Spreading depression (SD) is wave of profound depolarization that propagates throughout brain tissue and can contribute to the spread of injury after stroke or traumatic insults. The contribution of Ca2+ influx to SD differs depending on the stimulus, and we show here that Zn2+ can play a critical complementary role in murine hippocampal slices. In initial studies, we used the Na+/K+ ATPase inhibitor ouabain and found conditions in which SD was always prevented by L-type Ca2+ channel blockers; however, Ca2+ influx was not responsible for L-type effects. Cytosolic Ca2+ increases were not detectable in CA1 neurons before SD, and removal of extracellular Ca2+ did not prevent ouabain-SD. In contrast, cytosolic Zn2+ increases were observed in CA1 neurons before ouabain-SD, and L-type channel block prevented the intracellular Zn2+ rises. A slow mitochondrial depolarization observed before ouabain-SD was abolished by L-type channel block, and Zn2+ accumulation contributed substantially to initial mitochondrial depolarizations. Selective chelation of Zn2+ with N,N,N′,N′-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) abolished SD, implying that Zn2+ entry can play a critical role in the generation of ouabain-SD. TPEN was most effective when synaptic activity was reduced by adenosine A1 receptor activation, and a combination of Ca2+ and Zn2+ removal was required to prevent ouabain-SD when A1 receptors were blocked. Similarly, Zn2+ chelation could prevent SD triggered by oxygen/glucose deprivation but Zn2+ accumulation did not contribute to SD triggered by localized high K+ exposures. These results identify Zn2+ as a new target for the block of spreading depolarizations after brain injury.
Spreading depression (SD) is characterized as a wave of severe depolarization that spreads throughout CNS tissues at a rate of several millimeters per second. SD can be triggered by brief exposures to elevated extracellular K+, strong synaptic stimulation, inhibitors of Na+/K+ ATPase activity (e.g., ouabain), or ischemia models (Somjen, 2001). SD-like events are thought to be involved in the spread of injury after ischemic and traumatic brain injuries, and also contribute to migraine aura (Hossmann, 1996; Hadjikhani et al., 2001; Somjen, 2001; Hartings et al., 2003; Church and Andrew, 2005; Umegaki et al., 2005).
Large persistent intracellular Ca2+ increases follow SD, and it is clear that these increases contribute to neuronal death after some stimuli (Somjen, 2001). The contribution of Ca2+ to the initiation of SD is more complex, because Ca2+ accumulation is not required for initiation or progression of the SD event itself, at least when it is triggered in in vitro ischemia models or by ouabain (Ramos, 1975; Rader and Lanthorn, 1989; Young and Somjen, 1992; Basarsky et al., 1998; Somjen, 2001). However, SD generated by localized high K+ stimuli does appear to involve Ca2+ (Footitt and Newberry, 1998; Peters et al., 2003), likely because of influx via presynaptic P/Q type channels and stimulation of transmitter release (Ayata et al., 2000). Here, we examined whether Zn2+ accumulation might contribute to the initiation of SD, especially in cases in which Ca2+ removal is without effect.
Zn2+ can enter cells through several routes, including Ca2+ channels, and induce neuronal injury (Koh et al., 1996; Choi and Koh, 1998; Weiss et al., 2000; Calderone et al., 2004). Zn2+ can accumulate in mitochondria (Sensi et al., 1999; Jiang et al., 2001; Malaiyandi et al., 2005), and mitochondrial dysfunction has in turn been suggested to contribute to induction of some forms of SD (Bahar et al., 2000; Hashimoto et al., 2000; Gerich et al., 2006). A large and rapid mitochondrial depolarization has been reported coincident with SD generated by hypoxia, but a slow progressive mitochondrial depolarization was also noted before the onset of SD (Bahar et al., 2000). Because these effects were not prevented by the removal of extracellular Ca2+ (Bahar et al., 2000), we also examined the possibility that mitochondrial depolarization before SD could instead be a consequence of Zn2+ increases.
We examined first SD induced by the Na+/K+ ATPase inhibitor ouabain and report conditions in which L-type Ca2+ channel activation is essential for SD, and also for the mitochondrial depolarization that precedes ouabain-SD. Additional observations provide evidence that influx of Zn2+ rather than Ca2+ can be critically responsible for the onset of ouabain-SD. The relevance of this finding to other forms of SD was also tested and we show that Zn2+ accumulation is not required for SD generated by localized high K+ applications, but is an important contributor to SD in an in vitro model of ischemic injury. Some results have been presented in abstract form (Dietz et al., 2007a).
Materials and Methods
Male FVB/N mice were obtained from Harlan and were housed in standard conditions (12/12 h light/dark cycle) before being killed at 4–6 weeks of age. All procedures were performed in accordance with the National Institutes of Health guidelines for the humane treatment of laboratory animals, and the protocol for these procedures was reviewed annually by the Institutional Animal Care and Use Committee at the University of New Mexico School of Medicine. Acute slices (350 μm) were prepared as previously described (Dietz et al., 2007b). After cutting and holding for 1 h at 35°C, artificial CSF (ACSF) was changed, and slices were held at room temperature until used for recording. Individual slices were transferred to the recording chamber and were superfused with oxygenated ACSF at 2 ml/min. The recording temperature was maintained within 0.5°C by using a feedback controller (Warner TC344B) and was 30–35°C, depending on the specific experiments (see Results). Spontaneous burst-like or SD-like depolarizations were not observed under these recording conditions, and (except where noted for localized high-K+ applications) only a single challenge to a SD-triggering agent was tested in each slice.
