Arachidonic acid induces both Na+ and Ca2+ entry resulting in apoptosis

Authors


  • Drs Ming-Jai Su and Mei-Lin Wu have equal contribution to this work.

Address correspondence and reprint requests to Dr Mei-Lin Wu, Institute of Physiology, College of Medicine, National Taiwan University, No. 1, Section 1, Jen-Ai Road, Taipei, Taiwan.
E-mail: meilin@ntu.edu.tw

Abstract

Marked accumulation of arachidonic acid (AA) and intracellular Ca2+ and Na+ overloads are seen during brain ischemia. In this study, we show that, in neurons, AA induces cytosolic Na+ ([Na+]cyt) and Ca2+ ([Ca2+]cyt) overload via a non-selective cation conductance (NSCC), resulting in mitochondrial [Na+]m and [Ca2+]m overload. Another two types of free fatty acids, including oleic acid and eicosapentaenoic acid, induced a smaller increase in the [Ca2+]i and [Na+]i. RU360, a selective inhibitor of the mitochondrial Ca2+ uniporter, inhibited the AA-induced [Ca2+]m and [Na+]m overload, but not the [Ca2+]cyt and [Na+]cyt overload. The [Na+]m overload was also markedly inhibited by either Ca2+-free medium or CGP3715, a selective inhibitor of the mitochondrial Na+cyt-Ca2+m exchanger. Moreover, RU360, Ca2+-free medium, Na+-free medium, or cyclosporin A (CsA) largely prevented AA-induced opening of the mitochondrial permeability transition pore, cytochrome c release, and caspase 3-dependent neuronal apoptosis. Importantly, Na+-ionophore/Ca2+-free medium, which induced [Na+]m overload, but not [Ca2+]m overload, also caused cyclosporin A-sensitive mitochondrial permeability transition pore opening, resulting in caspase 3-dependent apoptosis, indicating that [Na+]m overload per se induced apoptosis. Our results therefore suggest that AA-induced [Na+]m overload, acting via activation of the NSCC, is an important upstream signal in the mitochondrial-mediated apoptotic pathway. The NSCC may therefore act as a potential neuronal death pore which is activated by AA accumulation under pathological conditions.

Abbreviations used
AA

arachidonic acid

ANT

adenine nucleotide translocase

ARC

AA-regulated Ca2+

CsA

cyclosporin A

CyP

cyclophilin D

cytC

cytochrome c

EPA

eicosapentaenoic acid

ER

endoplasmic reticulum

FFA

free fatty acid

mPTP

mitochondrial permeability transition pore

MTG

MitoTracker Green

NSCC

non-selective cation conductance

OA

oleic acid

rNa+-Ca2+ exchanger

reverse mode of the Na+-Ca2+ exchanger

ROS

reactive oxygen species

TMRM

tetra-methyl rhodamine-methyl ester

Z-DEVD.fmk

Z-Asp-Glu-Val-Asp-fluoromethyl ketone

In neurons, arachidonic acid (AA) plays many important roles under both physiological and pathological conditions. For example, tetanic stimulation or direct application of glutamate has been shown to activate NMDA receptors, resulting in dose-dependent release of AA (Miller et al. 1992), which is suggested to act as an retrograde messenger and is involved in sustained changes in synaptic efficacy and long-term potentiation (see Katsuki and Okuda 1995 for review). Moreover, AA has stimulatory or inhibitory effects on a variety of ion channels, including the NMDA receptor channel, voltage-gated Na+, Ca2+, and K+ channels, and Ca2+-dependent K+ channels (Miller et al. 1992; Denson et al. 2000; Liu and Rittenhouse 2003; reviewed by Meves 1994; Katsuki and Okuda 1995).

In non-excitable cells, AA directly opens the AA-regulated Ca2+ (ARC) channel, which is not permeable to Na+ ions in the presence of Ca2+ and Mg2+ ions (Mignen and Shuttleworth 2000; Mignen et al. 2001, 2003). The I/V curve for the ARC is non-linear and shows inward rectification with a very positive (> +30 mV) reversal potential.

During traumatic brain injury, brain ischemia, or convulsion, massive amounts of AA (10- to 13-fold increase) are released (Siesjo et al. 1982; Dhillon et al. 1997; reviewed by Katsuki and Okuda 1995) and are suggested to be involved in the ischemic-induced neuronal cell death (Dhillon et al. 1997; reviewed by Katsuki and Okuda 1995). Although it is difficult to determine the actual concentration of AA during ischemic brain injury, concentrations of 30–300 μmol/L have been suggested (Takeuchi et al. 1991; Lipton 1999).

In non-neuronal cell lines, AA induces apoptosis via a mitochondrial-mediated pathway (Scorrano et al. 2003; Penzo et al. 2004), which opens the mitochondrial permeability transition pore (mPTP), resulting in release of cytochrome c (cytC) (Scorrano et al. 2003), followed by activation of caspase 3-dependent nuclear condensation and fragmentation (Petronilli et al. 2001; Halestrap 2005). The mPTP is formed from a complex of the voltage-dependent anion channel (outer membrane), adenine nucleotide translocase (ANT; inner membrane), and cyclophilin D (CyP-D, matrix). Cyclosporin A (CsA), which prevents the binding of CyP-D to ANT (Crompton 1999; Halestrap 2005), has a marked protective effect against ischemia-induced CA1 hippocampal damage in the rat (Uchino et al. 1995).

