AMPA receptor activation, but not the accumulation of endogenous extracellular glutamate, induces paralysis and motor neuron death in rat spinal cord in vivo


Address correspondence and reprint requests to Ricardo Tapia, Instituto de Fisiología Celular, Universidad Nacional Autónoma de México, AP 70–253, 04510-México, D. F., México.


The mechanisms of motor neuron (MN) degeneration in amyotrophic lateral sclerosis (ALS) are unknown, but glutamate-mediated excitotoxicity may be involved. To examine directly this idea in vivo, we have used microdialysis in the rat lumbar spinal cord and showed that four- to fivefold increases in the concentration of endogenous extracellular glutamate during at least 1 h, by perfusion with the glutamate transport inhibitor l-2,4-trans-pyrrolidine-dicarboxylate, elicited no motor alterations or MN damage. Stimulation of glutamate release with 4-aminopyridine induced transitory ipsilateral hindlimb muscular twitches but no MN damage. In contrast, perfusion of α-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) did not modify glutamate levels but produced intense muscular spasms, followed by ipsilateral permanent hindlimb paralysis and a remarkable loss of MNs. These effects of AMPA were prevented by co-perfusion with the AMPA receptor antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo(F)quinoxaline. Perfusion with NMDA or kainate produced no motor effects or MN damage. Thus, the elevation of endogenous extracellular glutamate in vivo due to blockade of its transport is innocuous for spinal MNs. Because this resistance is observed under the same experimental conditions in which MNs are highly vulnerable to AMPA, these results indicate that excitotoxicity due to this mechanism might not be an important factor in the pathogenesis of ALS.

Abbreviations used

amyotrophic lateral sclerosis


α-amino-3-hydroxy-5-methyl-4-isoxazole propionate




Choline acetyltransferase


motor neuron





Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disease characterized by the progressive loss of motor neurons (MN) in the spinal cord and the cerebral cortex (Julien 2001; Cleveland and Rothstein 2002). Although a familiar type of ALS, linked to mutations of superoxide dismutase 1, has been described (Rosen et al. 1993), the great majority of cases (≥ 90%) are sporadic. The cause of ALS is unknown, but it has been related, among other possible mechanisms, to excitotoxicity induced by excessive glutamatergic neurotransmission. Some studies showed elevated glutamate levels in plasma and cerebrospinal fluid of a population of ALS patients (Plaitakis 1990; Rothstein et al. 1990; Shaw et al. 1995; Spreux-Varoquaux et al. 2002), and suggested that the death of MNs in the disease is associated with an abnormal glutamate metabolism leading to glutamate-mediated excitotoxicity. This hypothesis received strong support from the observation of a reduction of the astroglial glutamate transporter EAAT2 or GLT-1 in motor cortex and spinal cord of sporadic ALS patients (Rothstein et al. 1992, 1995). Loss or dysfunction of the glutamate transporters could theoretically result in an increase in the extracellular concentration of glutamate and in the consequent overactivation of glutamate receptors. In fact, chronic inhibition of glutamate uptake in spinal cord cultures by the transport blockers threo-hydroxyaspartate or L-2,4-trans-pyrrolidine-dicarboxylate (PDC) results in the accumulation of extracellular endogenous glutamate and MN loss (Rothstein et al. 1993; Carriedo et al. 1996). A similar neurotoxic effect of glutamate transport blockade has been shown in cortical neuronal cultures (Velasco et al. 1996).

Different from the above findings, in vivo microdialysis studies in the hippocampus and the striatum have shown that PDC induces notable and long lasting increases in the extracellular concentration of glutamate but this increase does not produce any sign of neuronal hyperexcitability and failed to cause neurodegeneration (Massieu et al. 1995; Massieu and Tapia 1997). In contrast, under identical microdialysis experimental conditions, short lasting elevations of extracellular glutamate in the hippocampus, induced by the K+ channel blocker 4-aminopyridine (4-AP), resulted in intense behavioral and electroencephalographic seizures and marked neurodegeneration, effects that were prevented by NMDA receptor antagonists and by blockers of glutamate release from nerve endings (Peña and Tapia 1999, 2000; Ayala and Tapia 2003). These results support the notion that, in the hippocampus in vivo, the endogenous extracellular glutamate accumulated by the blockade of its transport (by PDC) does not easily reach the post-synaptic receptors, whereas the stimulation of glutamate release from pre-synaptic terminals (by 4-AP) may indeed overactivate them (Tapia et al. 1999; Peña and Tapia 2000; Ayala and Tapia 2003).

