NADPH oxidase contributes substantially to acute Ca²⁺- but not Zn²⁺-dependent ROS generation

Cortical neuronal cultures were prepared as described and allowed to mature for at least 10 days before use in experiments. To assess ROS generation, cultures were loaded with the superoxide-preferring indicator, HEt, which is oxidized by superoxide radicals into a highly fluorescent compound, ethidium [1]. Ethidium intercalates and binds to DNA resulting in amplification of fluorescence, providing high sensitivity and resulting in predominant visualization of the signal in the nucleus. Because the oxidized dye accumulates, the ROS production rate was assessed as the rate of fluorescence increase over time, and net ROS production was assessed as the increase in fluorescence (∆F) over baseline. To compare ROS generation across experiments, all fluorescence readings for a given cell (F_x) were normalized to the average fluorescence of that cell during a 10 min baseline period (F₀).

Ca²⁺ loading was induced by exposure to NMDA (100 µM, 30 min) to strongly activate highly Ca²⁺-permeable NMDA-type glutamate channels, and was carried out in HEPES buffered salt solution containing physiological (1.8 mM) Ca²⁺, an exposure which predictably induces rapid and high rises in intracellular Ca²⁺ (estimated in the 10’s of µM) [20]. Zn²⁺ can enter neurons through a number of routes including voltage-gated Ca²⁺ channels (VGCC) or Ca²⁺-permeable α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) channels [8]. However, Ca²⁺-permeable AMPA channels are only present on small minorities of neurons in the cultures [21], and VGCC do not permit sufficiently rapid Zn²⁺ entry to cause strong and rapid ROS generation in most neurons [22]. Therefore, to rapidly induce large Zn²⁺ loads in virtually all neurons, the cultures were pre-exposed to the membrane-permeable Zn²⁺ ionophore, pyrithione (10 µM) prior to a 5 min exposure to Zn²⁺ (300 µM) with pyrithione. Although most of our prior studies examining effects of Zn²⁺ entry into neurons on mitochondrial function were carried out with Ca²⁺ present in the extracellular buffer [10,22], in the present study, to fully discriminate effects of Zn²⁺ and Ca²⁺, all Zn²⁺ exposures were carried out in Ca²⁺-free buffer.

In the absence of either NMDA or Zn²⁺ addition, HEt fluorescence gradually increased in a linear fashion throughout the recording episode, presumably reflecting slow oxidation of the indicator due to basal metabolic activity of the cell (data not shown). With NMDA exposure, however, a sharp HEt ∆F began several min after the addition of the NMDA, with fluorescence increasing over baseline (F_x/F₀) typically ~ 5 fold within 30 min. The Zn²⁺ exposures caused a similar sharp HEt ∆F, which began typically 3-4 min after washout of the Zn²⁺/pyrithione (Figure 1A, B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. NADPH oxidase (NOX) inhibition attenuates acute Ca²⁺ - but not Zn²⁺-induced ROS production.

HEt-loaded cultures were exposed to 100 μM NMDA (30 min) or 300 μM Zn²⁺/pyrithione (5 min) alone (red) or after pre-treatment with and in the presence of apocynin (500 µM), or in glucose-free media supplemented with pyruvate (15 mM) (blue) as described (see Materials and Methods).

A: Representative images of selected fields of neuronal cultures (“Brightfield”), and pseudocolor images (400x) of HEt fluorescence from these neurons before and 30 min following onset of exposure to NMDA (left) or Zn²⁺/pyrithione (right). The pseudocolor bar shows the 12-bit fluorescence intensity range.

B: Traces show time course of HEt ∆F, normalized to baseline values (F_x/F₀). Dashed lines show linear extrapolation of baseline. Traces show mean ± SD values from 4 experiments.

C: Quantification of HEt ∆F changes. Values show F_x/F₀ increases after subtraction of the extrapolated baseline value, 30 min after onset of the exposure. ∆F values each represent means (± SEM) of the 4 experiments; * indicates difference from NMDA alone (p< 0.001) by 2-tailed t test.

https://doi.org/10.1371/journal.pone.0083347.g001

Two interventions were used to block the contribution of NOX to the ROS generation. First, we used the nonselective NOX inhibitor, apocynin (500 µM), which blocks the membrane translocation and activation of multiple NOX isoforms, including the primary neuronal isoform, NOX-2 [4,23,24]. In addition, we carried out some experiments, in which pyruvate (15 mM) was substituted for glucose as energy substrate (see Materials and Methods), since nicotinamide adenine dinucleotide phosphate (NADPH), the substrate of NOX, is primarily generated by the hexose monophosphate shunt during glycolysis. Both of these maneuvers substantially attenuated both the rate and extent of HEt oxidation induced by NMDA, while having no detectable effect on the Zn²⁺-triggered signal over the 30 min period following the start of the exposures (Figure 1A, B, C). Thus, these data suggest that NOX contributes importantly to acute ROS generation triggered by Ca²⁺ influx but does not contribute substantially to acute Zn²⁺-dependent ROS generation.

