Glutamate mobilizes [Zn²⁺]_i through Ca²⁺-dependent reactive oxygen species accumulation

Abstract

Liberation of zinc from intracellular stores contributes to oxidant-induced neuronal injury. However, little is known regarding how endogenous oxidant systems regulate intracellular free zinc ([Zn²⁺]_i). Here we simultaneously imaged [Ca²⁺]_i and [Zn²⁺]_i to study acute [Zn²⁺]_i changes in cultured rat forebrain neurons after glutamate receptor activation. Neurons were loaded with fura-2FF and FluoZin-3 to follow [Ca²⁺]_i and [Zn²⁺]_i, respectively. Neurons treated with glutamate (100 μM) for ten minutes gave large Ca²⁺ responses that did not recover after termination of the glutamate stimulus. Glutamate also increased [Zn²⁺]_i, however glutamate-induced [Zn²⁺]_i changes were completely dependent on Ca²⁺ entry, appeared to arise entirely from internal stores, and were substantially reduced by co-application of the membrane-permeant chelator TPEN during the glutamate treatment. Pharmacological maneuvers revealed that a number of endogenous oxidant producing systems, including nitric oxide synthase, phospholipase A₂, and mitochondria all contributed to glutamate-induced [Zn²⁺]_i changes. We found no evidence that mitochondria buffered [Zn²⁺]_i during acute glutamate receptor activation. We conclude that glutamate-induced [Zn²⁺]_i transients are caused in part by [Ca²⁺]_i -induced reactive oxygen species that arises from both cytosolic and mitochondrial sources.

Introduction

Accumulation of extracellular glutamate and consequent over activation of glutamate receptors is thought to be a key event in neuronal death resulting from ischemia, seizure, and force trauma. At the level of the single cell, the main features of glutamate death include cytosolic calcium overload, mitochondrial dysfunction, and accumulation of reactive oxygen species (ROS) (Nicholls & Budd 2000). Since Olney first described the process of “excitotoxicity” (Olney 1969), a large body of research has greatly advanced our knowledge of the cellular and molecular mechanisms behind glutamate-induced neuronal death. At this time however our understanding has not translated into effective treatment of brain injury (O’Collins et al. 2006).

Because free zinc accumulates in dying neurons after excitotoxic stimulation, and because agents that reduce [Zn²⁺]_i can provide neuroprotection, there is interest in zinc as a clinically relevant target (Frederickson et al. 2000). However, there are large gaps in our understanding of how Zn²⁺ collects in neurons, from where it arises in the first place, and the mechanism by which it contributes to excitotoxic death. While Zn²⁺ released from vesicles might conceivably contribute to ischemic injury, mice lacking vesicular Zn²⁺ still show Zn²⁺-dependent neuronal injury, suggesting that most toxic Zn²⁺ arises from intracellular proteins (Lee et al. 2000). In vitro experiments demonstrated that oxidants can mobilize Zn²⁺ from purified metalloproteins (Knapp & Klann 2000, Maret & Vallee 1998), and a number of different groups using different cell culture models showed [Zn²⁺]_i accumulation in response to various exogenous oxidants (Aizenman et al. 2000, St Croix et al. 2002, Tatsumi & Fliss 1994). In contrast, the degree to which endogenous oxidant production modulates Zn²⁺ in normal or disease states is largely unknown. Downstream mechanisms suggested to account for Zn²⁺ toxicity include mitochondrial dysfunction, perturbed kinase activity, and, interestingly, ROS production (Malaiyandi et al. 2005a, McLaughlin et al. 2001, Noh & Koh 2000, Sensi et al. 1999b). Thus it appears that ROS generation and Zn²⁺ homeostasis are interdependent.

Many studies have exploited dual dye paradigms that monitor [Ca²⁺]_i with ROS or mitochondrial membrane potential in excitotoxicity, allowing a more precise picture of temporal and cause-and-effect relationships between these parameters (Vergun et al. 2001, Vesce et al. 2004). In contrast, live-cell detection of [Zn²⁺]_i has become feasible only through the recent development of Zn²⁺-sensitive and -selective fluorophores that are able to distinguish Zn²⁺ from Ca²⁺ (Gee et al. 2002), and to our knowledge no one has used dual parameter imaging to study [Zn²⁺]_i in injury paradigms. Previously we described a new technique that combined the low affinity, ratiometric Ca²⁺ indicator fura-2FF with the high affinity, Zn²⁺-selective indicator FluoZin-3 for the simultaneous detection of [Ca²⁺]_i and [Zn²⁺]_i (Devinney et al. 2005). The present report uses this approach to follow [Zn²⁺]_i in neurons stimulated with high levels of extracellular glutamate. We found that glutamate-induced [Zn²⁺]_i changes were completely dependent on Ca²⁺ influx, and that ROS inhibitors significantly reduced [Zn²⁺]_i transients. Specifically, we conclude that ROS derived from nitric oxide synthase (NOS), phospholipase A₂ (PLA₂) and mitochondria all contribute to glutamate-induced, [Ca²⁺]_i-dependent [Zn²⁺]_i transients. A previous abstract included portions of this work (Dineley et al. 2004).

