It is difficult to use the epidemiological literature to identify mechanisms, as long-term exposures to arsenic are likely compounded with exposures to pollution, poor diet, and low SES. Basic science research is poised to control for these confounding factors, including extent and timing of exposure. This research has determined that arsenic imparts its toxicity on the body via a number of mechanisms. These include the depletion of methyl groups affecting epigenetic profiles, the uncoupling of oxidative phosphorylation and increased reactive oxygen species, the inhibition of thiol-containing enzymes and proteins (including the depletion of glutathione), altered signal transduction and cell proliferation, and reduced DNA repair inducing genotoxicity [1, 6]. As these mechanisms have been discussed in detail in other reviews, we cover mechanisms particularly related to the brain. These include hippocampal dysfunction; glutamatergic, glucocorticoid, cholinergic, and monoaminergic signaling; pathways associated with Alzheimer’s disease; and synaptic plasticity, particularly neurogenesis. To begin, we briefly provide results from the few studies focusing on neural epigenetic patterns altered after arsenic exposure.
Arsenic accumulation and subsequent toxicity is likely mediated through multiple mechanisms of action. Of particular interest over the past few years is the impact of arsenic on the epigenome. While there have been several studies determining the impact of arsenic on epigenetic regulation in cancer cells, the liver, and other parts of the body, very few have focused the brain. Exposure to 3 and 36 ppm arsenic throughout gestation increased DNA methylation on two genes involved in neural plasticity in rat cortex and hippocampus at one month of age [54•]. Hypomethylation of these genes in both regions was observed after four months of cumulative exposure to arsenic; these animals also displayed deficits in fear memory as well, although the link between hypomethylation and memory was correlational. Since arsenic metabolism requires methyl groups derived from S-adenosylmethionine (SAM) for excretion, it is plausible that arsenic depletes SAM leading to alterations in DNA methylation. In vitro studies corroborate this assertion: 25 μM arsenic exposure for 24 hours depleted SAM concentrations, increased global DNA hypomethylation, and repressed Dnmt1 and Dnmt3a expression . Epidemiological work has shown that DNA methylation is affected by arsenic in human populations as well, including umbilical cord blood, but discussion of those studies is beyond the scope of this review [56–59]. We can determine that exposure to arsenic does induce epigenetic modifications to the DNA, which may result in aberrant gene expression even in the brain; however the link between DNA methylation on particular genes and cognitive deficits has yet to be elucidated.
Other epidemiological studies have determined the impact of arsenic ingestion on histone modifications. Arsenic compounds have been shown to alter gene expression and posttranslational modifications (PTM) of histones in vivo; interestingly, researchers found a differential effect of arsenic on global histone modifications among males and females . In vitro assessment of low doses of arsenic on histone modifications has also been performed; however, detailed discussion of this research is beyond the scope of this review [60–62]. Rodent studies with prenatal exposure to 100 μg/L arsenic demonstrated reduced global acetylation on lysine 9 of histone 3 (H3K9ac) in the cortex and hippocampus of postnatal day (PND1) pups, which was correlated to altered learning in adulthood . Exposure to 3 and 36 ppm arsenic throughout development up to four months reduced myelination (for which methylation is required) and dimethylation of arginine residues on histones . This could result from the altered expression and function of epigenetic modifiers or transcription factors, as arsenic impacts zinc-finger protein expression and function .
These studies demonstrate that arsenic impacts DNA methylation and histone modifications and alters the enzymes responsible for regulating these modifications. The effects of arsenic on the epigenome are related to the dose and extent of arsenic exposure as seen before, but also type of histone methylation mark, gene, and sex. As such, in determining mechanisms of arsenic toxicity it will be important to control for these variables both in epidemiological and molecular studies in the future. Additionally, cumulative low-level exposure to arsenic likely occurs over generations. The literature suggests that females are differentially more affected by arsenic than males. Since females are the key source of transgenerational effects (3rd generation ova are exposed to the in utero environment), arsenic could be impacting transgenerational epigenetics, including the imprinting of genes (from both males and females). This area of research could provide an insight into the effects of arsenic on the brain and body, yet there are no published reports on arsenic exposure and transgenerational epigenetic mechanisms to date.
Studies on the mechanisms of arsenic-induced toxicity have established that arsenic alters learning and memory in behavioral assays and impacts multiple neurobiological processes including those of neurogenesis and cholinergic, glutamatergic, and monoaminergic signaling pathways. Recent work using animal models has revealed potent alterations in hippocampal function, morphology, and signaling leading to altered cognitive behavior after arsenic exposure. While the exposure paradigms and concentrations of arsenic have been highly varied, the overall conclusions have been congruent between studies (see Table 3).
