| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Université de Caen, UMR 6232, CNRS-CEA, Cyceron, Caen, France; and
Department of Physiology and Neurobiology, Marquette University, Milwaukee, Wisconsin, USA
1Correspondence: UMR 6232 Centre National de la Recherche Scientifique–Université de Caen, GIP CYCERON, Bd Henri Becquerel, BP 5229 14074 Caen, France. E-mail: buisson{at}cyceron.fr
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: excitotoxicity • glutamate uptake • memantine • ERK signaling
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
. Under pathological conditions, Ca2+ can also trigger neurotoxic cascades (4)
. For example, excessive Ca2+ entry leading to neuronal Ca2+ overload is a key early step in glutamate-induced cell death during hypoxic/ischemic injury (5
, 6)
. Among the different routes of Ca2+ entry, N-methyl-D-aspartate receptors (NMDARs), a subtype of ionotropic glutamate receptors, are of particular interest because of their special ability to gate high levels of Ca2+ influx (7
, 8)
. These receptors have been extensively studied for their critical role in synaptic plasticity and neuronal damage occurring during acute brain injuries and neurodegenerative diseases (9
10
11)
. It is now well accepted that the Ca2+ overload induced by NMDARs overactivation leads to lethal cellular derangements ranging from perturbations in signaling pathways, mitochondrial function, and morphological alterations of dendrites and cell body (12
, 13)
. This unique relation between NMDARs activation and neuronal death is challenged by their parallel role in neuronal plasticity and neurotrophic processes (14
15
16)
. While NMDARs are mainly found on the postsynaptic membrane at excitatory synapses (17
, 18)
, they are also present at extrasynaptic positions (19)
. These distinct populations of NMDARs have been identified in primary cultures of cortical, hippocampal, and cerebellar granule neurons (19
20
21)
, as well as in hippocampal slices (22)
. Recent reports provide evidence that one of the key determinants of NMDAR versatility on neuronal fate may be their cellular location. Whereas Sattler et al. (20)
reported that synaptic and extrasynaptic NMDAR activation participates equally in excitotoxicity, Hardingham et al. (21)
suggested that only extrasynaptic NMDAR activation triggers neuronal death. More recently, the hypothesis that NMDARs’ cellular location governs their influence on neuronal survival was challenged by Liu et al. (23)
, who suggested that subunit composition of NMDARs, rather than cellular location, was a determinant for their effect on neuronal fate. Indeed, the localization of NMDARs seems to affect their functional properties; differences in NR2 subunit composition of synaptic and extrasynaptic NMDARs have been suggested. Although NR2A-containing receptors were originally described as located at the synapse and NR2B-containing receptors extrasynaptically (23
, 24)
, recent data have challenged this view and reported that NR2A- and NR2B-containing NMDARs were present at both synaptic and extrasynaptic sites (25)
. The aim of the present study was to design experimental paradigms to induce selective activation of either synaptic or extrasynaptic NMDARs. Then, we characterized the consequences of these activations on the beneficial aspect (ERK signaling) and on neurotoxic influences of NMDAR activation (mitochondrial function, morphological abnormalities, neuronal death). Finally, we studied the relation between synaptic and extrasynaptic NMDARs and determined the inhibition profile of the clinically well-tolerated NMDAR antagonist, memantine, a compound that was approved by the European Union and the U.S. Food and Drug Administration for the treatment of dementia.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cell cultures
Primary cortical neurons cultures were prepared from fetal mice at 15–16 days of gestation (26)
. Cerebral cortices were then dissected, dissociated, and resuspended in DMEM supplemented with 5% FBS, 5% horse serum and 2 mM glutamine and plated in 24-well plates or glass-bottom Petri dishes previously coated with poly-D-lysine and laminin. Cultures were kept at 37°C in a humidified atmosphere containing 5% CO2. Experiments were performed on cultures after 14 days in vitro (DIV).