Experimental Procedures

Cell Culture

All procedures using animals were in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Primary cultures of embryonic rat forebrain neurons were prepared using a modification of a previously described procedure (Brocard et al. 2001) and were prepared as follows: Embryonic day 17 pups were taken from pregnant Sprague-Dawley rats euthanized with CO₂. Pups were decapitated and heads were placed in a sterile culture dish on ice containing approximately 25mL of calcium, magnesium free medium (CMFM) (containing 5.4mM KCl, 116mM NaCl, 26mM NaHCO₃, 5.5mM glucose, 0.002% Phenol Red, and MEM amino acids, pH 7.4 and stored at 4°C). Whole brains were dissected from the head and placed in a second dish of CMFM on ice. Using forceps and lance, forebrains were separated from the midbrain (discarded) and meninges were removed. Forebrain tissue was placed in 2 mL of CMFM containing 0.04% trypsin, and incubated for 30min at 37° C, after which this mixture was triturated in a 15mL conical tube approximately 10 times with a disposable 10mL tissue culture pipette, then brought to a volume of 15mL by adding plating medium of Dulbecco’s modified eagle medium (DMEM) supplemented with 10% fetal bovine serum and penicillin/streptomycin. The resulting cell suspension was filtered through a 70μm nylon cell strainer into a 50mL conical tube. Cell density was counted with a hemocytometer using trypan blue to exclude dead cells. Suspension volume was increased with plating medium to a cell density of 1 × 10⁶ cells/mL. 1.5mL of cell suspension was added to each well which contained a glass coverslip. Typically, 24 pups yielded about thirty 6-well plates

Thirty-one mm glass coverslips were prepared by boiling in water for 60 minutes, placed in six-well plates, then incubated with poly-D-lysine (0.4 mg/mL in H₂O). After 2 hours, poly-D-lysine was aspirated out of wells, and the plates were stored at 4°C, but returned to room temperature before plating of cells. After plating of cells, 6-well plates were moved to a standard tissue culture incubator, and maintained at 5% CO₂ and 37°C for five hours, after which plating medium was replaced with 1.5 mL of neurobasal medium supplemented with N2 and penicillin-streptomycin (100 UmL⁻¹ and 100 μgmL⁻¹, respectively). After 4 days in vitro, 0.4mL of old media was removed from each well and 0.5mL of N2-NB media added. N2-NB media was made by adding 5mL of N2 supplement and penicillin/streptomycin to 500mL of Neurobasal Media (Gibco) Feedings were repeated on DIV 8 and 11 using Neurobasal media supplemented with 10 mL B27 instead of N2.

Live cell microscopy

Neurons were loaded with cell permeant (AM) dye derivatives for FluoZin-3, fura-2FF or fura-2 by incubating coverslips in 1mL of HBSS containing 5 mg mL⁻¹ of bovine serum albumin and 5μM dye for 30 min in the dark at 37°C. After loading, coverslips were rinsed thoroughly, mounted in a recording chamber, and continuously superfused at 10mL min⁻¹ at 37°C with HEPES-Buffered Salt Solution (HBSS) containing, in mM, 150 NaCl, 5 KCl, 0.9 MgSO₄, 1.4 CaCl₂, 20 HEPES, 5.5 glucose, and adjusted to pH 7.4 with NaOH. The membrane-permeant heavy metal chelator TPEN (from 25mM stock in DMSO) was used to decrease [Zn²⁺]_i. In Zn²⁺-free experiments, 1mM disodium EDTA and 1mM CaCl₂ was added to HBSS. This chelates extracellular background Zn²⁺, leaving Ca²⁺ and Mg²⁺ relatively unaffected (Koh et al. 1996). For Ca²⁺-free experiments, Ca²⁺ was omitted from and EGTA (20 μM) added to otherwise normal HBSS. For each coverslip, the responses of approximately 10–20 neurons were sampled.

The PC-based imaging system used in these experiments consisted of the following components: a Nikon Diaphot 300 microscope equipped with a 40x oil immersion objective, a CCD camera (Model# C4742-95; Hamamatsu Photonics, Hamamatsu City, Japan), a monochromator-driven xenon light source (ASI, Eugene, OR), and a software package (SimplePCI, Compix, Cranberry, PA). For time-lapse imaging, fura-2FF or fura-2 was excited alternately at 340 and 380 nm, and FluoZin-3 at 490 nm. Light was modified with a 515nm dichromatic mirror and a 525±12.5 nm bandpass emission filter. All optical filters were purchased from Chroma Technology (Brattleboro, VT). Fura-2FF and fura-2 values were expressed as a ratio of background subtracted F₃₄₀/F₃₈₀. FluoZin-3 fluorescence was background subtracted, normalized to starting values, and expressed as F/F₀.