Mitochondria are the primary source of ROS generated after acute Zn²⁺ loads, but only account for a portion of the ROS induced by Ca²⁺ loading

Mitochondria can buffer cytosolic Ca²⁺ and Zn²⁺ rises via electrogenically driven entry through the MCU into the matrix [10,25,26]. To examine the role of mitochondrial uptake on the magnitude of the acute Ca²⁺ and Zn²⁺ rises, we used the mitochondrial protonophore, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) to dissipate the proton gradient across the inner membrane, thus eliminating the driving force for the uptake while inducing release of Ca²⁺ or Zn²⁺ ions already present in the mitochondria. To image the acute cytosolic Ca²⁺ or Zn²⁺ rises we used the low-affinity Ca²⁺ indicator, Fura-2FF (K_d ~ 25 µM; Teflab manual), and the low-affinity Zn²⁺ indicator, Newport Green (K_d ~ 1 µM; Life Technologies manual). When FCCP (1 µM) was added 5 min before acute 100 µM NMDA or 300 µM Zn²⁺/pyrithione exposures as above, the magnitude of the resultant cytosolic Ca²⁺ and Zn²⁺ rises were significantly increased, indicating that in control conditions (without FCCP) a substantial amount of either of these ions is taken up into mitochondria (Figure 2A,B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Effects of mitochondrial inhibition on cytosolic Ca²⁺ and Zn²⁺ rises and consequent ROS generation.

Fura-2FF-, Newport Green-, or HEt-loaded cultures were exposed to 100 μM NMDA (30 min) or 300 μM Zn²⁺/pyrithione (5 min) alone (red) or after pre-treatment with and in the presence of FCCP (1 µM) or rotenone (1 µM) (blue) as described (see Materials and Methods).

A: Time course of Ca²⁺ rises (assessed as Fura-2FF 340/380 nm fluorescence ratios) after NMDA exposure (left), and of Zn²⁺ rises (assessed as Newport Green F_x/F₀) after Zn²⁺/pyrithione exposure (right). Traces show mean ± SD values from 4 experiments.

B: Quantification of ion-sensitive indicator (Fura-2FF or Newport Green) fluorescence changes. Values show % rise of the of the Fura2FF ratio (340 nm/380 nm) after NMDA exposure, or of Newport Green fluorescence after Zn²⁺/pyrithione exposure from baseline until 30 min after onset of exposure. Values represent means (± SEM) of the 4 experiments; * indicates difference from control condition (p< 0.05) by 2-tailed t test.

C: Time course of ROS generation (assessed as HEt F_x/F₀), after exposures to NMDA (left) or to Zn²⁺/pyrithione (right). Dashed lines show linear extrapolation of baseline. Traces show mean ± SD values from 4 experiments.

D: Quantification of HEt ∆F changes. Values show F_x/F₀ increases, after subtraction of expected increase in the absence of ion load (baseline extrapolated value), 30 min after onset of the exposure. Values represent means (± SEM) of the 4 experiments; * indicates difference from control condition (p< 0.01) by 2-tailed t test.

https://doi.org/10.1371/journal.pone.0083347.g002

Next, to assess the contribution of mitochondria to the ROS generated in response to the Ca²⁺ or Zn²⁺ loads, HEt loaded cultures were pretreated with either FCCP (1 µM, 5 min), or with the complex I inhibitor, rotenone (1 µM, 10 min) prior to initiating the NMDA or Zn²⁺/pyrithione exposures. While both FCCP and rotenone by themselves induced a small increase in HEt fluorescence (see Figure 2C, during pretreatment episodes, prior to initiating the NMDA or Zn²⁺/pyrithione exposures), they both largely eliminated the acute Zn²⁺ triggered HEt signal, but only partially attenuated the Ca²⁺ triggered signal (Figure 2C,D).

Zn²⁺ dependent ROS generation is specifically linked to Zn²⁺ uptake into mitochondria via the MCU

The observation that cytosolic Zn²⁺ loads appear to trigger ROS generation of mitochondrial origin does not necessarily mean that the Zn²⁺ acts within the mitochondria. However, present observations that treatment with FCCP not only depolarizes mitochondria and diminishes their Zn²⁺ uptake but also markedly decreases the resultant ROS production, is highly suggestive of an intramitochondrial site of the Zn²⁺ effects.

As discussed above, despite its sensitivity, HEt fluorescence signals do not provide information about the site of the ROS generation, as the fluorescence of the oxidized product is greatest in the nucleus, where interactions with DNA amplifies its emission. In order to obtain more information on the origin of the ROS produced by Zn²⁺ loads, we made use of a different indicator, MitoTracker Red CM-H₂XRos (MTR-CMH₂), which has previously been found to provide sensitive and specific visualization of mitochondrial ROS generation [27]. MTR-CMH₂, a derivative of X-rosamine, accumulates in active polarized mitochondria, is oxidized by ROS into the fluorescent form, and is retained in mitochondria upon depolarization. Cultures were co-loaded with this indicator along with the fluorescent mitochondrial marker, MitoTracker Green (see Materials and Methods). For these studies a confocal microscope (Olympus IX70 with Bio-Rad Radiance 2000 Laser Scanning System) was used to aid in the visualization of signal emanating from individual mitochondria, and high-resolution images obtained before, at the end of a 5 min exposure to 300 µM Zn²⁺/10 µM pyrithione, and again 5 min after washout. Under these conditions, discrete foci of increasing red fluorescence were seen 2-3 min after the addition of the Zn²⁺, which co-localized with MitoTracker Green (Figure 3), further supporting a mitochondrial origin of the Zn²⁺-triggered ROS generation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Mitochondrial localization of Zn²⁺ induced ROS production.

