy bodies, including Met oxidation, Tyr nitration (67), and formation of di-Tyr-linked α-synuclein dimers. Cu⁺-α-synuclein complexes have been implicated in these modifications, as they are capable of activating O₂, leading to Met oxidation and di-Tyr bond formation (5, 227) (Fig. 3a). In contrast, a recent report suggests that Cu⁺ complexes with oligomeric or fibrillar α-synuclein reduce metal-catalyzed ROS formation (264). Cu has also been shown to potentiate oxidative damage induced by the dopamine (DA) analog 6-hydroxydopamine (63). Cu also interacts with the early-onset PD recessive genes (protein) PARK7 (DJ-1) and PARK2(Parkin). Cu binding to DJ-1 protects against metal toxicity, possibly acting as a chaperone for SOD1 (35, 114); while Parkin mutations have been reported to increase the cytotoxic effect of heavy metals, including Cu (1).

AD is a neurodegenerative disease associated with the degeneration of hippocampal and cortical neurons and eventual loss of memory and progressive dementia (326). Decreased levels of Cu are found in AD brains (42), while the total Cu content and labile nonprotein-bound Cu fraction are increased in the plasma of AD patients (325). Interestingly, polymorphisms in ATP7B have been linked to AD (326). AD is associated with the formation of extracellular amyloid plaques composed of Cu bound to Aβ peptides 1–40 and 1–42 fragments produced by cleavage of the APP, as well as N-truncated forms 4–40 and 4–42, and to a lesser extent 11–40/42 (210, 273, 364). The Cu binding features of Aβ 4–40/42 and 11–40/42 are different from those of 1–40/42, leading to the formation of Cu-Aβ complexes with distinct redox properties (Fig. 3a) (229). Different aggregation properties have also been described, as illustrated by the faster fiber assembly rate of Aβ 11–40/42 when compared with 1–40/42 (21). Substoichiometric Cu²⁺ concentrations trigger Aβ aggregation through a different pathway that involves the formation of oligomers more neurotoxic than those generated by the peptide alone (25, 204). On the contrary, the redox activity of Cu-Aβ complexes has also been proposed to lead to the generation of ROS (52, 214), but contradictory results exist as well (264). Interestingly, interaction of Cu with N-truncated 4–40 and 4–42 peptides yields redox-inactive Cu-Aβ complexes (229, 350). A dysfunction in Cu homeostasis in AD is also evidenced by decreased levels of MT-III (390). MTs are capable of exchanging Cu²⁺ with Aβ1–40/42, reducing Cu and stabilizing it. This Cu exchange by MTs has been proposed as a redox-silencing mechanism that prevents ROS formation by Cu-Aβ complexes (223).

The impact of dysfunctional Cu homeostasis in AD might go beyond the neurotoxicity of Cu-Aβ complexes. APP, whose mutations are associated with AD, is a type 1 transmembrane protein that displays three Cu binding sites in its extracellular domain (22, 140). One Cu-binding site is located in the growth factor-like domain, and has been implicated in Cu-induced dimerization of APP, a process that would be important in cell adhesion and signaling (22). Cu was found to induce APP phosphorylation at Thr668 promoting its localization to the axonal membrane, suggesting an important link between Cu and APP functions at the synapse that might be perturbed in AD. Aβ peptides can also interfere with Cu-PrP^C interactions implicated in the regulation of NMDAR activity (Fig. 2b) (389). In addition, Cu has been reported to promote the degradation of the low-density lipoprotein receptor-related protein 1 (LRP1) via Tyr nitration and proteasomal degradation, which was linked to a decrease of Aβ clearance and its resultant accumulation in brain vasculature (319). Clearly, Cu plays important roles in neuromodulation and signaling processes, which would be perturbed in AD.

Iron

Fe is found in biological systems primarily as ferrous (2+) and ferric (3+) ions (Supplementary Table S1). Fe is a redox-active metal involved in several redox reactions that catalyze the formation of ROS. Fe is tightly bound to Fe storage and transport proteins, while <5% is present as labile redox-active Fe bound to low-affinity molecules. Fe is required as cofactor of several important enzymes for respiration and synthesis of neurotransmitters, including tryptophan hydroxylase (serotonin) and Tyr hydroxylase (norepinephrine and DA), cholesterol, and fatty acids; the latter particularly important for nerve myelination (161, 342). Electron transfer in many Fe enzymes, including the mitochondrial respiratory complexes, is facilitated by heme and Fe/S clusters (291). Fe is also important as cofactor for peroxidases and catalases, which are important for cellular redox homeostasis (6).

Fe is heterogeneously distributed in the brain; it is highly concentrated in the SN, hippocampus, striatum, interpeduncular nuclei (125), and myelin (329) (Supplementary Table S1). Fe homeostasis is regulated by communication between the BBB and astrocytes. In the blood stream, Fe³⁺ is found coordinated to transferrin (Tf) or ferritin (Ft), and as heme. Fe uptake into the BBB can occur by two pathways: (a) by direct transport of Fe²⁺ into the cytosol via DMT1; or (b) by endocytosis of Tf-bound Fe³⁺ via the Tf receptor (TfR), where the low pH of the endosome causes the release of Fe³⁺ from Tf. In both cases, Fe³⁺ is reduced to Fe²⁺ by the duodenal cytochrome b (Dcytb) or by the six transmembrane epithelial antigen of the prostate 2 (Steap2) ferrireductases and then transported by DMT1 (Fig. 4a) (215, 218).

OPEN IN VIEWER

FIG. 4.

Fe transport and redox signaling. (a) Fe transport across the BBB occurs via DMT1 or by endocytosis of Tf- or Ft- bound Fe³⁺; followed by Fe release via the concerted action of Fpn with a ferroxidase, either Hp or sCp. Astrocytes modulate Fe transport through the production of sCp, sAPP, and hepcidin. sAPP stabilizes Fpn, while hepcidin induces its internalization. (b) Mitochondria are the organelles with highest demand of Fe. Although Fe transport into mitochondria is not fully understood, the recently demonstrated interaction between mitochondria and TfR-containing endosomes supports the “kiss and run” hypothesis. Inside the mitochondria, Fe crosses the inner membrane by Mfrn2. (c) In hippocampal neurons, NMDAR activation induces Ca²⁺ entry and NO^• production by nNOS, causing S-nitros(yl)ation of Dexras1, which in turn induces Fe uptake through DMT1 and TfR. (d) Increased intracellular Fe also induces H₂O₂ production, inducing Ca²⁺ release from ER via RyR, which is important for neuronal plasticity. Dexras1, Ras-related dexamethasone induced 1; ER, endoplasmic reticulum; Fe³⁺, ferric; Fpn, ferroportin; Ft, ferritin; H₂O₂, hydrogen peroxide; Hp, hephaestin; Mfrn2, mitoferrin-2; NO^•, nitric oxide; PAP7, peripheral benzodiazepine receptor-associated protein 7; RyR, ryanodine receptor; sAPP, soluble fragment of APP; sCp, soluble ceruloplasmin; Tf, transferrin; TfR: transferrin receptor.