Reagents

Fura-2FF AM was purchased from TEFLABS; FluoZin-3 AM, fura-2 AM, and N, N, N′, N′-tetrakis(2-pyridalmethyl)ethylenediamine (TPEN) from Molecular Probes (Eugene, OR); all other reagents were from Sigma unless noted otherwise.

Statistics

All live-cell experiments were repeated on at least three different coverslips taken from three separate culture preparations, and 10–20 neurons were sampled from each coverslip. N for an experiment corresponds to the total number of individual neurons sampled from all coverslips in a given condition. Appropriate statistical analyses were performed using Graph Pad Prism, version 4.01 (San Diego, CA).

Results

FluoZin-3 detects oxidant labile zinc in primary cortical neurons

We and others previously showed that elevated [Zn²⁺]_i in neurons can be monitored with ion-sensitive fluorophores such fura-2, mag-fura-2, mag-fura-5 and Newport Green (Canzoniero et al. 1997, Canzoniero et al. 1999, Cheng & Reynolds 1998, Dineley et al. 2000, Sensi et al. 1997). Although these dyes and others can be used for monitoring [Zn²⁺]_i, none is ideal because of poor Zn²⁺ selectivity in the case of the fura family, or relatively low sensitivity in the case of Newport Green. In this study we used a relatively new Zn²⁺ indicator, FluoZin-3, which is (i) sensitive to low levels of Zn²⁺, (ii) selective for Zn²⁺ and insensitive to other metals including Ca²⁺ and Mg²⁺, (iii) stable in the presence of oxidants, and (iv) relatively unaffected by pH changes in the pathophysiological range (Devinney et al. 2005, Gee et al. 2002, Kay 2004). As the glutamate stimulation paradigm used here increases [Ca²⁺]_i, decreases pH, generates oxidants, and may change [Mg²⁺]_i (Brocard et al. 1993, Hartley & Dubinsky 1993, Reynolds & Hastings 1995), FluoZin-3 appeared exceptionally suited for measuring relatively small Zn²⁺ loads associated with glutamate stimulation and oxidative stress. In figure 1, we tested its usefulness for measuring oxidant-induced mobilization of [Zn²⁺]_i in rat primary cortical neurons treated with different oxidants: H₂O₂, 2,2′-dithiodipyridine (DTDP) and 2-methyl-4-isothiazolin-3-one (MIT). Panels A through C show example traces from 8–10 neurons on a single coverslip exposed to either 1 mM H₂O₂, 30 μM MIT, or 30 μM DTDP. In response to each oxidant, the FluoZin-3 signal increased throughout the 10-minute treatment, remained elevated during a 5-minute washout period, but decreased rapidly when TPEN (5 μM) was added. Responses to concentrations of MIT and DTDP higher than those presented here were particularly robust, often producing intense fluorescence that saturated the camera. H₂O₂ (1 mM) produced smaller and apparently more varied responses between cells, perhaps reflecting an intercellular difference in the capability of neurons to inactivate peroxide. Figure 1D shows summary data for the different oxidants, which collectively demonstrate the utility of FluoZin-3 for detecting oxidant-induced changes in [Zn²⁺]_i. We also noted that FluoZin-3 produced a homogeneous staining pattern in resting neurons, suggesting that [Zn²⁺]_i was distributed relatively evenly across different cellular compartments.

Figure 1.

FluoZin-3 detects oxidant-mobilized Zn²⁺ in cultured primary cortical neurons. Panels A–C show representative traces from neurons loaded with FluoZin-3 and exposed to H₂O₂ (1 mM), MIT (100 μM), or DTDP (30 μM) for 10 min. A washout period (5 min) followed the oxidant treatment, then TPEN (5 μM) was added at the time indicated by the arrow. Panel D shows average fluorescence change (F/F₀ as mean ± S.E.) for the three oxidants, each at two concentrations. Also, similar experiments for solvents DMSO and ethanol (0.3% for each) are shown. Each experiment was performed on at least three different culture preparations. The total number of neurons, n, for each condition is as follows: H₂O₂ (100 μM), 134; H₂O₂ (1 mM), 195; DTDP (10 μM), 130; DTDP (30 μM), 115; MIT (30 μM), 90; MIT (100 μM), 78; DMSO, 76; EtOH, 69.

