The Role of Reactive Oxygen Species in Arsenic Toxicity

Arsenic poisoning is a global health problem. Chronic exposure to arsenic has been associated with the development of a wide range of diseases and health problems in humans. Arsenic exposure induces the generation of intracellular reactive oxygen species (ROS), which mediate multiple changes to cell behavior by altering signaling pathways and epigenetic modifications, or cause direct oxidative damage to molecules. Antioxidants with the potential to reduce ROS levels have been shown to ameliorate arsenic-induced lesions. However, emerging evidence suggests that constructive activation of antioxidative pathways and decreased ROS levels contribute to chronic arsenic toxicity in some cases. This review details the pathways involved in arsenic-induced redox imbalance, as well as current studies on prophylaxis and treatment strategies using antioxidants.


Introduction
Arsenic is the 33rd element in the periodic table of elements. It displays features of both a metal and a non-metal, and thus called metalloid. however, it is referred as a heavy metal from a toxicological point of view [1]. Arsenic can be found in soil, water, and air from natural and anthropogenic sources. Over time, arsenic has accumulated a variety of uses, for example in cosmetics, wood preservatives, cotton desiccants, pesticides, and even in the treatment of acute promyelocytic leukemia (APL) [2][3][4]. however, evidence of arsenic poisoning has occurred from oral ingestion of arsenic, when found as a contaminant in food or potable water and when used as a therapeutic. The major route of exposure is via drinking water due to natural contamination of groundwater by inorganic arsenic in the earth's crust, which threatens the health of more than 140 million people worldwide [5,6]. Chronic exposure to arsenic has been associated with the development of a wide range of human cancers (e.g., lung, skin, liver, bladder, and kidney) [7], as well as other nonmalignant disorders (e.g., respiratory illnesses, cardiovascular diseases, diabetes, neurotoxicity, and renal diseases) [6,[8][9][10].
Arsenic exists in nature as both an inorganic trivalent (arsenite: iAs III) and inorganic pentavalent forms (arsenate: iAs V), and is metabolized via biomethylation involving a two-electron reduction of pentavalent arsenicals followed by oxidative methylation to form organic pentavalent arsenicals [11,12]. Inorganic arsenic can be methylated to organics by S-adenosyl-L-methionine (SAM), including monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA) in humans with trivalent (MMA III and DMA III, respectively) and pentavalent forms (MMA III and DMA III, respectively) [8].

ROS and Arsenic Toxicity
The toxic mechanisms of arsenic are complex and not fully understood. At a biochemical level, iAs V can replace phosphate in several reactions. Arsenite (iAs III) and trivalent organic (methylated) arsenicals react with thiols (-SH) in proteins and inhibit their activity. Other mechanisms include epigenetic alteration, oxidative stress, inflammation, and autophagic defects [24][25][26]. ROS are formed in biological systems during the reduction of molecular oxygen and include the superoxide radical anion (O 2 −• ), hydrogen peroxide (H 2 O 2 ), hydroxyl radical ( • OH), hydroperoxyl radical (HOO • ), singlet oxygen ( 1 O 2 ), and peroxyl radical (ROO • ) [27]. Arsenic induces formation of 1 [28]. Mechanisms responsible for generation of ROS induced by arsenic are proposed as follows ( Figure 1). (i) Mitochondria: Mitochondrial complexes I and III in electron transport chain are responsible for the production of O 2 −• . Arsenic shows mitochondrial toxicity by inhibiting succinic dehydrogenase activity and uncoupling oxidative phosphorylation with production of O 2 −• , which gives rise to other forms of ROS [29]. (ii) Nicotinamide adenine dinucleotide phosphate (NAD(P)H) oxidase (Nox): Nox is a membrane-associated enzyme involved in ROS generation in response to arsenic. The evidence mainly comes from endothelial cells and is reviewed by Ellinsworth [30]. (iii) Generation of ROS during formation of intermediate arsine species [31,32]. For example, dimethylarsenic peroxyl radical is formed during the metabolic processing of DMA [33]. (iv) Redox-active iron released from ferritin caused by methylated arsenic species [34].