In particular, behavioral studies using hippocampal-dependent tasks, including the Morris Water Maze and fear conditioning, have corroborated epidemiological evidence of reduced cognition observed in humans with arsenic exposure. Table 3 details the behavioral paradigms and arsenic exposures used in these studies. Results from this body of literature suggest that regardless of dose, timing, or extent of exposure, arsenic induces hippocampal-dependent behavioral deficits in rodent models, suggesting impaired spatial, working, long-term, and short-term memory [63, 66–68]. Interestingly, the effect extends to nonmammalian species as arsenic exposure in Danio rerio (zebrafish) induces deficits in long-term memory as well . Several of these behavioral deficits can be ameliorated with treatment; for example, sodium butyrate, a histone deacetylase inhibitor, attenuated deficits seen in the radial arm maze after two weeks of arsenic exposure of 100 μg/L during adulthood . More detail on therapeutic interventions after arsenic exposure will be discussed in later sections of this review. Overall, work with animal models has demonstrated that arsenic induces deficits in multiple learning paradigms, particularly those relying on proper hippocampal function. Yet, identifying the hippocampus as a sensitive area for arsenic’s effects does not necessarily identify a molecular target. Alterations in multiple signaling pathways are localized to the hippocampal formation, but it should be noted that mossy fiber terminals in the hippocampus contain substantial amounts of zinc. Thus, the sensitivity of the hippocampus may be due to the effect of arsenic on zinc either via displacement or substitution. To date, no studies investigating this mechanism have been published.
Arsenic impacts the synaptic activity of neurons localized to the hippocampus. Slices obtained from young and adult rats exposed to 100 μM arsenite in vitro had reduced amplitudes of excitatory post synaptic potentials (EPSPs) in the Schaffer collateral/CA1 synapses. Exposure inhibited long-term potentiation (LTP), a form of synaptic plasticity, in hippocampal slices from adult but not young rats; however, this effect was reversible after 20 minutes of washout. Arsenic exposure did not impair paired-pulse facilitation, indicative of presynaptic activity, suggesting that acutely applied arsenic does not affect presynaptic neurotransmitter release . However, components of NMDA receptors, specifically NR2A, were reduced after three months of arsenic exposure (2.72, 13.6, 68 mg/L arsenic) in mice. Additionally, these mice exhibited morphological changes in hippocampal neurons—reduced size with a condensed nucleus and cytoplasm—along with capillary edema and irregular vascular endothelial cell morphology . Similar morphological changes in hippocampal neurons were observed after a 3-month exposure of 8.2 mg/kg/day arsenic; both the striatum and hippocampus contained abnormally myelinated nerve fibers, while the hippocampus contained reduced mossy fiber terminals . Brief, two week exposure to 100 μM arsenic also altered expression levels of mRNA for synapse related genes, including increased Grin1, Syn2, and Stx6 expression, similar to results from our studies [63, 72].
In addition to altered synaptic activity and synapse-related gene expression, arsenic has been shown to impart alterations in central pathways involved in mediating learning and memory in the hippocampus. Using a three-month exposure model (from weaning until four months), arsenic-exposed animals had decreased NR2A expression, PSD-95, and p-CAMKIIα in the hippocampus with concurrent increased SynGAP expression, a known negative regulator of the Ras-MAPK pathway . Reduced p-ERK1/2 in the hippocampus was also observed in the arsenic-exposed animals. These findings corroborate our own work demonstrating reduced ERK2 in the hippocampus after perinatal exposure to arsenic . Thus, the Ras-MAPK/ERK pathway appears to be sensitive to arsenic damage; interestingly, zinc has been shown to alter this pathway as well . It is possible that arsenic is acting in a similar fashion. Overall, altered Ras-MAPK/ERK signaling, LTP, and synaptic regulation in the hippocampus could underlie behavioral deficits suggesting arsenic-induced alterations in long-term episodic memory, associative learning, spatial learning, and working memory.