In considering FluoZin-3 responsiveness, it is useful to compare responses here with those from our previous work using mag-fura-2 and Newport green, where DTDP (100 μM) caused relatively small changes in the signal of these dyes (Aizenman et al. 2000). In contrast, here we observed very large changes in FluoZin-3 fluorescence at considerably lower concentrations of DTDP. We adjusted basal FluoZin-3 fluorescence values to 2 to 3 times above background (background was about 30 arbitrary fluorescence units on an 8-bit scale). Thus, a strong FluoZin-3 response averaged across an individual cell corresponded to maximal F/F₀ values of about 5 to 6. These oxidant trials helped establish collection parameters for subsequent glutamate experiments in which FluoZin-3 typically did not rise above normalized values of 4, indicating that we stayed well within the reporting range of the dye.

Our previously established method for simultaneous monitoring of [Ca²⁺]_i and [Zn²⁺]_i with fura-2FF and FluoZin-3 (Devinney et al. 2005) was used here to investigate the relationship between [Ca²⁺]_i and [Zn²⁺]_i changes during intense glutamate exposure. In figure 2, neurons loaded with both fura-2FF and FluoZin-3 were stimulated with glutamate (100 μM) and glycine (10 μM) for 10 minutes for maximal activation of N-methyl-D-aspartate (NMDA) receptors. In response, fura-2FF ratio values typically rose to between 2 and 3 (Fig 2A). Fura-2FF ratios usually plateaued quickly and did not rise further despite continued glutamate application, suggesting that [Ca²⁺]_i reached concentrations that saturated fura-2FF. Consistent with a [Ca²⁺]_i-dominated signal, the membrane-permeant, heavy metal chelator TPEN (5 μM) did not change fura-2FF ratios. In these same neurons, glutamate stimulation also mobilized [Zn²⁺]_i, as reflected by an increased FluoZin-3 signal (Fig. 2B). FluoZin-3 responses differed from fura-2FF in that they were less uniform, sometimes declined during the glutamate stimulus, and, most notably, were reversed by TPEN. We then exploited TPEN in order to further resolve the [Ca²⁺]_i and [Zn²⁺]_i signals. Neurons were treated with a low concentration of TPEN (2 μM) 3 min before, during, and 2 min after a 10 min stimulation with glutamate and glycine (Fig 2C and D). This maneuver largely prevented [Zn²⁺]_i changes, as shown by the diminished FluoZin-3 response (Fig. 2D). In contrast, TPEN co-application had only a small effect on [Ca²⁺]_i changes detected by fura-2FF (Fig. 2C). Mean traces of glutamate with and without TPEN co-application are shown in Fig. 2E and F. Summary data is also presented in figure 5. Together these data show that fura-2FF and FluoZin-3 can be used to monitor simultaneously glutamate-induced [Ca²⁺]_i and [Zn²⁺]_i changes in the same cell, and that glutamate stimulation is associated with elevated [Zn²⁺ ]_i.

Figure 2.

Simultaneous detection of glutamate-induced [Ca²⁺]_i and [Zn²⁺]_i in cultured primary neurons. Neurons loaded with both fura-2FF and FluoZin-3 were exposed to glutamate (100 μM) and glycine (10 μM) for 10 min. A and B. Fura-2FF and FluoZin-3 traces, respectively, from individual neurons treated with glutamate + glycine, followed by 5 min wash, then TPEN (5 μM) was added as indicated. C and D. Fura-2FF and FluoZin-3 signals, respectively, from neurons stimulated with glutamate + glycine and co-treated with a low concentration of TPEN (2 μM) before, during, and after the glutamate stimulus. Panel E shows mean traces from a group of neurons treated with glutamate + glycine. Panel F shows mean traces from a group of neurons treated with glutamate + glycine with TPEN. Fura-2FF is indicated by solid trace and corresponds to left y-axis; FluoZin-3 is indicated by dashed trace and corresponds to right y-axis. Mean traces are derived from 20–30 individual neurons from the same coverslip, and each experiment was performed on at least three different culture preparations for a total of >40 individual neurons for each condition. See figure 5 for summary data.

Figure 5.

Summary data showing that glutamate-induced [Zn²⁺]_i changes are reduced by Ca²⁺ removal, antioxidants, depolarization of mitochondria, or TPEN. In contrast, [Ca²⁺]_i changes were robustly altered only when Ca²⁺ was removed from the extracellular buffer. Bars represent mean±S.E. n for each condition is as follows: Glu, 371; Glu + L-NAME, 53; Glu + Aristol. Acid, 112; Glu + FCCP, 190; Glu + TPEN, 176; Glu, Zn²⁺ free, 81; Glu, Ca²⁺/Zn²⁺ free, 47. * or # denotes statistical difference (p< 0.05) compared to glutamate-induced fura-2FF or FluoZin-3 values, respectively. Data were analyzed by ANOVA followed by Bonferroni post hoc.

