Measurement of Intracellular Free Zinc in Living Cortical Neurons: Routes of Entry

Abstract

We used the ratioable fluorescent dye mag-fura-5 to measure intracellular free Zn²⁺([Zn²⁺]_i) in cultured neocortical neurons exposed to neurotoxic concentrations of Zn²⁺ in concert with depolarization or glutamate receptor activation and identified four routes of Zn²⁺ entry. Neurons exposed to extracellular Zn²⁺ plus high K⁺ responded with a peak cell body signal corresponding to a [Zn²⁺]_i of 35–45 nm. This increase in [Zn²⁺]_i was attenuated by concurrent addition of Gd³⁺, verapamil, ω-conotoxin GVIA, or nimodipine, consistent with Zn²⁺ entry through voltage-gated Ca²⁺channels. Furthermore, under conditions favoring reverse operation of the Na⁺–Ca²⁺ exchanger, Zn²⁺ application induced a slow increase in [Zn²⁺]_i and outward whole-cell current sensitive to benzamil–amiloride. Thus, a second route of Zn²⁺ entry into neurons may be via transporter-mediated exchange with intracellular Na⁺. Both NMDA and kainate also induced rapid increases in neuronal [Zn²⁺]_i. The NMDA-induced increase was only partly sensitive to Gd³⁺ or to removal of extracellular Na⁺, consistent with a third route of entry directly through NMDA receptor-gated channels. The kainate-induced increase was highly sensitive to Gd³⁺ or Na⁺removal in most neurons but insensitive in a minority subpopulation (“cobalt-positive cells”), suggesting that a fourth route of neuronal Zn²⁺ entry is through the Ca²⁺-permeable channels gated by certain subtypes of AMPA or kainate receptors.

Zinc is required by all cells, playing a critical role in the control of gene transcription and metalloenzyme function (Vallee and Falchuk, 1993; Berg and Shi, 1996). In addition, Zn²⁺ likely serves as a mediator of cell–cell signaling in the CNS (Frederickson, 1989; Frederickson and Moncrieff, 1994). The mammalian brain contains a high concentration of histochemically reactive Zn²⁺, primarily localized to glutamatergic terminals (Danscher, 1984; Perez-Clausell and Danscher, 1985) and released with neuronal activity (Howell et al., 1984). Endogenously released Zn²⁺ may reach 100 μm concentrations in the extracellular space (Assaf and Chung, 1984) and alter the behavior of multiple channels and receptors (Harrison and Gibbons 1994; Smart et al., 1994), in particular inhibiting NMDA receptors (Peters et al., 1987; Westbrook and Mayer, 1987; Christine and Choi, 1990).

Furthermore, intense exposure to Zn²⁺ can be neurotoxic, killing cortical neurons after several minutes (Yokoyama et al., 1986). This neurotoxicity appears also to be mediated by Zn²⁺ influx, in large part through voltage-gated Ca²⁺ channels and also through NMDA receptor-gated channels (Koh and Choi, 1994) and Ca²⁺-permeable AMPA/kainate receptor-gated channels (Yin and Weiss, 1995). Pathophysiological relevance has been suggested by observations that brain Zn²⁺ translocates from presynaptic terminals into the postsynaptic cell bodies of neurons dying after prolonged seizures (Sloviter, 1985; Fredrickson et al., 1989) or transient global ischemia (Tonder et al., 1990; Koh et al., 1996), together with the finding that intraventricular administration of Ca-EDTA can attenuate global ischemia-induced neuronal death (Koh et al., 1996).

Although it would be useful to understand how neurons regulate intracellular cytosolic free Zn²⁺ concentrations ([Zn²⁺]_i), at present, the tools available for measuring dynamic changes in [Zn²⁺]_i are limited. The Zn²⁺-selective fluorescent dye 6-methoxy-8-p-toluene sulfonamide quinoline (TSQ) has provided valuable qualitative information about the location of chelatable Zn²⁺ in cells and tissues (Frederickson et al., 1987; Weiss et al., 1993), but limitations of solubility as well as possible complex interactions with membrane-bound Zn²⁺ have precluded quantitative measurement of [Zn²⁺]_i (Andrews et al., 1995). Recently, other sulfonamide derivatives of quinoline have been developed: (1) TFLZn, which has been used to detect [Zn²⁺] in synaptic terminals in hippocampal slices (Budde et al., 1997), in which high levels of Zn²⁺are present (Frederickson, 1989); and (2) zinquin, which has been used to detect [Zn²⁺]_i in thymocytes, pancreatic islet cells, and hepatocytes (Zalewski et al., 1993, 1994;Brand and Kleineke, 1996). However, significant limitations are posed by the low affinity of these indicators for Zn²⁺and, at least in the case of zinquin, by uneven intracellular distribution, perhaps reflecting sequestration. Other recently identified candidate Zn²⁺ indicators, Newport green and APTRA-BTC (Haugland, 1996), have promising specificity but low affinity.

Another approach to measuring [Zn²⁺]_iis to use fura-2-based, Ca²⁺-sensitive ratioable fluorescent dyes. Fura-2 binds transition metals such as Zn²⁺ with much higher affinity than Ca²⁺, producing a similar shift in emitted fluorescence (Grynkiewicz et al., 1985; Simons, 1993). Fura-2 itself has been used to monitor [Zn²⁺]_i in chromaffin cells (Vega et al., 1994), in myocytes and transfected cell lines (Atar et al., 1995), in bovine liver cells (Hechtenberg and Beyersmann, 1993), and in synaptosomes (Denny and Atchison, 1994), although confounding interference from changes in [Ca²⁺]_i is difficult to exclude.