Extracellular measurements of DC potentials were made using borosilicate glass microelectrodes, filled with ACSF (∼5 MΩ) and placed in stratum radiatum ∼50 μm below the surface of the slice. Ca2+ and Zn2+ measurements were made from individual CA1 pyramidal neurons. The procedures for intracellular recording/indicator injection were as described previously (Dietz et al., 2007b), with some modifications. Impalements were made using the step function of a Sutter manipulator (MP-225; Sutter Instrument), and neurons were visualized using a water-immersion objective [40×; numerical aperture (NA), 0.8; Olympus]. Neurons were impaled with sharp glass microelectrodes (initial resistance, ∼100–120 MΩ) containing charged fluorescent indicators for either Ca2+ (fura-2 or fura-6F) or Zn2+ (FluoZin-3). Microelectrodes were tip-filled with 10 mm indicator in 0.5 m KAc/0.5 m KCl, back-filled with 3 m KCl, and indicator was injected by passing hyperpolarizing current (350–450 pA; 10 min). In experiments coloading fura-6F and FluoZin-3, electrodes tips contained 5 mm of each indicator. All neurons included in the study showed stable resting potentials more negative than −60 mV and responded to depolarizing current pulses with trains of overshooting action potentials and Ca2+ transients, as described previously with these recording conditions (Shuttleworth and Connor, 2001; Dietz et al., 2007b). In all experiments, the recording/filling electrode was then slowly withdrawn (in 1–2 μm steps, applied every 10–15 s) from the neuron. Electrode withdrawal was monitored from the amplitude of evoked action potentials evoked by applying depolarizing test pulses (100 ms, 200 pA; every 10 s). When action potential amplitude had reduced to less than ∼20% of initial amplitude, a small membrane bleb had formed and withdrawal was paused. The impalement was then either spontaneously lost, or the electrode withdrawn with additional steps. Cytosolic Ca2+ levels were monitored throughout, and any neuron showing demonstrable Ca2+ increases during this procedure were discarded from the study. After electrode withdrawal, cells were allowed to recover for 20 min before application of any stimulus. Because high-resistance intracellular recording/filling electrodes were used, this procedure did not result in indicator leak or extracellular accumulation of any of the indicators examined.
Fura-2 and fura-6F were excited at 350/380 nm (400 nm dichroic; 50 ms) and emission detected at 510/50 nm using a monochromator-based imaging system (TILL Photonics, with Imago VGA CCD camera). FluoZin-3 was excited at 495 nm (505 nm dichroic; 15 ms), and emission was detected at 535/50 nm. When near-simultaneous fura-6F/FluoZin-3 measurements were made, coloaded neurons were excited sequentially at 350/380/495 nm (505 nm dichroic; 120/120/20 ms exposures) and emission of both indicators was detected at 535/50 nm. Estimation of Ca2+ concentrations and preparation of figures was as previously described (Shuttleworth and Connor, 2001; Dietz et al., 2007b).
Changes in mitochondrial potential were assessed using slices bulk-loaded with rhodamine 123 (Rh123). Slices were exposed to 26 μm Rh123 for 30 min at room temperature in a holding chamber before transfer to recording platform where the slices were washed with ACSF for 30 min before experimentation. Under these conditions, Rh123 is in the “quenched” mode, in which fluorescence increases are interpreted as mitochondrial depolarizations. Because of its relatively slow kinetics, mitochondrial Rh123 signals are not significantly complicated by plasma membrane effects under these conditions (Duchen et al., 2003). Fluorescence was excited at 488 nm, and emission was detected at 535/50 nm. The progression of SD was also monitored in some experiments by using autofluorescence excited at 360 nm (emission, >420 nm), as described previously (Shuttleworth and Connor, 2001). A 10× water-immersion objective (NA 0.3; Olympus) was used for Rh123 and autofluorescence imaging studies, which provided a field of view of ∼840 μm centered in area CA1. This region was not sufficient to assess progression of events into adjacent hippocampal subregions, such as area CA3.
Reagents and solutions.
Slice cutting solution contained the following (in mm): 2 KCl, 1.25 NaH2PO4, 6 MgSO4, 26 NaHCO3, 0.2 CaCl2, 10 glucose, 220 sucrose, and 0.43 ketamine. ACSF contained the following (in mm): 126 NaCl, 2 KCl, 1.25 NaH2PO4, 1 MgSO4, 26 NaHCO3, 2 CaCl2, and 10 glucose, and was equilibrated with 95%O2/5%CO2. Cutting and recording solutions were both 315–320 mOsm. ACSF was modified in zero-Ca2+ experiments by replacement of CaCl2 with MgSO4, and addition of chelators (0.5 mm EGTA or 1 mm BAPTA, as described below). In oxygen/glucose deprivation (OGD) experiments, ACSF was modified by equimolar replacement of glucose with sucrose, and equilibrated with 95% N2/5% CO2.