One potential weakness of this approach is saturation of fura-2FF by calcium. Because our goal was to study [Zn²⁺]_i changes during intense glutamate stimulation, we required a low-affinity Ca²⁺ indicator. Fura-2FF has a low Ca²⁺ affinity (Kd ~ 6 μM; (Hyrc et al. 2000)) and also features spectral compatibility with FluoZin-3. Mag-fura-2, which also has low Ca²⁺ sensitivity (Kd ~ 50 μM) and spectral compatibility, might be useful as well (Raju et al. 1989). As its name implies however, mag-fura-2 is also sensitive to magnesium, which probably changes during intense glutamate stimulation (Brocard et al. 1993).

In a study of hippocampal slices subjected to oxygen glucose deprivation followed by reperfusion, it was reported that large concentrations of TPEN reversed nearly all signal from the putative Ca²⁺ probe Calcium Green-1 (Stork & Li 2006). From this, the investigators suggested that Zn²⁺ constitutes the majority of cation accumulation in excitotoxicity paradigms. Our results differ considerably: in our hands TPEN eliminates only the [Zn²⁺]_i signal with little effect on [Ca²⁺]_i. These contradictions might be explained by the excessive TPEN treatments used in that study (100 μM, sustained for many minutes). At such high levels, TPEN may be working as a low affinity Ca²⁺ chelator (Arslan et al. 1985). Our results show that TPEN used in much smaller amounts achieves relatively selective chelation of [Zn²⁺]_i without altering [Ca²⁺]_i.

To be certain that fura-2FF was not interfering with FluoZin-3 detection of [Zn²⁺]_i, we performed control experiments where neurons loaded with FluoZin-3 only were stimulated with glutamate. Maximum FluoZin-3 values were similar in the presence or absence of fura-2FF, therefore fura-2FF apparently does not interfere importantly with FluoZin-3 detection of [Zn²⁺]_i in these conditions. (Mean + S.E. for maximum FluoZin-3 fluorescence change was 3.15 ± 0.099 (n=360) with both probes, 3.25 ± 0.332 (n=56) for FluoZin-3 alone. These values are not statistically different (p = 0.736) by two-tailed student’s t-test.)

We then tested whether small [Ca²⁺]_i loads change [Zn²⁺]_i. For this purpose, we exploited the propensity of our neurons to undergo spontaneous [Ca²⁺]_i oscillations (Figure 3). Pharmacological characterization shows that these oscillations depend on or are influenced by a myriad of receptors and ion channels, but do not require exogenous stimulation (K.E.D. & I.J.R. unpublished observations). Because these [Ca²⁺]_i changes are relatively small, we used the high affinity Ca²⁺ indicator fura-2 in combination with FluoZin-3. Figure 3 shows that small [Ca²⁺]_i changes do not elevate [Zn²⁺]_i. These data also confirm that FluoZin-3 does not disturb fundamental neuronal Ca²⁺ regulation. Interestingly, there is a slow but steady climb in FluoZin-3 fluorescence over time, which was also observed in the solvent controls in figure 1. This suggests that FluoZin-3 may be leaching Zn²⁺ from internal sites.

Figure 3.

Small fluctuations in [Ca²⁺]_i do not increase [Zn²⁺]_i. Mean traces taken from neurons loaded with both FluoZin-3 and the high affinity Ca²⁺ indicator fura-2 show that small, spontaneous [Ca²⁺]_i elevations do not increase [Zn²⁺]_i. Traces are a mean of 10 individual neurons from a single coverslip, and similar results were obtained from neurons from three different culture preparations.

We next tested what factors contribute to [Zn²⁺]_i changes. Zn²⁺ is a ubiquitous contaminant and it is therefore conceivable that adventitious extracellular Zn²⁺ could contribute to intracellular changes. By supplementing our HBSS with CaEDTA, which chelates extracellular Zn²⁺ without changing extracellular Ca²⁺ or Mg²⁺, we ruled out a contribution from external Zn²⁺ (See summary data of Figure 5). In contrast, removal of extracellular Ca²⁺ completely abolished FluoZin-3 changes, showing that [Ca²⁺]_i accumulation is critical for [Zn²⁺]_i mobilization after intense glutamate exposure (Mean traces in figure 4B, summary data in figure 5).

Figure 4.

Zn²⁺ mobilization requires elevated [Ca²⁺]_i and is partially dependent on ROS accumulation. Mean fura-2FF (solid) and FluoZin-3 (dashed) traces from neurons exposed to glutamate + glycine only (Panel A, reproduced from panel 2E for ease of comparison) or glutamate + glycine in Ca²⁺-free buffer (B), or with aristolochic acid (100 μM) (C), L-NAME (1 mM) (D), or FCCP (0.75 μM) (E). FCCP was applied for 5 min during the glutamate + glycine stimulus, while L-NAME and aristolochic acid were applied for 3 min before, during, and 3 min after glutamate + glycine. Experiments were concluded with a 5 min treatment with TPEN (5 μM), as indicated. Traces were averaged from 20–30 neurons from a single coverslip, and similar results were obtained from neurons from at least three different culture dates. See figure 5 for summary data.