Despite of ROS overproduction reported in earlier studies applying high-dose arsenic in acute exposure, emerging evidence suggests that ROS do not increase when cells are exposed to an environmentally relevant dose of arsenic [40,41], especially after chronic exposure [42]. Treatment with 1 µM of arsenite for 4 h did not alter the amount of ROS in human bronchial epithelial cells (Beas-2B) [43]. Similarly, acute exposure to arsenite (2 or 4 h) below 10 µM of arsenite did not alter the intracellular amount of ROS in various cell lines, even with a more sensitive ROS detection method (electron paramagnetic resonance spectroscopy) [40,41]. Reduced intracellular ROS levels were observed in arsenic-transformed Beas-2B cells, and moreover were indicated as contributing to the acquisition of malignant phenotypes [42]. Our recent study found that arsenite-transformed human keratinocytes showed dysregulated autophagy with enhanced p62-NRF2 (nuclear factor (erythroid-derived 2)-like 2) feedback loop and decreased intracellular ROS levels [44]. In this scenario, constitutive Nrf2-mediated antioxidant response is frequently observed (see below). Thus, it is debated that adaptive antioxidant Figure 1. Mechanisms of generation of reactive oxygen species (ROS) induced by arsenic. Arsenic (As) induces significant ROS generation mainly through the mitochondrial (Mit) electron transport chain. Activation of nicotine adenine disphosphonucleotide (NADPH) oxidase (Nox) also contributes to the generation of superoxide anion (O2 −• ). Additionally, arsenic metabolism leads to the generation of ROS in cells. 1 O2, singlet oxygen; H2O2, hydrogen peroxide; • OH, hydroxyl radicals; ER, endoplasmic reticulum; SOD, superoxide dismutase.

Nrf2-ARE Pathway
Nrf2 is a master transcription factor in antioxidant system. Under physiological conditions, the protein level of Nrf2 is low because it binds to its negative regulatory factor kelch-like epichlorohydrin-associated protein 1 (Keap1), which forms the E3 ubiquitin ligase complex and facilitates ubiquitination and subsequent degradation of Nrf2 by the 26S proteasome [51]. When excessive ROS are generated, certain cysteine residues (C273, C288, and C151) in Keap1 sense the stress and are S-alkylated [52,53], leading to impairment of Keap1-mediated Nrf2 degradation [54]. Nrf2 then accumulates in the cytoplasm, translocates in the nucleus, dimerizes with small musculo-aponeurotic fibrosarcoma (Maf) proteins, and binds to the ARE motif in the promotor region of target genes, including various antioxidant enzymes and detoxification enzymes [55][56][57]. Therefore, the role of Nrf2-ARE pathway attracts much more attention in studies on arsenic toxicity. Arsenic is an activator of Nrf2-Keap1 pathway [58][59][60][61][62]. Arsenite binds to the Ring finger domain of Ring-box 1 (Rbx1), which leads to the suppression of Cullin 3(Cul3)-Rbx1 E3 ubiquitin ligase activity, thereby activating the Nrf2-induced antioxidant signaling pathways [62]. Meanwhile, many research groups report that arsenic induces Nrf2 activation via the noncanonical mechanism, specifically by p62 accumulation due to dysregulated autophagy flux [43,[63][64][65][66]. Accumulation of p62 results in sequestration of Keap1 in the autophagosomes and impairs Nrf2 degradation [67]. On the other hand, p62 is a downstream gene of Nrf2, forming a positive feedback loop [63]. The loop may act as a critical molecular alteration in arsenic carcinogenesis [68]. In addition, arsenite is found to induce acetylation of Nrf2 by p300/CREB (cAMP response element binding protein) binding protein (CBP), which enhances Nrf2 binding capacity to promoter-specific DNA [69] (Figure 2).

Nrf2-ARE Pathway
Nrf2 is a master transcription factor in antioxidant system. Under physiological conditions, the protein level of Nrf2 is low because it binds to its negative regulatory factor kelch-like epichlorohydrin-associated protein 1 (Keap1), which forms the E3 ubiquitin ligase complex and facilitates ubiquitination and subsequent degradation of Nrf2 by the 26S proteasome [51]. When excessive ROS are generated, certain cysteine residues (C273, C288, and C151) in Keap1 sense the stress and are S-alkylated [52,53], leading to impairment of Keap1-mediated Nrf2 degradation [54]. Nrf2 then accumulates in the cytoplasm, translocates in the nucleus, dimerizes with small musculo-aponeurotic fibrosarcoma (Maf) proteins, and binds to the ARE motif in the promotor region of target genes, including various antioxidant enzymes and detoxification enzymes [55][56][57]. Therefore, the role of Nrf2-ARE pathway attracts much more attention in studies on arsenic toxicity.
