Human-induced pluripotent stems cells as a model to dissect the selective neurotoxicity of methylmercury

Abstract

Methylmercury (MeHg) is a potent neurotoxicant affecting both the developing and mature central nervous system (CNS) with apparent indiscriminate disruption of multiple homeostatic pathways. However, genetic and environmental modifiers contribute significant variability to neurotoxicity associated with human exposures. MeHg displays developmental stage and neural lineage selective neurotoxicity. To identify mechanistic-based neuroprotective strategies to mitigate human MeHg exposure risk, it will be critical to improve our understanding of the basis of MeHg neurotoxicity and of this selective neurotoxicity. Here, we propose that human-based pluripotent stem cell cellular approaches may enable mechanistic insight into genetic pathways that modify sensitivity of specific neural lineages to MeHg-induced neurotoxicity. Such studies are crucial for the development of novel disease modifying strategies impinging on MeHg exposure vulnerability.

1. Introduction

Mercury (Hg) is a global pollutant with profound neurological toxicity. Even the most stringent control measures will not eliminate exposure given its ubiquitous presence. Hg can be found in three chemical forms in the environment: elemental Hg vapor, inorganic Hg salts, and organic Hg [1]. The distribution, toxicity, and metabolism of Hg vary dependent on its chemical form. Methylmercury (MeHg) is one of the most extensively studied forms of Hg due to its relatively high accumulation in the brain [2, 3]. Exposure to Hg occurs primarily via the food chain in the form of MeHg, which accumulates in fish. Notably, fish consumption-safety is an important issue, considering as many as 3 billion people worldwide depend on fish as a daily dietary source of protein [4, 5]. Exposure to MeHg remains a major health concern, especially for the developing CNS. The effects on intellectual function and other neurobehavioral outcomes in children from prenatal MeHg exposure via maternal fish consumption are the subject of two major prospective longitudinal studies in the Seychelles [6–11] and the Faroe Islands [12–15]. A National Academy of Sciences expert panel reviewed these studies and concluded that the weight-of-evidence supports the adverse health effects MeHg [16]. The same panel concluded that Hg levels should not exceed 5.0 µg/L in whole blood or 1.0 µg/g in hair. These levels correspond to a reference-dose (RfD) of 0.1 µg/kg body-weight/day. In another study, referred to as The EPA’s Hg Study Report to Congress it was noted that 8% of US women of childbearing age have blood Hg concentrations in excess of the above RfD [17]. Translating these observations into real life, the implication is that approximately 300,000 newborns may be at increased risk of neurodevelopmental deficits, including learning disabilities, associated with in utero MeHg exposure in the United States alone.

Importantly, there have been conflicting results in the epidemiological literature as to the association between MeHg exposure and neurodevelopmental outcomes. Of note, the longitudinal studies conducted in the Seychelles Islands have found little evidence of such association [6–11], while the studies conducted in the Faroe Islands have associated MeHg with neurobehavioral deficits including motor function, learning, attention, and memory [12, 15, 18]. These conflicting results may be due to differences in developmental stage of exposure, genetics, and environmental modifiers, which influence background disease occurrence and impose differential sensitivity to MeHg [19–21]. Although epidemiological studies can give insights into the genetic and environmental modifiers of MeHg toxicity, many factors such as multiple exposures and timing of exposure are difficult to reconstruct and may influence the results. Therefore, a controlled model using relevant levels of MeHg exposure is necessary and crucial for defining the modifiers which lead to differential sensitivity.

Furthermore, MeHg toxicity has shown a specificity not only on the population level, but also on the cellular level, and the reasons for this specificity are largely unknown. The developing nervous system has been shown to be a primary target of MeHg, with a particular sensitivity in particular neurons in cerebellar and cerebral cortexes observed in cases of both adult and developmental MeHg exposure [22–26]. Additionally, the mechanisms of MeHg toxicity appear to be global in nature. Namely, MeHg neurotoxicity is associated with oxidative stress, mitochondrial dysfunction, activation of cell stress signaling pathways, disruption of microtubule assembly, disruption of Ca signaling, and binding to and sulfhydryl groups [2, 27–29]. Given that these targets of MeHg are ubiquitous, the reasons why specific tissues and cells show a sensitivity to MeHg remains an interesting and important question. We propose that new studies are needed to assess these mechanisms during in vitro neuronal differentiation of human induced pluripotent stem cells (hiPSCs), to allow comparative assessment of the genetic factors that influence sensitivity to chronic MeHg exposure across the human population.

