The interactive roles of zinc and calcium in mitochondrial dysfunction and neurodegeneration

Authors

  • Natalia B. Pivovarova,

    1. Laboratory of Neurobiology, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland, USA
    Search for more papers by this author
  • Ruslan I. Stanika,

    1. Laboratory of Neurobiology, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland, USA
    Search for more papers by this author
  • Galina Kazanina,

    1. Laboratory of Neurobiology, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland, USA
    Search for more papers by this author
  • Idalis Villanueva,

    1. Laboratory of Neurobiology, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland, USA
    Search for more papers by this author
  • S. Brian Andrews

    Corresponding author
    1. Laboratory of Neurobiology, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland, USA
    • Address correspondence and reprint requests to Brian Andrews, 49/3A62, 49 Convent Drive, National Institutes of Health, Bethesda, MD 20892-4477, USA. E-mails: sba@helix.nih.gov

    Search for more papers by this author

Abstract

Zinc has been implicated in neurodegeneration following ischemia. In analogy with calcium, zinc has been proposed to induce toxicity via mitochondrial dysfunction, but the relative role of each cation in mitochondrial damage remains unclear. Here, we report that under conditions mimicking ischemia in hippocampal neurons – normal (2 mM) calcium plus elevated (> 100 μM) exogenous zinc – mitochondrial dysfunction evoked by glutamate, kainate or direct depolarization is, despite significant zinc uptake, primarily governed by calcium. Thus, robust mitochondrial ion accumulation, swelling, depolarization, and reactive oxygen species generation were only observed after toxic stimulation in calcium-containing media. This contrasts with the lack of any mitochondrial response in zinc-containing but calcium-free medium, even though zinc uptake and toxicity were strong under these conditions. Indeed, abnormally high, ionophore-induced zinc uptake was necessary to elicit any mitochondrial depolarization. In calcium- and zinc-containing media, depolarization-induced zinc uptake facilitated cell death and enhanced accumulation of mitochondrial calcium, which localized to characteristic matrix precipitates. Some of these contained detectable amounts of zinc. Together these data indicate that zinc uptake is generally insufficient to trigger mitochondrial dysfunction, so that mechanism(s) of zinc toxicity must be different from that of calcium.

image

Zinc and calcium are both implicated in ischemic injury. While calcium toxicity is mediated by mitochondrial dysfunction, mechanisms of zinc toxicity are unclear. Here we show that under conditions mimicking ischemia in hippocampal neurons, glutamate-induced mitochondrial dysfunction – swelling (red mitochondrion), depolarization and increased ROS generation – is triggered only by calcium, while zinc mainly facilitates calcium-dependent mitochondrial damage.

Abbreviations used
AMPAR

α-amino-3-hydroxy-5-methylisoxazole-4-propionate receptor

HBSS

hepes-buffered saline solution

MMP

mitochondrial membrane potential

OGD

oxygen/glucose deprivation

ROS

reactive oxygen species

VGCC

voltage-gated calcium channel

Zn2+ plays an important role in regulating normal neuronal function. For example, Zn2+ that is sequestered in the synaptic vesicles of glutamatergic synapses and normally released together with neurotransmitter can raise the transient concentration of free Zn2+ at the synaptic cleft to ≥ 100 μM (Sensi et al. 2011), which is sufficient to effectively modulate synaptic function through a variety of mechanisms, both pre- and postsynaptic (Paoletti et al. 2009; Pan et al. 2011). However, synaptically released Zn2+ is also thought to be involved in ischemic cell death (Sensi et al. 2009; Shuttleworth and Weiss 2011). Thus, spreading depression occurring after stroke (Dohmen et al. 2008) – which can be mimicked by oxygen/glucose deprivation (OGD) or K+ depolarization in hippocampal slices – results in synaptic Zn2+ release and an increase of cytosolic Zn2+ in CA1 neurons (Carter et al. 2011).

Under normal conditions in the presence of physiological extracellular Ca2+, glutamate excitotoxicity is mainly mediated by excessive Ca2+ influx through NMDA receptors (NMDARs), which are highly permeable to Ca2+ but not to Zn2+. Indeed, Zn2+ on the order of 100 μM is a known inhibitor of the large majority of hippocampal NMDARs (Rachline et al. 2005). Unlike NMDARs, activation of AMPA/kainate receptors (AMPARs) (Carriedo et al. 1998; Rego et al. 2001) or voltage-gated calcium channels (VGCCs) (Stanika et al. 2012) in normal media generally results in limited Ca2+ loading and toxicity, whereas glutamate-mediated Zn2+ uptake primarily depends on VGCCs (Kerchner et al. 2000) and AMPARs, particularly GluR2-lacking, Ca2+-permeable AMPARs (Weiss 2011). These characteristic, partially overlapping routes of ion entry presage recent evidence indicating interdependent roles for Zn2+ and Ca2+ in ischemic neuronal injury. For example, an OGD-induced, N,N,N',N'-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN)-sensitive increase in cytosolic Zn2+ has been implicated in Ca2+ deregulation in CA1 (Medvedeva et al. 2009), while glutamate-induced mobilization of intraneuronal Zn2+ appears to be critically dependent on Ca2+-mediated intracellular acidification (Kiedrowski 2011) or the generation of reactive oxygen species (ROS) (Dineley et al. 2008).