Fe efflux from endothelial cells to the interstitial space occurs via the coordinated activity of the Fe²⁺ transporter ferroportin (Fpn) and the Cu-dependent ferroxidases hephaestin (Hp) and soluble ceruloplasmin (sCp), which oxidize Fe²⁺ to Fe³⁺. Astrocytes regulate the release of Fe from BBB by either secretion of sCp, which stimulates Fe release, or by production of hepcidin, a peptide that induces internalization and ubiquitination of Fpn and thus, decreased Fe efflux (216). Astrocytes also express a glycosylphosphatidylinositol (GPI)-anchored form of Cp, which interacts with Fpn and participates in Fe efflux (Fig. 4a) (144). In general, oligodendrocytes, astrocytes, microglia, and neurons have the same machinery for Fe efflux, involving the concerted action of Fpn with Cp or Hp (60).

The mechanisms of Fe uptake differ between brain cell types. Fe, Tf, and Ft are primarily found in oligodendrocytes (61). Most Tf in the brain is synthesized and secreted by oligodendrocytes as Fe-free Tf or apo-Tf and is required for Fe mobilization within the interstitial fluid brain (87). Although oligodendrocytes and astrocytes can accumulate high levels of Fe, they do not express TfR. Fe uptake into oligodendrocytes has been recently proposed to involve the internalization of H-Ft by a mucin-domain containing protein (Tim-2) (343). In astrocytes, ascorbate-dependent Fe²⁺ uptake is mediated by DMT1 (167) and transient-receptor potential channels (265).

In neurons, the mechanisms involved in Fe uptake are still unclear. DMT1 is found in human neurons and its expression levels are negatively regulated by Fe exposure via ubiquitination and degradation (134). However, DMT1 seems to colocalize in cytoplasmic vesicles with TfR and to primarily contribute to Tf-bound Fe uptake (232, 266). The ZIP8 and ZIP14 members of the ZIP family of metal transporters have also been demonstrated to mediate Fe²⁺ uptake (188) and to be expressed in the brain (113). A recent report demonstrates that ZIP8 at the plasma membrane is the primary transporter involved in non-Tf-bound Fe into neurons (145).

In brain cells, the ferrireductases Dcytb and stromal cell-derived receptor, SDR2, are expressed in astrocytes (192, 351), while SDR2 and Steap2 are found in neurons (145). In all cases, it is important to note that Fe uptake and efflux always require redox cycling between Fe²⁺ and Fe³⁺ oxidation states.

Cytosolic Fe is stored by Ft, which is a protein complex formed by 24 subunits of heavy-ferritin (H-Ft) and light-ferritin (L-Ft) chains. Ft genes are regulated by Nrf2 (272). While H-subunits have ferroxidase activity and participate in Fe uptake, L-subunits are involved in Fe mineralization and long-term storage. Ft can store ∼4500 Fe ions in its core. However, the mechanism by which Fe is released from Ft remains unclear (47). Although Fe is mainly stored in the cytosol, mitochondria are the organelles with the highest Fe demand, as they require Fe-S clusters and heme groups for electron transfer during respiration. Similarly, mitochondria are also considered the main sources of ROS under physiological conditions. Therefore, mitochondrial Fe homeostasis must be tightly regulated to prevent uncontrolled ROS production.

The mechanism(s) involved in mitochondrial Fe uptake are still unclear. Fe is delivered by a direct “kiss and run” interaction between the endosomes containing Tf-bound Fe and mitochondria (69). Subsequently, the metal crosses the inner membrane via the Fe importer mitoferrin-2 (Mfrn2) (Fig. 4b). In the mitochondrial matrix, Fe is either used for the biogenesis of prosthetic groups, Fe-S clusters, or heme, or it is stored by mitochondrial ferritin (FtMt), which is homologous to H-Ft, and it displays the same Fe uptake efficiency but lower ferroxidase activity. Fe distribution between mitochondria and cytosol depends on FtMt expression and the export of Fe prosthetic groups (106, 176).

Heme is a component of globins, a superfamily of heme-containing proteins involved in binding and/or transporting O₂, and a cofactor for cytochromes, catalase, NOX, NOS, and myeloperoxidase, which is also found in microglia (59, 151, 198). Hemoglobin (Hb) is involved in O₂, NO, and carbon dioxide (CO₂) transport in cells of erythroid lineage, but Hbα and β transcripts have been found in dopaminergic neurons, oligodendrocytes, and cortical or hippocampal astrocytes as well (32). Brain Hb levels are altered during neurodegeneration (94), but their functional consequences are unclear. Neuroglobin is another globin expressed in the CNS and PNS and found in both neurons and astrocytes. Neuroglobin has a higher affinity for O₂ than Hb and exerts a protective effect against oxidative and ischemic insults (7, 43, 81, 179, 331, 368). Oxidative stress has the potential to release heme from Hb, and labile heme can induce oxidative damage via Fenton reactions or NOX activation (165, 239). Heme oxygenases (HOs) catalyze heme degradation. HO-1 transcription is induced by oxidative stress, inflammation, hypoxia, and metal exposure, while HO-2 is constitutively expressed. HO activity has both antioxidant and pro-oxidant effects that relate to the ability of the cell to detoxify labile Fe released from heme (286, 369).

Cellular Fe trafficking is controlled by the iron regulatory proteins (IRP) IRP1 and IRP2 which regulate translation of proteins involved in Fe storage (H-Ft and L-Ft), Fe uptake (TfR), and Fe efflux (Fpn). Under conditions of Fe depletion, IRPs can bind to messenger RNAs (mRNAs), decreasing the levels of Ft and Fpn and promoting the translation of TfR via the stabilization of its mRNA. Conversely, Fe overload prevents mRNA binding to IRPs, promoting Ft and Fpn translation while reducing TfR levels. The ability of IRP1 and IRP2 to bind mRNAs exerts an important redox control via two distinct mechanisms. IRP1 has two conformations: (a) a closed one, triggered via an Fe-S cluster that prevents mRNA binding and has aconitase activity; and (b) an open conformation that is favored when NO^•, O₂, and H₂O₂ cause dissociation of the Fe-S cluster. In contrast, the ability of IRP2 to bind mRNAs is controlled by proteasomal degradation, involving the ubiquitin ligase F-box/LRR-repeat protein (FBXL5), which is also an Fe and O₂ sensing protein. FBXL5 has a binuclear nonheme Fe site, and it is stabilized on Fe and O₂ binding, promoting IRP2 degradation. Together, these two mechanisms illustrate the redox control of cellular Fe trafficking (164).

In neurons, Fe uptake is induced by a redox signaling cascade that starts with the activation of the NMDAR (Fig. 4c). Increased intracellular Ca²⁺ induces NO^• production by the neuronal (n) NOS, leading to S-nitros(yl)ation of the small GTPase Dexras1 (Ras-related dexamethasone induced 1), which in turn induces Fe uptake through DMT1 and TfR (51). In hippocampal neurons, NMDAR activation increases intracellular Fe and H₂O₂ production, which activates the redox-sensitive ryanodine receptor (RyR) and promotes Ca²⁺ release from the ER (237) (Fig. 4d). Thus, Fe uptake is clearly part of a redox signaling mechanism that might be important for neuronal plasticity.

Cell death induced by an increase in the labile Fe pool within cells has been defined as a specific entity named ferroptosis. Ferroptotic cell death is a necrotic-like cell death characterized by Fe-dependent lipid peroxidation due to either the formation of ^•OH and H₂O₂ via Fenton-like reactions or the activation of lipoxygenases. As such, ferroptosis is counteracted by Fe-chelators and the GSH/GPX4 system. GPX4 knockout in neurons induces motor neuron degeneration, paralysis (53), and cognitive impairment (121). During ferroptosis, Tf-dependent Fe uptake and release of Fe from lysosomal compartments have been shown to act as important sources for Fe. Interestingly, lysosomal permeabilization is a common phenomenon observed in a number of neurodegenerative disorders, including PD (40, 73). However, during pathological conditions such as neurodegeneration and hemorrhagic stroke, cell death is likely to involve a combination of different pathways and a complex balance between them, including apoptosis, necrosis, autophagic cell death, and ferroptosis as well (77, 143, 357, 398).