Although NMDA receptors are probably critical for neuronal injury in the hippocampus, several observations support the hypothesis that α-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) receptors may be of greater importance for the degeneration of MNs in ALS. MNs possess AMPA receptors that lack the GluR2 subunit, which makes them highly Ca2+-permeable (Carriedo et al. 1995, 1996; Williams et al. 1997; Shaw 1999). Indeed, AMPA and kainate induce preferential spinal MN loss in vitro (Rothstein and Kuncl 1995; Carriedo et al. 1996) and in vivo (Hugon et al. 1989), and AMPA/kainate receptor antagonists protect against MN degeneration caused by chronic blockade of glutamate uptake in both slice cultures (Rothstein et al. 1993) and dissociated cell cultures (Carriedo et al. 1996).

Here we tested whether the inhibition of glutamate transport or the stimulation of its release in the rat spinal cord in vivo might result in increased endogenous extracellular concentration of glutamate and whether such increase might cause MN death. Using microdialysis probes implanted in the lumbar spinal cord, we inhibited glutamate transport or stimulated its release. In addition, we compared the vulnerability of the spinal MNs to NMDA and AMPA/kainate receptor agonists. Our results indicate that large increases of extracellular glutamate induced by the blockade of its transport are innocuous to the MNs, whereas the activation of AMPA, but not NMDA or kainate receptors, produces remarkable MNs loss and paralysis of the hindlimbs. These observations do not support a role of glutamate-mediated excitotoxicity in ALS.

Materials and methods

Adult male Wistar rats, weighing 290–310 g, were used throughout and handled in accordance with the Rules for Research in Health Matters (Mexico), with approval of the local Animal Care Committee. They were anesthetized with 5% halothane in a 95% O2/5% CO2 mixture and placed in a stereotaxic spinal unit (David Kopf, Tujunga, CA, USA), on a heating pad. Rectal temperature was monitored in several experiments and was found not to vary significantly under any of the experimental conditions tested. A longitudinal incision of the lumbar region was made, and the surrounding muscles were retracted to expose the lumbar vertebrae. A ∼2 mm hole was drilled at the level of the second-third lumbar vertebrae and after careful removal of the dura, a microdialysis probe (dialysis membrane 1.2 mm long and 0.24 mm diameter, CMA/7, Solna, Sweden) was slowly lowered down into the right dorsal horn of the spinal cord (Sundström et al. 1995). Animals were maintained under low anesthesia (∼0.8% halothane) throughout the experiment. The probes were perfused continuously with a Krebs–Ringer solution containing 118 mm NaCl, 4.5 mm KCl, 2.5 mm MgSO4, 4.0 mm NaH2PO4, 2.5 mm CaCl2, 25 mm NaHCO3, and 10 mm glucose, pH 7.4, at a flux rate of 2 µL/min, using a microsyringe mounted on a microinjection pump (model CMA/100, Carnegie, Sweden). After a 60 min equilibration period, consecutive fractions of 25 µL (12.5 min) of perfusate were collected. After the first three fractions, which were used to determine the basal levels of amino acids, 100 mm K+ or the drugs studied were perfused during the time periods indicated in Results. PDC (Tocris, Bristol, UK, 25 and 50 mm) solution was prepared by dissolving the compound in 1 N NaOH, adjusting the pH to 7–7.5, and then adding the solution to the Krebs–Ringer perfusion medium to give the indicated concentrations. 4-AP (35 and 70 mm), NMDA (12 mm), kainate (8 mm), AMPA (6 and 12 mm) and 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo(F)quinoxaline (NBQX) (500 µm) were added to the medium after dissolving them in sterile saline. In all cases, the osmolarity of the medium was maintained by reducing the NaCl concentration proportionally. The concentrations of PDC and 4-AP were chosen on the basis of previous results from this laboratory (Massieu et al. 1995; Massieu and Tapia 1997; Peña and Tapia 1999), and those of the glutamate receptor agonists on the basis of work of other researchers (Hugon et al. 1989). However, it must be considered that the efficiency of the dialysis membrane for the low molecular weight compounds used and for the amino acids collected is 7–11% (Massieu et al. 1995; Morales-Villagrán and Tapia 1996); the amino acid results were not corrected for this efficiency. 4-AP was obtained from Sigma (St. Louis, MO, USA), and all other compounds from Tocris.