Oxidative stress is an established consequence of glutamate excitotoxicity. While it is certain that ROS accumulation after glutamate exposure depends on [Ca²⁺]_i accumulation, the degree to which ROS is responsible for key events in the excitotoxic sequelae, such as delayed [Ca²⁺]_i deregulation and mitochondrial failure, is controversial (Vesce et al. 2004). Two previous studies (Bossy-Wetzel et al. 2004, Sensi et al. 1999a) suggest that oxidant accumulation resulting from NMDA receptor activation mobilizes intracellular zinc. These studies are difficult to interpret in this context because extracellular zinc was not chelated, so that it is difficult to exclude the possibility that any zinc changes are due to the well-established Zn²⁺ influx that occurs with glutamate receptor activation (Cheng & Reynolds 1998, Koh & Choi 1994, Sensi et al. 1997).

We used our dual-dye approach to examine how inhibition of endogenous ROS mechanisms affects glutamate-induced [Ca²⁺]_i and [Zn²⁺]_i loads. Specifically, we used L-NAME (1 mM) to inhibit NOS, aristolochic acid (100 μM) to block phospholipase A₂, and FCCP (0.75 μM) to reduce mitochondrial calcium accumulation that promotes ROS generation (Reynolds & Hastings 1995). Figures 4 and 5 show that Ca²⁺ changes are relatively unaffected by maneuvers that prevent ROS buildup. This is consistent with the work of Nicholls’ group which suggests that ROS accumulation is a consequence and not a cause of glutamate-induced [Ca²⁺]_i dysregulation (Vesce et al. 2004). In contrast to [Ca²⁺]_i, all three maneuvers (L-NAME, aristolochic acid, and FCCP) substantially reduced FluoZin-3 changes. These results suggest that [Zn²⁺]_i mobilization after glutamate exposure is caused in part by ROS accumulation, which itself is caused by [Ca²⁺]_i accumulation. These experiments also indicate that several different sources of ROS are activated by glutamate. Figure 5 provides summary data corresponding to the mean traces in Figure 4. Cuvette-based experiments with a spectrofluorophotometer confirmed that neither FCCP, L-NAME, nor aristolochic acid chelated Zn²⁺ by themselves (data not shown).

With respect to [Ca²⁺]_i measurements, it is important to note that statistical analysis (One-way ANOVA followed by Bonferroni post hoc) revealed differences in fura-2FF changes observed between glutamate and some of the experimental conditions (Figure 5). As is apparent from the summary bar graphs, the differences are quite small, and because we are using fura-2FF at near-saturating Ca²⁺ levels, construing any meaningful differences between them would be inappropriate. In contrast, FluoZin-3 differences observed between various conditions are robust and obvious, and much more likely to represent differences of biological and not just statistical import.

Discussion

In this study we demonstrated that glutamate receptor activation results in mobilization of intracellular Zn²⁺ by a mechanism dependent on Ca²⁺ entry. Using a series of inhibitors of mechanisms involved in ROS accumulation, we showed that Zn²⁺ mobilization is likely the result of glutamate-induced oxidative stress. Thus, these results raise the possibility that glutamate-induced elevation of [Zn²⁺]_i contributes to the neurotoxic effects of glutamate, and demonstrate the feasibility of simultaneous detection of stimulus-induced changes in [Ca²⁺]_i and [Zn²⁺]_i.

Glutamate-induced [Zn²⁺]_i mobilization was critically dependent on Ca²⁺ entry, because [Zn²⁺]_i changes were completely abolished in nominally Ca²⁺-free buffer. Conversely, removal of extracellular Zn²⁺ with CaEDTA had no effect, therefore we concluded that adventitious extracellular Zn²⁺ makes no contribution to the intracellular change. Given the Zn²⁺ permeability of numerous ion channels and transporters and the ubiquity of Zn²⁺ contamination of buffer solutions (Kay 2004) this is an important issue that was not addressed in prior studies (Bossy-Wetzel et al. 2004, Sensi et al. 1999a). Unlike the relatively large and uniform [Ca²⁺]_i changes, [Zn²⁺]_i changes showed more heterogeneity, even amongst different neurons on the same coverslip, both in the magnitude and duration of the response (Fig. 2B). The variety of [Zn²⁺]_i responses reported here is interestingly reminiscent of heterogeneous oxidant production in cortical neurons after activation of NMDA receptors (Reynolds & Hastings 1995). In further contrast to [Ca²⁺]_i, [Zn²⁺]_i returned, if only partially, to pre-stimulus levels often before the glutamate stimulus was concluded. How cultured neurons export or sequester Zn²⁺ within this time frame is not well-understood.