Cultures were co-loaded with MitoTracker Green and MitoTracker Red CM-H₂XRos (MTR-CMH₂) and confocal images (1000x) obtained prior to Zn²⁺ treatment (t = 0 min), at the end of the 5 min 300 μM Zn²⁺/pyrithione exposure (t = 5 min), and 5 min after washout (t = 10 min). Note the marked increase in MTR-CMH₂ fluorescence that co-localizes with MitoTracker Green, strongly suggestive of a mitochondrial site of ROS production.

https://doi.org/10.1371/journal.pone.0083347.g003

Next, to further assess the dependence of Zn²⁺-triggered mitochondrial ROS generation upon specific Zn²⁺ entry into these organelles, we attempted to examine effects of pharmacological block of the Zn²⁺ entry on ROS generation. Zn²⁺ appears to enter mitochondria through the MCU, and in past studies, the MCU blocker ruthenium red (RR) has been found to inhibit Zn²⁺ uptake into isolated mitochondria [9,28], and antagonize downstream effects including mitochondrial depolarization and swelling [13,29]. As these prior studies were carried out in isolated mitochondria, however, we first sought to further confirm the ability of RR to penetrate the neurons and antagonize Zn²⁺ entry into mitochondria by assessing its effects on the magnitude of the acute Zn²⁺ rises. Cultures were loaded with Newport Green and exposed to 300 µM Zn²⁺/pyrithione as above, either with or without pre-exposure to RR (10 µM). As previously observed with FCCP, the RR pre-treatment resulted in a significant increase in the peak level of the cytosolic Zn²⁺ rise, strongly suggesting that RR antagonizes Zn²⁺ buffering by mitochondria (mean % increase ± SEM in 4 expts of 385±11.4 with Zn²⁺/pyrithione alone vs 452±5.6 with RR, p< 0.01; Figure 4). Next, we used HEt-loaded cultures to examine effects of RR on the Zn²⁺-triggered ROS generation, and found it to substantially attenuate the HEt ∆F, similarly to FCCP or rotenone. This provides new support for an intramitochondrial site from which Zn²⁺ exerts its effects to trigger the acute ROS generation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Blockade of the MCU increases cytosolic Zn²⁺ rises while attenuating consequent ROS production.

Newport Green or HEt-loaded cultures were exposed to 300 μM Zn²⁺/pyrithione for 5 min alone (red) or after pre-treatment with and in the presence of RR (10 µM) (blue) as described (see Materials and Methods).

A: Time course of Zn²⁺ rises (assessed as Newport Green F_x/F₀). Traces show mean ± SD values from 4 experiments.

B: Quantification of Newport Green fluorescence changes. Values show Newport Green fluorescence changes ( F_x/F₀) at the times indicated on the traces shown in A, above. Values represent means (± SEM) of the 4 experiments; * indicates difference from control condition (p< 0.01) by 2-tailed t test.

C: Time course of ROS generation (assessed as HEt F_x/F₀). Traces show mean ± SD values from 4 experiments.

D: Quantification of HEt fluorescence changes. Values show HEt fluorescence changes ( F_x/F₀) at the times indicated on the traces shown in C, above. Values represent means (± SEM) of the 4 experiments; * indicates difference from control condition (p< 0.01) by 2-tailed t test.

https://doi.org/10.1371/journal.pone.0083347.g004

Oxidative disruption of cytosolic Zn²⁺ buffering dramatically increases the potency through which Zn²⁺ loads impact mitochondria

The intense 300 µM Zn²⁺/pyrithione exposures, while useful to illustrate potential effects of very strong acute cytosolic Zn²⁺ rises, are of uncertain relevance to events that may occur in vivo in conditions like ischemia, as the magnitude of intracellular Zn²⁺ loading achieved may be greater than that readily induced upon mobilization of endogenous Zn²⁺ pools and stores. Neurons contain substantial cytosolic Zn²⁺ buffering capacity, much of which may be due to the presence of the metallothioneins or other Zn²⁺ binding metalloproteins [8]. Zn²⁺ binding to metallothioneins is destabilized under conditions of oxidative stress and acidosis [30,31], both of which occur prominently during ischemia. Indeed, metallothioneins appear able to mediate complex and disparate effects on Zn²⁺-dependent neurotoxicity depending on precise conditions, providing protection by buffering Zn²⁺ loads and serving as a source from which injurious Zn²⁺ can be mobilized under conditions of oxidative stress [19,32]. Indeed, it appears that oxidant-sensitive cytosolic Zn²⁺ pools and mitochondria can both sequester Zn²⁺, with release from one of them resulting in transfer of Zn²⁺ to the other [25]. The sulfhydryl-oxidizing agent, DTDP, has often been used to release Zn²⁺ from and prevent Zn²⁺ binding to cytosolic Zn²⁺ binding proteins, causing increases in free cytosolic Zn²⁺ levels in cultured neurons [33]; therefore, we used DTDP exposure to mimic the impaired cytosolic buffering that likely occurs during brain ischemia.