In addition to altered hippocampal-dependent behaviors, depressive-like symptoms have been observed in arsenic-exposed mice. We have demonstrated increased immobility in the forced swim task and increased latency for escape in the learned helplessness task, along with increased plasma corticosterone levels . Corticosterone (CORT), the rodent equivalent of cortisol, is a stress hormone that plays a role in mediating the effects of the hypothalamus-pituitary-adrenal (HPA) axis in response to “stressful” events ranging from fear to learning. In humans, significant alterations in the HPA axis have been connected to depressive-like symptoms, and patients with depression typically report stress as a major factor in the onset of their depression. Thus, decreased behavioral ability and increased depressive-symptoms in arsenic-exposed animals correlate with the epidemiological data on reduced cognition in humans; as such, arsenic may be mediating not only cognitive impairments but also mood disorders in humans via the glucocorticoid signaling pathway.
CORT signaling is mediated through the corticosterone receptors, specifically the mineralocorticoid (MR) and glucocorticoid (GR) receptors. When activated, MR and GR translocate to the nucleus to allow for binding to response elements (MRE or GRE) on genes for transcriptional activation or repression. CORT signaling, via its receptors in the hippocampus, is responsible for imparting an inhibitory tone for the HPA axis. Our work has demonstrated nuclear levels of GR are much lower in adult mice perinatally exposed to arsenic than levels observed in controls in hippocampal tissue . Additionally, we have shown decreased GR binding to and expression of H-Ras and Raf-1, genes involved in modulating the MAPK pathway with GRE binding sites . In vitro studies have demonstrated that various levels of arsenic impact GR-mediated transcription in a bidirectional manner: high arsenic levels (1-3 μM) induce an inhibitory effect, while low levels (0.05–1 μM) seem to produce a stimulatory effect . Arsenic’s impact on GR is predicated on the DNA binding domain (DBD) within the GR. Other steroid receptors, including the MR and the progesterone receptor respond to arsenic in a similar biphasic manner as the GR . Interestingly, arsenic’s effects on transcriptional regulation of the estrogen receptor seem to be only inhibitory . A recent study predicted that the GR pathway was a common mediator of metal-induced birth defects: indeed, arsenic-induced deficits, specifically neurodevelopmental toxicity, were prevented by inhibition of GR signaling in a chick embryo model . Oscillatory signaling of the HPA axis, in addition to proper GR localization for MAPK activation, is paramount for proper learning and memory. This signaling may be impaired in arsenic-exposed animals: while they have elevated circulating CORT, their ability to initiate a proper HPA response is blunted after a stressor [81•]. Thus, the susceptibility of the GR to arsenic may play an important role in hippocampal-related deficits, including reduced learning and increased depressive-like symptoms (observed in rodent models) and may underlie mood and cognitive deficits seen in human studies as well.
Motor learning, cholinergic signaling, and locomotion are all affected by arsenic exposure in rodent models. Early studies demonstrated impaired motor coordination and delayed spontaneous alteration in rats chronically consuming arsenic (36 mg/L) for four months . However the increased locomotion reported in this study has been challenged by more recent work: while altered locomotion is a common behavior seen in arsenic-exposed mice, whether this behavior is hypo- or hyperlocomotion seems to be dependent on sex and the arsenic exposure paradigm (see Table 3 for studies). Low levels of arsenic exposure seem to induce hyperactivity in male mice, while high levels induce hypoactivity [83, 84]. Conversely, female mice consuming any level of arsenic display hyperlocomotion . Subchronic exposures to arsenic increase locomotion but impair motor coordination in both sexes , but reports on in utero exposure are conflicting, with some suggesting impaired neural reflexes and others reporting no changes in locomotor behavior [86, 87]. A similar dose relationship between locomotion and arsenic level occurs in the zebrafish model, where moderate arsenic levels reduced locomotion (line crossing) and high levels increased the distance travelled . Details on these studies, including dose, timing of exposure, and age of behavioral assessment are provided in Table 3. While parsing out verifiable locomotor actions from anxiety is difficult and likely to impact these studies, we can ascertain that high concentrations of arsenic induce hypolocomotion, moderate levels of arsenic induce hyperlocomotion, and low concentrations may induce no change in locomotion.
Altered motor coordination and locomotion could arise from abberant cholinergic functioning. Several reports have noted reduced acetylcholinesterase (AchE) activity and choline acetyltransferase (ChAT) functioning after arsenic exposure. Female rats exposed to 20 mg/kg arsenic for 28 days displayed deficits in the transfer latency of the passive avoidance response and decreased labeling of muscarinic cholinergic receptors in the hippocampus and frontal cortex . These brain regions had reduced AchE activity and ChAT labeling after arsenic exposure as well. Interestingly, treatment with a bioactive component of the spice curcumin during arsenic exposure attenuated these observed effects in female rats . Exposure to less arsenic (5 mg/kg body weight) also inhibited AchE activity in the brain and was associated with poorer performance in operant learning ; another study demonstrated that AchE activity decreased with increasing arsenic concentrations in male rats after five days of exposure . Our analyses have shown that mRNA of AchE is increased in the adult dentate gyrus after developmental arsenic expoure to 50 μg/L arsenic, suggesting a compensatory mechanism for altered AchE activity . These studies provide support for arsenic in the etiology of Alzheimer’s disease, as more reports confirm reduced AchE and ChAT along with degeneration of cholinergic neurons in AD pathology .