Arsenic is an activator of Nrf2-Keap1 pathway [58][59][60][61][62]. Arsenite binds to the Ring finger domain of Ring-box 1 (Rbx1), which leads to the suppression of Cullin 3(Cul3)-Rbx1 E3 ubiquitin ligase activity, thereby activating the Nrf2-induced antioxidant signaling pathways [62]. Meanwhile, many research groups report that arsenic induces Nrf2 activation via the noncanonical mechanism, specifically by p62 accumulation due to dysregulated autophagy flux [43,[63][64][65][66]. Accumulation of p62 results in sequestration of Keap1 in the autophagosomes and impairs Nrf2 degradation [67]. On the other hand, p62 is a downstream gene of Nrf2, forming a positive feedback loop [63]. The loop may act as a critical molecular alteration in arsenic carcinogenesis [68]. In addition, arsenite is found to induce acetylation of Nrf2 by p300/CREB (cAMP response element binding protein) binding protein (CBP), which enhances Nrf2 binding capacity to promoter-specific DNA [69] (Figure 2). however, a significant down-regulation was found in cardiac Nrf2 and peroxisome proliferator-activated receptor-γ (PPAR-γ) mRNA expression in arsenic-treated Sprague-Dawley (SD) rats compared with control rats [70]. Different response of Nrf2 to arsenic exposure is related to the strains and age of murines, as well as exposure time and dose of arsenic.
Biomolecules 2020, 10, x FOR PEER REVIEW 7 of 30 Figure 2. Regulatory models of the Nrf2-ARE pathway induced by arsenic. Under basal condition, Nrf2 is associated with Keap1 and degraded by proteasomes. Under arsenic-exposed condition, Nrf2 is activated via the canonical and noncanonical mechanisms. Arsenic binds to the Ring finger domain of RING-box 1 (Rbx1), which leads to the suppression of Cul3-Rbx1 E3 ubiquitin ligase activity, thereby activating the Nrf2-induced antioxidant signaling pathway via the canonical mechanism. Arsenic induces Nrf2 activation via the noncanonical mechanism by p62 accumulation due to dysregulated autophagy flux. p62 is a downstream gene of Nrf2, forming a positive feedback loop with Nrf2.
Nrf2 is considered as a protective factor against arsenic toxicity by reducing oxidative stress. Our previous study found that stable knockdown (KD) of NRF2 in human keratinocytes (HaCaT) significantly increased the sensitivity to acute cytotoxicity of inorganic arsenite, whereas KEAP1-KD cells showed a significant resistance to arsenite toxicity [94]. When mouse macrophage cells (RAW 264.7) were exposed to arsenite, a marked increase in ROS occurred in Nrf2-KD cells compared to scramble cells [76]. However, abnormal activation of Nrf2 is suggested to be cancer-promoting [95,96]. On one hand, NRF2-dependent antioxidant and detoxification enzymes promote the detoxification and elimination of ROS to attenuate arsenic carcinogenesis. On the other hand, NRF2 activation may provide cell proliferation or survival advantage by mediating metabolic reprogramming [56,97] and contributing to apoptotic resistance [98][99][100][101], which are important events in the process of arsenic-induced malignant transformation. Long-term exposure to an environmentally relevant dose of arsenic may induce constitutive activation of Nrf2, leading to adaptive antioxidant response, and subsequently contributing to malignant transformation [60,74,102,103]. In chronic arsenic-exposed Chang human hepatocytes, protein levels of nuclear Nrf2 peaked at 8 weeks and significantly elevated afterwards, whereas cytosol Nrf2 did not show significant change, which suggests that chronic arsenic exposure may constitutively activate NRF2 by post-transcriptional mechanism [60]. Furthermore, downstream genes of NRF2; NAD(P)H dehydrogenase [quinone] 1 (NQO1); aldo-keto reductase family 1, member C2 (AKR1C2); and aldo-keto reductase family 1, member C3 (AKR1C3) were overexpressed in chronic arsenic-exposed HaCaT cells [102]. Our recent studies found that silencing NRF2 in HaCaT cells abolished Regulatory models of the Nrf2-ARE pathway induced by arsenic. Under basal condition, Nrf2 is associated with Keap1 and degraded by proteasomes. Under arsenic-exposed condition, Nrf2 is activated via the canonical and noncanonical mechanisms. Arsenic binds to the Ring finger domain of RING-box 1 (Rbx1), which leads to the suppression of Cul3-Rbx1 E3 ubiquitin ligase activity, thereby activating the Nrf2-induced antioxidant signaling pathway via the canonical mechanism. Arsenic induces Nrf2 activation via the noncanonical mechanism by p62 accumulation due to dysregulated autophagy flux. p62 is a downstream gene of Nrf2, forming a positive feedback loop with Nrf2.  