It has been suggested, in analogy with Ca2+, that Zn2+ induces toxicity via mitochondrial dysfunction and/or ROS generation (Sensi et al. 1999; Bossy-Wetzel et al. 2004; Bonanni et al. 2006; Medvedeva et al. 2009). In cultured neurons, kainate-mediated Zn2+ increases have been reported to induce mitochondrial Zn2+ uptake, resulting in a longer loss of mitochondrial membrane potential (MMP) compared to Ca2+-induced MMP depolarization, as well as a prolonged duration of ROS production (Sensi et al. 1999). However, recent reports demonstrate that toxic Zn2+ does not acutely depolarize mitochondria or initiate the mitochondrial permeability transition as does Ca2+ (Malaiyandi et al. 2005; Devinney et al. 2009), calling into question whether Zn2+ is an effective inhibitor of mitochondrial function.

This study examined the role of Zn2+ and Ca2+ in the degeneration of cultured hippocampal neurons stimulated with toxic concentrations of glutamate or kainate or with strong potassium depolarization, with the goal of dissecting the contribution of each cation to mitochondrial dysfunction. We found that although Zn2+ entering via VGCC or AMPARs is accumulated and precipitated in mitochondria in a manner reminiscent of Ca2+, it does not reach concentrations high enough to cause significant mitochondrial damage by any mechanism that would be reflected in swelling, depolarization, or ROS generation.

Materials and methods

Cell culture

All experiments and procedures, including animal euthanasia, as well as the design and analysis of the research, were performed in accordance with NINDS/NIH Animal Care and Use Committee protocol #1159-12 and are in compliance with ARRIVE guidelines for the use of animals in research. Primary cultures of rat hippocampal neurons were prepared as described previously (Pivovarova et al. 2004). In brief, suspensions of a mixed population of brain cells prepared by papain dissociation of hippocampi from ~ 20-day embryonic Sprague-Dawley rats were plated onto previously prepared glial layers in minimal essential medium supplemented with 5% horse serum, 1% fetal bovine serum, 2 mM Glutamax I (Life Technologies, Gaithersburg, MD), 136 μM uridine, 54 μM 2-deoxy-5-fluoro-uridine, and the growth factor cocktail N3 (Pivovarova et al. 2004). Half of the medium was replaced twice weekly. Experiments were performed on ~ 3-week-old cultures.

For experiments, cultures were transferred to a Hepes-buffered saline solution (HBSS) containing (in mM): 137 NaCl, 10 Hepes, 2.0 CaCl2, 1.0 MgSO4, 5.4 KCl, 0.3 Na2HPO4, 0.22 KH2PO4, 10 glucose, and 26 sucrose (320 mOsM, pH 7.4). Excitotoxic glutamate stimulation was carried out in Mg2+-free HBSS containing 10 μM glycine and 200 μM glutamate. VGCC activation was performed by depolarizing the plasma membrane with a high K+ (90 mM) solution (prepared by substituting K+ for Na+ in HBSS) in the presence of an AMPAR blocker (10 μM CNQX) and an NMDAR blocker (20 μM MK-801). AMPA/kainate receptors were activated with 100 μM kainate in HBSS in the presence of blockers of NMDARs (20 μM MK-801) and VGCCs (10 μM nimodipine). To achieve effectively toxic Zn2+ concentrations in incubation media, the concentration of added zinc was set at 300 μM. This compensates for the lower concentration of free zinc in solution compared to total zinc, presumably owing to complexation with inorganic phosphate (Rumschik et al. 2009).

For viability assays, cultures after stimulation were washed in HBSS and returned to the incubator in growth medium. Cell death was assayed at 6 h by propidium iodide (3.3 μg/mL) staining. (As a result of significant cell loss death assessment at 24 h was not reliable.) Immunostaining for NeuN, a neuron-specific protein, was used to assay the total number of neurons. Cell counting was carried out using ImageJ software (http://rsb.info.nih.gov/ij/index.html).

