Address correspondence and reprint requests to Menahem Segal, Department of Neurobiology, The Weizmann Institute, Rehovot 76100, Israel. E-mail: email@example.com
Networks of neurons express persistent spontaneous network activity when maintained in dissociated cultures. Prolonged blockade of the spontaneous activity with tetrodotoxin (TTX) causes the eventual death of the neurons. In this study, we investigated some molecular mechanisms that may underlie the activity-suppressed slow degeneration of cortical neurons in culture. Already after 3–4 days of exposure to TTX, well before the neurons die, they began to express markers that lead to their eventual death, 7–10 days later. There was a reduction in glutamate receptor (GluR2) expression, a persistent increase in intracellular calcium concentration, activation of calpain, and an increase in spectrin breakdown products. At this point, blockade of GluR2-lacking GluR1 or calpain (either with a selective antagonist or through the natural regulator of calpain, calpastatin), protected cells from the toxic action of TTX. Subsequently, mitochondria lost their normal elongated shape as well as their membrane potential. Eventually, neurons activated caspase 3 and PUMA (p53 up-regulated modulator of apoptosis), hallmarks of neuronal apoptosis, and died. These experiments will lead to a better understanding of slow neuronal death, typical of neurodegenerative diseases.
Spontaneous, ongoing network activity plays a critical role in neuronal development, in that it leads to formation of synaptic connections among neurons, and their operation as a network. Suppression of network activity has detrimental consequences to the extent that the silenced neurons die following a prolonged period of lack of activity (Baker and Ruijter 1991; Ramakers et al. 1991; Fishbein and Segal 2007). On the other hand, intense network activity and its subsequent acute rise of intracellular calcium concentration ([Ca2+]i) causes the triggering of apoptotic death signaling pathways which has been used as a model system for traumatic as well as neurodegenerative brain diseases (Dirnagl et al. 1999; Hardingham and Bading 2003). The cell death following suppression of activity is slower and thus resembles the actual slow course of cell death in neurodegenerative diseases especially during the progression of such diseases as neurons are likely to lose afferents and become gradually silent. The identity of the molecular mechanisms responsible for this slow degeneration has not been elucidated as yet. While it is intuitively clear that hyperactivation of neurons and excessive accumulation of free [Ca2+]i lead to their quick death, why do neurons that do not fire action potentials, and thus do not influx calcium, die? Furthermore, as there are several death pathways in neurons (Yuan et al. 2003), which of them is activated in the tetrodotoxin (TTX)-induced cell death?
We examined these questions in dissociated cultures of cortical neurons exposed to TTX for periods of 4–12 days. In an earlier study (Fishbein and Segal 2007), we found that TTX causes a slow cell death that can be prevented when miniature excitatory post-synaptic currents (mEPSC) are blocked by a glutamate antagonist. We now propose that the residual synaptic activity is sufficient to cause a sustained intracellular calcium rise, activation of calcium-dependent protease, and mitochondria dysfunction, leading to apoptotic cell death.
Materials and methods
Cultures were prepared as detailed elsewhere (Papa et al. 1995). Briefly, Wistar rat pups were decapitated on postnatal day 3 and their brains were removed and placed in a chilled (4°C), oxygenated Leibovitz L15 medium (Biological Industries, Beit Haemek, Israel) enriched with 0.6% glucose and gentamicin (20 μg/mL; Sigma, St Louis, MO, USA). Bilateral cortical tissue was mechanically dissociated and plated on 12-mm glass coverslips at 4 × 105 cells per well in a 24-well plate or on 35-mm culture dishes with 1.5–2 × 106 cells. The plating medium consisted of 5% heat-inactivated horse serum (HS) and 5% fetal calf serum and was prepared in minimal essential medium (MEM)-Earl salts (Biological Industries) enriched with 0.6% glucose, gentamicin, and 2 mM glutamax. Cells were left to grow in the incubator at 37°C, 5% CO2 for 4 days, at which time the medium was changed to 10% HS in enriched MEM, with a mixture of 5′-fluoro-2-deoxyuridine/uridine (20 and 50 μg/mL, respectively; Sigma). The medium was changed 4 days later to 10% HS in enriched MEM. TTX (at 1 μM; Alomone Labs, Jerusalem, Israel) was added to the growth medium at 4 days in culture for up to 14 days of incubation period.