Calcium imaging
Mice primary cortical neurons cultures (DIV 7 and 14) were loaded for 45 min at 37°C with 10 µM Fura-2/AM resuspended in 0.2% pluronic acid solution and incubated for an additional 15 min at room temperature in HEPES and bicarbonate buffered saline solution (HBBSS) containing (in mM) 116 NaCl, 5.4 KCl, 1.8 CaCl2, 0.8 MgSO4, 1.3 NaH2PO4, 12 HEPES, 5.5 glucose, 25 bicarbonate and 10 µM glycine at pH 7.45. Experiments were performed at room temperature with continuous perfusion at 2 ml/min with a peristaltic pump, on the stage of a Nikon Eclipse inverted microscope equipped with a 100-W mercury lamp and oil-immersion Nikon x40 objective with 1.4 numerical aperture (Nikon, Tokyo, Japan). Fura-2 (excitation: 340/380 nm, emission: 510 nm) ratio images were acquired every 2 s with a digital camera (Princeton Instruments, Trenton, NJ) using Metafluor 6.3 software (Universal Imaging Corporation, West Chester, PA, USA). Fluorescence ratios (340/380 nm) were converted to intracellular Ca2+ concentration using the following formula (27)
: [Ca2+]i = Kd[(R–Rmin)/(Rmax–R)] F0/Fs, where R is the measured ratio of 340/380 fluorescence, Rmin is the ratio measured in a Ca2+-free solution, Rmax is the ratio measured in a saturated Ca2+ solution, Kd = 135 nM (the dissociation constant for Fura-2), and F0 and Fs are the fluorescence intensities measured, respectively, in a Ca2+-free solution at 380 nm or in a saturated Ca2+ solution at 380 nm.
Immunocytochemistry
Cell cultures were fixed in a sucrose-containing solution (0.5 M) with 4% paraformaldehyde, then washed in phosphate-buffered solution (PBS) and incubated for 1 h in the presence of PBS containing 0.1% Triton-X100 and 4% bovine serum albumin (BSA) to saturate nonspecific sites. Primary antibodies against phosphorylated forms of ERKs and MAP-2 were incubated overnight at 4°C in PBS with 0.1% Triton-X100 and 1% BSA under gentle agitation. Alexa 488-conjugated anti-rabbit (1:500) and Alexa 555-conjugated anti-mouse (1:500) were further incubated for 1 h. Observations were performed on a Nikon Eclipse (TE2000-E) inverted C1 confocal microscope equipped with an oil immersion Nikon x60 objective with 1.4 numerical aperture, and images were acquired with Nikon EZC1 software.
Mitochondrial membrane potential measurements
Cortical neurons were incubated with 10 µg/ml rhodamine 123 for 15 min at 37°C in HBBSS supplemented with 10 µM glycine at pH 7.45. Experiments were performed on a Nikon Eclipse (TE2000-E) inverted C1 confocal microscope, and images were acquired with Nikon EZ-C1 Software. Fluorescence intensity measurements were normalized to the maximum rhodamine 123 fluorescence intensity obtained after exposing the neuronal cell cultures to carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP; 5 µM). FCCP is a protonophore (H+ ionophore), and uncoupler of oxidative phosphorylation in mitochondria that depolarizes mitochondrial membranes.
Excitotoxicity and assessment of neuronal cell death
Excitotoxic injuries were induced at 37°C by either 15 µM NMDA or 50 µM bicuculine + 2.5 mM 4-AP for 1 h in DMEM supplemented with 10 µM glycine. Cultures were washed and returned to standard DMEM. Twenty-four hours after the onset of glutamate agonist exposure, neuronal death was estimated by examination of the cultures under phase-contrast microscopy and quantified by measurement of lactate dehydrogenase (LDH) released by dead neurons into the bathing medium. The LDH level corresponding to the maximal neuronal death (without glial death) was determined in sister cultures exposed to 200 µM NMDA (LDHmax
75 AU). Background LDH levels were determined in sister cultures subjected to control wash (LDHmax10 AU). Experimental values were measured after subtracting LDHmin and then normalized to LDHmax – LDHmin in order to express the results in percentage of neuronal cell death.
Morphological analysis of GFP-labeled cortical neurons
Primary cortical neurons cultures (3 DIV) were transfected with enhanced-green fluorescent protein (eGFP) coding cDNA with the phosphate-calcium technique. Cells were exposed to 40 µg/ml DNA for 2 h and then shocked with dimethyl sulfoxide containing solution for 2 min. Cells were washed and returned to DMEM supplemented with 5% FBS, 5% horse serum and 2 mM glutamine. Experiments were performed on 7 and 14 DIV neuronal cultures. Spine quantification was performed on images acquired at a resolution of 1024 x 1024 pixels and was achieved by counting protrusions, whose head was localized at less than 5 µm from the dendritic shaft on five different dendritic segments of 50 µm. Dendrites and cell body alterations were quantified on time-lapse images acquired with Nikon EZ-C1 software at a resolution of 512 x 512 pixels. Thirty z sections of 1 µm thickness were taken to create a z stack from top to bottom of the specimen every 3 min for 30 min, and the brightest-points projection was generated. Quantification of cell body swelling and dendrite varicosity was performed on 3 independent experiments using MetaMorph 6.3 software (Universal Imaging Corporation) after image segmentation and morphometric filtering.