Global cognitive impairments assessed after arsenic exposure have been associated with Alzheimer’s disease (AD) in the human population; for an in-depth assessment of the relationship between arsenic and Alzheimer’s disease, see . For the purposes of this review, we will discuss arsenic exposure in the context of experimental studies. In vitro research using a cholinergic neuronal cell line has demonstrated that sodium arsenite and dimethylarsinic acid (DMA) have different effects on APP protein levels, β-amyloid formation, and altered activity of AchE and ChAT . These findings corroborate our own: we have found altered mRNA expression of genes associated with AD, including Appb1 and ApoE, and Ache in the dentate gyrus of adult male mice after developmental exposure to 50 μg/L arsenic .
For hallmark pathologies associated with AD, cell culture studies have demonstrated arsenic exposure increases β-amyloid protein and induces hyper-phosphorylation of tau protein, oxidative stress, inflammation, endothelial cell dysfunction, and angiogenesis . Interestingly, zinc has also been shown to induce tau phosphorylation through an ERK-sensitive mechanism ; again, this effect may be due to a cation sensitive system (to zinc or arsenic) involving the Ras-MAPK/ERK pathway. In 2012, Piacentini and colleagues provided further evidence supporting the role of arsenic in the etiology of AD. The authors found a significant correlation between a genetic polymorphism of the protein glutathione S-transferase GSTO1-1 in a population of 120 AD patients compared to 114 healthy controls (OR = 3.70); this transferase has been linked to both Alzheimer’s and Parkinson’s disease and is involved in arsenic metabolism, even in invertebrate systems [97–99]. While arsenic does not participate directly in REDOX reactions, it can induce oxidative stress via depletion of glutathione and impair REDOX reactions by inhibiting enzymes with sulfhydryl groups ; and oxidative damage is strongly associated with AD. In rodent models, arsenic exposure results in vascular endothelial dysfunction, impairment of learning and memory, altered nitrogen levels, and oxidative stress; all of which have been associated with symptoms of dementia . Treatment with a histone deacetylase inhibitor, sodium butyrate, attenuated these cognitive and vascular deficits observed in arsenic-exposed animals . Sodium butyrate has been used in the treatment of depression and improves cognitive skills in the Alzheimer’s mouse model (APP/PS1-21) . Overall, we confidently state that arsenic imparts cognitive deficits associated with AD in rodent studies of exposure; however, very few epidemiological studies have investigated this relationship to date.
Arsenic affects many transporter systems including the monoamines, dopamine (DA), serotonin (5-HT), and norepinephrine (NE). We have demonstrated that very low doses of arsenic (50 μg/L) during development increase 5-HT1A receptors in the dorsal hippocampus in adult offspring . Exposure to moderate levels of arsenic (1, 2, and 4 mg/L) in water for 60 days reduced levels of NE, DA, and 5-HT in both the cerebrum and cerebellum of 7-week old mice in a dose-dependent manner. mRNA levels of monoamine synthetases (including dopamine β-hydroxylase, tyrosine hydroxylase (TH), and tryptophan hydroxylase) were also all reduced after exposure to 4 mg/L arsenic . Evidence from other studies using low to moderate doses of arsenic suggest differential effects on monoaminergic signaling based on sex, as observed in locomotion tasks. Four months of chronic exposure to 0.05, 0.5, 5.0, or 50 mg/L arsenic in drinking water decreased levels of DA in the striatum and hypothalamus in females (who also exhibited increased locomotion) but not in males . TH and thioredoxin (Trx-1 A) levels were reduced in the striatum of males but not females; curiously, the opposite effect was observed in the nucleus accumbens where TH and Trx-1A were reduced in females but not males . Conversely, in a different study on rats ingesting 50 mg/L arsenic in water, DA content in the striatum was increased in males, although no changes in DA, its metabolites, or serotonin were found in the prefrontal cortex or the nucleus accumbens . This group also reported alterations in mRNA of several antioxidant genes including superoxide dismutase (SOD), Trx-1, and NE and DA receptor genes in arsenic-exposed animals dependent on dose and region of interest in the brain (nucleus accumbens, prefrontal cortex, or striatum). Upregulation of antioxidant factors (Trx-1 and SOD) indicate compensatory mechanisms to overcome the oxidative byproduct effects of arsenic toxicity on thiol containing enzymes related to arsenic metabolism . It is difficult to propose a mechanism of action for opposite effects based on sex for different brain regions. However, evidence from these studies, while conflicting, suggest that arsenic may induce a complex dose-response relationship, which may not be linear, on locomotion and the nigrostriatal dopaminergic system as seen in other studies.