A lipofectamine 2000™ (Invitrogen, Carlsbad, CA, USA) mix was prepared at 1 μL/well with 50 μL/well optimem™ (Invitrogen) and incubated for 5 min at 25°C. This was mixed with 1.5 μg/well total DNA in 50 μL/well optimem™ and incubated for 15 min at 25°C. The mix was then added to the cells and allowed to rest for 4–6 h until medium replacement. In most cases, at least several neurons were transfected. Using this method, cotransfection efficiency for several plasmids was nearly 100%. For large-scale transfection, the nucleofection kit of Amaxa Biosystems (Gaithersburg, MD, USA) was used. Three micrograms of DNA was added to cells resuspended in 100 μL of 1 : 5 mixture of supplement and nucleofactor solution (respectively), and later the cells were electroporated using the nucleofactor device for single cuvettes. Cells were diluted in plating medium, seeded, and 2–4 h later their medium was replaced.
The cultures were transferred to a recording chamber placed in a Nikon-inverted microscope (Nikon, Melville, NY, USA) and washed with standard recording medium containing (in mM) NaCl 129, KCl 4, MgCl2 1, CaCl2 2, glucose 10, and HEPES 10, pH was adjusted to 7.4 with NaOH, and osmolarity was adjusted to 320 mOsm with sucrose. TTX (0.5 μM) and picrotoxin (20 μM) were also added to this medium for the recording of spontaneous miniature excitatory post-synaptic currents (mEPSCs). Neurons were recorded at 25°C with patch pipettes containing (in mM) K-gluconate 140, NaCl 2, HEPES 10, EDTA 0.2, Na-GTP 0.3, Mg-ATP 2, and phosphocreatine 10, and pH 7.4 having a resistance in the range of 6–12 MΩ. Signals were amplified with Axopatch 200A (Axon Instruments Inc., Foster City, CA, USA) and were stored on IBM PC (Molecular Devices, Sunnyvale, CA, USA). PClamp analysis software (Molecular Devices) was used for the off-line analysis of voltage/current protocols. For the analysis of mEPSCs, a 600 Hz low-pass filter was first applied (in Clampfit analysis), and the events were then analyzed using Mini-analysis software (Synaptosoft Inc., Fort Lee, NJ, USA), with a threshold set at 9 pA.
Cultures were incubated for 1 h at 25°C with the standard recording medium containing TTX and 2 μM fura-2AM (or Fluo-4AM for recordings of spontaneous activity). Cells were imaged thereafter on the stage of an inverted Olympus microscope (Olympus, Center Valley, PA, USA), equipped with a Till Photonics light source and an Andor Technology Ixon CCCD camera (Belfast, UK). This microscope was used also for the recording of phase images. Basal fluorescence levels to illumination at 340/380 nm were recorded in several fields of each cover glass. The results were presented as the ratio of fluorescence at 340/380 nm, which reflected the [Ca2+]i.