Fluorescence microscopy

Intracellular Zn2+ imaging was carried out using low-affinity (Kd =1 μM) Newport Green (excitation 488 nm, emission 530 nm). For dye loading, cells were incubated for 25 min at 37°C in 2.5 μM Newport Green in HBSS and washed an additional 15 min before imaging. For imaging mitochondria, cells were incubated for 15 min at 37°C in 200 nM MitoTracker Green FM in HBSS. The ratio of major to minor axes, as determined on single confocal slices, was calculated as a semiquantitative measure of mitochondrial swelling. MMP was visualized with the membrane-permeant cationic dye rhodamine-123 (Rh123) (excitation 514 nm, emission 530 nm). Cultures were incubated for 15 min at 37°C in 25 μM Rh123 in HBSS. Cellular ROS production was assayed by recording the oxidation of 2 μM dihydroethidine (DHE; excitation 488 nm, emission 530 nm), which was present in the medium during imaging; prior to use pre-oxidized DHE was removed from stock solutions by exposure to a bed of Dowex 50WX4 ion exchange beads. Mitochondrial ROS was detected with MitoSOX (5 μM; excitation 488 nm, emission 530 nm), which was loaded for 10 min at 23°C and washed before imaging. In experiments with Rh-123, DHE and MitoSOX fluorescence was recorded in somatic areas. Imaging was performed at 23°C using a Zeiss LSM 510 confocal microscope equipped with 40× and 63× objectives (Carl Zeiss Microscopy, Thornwood, NY, USA).

Electron microscopy and electron probe X-ray microanalysis

Electron micrographs of conventionally processed cultures were recorded using a JEOL 1200 microscope (JEOL; Peabody, MA, USA). For electron probe X-ray microanalysis, cultures were rapidly frozen, cryo-sectioned, imaged, and analyzed in a Zeiss EM-912 analytical cryo-electron microscope (Carl Zeiss Microscopy, Thornwood, NY, USA) as described previously (Pivovarova et al. 2004). X-ray spectra were processed by established procedures (Pivovarova et al. 1999).

Chemicals and reagents

Nimodipine, CNQX, MK-801 were purchased from Tocris Bioscience (Ellisville, MO, USA) and NeuN from Chemicon (Temecula, CA, USA). The fluorescent probes FluoZin-3 AM, Newport Green DCF diacetate (Cat. No. N7991), MitoTracker Green, rhodamine-123, DHE, and MitoSOX were purchased from Life Technologies (Carlsbad, CA, USA). All other reagents were from Sigma-Aldrich (St. Louis, MO, USA).

Statistics

One-way anova followed by post hoc Dunnett multiple comparisons test or Kruskal–Wallis rank anova with post hoc Dunn's test were used for data analysis. Data are given as mean ± SEM. Analyses were carried out using InStat software (GraphPad Software, San Diego, CA, USA). In general, experiments were performed on at least three cultures prepared from different animals.

Results

Calcium reduces zinc accumulation and toxicity in hippocampal neurons

We first evaluated the dynamics of cytosolic Zn2+ in hippocampal neurons after activation of the major receptors or channels that underlie Zn2+ entry. In a standard 2 mM Ca2+ HBSS containing 300 μM exogenous Zn2+, glutamate induced Zn2+ elevations were small but detectable by the low-affinity Zn2+ probe Newport Green (Kd =1 μM) (Fig. 1a, blue traces). Application of Zn2+-specific ionophore sodium pyrithione at the end of the experiments – which floods cells with Zn2+, thereby providing an estimate of Fmax (Fig. 1a) – confirmed the small size of glutamate-induced Zn2+ elevations in normal Ca2+ medium. Much larger glutamate-induced Zn2+ elevations were seen in a Zn2+-containing but nominally Ca2+-free medium (Fig. 1a, red traces). Enhanced Zn2+ uptake in Ca2+-free medium suggests a competition or interaction between Zn2+ and Ca2+ at sites of ion entry.

Figure 1.

Glutamate, potassium depolarization, or kainate induce Ca2+-sensitive Zn2+ uptake in hippocampal neurons. (a–c) Traces of Newport Green fluorescence (NPG, F/F0) from neurons exposed to (a) 200 μM glutamate plus 10 μM glycine (Glu), (b) 90 mM K+ in the presence of 20 μM MK-801 and 10 μM CNQX (90 K+), or (c) 100 μM kainate in the presence of 20 μM MK-801 and 10 μM nimodipine (KA). Each panel compares responses in 2 mM Ca2+-containing hepes-buffered saline solution (HBSS) (blue) with those in Ca2+-free HBSS containing 300 μM Zn2+ (red). At the end of each experiment, neurons were exposed to 20 μM sodium pyrithione (Pyr). Both datasets in each plot show individual responses from 8 to 10 neurons in representative coverslips from three to four experiments at each condition. (d) Summary of maximal sustained level of NPG fluorescence 15 min after activation of Zn2+ uptake or 5 min after application of pyrithione in Zn2+ HBSS containing 2 mM Ca2+ (+Zn +Ca, blue), in nominally Ca2+-free HBSS (+Zn −Ca, red), or in Ca2+-free HBSS in the presence of 10 μM nimodipine and 10 μM CNQX (+Zn −Ca+Nim+CNQX, green). (e) – Neuronal death rate 6 h after 20 min exposure to glutamate, 90 mM K+, or kainate in Ca2+-containing HBSS without Zn2+ (−Zn +Ca, orange), with Zn2+ (+Zn +Ca, blue), or in Ca2+-free, Zn2+-containing HBSS with (+Zn −Ca+ Nim+CNQX, green) or without (+Zn −Ca, red) 10 μM nimodipine and 10 μM CNQX. In (b) and (c) data are given as mean ± SEM from 60 to 70 neurons from three to four experiments at each condition. Statistically significant differences (*p < 0.05) are indicated by bracketed pairs. Kruskal–Wallis rank anova with post hoc Dunn's test was used for data analysis.