2. Genetic and Environmental Modifiers of MeHg Susceptibility.

Polymorphisms in CYP3A and glutathione pathway genes may modify associations of mercury exposure and neurodevelopmental outcomes [30, 31]. Specifically, polymorphisms in both the modifier and catalytic subunits of glutathione-cysteine ligase (GCLM and GCLC, respectively), as well as in the glutathione s-transferases GSTP1 and GSTM1, have been found to modulate MeHg retention [32–35]. Furthermore, early exposures may not unmask themselves clinically for decades [36]. Thus, genetic modifiers of vulnerability to Hg may exert their influence during development and in the adult. Other genetic and environmental modifiers likely contribute to significant variation in neurodevelopmental outcomes associated with human exposures (e.g. the Seychelles study versus Faroe Islands and New Zealand studies) [6–15].

Environmental modifiers of MeHg susceptibility include diet; such as consumption of polyunsaturated fatty acids (PUFAs), vitamin E, and selenium; as well as co-exposures to other toxicants, primarily PCBs [20, 21]. These modifiers have been suggested to be a cause of the conflicting results seen in epidemiological studies examining the association between developmental MeHg exposure and neurodevelopmental deficits [19, 37]. For example, polyunsaturated fatty acids (PUFAs) are known to be a beneficial component of fish that are crucial for neural development. Some epidemiological studies have suggested that PUFA consumption may be protective against the effects of MeHg on neurodevelopmental outcomes [38, 39]. On the other hand, some marine species, particularly the pilot whale consumed more frequently in the Faroe Islands, contain higher levels of PCBs, which are thought to interact with MeHg [40]. Considering the array of dietary factors and co-exposures that may modify MeHg toxicity, epidemiological studies are limited in their capacity to identify genetic traits that lend tolerance or susceptibility to MeHg.

3. Mechanisms of MeHg-induced neurotoxicity.

Disruption of redox status and oxidative stress are hallmarks of MeHg poisoning [41–50]. A major source of MeHg-induced ROS is the mitochondrial electron transport chain (ETC) [51, 52]. MeHg impairs mitochondrial and cellular energetics by decreasing mitochondrial state-3 and increasing state-4 respiration, inhibiting glycolysis and tricarboxylic acid (TCA) cycle activity, and decreasing ATP utilization [51, 53]. MeHg-induced lipid peroxidation occurs by stimulation of mitochondrial ubiquinol:cytochrome c oxidoreductase complex [52] and inhibition of GPx [54–56]. Blockage of the mitochondrial transition pore by cyclosporine-A in brain synaptosomes lowers MeHg-induced ROS production [57]. Metabolically active cell types, particularly tonically active neurons such as those in the substantia nigra pars compacta (SNpc) require high ATP levels to optimally function and survive [58–61], which may render these brain areas especially susceptible to MeHg-induced oxidative and energetic stress.

Nuclear factor (erythroid-derived 2)-like 2 (Nrf2) (skn-1 homologue in C. elegans) and antioxidant response element (ARE)-driven genes combat oxidative stress. Nrf2 is a leucine zipper transcription factor that induces genes encoding antioxidant proteins, including those in the glutathione (GSH) family, in conditions of oxidative stress [62, 63]. Activated Nrf2 binds to the cis-acting ARE sequence in the regulatory regions of target genes encoding detoxifying and antioxidant enzymes/proteins, such as γ-glutamylcysteine ligase (GCL), glutathione (GSH) transferase (GST) A1 and A2, GSH peroxidase (GPx), heme oxygenase (HO), NAD(P)H:quinone reductase and the GLU-Cysteine (Cys) exchanger, (Xc^-) [64]. Upregulation of Nrf2 and its effector genes is a hallmark of MeHg poisoning [64–69]. In support of this, oxidative stress and mitochondrial dysfunctional associate with MeHg exposure has further been linked to genomic instability, DNA damage as well as glutathione metabolism [70, 71].

MeHg’s high affinity for sulfur and sulfhydryl groups (-SH) represents a major chemical mechanism for its toxicity. MeHg binds Cys-containing proteins via interaction with -SH [72]. Known cellular and molecular outcomes of MeHg neurotoxicity include (i) disrupted expression of neural differentiation markers, (ii) mitochondrial dysfunction, (iii) activation of cellular stress response pathways, especially those linked to redox status, (iv) disruption of calcium and potassium channels, and (v) oxidative stress [27, 73–80]. GSH is a major thiol-containing binding target of MeHg [81, 82]. The GSH MeHg interaction attenuates redox related toxicity. Decreased GSH levels parallel MeHg-induced increases in oxidative stress, and upregulation of GSH is neuroprotective [50, 55, 66, 83–85]. Thus, buthionine sulphoximine (BSO), an inhibitor of γ-glutamylcysteine synthetase, depletes GSH levels and increases sensitivity to MeHg toxicity [50, 86–89]. GSH-related genes contribute to human genetic variation of MeHg metabolism [32, 33, 90]. Therefore, differential cellular expression of GSH could also convey selective cellular sensitivity to MeHg.