A study using a brain slice model showed that ischemia induces intracellular [Na+]i and [Ca2+]i overload (Calabresi et al. 1999). Moreover, a microfluorometric study demonstrated that anoxia induces an increase in [Na+]i in hippocampal neurons by an unknown mechanism (Friedman and Haddad 1994). By inhibiting all possible AA-modulated ion channels, we show, for the first time in neurons, that AA induces cytosolic Na+ ([Na+]cyt) and Ca2+ ([Ca2+]cyt) overload, resulting in mitochondrial [Na+]m and [Ca2+]m overload, via a non-selective cation conductance (NSCC). Moreover, the AA-induced [Na+]m overload is an important upstream signal in mitochondrial-mediated, caspase 3-dependent apoptosis. A role for the AA-activated NSCC as a potential death pore is therefore suggested.

Material and methods

Preparation of cerebellar granule cells

All procedures were performed in accordance with the Animal Care Guidelines of the National Taiwan University. In brief, 6-day-old Wistar rats (both sexes) were killed by decapitation, and cerebellar granule cell cultures were prepared as previously described (Chen et al. 1999).

Chemicals and solutions

The normal Tyrode solution was 118 mmol/L NaCl, 4.5 mmol/L KCl, 1.0 mmol/L MgCl2, 2.0 mmol/L CaCl2, 10 mmol/L glucose, and 10 mmol/L HEPES, pH adjusted to 7.4. The Ca2+-free medium was normal medium lacking CaCl2 and containing 0.1 mmol/L EGTA. In Na+-free medium, the Na+ ions were isotonically replaced with N-methyl-d-glucamine. Moreover, 10 μmol/L MK-801, 10 μmol/L nifedipine, and 1 μmol/L ω-conotoxin GVIA were always present in the solution to abolish a potential activation of AA-activated NMDA receptors, or L-/N-types of Ca2+ channels (Miller et al. 1992; Liu and Rittenhouse 2003; see Meves 1994 for review), which are all expressed in granule cells (Miller et al. 1992; Pearson et al. 1995). As AA is easily to be oxidized resulting in reduction in its potency, 10 mmol/L stock of AA was freshly prepared in ethanol under argon and was used within 1 day.

All fluorescent indicators were purchased from Molecular Probes (Eugene, OR, USA). Z-Asp-Glu-Val-Asp-fluoromethyl ketone (Z-DEVD.fmk) and FITC-DEVD.fmk were obtained from BioVision (Mountain View, CA, USA). Primary polyclonal antibody to cleaved caspase 3 was purchased from Cell Signaling Technology (Beverly, MA, USA). Anti-cytC monoclonal antibody (clone 6H2.B4) was obtained from Promega (Madison, WI, USA). The TUNEL kit was purchased from Roche Molecular Biochemicals (Mannheim, Germany). All other chemicals were purchased from Sigma (St Louis, MO, USA).

Microfluorometry: intracellular fluorescence measurement of the [Ca2+]i and [Na+]i

The [Ca2+]i or [Na+]i was continuously monitored using fura-2AM- or SBFI-AM-loaded neurons. In vivo calibration was performed as described previously (Chen et al. 1999). In brief, cells were loaded for 60 min at 25°C with 5 μmol/L fura-2AM. The ratio of the emission at 510 nm with the excitation wavelengths, respectively, of 340 and 380 nm was calculated and converted to [Ca2+]i using the following equation (Grynkiewicz et al. 1985):

image

Where R is the ratio of the 510 nm fluorescence at 340 nm excitation over that at 380 mm.

The [Na+]i in neurons was measured by in vivo calibration (Chen et al. 1999). In brief, the cells were loaded with 5 μmol/L SBFI-AM for 90 min at 25°C, then excited alternately with 340 and 380 nm wavelength light, the emitted light being measured at 510 nm. Calibration solutions containing increasing concentrations of Na+ (0–140 mmol/L) were prepared by mixing different proportions of a high Na+ solution (115 mmol/L sodium gluconate, 25 mmol/L NaCl, 1 mmol/L EGTA, 10 mmol/L HEPES, and 11 mmol/L glucose) and a high K+ solution (115 mmol/L potassium gluconate, 25 mmol/L KCl, 1 mmol/L EGTA, 10 mmol/L HEPES, and 11 mmol/L glucose), both containing the Na+ ionophore cocktail, gramicidin D (5 μmol/L), monensin (40 μmol/L), and strophanthidine (100 μmol/L), and adjusting the pH to 7.4 at 37°C. Rmax (140 mmol/L Na+) and Rmin (Na+-free) represent, respectively, the maximum and minimum ratio values for the data curve. Using a Hanes plot of the data from 12 calibration experiments, the mean apparent dissociation constant (Kapp) was found to be 22.6 mmol/L, close to the value determined in other studies (Park et al. 1999). The following equation was used to convert the fluorescent ratio into the [Na+]i:

image

Simultaneous measurement of cytosolic and mitochondrial Ca2+ or Na+ changes in a single granule cell using time-lapse confocal microscopy

Thin optical sectioning (0.04 μm) was used to obtain independent mitochondrial and cytoplasmic fluorescent signals using a Leica TCS SP confocal laser scanning imaging system (Mannheim, Germany) equipped with a 63× oil-immersion objective. Mitochondria were identified by staining for 60 min at 25°C with a mitochondrial marker, MitoTracker Green (MTG; 200 nmol/L; excitation 488 nm/emission 525 ± 25 nm). As the nuclear membrane is no barrier to cytosolic ion movement (Collins et al. 2001), averaging the signal, respectively, over a small nuclear optical section [endoplasmic reticulum (ER)-/mitochondria-free] and the mitochondria simultaneously measures Ca2+/Na+ changes in both cytosol and mitochondria (Trollinger et al. 1997; Collins et al. 2001).