We next sought to determine whether impeding cytosolic Zn²⁺ buffering with DTDP would permit lower levels of Zn²⁺ loading to induce mitochondrial ROS generation. When cultures were exposed to 50 µM Zn²⁺/pyrithione, in the absence of DTDP, there was no HEt ∆F over baseline. DTDP (100 µM) alone also caused no HEt ∆F (data not shown). However, when the Zn²⁺ load was carried out in the presence of DTDP (before and during the Zn²⁺ exposure), a sharp HEt ∆F occurred, similar to that induced by 300 µM Zn²⁺/pyrithione. To confirm that this ROS generation was dependent upon mitochondrial Zn²⁺ uptake, identical exposures were carried out after exposure to RR, which almost completely blocked the HEt ∆F (Figure 5A,B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Under conditions of oxidative stress and disrupted Zn²⁺ buffering, lower levels of Zn²⁺ influx result in mitochondrial Zn²⁺ entry and ROS production.

A: Effect of DTDP on Zn²⁺-triggered ROS generation. HEt-loaded cultures were exposed to 50 μM Zn²⁺/pyrithione alone (red), in the presence of DTDP (100 µM, blue), or with both DTDP and RR (10 µM, black) as indicated. Note that the Zn²⁺/pyrithione exposure only induced ROS generation in the presence of DTDP, and that the ROS production was eliminated by RR. Traces show mean ± SD values from 4 experiments.

B: Quantification of HEt fluorescence changes. Values show HEt fluorescence changes (F_x/F₀) at the times indicated on the traces shown in A. Values represent means (± SEM) of the 4 experiments; * indicates difference from control condition (p< 0.01) by 2-tailed t test.

C, D: Disruption of cytosolic Zn²⁺ buffering by DTDP markedly increases cytosolic Zn²⁺ rises and uptake into mitochondria. Fluo-Zin3-loaded cultures were exposed to 1 μM Zn²⁺ with 90 mM K⁺ (“Zn²⁺/high-K⁺”, to trigger a low level of Zn²⁺ influx), to FCCP (1 µM), or to DTDP (100 µM) as indicated. In C, when the cultures were first exposed to Zn²⁺/high-K⁺ there was a very small FluoZin-3 ∆F, and subsequent FCCP exposure, to depolarize the mitochondria and release mitochondrially-sequestered Zn²⁺ into the cytosol, caused only a slight further increase. However, adding DTDP after the FCCP produced a large FluoZin-3 ∆F, suggesting that the Zn²⁺ entering during the Zn²⁺/high-K⁺ exposure had been largely buffered in the cytosol with little entering the mitochondria. In contrast (D), when the cultures were first exposed to DTDP (100 µM), there was a minimal FluoZin-3 ∆F, but when the DTDP exposure was followed by Zn²⁺/high-K⁺, the cytosolic ∆F was dramatically increased, and subsequent FCCP exposure resulted in a marked further ∆F, indicative of Zn²⁺ having accumulated within the mitochondria. Traces show mean ± SD values from 120 neurons from 4 experiments.

E: 50 μM Zn²⁺/high-K⁺ exposure cause mitochondrial ROS production only in the presence of DTDP. HEt-loaded cultures were exposed to 50 μM Zn²⁺/90 mM K⁺ alone (red) or after pre-treatment with and in the presence of DTDP alone (100 µM, blue) or with RR (10 µM, black) as indicated. Traces represent time course of HEt ∆F, normalized to baseline values (F_x/F₀) and show mean ± SD values from 4 experiments.

F: Quantification of HEt fluorescence changes. Values show HEt fluorescence changes (F_x/F₀) at the times indicated on the traces shown in E. Values represent means (± SEM) of the 4 experiments; * indicates difference from control condition (p< 0.01) by 2-tailed t test.

https://doi.org/10.1371/journal.pone.0083347.g005

Although pyrithione is useful for eliciting large Zn²⁺ loads into all neurons, a drawback of this approach is that it may be relatively non-physiologic, since pyrithione is membrane permeable and could directly facilitate Zn²⁺ entry into intracellular pools, including mitochondria, or Zn²⁺ exit from the cell. We therefore examined effects of triggering Zn²⁺ entry in a rapid and more physiologically relevant fashion, using brief exposures to high (90 mM) K⁺ buffer, which depolarizes neurons and causes opening of VGCC, through which Zn²⁺ can enter [34]. In prior studies, we found that 5 min high-K⁺ exposures with 300 µM Zn²⁺ failed to cause rapid HEt ∆F in most neurons (only causing such a response in the subpopulation of neurons that possessed large numbers of Ca²⁺ permeable AMPA channels, through which Zn²⁺ permeates more rapidly than through VGCC) [22].