Western blot analysis
Neuronal cultures were harvested in a lysis solution containing 50 mM Tris-HCl (pH 7.6), 1% Nonidet P-40 (Sigma), 150 mM NaCl, 2 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride (PMSF) in the presence of a protease inhibitor cocktail (1:100; Sigma). Cell lysates were centrifuged, and the supernatant was isolated. Protein concentration was determined by Bradford assay. Electrophoresis was performed on 12% NuPAGE Bis-Tris polyacrylamide gels (Invitrogen), and proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Polyscreen membrane, Perkin Elmer, Paris, France). Membranes were incubated in a 0.1% Tween 20 and Tris 200-mM buffered solution (TTBS) complemented with 5% nonfat dry milk for 1 h and incubated with the p-ERK antibody (1:1000) at 4°C overnight under gentle agitation. Incubation with the secondary peroxidase-conjugated anti-mouse antibody (1:10,000) (Sigma) was performed for 1 h at room temperature. Blots were detected with an enhanced chemiluminescence Western blot detection system (Western Lightning Chemiluminescence Reagent Plus, PerkinElmer). Blots were then incubated in a stripping buffer (62 mM Tris HCl, pH 6.8; 2% SDS; and 100 mM β-mercaptoethanol) for 30 min at 50°C, followed by incubation in TTBS containing 5% fat free dry milk, and probed with an antibody against ERK protein (1:1000) in the same conditions as described for p-ERK immunoblots.
Statistical analysis
Results are expressed as means ± SE. Statistical analysis was performed with StatView (Abacus, Berkeley, CA, USA) by two-way analysis of variance (ANOVA) followed by a PLSD Fisher test (P<0.05). Statistical analysis for Western blot was performed by a two-tail paired t test on Excel (Microsoft Corporation, Redmond, WA, USA). N represents the number of independent experiments and n the number of regions of interest analyzed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. By blocking GABAA receptor function with bicuculine (50 µM) in the presence of 4-aminopyridine (4-AP; 2.5 mM), a weak potassium-channel blocker, we revealed by Fura-2 calcium imaging that neurons displayed bursts of action potentials with transient Ca2+ increases while application of NMDA (25 µM) triggered a single peak increase in intracellular Ca2+ concentration (Fig. 1a, b
). Next, we estimated the amount of calcium that penetrates into neurons at two developmental stages: 7 and 14 days in vitro (DIV). The results showed that the developmental stage of mouse primary cortical neurons cultures greatly influences the [Ca2+]i modifications induced by synaptic stimulation. Indeed, at DIV 7, synaptic stimulation paradigm only induced a modest [Ca2+]i increase, whereas at DIV 14, similar stimulation induced larger [Ca2+]i increase (Fig. 1a-c
). Because the technique we used to monitor [Ca2+]i measures fluorescence intensity variations at the cell body level, we investigated whether these measurements reflected an increase in synapses number between 7 and 14 DIV. Quantification of spine density revealed a close correlation between developmental stage and synaptic-mediated Ca2+ response. At DIV 14, spine density was 5-fold higher than at DIV 7 and synaptic-mediated calcium response was 6-fold higher (Fig. 1d
). These results confirmed that the developmental stage greatly influences synaptic response and that changes in synaptic activity can be monitored at the cell body level with Fura-2 calcium imaging. To confirm that the blockade of GABAA receptor function induces an increase in presynaptic release of excitatory neurotransmitters and a subsequent increase of postsynaptic activity, we incubated neuronal cultures with tetanus toxin (1 µg/ml for 24 h), a neurotoxin characterized for its ability to prevent fusion of neurotransmitter-containing vesicles. This treatment abolished neuronal Ca2+ increase induced by bicuculine/4-AP application but poorly affected NMDA-mediated Ca2+ increase (Fig. 2a-c
). Similarly, Ca2+ increase induced by bicuculine/4-AP application was fully blocked by TTX (0.5 µM), a voltage-dependent sodium channel blocker (Fig. 2d
). These results confirm that Ca2+ increase measured during synaptic stimulation paradigm relies on action potential-driven synaptic release of neurotransmitters. Coapplication of AP-5 (100 µM), a selective NMDAR antagonist, induced a 90% inhibition of intracellular Ca2+ increase (Fig. 2e
), and coapplication of nimodipine (5 µM), a selective L-type voltage-sensitive Ca2+ channels blocker (Fig. 2f
) reduced by 10% the Ca2+ response. Together, these results demonstrate the major contribution of NMDARs in the bicuculine/4-AP-mediated Ca2+ response and that oscillatory elevation of [Ca2+]i is mediated by action potential-dependent presynaptic glutamate release.