During aging, accumulation of Fe in the frontal lobes and striatum is associated with motor dysfunction, loss of myelin sheaths, and memory decline (2, 327). Inflammation, a common hallmark of many brain disorders, increases the expression of DMT1 and hepcidin (354).

Abnormal accumulation of Fe is a common feature of many neurodegenerative diseases, including a group of twelve diseases known as neurodegeneration with brain iron accumulation (NBIA) (225). The most common clinical features of NBIA are movement disorders, such as ataxia, parkinsonism, and dystonia. Although in NBIA Fe is usually accumulated in the globus pallidus, and in some cases in the SN and cerebellum, only two types of NBIA involve the dysfunction of a protein that participates in Fe trafficking: aceruloplasminemia (lack of Cp) and neuroferritinopathy (loss of function mutations in L-Ft) (65, 261, 290, 360). Another neurodegenerative disease associated with disturbed Fe homeostasis is Friederich ataxia, which is linked to mutations in frataxin, an Fe-chaperone involved in Fe-S cluster biogenesis (55).

Altered Fe homeostasis and mitochondrial dysfunction are also hallmarks of PD. While no significant differences in Fe levels have been found in the blood, serum, and CSF (203), Fe is increased in the SN of PD patients (365). Neuromelanin is an Fe-rich pigment found in the dopaminergic neurons targeted in PD (A9) and it has been suggested that its presence makes this neuronal population vulnerable to oxidative damage (91). Decreased levels of serum Cp or its oxidation has also been proposed to exacerbate Fe accumulation in PD (149, 251). DMT1 is found increased in the SN of PD brains (294). Interestingly, Parkin regulates DMT1 expression levels via ubiquitination and proteasomal degradation (289).

Fe accumulation in SN might be associated with a dysfunctional delivery of Fe to mitochondria through the “kiss and run” interaction mentioned above (Fig. 4b) (69). Indeed, a Tf/TfR2-dependent mechanism for Fe transport into the mitochondria of dopaminergic neurons has been described (211), while a role for Tf and TfR2 genes in Fe accumulation and mitochondrial dysfunction in PD has been implicated as well (285).

Mitochondrial dysfunction in PD might lead to decreased synthesis of Fe-S clusters, which in turn would activate IRP1 binding to mRNAs, resulting in augmented Fe accumulation (142). Accordingly, knockdown of mitochondrial Grx2 impairs Fe-S cluster biogenesis in dopaminergic cells, decreases the activity of Complex I and aconitase, and increases the activation of IRP1 (171). Loss of PINK1/PARK6, another autosomal early-onset PD-related gene, has also been reported to inactivate Fe-S clusters via O₂ ^•− formation. As such, overexpression of FtMt exerts a protective effect on mitochondrial dysfunction and oxidative stress induced by loss of PINK1 (88).

HO-1 protects against neuronal cell death induced by PD-related mitochondrial toxins and α-synuclein, suggesting a role of heme in dopaminergic cell loss. However, HO-1 becomes toxic at high levels, but this effect is counteracted by FtMt (391). Interestingly, the pathogenic PINK1 mutation G309D impairs the induction of HO-1 on oxidative stress (56).

Fe²⁺ can interact with the negatively charged C-terminal region of α-synuclein, accelerating its amyloid aggregation (34). Conversely, α-synuclein overexpression also enables Fe accumulation (254). In PD, oxidative stress has been linked to intracellular Fe levels due to the redox activity of Fe-DA and Fe-neuromelanin complexes (400). Furthermore, overexpression of FtMt protects against neuronal cell death induced by 6-hydroxydopamine (313). Interestingly, a recent report demonstrates that depletion of Fpn has no consequence on the survival of dopaminergic neurons. In contrast, loss of TfR causes Fe deficiency and a PD-like neurodegeneration in mice (212). Previous findings have also suggested a link between Fe deficiency and predisposition to PD, while Fe overload seems to be protective (157, 190, 271). Thus, the exact role of Fe homeostasis in PD is still far from being understood.

A role of Fe in AD has also been proposed based on observations showing that Fe levels are decreased in the serum of AD patients (324), and increased in AD-related brain areas such as the hippocampus, neocortex, and basal ganglia (168). Mitochondrial dysfunction is also observed in AD, which might be related to a disruption in Fe homeostasis. Accordingly, overexpression of FtMt protects against the toxicity of Aβ (380). Fe is accumulated in amyloid plaques in the AD brain (38, 194), consistent with the observation that Aβ is able to bind Fe²⁺ in vitro (Fig. 3b) (39). Fe bound to Aβ plaques catalyzes H₂O₂ formation (321).

A novel mechanism for controlling Fe efflux from the BBB was recently discovered, involving the soluble fragment of APP (sAPP), which interacts with Fpn, stabilizing its membrane location (219) and counteracting the effects of hepcidin (216). APP expression is also negatively regulated by IRP1 (57), and production of sAPP and Aβ is increased in AD. Thus, the proposed role of sAPP in Fe movement from BBB into the brain is consistent with the accumulation of Fe in AD brains and the observation of higher levels of Aβ plaques surrounding the brain blood vessels. Indeed, BBB damage is a common feature in AD (217).

Heme metabolism also contributes to AD. APP binds and inhibits HO-1 activity and this effect is enhanced by pathogenic APP mutations (334). In addition, Aβ can form a complex with heme that displays peroxidase activity (16, 112). Conversely, neuroglobin has been proposed to protect against the toxicity of 1–42 Aβ peptides (180). Overall, disruption of Fe homeostasis in AD is likely linked to oxidative stress, and it may also impair NMDAR- and Fe- dependent redox signaling mechanisms (Fig. 4d) impacting neuronal plasticity and memory.

Manganese

Mn is an essential metal required for the activity of a plethora of enzymes, including hydrolases, isomerases, ligases, lyases, oxidoreductases, and transferases, involved in diverse metabolic functions such as amino acid (arginase and glutamine synthetase [GS]), lipid, protein, and carbohydrate metabolism (phosphoenolpyruvate decarboxylase), as well as protein glycosylation, energy production, and redox homeostasis (SOD2). The main source for Mn intake is food, but occupational/environmental exposures also occur associated with mining, smelting, welding, alloy, battery, pesticide, and electrical industries. Ingestion and inhalation are the primary routes of Mn exposure. Importantly, inhalation can transfer Mn directly to the brain. The brain is a major target for chronic Mn intoxication (manganism) where Mn is accumulated in nonheme Fe-rich regions. Manganism is defined as a parkinsonism that results in dystonia, hypokinesia, and rigidity as a consequence of impaired neurotransmitter function (133, 352).

Mn can be transported via Tfr and DMT1 as it competes with Fe for its binding sites (333). Other proposed Mn transporters include the Zn carriers ZIP-8 and ZIP-14, voltage-regulated, store-operated, and ionotropic Glu receptor Ca²⁺ channels, and the Mn-citrate complex shuttle (Fig. 5a) (352). Mn is efficiently detoxified by Fpn at the plasma membrane. In addition, Mn is also detoxified by sequestration in the Golgi via the solute carrier family 30 member 10 or human Zn transporter 1 (SLC30A10/hZnT1) transporter whose mutations are directly associated with manganism (178). Alternatively, the Ca²⁺/Mn²⁺ ATPases SPCA1 and SPCA2 also detoxify Mn via the secretory pathway (356). Finally, the autosomal recessive early-onset PD-related gene ATP13A2/PARK9 mediates sequestration of Mn in lysosomes (Fig. 5b) (336). Recently, direct comparison of different detoxification proteins demonstrated that hZnT1 and SPCA1, but not ATP13A2, are involved in Mn detoxification and resistance (246).