Fluorescent indicators and N,N,N′,N′-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) were obtained from Invitrogen, and all other reagents were obtained from Sigma-Aldrich. Ouabain was prepared as a 15 mm stock in H2O. Nimodipine and nicardipine were prepared as 10 or 100 mm stocks in ethanol. 1,3-Dipropyl-8-cyclopentylxanthine (DPCPX) was prepared as a 100 μm stock in EtOH. TPEN and (+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate (MK801) stocks were in DMSO. Vehicle controls were conducted throughout for matched levels of DMSO and showed no effect on SD. All other stocks were prepared in ACSF. Most drugs were applied for 30 min before beginning challenge with ouabain, or other SD stimuli. Freshly prepared TPEN was preexposed for 10 min. High-K+-SD was produced by delivering brief pulses from a patch electrode (3–5 MΩ) placed on the surface of the slice, ∼350 μm from the recording electrode. KCl (1 m) was ejected by using a single brief pressure pulse (10 psi, 100 ms; Picrospritzer II).
Significant differences between group data were evaluated using unpaired Student's t tests or one-way ANOVA. Bonferroni's multiple-comparison test was used for post hoc analysis in which the effects of multiple drug treatments were compared against each other. Dunnett's multiple-comparison test was used for post hoc comparisons of multiple time points of single drug treatment, when compared with responses immediately before drug treatment. A value of p < 0.05 was considered significant in all cases.
L-type channels can contribute to SD
A low concentration of ouabain (30 μm) reliably produced SD in acutely prepared hippocampal slices maintained at 35°C (7.9 ± 0.9 min latency; n = 13). Figure 1A shows an example of the rapid negative voltage deflection recorded with an extracellular electrode. This event is characteristic of SD and is similar to previous reports examining sustained ouabain exposures in rodent hippocampal slices (Basarsky et al., 1998; Obeidat and Andrew, 1998; Balestrino et al., 1999; Menna et al., 2000). The progression of the event was confirmed in all experiments by visualization of a spreading wave of tissue autofluorescence decrease that was coincident with the voltage deflection (Hashimoto et al., 2000; Shuttleworth et al., 2003) (supplemental data, available at www.jneurosci.org as supplemental material). Figure 1B shows that an L-type Ca2+ channel blocker (nimodipine; 10 μm) prevented SD in most (80%) preparations tested.
Because both presynaptic and postsynaptic events may contribute to the onset of SD, we tested whether the level of synaptic activation in the slice might influence the effectiveness of L-type block. We used an A1 receptor agonist [N6-cyclopentyladenosine (CPA); 300 nm] to strongly decrease synaptic efficacy. This concentration was sufficient to abolish evoked EPSPs in CA1 stratum radiatum (data not shown), and under these conditions, nimodipine always blocked SD (six of six preparations tested) (Fig. 1B). Conversely, when slices were pretreated with an A1 receptor antagonist (DPCPX; 100 nm) to increase presynaptic excitability, nimodipine never prevented ouabain-SD. A dependence of L-type sensitivity on endogenous A1 receptor activation was also supported by experiments at a lower recording temperature (30°C), under which conditions extracellular adenosine accumulation is reported to be reduced (Masino and Dunwiddie, 1999). At 30°C, SD was little affected by nimodipine, unless the exogenous A1 receptor agonist was coapplied. Thus, at 30°C, nimodipine alone blocked SD in only one of six preparations, but blocked SD in six of six slices when pretreated with CPA. Control experiments showed that CPA or DPCPX alone did not influence the propagation rate or time to ouabain-SD onset (n = 6 each).
The effects of nimodipine described above were mimicked by an alternative L-type Ca2+ channel blocker (nicardipine; 10 μm). In the presence of CPA at 35°C, nicardipine prevented SD in six of six slices and was without effect on ouabain-SD when A1 receptors were blocked (DPCPX; six of six slices).
Previous work (Basarsky et al., 1999) showed that NMDA receptor activation contributes to the onset of ouabain-SD in rat hippocampal slices. Consistent with this observation, we found that the noncompetitive antagonist MK-801 (50 μm) prevented SD produced by 30 μm ouabain. Inhibition (no SD during observations >30 min) was observed at 35°C in the presence of CPA (n = 6) and also in the presence of DPCPX [no SD in five of six slices, and delayed SD (14.3 min) in one of six slices]. Together, these results suggest that NMDA receptors are a major contributor to the onset of ouabain-SD but that L-type channel flux also contributes to reaching SD threshold. Removing L-type flux is not sufficient to prevent SD when synaptic transmission is increased with DPCPX. However, blocking L-type flux alone appears sufficient to prevent reaching SD threshold under conditions that are relevant for ischemic brain injury, in which extracellular adenosine levels are elevated (Rudolphi et al., 1992).
The next sets of studies therefore investigated mechanisms that link L-type Ca2+ channel function to the initiation of ouabain-SD. Except where noted, recording conditions were chosen that emphasized a contribution of L-type channels (i.e., experiments were done at 35°C, and CPA was included to “clamp” slices at a high level of adenosine A1 receptor activation, and thereby minimize the impact of slice–slice variation in endogenous adenosine levels).