To tease out contributions from various plausible routes of Zn2+ entry, we examined Zn2+ uptake through pharmacologically isolated channels. Activation of only the NMDA subtype of glutamate receptors did not induce any significant Zn2+ uptake, even in Ca2+-free medium (Figure S1). Similarly, blocking L-type VGCCs (10 μM nimodipine) and AMPARs (10 μM CNQX) strongly reduced glutamate-induced Zn2+ elevations in Ca2+-free medium (Fig. 1d), supporting previous evidence (Weiss et al. 1993; Sheline et al. 2002) that VGCCs and AMPARs are the main routes of Zn2+ uptake in neurons. Nonetheless, the significant residual Zn2+ elevations after blocking these channels indicates that contributions from other routes of entry, e.g., non-selective ion channels, cannot be ruled out.

In a Zn2+- and Ca2+-containing medium, VGCC activation by 90 K+ depolarization induced minimal Zn2+ uptake (Fig. 1b, d). However, in Ca2+-free medium Zn2+ uptake was greatly increased, becoming comparable in magnitude to that induced by glutamate (Fig. 1b, d). Similarly, kainate activation of AMPARs evoked stronger Zn2+ uptake in Ca2+-free medium, but only in a small fraction (~ 15%) of neurons, (Fig. 1c). This population of Zn2+-permeable cells likely consists of those expressing Ca2+-permeable AMPARs, as it correlates well with the fraction of kainate-responsive, Ca2+-permeable neurons (Figure S2). These observations indicate that Ca2+ attenuates Zn2+ uptake when Zn2+ enters through physiologically relevant channels but not, as indicated by the effect of pyrithione, via non-specific routes.

The results just described suggest that different routes and avidity of cation entry might lead to differences in strength and modes of toxicity. Under experimental conditions of glutamate exposure that fix control cell death at 40–50%, we find that 300 μM Zn2+ in an otherwise normal Ca2+-containing incubation medium was somewhat neuroprotective (Fig. 1e), probably because, as noted, Zn2+ at this concentration is an effective inhibitor of Ca2+ entry through NMDARs (Rachline et al. 2005). In contrast, 300 μM external Zn2+ strongly enhanced glutamate toxicity in Ca2+-free medium. These observations imply a fundamental change in the mechanism(s) of cell death when calcium overload via NMDARs is removed from available pathways, and are consistent with the idea that glutamate-induced Zn2+ toxicity is governed by Zn2+ entry mainly through VGCCs and/or AMPARs. The strong neuroprotective effect of nimodipine (Fig. 1e) also supports this conclusion.

In contrast to glutamate, depolarization or kainate exposure in a Zn2+- and Ca2+-containing medium dramatically enhanced neuronal cell death compared to normal Ca2+ medium. Zn2+ toxicity was further increased in Ca2+-free HBSS (Fig. 1e). The neuroprotective effect of external Ca2+ correlates well with the general effect of zero Ca2+ on Zn2+ uptake (Fig. 1a–d). To summarize, while elevated extracellular Zn2+ is protective against NMDAR-dependent Ca2+ overload, elevated extracellular Ca2+ inhibits VGCC-mediated Zn2+ entry and is protective against Zn2+ overload. Together, these mechanisms could conceivably become important under conditions that bathe the synaptic cleft in elevated concentrations of Zn2+.