Our data suggest that several oxidant producing systems contribute to glutamate-induced [Zn²⁺]_i mobilization. The PLA₂ inhibitor aristolochic acid substantially reduced [Zn²⁺]_i changes. PLA₂ metabolizes arachidonic acid from membrane phospholipids in cortical neurons treated with glutamate (Taylor & Hewett 2002), and was among the first proposed mechanisms for glutamate-induced oxidant production (Lafon-Cazal et al. 1993). Arachidonic acid metabolism proceeds via cyclo-oxygenase (COX) activity that generates hydrogen peroxide, so that liberation of arachidonic acid will result in an oxidative burden. Our prior studies in cultured forebrain neurons showed that COX inhibitors partially abrogate the glutamate-induced oxidant signal (Reynolds & Hastings 1995). A more recent study (Vesce et al. 2004) showed that cytoplasmic oxidant production, which was aristolochic acid-sensitive, was an early event following NMDA receptor activation in cerebellar granule neurons. The glutamate-induced [Zn²⁺]_i elevation observed here was also sensitive to NOS inhibition. Ostensibly, this is consistent with prior observations of NO-mediated Zn²⁺ release (St Croix et al. 2002). However, we were unable to raise [Zn²⁺]_i using NO donors such as PAPA NONOate (up to 200μM; data not shown). Yet it could be that NOS activation does not yield NO, because oxygen (rather than arginine) can accept electrons for NOS when either arginine or the tetrahydrobiopterin co-factor is inadequate. The result is superoxide instead of NO (Heinzel et al. 1992). The cultures used in this study clearly contain NOS and can oxidize arginine (Stout et al. 1998), but this does not establish NO as the prominent oxidant produced after Ca²⁺-dependent activation of NOS 1 or 3. Thus, it remains possible that either superoxide or hydrogen peroxide is the main species of endogenous oxidant responsible for Zn²⁺ release in neurons.

We used FCCP to abolish mitochondrial membrane potential, which decreased glutamate-induced [Zn²⁺]_i. With respect to [Ca²⁺]_i burden in neurons, FCCP is neuroprotective because it prevents Ca²⁺ accumulation in mitochondria and decreases accumulation of mitochondrially-derived ROS (Pivovarova et al. 2004, Reynolds & Hastings 1995, Stout et al. 1998). The mechanisms linking mitochondrial calcium accumulation to ROS generation are not clearly established, although we reported that Ca²⁺ potentiates ROS release from brain mitochondria mildly inhibited at complex I (Votyakova & Reynolds 2005), while others showed that oxidative phosphorylation is impaired in mitochondria prepared from glutamate-treated cultured neurons (Kushnareva et al. 2005). Thus, ROS accumulation from mitochondria, secondary to glutamate exposure, may be a reasonable mechanism to account for the FCCP-sensitive [Zn²⁺]_i signal.

Mitochondria also buffer [Ca²⁺]_i through uptake and release, and it has been suggested that they operate similarly in the regulation of [Zn²⁺]_i. However, in cultured neurons, mitochondrial depolarization with FCCP greatly exaggerates glutamate-induced cytosolic [Ca²⁺]_i loads, because mitochondrial Ca²⁺ uptake through the uniporter depends on the mitochondrial transmembrane potential (Δψ_m) (Stout et al. 1998). In this study, glutamate alone produced [Ca²⁺]_i that appeared to saturate the dye; therefore co-application of FCCP accelerated the Ca²⁺ rise but did not meaningfully increase the maximum fura-2FF ratios (see Fig. 5D). In the case of Zn²⁺ however, FCCP substantially reduced glutamate-induced [Zn²⁺]_i, suggesting that mitochondrial sequestering is not an important [Zn²⁺]_i buffering mechanism during intense glutamate exposure. This is consistent with our previous work in isolated neural mitochondria showing that matrix accumulation of Zn²⁺ is negligible in the presence of high Ca²⁺ (Malaiyandi et al. 2005b). Using cultured neurons, others have suggested that mitochondria accumulate Zn²⁺ hours after a [Zn²⁺]_i preloading protocol (Sensi et al. 2002). Thus it may be that Zn²⁺ accumulates in mitochondria when cytosolic Zn²⁺ persists at high levels. However, the homogeneous FluoZin-3 pattern throughout resting or stimulated neurons, and the failure of FCCP to exaggerate the [Zn²⁺]_i transient after glutamate, suggests that mitochondria do not collect appreciable quantities of Zn²⁺ at rest or during intense, acute glutamate stimulation. Instead, it seems that mitochondria influence Zn²⁺ homeostasis most importantly through ROS generation, and that FCCP retards glutamate-induced Zn²⁺ mobilization by limiting mitochondrially derived ROS.