Immunocytochemistry and imaging
Cultures were washed briefly with standard recording medium, fixed in 4%p-formaldehyde with 4% sucrose for 20 min, and washed with phosphate-buffered saline. For 3,3′-diaminobenzidine staining, cells were blocked for 1 h at 25°C in 10% HS with 0.1% Triton X-100 and incubated overnight at 4°C with the primary antibody against neuronal nuclei (NeuN) (1 : 1000; Chemicon, Temecula, CA, USA) followed by labeling using Vectastain ABC kit (Vector Laboratories, Inc., Burlingame, CA, USA). Neurons were visualized on a Nikon Eclipse E800 microscope, and digitized pictures were taken with Nikon Act-1 software. For fluorescence staining, cells were blocked for 1 h in 10% goat serum (GS) with 0.1% Triton X-100 and incubated overnight at 4°C with mouse α-NeuN (1 : 1000; Chemicon) followed by 1 h secondary antibody labeling (Alexa 488-labeled and Alexa 546-labeled goat anti-mouse, 1 : 200; Molecular Probes, Eugene, OR, USA). Coverslips were washed again, transferred onto glass slides, and mounted for visualization with anti-fading mounting medium. Confocal image stacks were recorded using a Zeiss LSM 510 laser scanning microscope, a Zeiss 40× and 100× oil immersion objective (Zeiss, Thornwood, NY, USA). About five fields were taken per 12-mm coverslip for statistical analysis at each case.
Caspase 3 activity assay
Following TTX treatment, cortical neurons were harvested into a lysis buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, 0.5% Triton, and 10 mM dithiothreitol), centrifuged at 14 000 rpm, and the supernatant was collected. Protein concentration in the cleared lysate was determined using the Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA), and samples were tested for caspase 3 activity using the kit reagents of Calbiochem (caspase 3 fluorogenic substrate and caspase 3 inhibitor II; San Diego, CA, USA). Following 1 h incubation at 37°C in the dark, the fluorescence of the samples was read by the GENios microplate reader (Tecan, Switzerland) and the activity was calculated as fluorescence per μg protein.
Western blot analysis
After TTX treatment, cortical neurons were harvested into a lysis buffer (20 mM Tris–HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, and 0.1% Triton), centrifuged at 14 000 rpm, and supernatant was collected. Protein concentration in the cleared lysate was determined using the Bradford assay (Bio-Rad Laboratories). Samples containing equal amounts of protein were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred electrophoretically onto nitrocellulose membranes. The membrane was blocked in 0.1% Tween 20 Tris-buffered saline (137 mM NaCl and 20 mM Tris–HCl, pH 7.6) containing 10% skimmed milk. Primary antibodies against neuron specific enolase (NSE, rabbit, 1 : 500; Zymed, San Francisco, CA, USA) and BDPn150 (rabbit, 1 : 400; Bahr et al. 1995) were added overnight at 4°C in 0.1% Tween 20 Tris-buffered saline and 1% bovine serum albumin and immunoreactive proteins were visualized with horseradish peroxidase-conjugated protein A and enhanced chemiluminescence. Quantification of the bands was performed using the ‘Image J’ software (NIH, Bethesda, MD, USA).
Total cellular RNA was isolated using Qiagen RNeasy mini kit (Valencia, CA, USA) according to manufacture’s instructions. Reverse transcription was performed with oligo dT primers and the RevertAid RT enzyme (Fermentas, Burlington, ON, Canada) at 42°C for 60 min. The RT product from 500 ng of total RNA was applied to each 20-μL PCR reaction mixture containing 1x Absolute Blue Cyber Green PCR Mix (Thermo Scientific, Waltham, MA, USA) and 250 nm of gene-specific forward and reverse primers. Each experimental set consisted of duplicate samples, non-template controls, and serial dilutions of standard templates. PCR was performed in Rotor-Gene 6000 apparatus of Corbett (Qiagen, Valencia, CA, USA) with the following thermal cycle conditions: 15 min enzyme activation at 95°C and subsequent 40 three-step cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s. As a reference, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and microtubule-associated protein 2 (MAP2) cDNA levels were used. The following rat-specific primers were used: GAPDH, fwd: 5′-TCAACGGCACAGTCAAGGC-3′, rev: 5′-CGCTCCTGGAAGATGGTGAT-3′; MAP2, fwd: 5′-GTCAAGTCCAAAATCGGATCAA-3′, rev: 5′-ACCTGACCCCCCTTAGGCT-3′; PUMA (p53 up-regulated modulator of apoptosis): fwd 5′-GCGGCGGAGACAAGAAGA-3′, rev: 5′-GGAGTCCCATGAAGAGATTATACATG-3′; glutamate receptor (GluR)1, fwd: 5′-CAGCAGGTGCGCTTCGA-3′, rev: 5′-GTCC GGCGCCCTTTCT-3′, and GluR2, fwd: 5′-ACCCCGGAAGATTG GGTACT-3′, rev: 5′-CCAGACGTGTCATTTCCTGATG-3′. Emission data were analyzed by the Rotor-Gene program and cycle threshold (Ct) number was calculated automatically by it. We also verified that there was no detectable amplification in non-template controls. Ct numbers showed a significant linear reverse correlation with logarithmic concentrations of standard templates. The original copy numbers of the first strand cDNA in RT samples were calculated from the Ct numbers by fitting to the obtained standard correlation.