|
|
Characterization of extrasynaptic NMDARs
Calcium measurement is an indirect measure of NMDAR activity. It has been shown that NMDA-mediated increase in intracellular calcium relies on NMDAR current and only for a small proportion (estimated at 10%) on voltage-sensitive calcium channel activation (28)
. This method represents an interesting alternative to a more direct method. Selective activation of extrasynaptic NMDARs was performed by NMDA bath application on primary cortical cultures (DIV 14) following by the permanent blockade of synaptic NMDARs with a noncompetitive irreversible NMDAR antagonist: MK-801 (10 µM). Because MK-801 is an open channel blocker, its coapplication during bicuculine/4-AP treatment ensures the selective blockade of the receptors activated during synaptic stimulation paradigm. Subsequently, NMDA bath application induced a [Ca2+]i increase fully blocked by AP-5 (100 µM) that confirms the extrasynaptic NMDAR activation (Fig. 3a
). After MK-801, the remaining Ca2+ response after bicuculine/4-AP application was fully blocked by the coapplication of nimodipine together with the AMPA receptor antagonist CNQX (Fig. 3b
). This result indicates that the remaining response was not mediated by synaptic NMDAR activation and that coapplication of MK-801 with bicuculine/4-AP fully blocked synaptic NMDAR-mediated calcium response. To control MK-801 recovery, the completeness of MK-801 blockade was retested after 60 min (Fig. 3c
). These results demonstrate the presence of a functional pool of NMDARs located at extrasynaptic sites on mouse primary cortical neuron cultures at DIV 14.
|
Next, we identified one downstream signaling pathway activated selectively by either synaptic or extrasynaptic NMDARs. Among the neuronal signaling pathways, the ERKs have been shown on rat hippocampal neuron cultures to respond differently depending on the location of activated NMDARs (29)
. On mouse cortical neuron cultures, we characterized the activation pattern of ERKs induced by synaptic or extrasynaptic NMDAR activation. We showed that 30 min after being exposed to selective synaptic NMDAR activation, 60–70% of neurons displayed a significant increase in p-ERK immunostaining, while selective activation of extrasynaptic NMDARs did not induce any significant changes (Fig. 4a
). These results were confirmed by the immunoblot analysis of p-ERKs. Bicuculine/4-AP application induced a rapid and sustained ERK phosphorylation (Fig. 4b
), and the contribution of action potential-dependent activation of synaptic NMDARs was confirmed since ERK phosphorylation was fully blocked by TTX or MK-801 application (Fig. 4b, c
). Moreover, treatments with either TTX or MK-801 alone induce a significant reduction of ERK phosphorylation’s basal level, suggesting that spontaneous excitatory postsynaptic potential and a subsequent activation of synaptic NMDAR activation promote ERK phosphorylation in control conditions (Fig. 4b, c
). We showed that synaptic NMDAR stimulation activated ERK pathways at all time points, whereas extrasynaptic NMDAR stimulation and NMDA bath application (15 µM) failed to activate ERK pathways (Fig. 4d-f
). The difference in magnitude of Ca2+ signal induced by extrasynaptic NMDA activation vs. synaptic NMDAR activation may result in the absence of ERK activation observed. However, NMDA bath application at a concentration that induces comparable calcium increase (data not shown) to synaptic NMDAR activation, could not reproduce bicuculine/4-AP-mediated ERKs activation profile. Together, these data illustrate that the ERK signaling pathway is activated only when synaptic NMDARs are stimulated and may serve as a signaling "fingerprint" for unveiling the participation of synaptic receptors in a mixed NMDAR activation.