4. Sensitivity of the developing nervous system to MeHg

The developing mammalian nervous system exhibits hypersensitivity to MeHg exposure relative to adult exposures [91, 92]. This sensitivity was seen in the accidental exposures at Minamata Bay, Japan and Iraq, where pregnant mothers and their fetuses were exposed to MeHg. Mothers presenting with mild symptoms such as paresthesia and muscle weakness; whereas their children presented with a cerebral palsy like disorder and severe neurological symptoms including mental retardation, dysarthria, and chorea [25, 93]. The lowest observable effect level in adults was observed with paresthesia, occurring at approximately 100ppm in the hair; whereas motor disturbances such as ataxia where seen at 200ppm and higher [3]. The estimated lowest observable effect level in developmental exposure in these poisoning was presented with motor symptoms, primarily delayed walking, occurring at approximately 10ppm Hg in maternal hair samples [3, 94]. Several epidemiological studies have also suggested that approximately 10ppm Hg in maternal hair may lead to neurodevelopmental deficits, including abnormal reflexes, poor leg coordination, learning, and memory [15, 18, 95]. The fetal brain dose has been estimated to be 20-fold lower than that of maternal hair [96], suggesting that lowest observable effect level occurs at brain doses of 0.5ppm. However, some studies have suggested that even lower concentrations of Hg in maternal hair affects cognition and behavior [38, 97].

Neurotoxicants such as MeHg are known to disrupt both the total number of neuronal cells in the developing brain, as well as the ontogeny and connectivity of these cells. Mouse embryonic stem cells undergoing differentiation exhibit MeHg dependent changes in expression levels of neuronal progenitor markers, numbers of differentiating neurons and neurite outgrowth [98]. Furthermore, a human neural stem cell-based model has shown the importance of neuronal developmental stage at the time of MeHg exposure, with more immature cells being the most vulnerable to MeHg toxicity [99, 100]. Additionally, several studies have shown neuroprogenitor cells (NPCs) are highly sensitive to MeHg, with 25–250 nanomolar concentrations of MeHg inducing apoptosis in primary mouse, rat, and human NPCs as well as in the C17.2 neural stem cell line [100–102]. Concentrations of 2.5 nanomolar, a dose which leads to accumulation of approximately 0.5ppm in the cells [103], were shown to inhibit differentiation of rat NPCs [101].

The neural effects of MeHg reveal themselves in the neurodevelopmental and behavioral deficits observed in animal models and human exposures. These deficits include motor impairment and psychological disturbances, sensory and cognitive issues, reductions in neuron number across multiple brain regions with cortex and cerebellum being the most effected [27, 104]. However, an understanding of the differential influence of MeHg across the developing and mature mammalian brain is lacking. Comparing concentrations of MeHg that induce cell death in various cell culture systems, neural stem cells and NPCs appear to be the most sensitive to MeHg [76]. However, if the neural stem cell and neuroprogenitor stages are the most sensitive to MeHg, this brings to question why there is regional and cell selective toxicity, with the cerebral and cerebellar cortices being the most sensitive. Additionally, within the cerebellum, disruption of glutamatergic granule cells is a hallmark of MeHg exposure; whereas the neighboring GABAergic Purkinje cells are largely unaffected [23].