To identify mitochondria, cells were either transfected with green fluorescent protein (GFP) conjugated to mitochondrial tag sequence, or incubated for 30 min at 37°C with MitoTracker deep red (Molecular Probes) diluted 1 : 10 000 in recording solution. For the assessment of membrane potential, cells were incubated with 100 nM tetramethyl rhodamine methyl ester (TMRM) in recording solution for 25 min at 37°C. Fixed or live neurons were imaged using the confocal Zeiss LSM 510 laser scanning microscope. Membrane potential was calculated based on the ratio between fluorescence of the nucleus and mitochondria as assessed using the Image J software and the formula presented by Koopman et al. (2008).
Every experiment was repeated at least three times with different dissections, where several wells/35 mm dishes were used for each treatment condition. The morphological and immunocytochemical data were summarized and analyzed automatically using Image-Pro Plus software (Media Cybernetics, Inc., Bethesda, MD, USA). The stained neurons were counted in several fields of view in each glass. The comparisons were made using anova or t-tests as the case required. Significance was set at p < 0.05. Further, Tukey’s multiple comparisons were performed when needed.
The first series of experiments was designed to determine the time, during the long exposure to TTX, at which the neurons began to activate an irreversible death cascade. We were able to replicate earlier observations to show that 12 days of exposure to TTX killed 85% of the neurons in a well (Fig. 1) TTX-treated neurons died gradually over a 2-week period; after 4 days of TTX only 13.5 ± 2.5% of the neurons died whereas after 12 days of treatment 87 ± 5% of the cells were dead (p < 0.01) (Fig. 1a). To study the reversibility of the TTX effect on the neuronal viability, TTX was washed out extensively 4 days after its initial application and the cells were left in regular growth medium for a period of eight more days. Surprisingly, 4 days of exposure to TTX was sufficient to trigger marked death of the neuronal population measured 8 days later. Thus, removal of TTX did not help the cells to recover from its toxic effect and neurons continued to die in the 8 days following washout of the drug (Fig. 1b and c). Compared with continued exposure to TTX where 85 ± 2% of the neurons died, removal of TTX after 4 days of exposure still caused 65 ± 4% of the neurons to die 8 days later, in contrast to only 23 ± 5% that died immediately following 4 days of treatment (p < 0.01). Persistent activation of sodium channels during the wash period using the sodium channel agonist, veratridine (1 μM; Sigma), could not prevent neuronal death, and following 8 days in the presence of veratridine and the absence of TTX, 57 ± 1% of the neurons still died (Fig. 1c). These experiments indicated that the irreversible death cascade began at least 4 days after onset of exposure to TTX, as indicated by the fact that little recovery was seen upon removal of the toxic agent.
These results raise the following concern: is it possible that the cell death that continues even after removal of TTX is caused in fact by hyperactivity of the cells, being released from the suppressive action of the drug? To examine this we imaged spontaneous activity of cultured neurons using Fluo-4AM calcium-sensitive dye at various times after removal of TTX. Indeed, immediately after removal of TTX the spontaneous activity of neurons was far more intense than that of controls (Fig. 2a). However, within 48 h after removal of the drug, spontaneous activity subsided back to almost control level, leaving no signs of hyperactivity that may lead to cell death. Furthermore, chronic treatment of the culture with a GABA antagonist bicuculline also prompted a massive increase in network activity yet it did not cause any apparent cell death (data not shown).