Cells were loaded for 1 h at 25°C with 5 μmol/L Rhod-2AM (a Ca2+-sensitive probe) or 30 min at 25°C with 100 nmol/L CoroNa Red (a Na+-sensitive probe), both of which are largely compartmentalized in mitochondria. A single neuron was excited at a wavelength of 568 nm using a 75 mW argon/krypton ion laser, and emission was measured at 630 ± 30 nm. The emitted light was separated into green (MTG) and red (Rhod-2/CoroNa Red) using a 565 nm dichroic mirror. To reduce photobleaching by laser illumination, the excitation intensity was reduced to ∼10–15%. The XY plane images were 512 by 512 pixels. Data analysis and processing were performed as described previously (Trollinger et al. 1997; Collins et al. 2001). Respective changes in [Ca2+]cyt/[Ca2+]m or [Na+]cyt/[Na+]m were simultaneously measured as peak/basal fluorescence ratios (F/F0).

In situ detection of apoptotic machinery

  • Tetra-methyl rhodamine-methyl ester (TMRM) was used to measure the persistent depolarization of the mitochondria, which is suggested to be an indicator of CsA-sensitive mPTP opening (Petronilli et al. 2001). Cells were loaded for 20 min at 25°C using a low dose of TMRM (10 nmol/L). Dynamically changes in the Ψm (ΔΨm) were then monitored by time-lapse confocal microscopy with excitation at 568 nm and emission at a wavelength longer than 580 nm.
  • • cytC release. After various treatments, the cells were fixed, permeabilized, incubated for 10 min at 25°C with blocking solution (5% non-fat milk), and then stained for 1 h at 37°C with mouse monoclonal anti-cytC antibody (1 : 500 dilution). They were then incubated for 1 h at 37°C with FITC-conjugated goat anti-mouse IgG antibody and for 15 min at 25°C with Hoechst 33342 (2 μg/mL), then examined by confocal microscopy.
  • • In situ labeling of activated caspase 3.
  • (i) Primary polyclonal antibody to cleaved caspase 3 (diluted 1 : 50), which recognizes 17/19 kDa of activated caspase 3 but not full-length caspase 3 (i.e. procaspase 3) or other cleaved caspases, was used according to the instructions of the manufacturer. Cells were then labeled for 1 h at 37°C with FITC-conjugated goat anti-rabbit IgG.
  • (ii) Caspase activation was examined using FITC-DEVD.fmk, which binds irreversibly to the catalytic site of activated caspase 3 in cells.
  • • Nuclear condensation or fragmentation. After various treatments, nuclei were labeled with Hoechst 33342 (nuclear condensation) or TUNEL (DNA/nuclear fragmentation) according to the manufacturer’s instructions.

Statistics

In fluorescence measurement and patch-clamp experiments, all results are expressed as the mean ± SEM for the stated number of animal preparations (n), each tested in duplicate. In immunostaining experiments, 200 cells from 10 randomly selected fields were scored on each coverslip, at least four animal preparations were used (n = 4) and duplicated in each test. Statistical differences were compared using the Mann–Whitney U-test, taking a p value of < 0.05 as significant.

Results

AA-induced intracellular Ca2+ and Na+ overload, which is not caused by AA metabolites or a reverse mode of the Na+-Ca2+ exchanger

The [Ca2+]i or [Na+]i was monitored using microfluorometry. The resting [Ca2+]i and [Na+]i was 75 ± 10 nmol/L (n = 28) and 7.7 ± 1.8 mmol/L (n = 21), respectively. Addition of 3–7 μmol/L AA in normal medium had only a small, or no, effect on intracellular levels of [Ca2+]i (Fig. 1a) or [Na+]i (Fig. 1b), but addition of 10–30 μmol/L AA resulted in a significant increase in both (Fig. 1a, b, and h). Using patch-clamp technique (Fig. S1, see also Supplementary Methods and Supplementary Results), we found that AA-induced [Ca2+]i and [Na+]i overloads were probably via an activation of a NSCC. Although concentrations of 30–300 μmol/L have been suggested to be released during ischemic brain injury (Takeuchi et al. 1991; Lipton 1999), the commonly used lower concentration of 10 μmol/L AA (reviewed by Meves 1994) was applied in the following experiments.

Figure 1.

 Arachidonic acid (AA), oleic acid (OA), and eicosapentaenoic acid (EPA) induce intracellular Na+ ([Na+]i) and Ca2+ ([Ca2+]i) overload. No Ca2+-ATPase or Na+-K+-ATPase inhibitors were added. (a–d) AA per se, but not AA metabolites, induces [Ca2+]i and [Na+]i overload. (a and b): Dose response of intracellular [Ca2+]i and [Na+]i overload induced by 3–30 μmol/L AA. (c–d): Effect of 10 μmol/L AA or 5,8,11,14-eicosatetraynoic acid (ETYA; in 0.1% ethanol) on the [Ca2+]i or of AA on the [Na+]i in the presence of 100 μmol/L indomethacin to inhibit cyclo-oxygenase, 10 μmol/L MK-886, and 10 μmol/L baicalein to inhibit, respectively, 5- or 12-lipoxygenase, and 30 μmol/L SK&F96365 to inhibit cytochrome P450. (e–h) OA and EPA induce a smaller increase in the [Na+]i and [Ca2+]i. In (h), where indicated, Na+-free medium was used to inhibit the rNa+-Ca2+ exchanger. The concentrations of AA, ETYA, OA, and EPA were all 10 μmol/L. The values shown in (h) are the mean ± SEM for at least four different animal preparations (= 5). *,#p < 0.05, compared with the basal level by the Mann–Whitney U-test.