Amino acids were measured in the dialysates by HPLC, as previously described (Salazar et al. 1994; Massieu et al. 1995). In brief, the 25-µL collected fractions were derivatized with the same volume of ο-phthaldialdehyde and 3 min later 20 µL was injected into a Beckman liquid chromatograph equipped with an ODS column (25 cm × 4 mm). The mobile phase was methanol/potassium acetate (0.1 m, pH 5.5) and was run at a rate of 1.5 mL/min in a 25% to 75% methanol linear gradient (15 min duration). Amino acid quantification was made by comparison with a standard mixture of amino acids processed in the same manner. This technique permits the quantification of aspartate, glutamate, glutamine, glycine, alanine, taurine and GABA, but results are given only for glutamate and aspartate. The changes in the other amino acids were comparatively smaller and were in general similar to those previously described in the hippocampus with PDC and 4-AP (Massieu and Tapia 1997; Peña and Tapia 1999).

At the end of the microdialysis experiment, the skin was sutured and anesthesia was discontinued. Rats were kept in individual cages with water and food ad libitum, and observed during the next hours and periodically every day after the operation. Five days later, the animals were anesthetized with sodium pentobarbital and transcardially perfused with 250 mL of 0.9% saline, followed by 250 mL of paraformaldehyde in 0.1 m phosphate buffer, pH 7.4. The lumbar portion of the spinal cord was removed, left in fixative for an additional 24 h, and transferred successively to 10, 20 and 30% sucrose (24 h each). Transverse sections (40–50 µm thick) were obtained in a cryostat and stained with cresyl violet. The correct location of the microdialysis probes was confirmed and the morphologically undamaged MNs (i.e. large, > 25 µm MNs, similar in appearance to those of the contralateral ventral horn) were counted in a 10 × microscopic field. Two sections, 40 µm apart, were counted in each ventral horn for each rat, covering the area showed in the square marked in the representative micrographs of control rats (Fig. 1). No further analysis was carried out when the mechanical lesion produced by the probe included not only the dorsal horn but invaded also the ventral horn. However, it is important to mention that when this happened the animals did not show significant motor alterations at any time.

Figure 1.

Left part: normal histological and ChAT immunocytochemical appearance of MNs in control rats perfused only with Krebs medium, 5 days after the experiment. (a, b and c) Cresyl violet staining, showing the mechanical lesion produced by the probe in the dorsal horn and the normal morphology of MNs in the ispilateral (b) and contralateral (c) ventral horns. (d and e) Negative control of ChAT immunocytochemistry, in the absence of primary antibody; note the total lack of immunostaining of MNs against the reddish background. (f and g) Positive ChAT immunocytochemistry of MNs in the ipsilateral (f) and contralateral (g) ventral horns; note the intense labeling of MNs. All micrographs were obtained from the same rat, but are representative of the results obtained in eight animals. In this figure and in Figs 2–4 , bars = 400 µm in (a), 100 µm in (b) and (c), and 50 µm in (d) to (g). Right part: changes in extracellular glutamate and aspartate induced by perfusion with 35 m m 4-AP and lack of damaging effect on MNs 5 days after the experiment (see Fig. 5 for quantitative data). The graph shows the time course of the amino acid changes during perfusion with 4-AP for two microdialysis fractions (horizontal bar). Means ± SEM for seven rats (* p <  0.05, ** p <  0.01). (a–c) are representative micrographs of cresyl violet staining and (d) and (e) of ChAT immunocytochemistry, obtained from the same rat.