OPEN IN VIEWER

FIG. 5.

Mn homeostasis and toxicity. (a) Mn can be transported via Tf and DMT1, as well as via Zn carriers, ion channels, and shuttle systems. (b) Mn is detoxified by Fpn at the plasma membrane or by its sequestration in the secretory system via the ZnT1 transporter or the Ca²⁺/Mn²⁺ ATPases SPCA1/2. ATP13A2 mediates sequestration of Mn in lysosomes. (c) Mn enhances ROS generation via Mn-catalyzed autoxidation of DA that involves the redox cycling of Mn²⁺ and Mn³⁺ and the generation of ROS and DA-o-quinones. (d) Mn accumulates in the mitochondria via the MCU and increases the accumulation of labile Fe. (e) Mn-induced neurotoxicity is linked to oxidative damage and mitochondrial dysfunction, and also to a decrease in Glu uptake by the EAATs in astrocytes (arrow in EAAT indicates reversal of Glu transport) leading to excitotoxicity in neurons, as well as the induction of inflammation. DA, dopamine; EAAT, excitatory amino acid transporter; Glu, glutamate; MCU, mitochondrial Ca²⁺ uniporter; Mn, manganese; ROS, reactive oxygen species; SPCA, secretory pathway Ca²⁺ ATPase; ZnT, zinc transporter.

Mn²⁺ is the predominant species found in cells that can be oxidized to the more reactive and toxic species Mn³⁺. Neither Mn²⁺ nor Mn³⁺ can generate free radicals via Fenton-type reactions. However, it has been proposed that Mn enhances ROS generation via the Mn-catalyzed autoxidation of DA that involves the redox cycling of Mn²⁺ and Mn³⁺ and the generation of ROS and DA-o-quinone (Fig. 5c) (79, 89). Mn accumulates in the mitochondria via the mitochondrial Ca²⁺ uniporter (MCU) (111) and increases the accumulation of labile Fe. Both mitochondrial dysfunction and Fe lead to ROS formation and oxidative damage (Fig. 5d) (54, 206). Mn specifically generates H₂O₂ but not O₂ ^•− in the mitochondria via complex II (92, 186, 322). Furthermore, Mn impairs oxidative phosphorylation and ATP production.

Astrocytes seem to have a high capacity to accumulate Mn (14). Mn-induced neurotoxicity has been linked to a decrease in Glu uptake by astrocytes (156) leading to excitotoxicity in neurons, as well as the induction of inflammation and increased activity of NOS (Fig. 5e) (95, 185).

Zinc

Zn is a redox-inactive transition metal ion with an oxidation state of +2. The majority of intracellular Zn is bound to proteins and is distributed in the cytoplasm (∼50%) and nucleus (∼40%) (304) (Supplementary Table S1). The function of Zn as enzyme cofactor is limited to structural roles (e.g., SOD1) (312), or as a Lewis acid that activates substrates for nucleophilic attack (e.g., carbonic anhydrase, where Zn²⁺ catalyzes the hydration of CO₂ to form bicarbonate [HCO₃ ⁻]) (138). Enzymes with Zn-dependent catalytic activity control many cellular processes, including DNA synthesis and brain development. Zn also plays an important role in cell signaling associated with development and learning. In the brain, Zn is highly concentrated in the hippocampus and cortex (304).

The key players in Zn homeostasis are the ZIP transporters that mediate Zn uptake into the cytosol, the zinc transporters (ZnT) that participate in Zn efflux, and MTs involved in Zn chelation. ZIP1 is expressed in astrocytes and microglia (303). In neurons Zn uptake is mediated by, voltage-gated Ca²⁺ channels, ZIP1 and 3 transporters, as well as Ca²⁺ and the Zn²⁺ permeable α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptor (AMPAR) (146, 278, 335). Interestingly, Zn has been reported to be transported into postsynaptic neurons through a complex formed by PrP^C, which is evolutionary linked to ZIP proteins, and the AMPAR (Fig. 6b) (373). ZnT1 is expressed in astrocytes, microglia, and oligodendrocytes, and its expression levels are directly modulated by Zn (247). Hypoxia decreases the levels of ZnT1 in astrocytes inducing the accumulation of cytosolic Zn (257). At the postsynaptic density, the ZnT1 transporter interacts with NMDAR and this complex is regulated during synaptic plasticity (Fig. 6c) (222, 373).

OPEN IN VIEWER

FIG. 6.

Zn trafficking at the synapse. (a) Zn ions are loaded into neurotransmitter vesicles by ZnT3, and are coreleased with Glu during synaptic transmission. In the synaptic cleft, Zn can bind to MT, Aβ, PrP^C, and NMDAR. (b) Recently, two links between synaptic activity and Zn transport have been discovered: Zn uptake by the postsynaptic terminal occurs via a complex formed by PrP^C and AMPAR, (c) while ZnT1 interacts with NMDAR and modulates spine morphology. AMPAR, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptor.

Cytosolic Zn is distributed to different organelles, including synaptic vesicles, Golgi, ER, and mitochondria. Similarly, Zn uptake by these organelles is performed by ZnTs, while ZIPs participate in Zn efflux into the cytosol (155). Zn modulates cellular signaling pathways and acts as a neuromodulator. ZnT3 participates in the transport and accumulation of Zn within synaptic vesicles of glutamatergic terminals (Fig. 6a) (125). During synaptic transmission, Zn is released with Glu, where it inhibits the activity of NMDAR and AMPAR (9, 154), modulating neuronal excitability and long-term synaptic plasticity (long-term potentiation [LTP] and long-term depression). In addition, Zn regulates the activation of the tropomyosin kinase receptor B (TrkB) (304). A metabotropic Zn-sensing receptor (mZnR) has also been reported (31). At the synapse, Zn can also bind to MTs, Aβ, and PrP^C (Figs. 3c and 6).

Although Zn²⁺ is a nonredox-active metal ion, it participates in redox signaling through several mechanisms. MTs bind about a fifth of the intracellular Zn with a stoichiometry of 1:7. MTs and proteins containing Zn-finger domains use Cys residues to bind to Zn²⁺ ions. Zn coordination stabilizes the reduced state of Cys thiol groups preventing their oxidation and subsequent formation of disulfide bonds. Transcription of MTs is induced by oxidative stress via Nrf2, and by heavy metal exposure via metal-response elements (MREs). MREs are recognized by the Zn-finger domain containing protein metal-responsive transcription factor-1 (MTF1, also known as MRE-binding transcription factor-1 or metal regulatory transcription factor-1). MTF1 senses Zn levels. Furthermore, via Zn displacement from MTs or Cys oxidation, MTF1 also senses heavy metal toxicity (Cd) and oxidative stress. In addition to MTs, MTF1 regulates the transcription of a number of genes involved in redox homeostasis and metal ion detoxification, including Zn transporters, Fr, Fpn, ATP7(A/B), Trx, selenoproteins, and γ-glutamate-cysteine ligase (GCL), a rate-limiting enzyme in GSH synthesis (119). The role of MTF1 in brain function and redox homeostasis is unclear. MTF1 has been shown to regulate the expression of β-synuclein (220), which is thought to act as a negative regulator of its homologue α-synuclein (126). Interestingly, deletion of MTF1 induces lethality in Parkin-deficient flies (Drosophila melanogaster) (293).