In the present study, we explored the use of the related dye mag-fura-5 for real-time measurements of [Zn²⁺]_iin living cortical neurons. Simons (1993) reported that a similar analog, mag-fura-2, could be used to measure Zn²⁺ in solution over the 1–100 nm range, even in the presence of moderate Ca²⁺, reflecting the higher affinity of the dye for Zn²⁺ (K_d = 20 nm) over Ca²⁺ or Mg²⁺. We selected mag-fura-5 over mag-fura-2 because the former has additionally reduced affinity for Ca²⁺ or Mg²⁺ compared with mag-fura-2 (Haugland, 1996).

MATERIALS AND METHODS

Cell culture. Mixed cultures were prepared as described previously (Rose et al., 1993). Briefly, dissociated neocortices from embryo mice at 15–16 d gestation were plated in glass bottom 35 mm dishes (MatTek, Ashland, MA) at a density of three hemispheres/10 ml of plating medium containing 5% fetal bovine serum and 5% horse serum. Cultures were fed once a week with growth medium (5% horse serum). Videomicroscopy experiments were performed at 12–17 d in vitro.

[Zn²⁺]_i measurement.To monitor [Zn²⁺]_i, cells were loaded with mag-fura-5 by incubation with mag-fura-5 AM at the concentration of 3 μm for 10 min in a HEPES-buffered solution containing (in mm): NaCl 120, KCl 5.4, MgCl₂ 0.8, HEPES 20, glucose 15, CaCl₂ 1.8, and NaOH 10, pH 7.4, at room temperature. Cells were washed and incubated for an additional 20 min in the HEPES-buffered solution. After loading, neurons were washed twice with the same solution, but lacking Mg²⁺ and Ca²⁺. Na⁺-free solutions were made by substituting NaCl and NaOH with equimolar amounts ofN-methyl-d-glucamine (NMDG) and adjusting the pH to 7.4 with HCl.

All experiments were performed at room temperature under constant perfusion (2 ml/min) on the stage of a Nikon Diaphot inverted microscope equipped with a 75 W Xenon lamp and a Nikon 40×, 1.3 numerical aperture, epifluorescence oil immersion objective. Light was passed through 340 and 380 nm excitation filters mounted on a filter wheel with a computer-controlled filter driver. To avoid photobleaching of the cells under observation, the fluorescence intensity was attenuated with neutral density filters. The light was then reflected off a dichroic mirror (450 nm) and passed through a 510 nm emission filter.

Images were acquired with an intensified CCD camera (Quantex) and digitized using a Metafluor 2.5 software (Universal Imaging, West Chester, PA). Background fluorescence was subtracted from both (340 and 380 nm) images at the beginning of each experiment.

To determine the K_d of mag-fura-5 for Zn²⁺, we recorded excitation spectra of mag-fura-5 free acid (1 μm) solution containing (in mm): KCl 100, EGTA 0.1, and 4-morphoinepropanesulfonic acid 10, pH 7, in a spectrofluorimeter (LS 50; Perkin-Elmer, Norwalk, CT) at room temperature. Various ZnCl₂ concentrations (from 0 to 0.2 mm) were added to the calibrating solution via a micrometer syringe (Hamilton, Reno, NV).

To calculate free Zn²⁺ concentration, we used Microsoft Excel (Solver) and the following equation: [Zn²⁺] = K_d[ZnEGTA]/[EGTA], where a dissociation constant (K_d) of EGTA for Zn²⁺ of 7.09 nm at pH 7 was taken and modified by 0.02 log unit per 0.01 increase in pH (Grynkiewiecz et al., 1985). K_d of mag-fura-5 for Zn²⁺ was obtained by plotting the log [(F −F_min)/(F_max− F)] versus the log of free [Zn²⁺] (Grynkiewicz et al., 1985), whereF, F_min, andF_max are the fluorescence intensity values obtained with an excitation of 380 nm at each [Zn²⁺], at [Zn²⁺] = 0, and at saturating [Zn²⁺], respectively. This plot gives an x-intercept of 7.56, so K_d= 27 nm (Fig. 1).

Fig. 1.

K_d of mag-fura-5 for Zn²⁺. Hill plot analysis derived from plotting the log [(F −F_min)/(F_max −F)] versus the log of free [Zn²⁺]. Values were obtained by using a mag-fura-5-free acid solution at various [Zn²⁺], as described in Materials and Methods. Each pointrepresents the mean ± SEM from four different experiments.

This K_d value of 27 nm was used in the ratio formula: [Zn²⁺] =K_d · S_f2/S_b2 · [(R−R_min)/(R_max− R)], where R is the observed 340/380 fluorescence ratio (Grynkiewicz et al., 1985),R_min is the 340/380 fluorescence ratio value determined in mag-fura-5-loaded cortical neurons exposed to 0 Zn²⁺ and 100 μmN,N′,N′-tetrakis (2-pyridylmethyl) ethylenediamine (TPEN), andR_max is 340/380 fluorescence ratio value determined in mag-fura-5-loaded cortical neurons exposed to 20 μm of the selective Zn²⁺ ionophore Na⁺-pyrithione in the presence of 1 mmZn²⁺. S_f2 indicates the fluorescence intensity at 380 atR_min; S_b2indicates the fluorescence intensity at 380 atR_max.

Zinquin-ethyl ester (excitation, 380 nm; emission, 510 nm), APTRA-BTC (excitation, 380/440; emission, 510 nm), and Newport green diacetate (excitation, 495 nm; emission, 530 nm) were tested by using the same microscope setting equipped with the appropriate set of filters.