In all rats, immunocytochemistry of choline acetyltransferase (ChAT) was carried out in alternate sections to those used for cresyl violet staining. Free floating sections were rinsed in 0.1 m phoshate-buffered saline pH 7.4, pre-incubated with phosphate-buffered saline−0.3% Triton-X-100 containing 5% bovine serum albumin for 2 h, and incubated with a goat polyclonal anti-ChAT antibody (1 : 200, Chemicon, Temecula, CA, USA) overnight at room temperature. The next day, after two washings in phosphate-buffered saline–Triton-X-100 for 10 min, sections were incubated with biotinylated-conjugated mouse anti-goat IgG (1 : 200, Vector, Burlingame, CA, USA) during 1 h. After two more 10-min washings, incubation with avidine-Texas Red conjugate (1 : 200, pH 8.2, Vector), was carried out for 1 h and then washed twice. Finally, sections were coverslipped with Vectashield fluorescent mounting medium and examined in a Nikon microscope equipped with an epifluorescence attachment. To assess the specificity of the procedure, some parallel sections were processed in the absence of primary antibody.

Statistical analysis of amino acid changes in the dialysates was carried out using paired Student's t-test, and anova and post hoc Fisher test was used for the quantitative analysis of MNs. A value of p < 0.05 was considered statistically significant.


Stimulating glutamate release with high K+ or 4-AP does not induce MN death

To test the effect of increasing the concentration of synaptic glutamate on MN viability, we first stimulated its release in two different ways, depolarizing the tissue with 100 mm K+ and using 4-AP to block K+ channels. No motor alterations were observed at any time in any of the rats treated only with vehicle (Krebs medium) or with 100 mm K+. In both groups viability of MNs in the ventral horn, assessed by cresyl violet staining and by ChAT immunocytochemistry, was not affected, 5 days after the microdialysis procedure (Fig. 1, left). Figure 1(left) shows also a representative example of the negative control of ChAT immunocytochemistry, in which the primary antibody was omitted. No labeled MNs were observed under these conditions.

The concentration of extracellular glutamate and aspartate in the control animals was similar to that observed in the first three microdialysis fractions of the experiments with drugs (Figs 1–4) and did not vary significantly during the 2 h of microdialysis. When the rats were perfused with 100 mm K+ during two microdialysis fractions (25 min), the extracellular concentration of glutamate increased in the first dialysis fraction to about twice the basal value and then returned to the basal concentration (n = 7, data not shown). In these animals, the number and histological and immunocytochemical appearance of MNs was normal (not shown), similar to the control animals perfused only with Krebs medium (Fig. 1).

Figure 2.

Changes in extracellular glutamate and aspartate induced by perfusion with 50 m m PDC during five microdialysis fractions (horizontal bar in graph), and lack of damaging effect on MNs (see Fig. 5 for quantitative data). Details as for Fig. 1 ( n  = 15, * p <  0.01).

Figure 3.

Left part: perfusion of 6 m m AMPA during two microdialysis fractions (horizontal bar in graph) did not modify extracellular glutamate and aspartate levels, but induced marked MNs loss in the ipsilateral ventral horn (see Fig. 5 for quantitative data). The bottom photograph shows the ipsilateral hindlimb paralysis developed in all rats treated with AMPA, which appeared about 3 h after treatment and lasted for at least 5 days, when the rats were killed for histology. Details as for Fig. 1 ( n  = 8). Right part: perfusion of 500 µ m NBQX during four microdialysis fractions (lower bar in graph) prevents the MN loss produced by 6 m m AMPA during two microdialysis fractions (see Fig. 5 for quantitative data). These animals did not show paralysis or any other motor alteration. Details as for Fig. 1 ( n  = 9).

Figure 4.

Left part: changes in extracellular glutamate and aspartate induced by perfusion with 12 m m NMDA during two microdialysis fractions, and lack of damaging effect on MNs (see Fig. 5 for quantitative data). Details as for Fig. 1 ( n  = 7, * p <  0.05, ** p <  0.01). Right part: changes in extracellular glutamate and aspartate induced by perfusion with 8 m m kainate during two microdialysis fractions, and lack of damaging effect on MNs (see Fig. 5 for quantitative data). Details as for Fig. 1 ( n  = 8, * p <  0.05, ** p <  0.01).