Altered Zn levels have been reported to promote neuronal injury. Upon Cys oxidation, Zn is released from MTs (255). Cellular acidification also releases intracellular Zn in neurons (159). In a recent report, AMPA-induced oligodendrocyte cell death was shown to be linked to Zn mobilization from mitochondria and protein-bound pools that were mediated by cytosolic acidification, independently from ROS (213). Exposure of mitochondria to Zn promotes increased ROS formation (305). While no specific mitochondrial Zn transporter(s) has been identified, potential candidates include the MCU, ZIP8, and Znt2 transporters (30, 199, 307). Zn also increases NOX-derived ROS formation and NOS activity (162). High extracellular Zn enhances microglia activation and ROS formation (130). Zn deficiency also induces oxidative stress via a reduction in the activity of SOD1 (378), and an impairment in the transcriptional regulation of GCL by Nrf2 (253).

Alterations in Zn homeostasis are associated to neurodegenerative diseases. Serum levels of Zn are decreased in AD patients (358), while Zn is enriched in Aβ plaques (364). Furthermore, a decrease in the levels of Znt3 and MT-III (390) is found in AD. Importantly, the predominant localization of Aβ plaques in Zn-containing glutamatergic synapses might explain why they are primarily found in the neocortex (304). Zn promotes a rapid, but reversible, aggregation of Aβ that is different to the aggregation of Aβ or Aβ-Cu complexes (25). Zn also reduces the toxicity of Cu-induced Aβ aggregates (214). ZnT3 knockout increases soluble Aβ in transgenic APP mice corroborating the role of extracellular Zn in plaque formation (172).

Elevated Zn levels have been found in the SN of PD brains (74), while reduced levels of Zn in serum and plasma have been linked to an increased risk for PD (80). Zn has been shown to potentiate the toxicity of DA as well (189). Furthermore, Zn chelation reduces the toxicity of mitochondrial PD-related toxins (310). Recently, ATP13A2 was identified as a Zn transporter localized to multivesicular bodies. Loss of function mutations of ATP13A2 induces alterations in Zn homeostasis and mitochondrial dysfunction (259).

Arsenic

As is naturally present in air, water, and soil and is the 20th most abundant element in the earth's crust and 12th in the human body. This metal is named inorganic As (iAs) when found combined with other elements such as O₂, chlorine (Cl), and sulfur (S). Combined with carbon (C) and hydrogen (H) is referred to as organic As. In the environment and within the human body, iAs predominantly exists in two oxidation states: arsenite +3 (or AsIII, found as arsenic trioxide [As₂O₃], sodium arsenite [NaAsO₂], and arsenic trichloride [AsCl₃]), and arsenate +5 (or AsV found as arsenic pentoxide [As₂O₅], arsenic acid [H₃AsO₄], and arsenates [PbHAsO₄, Ca₃(AsO₄)₂]).

iAs has been widely used as a therapeutic agent to treat leukemia. Currently, iAs compounds are predominantly used in pesticides, herbicides, cotton desiccants, wood preservatives, alloys for batteries, and in semiconductors and light-emitting diodes. Millions of individuals are currently exposed to iAs across the world due to natural groundwater contamination. Fish and crustaceans contain very high levels of organic arsenobetaine but no toxicity has been reported in vivo (382). The concentration of iAs in natural surface and groundwater is generally about 1 parts per billion (ppb) of water but it may exceed 1000 ppb in contaminated areas or where iAs soil levels are high (118, 200) (Supplementary Table S1).

While As is considered a carcinogen, in the brain, acute exposure to iAs can induce encephalopathy, with symptoms such as confusion, hallucinations, reduced memory, and emotional lability (exaggerated changes in mood or affect). Long-term exposure to lower levels of iAs can lead to the development of peripheral neuropathies.

There are reports of neurobehavioral alterations (cognitive function, verbal abilities, long-term memory, and motor skills) in children exposed to As concentrations ranging from 5 to 50 ppb in water, in Bangladesh (260), Mexico (45, 287), and in the United States (370). Although scientific understanding of the developmental neurotoxicity of As is still evolving, epidemiological and toxicological studies clearly show that As is a developmental neurotoxicant that affects intellectual function. Moreover, exposures even below current safety guidelines are associated with decrements in full-scale intelligence quotient (IQ) and memory (90, 347). Evidence in experimental models, including mice, rats, Caenorhabditis elegans (worm), and Danio rerio (zebrafish), has replicated many of the observations in humans supporting the notion that As can lead to cognitive, locomotor, and neurological impairment (85). Gestational exposure to NaAsO₂ leads to a significant iAs accumulation in the mice offspring's brain (280). As neurotoxicity has been linked to changes in neurotransmitter metabolism and synaptic transmission (85, 276, 280). However, the mechanisms involved remain unclear.

In the environment, oxygenated water contains iAsV species, while in reducing environments iAsIII species are prevalent. iAsV enters cells through phosphate transporters to be subsequently reduced to iAsIII, while iAsIII is transported via aqua(glycerol)porins (AQP), organic anion transporters, and glucose transporters (GLUT) (Fig. 7a) (44, 187, 348). Once in the cytoplasm, iAsIII is methylated by different mechanisms. Oxidative methylation (Fig. 7b) is mediated by arsenite methyltransferase (AS3MT) that uses S-adenosylmethionine (AdoMet) as a cosubstrate. AS3MT methylates iAsIII to monomethylarsonic acid or arsonate (MMAV) that is reduced to monomethylarsonous acid (MMAIII) before being methylated again to dimethylarsinic acid (DMAV) by AS3MT (353, 372). Finally, DMAV is reduced generating dimethylarsinous acid (DMAIII). The reduction of pentavalent arsenicals (iAsV, MMAV, and DMAV) in this pathway is now well recognized to be mediated by the Trx/TR system, but GSH seems to increase the methylation rates by an unknown mechanism (76).

OPEN IN VIEWER

FIG. 7.