Whole-cell recording of Na⁺–Ca²⁺ exchanger current.A 35 mm culture dish was used as a recording chamber on the stage of an inverted microscope (Nikon). Neurons of 15–20 μm diameter, with phase-bright cell bodies, were chosen for recordings. Neuronal identity has been confirmed previously by Nissl staining and electrophysiological characterization (Choi et al., 1987). Patch electrodes had a tip resistance of 10–15 MΩ (fire-polished). Voltage clamp was achieved by using an EPC-7 amplifier (List Electronic, Darmstadt, Germany). Series resistance compensation and capacitance compensation were routinely applied. Whole-cell current was sampled digitally at 500 μsec. The current signals were filtered by a 3 kHz, three-pole Bessel filter. Current traces were displayed and stored on a Macintosh computer (Quatra 950; Apple, Santa Clara Valley, CA) using the data acquisition and analysis program PULSE (HEKA Electronik, Pfalz, Germany). To test Zn²⁺-activated membrane current, solutions were delivered to the selected neuron body via pressure ejection from a delivery pipette of 100 μm internal diameter placed ∼100 μm from the cell body, using the DAD-12 superfusion system (Adams & List, Westbury, NY). The extracellular solution contained (in mm): NMDG 120, HEPES 10, andd-glucose 10, with or without ZnCl₂ 1. The pipette solution contained (in mm): NaCl 60, Cs-acetate 70, CaCl₂ 4.2, Mg₂-ATP 2, BAPTA 10, and HEPES 10. In addition, TTX (0.5 μm), TEA (5 mm), nifedipine (10 μm), and ouabain (20 μm) were included regularly in extracellular solutions to block Na⁺ channels, K⁺ channels, dihydropyridine-sensitive Ca²⁺ channels, and Na⁺/K⁺ ATPase, respectively. Solution pH was adjusted to 7.3 by adding HCl or NaOH as needed. Neuronal input resistance was measured at −70 and −10 mV, using an external solution containing (in mm): NaCl 124, KCl 5, TTX 0.002, HEPES 20, and glucose 15, with or without MgCl₂ 0.8 and CaCl₂ 2; and an internal solution containing (in mm): KCl 140, MgCl₂ 2, HEPES 10, and BAPTA 0.3.

Cobalt staining. After Zn²⁺-imaging experiments, Co²⁺ uptake was performed as described previously (Turetsky et al., 1994). Cells were exposed to 100 μm kainate plus 5 mm CoCl₂ in uptake buffer containing (in mm): sucrose 139, NaCl 57.5, KCl 5, MgCl₂ 2, CaCl₂ 1, glucose 12, and HEPES 10, pH 7.6, for 30 min at room temperature. Cultures were then washed with uptake buffer plus 3 mm EDTA to remove extracellular Co²⁺ and incubated in 0.12% (NH₄)₂S for 5 min to precipitate intracellular Co²⁺. After three washes in uptake buffer, cells were fixed in 4% paraformaldehyde for 30 min. For silver enhancement, cultures were washed three times in development buffer (in mm: sucrose 292, hydroquinone 15.5, and citric acid 42) and incubated in 0.1% AgNO₃ in development buffer at 50°C on a slide warmer. This solution was changed at 15 min intervals, and enhancement was monitored periodically by microscopic observation. When enhancement was complete (usually after 45–50 min), the reaction was terminated by washing three times in warm development buffer. Once stained, fields were relocated by an independent observer lacking knowledge of the fluorescence results.

Statistics. Comparisons were obtained by ANOVA followed by Student–Newman–Keuls’ test (p < 0.05).

Materials. APTRA-BTC-AM, mag-fura-5 AM, fura-2, Newport green diacetate, and 4-bromo (br)-A23187 were obtained from Molecular Probes (Eugene, OR). Na⁺-pyrithione, Ca-EDTA, ZnCl₂, and TPEN were obtained from Sigma (St. Louis, MO). Zinquin-ethyl ester was obtained from Dojindo (Kumamoto, Japan).

RESULTS

Measurements of intracellular free Zn²⁺ in cortical neurons

Before settling on mag-fura-5, we explored the possible use of zinquin, Newport green, APTRA-BTC, or fura-2 in measuring [Zn²⁺]_i in cultured cortical neurons.

After 1 hr loading at 37°C with 25 μm zinquin-ethyl ester, cortical cultures exposed to 300 μm kainate plus 300 μm Zn²⁺ did not show a detectable increase in neuronal somatic zinquin fluorescence. Even exposure to the Zn²⁺ ionophore 20 μmNa⁺-pyrithione plus 1 mmZn²⁺ did not yield a zinquin signal, suggesting that the dye was not adequately loaded or cleaved. As a control, zinquin-loaded HEK 293 cells exposed to the same ionophore plus Zn²⁺ did exhibit a large increase in fluorescence, albeit distributed unevenly within the cells (data not shown).

Two other Zn²⁺-selective dyes, Newport green and APTRA-BTC, were loaded successfully into cortical neurons and exhibited strong responses to Zn²⁺ ionophore treatment. However, consistent with their low affinities for Zn²⁺ (see above), minimal responses were detectable after the physiological (probably pathophysiological) stimulus, 500 μm kainate plus 300 μmZn²⁺. We also examined the use of fura-2 to detect intracellular Zn²⁺. However, beside the high sensitivity of fura-2 for Ca²⁺, fura-2 also appeared too sensitive to Zn²⁺ for our purposes (K_d for Zn²⁺, ∼2 nm) (Grynkiewicz et al., 1985). We found that fura-2-loaded neurons exhibited a near saturating response to exposure to 300 μm kainate plus 300 μmZn²⁺ (in the absence of extracellular Ca²⁺).