|
Ca2+ overload through synaptic NMDARs is not neurotoxic
Synaptic NMDAR activation and NMDA bath application induce Ca2+-mediated signaling pathways differently. This result may be due to different levels of intracellular Ca2+ concentration reached during these experimental conditions. Because intracellular Ca2+ overload plays a pivotal role during excitotoxicity, it was essential to compare the neurotoxic potential of bicuculine/4-AP or NMDA bath application when both paradigms of activation induce equivalent intraneuronal Ca2+ accumulation. Time-lapse Ca2+ imaging with Fura-2 was used to estimate intracellular Ca2+ accumulation evoked by stimulation of synaptic NMDARs or by NMDA bath application (15 µM; alone or in combination with 0.5 µM TTX) for 1 h. This exposure for 1 h to bicuculine/4-AP produced an sustained increase of [Ca2+]i that reached a plateau with bursts of action potentials, while NMDA bath application (with or without TTX) produced a linear increase of [Ca2+]i (Fig. 5a
). Despite differences in intracellular Ca2+ dynamic profiles, measurements of the area under curve revealed similar intracellular Ca2+ accumulation induced by these different stimulation paradigms (Fig. 5a
). Next, we exposed sister cortical neuron cultures to bicuculine/4AP or NMDA 15 µM for 1 h and measured the neuronal death 24 h later. At 15 µM, NMDA bath application induced 40% neuronal death, whereas bicuculine/4-AP treatment did not induce any neurotoxicity (Fig. 5b
). These results demonstrate that Ca2+ increase induced by sustained synaptic NMDAR activation has no deleterious effect on neurons, whereas similar Ca2+ increase induced by NMDA bath application induces neurotoxicity.
|
Only extrasynaptic NMDAR activation induces characteristic features of excitotoxicity
Loss of mitochondrial membrane potential (
m) is an early feature of excitotoxicity that participates in the deleterious cascade leading to neuronal cell death (30)
. We used rhodamine 123 to monitor m changes during synaptic or extrasynaptic NMDAR activation. We observed that Ca2+ influx through synaptic NMDARs did not trigger any significant modification of the m, whereas Ca2+ entry through extrasynaptic NMDARs caused a significant depolarization of the mitochondrial membrane (Fig. 6a, bottom left panel; c, e
). In parallel, bath application of NMDA induced a rapid m breakdown similar to the one observed after selective activation of the extrasynaptic NMDARs (Fig. 6d, e
), demonstrating that mitochondrial dysfunction is associated with extrasynaptic NMDARs activation, whereas synaptic NMDAR activation does not induce any alteration of the mitochondrial function. Dendrites and cell body alterations are another hallmark of excitotoxicity, which are characterized by formation of dendritic varicosity and cell body swelling (13)
. Morphological analysis was performed to quantify cell body swelling and dendritic varicosity on eGFP-expressing cortical neurons (14 DIV) exposed to selective stimulation of synaptic, extrasynaptic NMDARs or NMDA bath application for 30 min. While stimulation of synaptic NMDARs did not significantly alter neuronal morphology, extrasynaptic NMDAR stimulation or NMDA bath application triggered the appearance of dendritic varicosity and cell body swelling with similar kinetics (Fig. 7a, b
). Thus, neuronal morphology alterations mediated by NMDAR activation can be mainly attributed to extrasynaptic receptors.
|
|
Extrasynaptic NMDAR activation shuts off synaptic NMDAR signaling
Our results revealed a functional dichotomy between synaptic NMDAR activation and extrasynaptic NMDAR activation or NMDA bath application. To investigate the origin of such dichotomy, we investigated whether glutamate transporters play a key role in preventing glutamate spillover and subsequent extrasynaptic NMDAR activation during synaptic stimulation. To validate this hypothesis, we used a pharmacological compound characterized for its ability to block glutamate uptake systems. After blocking synaptic NMDARs by coapplication of bicuculine/4-AP and MK-801, we studied the effect of the competitive blocker of the glutamate transporters, DL-TBOA (30 µM). As expected, and after synaptic NMDAR blockade, synaptic stimulation only produced a modest Ca2+ response mediated by non-NMDA receptor activation, whereas similar application of bicuculine/4-AP in the presence of DL-TBOA triggered an important Ca2+ response fully blocked by AP-5. These results demonstrate that in the presence of DL-TBOA synaptic NMDAR activation protocol triggers NMDAR activation located extrasynaptically by glutamate spillover (Fig. 8a
). Thereafter, we evaluated ERK phosphorylation mediated by bicuculine/4-AP treatment in the absence or presence of DL-TBOA. Our data showed that the presence of DL-TBOA antagonizes ERK activation induced by synaptic NMDARs. Together, these results suggest that when glutamate reaches extrasynaptic NMDARs, it turns off ERK signaling pathways (Fig. 8a, b
). To confirm these observations, we exposed our cortical neuron cultures to simultaneous synaptic and extrasynaptic NMDAR activation by using both bicuculine/4-AP and NMDA bath application. We observed that NMDA bath application blocked the bicuculine/4-AP-mediated activation of the ERK pathway (Fig. 8c
). Thus, these results showed that, by preventing glutamate spillover, glutamate transporters control the neuronal fate from beneficial synaptic NMDAR activation to deleterious extrasynaptic NMDAR activation.