To examine intracellular Zn²⁺ dynamics after inducing low sub-toxic Zn²⁺ entry through VGCC, we used the relatively high affinity Zn²⁺ indicator, FluoZin-3 (K_d ~ 15 nM) [35]. Exposure to 1 µM Zn²⁺ in the presence of high-K⁺ caused a very small FluoZin-3 ∆F, and subsequent FCCP exposure (to depolarize the mitochondria and release mitochondrially-sequestered Zn²⁺ into the cytosol) caused only a slight further increase. However, adding DTDP after the FCCP produced a large FluoZin-3 ∆F, suggesting that the Zn²⁺ had been largely buffered in the cytosol with little entering the mitochondria (Figure 5C). When the culture was exposed to DTDP (100 µM) before the 1 µM Zn²⁺/high-K⁺ exposure, a very slight ∆F occurred. However, when this DTDP exposure was followed by the 1 µM Zn²⁺/high-K⁺ exposure, the cytosolic ∆F dramatically increased. Furthermore, subsequent FCCP exposure resulted in a marked further ∆F, indicative of Zn²⁺ having accumulated within the mitochondria (Figure 5D). Thus, impairment of cytosolic Zn²⁺ buffering dramatically increases the amount of Zn²⁺ entering mitochondria upon relatively low level Zn²⁺ entry into the cell.

Finally, we carried out one additional assessment of ROS generation, using high-K⁺ to trigger Zn²⁺ entry as above, in the presence of 50 µM Zn²⁺. As with 50 µM Zn²⁺/pyrithione exposures, 50 µM Zn²⁺/high-K⁺ exposures by themselves induced no HEt ∆F, but a distinct HEt ∆F occurred when the cultures were pre-exposed to DTDP, which, as in the case of the 50 µM Zn²⁺/pyrithione/DTDP exposures, was substantially blocked by RR (Figure 5E,F). Thus, these data lend support to the idea that under conditions of impaired cytosolic Zn²⁺ buffering as likely occurs as a result of oxidative stress and acidosis during brain ischemia, relatively low levels of Zn²⁺ entry may result in Zn²⁺ passage through the MCU into mitochondria at levels sufficient to induce acute mitochondrial dysfunction with excess ROS generation.

Summary of principal findings

The present study seeks to compare acute ROS generation induced in cortical neurons by strong cytosolic Ca²⁺ loading with that induced by Zn²⁺ loading, events that are of likely relevance to neurodegeneration occurring in conditions of strong excitotoxic activation as occurs in ischemia or prolonged seizures. The rationale for the specific comparisons addressed by the studies (Ca²⁺ vs Zn²⁺; NOX vs mitochondria) reflect persistent uncertainties as to the key early events underlying excitotoxic neurodegeneration under differing circumstances. First, whereas early in vitro studies focused on Ca²⁺ overload as the key trigger of excitotoxic injury, subsequent studies in native tissue preparations (in vivo and slice models) found compelling evidence for important contributions of endogenous Zn²⁺ accumulation in triggering injury after ischemia or prolonged seizures. In addition, whereas imaging studies with “Ca²⁺ indicators” had long been interpreted as demonstrating Ca²⁺ transients and associated Ca²⁺ triggered effects, it is now known that Ca²⁺ indicators almost all bind and respond to Zn²⁺ with far higher affinity than Ca²⁺, raising the strong possibility that some effects previously attributed to Ca²⁺ might in fact be Zn²⁺-mediated [36]. Thus, as past in vitro studies have clearly demonstrated that cellular loading with either of these ions can trigger ROS generation [2,3,22], there is persistent uncertainty as to the likely dominant ionic “instigator” of excitotoxic injury in ischemia and prolonged seizures.

We find, in line with a recent study suggesting that Ca²⁺ causes rapid translocation and activation of NOX and consequent superoxide generation, that the Ca²⁺-dependent ROS signal was substantially (but not completely) NOX-dependent. In contrast, the Zn²⁺-triggered ROS signal was markedly diminished by mitochondrial inhibitors or by MCU inhibition, but appeared to be NOX-independent. As the strong Zn²⁺ exposures initially employed are of uncertain direct physiological relevance, we next examined effects of the sulfhydryl-oxidizing agent DTDP, which impairs Zn²⁺ binding to cytosolic Zn²⁺ buffering proteins, and thus reproduces deficiencies in cytosolic Zn²⁺ buffering likely caused by oxidative stress and acidosis during in vivo cerebral ischemia. Under these conditions, far lower Zn²⁺ loads resulted in substantial mitochondrial Zn²⁺ uptake and ROS generation, suggesting that during ischemia, physiological levels of Zn²⁺ accumulation might well enter mitochondria and induce deleterious effects on their function with release of ROS.

Ca²⁺ and excitotoxic neurodegeneration

The key finding that removal of Ca²⁺ from the media during brief glutamate or NMDA exposures to cultured neurons markedly diminished the injury that evolved over the subsequent hours has implicated Ca²⁺ entry through NMDA channels as a critical trigger of excitotoxic neurodegeneration. Consequently, it was predicted that NMDA receptor blockade would have potent protective effects against ischemic neurodegeneration [37], and a large number of studies have examined mechanisms of this Ca²⁺-dependent injury.