The conversion of AA to active metabolites which induce NSCC activation, resulting in the [Ca2+]i and [Na+]i overload, is unlikely, as 5,8,11,14-eicosatetraynoic acid (red line, Fig. 1c), a non-metabolized AA analog, had a similar effect to AA (Fig. 1a). Moreover, a mixture of metabolic inhibitors of cyclo-oxygenase (indomethacin), 5- and 12-lipoxygenases (MK-886 and baicalein), and cytochrome P450 (SK&F96365) did not inhibit the entry of either Ca2+ (Fig. 1c) or Na+ (Fig. 1d). When these inhibitors were used singly, none had an inhibitory effect on the AA-induced [Ca2+]i overload (not shown). Under Na+-free conditions [which completely inhibit the reverse mode of the Na+-Ca2+ exchanger (rNa+-Ca2+) exchanger in the plasmalemma], the [Ca2+]i overload was not reduced (Fig. 1h), indicating that activation of the rNa+-Ca2+ exchanger, resulting from the [Na+]i overload, was not involved in the AA-induced [Ca2+]i overload (Fig. 1a). It has been shown in non-excitable cells that both Gd3+ (< 1 μmol/L) and La3+ (< 10 μmol/L) inhibit the AA-activated ARC (Mignen et al. 2001, 2003). However, neither Gd3+ nor La3+ (both were 100 μmol/L) inhibit the AA-induced [Ca2+]i overload in neurons (data not shown).

In addition to AA (C20:4), we tested another two unsaturated free fatty acids (FFAs), oleic acid (OA; C18:1), and eicosapentaenoic acid (EPA; C20:5). Compared with AA (Fig. 1h), 10 μmol/L OA induced a smaller increase in the [Ca2+]i and [Na+]i (Fig. 1e, f, and h). Interestingly, 10 μmol/L EPA induced a small change in the [Ca2+]i, but a significant increase in the [Na+]i (Fig. 1g and h). Whether these two FFAs open a similar NSCC to that opened by AA (see Fig. S1) requires further investigation.

Simultaneous measurement of cytosolic and mitochondrial Ca2+ or Na+ changes using confocal microscopy in live neurons

As AA induced intracellular [Ca2+]i and [Na+]i overload (Fig. 1), we next investigated whether [Ca2+]i and [Na+]i overload resulted in mitochondrial Ca2+ ([Ca2+]m) and/or Na+ ([Na+]m) overload in neurons.

Confocal microscopy of a thin optical section (0.04 μm) can be used to simultaneously monitor changes in different intracellular compartments of live cells (Trollinger et al. 1997). As the nuclear membrane does not represent a barrier to cytosolic ion movement (Collins et al. 2001), the mean fluorescent intensity over a small nuclear optical section (ER- and mitochondria-free; ‘N’ in the last frames in Fig. 2a and c) allows dynamic changes in the [Ca2+]cyt to be recorded, while averaging small optical section over MTG(+) (a specific mitochondrial marker; last frames in Fig. 2a and c) allows the [Ca2+]m to be recorded. Because variability in dye loading causes differences in absolute fluorescent intensity between neurons, the signal increase is shown as peak/basal fluorescence ratio (the F/F0).

Figure 2.

 AA induces [Ca2+]cyt and [Ca2+]m overload and the [Ca2+]m overload is blocked by RU360: time-lapse confocal microscopy in live single neurons. (a and b) AA (10 μmol/L) induced an increase in the [Ca2+]m and [Ca2+]cyt in two Rhod-2AM-loaded neurons, using nuclear levels to represent cytosolic Ca2+ levels ([Ca2+]cyt) (see text). (c and d) RU360, a mitochondrial Ca2+ uniporter inhibitor, markedly inhibited the AA-induced [Ca2+]m overload, but not the [Ca2+]cyt overload. RU360 (10 μmol/L) was used as a 60 min pre-treatment and was present during recording. The bottom right panels in (a and c) show the localization of mitochondria (MTG, green; ‘N’ for nucleus) and the numbers indicate the optical section area of the mitochondria analyzed. The color scale is shown below the left column in (a and c). The arrows in (a and c) indicate the time of AA addition. (e) Summary histogram (data from a to d). Under Na+-free conditions, the AA-induced [Ca2+]cyt and [Ca2+]m overload were not affected. Note the marked reduction in the [Ca2+]m, but not the [Ca2+]cyt, overload in RU360-pre-treated cells. Ca2+-ionophore (1 μmol/L ionomycin in Na+-free medium) also induced [Ca2+]cyt and [Ca2+]m overload. The mean value of F/F0 for three to four regions of the mitochondria in each of neuron was calculated. The values shown are the mean ± SEM for at least five different animal experiments. *p < 0.05 by the Mann–Whitney U-test.

In normal medium, the resting levels of the [Ca2+]cyt and the [Ca2+]m were both F/F0 = 1. When 10 μmol/L AA was added at 300 s, sustained increases in the [Ca2+]cyt (F/F0 = 6) and the [Ca2+]m (F/F0 = 7) were seen (Figs 2a and b, Video S1). In Ca2+-free medium, AA had little effect on either the [Ca2+]cyt or the [Ca2+]m (bottom trace in Fig. 2b; Fig. 2e), indicating that AA induced little Rhod-2 bleach or membrane leak within the recording period. Pre-treatment with RU360, a potent mitochondrial Ca2+ uniporter inhibitor (Gincel et al. 2001), markedly inhibited the AA-induced [Ca2+]m overload (Fig. 2c and d; p < 0.05, blue bars in Fig. 2e), confirming that the mitochondrial Ca2+ uptake is mainly via the Ca2+ uniporter (Rizzuto et al. 2000; Brustovetsky et al. 2002). As in Fig. 1g, Na+-free conditions had little inhibitory effect on either AA-induced [Ca2+]cyt or [Ca2+]m overload (Fig. 2e). Ionomycin (1 μmol/L), a Ca2+-ionophore, induced over 10-fold increase in both [Ca2+]cyt and [Ca2+]m (in Na+-free medium, Fig. 2e), confirming that mitochondria are a strong buffer for [Ca2+]cyt overload in neurons (Budd and Nicholls 1996; Wang and Thayer 1996; Murchison and Griffith 2000).