As described above, we loaded mixed neuronal and glial cortical cell cultures with mag-fura-5 AM at a concentration of 3 μm for 10 min at room temperature. Dye concentrations between 1 and 10 μm were explored, 1 μmbeing too little for reliable measurements, and 10 μmyielding reduced responses compared with 3 or 5 μm, suggestive of buffering (data not shown). Because neuronal cell bodies sat well above the underlying glial layer, it was possible to detect the 340/380 fluorescence ratio originating in these cell bodies with minimal interference from underlying glia. At rest, the neuronal somatic 340/380 fluorescence ratio was ∼1 and insensitive to application of the membrane-permeant selective Zn²⁺chelator TPEN (100 μm) (Arslan et al., 1985) (Fig.2, R_min). Application of the selective Zn²⁺ ionophore 20 μm Na⁺-pyrithione (Zalewski et al., 1993), together with 1 mm Zn²⁺ in HEPES buffer lacking Ca²⁺ and Mg²⁺, produced an increase in the neuronal somatic 340/380 fluorescence ratio to 5.6 (Fig. 2, R_max). If extracellular Zn²⁺ was omitted, no signal was seen (data not shown). Reapplication of 100 μm TPEN produced a rapid fall in the 340/380 fluorescence signal back to baseline (Fig. 2).

Fig. 2.

Calibration of mag-fura-5 detection of [Zn²⁺]_i in cortical neurons. Cortical neurons were loaded with mag-fura-5 and bathed in medium lacking Ca²⁺ or Mg²⁺. At the indicated times, 20 μm Na⁺-pyrithione plus 1 mm Zn²⁺ or 100 μm TPEN were added by bath perfusion, and the 340/380 nm fluorescence ratio was determined (mean ± SEM, n = 24 cells). Experiment is representative of three.

We considered it unlikely that this somatic mag-fura-5 signal was confounded by Ca²⁺ detection, given the absence of extracellular Ca²⁺ and the low affinity of the dye for Ca²⁺ (see above). Indeed, previous studies with mag-fura-2 or mag-fura-5 did not detect neuronal somatic Ca²⁺ after glutamate exposure or electrical stimulation, even with extracellular Ca²⁺ present (Brocard et al., 1993; Petrozzino et al., 1995). A more serious possible confounding effect could derive from changes in intracellular Mg²⁺, which are in the millimolar range and may be increased by glutamate receptor activation even in the absence of extracellular Mg²⁺ (Brocard et al., 1993). However, consistent with the studies of Illner et al. (1992), even exposure to 30 mm Mg²⁺ in the presence of an ionophore, br-A23187, produced only a small change in the neuronal somatic 340/380 fluorescence ratio compared with the large change produced by Zn²⁺ in the same cells (Fig.3). As a control, we demonstrated that the large Zn²⁺ signal, but not the small Mg²⁺ signal, was sensitive to application of 50 μm TPEN (Fig. 3).

Fig. 3.

Comparison of mag-fura-5 sensitivity to Mg²⁺ and Zn²⁺. A, Exposure to br-A23187 (10 μm) for 60 min in the presence of 30 mm Mg²⁺ (open symbols) or 1 mm Zn²⁺(solid symbols). TPEN (50 μm) was added at the indicated time to quench the Zn²⁺ signal.B, Inset showing the same Mg²⁺ signal as in A, but at higher vertical gain; TPEN has no effect. Each point is the mean ± SEM of at least 30 cells. Experiment is representative of three.

Depolarization and Na⁺-dependent pathways involved in [Zn²⁺]_i elevation

Next we used mag-fura-5 to measure in cortical neurons accumulating Zn²⁺ under depolarizing conditions that might have pathophysiological relevance. Mixed cortical cell cultures were loaded with mag-fura-5 as above and placed into HEPES buffer lacking Ca²⁺ and Mg²⁺ (no chelators were added). The removal of external Ca²⁺and Mg²⁺ produced no significant change (<20% change in the mean. p > 0.05 with n = 4 cells per condition) in the input resistance of the neurons either in resting condition (−70 mV) or after depolarization (−10 mV). Application of 30–90 mm KCl plus 300 μmZn²⁺ for 15 sec elicited a K⁺concentration-dependent and persistent (lasting >60 min) increase in [Zn²⁺]_i to 48 ± 1.02 nm that was terminated by application of 50 μm TPEN, whereas application of 90 mm KCl alone produced no response (Fig.4A,B). The response to 90 mm KCl could be blocked completely by concurrent application of the nonselective voltage-gated Ca²⁺ channel blockers gadolinium (Gd³⁺, 10 μm) and verapamil (100 μm) and partially blocked by the selective blockers ω-conotoxin GVIA (100 nm) and nimodipine (1 μm) (Fig. 4C). The observed reduction of the fluorescence produced by Gd³⁺ was not attributable to quenching, because the signal induced by exposure to Na⁺-pyrithione was not affected by 10 μm Gd³⁺ (data not shown).

Fig. 4.

KCl evoked increase in [Zn²⁺]_i. A, KCl (90 mm) was applied for 15 sec alone or in combination with 300 μm Zn²⁺. TPEN (50 μm) was added 5 min after the exposure to Zn²⁺. Eachpoint is the mean ± SEM of 49 cells. Experiment is representative of nine. B, Concentration–response curve of KCl-induced [Zn²⁺]_i. Neurons were exposed for 15 sec to indicated KCl concentrations. Eachpoint is the mean ± SEM of the peak [Zn²⁺]_i levels obtained from at least three experiments. C, Pharmacological modulation of 90 mm KCl-induced increase in [Zn²⁺]_i in cortical neurons. [Zn²⁺]_i levels were measured before (Basal) or at the peak response after a 15 sec exposure to 90 mm KCl, alone or in the presence of 10 μm Gd³⁺, 100 μmverapamil (VER), 100 nm ω-conotoxin-GVIA (ω-CTx), 1 μm nimodipine (NIMO), 1.8 mm Ca²⁺, or 1.8 mm Ca²⁺ plus 0.8 mmMg²⁺. Each column is pooled from three to nine experiments involving at least 40 cells per experiment.Asterisks indicate difference from basal level; #, difference from KCl-stimulated level, as determined by ANOVA and Student–Newman–Keuls’ test (p < 0.05).