iAs biochemistry, redox signaling, and neurotoxicity. (a) iAsIII species prevail in anaerobic conditions, while aerobic environments contain iAsV species. iAsIII enters cells through AQP9/7 and GLUT, while iAsV uses phosphate transporters. Three mechanisms for iAs methylation have been proposed. (b) Oxidative methylation refers to the progressive reduction of AsV to AsIII metabolites by the Trx/TR system and the oxidative methylation of AsIII species by AS3MT. (c) The GSH conjugation pathway is based on the formation of As(GS)₃ by the GSTs. As(GS)₃ is subsequently methylated by AS3MT. (d) The protein thiol conjugation mechanism suggests that instead of As(GS)₃, iAsIII binds to protein thiols and is methylated while still being conjugated to proteins. iAsIII methylation is likely to alter epigenetic signatures by alterations in AdoMet, a cosubstrate required for methyl group transfer. (e) Methylated metabolites and GSH/protein complexes are highly reactive and are exported through the MRPs. (f) DMA(GS) can form DMAH and react with O₂ to form DMAH^• and DMAOO^•. DMAIII also reacts with O₂ and forms DMAOOH. These reactive species lead to lipid peroxidation and protein carbonylation (oxidative stress). (g) AsV can replace phosphate in several metabolic pathways (arsenylation) where the end product is reduced AsIII. AsV binds ADP via ATP-synthase and uncouples oxidative phosphorylation and ATP formation in the mitochondria. AsV is also arsenylated by GAPDH to generate iAsV-3-P-glycerate from G3P. (h) iAsIII toxicity generates ER stress likely via thiol depletion and alterations in redox balance. (i) AsIII binds to thiol containing molecules (coenzyme A, DLA, GSH) and protein thiols inactivating enzymatic function. (j) iAsIII toxicity is counteracted by Nrf2 and MTF1 that mediate the transcriptional-dependent induction of antioxidant systems and MTs. AdoMet, S-adenosylmethionine; AQP, aqua(glycerol)porin; AS3MT, arsenite methyltransferase; AsIII, arsenite +3; As(SG)₃, arsenic triglutathione; AsV, arsenate +5; DLA, dihydrolipoamide; DMA(GS), dimethylarsinic GSH; DMAH, dimethylarsine; DMAH^•, DMAH radical; DMAIII, dimethylarsinous acid; DMAOO^•, dimethylarsine peroxyl radical; DMAOOH, dimethylated arsenic peroxide; G3P, glyceraldehyde 3-phosphate; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GLUT, glucose transporters; GSTs, glutathione-S transferases; iAs, inorganic As; iAsV-3-P-glycerate, 1-arsenato-3-phospho-D-glycerate; MRPs, multidrug resistance proteins; MTF1, metal-responsive transcription factor-1; Nrf2, nuclear factor erythroid-2-related factor 2; O₂, molecular oxygen; Trx, thioredoxin.

Developmental exposure to As alters the methylation patterns of genes involved in neuroplasticity likely due to changes in AdoMet, but its long-term implications are unclear (207). Recent in vivo studies demonstrated that the alterations in synaptic plasticity (LTP), memory, and learning induced by gestational exposure to iAs were associated with an increase in extracellular Glu levels and downregulation of AMPAR subunits (244).

As methylation via the GSH conjugation mechanism is based on the formation of GSH complexes with iAsIII resulting in arsenic triglutathione [As(SG)₃] (Fig. 7c). Conjugation of iAsIII with GSH has been proposed to occur nonenzymatically, but enzymatically as well by the activity of glutathione-S transferases (GST isoforms GSTO1, GSTM1, or GSTP1) (173, 372). As(GS)₃ is subsequently methylated by AS3MT to form monomethylarsinic diglutathione [MMA(GS)₂] and then again to generate dimethylarsinic GSH [DMA(GS)]. At low GSH levels, As(GS) conjugates are hydrolyzed and then oxidized to generate MMAV and DMAV (372). A third mechanism for iAsIII methylation has been recently proposed, where instead of As(GS) conjugate formation, iAsIII binds to protein-Cys (thiol) and is methylated while still being conjugated to proteins (Fig. 7d). This hypothesis is supported by the preferential binding of iAsIII to protein-Cys when compared to GSH (284).

Methylated (and maybe unmethylated) As metabolites are exported through the multidrug resistance proteins (MRP1, MRP2, or MRP4) (173, 317, 388) (Fig. 7e). AS3MT is ubiquitously expressed in all brain regions, and animal studies have shown that the iAs that crosses the BBB is methylated and accumulated across the brain, with the highest accumulation observed in the pituitary gland (297). Interestingly, knockout mouse for P-glycoprotein accumulates more As in the brain (183). Endothelial cells and astrocytes feet surrounding capillaries are the first barrier of detoxification of xenobiotics entering from the circulation. We (unpublished data) and others have observed that the resistance of astrocytes to iAsIII is mediated by MRPs (332).

iAs generates ROS and dimethylarsenic or peroxyl radicals that in turn lead to lipid peroxidation and the accumulation of oxidized by-products (MDA and HNE) (Fig. 7f). Importantly, MMAIII and DMAIII are proposed to be more potent toxicants than iAsIII due to their increased ability to generate radicals (392). Oxidative stress has been reported in brain regions of different animal models and in neurons and glial cell cultures exposed to As compounds (48, 107, 108, 243, 394).

Mitochondria have been proposed to be a primary source for ROS formation by iAs (Fig. 7g) (97, 150). Chronic iAs exposure generates mitochondrial oxidative stress in the rat brain by impairment of mitochondrial complexes I, II, and IV activities followed by increased ROS generation, lipid peroxidation, and protein carbonylation (275). Mitochondrial pyruvate dehydrogenase is also directly inhibited by iAs (136). In addition, iAs reduces the levels of peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α), downstream targets Nrf1 and Nrf2, and the mitochondrial transcription factor A (TFAM) decreasing mitochondrial biogenesis (274). ER stress has also been shown to contribute to iAs toxicity, but the mechanisms involved remain unclear (Fig. 7h) (182).

iAs toxicity has also been attributed to the ability of AsV to replace phosphate in several metabolic pathways (arsenylation) where the end product is the reduction of AsV to AsIII, because the arsenylated by-product is more readily reduced than AsV itself (Fig. 7g). AsV uncouples oxidative phosphorylation and ATP formation in the mitochondria by binding to ADP via ATP synthase. Replacement of phosphate in glycolysis also impairs carbon flux and ATP production. Reaction of AsV with glucose generates glucose 6-arsenate, an analog of glucose 6-phosphate that is suggested to act as an inhibitor of hexokinase. AsV is also arsenylated by glyceraldehyde 3-phosphate dehydrogenase (GAPDH) to produce the unstable product 1-arsenato-3-phospho-D-glycerate (iAsV-3-P-glycerate). Purine nucleoside phosphorylase, glycogen phosphorylase, and mitochondrial ornithine carbamoyl transferase (OCT) have also been shown to arsenylate AsV (139, 245, 340). Thus, energy failure, alterations in central carbon metabolism, and mitochondrial dysfunction are consequences of AsV toxicity.

AsIII binds to thiol containing molecules (coenzyme A, GSH, and dihydrolipoamide also known as dihydrolipoic acid [DLA]) and protein-Cys thiols inactivating enzyme function (Fig. 7d, i). AsIII has higher affinity for dithiols than monothiols as demonstrated by the transfer of AsIII from the GSH-adduct to 2,3-dimercaptosuccinic acid (DMSA) a sulfhydryl-containing metal chelator used to treat heavy metal toxicity. In addition, AsIII conjugated with GSH has the ability to bind protein thiols, which highlights the importance of detoxification of GSH-As adducts from the cell (230). Dithiol molecules such as the cofactor DLA and dithiol oxidoreductases Trxs, Trx reductase (TrxR), Prxs (except for monothiol Prx6), Grx, and GR, as well as proteins with adjacent Cys (MT), have been reported to avidly bind AsIII (Fig. 7d, j) (50, 279, 311, 393, 397). In addition to binding AsIII, Zn finger domains have been shown to be oxidized upon As binding (396). Binding of AsIII to DLA (Fig. 7i) is expected to interfere with the TCA cycle and energy production as DLA is a cofactor for the pyruvate dehydrogenase and 2-oxoglutarate dehydrogenase complexes that catalyze the synthesis of acetyl-CoA and succinyl-CoA, respectively. DLA reverses protein oxidation and loss of protein-SHs in the brains of rats exposed to high levels of iAs (82, 296, 315, 316).