5. MeHg may preferentially target the glutamatergic (GLUergic) and dopaminergic (DAergic) systems

Although astrocytes exhibit preferential MeHg accumulation [105–108], neurons are more susceptible to MeHg-induced toxicity. Compelling evidence invokes cortical GLUergic dysfunction in MeHg-induced developmental neurotoxicity with persistence into adult life [22, 44, 71, 109–117]. Thus, neuronal cells from diverse brain structures are potential targets in developmental exposure to MeHg [118–123]. In terms of specific evidence for neurotoxicity to GLUergic neurons, cell culture experiments have demonstrated inhibition of glutamate (GLU) uptake by cultured astrocytes exposed to MeHg [124, 125]. In addition, GLU uptake into synaptic vesicles in both cerebral cortical slice culture [114] and primary culture of rat neurons [126] is reduced, which may suggest an increase in extracellular GLU could occur due to this decreased uptake into both astrocytes and neurons. Consistent with this, ex vivo studies have revealed decreased GLU uptake in adult [116] and weanling [127] mice cerebral cortex slices exposed to MeHg. Extracellular GLU levels may also be increased by MeHg effects to increase the spontaneous release of GLU, as observed in cerebellar slices [128] and cultured neuronal cells [129]. Increased levels of hydrogen peroxide during MeHg exposure may be involved in the oxidative and inhibitory effects of MeHg on astrocyte GLU transporters [130]. Whereas, the observed increase in GLU release may be at least partially explained by decreased vesicular uptake of GLU, perhaps from direct inhibition of the H⁺-ATPase activity [126]. Finally, an increase in extracellular brain GLU levels have been confirmed with in vivo studies by microdialysis of the frontal cortex of MeHg exposed adult Wistar rats [131].

Substantial evidence also invokes nigral DAergic dysfunction in MeHg developmental neurotoxicity and persistent/adult neurotoxicity [42, 68, 132–138]. Specific evidence for neurotoxicity to mesencephalic DAergic neurons includes data showing that MeHg inhibited dopamine (DA) uptake and metabolism [132, 139, 140], and developmental MeHg administration decreased rat striatal DA synthesis and turn-over [138]. In vitro rat synaptosomes show age-dependent [postnatal day (PND) 7>PND 21] sensitivity to MeHg [42], characterized by higher DA release and lower DA transporter (DAT) activity. A MeHg-induced increase in DA efflux has been supported with in vitro studies of mouse striatal slices as well as in an in vivo rat model [141–143]. Additionally, one study using rat striatal, cortical, and hypothalamic synaptosomes, showed a MeHg dose-dependent release of serotonin (5-HT), dopamine (DA), and noradrenaline (NA), with MeHg causing a significant increase in DA at 1µM, a concentration at which neither 5-HT or NA efflux was affected, suggesting a sensitivity of DA efflux to MeHg toxicity [144]. However, other studies have suggested that the increase in DA efflux may be a secondary effect of MeHg activating the glutamate receptor, NMDA, potentially through the increased efflux of glutamate, as NMDA receptor antagonists protect against MeHg-induced DA release [145].

Behavioral changes indicative of altered DAergic neurotransmission are also evident both in pre-pubertal and adult rats after chronic perinatal exposure to MeHg [136]. Increased sensitivity to the DA reuptake inhibitor, cocaine [135], and transient effects on DA receptor number associated with behavioral dysfunctions are noted in rat pups exposed to a single high dose of MeHg at late gestation [136, 137, 146, 147]. These behavioral alterations are accompanied by significant reduction in D2 receptor binding in the caudate/putamen [140, 148]. Furthermore, in the worm, pdr-1 (homologue of mammalian parkin/PARK2) and MRP-7 (multidrug resistance protein 7) loss-of-function mutations increase DAergic neuronal susceptibility to MeHg, attesting to gene x environment interactions [133, 134], which may exacerbate neurobehavioral outcomes associated with MeHg toxicity.

6. Stem cells as a translational model of neurotoxicity.

Quantitative and qualitative risk assessment of environmental and pharmacological exposures for public and personal health advice is a major goal of neurotoxicological research. However, this goal can be difficult to achieve given the challenges of human subject research and the real possibility of species-specific toxicological responses in animal models. Therefore, neuronal lineages derived from human stem cells or human neuronal precursors are an attractive solution, allowing the validation of neurotoxicological model systems data in humans. Stem cell technology has rapidly advanced in the past two decades and has begun to be applied to the field of toxicology. One of the first major successful use of stem cells in the study of toxicity was the Embryonic Stem Cell Test (EST) developed by Speilmann and colleagues [149–151]. This approach differentiated mouse embryonic stem cells (ESCs) into cardiomyocytes in the presence of toxic agents that could impinge upon developmental processes [149, 150]. Although this method used murine stem cells, and focused on differentiation into beating cardiomyocytes, the method was broadly recognized for its ingenuity [152–154]. However, this methodology had notable shortcomings in its application to neurotoxicology; although the EST correctly classified the majority of known embryotoxic chemicals tested, it sometimes failed to correctly classify MeHg as a developmental toxicant [155]. Potential reasons for these shortcomings of the EST include species-specific and tissue-specific toxicities.