Arsenic causes ROS over-production and induces activation of ERK, JNK, and p38 MAPK, as well as expression of IL-6 and VEGF. [75] ROS/ERK1/2/Beclin1, PINK1, Parkin 1, LCIIIB Male Wistar rats NaAsO 2 (10 mg/kg) orally for 3 months PKCδ is activated in the arsenic-intoxicated aged brains, which increases the expression of ERK1/2. ERK1/2 activates its downstream autophagic molecules Beclin1, PINK1, Parkin 1, and LCIIIB. Nrf2 is considered as a protective factor against arsenic toxicity by reducing oxidative stress. Our previous study found that stable knockdown (KD) of NRF2 in human keratinocytes (HaCaT) significantly increased the sensitivity to acute cytotoxicity of inorganic arsenite, whereas KEAP1-KD cells showed a significant resistance to arsenite toxicity [94]. When mouse macrophage cells (RAW 264.7) were exposed to arsenite, a marked increase in ROS occurred in Nrf2-KD cells compared to scramble cells [76]. however, abnormal activation of Nrf2 is suggested to be cancer-promoting [95,96]. On one hand, NRF2-dependent antioxidant and detoxification enzymes promote the detoxification and elimination of ROS to attenuate arsenic carcinogenesis. On the other hand, NRF2 activation may provide cell proliferation or survival advantage by mediating metabolic reprogramming [56,97] and contributing to apoptotic resistance [98][99][100][101], which are important events in the process of arsenic-induced malignant transformation. Long-term exposure to an environmentally relevant dose of arsenic may induce constitutive activation of Nrf2, leading to adaptive antioxidant response, and subsequently contributing to malignant transformation [60,74,102,103]. In chronic arsenic-exposed Chang human hepatocytes, protein levels of nuclear Nrf2 peaked at 8 weeks and significantly elevated afterwards, whereas cytosol Nrf2 did not show significant change, which suggests that chronic arsenic exposure may constitutively activate NRF2 by post-transcriptional mechanism [60]. Furthermore, downstream genes of NRF2; NAD(P)H dehydrogenase [quinone] 1 (NQO1); aldo-keto reductase family 1, member C2 (AKR1C2); and aldo-keto reductase family 1, member C3 (AKR1C3) were overexpressed in chronic arsenic-exposed haCaT cells [102]. Our recent studies found that silencing NRF2 in haCaT cells abolished arsenic-induced acquisition of invasion capacity [44]. These data provide direct proof for the oncogenic role of Nrf2 in arsenic carcinogenesis. Taken together, Nrf2 pathway may exert dual roles in arsenic toxicity depending on the dose, exposure time, and cell types. Thus, concerns about strategy of using natural compounds, such as daphnetin (Daph) as an Nrf2 activator, for arsenic detoxification have been raised [104].

microRNAs
Epigenetic modifications contribute to toxic effects by arsenic exposure [105]. Alteration in miRNAs is one of these modifications and is closely related with intracellular ROS levels. he et al. found that chronic arsenic exposure lead to an overproduction of ROS, which induced activation of the miR-199a-5p/hypoxia inducible factor-1α (HIF-1α)/cyclooxygenase-2 (COX-2) pathway [77]. ROS inhibited miR-199a expression through increasing the promoter methylation of the miR-199a gene by DNA methyltransferase 1 [106]. miR-214 expression was transcriptionally repressed by Nrf2 through ARE within its promoter region in response to arsenic exposure in erythroid cells, and this repression was ROS dependent [78]. Not all alterations in miRNAs in arsenic-induced malignant transformation are related with ROS. Chen et al. found that Nrf2 was modulated by miR-155 in the process of malignant transformation induced by arsenite in human bronchial epithelial cells (16-HBE). however, there was no significant alteration in ROS production as determined by dichlorodihydrofluorescein diacetate (DCFH-DA) probe in the arsenic-transformed cells [107].
The levels of ROS can be regulated by miRNAs, as found in diseases such as cerebral ischemia, ischemia/reperfusion injury, and spinal cord injury. Overexpression of miRNA-20b increased the levels of adenosine 5' triphosphate (ATP) and ROS in the cerebral ischemia of SD rats, whereas suppression of miRNA-20b decreased the levels of ROS [108]. Overexpression of miR-451 decreased apoptosis rate, ROS levels, and cleaved caspase-3 expression in the oxygen and glucose deprivation/reoxygenation cells [109]. Lycium barbarum polysaccharides (LBPs) reduced levels of ROS and nitric oxide (NO) induced by h 2 O 2 through down-regulating miR-194 in PC-12 cells [110].
Understanding the molecular mechanisms of arsenic-induced toxicity, such as dynamics of ROS generation, miRNA expression, and the relationship between ROS and miRNAs, will certainly shed new light for future strategies against arsenic toxicity. however, so far, there are only a few studies exploring the molecular mechanism of arsenic toxicity in this regard. More studies are preferably needed to understand the potential relationship between ROS alteration and miRNA expression in response to arsenic exposure, of which research on the role of ROS in epigenetic dysregulation would be top priority.