Calcium but not zinc induces mitochondrial structural damage

VGCC activation in normal Ca2+-containing medium leads to mitochondrial damage in only a small fraction of responsive neurons (Stanika et al. 2012). Adding Zn2+ resulted in a major stimulus-dependent increase (from ~ 10% to > 50%) in the fraction of neurons with damaged, swollen mitochondria (Fig. 2a). In responsive neurons, there was a significant increase in mitochondrial total calcium, as determined by electron probe X-ray microanalysis, compared to the average calcium load of damaged mitochondria in neurons depolarized in a Zn2+-free medium (Fig. 2b). Mitochondrial calcium localized, as usual, to characteristic matrix precipitates (Fig. 2a, right panel, arrows). VGCC-mediated Zn2+ uptake did not lead to changes in intracellular sodium or potassium (Fig. 2d–e), indicating that Zn2+ accumulation does not compromise Na+/K+ homeostasis. Presumably, it also does not compromise ATP production, as this would be reflected in Na+ and K+ redistribution owing to the high ATP demand of Na+/ K+-ATPases (Jekabsons and Nicholls 2004).

Figure 2.

Zinc facilitates mitochondrial calcium accumulation and colocalizes with calcium. (a) Electron micrographs of freeze-dried cryosections prepared from hippocampal neurons rapidly frozen after 20-min exposure to 90 mM K+ in the presence of 1 μM Bay K 8644, 20 μM MK-801, and 10 μM CNQX in a Ca2+- and 300 μM Zn2+-containing hepes-buffered saline solution (HBSS). Left panel illustrates a non-responsive cell with structurally normal mitochondria; right panel is a responsive cell with swollen mitochondria that contain calcium-rich precipitates (arrows). Bar = 1 μm. (b) Depolarization-induced Ca2+ accumulation in mitochondria of responsive cells, as measured by electron probe X-ray microanalysis, increased in Ca2+- and Zn2+-containing medium (= 16) compared to a Ca2+ only medium (= 15; *p < 0.05). (c) X-ray spectra from a normal mitochondrion of a neuron exposed to 90 mM K+ for 20 min in Ca2+-containing medium (blue) and in a mitochondrial precipitate from a neuron depolarized in a Ca2+- and Zn2+-containing medium (red). Red spectrum contains large peaks for P and Ca and a distinct Zn peak at 8.64 keV; the latter peak is absent in the blue spectrum. (The Cu peak at 8.91 keV is a well-known artifact arising from stray radiation exciting the copper support grid.) (d) X-ray spectra from cytoplasm of neurons under control conditions (green) or after exposure to 90 mM K+ [as in panel (a), red] or 200 μM glutamate (blue) in Ca2+- and Zn2+-containing medium. (e) Summary analysis of cytoplasmic Na and K concentrations in neurons shows that the Na/K ratio does not change significantly after exposure to 90 mM K+ (with or without Zn2+, = 30 and = 35, respectively) compared with control (= 45), but is reversed after exposure to 200 μM glutamate (n =20; *p < 0.001). Values in panels (b and e) are mean ± SEM from five to six cells.

Interestingly, in addition to the expected high concentrations of calcium and phosphorus, a small but detectable amount of zinc was found in the mitochondrial precipitates after VGCC activation in Zn2+- and Ca2+-containing medium (Fig. 2c). No zinc-containing mitochondrial precipitates were present in Ca2+-free medium, nor was zinc detected in any other location regardless of the media composition. Although the amount of zinc in precipitates is estimated to be < 5% of the amount of calcium, the presence of zinc in these structures is consistent with a considerable literature indicating that accumulated zinc translocates to mitochondria, and prompted us to ask whether zinc overload, like calcium overload, could lead to swelling, depolarization, and ROS production, that is, to the typical hallmarks of mitochondrial dysfunction.

Toxic glutamate exposure in normal Ca2+-containing medium induces strong mitochondrial swelling (Fig. 3, compare upper and middle rows). Added external Zn2+ had minimal impact on Ca2+ overload-induced swelling (data not shown). In contrast, glutamate exposure in a Zn2+-containing but Ca2+-free medium does not lead to significant mitochondrial swelling (Fig. 3, lower row), despite the massive Zn2+-dependent death seen under these conditions (Fig. 1c). Axial/equatorial ratios, calculated as a measure of mitochondrial swelling, were 7.2 ± 0.5 (SEM) for controls, 1.3 ± 0.7 for glutamate activation in Ca2+-containing medium (p < 0.001), and 7.7 ± 0.5 in Zn2+-containing medium (n.s.).

Figure 3.

Glutamate-stimulated Zn2+ accumulation does not induce mitochondrial structural damage. Representative confocal fluorescence images of MitoTracker Green (left column) and electron micrographs (right column) of neurons in normal Ca2+-containing medium (control, top) or after 20-min exposure to 200 μM glutamate plus 10 μM glycine in Ca2+-containing, Zn2+-free medium (middle) or in Ca2+-free, Zn2+-containing medium (bottom). Glutamate induces mitochondrial swelling in Ca2+-containing medium, but not in Zn2+-containing, Ca2+-free medium. Bar = 1 μm.