|
Memantine selectively blocks extrasynaptic NMDARs
Memantine is a low-affinity NMDAR antagonist that has the ability to prevent the consequence of NMDAR activation. First, we measured the neuronal cell death triggered by NMDA bath application (15 µM) in the presence of different concentrations of memantine, ranging from 0.1 to 10 µM. We observed that coapplication of memantine induced a neuroprotection against excitotoxicity when applied at 1 and 10 µM (Fig. 9a
). Then, we evaluated the effect of memantine treatment (at 1 µM) on synaptic or extrasynaptic NMDAR-mediated response. Although we did not observe any significant modification of the synaptic NMDAR-mediated calcium response (Fig. 9b
) memantine significantly blocked the extrasynaptic NMDAR response (Fig. 9c
), as well as the calcium response induced by NMDA bath application (Fig. 9d
). These results demonstrate that memantine at 1 µM is a selective NMDAR antagonist that preferentially targets extrasynaptic NMDARs.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 31). However, it has been demonstrated that NMDARs located at the synapse or at extrasynaptic sites affect intracellular signaling pathways and neuronal survival differently (21
, 29)
. Here, we studied whether the location of Ca2+ entry determines its effect on neuronal viability. We reported the following observations. 1) Mature mouse cortical neuron cultures exhibit two populations of NMDARs: synaptic and extrasynaptic receptors. Selective activation of each population produces sustained intracellular Ca2+ increase, but the qualitative effect of this Ca2+ increase on downstream signaling pathways differs: synaptic NMDARs produce a sustained ERK phosphorylation, whereas extrasynaptic NMDA receptors do not activate ERKs. 2) Intracellular Ca2+ increase triggered by synaptic NMDARs activation is not neurotoxic, whereas similar Ca2+ increase induced by NMDA bath application induces neuronal death. 3) NMDA bath application and extrasynaptic NMDAR activation share identical neuronal alterations, ranging from mitochondrial dysfunction to segmental dendritic beading and cell body swelling. These results suggest that only the extrasynaptic pool of NMDARs mediates the deleterious effects of glutamate in neurons. 4) Activation of extrasynaptic NMDARs exerts a negative feedback on signaling pathways activated by synaptic NMDARs. 5) Finally, memantine, the only NMDAR antagonist well tolerated in clinic, targets preferentially extrasynaptic NMDARs.
As described previously in primary hippocampal and cerebellar granule neuron cultures (19
, 32)
, we identified two populations of NMDARs in mouse cortical neurons: synaptic receptors activated by presynaptic glutamate release and extrasynaptic receptors. The relative distribution of these receptors evolves during synaptogenesis, as in our culture model both spine number and synaptic NMDAR response show a 5- to 6-fold increase between DIV 7 and DIV14. Thus, to evaluate the effect of both synaptic and extrasynaptic NMDAR populations on neuronal fate, we used mature primary neuron cultures at 14 DIV, which exhibit high susceptibility to glutamate toxicity (12)
, and a large number of dendritic spines. By comparing the Ca2+ response mediated by selective activation of synaptic NMDARs or NMDA bath application, we showed that synaptic NMDAR stimulation was dependent on firing of action potentials and presynaptic release of glutamate and that subsequent selective activation of synaptic NMDARs was controlled by the glutamate uptake systems that prevent glutamate spillover. Because the method used to generate selective activation of synaptic NMDARs (application of 4-AP and bicuculine) may produce high-frequency input and subsequently leads to a global activation of NMDARs, it was important to determine whether the population of NMDARs activated is only synaptic. When we combined bicuculine + 4AP into a pharmacological compound that inhibits glutamate transporters, to pharmacologically generate a glutamate spillover and subsequently extrasynaptic NMDAR activation, we obtained a p-ERK signaling profile distinct from those obtained following a selective activation protocol of synaptic NMDARs, suggesting that under bicuculine and 4-AP application, only a marginal extrasynaptic NMDAR occurs.