We have already shown that TTX induces a persistent elevation of [Ca2+]i which can be the signal that leads to the activation of cell death cascade (Fishbein and Segal 2007). We therefore measured [Ca2+]i following 4 days of exposure to TTX. Indeed, there was a significant (p < 0.001) elevation of 20% in [Ca2+]i in these cells relative to controls (Fig. 2b), measured with fura-2AM dye. Following 8 days of treatment, the [Ca2+]i was still 20% higher than control (p < 0.001), and removal of TTX for at least 3 days after 5 days of treatment was not associated with a full recovery of [Ca2+]i back to control levels, although it was significantly lower (p < 0.001) by 13% than the [Ca2+]i of the continuously treated cells (Fig. 2c). This indicates that the sustained elevation of [Ca2+]i may indeed be involved in subsequent activation of cell death cascade.
Calcium may enter cells via different voltage-gated and receptor-coupled ion channels, one of which is the α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptor. The AMPA receptor is composed of four subunits (GluR1–4) and the GluR2 subunit is the one to determine the receptor permeability to Ca2+ ions because of editing process which confer the channel with Ca2+ impermeability (Pellegrini-Giampietro et al. 1997; Tanaka et al. 2000). As a result, GluR2-lacking AMPA receptors are calcium permeable, and this is served as an additional route for Ca2+ entry into the cell. Real-time PCR analysis revealed a consistent reduction in the mRNA expression level of the GluR2 subunit in correlation with the duration of the TTX treatment (Fig. 3a). While after 3 days of treatment only 10% of the neurons had died (as seen by the MAP2 mRNA level), the normalized GluR2 mRNA level was already reduced by 30% compared with control. At later stages, when the neuronal population was reduced to 58% of control, the normalized GluR2 level continued to stay low relative to control, and accounted for only 65% of control, meaning that the degree of loss of the GluR2 subunit exceeded the one of the neuronal-specific gene, indicating that the effect seen is not just because of the reduction in neuronal numbers, but rather a genuine loss of this subunit in the destined-to-die neurons. In contrast to the GluR2, the GluR1 subunit expression was elevated at the early stages of TTX treatment by 50% (supporting previous observations; Fishbein and Segal 2007; Fig. 3b), but later it was reduced back to baseline levels (p < 0.001), and after 8 days of TTX treatment there was no change in the expression level of GluR1 between treated cells and controls. The ratio of GluR2/GluR1 which was 0.6 after 3 days of TTX treatment (41% of the corresponding control, p < 0.05) compared with 1.5 ± 0.2 in control culture of the same age and 0.55 ± 0.15 after 8 days of TTX (61% of the corresponding control) further demonstrated that GluR2 was highly down-regulated in the TTX-treated cells compared with GluR1 (Fig. 3c). If indeed the down-regulation of the GluR2 subunit and the subsequent Ca2+ entry trigger the death cascade in the neurons, then blocking these Ca2+-permeable channels should attenuate the death process. Indeed, application of 100 μM of the GluR2-lacking GluR1 receptor antagonist, 1-naphthyl acetyl spermine (Sigma), for 8 days of TTX treatment enhanced cell survival rates, from 33 ± 0.5% in its absence to 57 ± 4% in its presence (Fig. 3d, p < 0.01). This result is similar, albeit less striking than the effect of AMPA receptor blockade reported before (Fishbein and Segal 2007).