Unlike glutamate activation, the majority of neurons retain normal mitochondrial morphology after VGCC activation in normal Ca2+-containing medium without Zn2+ (Fig. 4, upper row), although a minor fraction of neurons, those that highly express VGCCs (Stanika et al. 2012), exhibit significant Ca2+ loading and mitochondrial swelling. However, identical stimulation in the presence of added Zn2+ leads to mitochondrial swelling in a much larger fraction (> 50%) of neurons (Fig. 4, middle row), a result that reflects Zn2+ enhancement of VGCC-dependent cytosolic and mitochondrial Ca2+ loading (see Fig. 2b). As with glutamate, depolarization in Ca2+-free medium, even in the presence of Zn2+, essentially eliminated mitochondrial swelling (Fig. 4, lower row). Axial/equatorial ratios were 7.8 ± 0.5 (SEM) for 90 K+ activation in Ca2+-containing medium, 1.6 ± 0.2 in Ca2+- and Zn2+-containing medium (p < 0.001), and 7.8 ± 0.5 in Ca2+-free, Zn2+-containing medium (n.s.).

Figure 4.

Depolarization-induced Zn2+ accumulation facilitates Ca2+-dependent mitochondrial structural damage. Representative fluorescence images of MitoTracker Green (left column) and electron micrographs (right column) of neurons after 20-min exposure to 90 mM K+ in the presence of 1 μM Bay K 8644, 20 μM MK-801, and 10 μM CNQX. Incubation media contained either Ca2+ without external Zn2+ (top), both Ca2+ and Zn2+ (middle), or Zn2+ without Ca2+ (bottom). Mitochondrial swelling is apparent in Ca2+- and Zn2+-containing media, but is minimal in Ca2+-containing medium or in Zn2+-containing, Ca2+-free media. Bar = 1 μm.

Calcium but not zinc induces mitochondrial depolarization and ROS generation

Glutamate induces fast and sustained mitochondrial depolarization – as detected by Rh-123 fluorescence increases – in a normal Ca2+-containing medium (Fig. 5a). Application of FCCP following glutamate exposure does not cause additional depolarization. However, in Ca2+-free, Zn2+-containing medium, glutamate does not induce significant mitochondrial depolarization (the ratio F/F0 at 15 min is 1.10 ± 0.06, p = 0.14, = 40) as confirmed by the response to FCCP (Fig. 5b). Similarly, in a normal Ca2+-containing medium, kainate activation strongly depolarizes mitochondria (F/F0 > 2.0), but only in that fraction (~ 20%) of neurons, thought to express Ca2+-permeable AMPARs (Fig. 5c, f). As with Zn2+ accumulation, this fraction is consistent with the fraction of neurons that avidly take up Ca2+ during kainate exposure (Figure S2). No significant kainate-induced mitochondrial depolarization (F/F= 1.03 ± 0.06, p = 0.64, = 57) was observed in a Zn2+-containing, Ca2+-free medium (Fig. 5d, f). Together, these results indicate that Ca2+ is much more effective than Zn2+ at triggering mitochondrial depolarization.

Figure 5.

Glutamate- or kainate-induced Zn2+ uptake does not depolarize mitochondria. (a–d) Time course of Rh123 fluorescence (F/F0) as a measure of mitochondrial membrane depolarization. Neurons exposed to glutamate or kainate (as in Fig. 1a) show strong mitochondrial depolarization only in Ca2+-containing hepes-buffered saline solution (HBSS) (a, c). The same stimuli evoke minimal depolarization in Zn2+-containing, Ca2+-free media until subsequently exposed to 1 μM FCCP (b, d). (e) Sodium pyrithione (20 μM) induced slow, strong mitochondrial depolarization in Ca2+-free, Zn2+-containing HBSS. (f) Scatter plots of maximal sustained levels of Rh123 fluorescence 10 min after kainate application show significant depolarization in Ca2+ HBSS in only a small fraction of neurons. Panels (a–e) show traces of 12–15 neurons from representative coverslips taken from three to four experiments at each condition; panel (f) includes combined data from three experiments at each condition.

As there is a considerable literature indicating that Zn2+ ions can indeed depolarize mitochondria, we tested the possibility that depolarization was not observed here because intracellular zinc levels were not sufficiently high. This can be tested using the Zn2+-specific ionophore sodium pyrithione (Malaiyandi et al. 2005), as exposure to this reagent in a Ca2+-free, Zn2+-containing medium induces essentially maximal cytosolic Zn2+ uptake (Fig. 1). Here, we find that the same ionophore treatment resulted in maximal, albeit slowly developing (Devinney et al. 2009), mitochondrial depolarization (Fig. 5e). These results demonstrate that Zn2+ can indeed disrupt mitochondrial function if Zn2+ elevations are high enough, although it is important that the threshold for Zn2+-induced cell death via glutamate-dependent routes is much lower than the level needed to damage mitochondria (compare, e.g., responses to glutamate vs. pyrithione in Fig. 1b).