L-type flux causes mitochondrial depolarization before SD
Changes in mitochondrial inner membrane potential (ΔΨm) during ouabain exposure were assessed in slices loaded with Rh123. Rh123 was used in the quenched mode, in which fluorescence increases represent ΔΨm depolarization (see Materials and Methods). Figure 2 shows results from stratum pyramidale, in which a large and rapid ΔΨm depolarization was observed coincident with the onset of SD, but in addition, a significant progressive depolarization was observed for ∼5 min before ouabain-SD (Fig. 2A). Figure 2B shows that nimodipine prevented both phases of ΔΨm changes during ouabain exposures. The slow initial ΔΨm depolarization was abolished, and because SD was blocked, the large ΔΨm depolarization associated with ouabain-SD was prevented.
Comparison of neuronal Ca2+ and Zn2+ dynamics
We examined whether intracellular Ca2+ and/or Zn2+ increases were likely responsible for the events described above. Although many Ca2+ indicators also demonstrate high-Zn2+ sensitivity when measured in free solution, a combination of factors appears to greatly diminish the apparent Zn2+ sensitivity of fura-based indicators when recordings are made in intact cells. These factors include the effects of relatively high indicator concentrations, competition with Ca2+ (Dineley et al., 2002), and possibly the influence of intracellular constituents (Marchi et al., 2000). Thus, intracellular Ca2+ and Zn2+ transients within neurons can be effectively distinguished by using combinations of fluorescent indicators (Devinney et al., 2005), and a similar approach was used here. Figure 3A shows an example of simultaneous measurements of intracellular Zn2+ and Ca2+ increases in the soma of a CA1 neuron coloaded with the high-affinity Zn2+ indicator FluoZin-3 and the low-affinity Ca2+ indicator fura-6F. Before the onset of ouabain-SD, there was no detectable increase in fura-6F ratio, and in addition there was no evidence of fluorescence increases in the individual signals after 350/380 nm excitation. In contrast, there was a significant increase in neuronal FluoZin-3 fluorescence, which was seen for ∼5 min before SD propagated through the region (Fig. 3). Coincident with the onset of ouabain-SD, large increases in both fura-6F ratio and FluoZin-3 fluorescence were detected.
Because fura-6F might miss small Ca2+ elevations before SD, experiments were repeated in neurons loaded with only the high-affinity indicator fura-2. Figure 3B shows there was no detectable cytosolic Ca2+ elevation before ouabain-SD measured in these fura-2-loaded neurons, in contrast to mean FluoZin-3 fluorescence increases, recorded from a population of different neurons. We acknowledge that the differences in apparent affinities of the indicators for Zn2+ and Ca2+ may mean that small Zn2+ increases may be detected by FluoZin-3 (KD,Zn ∼ 15 nm), but similar amplitude Ca2+ responses may still be missed in fura-2 measurements (KD,Ca ∼ 225 nm). It is also possible that somewhat larger Ca2+ increases may be more effectively sequestered and/or extruded by neurons (when compared with Zn2+) and such differences may contribute to the differential ability to detect cytosolic Zn2+ versus Ca2+ accumulation under these conditions.
Figure 4 shows that the Zn2+-selective chelator TPEN (50 μm) abolished ouabain-induced FluoZin-3 increases before SD and prevented ouabain-SD in all preparations tested under these conditions. These results suggest that (1) Zn2+ increases are indeed responsible for FluoZin-3 increases before SD and (2) Zn2+ (rather than Ca2+) increases are required for SD initiation with ouabain. Consistent with these suggestions, superfusion of slices with nominally Ca2+-free ACSF did not prevent ouabain-SD in six of six slices tested.
Sources of Zn2+ and strong correlation of early Zn2+ rises with subsequent SD induction
TPEN is membrane permeable and thus cannot distinguish between intracellular and extracellular sources of Zn2+ that might contribute to ouabain-SD. When both Ca2+ and Zn2+ were removed from the extracellular space (superfusion with Ca2+-free ACSF supplemented with 0.5 mm EGTA; KD,Ca = 10−8 m; KD,Zn = 10−12 m), there was still a significant initial FluoZin-3 increase and SD was not prevented (Fig. 4). Selective extracellular Zn2+ chelation with Ca2+-EDTA (1 mm; KD,Zn = 10−16 m) was also ineffective at preventing either Zn2+ rises or SD (six of six preparations). Despite the lack of effect of these extracellular Zn2+ chelators, increases in FluoZin-3 fluorescence before SD were completely prevented by preexposure with nimodipine. This was demonstrated in experiments with nimodipine in Ca2+-free media containing EGTA, and under these same conditions SD was always prevented (Fig. 4). Together, these results suggested that Zn2+ increases before ouabain-SD were mediated by L-type Ca2+ channels, but that the Zn2+ flux was not accessible to chelation by EGTA or Ca2+-EDTA.
EGTA and Ca2+-EDTA both have relatively slow on-rates for divalent cation binding (Smith et al., 1984; Vogt et al., 2000). We therefore tested the effects of BAPTA, which chelates both Ca2+ and Zn2+ (KD,Ca = 10−7 m; KD,Zn = 10−9 m) but with significantly faster binding kinetics (Adler et al., 1991). BAPTA blocked SD in 6 of 10 cases and significantly decreased FluoZin-3 increases before ouabain-SD (Fig. 4). These observations suggest that Zn2+ may accumulate outside neurons and enter through L-type channels before it can be bound by the slower extracellular chelators. One possible explanation for these observations would be a very close proximity between CA1 L-type channels and Zn2+ release sites.