As activates the cystine/Glu exchanger system (xCT) in microglia to increase extracellular Glu levels (320), while in astrocytes it decreases the expression levels and activity of GS and Glu transporters (GLAST/excitatory amino acid transporter (EAAT) 1 and GLT-1/EAAT2) (48, 395). Importantly, these effects were linked to an increase in GSH levels and Nrf2 activity (Fig. 7i), but not oxidative stress (48). Accordingly, activation of Nrf2 by As seems to involve a noncanonical pathway where inhibition of autophagy leads to the accumulation of the ubiquitin-binding protein/adaptor p62 that sequesters Keap1 (169). iAsIII toxicity is also counteracted by the transcriptional regulation of MT via MTF1 (Fig. 7j) (129).

Lead

Inorganic Pb remains one of the most studied toxic elements due to several reasons. To begin with, human contact with Pb started very early in human civilization, and its toxic effects were also known since then. However, more importantly, Pb is neurotoxic leading to lower IQ even at lower doses than those recommended by the World Health Organization (10 ppb in drinking water). Human exposure to Pb not only occurs occupationally but also environmentally. The presence of Pb in the environment has multiple sources such as gasoline, industrial processes, paint, water pipes, and solder in canned food. It is present in air, household dust, soil, water, and food (Supplementary Table S1). Environmental Pb levels have fortunately decreased, especially in those countries where the Pb addition to gasoline and paints was banned. This prohibition was enforced after several studies associated the presence of high blood levels of Pb with impaired or diminished cognitive functions. Epidemiological studies have clearly shown that exposure to Pb in early stages of development is associated with significant deficits in neurobehavioral performance, including lower IQ, attention deficits, and aggressiveness later in life. Despite all the enforced restrictions, Pb contamination is still a major public health concern. For example, in November 2000 in Washington DC, there was a “lead drinking water crisis” triggered by a change in the disinfectant used to clean the water, this contamination affected hundreds of kids for 3 years. The health consequences of the recent crisis of Pb-contaminated water in Flint Michigan (United States, 2015) are still to be revealed in the future (68, 84, 124, 298).

Inhalation and ingestion of Pb and Pb-containing particles or products are the main routes of Pb entry into the body. Young children are especially vulnerable because they show higher gastrointestinal absorption than adults. Inhaled Pb particles are quickly absorbed in alveoli and distributed to other organs through the circulation. Thus, blood lead levels (BLL) are reliable biomarkers of exposure and risk. However, BLL do not reflect the total Pb body burden because Pb is absorbed in bones where it can be stored for several years (68). Currently, the acceptable BLL for children is lower than 10 μg/dl (0.48 μM) (49, 376), but due to the devastating effects that might occur later in life, there is a consensus to recommend efficient surveillance methods for children protection to reduce BLL to the lowest possible level (141). Pb binds with high affinity to erythrocytes' δ-aminolevulinic acid dehydratase (ALAD) that catalyzes the second step in the porphyrin and heme biosynthetic pathway, causing the accumulation of aminolevulinic acid (ALA) in both plasma and urine, which is used as a biomarker of exposure (248).

Pb can cross the BBB and cell membrane because of its ability to mimic Ca²⁺ and Fe²⁺ ions (Fig. 8a) (205, 298). In children, due to a more permeable BBB and a lower bone storage capacity for Pb, the amount of Pb passing into the nervous system is higher than in adults. The highest accumulation of Pb has been reported in the hippocampus, amygdala (116), and choroids plexus (201). The PNS may accumulate considerably more Pb than the CNS. Animals chronically exposed to Pb had impaired dendritic spines and synapse formation (306). Developmental exposure to Pb impacts the prefrontal cerebral cortex, hippocampus, and cerebellum regions, which can lead to neurological disorders, mental retardation, behavioral problems, and nerve damage (242). Early life exposure to Pb has also been linked to neurodegenerative diseases such as AD and PD (62, 209).

OPEN IN VIEWER

FIG. 8.

Pb neurotoxicity and synaptic signaling. (a) Pb enters cells due to its ability to mimic Ca²⁺ and Fe²⁺ ions. (b) Inside the cell, Pb can trigger oxidative stress by different mechanisms, including the accumulation of ALA; the formation of Pb²⁺–O₂ ^•− complexes; and the release of labile Fe²⁺ from Ft. (c) Pb can also form complexes with GSH [Pb(GSH_2–3)] and bind to protein thiols. (d) Mitochondrial dysfunction and oxidative stress induced by Pb have also been linked to MCU. (e) Pb can displace Zn²⁺ and impair Zn²⁺ binding protein function. Furthermore, Pb interferes with the Se-dependent redox processes. (f) Finally, Pb can prevent CaM activation and thus reduce NOS activity. ALA, aminolevulinic acid; CaM, Ca²⁺/calmodulin; NOS, nitric oxide synthase; O₂ ^•−, superoxide anion radical; Pb, lead; Se, selenium.

The mechanisms of Pb toxicity include the ability of Pb to bind SH groups of proteins Cys and to mimic or compete with Ca²⁺, Fe²⁺, and Zn²⁺ (Fig. 8b, c, e, f) (99, 283). Zn deficiency increases the toxicity of Pb (4). The generation of oxidative damage by Pb in vitro and in vivo suggests that ROS also participate in Pb toxicity. For example, Pb acetate induces the opening of the mitochondrial permeability transition pore in human neuroblastoma SH-SY5Y cells via ROS (384). Pb can form a Pb²⁺–O₂ ^•− complex with higher oxidizing capacity than O₂ ^•− (3). In addition, accumulated δ-ALA by Pb-induced ALAD inhibition can be subsequently oxidized to generate O₂ ^•−, ^•OH, and H₂O₂. Pb per se has been reported to stimulate Fe²⁺-initiated lipid peroxidation (Fig. 8b) (337). Early postnatal exposure of rats to Pb leads to a higher accumulation of oxidative DNA damage in the cerebral cortical tissue when compared with aged controls or aged mice exposed acutely to Pb (36).

Perinatal exposure to Pb acetate inhibits the activity of brain acid and alkaline phosphatases, catalase, acetylcholinesterase, and ATPases (12). Similar observations have been made for the activities/levels of SOD1, GPX1, and GPX4 in the hippocampus, and for mitochondrial SOD2 and GSH, both in the cortex and hippocampus (19). Antioxidant nutrients such as vitamin E, vitamin C, vitamin B6, β-carotene, and DLA, as well as metal chelators such as DMSA, or replenishment of displaced metals has been shown to be beneficial against Pb-induced oxidative stress in the brain (98, 236, 256, 258, 277, 367). Diet supplementation with Zn and Se, which participates in the regulation of the GSH and Trx antioxidant systems, can effectively outcompete Pb binding to Zn- and Se-binding sites (Fig. 8e) (135).

Pb interferes with and disrupts Ca²⁺ signaling and homeostasis leading to excitotoxicity. In addition, Glu potentiates Pb-induced cell death in PC12 cells (267). Recently, oxidative stress induced by Pb has been shown to be linked to changes in the levels of MCU (Fig. 8d) (383). Other important intracellular targets of Pb in the brain are both neural NOS and endothelial NOS due to an impairment in their Ca²⁺/calmodulin (CaM)-dependent activation (Fig. 8f) (241). Importantly, Pb amplifies Glu-induced oxidative stress in a Ca²⁺-independent manner, but neither Ca²⁺ nor ROS seem to be essential for the enhanced cytotoxicity of combined exposure to Glu and Pb (191, 238).