Increased generation of ROS, at least partly derived from the mitochondrial respiratory chain, reflects oxidative stress and is another known indicator of impending cell damage. In normal Ca2+-containing media with no added Zn2+, over-activation of Ca2+ uptake by either glutamate, kainate or depolarization induces significant ROS production, as detected by the rate of dihydroethidium (DHE) fluorescence increase (Fig. 6a–d, red traces and bars). (In the case of kainate, significant ROS is seen in only a fraction of cells corresponding to the fraction expected to express Ca2+-permeable AMPARs.) These data are consistent with numerous previous reports linking Ca2+-dependent up-regulation of respiration to increased, and potentially deleterious, ROS production (Reynolds and Hastings 1995; Abramov et al. 2007; Duan et al. 2007). Conversely, these same three stimuli in a Ca2+-free, Zn2+-containing medium – conditions that induce strong Zn2+ accumulation (Fig. 1a, b) – generate only limited acute ROS (Fig. 6a–d, blue traces and bars). Similar results were obtained when ROS generated by mitochondria was detected using the mitochondrial ROS indicator MitoSOX (Fig. 6e–h), which reinforces the conclusion that much of the ROS generated by mitochondria is mediated by Ca2+, not Zn2+. As with conclusions based on swelling and depolarization, ROS production suggests that mitochondrial damage depends specifically on calcium accumulation, and zinc cannot substitute for calcium in this role.

Figure 6.

Ca2+- and Zn2+-dependent reactive oxygen species (ROS) production. Traces of ethidium fluorescence (F/F0) (left column) or MitoSox fluorescence (right column) in 2 mM Ca2+ hepes-buffered saline solution (HBSS) (red) or Ca2+-free HBSS containing 300 μM Zn2+ (blue) from neurons exposed (as in Fig.  1) to glutamate (a, e), kainate (b, f), or 90 mM K+ (c). Each panel shows traces of 18–20 representative neurons from three to four experiments at each condition. The rate of compartmentally non-specific, Zn2+-dependent ROS production reported by ethidium (d, blue bars) is detectable but much less than Ca2+-dependent ROS production after corresponding stimuli (d, red bars). Similarly, the rate of Zn2+-dependent, glutamate- or kainate-induced (in Ca-responsive cells, KACa) mitochondrial ROS production reported by MitoSox is much less than Ca2+-dependent ROS production (g). Representative confocal image (h) shows maintained mitochondrial localization of MitoSox 20 min after glutamate activation in Ca2+-containing medium. Summary data are given as mean ± SEM from 40 to 60 neurons from three to four experiments under each condition. Statistical analyses, as determined by Kruskal–Wallis rank anova with post hoc Dunn's test, compare differences between ROS production in Ca2+ medium versus Zn2+ medium (*p < 0.05) or between stimulated versus basal ROS production in Zn2+ medium (p < 0.001).

Discussion

Synaptically released Zn2+ is known to be involved in neurodegeneration. For example, much evidence implicates Zn2+ accumulation through voltage-gated calcium channels or AMPA/kainate receptors in neuronal death following ischemic injury (Shuttleworth and Weiss 2011). Under pathological conditions, the transient concentration of extracellular Zn2+ in the vicinity of the synaptic cleft can reach 100 μM largely owing to the release of Zn2+ from glutamatergic synaptic vesicles (Sensi et al. 2011). This surge of Zn2+ provides a source for the harmful elevation of intracellular Zn2+, as, for example, occurs during spreading depression induced by OGD in hippocampal slices (Carter et al. 2011). The simultaneous availability of extracellular Ca2+ and Zn2+ leads to some important functional consequences attributable to competition and/or interaction between these two cations. Thus, high glutamate induces only a modest accumulation of exogenous Zn2+ in normal Ca2+-containing medium spiked with potentially toxic levels (300 μM) of Zn2+. However, under these conditions Ca2+ uptake and Ca2+-dependent excitotoxicity were actually attenuated because Zn2+ at concentrations in the micromolar range blocks the large majority of hippocampal NMDARs (Paoletti et al. 2009), which are the dominant routes of Ca2+ entry.

In a nice example of reciprocity, extracellular Ca2+ similarly appears to protect against Zn2+ toxicity, as evidenced by the strongly increased glutamate-induced Zn2+ uptake and toxicity in Ca2+-free medium. In this case, however, the basis of this effect is likely to be different, namely, relief from Ca2+ block of Zn2+ permeation at VGCCs, the main route of Zn2+ entry (Shuttleworth and Weiss 2011). More direct evidence implying ion–ion interaction at VGCCs comes from our experiments showing that extracellular Ca2+ inhibits VGCC-mediated Zn2+ uptake and therefore is highly protective against Zn2+ toxicity, which is consistent with earlier reports (Weiss et al. 1993; Sheline et al. 2002).