Inclusion of 1.8 mm Ca²⁺, alone or together with 0.8 mm Mg²⁺, in the bathing medium reduced by approximately half the peak [Zn²⁺]_i response (sensitive to TPEN) induced by application of 90 mm KCl plus 300 μm Zn²⁺ (Fig. 4C). To test the hypothesis that Zn²⁺ might also enter neurons via reverse operation of an exchanger similar (or identical) to the Na⁺–Ca²⁺ exchanger, we loaded neurons with Na⁺ by exposing the cultures to 1 mm ouabain for 15 min. Removal of extracellular Na⁺ (replaced with 130 mm NMDG), together with application of 100 μmZn²⁺, still in the presence of ouabain, then produced over several minutes a slowly progressive increase in neuronal somatic [Zn²⁺]_i (Fig.5B) that was larger than that produced by application of 100 μm Zn²⁺alone on control neurons (Fig. 5A). This enhanced increase in [Zn²⁺]_i was inhibited by the Na⁺–Ca²⁺ exchanger blockers benzamil-amiloride (BNZ) (Fig. 5C) andd-methyl-benzamil-amiloride (100 μm) (data not shown). To avoid activation of voltage-gated Ca²⁺ channels and glutamate receptors, 10 μm Gd³⁺, 10 μm MK-801, and 10 μm NBQX were present throughout the experiment. No detectable increase in [Zn²⁺]_i was produced by brief (15 sec) application of an Na⁺-free solution containing 100–300 μm Zn²⁺.

Fig. 5.

Na⁺-dependent Δ[Zn²⁺]_i. A, Fifteen minute exposure to 100 μm Zn²⁺ in the presence of 10 μm Gd³⁺, 10 μm MK-801, and 10 μm NBQX produced a gradual increase in [Zn²⁺]_i. Eachpoint is the mean ± SEM of 58 cells in a single experiment, representative of three. B, Removal of extracellular Na⁺ (equimolar substitution with NMDG) after the addition of 1 mm ouabain increases Δ[Zn²⁺]_i induced by exposure to 100 μm Zn²⁺. Each point is the mean ± SEM of 57 cells in a single experiment, representative of six. C, Same as in B, but with the addition of 100 μm BNZ. Each point is the mean ± SEM of 56 cells in a single experiment, representative of four.

To confirm that reversed operation of a neuronal exchanger like the Na⁺–Ca²⁺ exchanger could mediate Zn²⁺ influx, we used whole-cell voltage clamp to record the membrane current associated with the electrogenic exchanger operation. Under conditions favoring reverse operation, but limited by lack of extracellular Ca²⁺ (0 Na⁺, 0 Ca²⁺ in the bath; 60 mm Na⁺, 0 Ca²⁺ in the pipette), local application of 1 mm extracellular Zn²⁺ produced a reversible outward current shift (23.0 ± 4.4 pA at −40 mV, SEM; n = 4 cells) (Fig. 6), resembling the Na⁺–Ca²⁺ exchanger current induced by Ca²⁺ under the same conditions (Yu and Choi, 1997). This Zn²⁺-induced membrane current was blocked 70% by 100 μm BNZ (7.7 ± 3.2 pA, SEM;n = 4 cells) (Fig. 6) and was totally blocked by 300 μm BNZ (n = 7 cells) (data not shown). Furthermore, the current was not observed if Na⁺ was removed from the pipette solution (data not shown).

Fig. 6.

Exchanger current activated by external Zn²⁺ application. Left, Whole-cell recording from a cortical neuron in a bathing medium lacking Na⁺ or Ca²⁺ (see Materials and Methods); the pipette solution contained 60 mmNa⁺. ZnCl₂ (1 mm) was applied in the bath as indicated. Holding potential = −40 mV. Representative of four cells. Right, Same as inleft, but with 100 μm BNZ in the bath. Representative of four cells.

Modulation of [Zn²⁺]_i in response to glutamate agonists

Application of NMDA (10–300 μm, 15 sec) in the presence of 300 μm extracellular Zn²⁺produced a quick increase in neuronal [Zn²⁺]_i in an NMDA concentration-dependent manner that was sensitive to the competitive NMDA antagonist d-APV (100 μm) (Fig.7). A maximal [Zn²⁺]_i of 31 ± 0.97 nm was induced by 300 μm NMDA. This NMDA-induced increase in [Zn²⁺]_i was only partially sensitive to application of 10 μmGd³⁺ or removal of extracellular Na⁺ (NMDG replacement).

Fig. 7.