Recently, Stummann et al. sought to address these issues by adapting the principles of the EST to toxicity testing in human ESCs (hESCs) undergoing neuronal differentiation [156]. They showed greater sensitivity of early-developing neural precursors versus maturing neuronal cells to MeHg toxicity by expression analysis of key early versus mature neurodevelopmental markers [156]. Edoff et al. found similar results with MeHg having an age-dependent effect and inhibiting differentiation in human neural progenitor cells at 8.5 and 16 weeks [100]. Other groups have also provided proof-of-principle experiments demonstrating the potential of hESCs to evaluate toxicity across different developmental stages [157]. However, ethical and regulatory concerns about the use human-embryo derived cells have limited hESC-based toxicity testing [158, 159].

The development of human induced pluripotent stem cell (hiPSC) technology has offered a new approach to human toxicological risk assessment [160–162]. Moreover, since hiPSCs are not derivatives of human embryos, many of the ethical and regulatory concerns associated with hESCs are avoided. This critical advance is widely expected to facilitate analysis of cellular physiological pathways in the context of human neurons and the underlying genetic factors that lead to disease [163]. Thus, hiPSC technology provides an opportunity to characterize toxicological outcome measures of human neurons with identical genetic determinants of actual human subjects for which exposure and risk analysis information can be obtained. This opportunity may be especially advantageous in determining the risk of exposure to a legacy pollutant such as MeHg for which, despite being studied extensively, many questions remain as to the genetic and environmental modifiers of toxicity as well as to its selective toxicity.

It remains to be seen if neural lineage developing from hiPSCs will display development stage and lineage selective toxicity reflective of the apparent selective toxicity seen in vivo. We propose that if this in fact occurs, that hiPSC models are exceptionally suited for mechanistic dissection of the selective neurodevelopmental toxicity of MeHg. Previously, the selective toxicity of MeHg and its mechanisms of action have been studied in both in vitro and in vivo models, which may be confounded by factors such as species and cell-line differences. Additionally, neuron-neuron and neuron-glia interactions in in vitro models may further complicate results, as primary versus secondary effects may be difficult to deduce. For example, as described above, it is currently unknown whether or how much a MeHg-induced increase in GLU efflux contributes to the MeHg-induced stimulation of DA efflux. The hiPSCs model will allow for the dissecting out of these mechanisms with a controlled and high-throughput model of isolated neurons from human subjects. Differentiated neurons can be generated within 30 days and mature neurons within 90 days, using well-established protocols, including generation of cortical pyramidal GLUergic neurons [164, 165] and nigral-like mesencephalic neurons [166, 167]. These protocols also allow for exposure during precise neurodevelopmental windows and for the determination of neurodevelopmental outcomes; such as expression of key lineage markers, morphology, and neural activity, at specific stages of development.

7. Significance and conclusion

Given the strong link between genetic variation and developmental MeHg-induced neurotoxicity new research is needed to (1) develop improved research models for MeHg-induced neurological disease, and (2) to elucidate molecular and genetic bases of MeHg-induced cell-selective (i.e. GLUergic and DAergic) neurodegeneration, and (3) to identify mechanistic-based neuroprotective strategies to mitigate human exposure risk. We hypothesize that genetic pathways that alter susceptibility to MeHg-induced neurotoxicity will display neural lineage- and developmental stage-dependent activity. Future studies to test this hypothesis have the potential to reveal novel insight into the mechanistic basis of selective neurotoxicity of MeHg. There is also great potential for linking forward and reverse genetic screening approaches in genetically tractable model systems such as the fly, worm and zebrafish to identify candidate pathways to then study in human based neuronal stem cell model systems discussed here. Such approaches would allow examination of how specific modifiers of MeHg neurodevelopmental toxicity act at distinct developmental lineages and stages. In addition, genomic analysis of the transcriptome, proteome, or metabolome changes coinciding with developmental exposures may also reveal human-specific pathways that mediate selective neurotoxicity. As selective neuropathology is poorly understood, the utilization of developing human neural lineages of differential sensitivity to neurotoxicants provides a highly rationale scientific approach to mechanistic studies of environmental hazards.

Highlights.

• Methylmercury (MeHg) is a potent neurotoxicant affecting both the developing and mature central nervous system (CNS)

• MeHg displays developmental stage and neural lineage selective neurotoxicity.

• We propose that human-based pluripotent stem cell cellular approaches may enable mechanistic insight into genetic pathways that modify sensitivity of specific neural lineages to MeHg-induced neurotoxicity

• Genetic and environmental modifiers contribute significant variability to neurotoxicity associated with human exposures.