The occurrence of transmembrane Zn2+ flux was further supported by measurements of extracellular FluoZin-3 fluorescence (Fig. 5). FluoZin-3 was included in ACSF, together with 1 mm Ca2+-EDTA to reduce background fluorescence (Qian and Noebels, 2005) (supplemental data, available at www.jneurosci.org as supplemental material). Under these conditions, fluorescence was stable at baseline but was noted to decrease before ouabain-SD, with a time course that corresponded well with the intracellular FluoZin-3 increases signals shown above (Fig. 3). Furthermore, nimodipine preexposures that prevented SD also abolished this FluoZin-3 fluorescence decrease, consistent with the hypothesis that extracellular Zn2+ decreases were attributable to flux through activated L-type channels. In slices in which SD was not blocked, a large extracellular FluoZin3 increase propagated across the slice after the onset of the ouabain-SD response. Control studies showed that slice autofluorescence changes did not contaminate extracellular FluoZin-3 measurements (see supplemental data, available at www.jneurosci.org as supplemental material).
Zn2+ and Ca2+ can both contribute to the depolarization of mitochondrial inner membrane potential before SD
The findings illustrated in Figure 4 provide evidence that Zn2+ entry through L-type Ca2+ channels can play a crucial role in the induction of ouabain-SD, but do not indicate the ionic contributions to the slow ΔΨm preceding SD. Indeed, as shown in Figure 6, a slow ΔΨm still occurs in the presence of TPEN, which is presumably Ca2+ dependent, because the addition of Ca2+-free/EGTA in addition to TPEN abolished the ΔΨm completely. To separate Zn2+- from Ca2+-dependent components to the slow ΔΨm, we compared effects of selective Zn2+ removal (by TPEN) with Ca2+-free/EGTA ACSF. Figure 6B shows that, with Ca2+-free/EGTA, the slow ΔΨm occurred early, whereas with selective Zn2+ removal it was more delayed, indicating that Zn2+ is the primary contributor to the earliest phase of the ΔΨm. Over the first 5 min after ouabain exposure, there was no significant increase in Rh123 fluorescence with TPEN, but there was a significant increase in Ca2+-free/EGTA (p < 0.01, Dunnett's post hoc test). At 7 min, ΔΨm significantly increased under both conditions (p < 0.01 for both). Indeed, suggesting that Zn2+ from the same extracellular source accounts for the measured Zn2+ rises that correlates with SD in Figure 4 and the early ΔΨm, both are blocked by TPEN and the fast extracellular chelator, BAPTA (data not shown for ΔΨm), but are not blocked by the slow extracellular chelator, EGTA.
Taking together the results from Figures 2, 3, and 6, this suggests that ouabain can lead to Ca2+ and Zn2+ influx via L-type channels and uptake by mitochondria. The high-affinity Zn2+ indicator FluoZin-3 appears to detect small amounts of Zn2+ that remain available in the cytosol, whereas cytosolic Ca2+ increases are not detectable. Ca2+-dependent ΔΨm by itself is not sufficient to cause SD. In contrast, mitochondrial uptake of Zn2+ is closely correlated with the onset of ouabain-SD
Combined effects of Zn2+ and Ca2+ when A1 receptors were blocked
We next investigated whether Zn2+ was still critically required for ouabain-SD under conditions in which L-type flux was not involved. As described above (Fig. 1), nimodipine did not block SD when preparations were preexposed to DPCPX, an antagonist of A1 receptors. This preexposure is expected to maximize presynaptic excitability and possible contributions of glutamate release to the onset of ouabain-SD. Despite the fact that L-type block was ineffective, TPEN was still quite effective in blocking SD, preventing the event in four of six preparations tested (Fig. 7). This observation suggests that, in some cases, Zn2+ can contribute to ouabain-SD via mechanisms other than L-type influx.
Previous studies have shown that Ca2+-permeable AMPA receptors flux Zn2+ and can contribute to Zn2+-mediated toxicity in cortical neurons (Weiss et al., 2000). We therefore tested the effects of a nonselective AMPA/KA receptor antagonist (CNQX; 30 μm) under conditions in which glutamate release should be enhanced. CNQX did not prevent ouabain-SD in DPCPX in six of six preparations (delay, 7.48 ± 0.44 min). In contrast, as noted above (see L-type channels can contribute to SD), the NMDA receptor antagonist MK801 was effective in preventing ouabain-SD in DPCPX. These results imply that Zn2+ influx via Ca2+-permeable AMPA receptors are not responsible for Zn2+ increases triggering SD, and suggest that consequences of NMDA receptor activation are likely contributors for triggering ouabain-SD under these conditions. We examined Zn2+ increases in CA1 neurons loaded with a combination of fura-6F and FluoZin3, under conditions designed to isolate Zn2+ accumulation caused by non-L-type influx (DPCPX; nimodipine; Ca2+-free/EGTA; 35°C). Under these conditions, a significant FluoZin-3 increase was observed in CA1 somata before SD onset (peak ΔF/F0, 16.43 ± 3.48; p < 0.01; n = 5). Consistent with the notion that these increases were attributable to NMDA receptor activation, preexposure of slices to MK801 abolished FluoZin-3 increases (mean loss of fluorescence, −8.12 ± 3.51% over the same time course; n = 6). The kinetics of these responses is shown in supplemental Fig. 2 (available at www.jneurosci.org as supplemental material).