NMDA-induced increase in [Zn²⁺]_i. A, NMDA (100 μm, 15 sec) was applied together with 300 μm Zn²⁺, with or without 100 μmd-APV as indicated. TPEN (50 μm) was added as indicated. Each point is the mean ± SEM of 58 cells in a single experiment, representative of four. B, Concentration–response curve of NMDA-induced Δ[Zn²⁺]_i. The indicated concentrations of NMDA were applied for 15 sec in the presence of 300 μm Zn²⁺, and the resultant peak increase in [Zn²⁺]_i was measured. Eachconcentration point was pooled from three experiments involving at least 40 cells per experiment; SEM bars are lost in the symbols. C, Pharmacological modulation of NMDA-induced Δ[Zn²⁺]_i. [Zn²⁺]_i levels were measured before (basal) or at the peak response after a 15 sec exposure to NMDA, in presence of 10 μm Gd³⁺ or in absence of extracellular Na⁺ (NMDG). Eachcolumn is pooled from three to nine experiments involving at least 40 cells per experiment. Asterisksindicate difference from basal level; #, difference from NMDA-stimulated levels, as determined by ANOVA and Student–Newman–Keuls’ test (p < 0.05).

Application of kainate (10–300 μm, 15 sec) in the presence of 300 μm extracellular Zn²⁺also produced a quick increase in neuronal [Zn²⁺]_i in a concentration-dependent manner, sensitive to 10 μm NBQX (Fig.8). To determine the extent to which the Zn²⁺ entry triggered by kainate receptor stimulation was mediated indirectly via voltage-gated Ca²⁺channels, we exposed neurons to kainate in an Na⁺-free solution (NMDG) (to prevent Na⁺ entry-mediated membrane depolarization and consequent activation of voltage-gated Ca²⁺channels).

Fig. 8.

Kainate-induced Δ[Zn²⁺]_i. A, Kainate (100 μm, 15 sec) was applied together with 300 μm Zn²⁺, with or without 10 μm NBQX as indicated. TPEN (50 μm) was added as indicated. Each point is the mean ± SEM of 61 cells in a single experiment representative of three.B, Concentration–response curve of kainate-induced elevation in [Zn²⁺]_i. The indicated concentrations of kainate were applied for 15 sec in the presence of 300 μm of Zn²⁺, and the resultant peak increase in [Zn²⁺]_i was measured. Eachconcentration point was pooled from three experiments involving at least 40 cells per experiment; SEM bars are lost in the symbols.

Removal of Na⁺ inhibited kainate-induced increase in [Zn²⁺]_i in most, but not all, neurons. In three pooled experiments, application of 50 μm kainate in the Na⁺-free solution caused [Zn²⁺]_i to rise strongly in 17 of 156 (10.8%) neurons (Fig.9A,B). The microscope fields were marked, and the cultures were subsequently subjected to staining for kainate-activated cobalt uptake (“Co²⁺-positive neurons”), a marker for the functional expression of AMPA or kainate receptors gating channels with high Ca²⁺ permeability (Pruss et al., 1991, Turetsky et al., 1994; Yin et al., 1994). There was a one-to-one correlation between strong [Zn²⁺]_i-responsiveness in Na⁺-free conditions and Co²⁺positivity (Fig. 9, C,D).

Fig. 9.

[Zn²⁺]_i elevation in cobalt-positive neurons. A, B, Pseudocolor images depicting [Zn²⁺]_i, as mag-fura-5 340/380 ratio values as indicated in the scale, in cortical neurons before (A) and after (B) exposure (15 sec) to 50 μmkainate in the presence of 100 μm Zn²⁺but in the absence of extracellular Na⁺ (NMDG replacement). [Zn²⁺]_i elevation in a minority subpopulation of neurons can be seen. C, Phase-contrast micrograph of the same field. D, Kainate-activated cobalt staining of the same field. A one-to-one correlation is seen between neurons responding with a large Δ[Zn²⁺]_i in the absence of extracellular Na⁺ and cobalt-positivity. Scale bar, 50 μm.

Like the removal of extracellular Na⁺, inhibition of voltage-gated Ca²⁺ channels with 10 μmGd³⁺ or 1 μm nimodipine strongly attenuated the kainate-induced increase in [Zn²⁺]_i in most of the cells (Fig.10). However, in a small number of neurons, the increase in [Zn²⁺]_iremained near the control value (Fig. 10), supporting the idea that Zn²⁺ influx also occurred directly through AMPA/kainate receptor-gated channels, possessing high Ca²⁺ permeability.

Fig. 10.

Pharmacological modulation of kainate-induced elevation in [Zn²⁺]_i.Histograms show the fraction of the cortical neuronal population exhibiting the indicated peak [Zn²⁺]_i response during a 15 sec exposure to 100 μm kainate in the presence of 300 μm Zn²⁺. A, Normal extracellular Na⁺ (mean ± SEM, 23.5 ± 0.4 nm; n = 210 cells);B, in the absence of extracellular Na⁺ (12.9 ± 0.8; n = 235);C, in the presence of 1 μm nimodipine (9.6 ± 0.5; n = 289 cells); D, in the presence of 10 μm Gd³⁺ (11.2 nm ± 0.5; n = 305 cells). For each condition, the data were pooled from at least three experiments. All the treated conditions were statistically different from the control condition, as determined by ANOVA and Student–Newman–Keuls’ test (p < 0.05).

DISCUSSION

Taking a lead from Simons (1993), who used mag-fura-2 to measure Zn²⁺ concentrations in solution, we demonstrate here that mag-fura-5 can be used to follow [Zn²⁺]_i in living cortical neurons. Three lines of evidence indicate that the mag-fura-5 ratio signal we observed indeed reflects increases in [Zn²⁺]_i. First, the signal was induced by application of a Zn²⁺-selective ionophore, Na⁺-pyrithione, only when extracellular Zn²⁺ was present. Second, this Na⁺-pyrithione/Zn²⁺ signal was quenched rapidly by the specific cell-permeable Zn²⁺ chelator TPEN. Third, conclusions regarding routes of Zn²⁺entry reached here are consistent with previous observations using TSQ fluorescence or ⁶⁵Zn²⁺ accumulation to assess Zn²⁺ influx.