A similar method to that above, but using CoroNa Red, was used to simultaneously measure changes in the [Na+]cyt and [Na+]m. In normal medium, the basal levels of the [Na+]cyt and the [Na+]m were both F/F0 = 1. After exposure to AA in normal medium, the [Na+]cyt and [Na+]m both markedly increased (Fig. 3a, b, and e; Video S2), whereas, in Na+-free medium, AA had little effect (bottom trace, Fig. 3b), indicating that CoroNa Red is a Na+-sensitive and Na+-selective probe. Surprisingly, RU360 caused a significant, but only partial, decrease in the AA-induced [Na+]m overload, but not in the [Na+]cyt overload (Fig. 3e).

Figure 3.

 AA induces both [Na+]cyt and [Na+]m overload and the [Na+]m overload is markedly inhibited by CGP3715 or Ca2+-free medium. (a–d) The recording methods were similar to those in Fig. 2, but measured Na+ (see Materials and methods). The arrow indicates the time of addition of 10 μmol/L AA (a and c). The bottom right panels in (a and c) show the localization of mitochondria (MTG, green; ‘N’ for nucleus) and the numbers indicate the mitochondria analyzed. (c and d) Under Ca2+-free conditions, Na+-ionophore cocktail also induced both [Na+]cyt and [Na+]m overloads. The cocktail consisted of 5 μmol/L gramicidin D, 40 μmol/L monensin, and 100 μmol/L strophanthidine in 100 mmol/L Na+o and clamped the [Na+]i at the level of the [Na+]o (i.e. [Na+]o = [Na+]i). (e) Summary histogram (data from a to d). Both CGP3157 (10 μmol/L, a selective inhibitor of the mitochondrial Na+-Ca2+ exchanger) and Ca2+-free medium had a significant inhibitory effect on the AA-induced [Na+]m overload, but not the [Na+]cyt overload. The values shown are the mean ± SEM for at least five different animal preparations. *p < 0.05 by the Mann–Whitney U-test.

The mechanism by which RU360 inhibits both [Ca2+]m and [Na+]m overload in neurons is unclear. It has been shown that the mitochondrial Ca2+m-Na+cyt exchanger (mitochondrial matrix Ca2+ exchange for cytosolic Na+) is activated when the [Ca2+]m increases (Crompton et al. 1978; Murchison and Griffith 2000). A role of the exchanger in the AA-induced [Na+]m overload was therefore investigated using CGP37157, a selective inhibitor of the Ca2+m-Na+cyt exchanger (Rizzuto et al. 2000), or using Ca2+-free/EGTA medium to completely abolish the AA-induced [Ca2+]m overload. Both treatments markedly inhibited the AA-induced [Na+]m overload, but had little effect on the [Na+]cyt overload (Fig. 3e). These results suggest that the inhibitory effect of RU360 on the [Na+]m overload is secondary to a reduction in the Ca2+m uptake by RU360 and that activation of the Ca2+m-Na+cyt exchanger is involved in the AA-induced [Na+]m overload.

Next, we used the Na+-ionophore cocktail in Ca2+-free/EGTA medium to clamp the [Na+]i at the level of the [Na+]o (i.e. [Na+]o = [Na+]i, see Materials and methods). Using this cocktail medium containing 100 mmol/L Na+, the [Na+]cyt and [Na+]m showed a marked increase (Fig. 3e). However, RU360 did not inhibit the Na+-ionophore-induced [Na+]cyt or [Na+]m overload (not shown), so the Na+ influx pathway is possibly different from that seen using AA.

The [Na+]m overload is involved in AA-induced mPTP opening, resulting in cytC release and activation of the caspase 3-dependent apoptotic machinery

As [Ca2+]i overload has been suggested to induce cell necrosis (membrane disruption) (McConkey 1998), we tested whether [Ca2+]i overload evoked by AA-induced neuron necrosis. Either propidium iodide (25 μg/mL) or ethidium homodimer (1 μmol/L) was added in the bath solution and we did not observe any nucleus staining after 10 μmol/L of AA treatment for 24 h (data not shown). Thus, AA-induced [Ca2+]i overload did not result in neuron necrosis. Moreover, there was little membrane leak induced by long-term AA application.

There is also evidence that [Ca2+]m overload is one of the major causes of mPTP opening resulting in cytC release in cortical neurons (Gunter et al. 2000; Brustovetsky et al. 2002). The following experiments investigated whether [Na+]m and/or [Ca2+]m overload played a role in the activation of mPTP-mediated neuronal apoptosis.

Tetra-methyl rhodamine-methyl ester was used to measure the persistent depolarization of the mitochondria, which is suggested to be an indicator of CsA-sensitive mPTP opening (Petronilli et al. 2001). FCCP, a potent proton ionophore, induces CsA-/mPTP-insensitive mitochondrial depolarization (Abramov et al. 2004), a fact also demonstrated in the present study (100%ΔΨm depolarization, Fig. 4a). Addition of AA, however, evoked CsA-sensitiveΔΨm depolarization [Fig. 4a(iii)], indicating that AA-induced mPTP opening. RU360, which inhibited both the [Ca2+]m and [Na+]m overload (Figs 2 and 3), largely abolished the AA-induced ΔΨm depolarization [Fig. 4a(ii and iii)], indicating that [Ca2+]m and/or [Na+]m overload might play a role in mPTP opening. Interestingly, Na+-ionophore/Ca2+-free medium also induced CsA-sensitive mitochondrial depolarization, indicating that Na+ overload per se induces mPTP opening [Fig. 4a(iii)].