Consistent with the observations above, the slow extracellular Zn2+ chelator Ca2+-EDTA did not block nor delay ouabain-SD during DPCPX exposure (latency, 8.45 ± 0.60 min; n = 6; 35°C), suggesting that the response is not attributable to contaminating Zn2+ in the superfusate, but rather is attributable to endogenous Zn2+ mobilized after NMDA receptor activation, either from intracellular sites or from extracellular sites close to influx routes.
The incomplete block of SD with TPEN (see above) also implies some involvement of Zn2+-independent mechanisms, revealed by increasing presynaptic excitability with DPCPX. Under these conditions, Ca2+-free/EGTA solution remained ineffective (no block or delay of SD; occurring in six of six preparations with a delay of 6.99 ± 0.93 min). However, a combination of Ca2+-free/EGTA and TPEN blocked SD in six of six preparations tested. Together, these findings suggest that, although Zn2+-dependent mechanisms predominate, either Zn2+- or Ca2+-dependent mechanisms can be sufficient to trigger SD, and a combination of these mechanisms may become more apparent when presynaptic excitability is increased.
SD triggered by OGD or high K+
Having established that Zn2+ can contribute to SD generated by ouabain, a final series of studies examined whether Zn2+ was also a contributor to SD triggered by other types of stimuli. The following two test stimuli were chosen: (1) OGD and (2) localized high-K+ applications. OGD is a model of some aspects of in vivo ischemic injury and produces an SD event. Previous work has shown that SD produced by either hypoxia alone (Balestrino and Somjen, 1986; Bahar et al., 2000) or OGD (Rader and Lanthorn, 1989; Tanaka et al., 1997; Obeidat and Andrew, 1998) can lead to irreversible neuronal damage, but that the propagation of these types of SD is generally not prevented by Ca2+ removal. In contrast, high-K+-induced SD can generally be generated repetitively in the same preparation without obvious tissue damage (Buresová and Bures, 1969; Nedergaard and Hansen, 1988) and Ca2+ removal can effectively prevent the propagation of these events (Footitt and Newberry, 1998; Peters et al., 2003). We therefore tested the hypothesis that OGD-SD would be prevented by Zn2+ (but not Ca2+ removal), and the opposite would be the case for high-K+-SD. Based on our observations with ouabain-SD, our initial studies here used concomitant A1 receptor activation (1 μm CPA; 32°C), in an attempt to make a clear distinction between Zn2+- and Ca2+-dependent mechanisms.
Figure 8 shows that both OGD and localized high K+ elicited negative shifts in extracellular potential that were similar in waveform and amplitude. Additionally, the propagation rates [as assessed by the spreading wave of tissue autofluorescence decrease described above (see L-type channels can contribute to SD)] were not significantly different between these two stimuli. Figure 8 also shows that OGD-SD Ca2+ signals were similar to ouabain responses. Before OGD-SD, there was no detectable increase in Ca2+ in six CA1 neurons, but the arrival of SD was associated with a large, irrecoverable Ca2+ increase that originated in somata and rapidly progressed toward apical dendrites in all neurons tested. Somatic Ca2+ elevations were estimated at 24.1 ± 1.1 μm (n = 6), and Ca2+ elevations throughout the neuron showed no indication of recovery after OGD-SD was generated and resulted in rapid neuronal injury. In contrast, high-K+-SD produced a transient Ca2+ elevation (average increase 7.8 ± 1.9 μm in soma and 25.3 ± 2.6 μm measured in apical dendrites 40 μm from soma) that lasted <5 s. After this transient Ca2+ increase, an advancing front of high Ca2+ traveled from distal dendritic sites toward the soma. The advancing front of Ca2+ never fully involved the soma but instead quickly retreated out along the dendrites toward their sites of origin. Within an average of 1.7 ± 0.2 min, intracellular Ca2+ had recovered to >95% of resting levels. Stimulation of SD by high K+ did not result in irreversible injury because SD could be generated repetitively by high K+ in these preparations after 45 min recovery and a similar Ca2+ response was produced with each SD (three of three preparations tested).
Effects of Ca2+ or Zn2+ removal on OGD-SD and high-K+-SD
Selective removal of extracellular Ca2+ (plus EGTA) did not prevent the generation of OGD-SD, nor did it change the propagation rate of the spreading event when compared with responses in 2 mm Ca2+ in interleaved experiments (4.0 ± 0.2 vs 4.1 ± 0.4 mm/min, Ca2+-free and control, respectively; n = 6 each; p = 0.75). In contrast, Ca2+ removal caused a profound inhibition of the propagation of SD evoked by K+ application, in all preparations tested (Fig. 9A) (supplemental Fig. 3, available at www.jneurosci.org as supplemental material).
Figure 9B summarizes experiments showing that preexposure to the Zn2+ chelator TPEN (50 μm) abolished OGD-SD in all slices tested (six of six slices). In contrast, high-K+-SD was unaffected by TPEN. High-K+-SD occurred in all slices tested (six of six) and the time before SD onset was not changed (0.17 ± 0.04 vs 0.15 ± 0.08 min, control and TPEN, respectively; n = 6 each; p = 0.75). Thus, under these recording conditions (with A1 receptor activation), there was a complete discrimination between Ca2+ and Zn2+ dependence of the two forms of SD; OGD-SD was entirely dependent on Zn2+ but unaffected by Ca2+ removal, whereas high-K+-SD was entirely dependent on Ca2+ but unaffected by Zn2+ chelation.