The idea that depolarizing stimuli can drive Zn²⁺entry through Gd³⁺- and nimodipine-sensitive voltage-gated channels fits with toxicity experiments (Weiss et al., 1993; Freund and Reddig, 1994; Manev et al., 1997), as well as with recent experiments from our laboratory tracking depolarization-induced neuronal ⁶⁵Zn²⁺ influx in the same cultures (H. S. Ying, S. Farhangrazi, C. S. Ling, D. Lobner, L. M. T. Canzoniero, S. L. Sensi, J. Y. Koh, C. T. Sheline, and D. W. Choi, unpublished observations). Although Zn²⁺, like other transition metals, is commonly considered an inhibitor of Na⁺ and Ca²⁺-voltage-gated channels (Harrison and Gibbons, 1994), such inhibition does not exclude slow permeation. Zn²⁺ sheds its water of hydration at a speed intermediate between that of Ca²⁺ and Mg²⁺ (Diebler et al., 1969) and permeates through Ca²⁺ channels on invertebrate neurons (Fukuda and Kawa, 1977), chromaffin cells (Vega et al., 1994), and myocytes (Atar et al., 1995). Wang and Quastel (1990) attributed the complex effects of extracellular Zn²⁺ on acetylcholine release at the neuromuscular junction to permeation through nerve terminal Ca²⁺ channels. It will be interesting in the future to see whether any subset of the voltage-gated channels passing Zn²⁺ prefer Zn²⁺ over Ca²⁺ and are really thus “Zn²⁺channels.”

The idea that Zn²⁺ can also permeate through NMDA receptor-gated Ca²⁺ channels (Koh and Choi, 1994) is supported by present data showing that the NMDA-induced [Zn²⁺]_i signal was only partly sensitive to Gd³⁺ addition or extracellular Na⁺ removal. The speed of this response (seconds) argues against any large mediation by exchange with Na⁺, a mechanism requiring several minutes to produce substantial elevations in [Zn²⁺]_i (see below). Zn²⁺ mediates a voltage-sensitive, fast flicker block of NMDA receptor-gated channels (Christine and Choi, 1990). Possibly, the onset of this block reflects plugging of the channel by Zn²⁺, and relief of block reflects permeation of Zn²⁺.

It is likely that Zn²⁺ similarly passes through the Ca²⁺-permeable AMPA/kainate receptors expressed on a small minority of cortical neurons, the “cobalt-positive” cells (Pruss et al., 1991; Turetsky et al., 1994), reflecting (in the case of AMPA receptors) lack of an edited form of GluRB/GluR2 (Hollmann et al., 1991; Verdoorn et al., 1991). This conclusion fits with that reached byYin and Weiss (1995), who used TSQ to assess [Zn²⁺]_i qualitatively. Recent evidence suggests that cobalt-positive cortical neurons are primarily GABAergic interneurons (Jonas et al., 1994; Yin et al., 1994), raising the interesting possibility that inhibitory interneurons may be especially vulnerable to Zn²⁺-induced death under conditions in which AMPA receptors are activated. Furthermore, the ability of certain AMPA receptors to mediate toxic Zn²⁺ entry may provide a link between previous observations implicating AMPA receptor overactivation (Pellegrini-Giampietro et al., 1992; Sheardown et al., 1993) and Zn²⁺ entry (Koh et al., 1996) in selective neuronal death after transient global ischemia.

Beside entry through voltage- and agonist-gated Ca²⁺channels, we provide evidence here for the possibility that Zn²⁺ may also enter depolarized neurons via BNZ-sensitive exchange with intracellular Na⁺. Although it is possible that a unique “Na⁺–Zn²⁺” exchanger mediates Zn²⁺ entry in exchange for Na⁺ efflux (or vice versa under different conditions), it is conservative to postulate that Zn²⁺ can substitute for Ca²⁺ in the operation of the known neuronal Na⁺–Ca²⁺ exchanger (Blaustein et al., 1991, 1996; Yu and Choi, 1997), as shown for Sr²⁺ and Ba²⁺ in cardiac sarcolemmal vesicles (Tibbits and Philipson, 1985) and for Mn²⁺ and Cd²⁺ in ferret red blood cells (Frame and Milanick, 1991). Such exchanger-mediated Zn²⁺ entry would be favored when the normal membrane Na⁺ gradient is diminished and membrane potential is depolarized, as occurs during brain ischemia. The outward current detected here in association with Zn²⁺ entry is consistent with the electrogenic nature of the exchanger, with a typical stochiometry of one Ca²⁺ (or Zn²⁺ ion) exchanged for three Na⁺ions.

We think it is unlikely that the mag-fura-5 signal measured here was substantially confounded by detection of Ca²⁺ or Mg²⁺. In the absence of extracellular Ca²⁺, previous studies have not suggested that [Ca²⁺]_i in neuronal soma would rise to the high micromolar levels needed to activate mag-fura-5 (K_d, ∼20 μm) (Petrozzino et al., 1995). Mg²⁺ is a greater possibility, because intracellular levels at rest are in the 0.5–1 mmrange (Brocard et al., 1993), already approaching theK_d of the dye for Mg²⁺ (2.6 mm) (Haugland, 1996). However, although some baseline Mg²⁺ binding is therefore probable, when we forced Mg²⁺ entry by increasing extracellular Mg²⁺ to 30 mm in the presence of a suitable ionophore, br-A23187, only small ratio shifts in mag-fura-5 fluorescence occurred, and these shifts, unlike the shifts induced by depolarizing neurons in the presence of extracellular Zn²⁺, were insensitive to TPEN. We cannot absolutely rule out the complex possibility that Zn²⁺ entry somehow caused massive intracellular Mg²⁺ release analogous to the Ca²⁺-triggered intracellular Mg²⁺ release delineated by Brocard et al. (1993). However, no elevation of [Mg²⁺]_i was seen by Brocard et al. (1993) in cortical neurons exposed to depolarizing conditions such as veratridine or kainate.