One target of the Ca²⁺ effects is mitochondria, which clearly can buffer large Ca²⁺ loads [26,38,39]. Furthermore, a number of studies concluded that rapid Ca²⁺ uptake into mitochondria caused ROS generation [1-3]. However, effects of Ca²⁺ on mitochondria are complex and mechanisms through which it increases mitochondrial ROS release are poorly understood and depend critically upon the paradigm employed. Mechanisms that have been suggested to be involved include stimulation of certain mitochondrial dehydrogenases, respiratory inhibition at complex I or III, and mitochondrial permeability transition pore induction, with consequent direct release of ROS, loss of cytochrome c, or loss of intramitochondrial antioxidant enzymes and glutathione [40-42].

Further studies have highlighted extramitochondrial routes through which cellular Ca²⁺ loading might trigger excitotoxic injury. One of these is the activation of nitric oxide synthetase (NOS), a free radical generating enzyme, which is linked intracellularly to the NMDA receptor complex. Interestingly, for this injury pathway, the critical factor does not appear to be the amount of Ca²⁺ entering the neuron, but rather the specific activation and passage of Ca²⁺ through NMDA channels [43]. Other studies have highlighted Ca²⁺-dependent NOX activation via a pathway involving phosphoinositide 3-kinase and protein kinase c-zeta activation [4,44]. Furthermore, NOX activation during in vivo ischemia, driven in part by high glucose [45], appears to contribute to degeneration after stroke [46]. Both NOS and NOX appear to mediate degeneration in part via a pathway in which oxidative DNA damage causes activation of the DNA repair enzyme, poly(ADP-ribose) polymerase-1 (PARP), resulting in depletion of NAD⁺ and ATP [47].

Zn²⁺ and excitotoxic neurodegeneration

In light of evidence that neuronal Zn²⁺ accumulation contributes to neurodegeneration in disease conditions, a large number of studies over the past two decades have examined relevant mechanisms. One key question concerned the source of the injurious neuronal Zn²⁺ accumulation. One source appears to be presynaptic vesicular Zn²⁺, which is co-released with glutamate from presynaptic terminals and enters post-synaptic neurons through routes, including VGCC and Ca²⁺-permeable AMPA channels [22,34,48]. However, it has become apparent that Zn²⁺ does not need to enter neurons from outside in order to injure them. There are large pools of Zn²⁺ within neurons bound to buffering proteins like metallothioneins. Zn²⁺ binding to these proteins is sensitive to both pH and oxidative stress [30,31], and these metabolic changes occur prominently in pathological conditions of ischemia or prolonged seizures, likely contributing to free Zn²⁺ accumulation in these conditions. Of note, even cultured neurons contain sufficient oxidant-releasable Zn²⁺ stores to induce injury, as exposure to the sulfhydryl-oxidizing agent DTDP induced both cytosolic Zn²⁺ rises and Zn²⁺-dependent injury [33], and it is apparent that these intracellular Zn²⁺ pools can contribute to neurodegeneration in vivo as well [32,49].-

Subsequent studies concerned mechanisms of Zn²⁺ toxicity. Interestingly, a number of studies indicated prominent parallels with Ca²⁺. Early studies found that Zn²⁺ exposures less intense than those used in this study to induce acute ROS generation still caused slowly evolving neurodegeneration with oxidative features [50], resulting in part from protein kinase C-dependent induction and activation of NOX [18]. Further studies indicated that this NOX activation, along with induction of NOS, resulted in activation of PARP, NAD⁺ and ATP depletion and cell death [17], much as in the case of Ca²⁺-dependent excitotoxicity. Thus, the role of NOX in more slowly developing Zn²⁺-dependent neurotoxicity is in contrast to present findings that did not show evidence of NOX contribution in acute Zn²⁺-triggered ROS production.

In other studies we correlated Zn²⁺ rises after entry through different routes with ROS generation, and found that rapid Zn²⁺ entry through Ca²⁺-permeable AMPA channels appeared to cause acute mitochondrial ROS generation that correlated with potent induction of toxicity [22]. Further studies showed that Zn²⁺ was taken up into mitochondria, causing depolarization and swelling in a MCU dependent fashion, and that it induced swelling of isolated mitochondria with far greater potency than Ca²⁺ [10,13,29].

Whereas Zn²⁺ was long ago found to inhibit electron transport [51], more recent studies have suggested mechanisms of its effects. Like Ca²⁺, it can enter mitochondria through the MCU [9], can antagonize the bc1 center of the electron transport chain [52], and appears, unlike Ca²⁺, to potently and irreversibly inhibit major enzymes of mitochondrial energy production and antioxidant defense [12,53], triggering induction of the mitochondrial permeability transition pore [12,13,54].

Possible relevance to disease

Whereas studies described above regarding Zn²⁺ effects on mitochondria all employ addition of exogenous Zn²⁺, recent studies provide strong support to the idea that mobilization of endogenous Zn²⁺ in pathological conditions yields levels sufficient to induce mitochondrial dysfunction. Specifically, recent studies suggest that endogenous Zn²⁺ contributes to mitochondrial dysfunction after in vivo ischemia [15,16]. Also, in hippocampal slices subjected to an oxygen glucose deprivation model of acute ischemia, we found that Zn²⁺ rises in CA1 neurons preceded and contributed to the occurrence of both irreversible mitochondrial depolarization and to the onset of a terminal Ca²⁺ deregulation [14].