Mitophagy
The mitochondrion is the major site for ROS production and leakage [111,112]. Moderate ROS levels are essential for cell proliferation and survival by mitophagy [113,114]. Excessive levels of ROS induce apoptotic signaling pathways. Furthermore, unceasing generated ROS in mitochondria lead to autophagy, apoptosis, or necrosis. On one hand, arsenic has mitochondrial toxicity, resulting in mitochondrial swelling and crista fragmentation, disturbing respiratory complex, and giving rise to ROS [115]. On the other hand, excessive ROS generation causes mitochondrial dysfunction [47,49].
Arsenic inhibits complex I in the mitochondrial electron transport chain, which leads to excessive generation of ROS, giving rise to lipid peroxidation and protein damage and the subsequent formation of mitochondrial permeability transition (MPT) [122]. Arsenite induces mitophagy via mitochondrial ROS and MPT [80]. PINK1/Parkin is activated upon mitochondrial membrane depolarization, a signal of mitochondrial dysfunction that results from multiple causes including hypoxia and impaired electron transport [123]. In arsenic toxicity, mitophagy exerts a dual role, facilitating cell survival either by eliminating damaged mitochondria or causing cell death. Arsenic-induced apoptosis in the pancreas of rats and insulinoma β-cell (INS-1) through impairment of mitophagy mediated by the ROS/Pparγ/PINK1/Parkin pathway [79]. Mitochondrial B cell lymphoma/leukemia-2 (Bcl-2)/adenovirus E1B 19 kDa-interacting protein 3 (BNIP3), a pro-apoptotic Bcl-2 homology 3 (BH3)-only protein, was activated as an upstream signal to increase the expression of caspase 3 and sequestosome 1 (SQSTM1), and contributed to increased cell death caused by arsenic [124]. BNIP3 also interacted with LC3 to target the damaged mitochondria and initiate mitophagy [125].

Tyrosine Phosphorylation
Tyrosine phosphorylation is an important posttranslational modification that is known to regulate receptor kinase (RK)-mediated signaling in mammals [126,127]. Auto-phosphorylation of specific tyrosine residues increases catalytic efficiency of the RK itself, whereas phosphorylation of additional tyrosine residues creates docking sites for downstream signaling molecules [128,129]. The tyrosine phosphorylation system is mediated by two important classes, receptor tyrosine kinases (RTKs) and nonreceptor tyrosine kinases (NTKs). The former includes growth receptors such as epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor (PDGFR), and vascular endothelial grow factor (VEGF). The latter includes Src-family protein kinases. Activation of EGFR leads to activation of MAPK pathways. Arsenic exposure increases tyrosine phosphorylation in numerous proteins, with phosphorylation of EGFR as a central target [81,130,131]. The proposed mechanisms involve interaction of -SH groups on EGFR with arsenic and structural changes or dimerization of EGFR caused by ROS [81,[130][131][132]. In addition, ROS may inactivate negative regulators of EGFR (namely protein tyrosine phosphatases, PTPs) via oxidation of cysteine residues in the active sites of these enzymes [133,134]. Following EGFR activation, Shc is recruited and phosphorylated at its tyrosine residues, which leads to enhanced Shc-Grb2 interaction and downstream signaling transduction including MAPKs [81,133]. In vivo and in vitro studies also found that arsenic induced Src activation in various cell lines [83,84,135], which is proposed to be mediated by direct interactions between arsenic and vicinal -SH groups of Src [135]. ROS may participate in arsenic induced Src activation, but the evidence remains unclear. In addition, Src acts as the upstream of EGFR and MAPK signaling in response to arsenic exposure [135].

AP-1 Pathway
AP-1 is a complex composed of homodimers or heterodimers of Jun and Fos proteins. It has a vital role in cell growth and apoptosis [162]. All MAPK cascades can induce AP-1 activation in response to arsenic [84,[163][164][165]. Effects of arsenic on AP-1 are closely related to arsenic species, as well as time and concentration of exposure [163,166]. Trivalent methylated arsenicals were more potent than trivalent inorganic arsenic as inducers of c-Jun phosphorylation and AP-1 activation [163]. In cells transiently transfected with an AP-1-dependent promoter-reporter construct, methylated trivalent arsenicals methylarsine oxide (MAs III O) was more effective than iAs III in inducing the AP-1-dependent gene transcription [163]. Acute arsenic exposure increased AP-1 binding to DNA via c-Jun and c-Fos, whereas chronic exposure attenuated DNA-binding capacity of AP-1 [163,166,167]. c-Jun/AP-1 pathway-mediated cyclin D1 was indicated as one of the key events in cell malignant transformation caused by low-dose arsenic exposure [168].