The high calcium levels observed in the affected neurons made the calcium-dependent protease calpain an obvious target candidate for involvement in this death process, as was shown to be the case in other processes of neurodegeneration and cell death (Higuchi et al. 2005; Raynaud and Marcilhac 2006; Wales et al. 2008). Several experiments were designed to examine the involvement of calpain in the death cascade. First, the amount of the 150 kDa calpain-mediated spectrin breakdown product was assessed using a specific antibody in western blot (Fig. 4a). Following 5 days of TTX treatment, cells were harvested and protein was extracted and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) analysis. The results indicated that during the course of treatment, cultures accumulated 90% more calpain-specific spectrin breakdown products than controls, indicating an increased calpain activity in the treated cells. To further test the involvement of calpain activation in the death process, we exposed the TTX-treated cultures to the pharmacological blocker of calpain, PD150606 (Calbiochem, Darmstadt, Germany) (Fig. 4b). PD150606 at 100 μM was able to block almost completely the elevation in the spectrin breakdown product in treated cells compared with controls as assessed on western blot. Following 10 days of treatment cells were fixed and stained with the neuronal-specific marker, NeuN. While the drug had no effect of its own on survival of neurons in culture, it induced a marked protection from TTX toxicity and elevated significantly (p < 0.001), the survival rates of the neurons from 10 ± 1% without the inhibitor to 49 ± 3% in its presence (Fig. 4c).
A more specific way to examine the role of calpain in the death process involved the transfection of neurons with the endogenous inhibitor of calpain, calpastatin (Goll et al. 1992), and comparison of these cells with cells transfected with a control GFP plasmid. This procedure allowed us to follow the morphology of the transfected neurons in the course of exposure to TTX as well as compare their viability in culture. Figure 5 demonstrates that cells over-expressing calpastatin indeed show better survival rates following 12 days of TTX treatment relative to control GFP-transfected neurons as assessed by NeuN staining (52 ± 4% cells survived, compared with 24 ± 5%, Fig. 5a and b, p < 0.05). Furthermore, the percentage of cells expressing the calpastatin plasmid among the TTX-treated cells was twice as large as that of the other groups (Fig. 5c, p < 0.01) which supported our assumption that under death-promoting conditions, calpastatin exerted a protective effect on the neurons.
In the process of their death, TTX neurons retract their dendrites. This aspect of dendritic retraction was also governed by calpain as indicated in the sholl analysis presented in Fig. 6a and b. Cultures transfected with either GFP or GFP/calpastatin were treated with TTX for 9 days, after which live neurons were examined under the confocal microscope and their dendrites were counted at fixed distances from the soma. The analysis showed that calpastatin-expressing neurons experienced reduced levels of dendritic loss following exposure to TTX compared with controls, indicating that calpain activation contributes to the shrinkage of dendrites.
We then examined if calpain activation is instrumental in the initial increase in miniature synaptic currents following exposure to TTX, seen by us and others (Turrigiano et al. 1998; Fishbein and Segal 2007). Strikingly, while there was no difference in mEPSC amplitudes among the four groups (data not shown), in both calpastatin- and GFP-transfected neurons, exposure to TTX for about a week caused a marked increase in mEPSCs amplitude, and in that respect there was no difference between calpastatin and GFP-transfected neurons (Fig. 6c and d). These results indicate that calpain is not likely to be involved in the early homeostatic response of neurons to the blockade of action potential discharge, and it becomes instrumental only in subsequent stages, resulting from the mEPSC-induced increase in [Ca2+]i in the affected neurons. These converging experiments also indicate that calpain activation plays a significant role in the death cascade evoked by suppression of activity induced by TTX.
Mitochondrial damage is one of the hallmarks of the apoptotic cascade (Newmeyer and Ferguson-Miller 2003). We suspected that TTX-induced cell death was also apoptotic as the dying cells lack of necrotic characteristics. We therefore searched for evidence for mitochondria defects in the TTX-treated cells. Figure 7a and b focus on the cell soma of TTX-treated cells compared with controls, using either mito-GFP (mitochondrial-directed GFP) or the mitochondria-specific dye mito-tracker, and the same morphological changes can be observed in both cases.