Mitochondria play an important role in Zn2+-dependent neurodegeneration, but whether these organelles are the main targets or effectors of Zn2+ toxicity, as has been suggested (Sensi et al. 1999; Bossy-Wetzel et al. 2004; Medvedeva et al. 2009), is an open question. Mitochondrial dysfunction caused by high concentrations of Zn2+, i.e., concentrations in the micromolar range, has been demonstrated in preparations of isolated mitochondria; for example, 10 μM Zn2+ promoted swelling of isolated liver mitochondria and increased respiratory H2O2 production (Bossy-Wetzel et al. 2004). The inference follows that Zn2+ toxicity is at least partly mediated by mitochondria-dependent ROS generation (Sensi et al. 1999; Bossy-Wetzel et al. 2004). On the other hand, mitochondria in similar preparations were reportedly depolarized only partially and reversibly by 30 μM Zn2+ (Devinney et al. 2009), while in forebrain neurons even artificially high, supra-toxic intracellular Zn2+ concentrations minimally depolarized mitochondria, further suggesting that in neurons Zn2+ exposure does not significantly compromise mitochondrial function (Malaiyandi et al. 2005).

Present results contribute several new, important pieces of information to this debate. First, when both Ca2+ and high Zn2+ are present in the extracellular medium, as might well be the case after an ischemic event, VGCC-mediated Zn2+ uptake significantly increases mitochondrial Ca2+ accumulation, mitochondrial swelling, and cellular toxicity relative to a Zn2+-free medium. In our cells, excess Ca2+ uptake was accompanied by mitochondrial Zn2+ accumulation, which result provides the first direct evidence that Zn2+ taken up into neurons is translocated to mitochondria, where it colocalizes with Ca2+ in characteristic phosphate-rich precipitates. Although the amount of Zn2+ in these precipitates is relatively small – the Ca2+/Zn2+ ratio is estimated to be ~ 20 – it is associated with a significant increase in the fraction of damaged cells. Thus, it is conceivable that mitochondrial Zn2+ accumulation potentiates Ca2+-dependent mitochondrial damage.

This is reinforced by the observation that under all ischemia-relevant stimulus protocols examined, mitochondrial swelling, depolarization, and ROS generation were only observed in the presence of external Ca2+, and never in its absence. This indicates that mitochondrial dysfunction, at least in hippocampal neurons, is specifically dependent on Ca2+ accumulation. Zn2+ appears to disrupt mitochondrial function only after excessive, ionophore-induced Zn2+ uptake, but not through physiological loading mechanisms. These data clearly illustrate the primacy of Ca2+ loading relative to Zn2+ with respect to mitochondrial dysfunction. However, they do not address the mechanism of direct Zn2+ toxicity. This is evidently an important issue as there are relevant scenarios, mainly characterized by low levels of Ca2+, in which accumulated Zn2+ ions alone are the toxic agents. For example, glutamate was significantly more toxic in high Zn2+ but Ca2+-free media even though only minimal mitochondrial damage was seen. Similarly, enhanced Zn2+ toxicity in Ca2+-free media induced by depolarization or kainate was accompanied by insignificant mitochondrial swelling, depolarization, or ROS production. Taken as a whole, our results suggest that mitochondrial dysfunction is not the primary mechanism of Zn2+ toxicity in neurons.

In this case, alternative mechanisms of action must be considered. There is limited evidence in this regard, but some possibilities stand out. First, toxic Zn2+ exposure (depolarization in 300 μM Zn2+) of forebrain neurons did not compromise mitochondrial function, but did inhibit mitochondrial movement (Malaiyandi et al. 2005) which is known to play a role in apoptosis-like death mechanisms (see Youle and van der Bliek 2012). Zn2+-induced inhibition of mitochondrial movement and attendant neurotoxicity were prevented with blockers of phosphatidyl inositol 3-kinase (Malaiyandi et al. 2005). Independently, Zn2+ has been shown to accumulate in lysosomes and mediate H2O2-induced lysosomal membrane permeabilization, which leads to neuronal cell death (Hwang et al. 2008). Considering the potential clinical importance of understanding Zn2+ toxicity, this is clearly an area of research that warrants more attention.

Acknowledgments

We would like to thank Ms. Christine A. Winters, and the staff of the NINDS EM facility, Dr. Jung-Hwa Tao-Cheng, Director, for excellent technical assistance. This research was supported by the Intramural Research Program of the NIH, NINDS (Z01 NS002610). The authors declare no conflicts of interest.

Author contributions

NBP and SBA designed research; RIS, GK, IV, and NBP performed research; RIS, GK, NBP, and SBA analyzed data; and NBP and SBA wrote the manuscript.

Ancillary