In contrast to the rats treated with high K+, immediately after the perfusion of 35 mm 4-AP during two microdialysis fractions rats showed frequent twitches of the hindlimb ipsilateral to the perfused side of the spinal cord, which became more frequent and intense after recovery from anesthesia and lasted for about 2 h. After this period the animals recovered completely. In addition, all 4-AP-treated animals had transient episodes of slamming their tails and of ipsilateral turns during the first hour. The perfusion with 4-AP induced a five- to sixfold increase in the extracellular concentration of glutamate and three- to fourfold in aspartate. This increment lasted for only two fractions and then returned to the basal levels. However, similarly to the high K+-treated rats, no damage to MNs was observed (Fig. 1, right, and Fig. 5). We tested also a higher concentration of 4-AP (70 mm) during two microdialysis fractions in 10 rats. These animals showed more intense ipsilateral spasms and turns, and half of them showed generalized seizures and died a few hours after the experiment. However, no MN loss was observed in the surviving rats, 5 days later (not shown).

Figure 5.

Number of MNs in the ipsilateral and contralateral ventral horns of control rats perfused only with Krebs medium (not differing from intact rats) and of rats treated with PDC, 4-AP, NMDA, kainate, 6 and 12 m m AMPA, and 6 m m AMPA + NBQX. Mean values ± SEM for the number of animals indicated in Figs 1–4 ( n  = 11 for 12 m m AMPA; no MNs could be identified in the ipsilateral horn of these rats). * p <  0.001.

Glutamate increase by inhibiting glutamate uptake does not induce MN death

The mechanism whereby high K+ and 4-AP increase extracellular glutamate levels involves the release of the amino acid. An alternative way to achieve the same results, which is particularly germane to the ALS debate, is to reduce the activity of the glutamate uptake mechanisms. So, we tested the effect of the glutamate transport inhibitor PDC in a different group of rats. PDC was perfused during five microdialysis fractions (62 min), at 25 and 50 mm, but we show the data with 50 mm, as the changes in glutamate and aspartate were less intense with 25 mm PDC and no MN damage was observed. Similarly to the control and high K+ groups, no motor alterations were observed at any time, during or after the experiment. As shown in Fig. 2, 50 mm PDC induced a four- to fivefold increment in the extracellular concentration of glutamate and aspartate, which was apparent two fractions after the beginning of PDC perfusion and tended to increase further through the 1-h experimental period. No significant morphological alterations of MNs were noted in the PDC-treated animals, in spite of the long lasting high levels of extracellular glutamate (Figs 2 and 5). In some rats in which the microdialysis probe reached the ventral horn (see Methods), the levels of glutamate were similarly increased but neither motor alterations nor MN loss apart from the mechanical lesion were observed.

AMPA elicits paralysis and MN death

As AMPA-receptor activation has been shown to cause MN death (Hugon et al. 1989; Rothstein and Kuncl 1995; Carriedo et al. 1996), as a positive control we tested if AMPA had any affect in our experimental model. We perfused 6 and 12 mm AMPA during two microdialysis fractions. With 6 mm, rats did not show motor abnormalities immediately after the perfusion, but after recovery from anesthesia frequent ipsilateral hindlimb twitches, similar to those occurring after 4-AP, were observed in all animals during 2–3 h. The twitches gradually disappeared as the limb became progressively weak, until reaching total paralysis at about 3 h (Fig. 3, left), which persisted until the fifth post-operative day, when the animals were killed for the histological studies. With 12 mm AMPA, the ipsilateral hindlimb twitches were more intense than with 6 mm or with 4-AP, and were accompanied by tail slamming and ipsilateral turning. As with 6 mm AMPA, the 11 rats treated with 12 mm AMPA showed total ipsilateral hindlimb paralysis and in three of these animals the paralysis extended to the contralateral hindlimb. Neither 6 mm (Fig. 3) nor 12 mm (not shown, the results were similar to those of the graph in Fig. 3) AMPA significantly modified the extracellular levels of glutamate or aspartate. However, these treatments resulted in a severe loss of MNs. With 6 mm, nearly 90% MNs were lost in the ipsilateral ventral horn, and a non-significant 10% was lost in the contralateral horn. With 12 mm, a total loss of MNs occurred in the ipsilateral horn and a 50% loss in the contralateral one (Fig. 3 left and Fig. 5). The neuronal destruction caused by AMPA was most evident at the site of the probe and, particularly with the highest concentration used, extended both rostrally and caudally, with progressively less intensity, for about 500 µm.

