We find that treatment of NHA with ethylmercury causes an increase in mitochondrial superoxide generation as shown in . However, the increase in superoxide generation is identical to the increase in the levels of protein carbonyls as shown in . H2
-induced formation of dichlorofluoresein from H2
DCF-AM is only approximately 20% greater than superoxide/carbonyl formation, which suggests that the loss of peroxidase function is not a feature of NHA ethylmercury toxicity. This is consistent with the effect of methylmercury on HeLa cells, where mitochondrial matrix generation of superoxide was implicated as the most damaging ROS [39
]. HeLa cells can be protected from methylmercury toxicity by upregulating mitochondrial Mn-SOD but not cytosolic Cu/Zn-SOD, GPx or catalase.
The majority of protein carbonyls in controls and in ethylmercury-treated NHA are also colocalized with mitochondria, as shown in . The peroxides measured via H2
DCF-AM and protein carbonyls are derived from mitochondrial ROS generation, as shown by colocalization of signals with the specific mitochondrial superoxide probe, MitoSox, as shown in . These findings are in broad agreement with the known generation of ROS on either side of the inner mitochondrial membrane in normal mitochondria [40
] and effects of methylmercury on rodent astrocytes observed by Shanker and coworkers [30
], as they too identified that mitochondria are the main production sites of increased superoxide generation.
In addition to measuring peroxide/superoxide generation we also examined the formation of HO•
using the specific probe HPR and using the Fpg
TUNEL assay which measures oxidized DNA bases. The conversion of guanine to 8-hydroxyguanine and 8-hydroxyguanine to more oxidized DNA hydantoin lesions, spiroiminodihydantoin and guanidinohydantoin, is generally believed to be due to HO•
or to Fenton's reagent (oxy-ferry; Fe(IV)
) and oxy-cupryl Cu(III)
8-hydroxyguanine, spiroiminodihydantoin, and guanidinohydantoin are substrates from the Fpg
TUNEL assay [35
]. In we demonstrated that while the levels of damaged nuclear DNA and mtDNA are very low in untreated cells, ethlymercury induces a large increase in oxidized mtDNA lesions. The highest levels of damaged mtDNA and protein carbonyls occur in structures that appear to be flocculated mitochondria. These grainy, oxidized, structures are not present as bright grains when viewed using Mitotracker, when carbonyl rich grains can be identified, shown in . These same vermiform structures are also identified in treated cells labeled with specific probes for both superoxide and HO•
seen in . However, although we observe an increase in the levels of cytosolic (hence mitochondrial) blunt-ended breaks and nicks in , very high levels of DNA breaks are not present in granular form. Thus, these flocculated mitochondria represent a dead-end mitochondrial state and given the close correlation between Fpg
TUNEL and Caspase-3 upregulation shown in , it is reasonable to conclude these are mitochondria that have undergone the permeability transition [20
], resulting in the release of proapoptotic proteins like cytochrome c
and DIABLO from the intermembrane space, mitoposis, and the initiation of the Caspase-3 apoptotic cascade [4
4.1. The Mechanism of Superoxide, Peroxide, and HO• Generation in NHA
It has long been known that organomercury reacts with iron sulfur centers [43
]; indeed methylmercury has been used as an aid to identify mercury adducts in iron-sulfur protein crystal structures for decades. The reaction of organomercury with iron sulfur centers in proteins such as aconitase results in loss of enzymatic function, the formation organomercury thioether adducts, and exposure to the bulks aqueous phase to redox active iron or release of free iron. It has been shown that, in mouse brain, the mitochondrial iron-sulfur complex rich enzyme NADH/Quinone oxidoreductase (Complex I) is highly sensitive to methylmercury [44
]. In a study by LeBel et al. it was found that methylmercury neurotoxicity was partially iron mediated [37
]. The potent iron-chelator, Deferoxamine, protected rat cerebellum from ROS following an injection of methylmercury. Iron chelation also protected neurons from ROS following invitro
exposure to methylmercury, but there was no evidence of deferoxamine-mercurial complex formation [37
]. Methylmercury treatment of isolated mitochondria, from the cerebrum, the cerebellum and from liver, causes an inhibition of respiration and increased superoxide/hydrogen peroxide formation [29
], mostly via damage to succinate dehydrogenase. The three iron-sulfur centers of succinate dehydrogenase on the matrix side of the inner mitochondrial membrane are the likely site of inhibition and possible iron release given that these clusters are sulphide/iron labile towards the thiophilic reagent, p
Based on the work reported here and by others, we suggest a mechanism for the toxicity of organomercury, which is shown in diagrammatic form in . As a lipophilic cation, ethylmercury will become concentrated inside astrocytes, with respect to the bulk extracellular phase, following the plasma membrane potential of 45
] by a factor of 5.6 fold, and cytosolic ethlymercury will partition into the mitochondria by a factor of 1,000 fold, its accumulation driven by the approximate 180
mV mitochondrial membrane potential [25
Figure 7 Proposed mechanism for the toxicity of organomercury. (a) As a lipophilic cation, ethylmercury will become concentrated inside astrocytes, following the plasma membrane potential of 45mV , by a factor of 5.6 fold, and cytosolic ethlymercury (more ...)
Inside the mitochondria the ethylmercury will react with iron-sulfur centers, causing the release of iron into the mitochondrial matrix, .
The role of ethlymercury in ROS species formation and detoxification is shown in . The iron-sulfur centers of oxidoreductases (e.g., succinate dehydrogenase) when damaged by organomercury not only generate free iron, ( I), but also form intraenzyatic carbon radical species ( II) that will react with molecular oxygen to give rise to superoxide, ( III). Superoxide can react with either free iron generation, the ferrous ion, or be dismutated into hydrogen peroxide by the mitochondrial Mn-SOD. Ferrous ion, and hydrogen peroxide react to generate the highly oxidizing radical, hydroxyl radical, ( IV), an agent implicated in pathology and ageing [47
]. The levels of hydrogen peroxide would be generally lowered by the mitochondrial antioxidants, including glutathione-dependent selenol/thio-based peroxidases, like GPx and TrxR. However, these enzymes are inhibited by organomercury indirectly by depletion of glutathione, ( V), and directly by the capping of the active site selenol/thiol by organomercury, ( VI).
Thus, the release of iron catalyzes Fenton/Haber-Weiss chemistry leading to the formation of the highly oxidizing HO•
has multiple targets, including sensors of the permeability transition complex and also mtDNA. High levels of HO•
cause Mitoposis, leading to cytochrome c
release from the mitochondria and the initiation of apoptosis. We find that a consequence of ethylmercury exposure to NHA is damage to the mitochondrial genome. We find an increase in DNA nicks, breaks and most importantly, in the level of oxidized bases. Mitochondria typically have 150 copies of mtDNA and during aging or with exposure to environmental stressors, the number of error free copies of mtDNA undergoes a decline. According to Harman's free radical/mitochondrial theory of aging [47
], the production of ROS by mitochondria leads to mtDNA damage and mutations. These in turn lead to progressive respiratory chain deficits, which result in yet more ROS production, producing a positive feedback loop.
The results of this study suggest that ethylmercury is a mitochondrial toxin in human astrocytes. We believe that this finding is important, particularly since the number of diseases in which mitochondrial dysfunction has been implicated are rapidly increasing.