Inversely, Ca2+ response mediated by NMDA bath application was not modified by either action potentials or presynaptic glutamate release blockades, raising the question about the relation between synaptic and extrasynaptic NMDARs during NMDA bath application. Then, we used a well-characterized protocol to stimulate only extrasynaptic NMDARs by blocking synaptic NMDARs with MK-801 before NMDA bath application (21)
. First, we controlled whether extrasynaptic NMDARs move laterally into the synapse during our experimental protocol. Indeed, if several reports showed that synaptic currents can progressively recover after synaptic receptor blockade by MK-801, suggesting (19
, 33)
that synaptic and extrasynaptic NMDARs can switch between these two receptors pools (34
, 35)
, others demonstrated an absence of extrasynaptic NMDAR lateral mobility (22
, 36)
. Restricted to short periods of time (
60 min), our experimental protocols performed at 14 DIV revealed a stability of extrasynaptic NMDAR-mediated signals that suggests an absence of NMDAR lateral mobility.
In rat hippocampal neurons, synaptic NMDAR activation has been shown to induce brain-derived neurotrophic factor (BDNF) expression through cAMP response element-binding protein (CREB) phosphorylation, whereas extrasynaptic NMDAR activation shuts off CREB activation (21
, 37)
. In a similar way, we studied the consequence of selective receptor activation on the ERK pathway, a signaling pathway that has been tightly associated with NMDAR activation in hippocampal neurons (29
, 38)
. We observed that activation of synaptic or extrasynaptic receptor populations affects the ERK signaling pathway differently even if both activations induce a massive increase in intracellular Ca2+ concentration. Synaptic NMDAR activation induces a sustained increase of p-ERKs, whereas extrasynaptic NMDAR activation does not modify p-ERK levels. Interestingly, we showed that NMDA bath application induces an activation profile similar to a selective extrasynaptic NMDAR activation. Again, this last observation raises the question about the functional relation between synaptic and extrasynaptic NMDARs during NMDA bath application.
How does cellular location influence the effect of NMDAR-mediated Ca2+ increase on neuronal death? To answer this question, we determined the NMDA concentration (applied in the bathing medium) that induces an intracellular Ca2+ load comparable to that mediated by prolonged synaptic NMDAR activation. Then, we evaluated the effect of these two experimental protocols on neuronal fate. We observed that NMDA bath application triggers neuronal death, whereas synaptic NMDAR activation does not induce neurotoxicity. During NMDA bath application, neurons display a linear Ca2+ increase devoid of any oscillation and not modified by TTX treatment. These results suggest that action potential-dependent activation of synaptic NMDARs does not significantly participate in the Ca2+ load evoked by NMDA bath application. To further understand the influence of NMDARs cellular location, we compared the effect of synaptic or extrasynaptic NMDARs activation and NMDA bath application on mitochondrial function, as well as cell body and dendrite morphology. In neurons, mitochondria represent the major source of energy and participate in the regulation of Ca2+ homeostasis (39)
. By monitoring the mitochondrial membrane potential (m), we estimated the mitochondrial dysfunction that has been described as an early feature of excitotoxic neuronal death (30
, 40)
. We observed that synaptic NMDAR activation does not modify the m, whereas extrasynaptic NMDAR activation and NMDA bath application lead to a rapid loss of m. Another early feature of excitotoxicity is the rapid degeneration of dendrites and cell body (41
42
43)
that occurs during ischemic-hypoxic injury (44
, 45)
. These alterations can be reproduced in primary cortical neuron cultures by exposure to glutamate receptor agonists (43)
. By performing time-lapse confocal microscopy on eGFP-expressing neurons, we observed that both extrasynaptic NMDAR activation and NMDA bath application induce cell body swelling and dendritic varicosity, whereas synaptic NMDAR activation does not alter neuronal morphology. Together, these data corroborate the observation that extrasynaptic NMDARs mediate the deleterious effect of glutamate, whereas synaptic NMDARs are involved in the neurotrophic effect of glutamate (21)
.