p53 Pathway
p53 is a well-known tumor suppressor and plays an importance role in DNA repair, cell cycle, and apoptosis. Reports on the role of arsenic exposure on p53 are controversial, varying with arsenic species, exposure time, and cell types. In a human lyphoblastoid cell line, arsenic induced p53 expression by an ataxia telangiectasia mutated (ATM) (a member of PI3-kinase-related protein kinase)-dependent pathway, which phosphorylated p53 at serine 15 [169,170]. In haCaT cells, chronic arsenic exposure inactivated p53 via poly(ADP-ribosyl)ation [171]. however, haCaT cells are immortalized with SV40, which are known to interfere with p53 expression. In human telomerase reverse transcriptase (hTERT)-immortalized human keratinocytes, exposure to low-dose arsenite inhibited p53 expression by transcriptionally upregulating murine double minute 2 (MDM2) or ERK2-mediated overexpression of MDM2 [172]. In arsenic-exposed MCF-7 cells, an S-phase cell cycle arrest was found to depend on activation of p53 downstream cellular defense enzymes (i.e., sestrin 1 (SESN1) and activating transcription factor 3 (ATF3)) that was triggered by ROS generation in the early stage [93]. Recently, it has been shown that MDM2 is a downstream gene of Nrf2 and serves a link between Nrf2 and p53 in pancreas cancer [173]. Yet, involvement of Nrf2-MDM2-p53 pathway in arsenic-induced cancer is not clarified.

Stress Granules (SGs) Pathway
The overproduction of ROS induced by arsenic exposure causes tremendously harmful outcomes to cells, organs, and organisms [174]. To protect against arsenic toxicity, cells quickly activate antioxidant systems. Stress granules (SGs), the non-membranous cytosolic structures consisting of mRNAs and proteins, are formed and have antioxidant activity during arsenic exposure [175][176][177]. After the formation of SGs, the elevation of ROS was suppressed and ROS-dependent apoptosis was inhibited [175]. Beyond their function in defencing arsenic toxicity, SGs have been proposed to alter multiple signaling pathways, such as the JNK, Wnt, and mammalian target of rapamycin (mTOR) pathways, by intercepting and sequestering signaling components [178]. Accordingly, SG formation is a marker of chemoresistance and is upregulated by the production of a prostaglandin (15d-PGJ2), which is controlled by NRF2, in mutant v-Ki-ras 2 Kirsten rat sarcoma viral oncogene homolog (KRAS) cells [179]. The two antioxidant systems mentioned above, NRF2 and SGs, may intertwine in response to environmental stress, although the underling mechanism is not fully understood. Further understanding of the role of ROS along with NRF2 and SGs in arsenic toxicity is needed.

Metabolism Pathway
Metabolic reprograming is a feature of cancer cells, which usually show a strong dependence on aerobic glycolysis (the Warburg effect), increase in glutaminolysis, enhancement of macromolecule production and mitochondrial biogenesis, activation of the pentose phosphate pathway, and upregulation of amino acid and lipid metabolism [180]. Although metabolic reprograming has been implicated in carcinogenesis [181], few studies have been performed to investigate the role of ROS in arsenic-relevant cancer metabolism. The feature of metabolic reprograming in response to arsenic challenge is still not fully understood, albeit there is some evidence suggesting that arsenic induces overproduction of ROS and aerobic glycolysis. Chronic arsenite (75 ppb) exposure was shown to induce aerobic glycolysis while inhibiting mitochondrial oxidative phosphorylation in primary human cells and multiple cell lines (BEAS-2B, human prostate epithelial cell line (RWPE-1), human pulmonary epithelial carcinoma cell line (A549), primary human urothelial cells (HUC), and human dermal fibroblasts (HDF)) [182]. A similar phenomenon was also observed in Caenorhabditis elegans following 48 h arsenite exposure (50 to 500 µM) [183]. When human hepatocyte cells (HL-7702) were treated with different concentrations of arsenite (1 to 5 µM, 12 h), overproduction of ROS resulted from activated nicotine adenine disphosphonucleotide (NADPH) oxidase-mitochondria axis inactivated prolyl hydroxylases (PHDs), which led to protein accumulation of hIF-1α [184]. The latter is recognized as an inducer of aerobic glycolysis [182,185]. Reciprocal crosstalk between ROS and metabolism is vital to function and fate of cancer cells [186][187][188]. Thus, the mitochondria, a major source of ROS production and ATP synthesis, represent a potentially target for cancer therapy [188]. Some antioxidants (e.g., NADPH and GSH) and redox cofactors (e.g., Nicotinamide adenine diuncleotide hydrogen (NADH) and reduced Flavin adenine dinucleotide (FADH)) act as a bridge in redox regulation and metabolic reprograming [189][190][191]. ROS can consume reducing agents (NADPH and GSH) key for cell metabolism [192,193], and meanwhile activate Nrf2, AMPK, and hIF-1, which regulate metabolism and in turn fine tune ROS levels [194][195][196]. Therefore, the overproduction of ROS induced by arsenic is linked to metabolic reprogramming by direct or indirect ways. however, the exact order for the evolution of ROS and cell metabolism and their specific roles in arsenic carcinogenesis remain to be further investigated.