A number of reports have suggested that Zn²⁺ causes ROS accumulation. For instance, high extracellular Zn²⁺ leads to superoxide accumulation through PKC-dependent activation of NADPH oxidase (Noh & Koh 2000), and Zn²⁺ can augment ROS production in isolated brain mitochondria (Dineley et al. 2005). While it is clear that elevated [Zn²⁺]_i can promote oxidant accumulation, it is equally clear that oxidative stress elevates [Zn²⁺]_i in a variety of cell types. For example, the sulfhydryl oxidants DTDP and MIT liberate toxic [Zn²⁺]_i from intraneuronal stores (Aizenman et al. 2000, Du et al. 2002), and other reports show Zn²⁺ mobilization in response to NO, HOCl, ONOO⁻ (Sah & Schwartz-Bloom 1999, St Croix et al. 2002, Tatsumi & Fliss 1994). Here we show that endogenous ROS production, resultant of glutamate exposure, increases [Zn²⁺]_i in cortical neurons. Thus is appears that the mobilization of Zn²⁺ from intracellular sites is a general response to oxidative stress that may contribute to neuronal injury. While this work does not consider toxicity per se, previous reports showed that metal chelation can improve neuronal survival after oxidant-induced Zn²⁺ burden (Aizenman et al. 2000). Therefore it seems reasonable to propose that [Zn²⁺]_i loads observed here contribute to glutamate-induced neuronal death, perhaps by inciting further ROS accumulation in the manner of a vicious cycle; however it is possible that increased [Zn²⁺]_i is merely an early marker of severe excitotoxicity occurring downstream of large [Ca²⁺]_i loads.

Determining exactly how much Zn²⁺ is present in resting or stimulated cells is an area of active research and considerable debate. One recent report suggested resting [Zn²⁺]_i of 5–10 picomolar in PC 12 and CHO cells (Bozym et al. 2006). Using HT-29 cells, others reported [Zn²⁺]_i in resting cells at ~600pM, which approximately doubled upon differentiation (Krezel & Maret 2006). In neurons depolarized with high K⁺ in the presence of elevated extracellular Zn²⁺, [Zn²⁺]_i climbed to levels estimated in the hundreds of nanomolar over the course of several minutes (Canzoniero et al. 1999). It is difficult to know the accuracy of such estimations however, because calculating precise Zn²⁺ concentrations from intracellular fluorescence is fraught with difficulties (Dineley et al. 2002). The largest problem arises from the scarcity of Zn²⁺ relative to an abundant, high-affinity probe, and the upshot is that depletion of free ion under conditions of relatively high dye concentrations renders inaccurate the classic calibration approach (Grynkiewicz et al. 1985). When intracellular probe concentration is known, it is possible to correct for ion depletion (Kenakin 1993). However determining probe concentration is itself a non-trivial matter. Thus we prefer the more conservative approach of reporting fluorescence values without conversion to [ion]_i.

As shown here and in other reports, FluoZin-3 is particularly useful for monitoring oxidant-induced mobilization of [Zn²⁺]_i because it (i) is sensitive to small [Zn²⁺]_i loads, (ii) is insensitive to Ca²⁺, and (iii) does not fluoresce in response to iron or copper, two other abundant and redox labile heavy metals (Devinney et al. 2005, Gee et al. 2002, Kay 2003) The sulfhydryl oxidizing agents MIT and DTDP produced large FluoZin-3 responses, while the ‘biological’ oxidant H₂O₂ was comparatively smaller. To measure [Ca²⁺]_i and [Zn²⁺]_i simultaneously, we used our previously established method that combines fura-2FF with FluoZin-3 (Devinney et al. 2005). This approach confirms the well-established role for Ca²⁺ in excitotoxicity while revealing the release of Zn²⁺. Because fura-2FF has a high affinity for zinc (K_D ~ 12 nM; (Hyrc et al. 2000)), the two dye combination could lead to an underreporting of [Zn²⁺]_i by FluoZin-3, and/or spurious reporting of [Ca²⁺]_i due to [Zn²⁺]_i contamination of the fura-2FF signal. However, our prior in vitro work showed that, in the presence of high [Ca²⁺]_i and [Zn²⁺]_i, fura-2FF responds largely to Ca²⁺, while FluoZin-3 responds to Zn²⁺ (Devinney et al. 2005). Moreover, data here demonstrate that chelating Zn²⁺ with TPEN markedly reduced FluoZin-3 changes with little effect on fura-2FF, and Figure 3 shows that FluoZin-3 responds to glutamate similarly in the presence or absence fura-2FF. Therefore it seems reasonable to conclude that, in the presence of high [Ca²⁺]_i and a high-affinity Zn²⁺ chelator, fura-2FF interaction with [Zn²⁺]_i is relatively minor. Finally, FluoZin-3 altered neither large [Ca²⁺]_i changes measured with fura-2FF, nor small ones measured by fura-2; thus there is no evidence that FluoZin-3 has any significant impact on neuronal calcium homeostasis.