Mitochondrial-directed GFP-transfected neurons were treated with TTX for a week after which they were imaged at high resolution and a close look at their dendrites revealed that their mitochondria were round and small compared with the long strips-like mitochondria of control cells (Fig. 7c).
The fact that the mitochondria are disrupted still does not mean that they are non-functional. To examine this more directly we used the TMRM dye which is known to equilibrate between functional mitochondria and cytosol in a membrane potential-dependent manner, and thus it allows for quantification of mitochondrial membrane potential (Koopman et al. 2008). Figure 8 shows that among cells that were exposed for 5 days to TTX the dye was distributed throughout the cytosol, in contrast to its specific localization inside the mitochondria of healthy, control cells. Quantification of the membrane potential of the mitochondria using the fluorescence ratio of TMRM inside mitochondria and nucleus and the Nernst equation showed that TTX reduced the membrane potential by 33% compared with control (−80 mV relative to −120 mV, respectively, p < 0.001). The exposure to the calpain inhibitor PD150606 could not prevent mitochondrial damage and the deterioration of its membrane potential (p < 0.001), indicating that mitochondrial deterioration was upstream or parallel to calpain activation.
As mitochondrial dysfunction also appeares in other types of cell death we searched for induction of unique cellular pathways characteristic of apoptosis in the TTX-treated cells. Two of these pathways found in the cells were activation of caspase 3 and induction of the pro-apoptotic gene, PUMA, both of which were shown already to participate in neuronal death (Yuan and Yankner 2000; Niizuma et al. 2009). First, cells grown in the presence or absence of the calpain inhibitor, PD150606, were harvested following 6 days of TTX treatment and their supernatant was extracted and used for measurement of caspase 3 activity and for protein amount determination. TTX induced a significant (p < 0.01) 5 (±0.3)-fold increase in the activity of caspase 3 relative to control, which was reduced to 2 ± 0.06-fold when calpain activity was blocked (Fig. 9a). Second, cells were grown in the presence or absence of the caspase 3 inhibitor ZVAD [Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone]-fmk (Sigma) and were subjected to 7 days of TTX treatment following which control and treated cultures were fixed and stained for the specific neuronal marker, NeuN. Cell counts indicated that the application of the caspase 3 inhibitor was able to elevate cell survival rates relative to control from 55 ± 2% to 88 ± 16% (Fig. 9b). These complementary experiments indicated that caspase 3 was activated by TTX, and that this activation contributed to cell death.
Real-time PCR analysis showed that already at 3 days of TTX treatment when almost no neuronal loss could be detected (94 ± 12% MAP2 expression in treated cells compared with control), PUMA expression level was already elevated by 77 ± 15% in treated cells compared with control (Fig. 9c). This elevation was persistent throughout the TTX treatment, and after 8 days of treatment when 40 ± 9% of the neurons had died, there was still an increase in the amount of the gene expressed in the remaining neurons. To determine if calpain was responsible for this p53-dependent activation, we repeated the experiment in the presence of the calpain blocker, PD150606 (Fig. 9d). Surprisingly, calpain blockade was not able to reduce back to normal the normalized levels of PUMA which remained high (96 ± 8% more without PD150606, p < 0.05; 74 ± 1% more with PD150696, p < 0.001).
The present results indicated that a prolonged exposure to TTX caused a sustained suppression of network activity and resulted in a massive death of cultured cortical neurons. Taken together with our previous results (Fishbein and Segal 2007), this data indicates that the suppression of network activity, by itself was not the main cause of the cell death, as suppression of activity using glutamate antagonists did not mimic the toxic effect of TTX, and in fact protected the cells from the toxic action of TTX (Fishbein and Segal 2007). The suppression of activity is likely to cause an imbalance between GluR1 and GluR2 such that the spontaneous activation of GluRs via TTX-insensitive release of miniature synaptic quanta can cause a sustained increase in [Ca2+]i, which is sufficient to trigger a death cascade. The death cascade continues with activation of the calcium-dependent protease calpain and the p53-dependent gene, PUMA, leading to loss of mitochondrial membrane potential, and ending with the activation of caspase 3.