Potential Application of Antioxidants to Rescue Arsenic Toxicity
The first line of defense against acute arsenic toxicity is to reduce the amount of arsenic in the body [197]. The ideal arsenic-removal drug can interfere with the interactions of arsenic and molecules in the tissue. 2,3-dimercaptopropane-1-sulfonic acid (DMPS) and meso-2,3-dimercaptosuccinic acid (DMSA) are hydrophilic and belong to the mercapto family, which have vicinal dithiol moiety for the binding of metals [198,199]. These drugs or their analogs offer therapeutic benefit in acute arsenic poisoning when administered promptly [200]. Due to the limited ability of crossing the blood-brain barrier, loss of essential metals in the body, low cellular membrane penetration, and potential side effects in the kidney and liver, the use of metal chelators is limited [197,201].
Because it is well accepted that excessive generation of ROS plays an important role in the molecular mechanism of arsenic-induced toxicity and related diseases, application of antioxidants, especially extracts from plants, has been widely studied as therapeutics to counteract arsenic-induced toxicity. Some antioxidants involved in the methylation-mediated arsenic detoxification-excretion process, for example, GSH, can mitigate toxicity [202]. Some antioxidants decrease intracellular ROS levels via inhibiting mitochondrial respiratory chain complex I (e.g., metformin) [203]. Others present a protective role against arsenic-induced toxicity by regulating apoptosis-related molecular changes (e.g., diallyl trisulfide) [204,205]. Different types of antioxidants used for rescuing arsenic toxicity and their possible mechanisms are listed in Table 2. These antioxidants are classified as ROS scavengers, oxidative enzyme inhibitors, metal chelators, and antioxidant enzyme cofactors. The protective mechanisms of antioxidants extracted from the natural plants against arsenic-induced toxicity are shown in Figure 3. Besides reducing ROS, these antioxidants are involved in regulating signaling pathways such as Nrf2, NF-κB, MAPKs, transforming growth factor beta/Smad (TGF-β/Smad), and mammalian target of rapamycin/Akt (mTOR/Akt) [104,203,[206][207][208][209][210][211]. protein 2 (ARS2), which contributed to its nephron protective effects [209]. As a prospective remedial agent, all-trans retinoic acid (ATRA) (0.5 mg/kg) reversed arsenic-induced oxidative stress and apoptosis by inhibiting the MAPK signaling pathways and repressing p53-dependent apoptosis in the rat uterus [206]. When hepatic cells were exposed to arsenic, p38 and JNK signaling pathways were activated. Carnosic acid (CA), which was commonly found in Rosmarinus officinalis and Salvia officinalis, significantly attenuated arsenic-induced phosphorylation of p38 and JNK [235]. EGCG, (-)-Epigallocatechin-3-gallate; GSE, grape seed extract; TGFβ, transforming growth factor-β; NOX, Nicotinamide adenine dinucleotide phosphate oxidase; SMADs, drosophila mothers against decapentaplegic protein.
Generally speaking, accumulating evidence suggests that antioxidants have the potential to alleviate arsenic toxicity through reducing ROS generation, enhancing antioxidant capacity, Interestingly, different natural compounds used as antioxidants exert varied roles in Nrf2 activation in reducing arsenic toxicity. For example, tetramethylpyrazine (TMP) (50 µM or 100 µM) protected against arsenic-induced nephron toxicity by inhibiting Nrf2 activation, and accordingly reducing Heme oxygenase-1 (HO-1) expression [209]. Pomegranate fruit extract (PFE) (0.2 mL of 0.2% of extract) reduced ROS generation in hepatocytes, thereby reducing arsenic-induced Nrf2 activation [207]. however, most antioxidants showed an effect on promoting Nrf2 activation in response to arsenic [26,70,104,211]. Dietary supplementation with SF (80 mg/kg BW) protected against arsenic-induced nephrotoxicity via the Phosphoinositide 3-kinase (PI3K)/Akt-mediated Nrf2 signaling pathway in the rat kidney [212]. Grape seed proanthocyanidin extract (GSPE) (10, 25, and 50 mg/L) activated Nrf2 signaling pathway to antagonize arsenic-induced oxidative damage, promoted arsenic methylation metabolism, and relieved arsenic-induced hepatotoxicity [26]. Alleviated arsenic toxicity in the liver and reproductive system by eriodictyol and lutein was via activating Nrf2 signaling pathway [213,214]. Co-treatment of antioxidant vitamins L-ascorbic acid and α-tocopherol attenuated toxicity induced by arsenic trioxide in h9c2 cardiomyocytes through activation of Nrf2 and Bcl2 transcription factors [215]. In contrast, PFE (0.2 mL of 0.2% of extract) reversed arsenic-induced hepatotoxicity with reduction in arsenic-induced Nrf2 activation [207], suggesting that other Nrf2-independent mechanisms are involved in attenuation of arsenic toxicity by this antioxidant.