In order to assess the specificity of the motor and neurotoxic effects of AMPA, the AMPA receptor antagonist NBQX was perfused at a 500 µm concentration during four microdialysis fractions, including one before and one after the perfusion of 6 mm AMPA. Remarkably, these animals did not show any of the motor alterations observed with AMPA alone, during or after the experiment and, as shown in Fig. 3 (right) and Fig. 5, the damage of MNs produced by this agonist in the ipsilateral ventral horn was notably prevented, from 90% to about 50% neuronal loss. As with AMPA alone, the extracellular levels of glutamate and aspartate were not significantly altered in these animals (Fig. 3). Perfusion with NBQX alone during four microdialysis fractions did not modify any of the parameters that we studied (n = 5).

NMDA and kainate do not induce MN death

In order to study the possible involvement of other glutamate receptor subtypes in spinal cord damage, in other experiments we perfused NMDA and kainate. Neither 12 mm NMDA nor 8 mm kainate perfused during two microdialysis fractions produced motor alterations at any time. With NMDA, a significant but transient increase in the extracellular concentration of aspartate, but not glutamate, was observed, whereas kainate induced an increase of both amino acids, which returned to the basal levels after two microdialysis fractions (Fig. 4). NMDA did not produce any significant damage of MNs, whereas kainate caused a small MN loss (about 10%) in the ipsilateral ventral horn (Figs 4 and 5).


To our knowledge, this is the first study in which drugs were administered locally in the spinal cord through reverse microdialysis, and the extracellular concentration of glutamate and other amino acids was measured and correlated with MN viability. With this procedure the effects of the drugs are restricted to a small segment of the spinal cord, unilaterally, therefore permitting the comparison with the contralateral half. We used three strategies to increase the concentration of extracellular glutamate in the spinal cord: inhibition of its transport by PDC, stimulation of its release by 4-AP and depolarization by high K+ concentration. High K+ induced a moderate and transient increase in the extracellular concentration of aspartate and glutamate, which did not produce any motor or neuropathological alteration, most probably because the glutamate released as a result of the depolarization cannot easily gain access to glutamate receptors, similarly to our previous findings in the hippocampus (Peña and Tapia 1999). In another work on the spinal cord, perfusion with 100 mm K+ resulted in a two- to fourfold transient increase of glutamate and aspartate, and a normal gross morphology of the perfused area was reported (Sundström et al. 1995).

Whereas microdialysis perfusion of 4-AP in the hippocampus induced intense epileptic seizures and neurodegeneration (Peña and Tapia 1999, 2000), when this drug was administered in the spinal cord, only transient muscular ipsilateral spasms occurred, but no neurodegeneration, in spite of extracellular glutamate elevations very similar to those observed in the hippocampus. This indicates that the stimulation of glutamate release by 4-AP was capable of exciting MNs, but not enough to produce excitotoxic damage. This might be related to the fact that, whereas in the hippocampus the main glutamate receptor involved in 4-AP-induced excitotoxicity is the NMDA subtype (Fragoso-Veloz and Tapia 1992; Morales-Villagrán et al. 1996; Tapia et al. 1999; Peña and Tapia 2000), in the spinal cord the AMPA receptor seems to be much more important, as reported in this work. In fact, studies on the distribution of the expression of the different subunits of glutamate receptors show that in the ventral horn the AMPA receptors predominate over the NMDA or kainate type (Petralia et al. 2000; Wisden et al. 2000). It seems possible that the glutamate released by 4-AP does not reach the synaptic concentration necessary to induce neurodegeneration through the overactivation of AMPA receptors, whereas the potent and direct action of AMPA clearly was capable of doing so, as we discuss below.