Figure 10 shows that when A1 receptors were blocked (DPCPX; 100 nm), Zn2+ remained a critical contributor to the generation of OGD-SD, but that Zn2+ chelation alone was no longer sufficient to prevent OGD-SD. Thus, in the presence of TPEN and DPCPX, SD was generated in five of six preparations tested. Ca2+ removal (plus EGTA) remained ineffective (zero of six preparations tested), but a combination of Zn2+ and Ca2+ removal blocked SD in six of six preparations tested. Together, these results suggest cooperative actions of Zn2+ and Ca2+ in triggering OGD-SD, under conditions in which synaptic activity is enhanced. An exclusive requirement for Zn2+-dependent initiation can be demonstrated when presynaptic activity is reduced by either A1 receptor activation or previous removal of extracellular Ca2+.
Zn2+ can contribute to SD
SD can be generated by a number of diverse stimuli and involves different mechanisms. The results here provide the first evidence that Zn2+ accumulation can contribute to SD initiated by the Na+/K+ ATPase inhibitor ouabain. Using this model, we characterized a predominant route of endogenous Zn2+ influx and subsequent neuronal accumulation with SD. The relevance of this finding to other types of SD was tested and showed that Zn2+ could be critically involved in OGD-SD, but not high-K+-SD. The relative contributions of Zn2+- and Ca2+-dependent mechanisms appear to be dependent on the presynaptic excitability, because purely Zn2+ mechanisms are more apparent when presynaptic activity is suppressed by A1 receptor activation, and in contrast a combination of Ca2+ and Zn2+ removal is required for abolition of SD when synaptic excitability is enhanced.
Zn2+, rather than Ca2+ flux via L-type channels in ouabain-SD
Zn2+ first emerged as a candidate for SD initiation because of the mismatch between the effects of Ca2+ channel blockers and Ca2+ removal during ouabain-SD experiments (Figs. 1, 2, 4). We found conditions in which L-type channel blockers always prevented ouabain-SD, but the effects of the L-type blockers were not mimicked by removal of Ca2+ from the bathing medium. Likewise, mitochondrial depolarization before SD was also dependent on L-type channels, but not blocked by Ca2+ removal. Zn2+ is well established to permeate L-type channels (Weiss et al., 1993; Kerchner et al., 2000), and can contribute to mitochondrial depolarization (Sensi et al., 1999; Dineley et al., 2005), and therefore Zn2+ influx was considered a candidate mechanism for ouabain-SD initiation.
Neuronal Zn2+ increases were blocked by L-type channel blockers and were substantially attenuated by the fast extracellular Ca2+ and Zn2+ chelator, BAPTA (Adler et al., 1991), but were not affected by extracellular chelators with slow kinetics (EGTA; Ca-EDTA). Thus, these increases appear to result from Zn2+ accumulation in the extracellular space and rapid entry through L-type channels, before being bound by the slower chelators. Fluorescence measurements of extracellular Zn2+ (Fig. 5) also support extracellular sources of Zn2+ for the initiation of ouabain-SD, rather than liberation from intracellular binding sites. The relatively slow inactivation kinetics of L-type channels may underlie the large contribution of this particular voltage-dependent channel to Zn2+ influx, during prolonged depolarizations generated by ouabin exposure.
The source of the extracellular Zn2+ accumulation is unknown but is likely to be made up at least in part by vesicular release (Frederickson et al., 2005), for which evidence has previously been presented in the CA1 region of murine hippocampal slices (Qian and Noebels, 2005). It is also possible that other extracellular sources of Zn2+ contribute (Kay, 2003), but the relative contribution of different potential sources remains to be determined. Zn2+ accumulation was readily detected with FluoZin-3 in CA1 somata, but these experiments did not have the resolution to investigate possible increases in structures close to synaptic release sites (dendrite shafts and spines), before accumulation in somata.
Coupling Zn2+ flux to SD
Mitochondrial accumulation is a possible mechanism by which Zn2+ contributes to the induction of ouabain- or OGD-SD. Zn2+ uptake can induce depolarization of isolated mitochondria (Jiang et al., 2001) and mitochondria within cultured cortical neurons (Sensi et al., 2000). Notably, the studies on isolated mitochondria indicate that Zn2+ induces this effect with far greater potency than Ca2+ [10 nm for Zn2+ vs ∼100 μm for Ca2+ (Jiang et al., 2001)], consistent with present indications of Zn2+-dependent mitochondrial depolarization. In addition to effects on mitochondrial potential, previously described consequences of mitochondrial Zn2+ uptake also include reactive oxygen species production (Sensi et al., 2000), increased mitochondrial membrane permeability (Bonanni et al., 2006), and possibly compromised ATP production (Dineley et al., 2003).
NMDA receptor contributions to ouabain-SD
A prominent effect of A1 activation is to reduce synaptic transmitter release, and the concentration of agonist (CPA) used here was sufficient to abolish evoked postsynaptic potentials. An A1 receptor agonist was useful for characterizing the mechanisms of Zn2+ involvement in ouabain-SD, because it revealed a clear dependence on L-type Ca2+ channels and Zn2+ influx, without requirement for Ca2+-dependent mechanisms. In contrast, when synaptic efficacy was deliberately enhanced, preventing L-type flux no longer blocked ouabain-SD, and the partial TPEN sensitivity of SD under these conditions must be explained by Zn2+ accumulation via other routes.