In summary, as time passes, the “landscape” of excitotoxic contributions in conditions like ischemia is growing progressively complicated, and the early focus on NMDA receptor-mediated Ca²⁺ overload is clearly too limited. Contributions of Zn²⁺ may have been underestimated for a number of reasons, including the presumption that signals from “Ca²⁺ indicators” were all due to Ca²⁺, lack of awareness of the degree to which metabolic derangements in these disease conditions may promote intracellular Zn²⁺ accumulation, and the primary focus of disease studies on delayed events associated with “reperfusion” that paid inadequate attention to early and possibly highly impactful events occurring during acute ischemia. It is most likely that, the true evolution of “excitotoxic” injury after ischemia involves a host of events, some Ca²⁺- and some Zn²⁺- initiated, and some occurring acutely (perhaps including mitochondrial dysfunction), and, if the neurons survive the acute insult, others, likely including activation of NOX and downstream injury pathways, playing roles in the determination of cell fate over hours to days after “reperfusion”.

In light of recent evidence for early Zn²⁺-dependent mitochondrial dysfunction in ischemia, present observations lend support to the idea that under conditions of impaired cytosolic Zn²⁺ buffering, as occurs in ischemia, physiologically relevant levels of Zn²⁺ entry and release from cytosolic buffers likely result in Zn²⁺ passage into mitochondria via the MCU. Present findings, together with emerging indications of the high potency of Zn²⁺ effects on mitochondria, lead us to suggest the hypothesis that such early mitochondrial Zn²⁺ accumulation provides a potent assault on mitochondrial function with an accompanying burst of ROS generation. The ROS could further destabilize cellular Zn²⁺ buffering, resulting in a cascading positive feedback cycle leading to mitochondrial disruption. Understanding and prevention of such a cycle in the early stages of ischemia may markedly increase prospects for neuronal survival beyond reperfusion into the post-ischemic period.

Ethics Statement

This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of the University of California, Irvine (Protocol Number: 1997-1267). All procedures were terminal and all animals were deeply anesthetized before starting any procedures to minimize suffering.

Chemicals and Reagents

Hydroethidine (HEt) was purchased from Assay Biotech (Sunnyvale, CA). Newport Green, FluoZin-3, AM, MitoTracker Green, MitoTracker Red CM-H₂XRos, Pluronic F-127, tissue culture media and horse serum were purchased from Life Technologies (Grand Island, NY). Fura-2FF was obtained from Teflabs (Austin, TX). NMDA, rotenone, pyrithione, DTDP, ruthenium red and fetal bovine serum were purchased from Sigma-Aldrich (St. Louis, MO). FCCP was purchased from Tocris Bioscience (Ellisville, MO) and apocynin was obtained from Acros Organics (Morris Plains, NJ). All other chemicals and reagents were purchased from common commercial sources.

Cortical Cultures

Cultures were prepared generally as previously described [21]. Briefly, dissociated mixed neocortical cell suspensions were prepared from 15-16 day CD-1 mouse embryos and plated on previously established astrocytic monolayers in glass-bottomed dishes, in media consisting of Eagle's Minimum Essential Medium (EMEM; Earle's salts prepared glutamine-free) supplemented with 10% heat-inactivated horse serum, 10% fetal bovine serum, 2 mM glutamine, and 25 mM glucose, and kept in a 37°C/5% CO2 incubator. After 3-4 DIV (days in vitro), non-neuronal cell division was halted by exposure to 10 μM cytosine arabinoside for 24 h, and the cultures were switched to an identical maintenance medium lacking fetal serum. The same procedure was used to prepare glial cultures, except that tissue was obtained from early postnatal (1–3 d) mice, media was supplemented with epidermal growth factor (10 ng/ml), and cell suspensions were plated directly on the polylysine- and laminin-coated coverslips.

Imaging Studies

10–13 DIV cultures were mounted on the stage of a Nikon Diaphot inverted microscope equipped with a 75 W xenon lamp, a computer-controlled filter wheel, a 40 X 1.3 numerical aperture epifluorescence oil-immersion objective along with a green FITC fluorescence cube (Ex: 480 nm, dichroic: 505 nm, Em: 535 nm), a red TRITC fluorescence cube (Ex: 540 nm, dichroic: 565 nm, Em: 605 nm), and an ultraviolet Fura-2 fluorescence cube (Ex: NA, dichroic: 400 nm, Em: 510 nm). Emitted signals were acquired with a Sensys Photometrics intensified charge-coupled device camera and digitized by using MetaFluor Version 7.0 software (Molecular Devices LLC, Sunnyvale, CA). Background fluorescence was subtracted from images at the beginning of each experiment and all experiments were carried out at room temperature in HEPES-buffered medium (HSS), consisting of (in mM): 120 NaCl, 5.4 KCl, 0.8 MgCl₂, 20 Hepes, 15 glucose, 1.8 CaCl₂, 10 NaOH, pH 7.4. Camera gain and exposure were adjusted to give baseline maximal fluorescence levels of 200-300 arbitrary units of a maximal 12-bit signal output of 4,096 for all fluorophores.