While the activation of apoptotic death cascades has been studied extensively in many neuronal and non-neuronal cell populations, the current results are unique in that they represent a slow process of cell death that once is triggered, many days before the actual death of these neurons, they are destined to die, even if the triggering signal is long gone. This indicates that the cascade of events begins with a fairly robust activation that takes several days to develop but once it evolves, it is irreversible. In that respect, the TTX-induced reduction of GluR2 provides an important clue: GluR2-lacking neurons are ubiquitous in the brain, primarily as interneurons, but these are endowed with most efficient calcium buffering mechanisms that prevent their death following synaptic activation (McBain and Dingledine 1993; Vissavajjhala et al. 1996). On the other hand, once normal pyramidal neurons lose GluR2, they are unable to cope with even small elevations of [Ca2+]i, to the extent that they eventually die (Oguro et al. 1999; Gorter et al. 1997; Liu et al. 2004). This small but persistent increase in [Ca2+]i may not be as dramatic as that produced by exposure of neurons to high concentration of glutamate, but it is probably sufficient to cause an eventual degeneration of the exposed neurons. In that respect, the slow degeneration of neurons is more likely to model the slow death seen in neurodegenerative diseases than the acute glutamate intoxication.
One of the immediate consequences of a calcium rise in the cell is the activation of calcium-dependent protease, calpain. In the dying neurons, calpain is shown to actively participate in the death process, and its inhibition by either PD150606 or calpastatin, protects the cells against TTX toxicity, as was shown already to happen in an Alzheimer’s disease mouse model (Rao et al. 2008). Calpain activation seemed to be downstream to the activation of p53 and to mitochondrial damage as calpain inhibition could not prevent the elevated PUMA expression and the reduction in mitochondrial membrane potential (Figs 8 and 9). It was shown recently that synaptic activity suppresses p53 activation and inhibits mitochondrial permeability transition (Lau and Bading 2009), so it may well be that the initial amount of Ca2+ entering the cells through the calcium permeable AMPA receptors is not sufficient to activate calpain but is able and activate p53 pathway. Once the mitochondria are damaged, Ca2+ leaks into the cytosol and activates calpain. The involvement of calpain makes once again the comparison between neuronal death induced by hyperactivity to the one caused by lack of activity particularly relevant, as different amounts of [Ca2+]i can both cause the activation of the same intracellular pathways leading eventually to a similar outcome of cellular death, although in different periods of time and intensity.
The slow neuronal death observed in our experiments seems to be apoptotic in its nature because of the mitochondrial dysfunction, the caspase 3 activation, and the expression of the pro-apoptotic p53-dependent gene, PUMA. Mitochondrial damage expressed both as morphological fragmentation and functional depolarization, is a key feature of the neuronal pathology (DiMauro and Schon 2008). TTX-treated neurons clearly experience both aspects of mitochondria dysfunction, already at a fairly early stage (following 5 days of treatment) when morphologically the mitochondria is still intact, the membrane potential in the treated cells is depolarized relative to controls, which may imply that the fragmentation process and the loss of the intact mitochondrial network of the cell is a later stage in the degeneration process, which reflect the non-functionality of the affected mitochondria.
As for the early induction of PUMA, it was shown already that it plays a role in delayed death of hippocampal neurons (Niizuma et al. 2009) and that its transcriptional activation is sufficient to mediate p53 apoptotic effects in cortical neurons (Uo et al. 2007). Hence, it strengthens the assumption that the course of death observed in TTX-treated cortical neurons is indeed apoptotic and may at least in part be mediated by p53.
We thank V. Greenberger and S. Shtreim for the preparation of the cultures, Dr E. Korkotian for help with the imaging system, Dr N. Kosower for the calpastatin plasmid, and Dr B. Bahr for the antibody to the BDPn150. This study was supported by a grant from the Israel Science Foundation.