Figure 4.

 The AA-induced [Ca2+]m and [Na+]m overload results in CsA-sensitive mPTP opening, cytC release, and caspase 3 activation. (a) AA or Na+-ionophore/Ca2+-free medium induced CsA-sensitive mPTP opening. [a(i-iii)] 10 μmol/L AA induced RU360- [a(ii)] and CsA-sensitive [a(iii)] mitochondrial depolarization. The complete depolarization (100%) of the mitochondrial potential [a(i and ii)] caused by 1 μmol/L FCCP was CsA (1 μmol/L)-insensitive [a(iii)]. Changes in mitochondrial potential (ΔΨm) was measured by TMRM. The values shown are the mean ± SEM for at least five different animal preparations. (b and c) Cells were exposed to 10 μmol/L AA, OA, or EPA for 90 min, washed for 4 h (b) or 16 h (c), and labeled with either anti-cytC antibody (b), FITC-conjugated DEVD.fmk [c(vii)] or anti-cleaved caspase 3 antibody [c(i–v and viii)]. When cells were treated with Na+-ionophore/Ca2+-free medium for 30 min, then transferred to Ca2+-free medium, similar cytC release and caspase 3 activation were seen [b(vii) and c(vii)]. Ca2+-ionophore/Na+-free medium also induced CsA-sensitive cytC release and caspase 3 activation [b(vii) and c(vii)]. CytC release (arrows in b) and cleaved caspase 3 (arrows in c) were examined using confocal microscopy. Noted that either OA or EPA also induced less cleaved caspse-3 (+) staining. The control experiment [first bar, b(vii) and c(vii and viii)] was performed in the presence of 0.1% of ethanol as vehicle. The AA- or Na-ionophore-induced cytC release [reduction in green fluorescence in the mitochondria, arrows in b(ii and vi)] were markedly inhibited by 10 μmol/L RU360, Ca2+-free-, Na-free-treatment, or 1 μmol/L CsA, but not by a mixture of ROS inhibitors (100 μg/mL of catalase, 1 mmol/L phenanthroline, 150 U/mL of SOD, and 10 mmol/L tiron). Z-DEVD.fmk (a caspase 3 inhibitor, 100 μmol/L pre-treatment for 2 h) inhibited AA-induced caspase 3 activation [c(v)]. In band c(vii, viii), at least 200 cells from 10 randomly selected fields were scored in each experiment to determine the percentage; the values shown are the mean ± SEM for at least five rats. *,#p < 0.05 compared with the AA [*b and c(vii and viii)], ionophores [#b and c(vii)], or vehicle [#c(viii)] group by the Mann–Whitney U-test.

After exposure to AA for 90 min followed by wash-off of AA for 4 h (cytC studies) or 16 h (caspase studies), both cytC release [Fig. 4b(ii and vii)] and caspase 3 activation [measured by DEVD-FITC, Fig. 4c(vii)] were seen. Both effects were markedly inhibited by pre-treatment with RU360, Ca2+-free medium, Na+-free medium, or CsA [Fig. 4b(vii) and c(vii)]. CGP37157 was not used because the neuronal cells became unhealthy when they were exposed to CGP37157 for more than 2 h. A caspase 3 inhibitor, Z-DEVD.fmk, inhibited the AA-induced caspase 3 activation [Fig. 4c(v)], but not the cytC release (not shown), indicating that cytC release is upstream of caspase 3 activation. Ionomycin (1 μmol/L; a Ca2+ ionophore in Na+-free medium) evoked CsA-sensitive cytC release and caspase 3 activation [Fig. 4b(vii) and c(vii)]. Interestingly, Na+-ionophore (containing 50 mmol/L or 100 mmol/L Na+ in Ca2+-free medium) also induced CsA-sensitive cytC release and caspase 3 activation [Fig. 4b(vii) and c(vii)].

It has been suggested that, in glial cells, AA induces reactive oxygen species (ROS) overproduction, resulting in cell toxicity (Wang et al. 2004). We also observed that AA-induced O2˙, but not OH˙ or H2O2, increase (Fig. S2b, see also Supplementary Methods and Supplementary Results), However, when ROS inhibitor was used, little inhibition of either [Ca2+]i or [Na+]i overload was seen (Fig. S2f). Not surprisingly, AA-induced either cytC release, caspase 3 activation or nuclear condensation was not inhibited by a mixture of ROS inhibitors [Figs 4b(vii) and c(vii) and 5a(vii)].

DEVD-FITC is not specific for caspase 3 staining, as it may also bind to caspase 7. Anti-cleaved caspase 3 antibody, which recognizes the 17/19 kDa fragments of caspase 3, but not procaspase 3 or other cleaved caspases, was therefore used. Figure 4c(i–iv) show that AA induced Na+- and Ca2+-sensitive caspase 3 activation (arrows). Moreover, OA and EPA [both 10 μmol/L, Fig. 4c(v–vi), arrows] also induced positive staining for caspase 3. The statistical data are shown in Fig. 4c(viii).

After exposure to AA for 90 min and washing for 24 h, significant nuclear condensation [arrows in Fig. 5a(ii)], Hoechst staining) and DNA fragmentation [Fig. 5c, green for TUNEL (+) staining] were seen. Both effects were significantly inhibited by treatment with RU360, Ca2+-free medium, Na+-free medium, or CsA, but not by a mixture of ROS inhibitors (Fig. 5a and c), and were also inhibited by Z-DEVD.fmk, indicating the nuclear changes were caspase 3-dependent. As OA and EPA induced a smaller increase in the [Na+]i and [Ca2+]i (Fig. 1h), not surprisingly, they also induced less nuclear condensation, which could be partially inhibited by Na+-free and Ca2+-free medium [Fig. 5b(vii), Hoechst staining]. Ionomycin (1 μmol/L), again, evoked marked CsA-sensitive apoptotic nuclear changes [Fig. 5a(vii) and c(vii)]. Furthermore, Na+-ionophore/Ca2+-free medium also induced nuclear condensation and fragmentation and both effects were CsA-dependent.