Blocking glutamate transport with PDC is a very effective mechanism for augmenting the extracellular levels of the amino acid in the spinal cord: with 50 mm PDC, glutamate concentrations attained values four to fivefold higher than the basal, that persisted for at least 1 h, when the experiment was terminated. However, in spite of this notable increment no motor alterations or MNs loss was observed. This lack of excitotoxic effect was similar to our previous finding in the hippocampus and in the striatum (Massieu et al. 1995; Massieu and Tapia 1997; Tapia et al. 1999), and leads us to conclude that increased levels of extracellular glutamate in the spinal cord in vivo due to the blockade of its transport are not capable of significantly activating the MN post-synaptic receptors. It could be argued that this lack of excitotoxic effect may be due to a limited diffusion of glutamate to the ventral horn, where MNs are located, because the microdialysis probe was placed in the dorsal horn. Three findings, however, rule out this possibility. First, in rats in which the probe was lowered to reach the ventral horn, glutamate levels were similarly elevated by PDC and no signs of hyperexcitation or MN loss were observed. Second, perfusion of AMPA in the dorsal horn produced almost immediate ipsilateral hindlimb twitching, delayed complete paralysis, and remarkable MN loss, effects that were prevented by NBQX perfused also in the dorsal horn, indicating that diffusion to the ventral horn of compounds structurally similar to glutamate occurs easily and rapidly. Moreover, the damage extended rostrally and caudally for about 500 µm, and the highest AMPA concentration used (12 mm) induced MN damage also in the contralateral ventral horn and, correspondingly, in some of these rats the paralysis extended to the contralateral hindlimb. Third, although the perfusion of 4-AP in the dorsal horn did not induce MN death, it produced immediate twitches of the ipsilateral hindlimb, probably due to the release of glutamate from nerve endings (Morales-Villagrán and Tapia 1996; Peña and Tapia 1999, 2000), also supporting the conclusion that the lack of excitotoxic effects of glutamate after PDC perfusion cannot be ascribed to a limited diffusion of the amino acid into the ventral horn. So, our experiments with PDC do not support the hypothesis that a deficiency of glutamate transporters is involved in the mechanisms of MN death induction in ALS.

Although we cannot exclude the possibility that a chronic blockade of glutamate transport, leading to a persistent elevation of extracellular glutamate for several days or weeks, may be neurotoxic for MNs, the rapid excitotoxic effect of AMPA under identical experimental conditions argues against it. In fact, the results of experiments in vitro, in spinal organotypic (Rothstein et al. 1993) or dissociated cultures, show that long times of exposure to both AMPA receptor agonists and glutamate transport blockers are required to cause MN damage, unless the concentration of external Ca2+ is increased to 10 mm (Carriedo et al. 1996; Van Den Bosch et al. 2000; Vandenberghe et al. 2000b). Hence, the conditions in vivo seem to be different from those in cultures, inasmuch as under physiological Ca2+ concentrations the acute administration of AMPA rapidly produces neuronal excitation and MN death, whereas augmented glutamate did not. Furthermore, chronic infusion with mini-osmotic pumps of the glutamate transport inhibitor threo-hydroxyaspartate (up to 300 nmol/h during 7 days) into the subarachnoid space failed to cause motor alterations or MN damage (Hirata et al. 1997). Thus, our experiments clearly demonstrate that an excess of extracellular glutamate is not likely to be toxic by direct activation of AMPA receptors, although they do not rule out other possible effects of chronic glutamate elevations. Under the latter conditions certain complex effects of glutamate that could produce delayed damage to motor neurons, including an action on receptors located on interneurons, cannot be discarded.