ROS generation was assessed by examining fluorescence changes in cells loaded with the oxidation sensitive dye HEt (Ex: 510-560; Em: >590), using the red fluorescence cube. For the NMDA studies, cultures were loaded in the dark at room temperature with 5 μM HEt in HSS for 45 min and then washed into a static bath of HSS containing 5 μM HEt for the 10 min baseline recording. For the Zn²⁺ experiments, treatments were the same except that all exposures were in Ca²⁺-free HSS. HEt fluorescence measurements for each cell (F_x) were normalized to the average fluorescence intensity for that cell during a 10 min period of baseline recording (F₀).

For pharmacological treatments, ruthenium red (RR, 10 μM), was added during the HEt loading, pyrithione (10 μM) was added at the onset of baseline recording and maintained through the Zn²⁺ exposure, and other drugs (500 μM apocynin, 1 μM FCCP, 1 μM rotenone, or 100 μM DTDP) were added prior to the NMDA (100 μM, 30 min) or Zn²⁺ (with 10 μM pyrithione or 90 mM K⁺, 5 min) exposures, and maintained throughout the duration of the recordings, as indicated. For pyruvate treatment, cells were switched to glucose-free HCSS containing 15 mM pyruvate during the HEt loading and maintained in this media throughout the imaging run. Control experiments showed that in the absence of NMDA or Zn²⁺ treatment, HEt fluorescence slowly increased in a linear fashion over the 60 min recordings. We thus obtained linear regressions of the 10 min baseline for each set of data, enabling comparison of endpoint fluorescence with expected values in the absence of Ca²⁺ or Zn²⁺ loading.

For [Zn²⁺]_i and [Ca²⁺]_i imaging, cultures were loaded in the dark with 5 μM of either the relatively low-affinity Zn²⁺ indicator, Newport Green diacetate (K_d ~ 1 µM, Ex: 490 nm, Em: 530 nm), the high affinity Zn²⁺ indicator, FluoZin-3 (K_d ~ 15 nM, Ex: 494 nm, Em: 516 nm), or the low-affinity Ca²⁺ indicator, Fura-2FF (K_d ~ 25 µM, Ex: 340-380 nm, Em: 510 nm) in HSS containing 0.2 % Pluronic F-127 and 1.5 % dimethlyl sulfoxide (DMSO) for 30 min at 25°C, then washed into HSS for Ca²⁺ imaging or Ca²⁺-free HSS for Zn²⁺ imaging, and kept in the dark for an additional 30 min. For Newport Green and FluoZin-3, the green fluorescence cube was used, while for Fura-2FF, excitation was via the filter wheel (340 and 380 nm filters), and emission via the Fura-2 filter cube. Experiments were carried out in a static bath of HSS or Ca²⁺-free HSS and drugs were added to the buffer for pre-treatments as indicated. For intracellular Zn²⁺, fluorescence measurements for each cell (F_x) were normalized to the average fluorescence intensity for that cell during the first 10 min of the experiment (F₀). Changes in intracellular Ca²⁺ were assessed as the ratio of the emission intensity (at 510 nm) when excited at 340 nm to that upon excitation at 380 nm (“340/380 emission ratio”).

Confocal microscopy of live dissociated cell culture was performed using an Olympus IX70 microscope equipped with a Bio-Rad Radiance 2000 laser scanning system. Images were collected with a 100 X oil-immersion objective using Zeiss LaserSharp 2000 software. Briefly, cells were loaded with 200 nM MitoTracker green (Ex: 490 nm, Em: 516 nm) and 500 nM MitoTracker Red CM-H₂XRos (MTR-CMH₂) (Ex: 579 nm, Em: 599 nm) for 30 min at 37°C in the dark, and then switched to 0 Ca²⁺ HCSS and 10 µM pyrithione for the imaging experiment using the confocal. An image was acquired prior to Zn²⁺ treatment, immediately after a 5 min exposure to 300 µM Zn²⁺/pyrithione, and again 5 min after washout. Images were scanned sequentially with a 488 nm line Argon laser for MitoTracker Green followed by a 543 nm line Green HeNe laser for MTR-CMH₂.

Statistical Analysis

Traces (with the exception of Figure 5 C, D) all represent the mean values from 4 independent experiments. In each experimental condition in each experiment a field containing many healthy appearing neurons was selected for imaging, and ~ 25-35 healthy-appearing neurons were prospectively chosen and marked for fluorescence measurements; after the exposure the fluorescence changes in these 25-35 cells was averaged to yield a single value constituting one independent replication of one condition in the experiment. In addition, we always carried out paired examinations of control or treatment exposures using sister cultures from the same preparation and studied on the same day. All compiled data are derived from at least 4 fully independent experiments, and statistics are based upon the number of independent experimental repetitions rather than the total numbers of cells studied for each condition in order to provide the most rigorous and conservative test of differences. Error bars on traces show means ± SD for the 4 experiments (or for 120 neurons from 4 dishes studied in 5 C, D). The bar graphs display the mean values of 4 experiments (± SEM), and significance values are based on n=4 for each condition. All significance values are based upon 2-tailed t-tests.