Some antioxidants attenuated arsenic-induced toxicity via MAPK activation. TMP inhibited arsenic-induced activation of MAPK family, and further reduced the expression of arsenic response protein 2 (ARS2), which contributed to its nephron protective effects [209]. As a prospective remedial agent, all-trans retinoic acid (ATRA) (0.5 mg/kg) reversed arsenic-induced oxidative stress and apoptosis by inhibiting the MAPK signaling pathways and repressing p53-dependent apoptosis in the rat uterus [206]. When hepatic cells were exposed to arsenic, p38 and JNK signaling pathways were activated. Carnosic acid (CA), which was commonly found in Rosmarinus officinalis and Salvia officinalis, significantly attenuated arsenic-induced phosphorylation of p38 and JNK [235].
Generally speaking, accumulating evidence suggests that antioxidants have the potential to alleviate arsenic toxicity through reducing ROS generation, enhancing antioxidant capacity, regulating ROS-related signaling pathways, and keeping the balance of inflammation and immunomodulation. Natural antioxidants extracted from plants are more promising due to their rich sources, diversity, and few side effects. however, many studies on the role of plant extracts have not been systematically conducted. In vitro assessment results do not provide exact therapeutic implications because effectiveness plant extracts may be influenced by several physiopharmacological processes, such as absorption, distribution, metabolism, storage, and excretion, as well as bioavailability, and presence of co-antioxidants and ions [237]. hence, a commonly suggested strategy, in which in vitro mechanic analysis goes after in vivo effect evaluation [238], is proposed to assess therapeutic potential of plant extracts in arsenic toxicity. In addition, there is a need to first identify the major target of arsenic-induced pathophysiological alterations. Many plant extracts have multiple functions in addition to severing as an antioxidant. Finally, plant extracts should be screened for low molecular antioxidants that are able to cross blood-brain barriers and reach other target organs. In addition to the above, more and more physiologically-based pharmacokinetic (PBPK) models have been used to predict the pharmacokinetic behavior of drugs in humans on the basis of preclinical data [239,240]. PBPK modeling, a good tool for evaluating and optimizing clinical trial design, provides an approach that enables the plasma concentration-time profiles to be predicted from in vitro and in vivo data [239,241,242], which can be used for choosing the dose of antioxidants.

Conclusions
An effectively preventive and therapeutic strategy for arsenic poisoning is still a challenge worldwide due to the incomplete understanding of underlying mechanisms. ROS alteration is a widely accepted key event in arsenic toxicity. Earlier studies with acute exposure and relatively high dose arsenic suggest that excessive levels of ROS disturb cellular signaling pathways, as well as damage macromolecules. The effects of antioxidants on arsenic toxicity are also assessed in a scenario of relatively high dose of arsenic. however, the mechanism and function of ROS may be different in the realistic environment, in which a relatively low-dose and long-term exposure to arsenic affects millions of people. In addition, in vitro assay concentrations of arsenic may misrepresent potential in vivo toxic effects and do not provide dose-response data that can be used for a risk assessment [243,244]; hence, the in vitro-to-in vivo extrapolation (IVIVE) by physiologically based toxicokinetic (PBTK) modeling was used to serve in vitro-in silico-based risk assessment [245]. It is of great importance to apply physiologically relevant doses during toxicological research. The role of ROS in arsenic toxicity should be fully clarified before the application of antioxidants. Of all the antioxidants investigated in arsenic toxicity, natural antioxidants extracted from plants are promising due to their rich sources, diversity, and few side effects, especially those with low molecular weight. More effective therapeutic value from plant extracts is expected on the basis of arsenic-induced pathophysiology targeting combined with in vitro and in vivo assessment.
Author Contributions: Y.H., J.L., and B.L. designed the outline of the article and wrote it. R.W., G.W., and C.L. designed the outline of the article initial draft, and revised and expanded the manuscript. h.W., J.P., and Y.X. helped with intellectual contributions. All authors have read and agreed to the published version of the manuscript.