Figure 5.

 The [Na+]m- and [Ca2+]m-overloads are involved in the AA-, OA-, or EPA-induced nuclear condensation and fragmentation. Cells exposed to 10 μmol/L AA, OA, or EPA for 90 min, then returned to AA-/FFAs-free medium for 24 h showed Z-DEVD.fmk-sensitive nuclear (Nu.) condensation (arrows in a, b, and c; Hoechst staining) and nuclear fragmentation [arrows in C(ii), TUNEL (+) staining, green fluorescence]. The control condition [first bar, a(vii), b(vii), and c(vii)] was carried out in the presence of 0.1% of ethanol as vehicle. RU360, Ca2+-free medium, Na+-free medium, or CsA all markedly inhibited the AA-, OA-, or EPA-induced nuclear changes. A mixture of ROS inhibitors (catalase, phenanthroline, SOD, and tiron) again had little inhibitory effect. Na+-ionophore/Ca2+-free medium or ionomycin (1 μmol/L, a Ca2+ ionophore) induced CsA-sensitive nuclear condensation and DNA fragmentation (both treatments for 30 min followed by 24 h wash-off of the ionophore). The cells were double-labeled with (a) Hoechst 33342 (blue) or (b) TUNEL reagent (green). In [a, b, and c(vii)], the values shown are the mean ± SEM for at least four different animal preparations. *,#p < 0.05 compared with the AA (*), OA (*), EPA (#), or ionophore (#) group by the Mann–Whitney U-test.

These results indicate that not only the [Ca2+]m overload, but also the [Na+]m overload, plays an important role in mPTP opening, resulting in cytC release and caspase 3-dependent apoptosis, as shown in the schematic diagram in Fig. 6.

Figure 6.

 AA or other types of FFAs possibly opens a death pore, resulting in neuronal apoptosis. AA activates an NSCC, resulting in both cytosolic and mitochondrial [Na+] and [Ca2+] overload. Activation of the mitochondrial Na+cyt-Ca2+m exchanger and an undefined [Na+]cyt influx pathway are both involved in the AA-induced [Na+]m overload. The RU360-sensitive [Na+]m and [Ca2+]m overloads activate the mitochondrial-mediated apoptotic machinery, including CsA-sensitive mPTP opening, cytC release, and caspase 3-dependent nuclear condensation and fragmentation in neurons.

Discussion

During brain ischemia, a large amount of AA (30–300 μmol/L) is released and this has been suggested to be involved in neuronal cell death (Siesjo et al. 1982; see reviews by Katsuki and Okuda 1995 and Dhillon et al. 1997). Using brain slice model, ischemia-induced membrane depolarization is associated with an increase of both [Na+]i and [Ca2+]i, which are not affected by tetradotoxin or by glutamate receptor antagonists in neurons, whereas ischemia-induced depolarization is reduced by lowering the external Na+ concentration (Calabresi et al. 1999). Moreover, microfluorometric studies have demonstrated that anoxia also induces an increase in [Na+]i in rat CA1 hippocampal neurons (Friedman and Haddad 1994).

Our results show that AA-induced Na+ influx may be involved in the neuronal cell death seen during brain ischemia, as (i) 10 μmol/L AA induces both cytosolic and mitochondrial Na+/Ca2+ overload (Figs 1–3), possibly via activation of an NSCC (Fig. S1); (ii) The mitochondrial Ca2+ uniporter and Na+cyt-Ca2+m exchanger play important roles in the AA-induced [Na+]m overload (Fig. 3); and (iii) Although mitochondria have a huge capacity for removing high levels of cytosolic Ca2+/Na+, accumulation of Ca2+/Na+ ions in the matrix can eventually damage them, resulting in CsA-sensitive mPTP opening, cytC release, and induction of caspase 3-dependent neuronal apoptosis (Figs 4 and 5). A summary diagram is shown in Fig. 6.

There is ample evidence that AA can modulate NMDA and voltage-gated ion channels in neurons (Miller et al. 1992; Denson et al. 2000; Liu and Rittenhouse 2003; for reviews, see Meves 1994; Katsuki and Okuda 1995). Under conditions in which all possible AA-modulated ion channels in granule cells were inhibited, the biophysical properties of the AA-activated NSCC were different from those of the ARC channel in non-excitable cell lines, because (i) the I/V curve for the ARC channel shows significant inward rectification and its reversal potential is +30 mV, whereas that for the NSCC was linear without rectification and its reversal potential was close to zero, and the reversal potential was shifted to the left in Na+-free medium and (ii) the Ca2+/Na+ permeability ratio of the NSCC was low (Ca2+/Na+ = 0.5), indicating that the Na+ permeability of the NSCC is much higher than the Ca2+ permeability, whereas ARC channels show Ca2+ and only very little Na+ permeability in the presence of Ca2+ and/or Mg2+ ions (Mignen and Shuttleworth 2000; Mignen et al. 2001, 2003). The differences between the NSCC and ARC are probably because of the expression of different channel isoforms. However, there is possibility that the NSCC activated by AA is not a protein channel, as it seemed to behave like a non-selective cation ‘pore’ which could be opened when AA accumulated (30–300 μmol/L), as occurs in certain pathological conditions (Takeuchi et al. 1991; Lipton 1999), as (i) no current inactivation of the NSCC was seen within the 400 ms voltage-clamp period (Fig. S1) and the AA-induced Ca2+/Na+ overload was long-lasting (Fig. 1) and (ii) the current density of the AA-induced NSCC was much higher than normally seen for non-selective cation channels. However, the AA-induced currents were not leak currents, as there were no fluorescent probe leaks during ∼60 min of AA application (Fig. 1a and e), a longer recording period than normally used in patch-clamp recording. Furthermore, after treatment with AA for 24 h, neither propidium iodide nor ethidium homodimer influx was seen (unpublished observations), indicating that AA did not induce ‘leak’ of the plasma membrane. As at least two other FFAs, OA and EPA, also induced a similar, but smaller, increase in the [Na+]i and the [Ca2+]i, indicating that the pore (or NSCC) is probably not specifically opened by AA. The ion permeability and biophysical properties of these AA-/FFA-induced NSCCs require further investigation.