Mercury

Hg is a transition metal that exists as elemental, inorganic, and organic Hg (Fig. 9a). Hg is ubiquitously found in the environment as sulfide compounds generated from volcanic activity and erosion, or released by anthropogenic sources such as fuel combustion, waste disposal, and industrial activities (Supplementary Table S1). Elemental or metallic Hg (Hg⁰) used in thermometers and amalgams is primarily absorbed via inhalation, while inorganic mercury (Hg^{1+ or 2+}) used in medicine and everyday life products is partially absorbed through the gut. Organic Hg (ethylmercury [EtHg or C₂H₅Hg] and methylmercury [MeHg or CH₃Hg]) is originated from atmospheric sources that are deposited in water body surfaces to be biomethylated and magnified in the food chain (Fig. 9a). Around 95% of MeHg is absorbed by the gastrointestinal tract making it the most toxic Hg species. Neurotoxic signs of Hg intoxication are vast and include ataxia, dizziness, insomnia, speech impairment, arthralgia, cognitive and behavioral changes, seizures, fatigue, and sensory disruption. While there has been an association between Hg exposure and neurodegeneration or autism, the neurological effects of chronic exposure to Hg are largely unclear. However, research has clearly demonstrated that Hg impairs neuronal development, communication, and myelination (89).

OPEN IN VIEWER

FIG. 9.

Hg redox-related toxicity mechanisms in brain cells. (a) Hg species are highly transformed (oxidation or methylation) before they are absorbed into organisms. MeHg and EtHg form a complex with Cys (CH₃HgCys) and are transported across the membranes via LAT. (b) MeHg suffers dealkylation and thiol exchange to form Hg²⁺. (c) Thiol exchange reactions of CH₃HgCys complexes with low-molecular-weight thiols (GSH) and protein thiols alter the cellular redox balance. MeHg also has a strong affinity for selenol groups. (d) GSH-MeHg adduct formation (CH₃HgGS) is catalyzed by the GSTs and adducts are detoxified by MRPs. (e) MeHg also induces mitochondrial ROS and energy failure. Furthermore, MeHg-induced neurotoxicity has been ascribed to a reduction in Glu uptake by astrocytes via EAAT triggering neuronal excitotoxicity (arrow in EAAT indicates reversal of Glu transport). Cys, cysteine; Hg, mercury; EtHg, ethylmercury; Hg¹⁺, inorganic mercurous ions; Hg²⁺, inorganic mercuric ions; LAT, L-type neutral amino acid transporter; MeHg, methylmercury.

Most of the studies regarding the mechanisms involved in Hg neurotoxicity have been done using MeHg. MeHg and EtHg are potent electrophiles that form a complex with Cys (CH₃HgCys or C₂H₅HgCys), and then transported across the BBB and into neuronal cells via L-type neutral amino acid transporters (LAT1 and 2) (Fig. 9a) (318, 385, 399).

A high percentage of Hg in individuals intoxicated with MeHg is found as Hg²⁺, suggesting that dealkylation of MeHg is an important mechanism for the high persistence of Hg in the brain (Fig. 9b) (72). Thiol exchange from CH₃HgCys to low-molecular-weight thiols (GSH) and protein thiols has been proposed to be central mechanisms by which MeHg induces GSH depletion, inhibition of thiol-dependent antioxidant systems, and alters the activity or function of proteins with redox-sensitive Cys (signaling proteins, metabolic enzymes, neurotransmitter receptors, and transporters) (Fig. 9c) (89). MeHg also has a stronger affinity for selenol groups (selenohydryl groups in selenocysteines) compared with thiol groups. As such, selenoproteins are important targets for direct electrophilic attack of MeHg or transfer from thiol adducts (CH₃HgCys, CH₃HgGS, or CH₃HgPS [protein-Cys adduct]) (Fig. 9c) (104, 221). GSTs have been proposed to mediate the formation of CH₃HgGS adducts, which are detoxified by MRP1-mediated transport (Fig. 9d). GSH synthesis, GST, and MRP1 levels are regulated transcriptionally by the Nrf2 antioxidant system (152, 292, 349).

MeHg induces mitochondrial ROS and energy failure (175, 233). Neurotoxicity induced by MeHg has also been ascribed to its inhibitory effect on Glu uptake by astrocytes, triggering neuronal excitotoxicity (15, 235) (Fig. 9e).

Hg⁰ absorbed through the respiratory tract is oxidized to inorganic mercurous (Hg¹⁺) and mercuric ions (Hg²⁺) (Fig. 9a). While inorganic Hg ions have limited access to the CNS, they induce profound neurotoxic alterations that seem to be mediated as well by their binding to thiol groups (89). Accordingly, MTs exert protective effects against Hg⁰-induced neurotoxicity (387) (Fig. 9b).

Other xenobiotic metals

Aluminum

Al is one of the most abundant metals in the earth's crust (8.1%). Al has a plethora of uses in industry and manufacturing, as well as in food additives. As such, human exposure is primarily originated from food and drinking water. Importantly, pharmaceuticals have higher levels of Al compared to food. Occupational exposures to Al are related to mining, processing, and welding (359) (Supplementary Table S1). While Al is poorly absorbed in the gut, inhalation mediates direct transfer to the brain via the olfactory system (341). Importantly, ∼85% of Al in blood is bound to Tf, which is considered to mediate its transport across the BBB (Fig. 10a) (288), but Tf-independent Al transport also exists. Interestingly, monocarboxylate and xCT transporters have also been proposed to mediate the transport of Al-citrate complexes (240, 386). Acute Al toxicity occurs as a result of occupational exposure or chronic renal failure and is known to target the nervous system. Al is neurotoxic in animal models triggering the accumulation of neurofibrillary tangles and impairment of cognitive, behavioral, and motor functions. Al promotes Aβ aggregation, mitochondrial dysfunction, and triggers neuroinflammation (Fig. 10b, c) (24, 202, 309). However, conflicting results exist regarding the association of Al with any human disease, including AD (37, 359, 377).

OPEN IN VIEWER

FIG. 10.

Mechanisms of Al and Cd neurotoxicity. (a) Al is bound to Tf, which mediates its transport across the BBB via the Tfr, but a Tf-independent Al transport also exists. (b) Al promotes Aβ aggregation and (c) mitochondrial dysfunction trigging neuroinflammation and NO^• formation. (d) Al³⁺ has been proposed to react with H₂O₂ to produce AlO₂ ^•− that can promote oxidative damage. (e) Al toxicity is also related to its ability to displace other metals (Ca or Fe). (f) Cd transport across membranes is through molecular mimicry via several transporters such as Cu/Zn transporters, DMT1, and Ca²⁺-channels. (g) High concentrations of Cd can block Ca²⁺ currents as well. (h) Cd toxicity is linked to its ability to bind thiol containing molecules such GSH, and protein-Cys (Trx and MT), as well as the displacement of redox-active metals such as Fe. Cd detoxification of cells is facilitated by the activity of GST and the detoxification of GSH-Cd adducts via MRPs. Al, aluminum; AlO₂ ^•−, aluminum superoxide radicals; Ca, calcium; Cd, cadmium.

Al exists primarily in a trivalent state (Al³⁺). While Al has no redox capacity, Al toxicity is linked to oxidative damage. Al³⁺ has been proposed to react with H₂O₂ to produce Al superoxide radicals (AlO₂ ^•−) that can deplete mitochondrial Fe and promote generation of ROS (Fig. 10d) (166). However, because of its high reactivity, Al is primarily found forming insoluble oxides whose toxicity seem to be related with the displacement of other biological cations (Ca²⁺, Fe²⁺, or Mg²⁺) (Fig. 10e) (377).