The nominal absence of Ca²⁺ or Mg²⁺ used in many of the present experiments to minimize confounding detection of Ca²⁺ or Mg²⁺ likely exaggerated Zn²⁺influx by over that which would normally occur in the presence of physiological concentrations of these ions. The absence of extracellular Ca²⁺ can reduce the selectivity of calcium channels on frog muscle cells (Almers et al., 1984) or potassium channels on squid neurons (Armstrong and Barneo, 1987), although because we performed our experiments without added chelators, enough Ca²⁺ and Mg²⁺ may have been present to maintain channel selectivity. In any case, previous studies demonstrated that [Ca²⁺]_oinversely affects Zn²⁺ toxicity (Koh and Choi, 1994). Moreover, a direct inhibition of⁶⁵Zn²⁺ influx by extracellular Ca²⁺ (IC₅₀ = 416 μm; Hill coefficient = 1.01) has been demonstrated in our culture system (H. S. Ying, S. Farhangrazi, C. S. Ling, D. Lobner, L. M. T. Canzoniero, S. L. Sensi, J. Y. Koh, C. T. Sheline, and D. W. Choi, unpublished observations), and we found here a substantial decrease (∼50%) in peak [Zn²⁺]_i when depolarization by 90 mm KCl was evoked in the presence of a physiological concentration (1.8 mm) of extracellular Ca²⁺ with or without 0.8 mmMg²⁺. However, present experiments may model conditions occurring in the ischemic brain in vivo, in which extracellular calcium dramatically drops (Hansen, 1985) and extracellular Zn²⁺ likely rises (Koh et al., 1996). That is, if we are correct in postulating that most of the mag-fura-5 fluorescence shift observed in the present experiments can be interpreted as an increase in [Zn²⁺]_i, then present observations provide a direct estimate (10–30 nm) of neuronal somatic [Zn²⁺]_i under pathophysiological conditions of NMDA or kainate activation when extracellular Zn²⁺ is abundantly present and extracellular Ca²⁺ is low (Figs. 7, 8).

Such increases would be produced easily by the total Zn²⁺ accumulating in cultured cortical neurons exposed to high K⁺ and 1 mmZn²⁺, as measured with atomic emission spectroscopy, which is sufficient to produce a calculated intracellular Zn²⁺ concentration of ∼1 mm in the absence of Zn²⁺ sequestration or binding (H. S. Ying, S. Farhangrazi, C. S. Ling, D. Lobner, L. M. T. Canzoniero, S. L. Sensi, J. Y. Koh, C. T. Sheline, and D. W. Choi, unpublished observations). Most likely, the bulk of the Zn²⁺entering depolarized neurons does not remain free in the cytosol, but rather is taken up into organelles (Palmiter et al., 1996) or binds to macromolecules such as metallothioneins (Masters et al., 1994; Ebadi et al., 1995; Maret, 1995). Erickson et al. (1997) demonstrated that deletion of the gene coding for metallothionein-III increased the CA3 neuronal death associated with kainate-induced seizures, a process likely mediated at least in part by toxic zinc translocation from mossy fiber terminals into CA3 neurons (Sloviter, 1985; Koh et al., 1996) (S. W. Suh, J. Y. Koh, and D. W. Choi, unpublished observations). Furthermore, present measurements of somatic fluorescence would not be expected to detect Zn²⁺ entering in neurites.

Previous determinations of [Zn²⁺]_i in nonneuronal cells have generally reported concentrations lower than those found here. In erythrocytes, Simons (1991) used⁶⁵Zn²⁺ flux to arrive at an estimate of 1.5–32 pm for [Zn²⁺]_i, depending on the ionic composition of the extracellular solution; in human leukemic cells, a [Zn²⁺]_i of ∼1 nm was determined by ¹⁹F nuclear magnetic resonance (Adebodun and Post, 1995). Additionally, fura-2-monitored [Zn²⁺]_i increased from 0.4 to 2 nm in chromaffin cells on electrical stimulation (Atar et al., 1995). However, recently zinquin was used to estimate concentrations of ∼20–50 μm for [Zn²⁺]_i in splenocytes and thymocytes at rest (Zalewski et al., 1993) and of 0.6 to 2.7 μm in hepatocytes removed from rats fed a high-zinc diet (Brand and Kleineke, 1996).

In summary, we make one comment about methods and one comment about cellular Zn²⁺ homeostasis. First, although mag-fura-5 appears to provide a reasonable tool for measuring [Zn²⁺]_i dynamically in living neurons, especially when extracellular Ca²⁺ and Mg²⁺ are absent, such measurements will need to be interpreted cautiously, and all measurements made with mag-fura-5 will need confirmation once truly Zn²⁺-selective ratioable dyes are available. Second, the observations presented here add to available evidence suggesting that Zn²⁺ can enter central neurons by multiple routes, likely in large part overlapping with routes of Ca²⁺ entry. Although the significance of this Zn²⁺ entry is most clearly defined in terms of contributing to the pathophysiology of neuronal degeneration in certain disease states, these mechanisms may also facilitate Zn²⁺ entry during normal synaptic transmission, permitting “transsynaptic messenger” Zn²⁺ to modulate a variety of processes in the postsynaptic neurons, such as signal transduction, transport processes, and gene expression (O’Halloran, 1993; Vallee and Falchuk, 1993).