Inhibition of NMDA receptors has been reported to inhibit ouabain-SD (Basarsky et al., 1999), and we found that NMDA receptor block prevented SD triggered by 30 μm, both under conditions of increased (DPCPX) or decreased (CPA) synaptic excitability. This suggests a more dominant contribution of NMDA receptors to SD, which operates together with L-type flux. However, the relative contributions of the two pathways shifts significantly depending on the degree of synaptic excitability, such that simply removing the contribution of L-type channels is sufficient to prevent reaching SD threshold when synaptic excitability is suppressed by adenosine receptor activation. Such conditions are particularly relevant to ischemia, in which extracellular adenosine levels and A1 activation increase significantly (Rudolphi et al., 1992).
Consistent with the hypothesis that NMDA receptor activation is primarily responsible for Zn2+ increases under conditions of increased excitability, MK801-sensitive FluoZin-3 increases were demonstrated in CA1 neurons, before the onset of ouabain-SD in DPCPX. These increases could involve either influx and/or liberation from intracellular Zn2+ buffers such as metallothionein III (Lee et al., 2003) as a consequence of NMDA receptor activation. In addition, the results suggest that additional consequences of NMDA activation (i.e., dependent on Ca2+ permeability) are likely to explain the triggering of SD in cases in which TPEN did not block SD, but a combination of Ca2+ and Zn2+ chelation was shown to be effective.
SD triggered by oxygen/glucose deprivation and high K+
The relevance of Zn2+-dependent mechanisms was tested using two different models, high K+ and OGD. The SD produced by these stimuli appears very similar as it propagates across the brain slice, but the latency in onset and consequences of the SD are quite different. Similar findings have been described previously between SD evoked by hypoxia or high K+ (Aitken et al., 1998). Figure 8 illustrates the long delay in OGD-SD and sustained Ca2+ overload that follows establishment of the response, in contrast to the very rapid onset of high-K+-SD and the reversible nature of its Ca2+ transient. This is consistent with the irrecoverable nature of damage after persistent OGD (Rader and Lanthorn, 1989; Obeidat and Andrew, 1998; Jarvis et al., 2001) and the lack of injury after high-K+-SD (Buresová and Bures, 1969; Nedergaard and Hansen, 1988). Previous work has established that synaptic release mechanisms are involved in recoverable forms of SD (Ayata et al., 2000; Kunkler and Kraig, 2004; van den Maagdenberg et al., 2004), and consistent with this, Ca2+ removal inhibited high-K+-SD. Furthermore, Zn2+ chelation had no effect, implying that Zn2+ accumulation was not required, even when synaptic efficacy was dampened by A1 receptor agonists.
In contrast, Zn2+ accumulation can be essential for OGD-SD, and this was shown when synaptic activity was suppressed by A1 receptor activation. Under these conditions, TPEN always blocked OGD-SD, but Ca2+ removal was completely without effect, pointing to an essential role for Zn2+ under these conditions. Alternatively, when adenosine A1 receptor activation was deliberately prevented (using DPCPX), Zn2+ mechanisms were no longer the only contributor required for OGD-SD, and a combination of Ca2+ and Zn2+ removal was required to prevent SD. This suggests that both Zn2+-dependent and Zn2+-independent mechanisms are involved and (as with ouabain) the relative contribution depends on the degree of synaptic activation.
Together, it is likely that progressive accumulation of Zn2+ in neurons could be an important contributor to SD in which initial membrane depolarizations are slow and progressive (OGD; ouabain), but not necessarily in other SD models in which the time course of initial depolarization is very rapid (high K+). Previous studies have provided strong evidence that intracellular Zn2+ accumulation contributes substantially to injury after in vivo ischemia (Koh et al., 1996; Calderone et al., 2004). It is tempting to consider the possibility that one of the ways that Zn2+ contributes to postischemic injury is by facilitating the induction of SD-like events in vivo. Periinfarct depolarizations have been described after in vivo ischemia and propagate from the edges of infarcted regions to contribute to the spread of ischemic injury in the hours or days after an insult (Nedergaard and Astrup, 1986; Hossmann, 1996; Hartings et al., 2003). It will be of interest to determine whether Zn2+ accumulation is critical for the generation of periinfarct depolarizations after ischemia in vivo, and whether prevention of these events could contribute to the effectiveness of delayed Zn2+ chelation, provided many hours after the initial insult (Calderone et al., 2004).
The present work provides the first evidence that endogenous Zn2+ can contribute to SD, and could provide a key to distinguishing between different types of SD, in different pathophysiological settings. Additional understanding of mechanisms involved in the induction of SD by Zn2+ and its role in propagation of injury in conditions including ischemia and trauma could help in the development of new neuroprotective strategies.
This work was supported by National Institutes of Health Grants NS051288 (C.W.S.) and NS36548 (J.H.W.). We thank Drs. J. A. Connor and C. F. Valenzuela for helpful comments on the experiments and this manuscript.
- Correspondence should be addressed to Dr. Claude W. Shuttleworth, Department of Neurosciences, MSC08 4740, University of New Mexico School of Medicine, 1 University of New Mexico, Albuquerque, NM 87131-0001.