We also showed that AA induces [Ca2+]cyt/[Na+]cyt and [Ca2+]m/[Na+]m overload and apoptosis, including cytC release and caspase 3-dependent nucleus condensation and fragmentation, and that either Ca2+-free or Na+-free medium blocked induction of apoptosis (Figs 4 and 5). These results indicate that either [Ca2+]cyt/[Ca2+]m or [Na+]cyt/[Na+]m overload are involved in the apoptosis. Moreover, RU360, which prevented mitochondrial [Ca2+]m/[Na+]m overload, but not cytosolic [Ca2+]cyt/[Na+]cyt overload (Figs 2 and 3), blocked the AA-induced apoptotic pathway (Figs 4 and 5), this shows that [Ca2+]m and/or [Na+]m overload, but not cytosolic [Ca2+]cyt and [Na+]cyt overload, is involved.

Possible molecular mechanisms suggested for AA-induced apoptosis are that (i) AA induces ER Ca2+ release followed by mitochondrial Ca2+ uptake, resulting in opening of the CsA-sensitive mPTP in non-neuronal cell lines (Scorrano et al. 2003), or (ii) in the absence of [Ca2+]m overload, AA directly activates the voltage sensor in the mPTP in isolated mitochondria (Scorrano et al. 2001). In neurons, AA-induced ER Ca2+ release was negligible, as there was little increase in the [Ca2+]cyt or [Ca2+]m when Ca2+o influx was inhibited under Ca2+-free conditions (Fig. 2e). Direct activation of the voltage sensor in the mPTP by AA also seems unlikely, as RU360, Ca2+-free medium, or Na+-free medium prevented the AA-induced CsA-sensitive mPTP opening and cytC release (Fig. 4 and 5).

It has been proposed that [Ca2+]m overload per se triggers a conformational change in the ANT (inner membrane) and that this is facilitated by bound CyP-D, resulting in mPTP opening, and that CsA prevents the binding of matrix CyP-D to the ANT (Crompton 1999; Halestrap 2005). Not surprisingly, ionomycin, which evoked [Ca2+]m overload in Na+-free medium (Fig. 2e), also caused CsA-sensitive cytC release and caspase 3-dependent apoptosis [Figs 4c(vii)-5b(vii)]. Interestingly, Na+ ionophore cocktail (containing 50 or 100 mmol/L NaCl in Ca2+-free medium) also stimulated similar apoptotic response, indicating that [Na+]m overload per se can also activate mitochondrial-dependent apoptosis. An important unanswered question is how mitochondrial Na+ overload directly or indirectly activates the mPTP complex. Clearly, additional experiments are needed to elucidate the mechanism.

Both Ca2+-free medium and Na+-free medium largely inhibited AA (or OA) activation of the apoptotic pathway (Figs 4 and 5). This raises the important question of why Na+-free conditions, in which AA-induced [Ca2+]i/[Ca2+]cyt/[Ca2+]m overload was not reduced (Figs 1g and 2e), totally abolished AA-induced apoptosis. Two possible explanations are that (i) although the [Ca2+]m in Na+-free medium was not reduced (F/F0 = 6), this [Ca2+]m is probably not high enough to activate CsA-sensitive apoptosis, in contrast to the ionomycin-induced [Ca2+]m (F/F0 > 10-fold, Fig. 2e) and (ii) under Ca2+-free conditions, the AA-induced [Na+]m overload was significantly reduced, probably by inactivation of the Na+cyt-Ca2+m exchanger (Fig. 3e). The inhibitory effect on cytC release, caspase 3 activation and nuclear changes seen in Ca2+-free medium is probably because of a reduction in the [Na+]m overload. Our results therefore suggest that the [Na+]m is an essential upstream signal for AA-/NSCC-induced neuronal apoptosis and that inhibition of the Ca2+ uniporter and the Na+cyt-Ca2+m exchanger is important in preventing mitochondrial-mediated neuronal apoptosis.

In summary, the present study shows that AA (and other types of FFAs) activates an NSCC, resulting in both cytosolic and mitochondrial Ca2+ and Na+ overload in neurons. The mitochondrial Ca2+ uniporter and the Na+cyt-Ca2+m exchanger are both involved in the [Ca2+]m and [Na+]m overload. In addition to the [Ca2+]m overload, the [Na+]m overload is an important upstream signal in AA-induced neuronal apoptosis. The NSCC may therefore act as a potential death pore, which is activated by AA accumulation in brain ischemia or trauma, and may be involved in neuronal cell death under pathological conditions.

Acknowledgements

We gratefully thank the 2nd Core Laboratory of the Department of Medical Research, National Taiwan University Hospital, and Major Instruments Co. Ltd. (Chin-Yung Wang, Chin-Hsiang Wang, and Wen-Chao Hsien) for expert technical assistance. This study was supported by the National Science Council of Taiwan (NSC 94-2320-B-002-005 and NSC 94-2320-B-002-061).

Ancillary