As stated before, under conditions of NMDA bath application, the extrasynaptic NMDAR activation profile is overwhelmingly dominated by synaptic-mediated signals. Thus, it was necessary to further understand the relation between synaptic and extrasynaptic NMDARs in those conditions. We reported that bath application of neurotoxic concentrations of NMDA induces the same profile of ERK signaling, mitochondrial function, and dendrite alteration as that observed following a selective extrasynaptic NMDAR activation. These results corroborate recent reports from H. Bading’s group (46)
and G. Hardingham’s group (47)
, which both show that bath application of glutamate (20 µM) mimics extrasynaptic NMDAR activation by selectively inducing a prodeath gene or failing to induce a synaptic-dependent prosurvival genes. The identification of a reciprocal influence between these two pools of receptors brings interesting perspectives on the roles of extrasynaptic NMDARs in physiological and pathological conditions. A recent report has shown that under high-frequency stimulation, extrasynaptic NMDAR activation could contribute to synaptic transmission by transiently increasing synaptic strength (48)
. According to our data, extrasynaptic activation occurs only when glutamate overwhelms the synaptic cleft. If sustained, this activation will trigger excitotoxic injury. This specific feature of extrasynaptic NMDARs may provide the pharmacological basis to design specific antagonists of this receptor pool. Since controversial results have been reported on the role of subunit composition of NMDARs and the use of selective NMDAR antagonists to NR2B-containing NMDARs in neuroprotection strategy (23
24
25
, 49)
, the potential use of the low-affinity antagonist memantine in targeting extrasynaptic NMDARs has drawn growing interest. Here, we demonstrated that memantine, at a clinically achievable concentration (50)
, exerts a preferential blockade of extrasynaptic NMDARs. The fact that memantine does not affect synaptic NMDAR response is in accordance with previous results showing that memantine only acts under pathological activation of NMDARs, without affecting normal synaptic neurotransmission (for a review, see ref. 50
). This pharmacological feature may explain the favorable safety profile of this compound in humans. In summary, we demonstrated that NMDAR-mediated Ca2+ increase does not systematically trigger neurotoxicity. Indeed, only Ca2+ influx through extrasynaptic NMDARs triggers dramatic perturbations of neuronal physiology, leading to excitotoxic neuronal death. Compounds that specifically inhibit the extrasynaptic NMDAR-mediated response represent a promising therapeutic strategy for treatment of glutamatergic dysfunction observed in dementia, acute brain injuries, and neurodegenerative diseases.
|
ACKNOWLEDGMENTS |
|---|
Received for publication April 3, 2008. Accepted for publication July 24, 2008.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Xiao, C. Feng, and Y. Chen Glucocorticoid Rapidly Enhances NMDA-Evoked Neurotoxicity by Attenuating the NR2A-Containing NMDA Receptor-Mediated ERK1/2 Activation Mol. Endocrinol., March 1, 2010; 24(3): 497 - 510. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Lopez-Menendez, S. Gascon, M. Sobrado, O. G. Vidaurre, A. M. Higuero, A. Rodriguez-Pena, T. Iglesias, and M. Diaz-Guerra Kidins220/ARMS downregulation by excitotoxic activation of NMDARs reveals its involvement in neuronal survival and death pathways J. Cell Sci., October 1, 2009; 122(19): 3554 - 3565. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Xu, P. Kurup, Y. Zhang, S. M. Goebel-Goody, P. H. Wu, A. H. Hawasli, M. L. Baum, J. A. Bibb, and P. J. Lombroso Extrasynaptic NMDA Receptors Couple Preferentially to Excitotoxicity via Calpain-Mediated Cleavage of STEP J. Neurosci., July 22, 2009; 29(29): 9330 - 9343. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Y. Heng, P. J. Detloff, P. L. Wang, J. Z. Tsien, and R. L. Albin In Vivo Evidence for NMDA Receptor-Mediated Excitotoxicity in a Murine Genetic Model of Huntington Disease J. Neurosci., March 11, 2009; 29(10): 3200 - 3205. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. R. Joshi, N.-P. Wu, V. M. Andre, D. M. Cummings, C. Cepeda, J. A. Joyce, J. B. Carroll, B. R. Leavitt, M. R. Hayden, M. S. Levine, et al. Age-Dependent Alterations of Corticostriatal Activity in the YAC128 Mouse Model of Huntington Disease J. Neurosci., February 25, 2009; 29(8): 2414 - 2427. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