High levels of arsenic exposure, which have been reviewed elsewhere, produce more definitive deficits on dopaminergic and serotonergic signaling in the corpus striatum, hippocampus, and frontal cortex [85, 103]. Interestingly, treatment with curcumin in these studies attenuates deficits in monoamines and increases nitric oxide (NO) and TH expression, in both sexes. More details on therapeutic strategies to combat arsenic toxicity will be provided later.
Deficits in hippocampal-related behavioral tasks and increased depressive-like symptoms suggest that arsenic exposure induces changes in hippocampal morphology. Recent studies have demonstrated deficits in adult neurogenesis in the dentate gyrus of the hippocampus in both developmental and adult exposures to arsenic. Treatment for four months with 4 μg/L arsenic in drinking water reduced proliferation of neural progenitor cells and the number of mature neurons [104•]. Developmental exposure of 50 μg/L arsenic (in utero and postnatal) altered differentiation but not proliferation of neural progenitor cells in the adult hippocampus at PND63 . In both studies, deficits in adult neurogenesis were ameliorated either after cessation of the use of arsenic in water [104•] or with experience in an enriched environment . In vitro studies using P19 pluripotent cells cultured with varying concentrations of arsenic (7.5-75.0 μg/L) demonstrated that arsenic inhibited the formation of muscle and neuronal cells during P19 cell differentiation in a dose dependent manner. Deficits in neuronal differentiation may have resulted from reduced expression of transcription factors including neurogenin 1, neurogenin 2, and NeuroD as compared to control P19 cultures. Further, Nanog expression increased during cell differentiation, suggesting that arsenic impacts differentiation but not proliferation .
Altered hippocampal morphology in CA1, CA3, and the dentate gyrus was also observed during postnatal development after I.P. injection of 1, 1.5, and 2.0 mg/kg arsenic from PND4-11 . Two weeks of 100 μM exposure to arsenic reduced GFAP staining in the hippocampus in another study, indicative of less neural stem cells , while exposure to arsenic induced deficits in cell proliferation and increased apoptosis in the brains of zebrafish . These studies suggest many different types of exposures to arsenic, including brief, chronic, adult, or developmental, can impact the morphology of the hippocampus. The effect of arsenic on neurogenesis may be mediated through a signaling mechanism already discussed: altered glucocorticoid and HPA axis function. Neurogenesis is a particularly sensitive process in the brain, and as such, is prone to insults like stress activity. Recent in vitro studies done in human hippocampal progenitor cells have demonstrated biphasic responses of adult neurogenesis to cortisol; of particular interest is that increased cortisol levels inhibit proliferation and differentiation possibly mediated by the GR . However, the effect of the GR on adult neurogenesis is complex, as proposed by work showing antidepressants mediate an increase in adult hippocampal neurogenesis by activating the GR  as well. Arsenic has an impact on adult neurogenesis and hippocampal-dependent learning and memory and induces cognitive deficits and depressive-like symptoms; altered HPA axis regulation, and particularly GR signaling in the hippocampus could underlie all of these alterations seen in arsenic-exposed animals.
Arsenic can alter a multitude of systems in the brain. Of particular interest is HPA axis dysregulation that may underlie several behavioral deficits, particularly related to the hippocampus including alterations in adult neurogensis and Ras-MAPK/ERK signaling. Additionally, arsenic seems to have an impact on cholinergic and monoaminergic signaling, though the mechanisms are not well understood at this point. Rodent studies have provided useful corroboration of the epidemiological evidence suggesting that a number of mechanisms could underlie cognitive deficits and mood disorders observed in human populations. More research focused on the dynamics of epigenetics, particularly on mechanisms of learning and memory and mood, will be important for understanding the impact of arsenic on the brain. While it is unlikely one common unifying mechanism for arsenic’s effects will be identifed, some clues on how naturally present cations, like zinc, interact with the system, may provide insight.