In contrast to high K+ and PDC, perfusion with AMPA did not modify the extracellular levels of glutamate and aspartate, but produced a remarkable excitotoxic effect, characterized by initial contractions of the muscles of the ipsilateral hindlimb, followed by complete paralysis that correlated with the death of MNs. We do not know whether the paralysis was permanent, but no signs of recovery were observed during the five days studied, and the damage of ChAT-positive MNs was severe, particularly with the highest concentration of AMPA studied (12 mm), which produced a total loss of these cells in the segment of the lumbar cord perfused with the agonist. That this excitotoxic action of AMPA is due to its interaction with AMPA-like receptors was demonstrated by the remarkable protection exerted by the AMPA receptor antagonist NBQX, which completely prevented the muscular twitches and the limb paralysis and also reduced notably the loss of MNs. So, these results indicate that the excitotoxic action of AMPA is due to the direct activation of AMPA receptors, which seems to be very specific, as our results with kainate and NMDA show that their receptor subtypes scarcely participate. Both kainate and NMDA, perfused at concentrations equal to or higher than those of AMPA, failed to induce motor alterations or significant MN loss, in spite of the fact that, besides a possible direct action on their receptors, they induced a moderate increase in the extracellular concentration of glutamate and/or aspartate. This effect of NMDA and kainate on excitatory amino acids in vivo has been previously described in the striatum (Young et al. 1988; Young and Bradford 1991) and in the spinal cord (Sundström et al. 1995), and might be due to trans-synaptic or pre-synaptic stimulation of their release.

Assuming a 10% efficiency of the dialysis membrane (see Methods), during 25 µL perfusion of medium containing these compounds at the concentrations used, the amount of kainate, NMDA and AMPA reaching the tissue would be 20, 30 or 15–30 nmoles, respectively. Previous work using intrathecal microinjection (Hugon et al. 1989; Urca and Urca 1990), or mechanical infusion (Kwak and Nakamura 1995) of 15–32 nmoles of kainate induced non-selective neuronal damage and in some rats paraplegia, whereas NMDA was ineffective, even at a dose threefold higher (Urca and Urca 1990). In other work (Ikonomidou et al. 1996), kainate, NMDA and AMPA (7.5–9 nmoles) were applied directly on spinal cord surface during 30 min in 21-day old rats; whereas NMDA caused neuronal damage in the dorsal horn neurons, kainate and AMPA were so toxic that the animals did not survive longer than 4 h. In the majority of these studies AMPA was the agonist with the most neurotoxic action. Its mechanical or mini-osmotic pump intrathecal infusion induced hindlimb paralysis and generalized neuronal loss, at cumulative doses as high as 90–168 nmoles, which were prevented by AMPA receptor antagonists (Nakamura et al. 1994; Kwak and Nakamura 1995). When injected directly in spinal cord tissue, both AMPA and NMDA induced general tissue necrosis, which cannot be easily related to an excitotoxic action on MNs (Liu et al. 1997).

The selective excitotoxicity of AMPA, as compared to NMDA or kainate, is in accordance with the selective vulnerability of MNs to AMPA described in organotypic and dissociated spinal cord cultures (Rothstein et al. 1993; Carriedo et al. 1996). This selective MN sensitivity has been ascribed to the predominance of AMPA receptors lacking the GluR2 subunit, which makes them highly permeable to calcium (Carriedo et al. 1995, 1996; Williams et al. 1997; Shaw 1999). Cultured MNs are more sensitive to AMPA or kainate than dorsal horn neurons, and this selective vulnerability depends on the presence of receptors lacking the GluR2 subunit, on their density rather than on desensitization, and on extracellular calcium (Carriedo et al. 1995; Van Den Bosch et al. 2000; Vandenberghe et al. 2000b), although no significant difference in Ca2+ permeability was found between cultured MNs and dorsal horn neurons (Vandenberghe et al. 2000a).

Our present findings constitute the first demonstration that large and long lasting increases in the extracellular concentration of endogenous glutamate in the spinal cord in vivo do not produce motor hyperexcitation or MN degeneration, and therefore do not support the notion that glutamate-mediated excitotoxicity resulting from the loss of glutamate transporters is involved in the pathogenesis of ALS. This conclusion is strongly supported by our finding that under similar acute experimental conditions spinal MNs are selectively vulnerable to AMPA receptor agonists, and that the resulting loss of MNs induces paralysis of the ipsilateral hindlimb. This experimental model should be useful for studying the mechanisms of MN death caused by AMPA receptor overactivation, the development of the paralysis and its muscular consequences, and the discovery of possible preventive or therapeutic strategies.


This work was supported by CONACYT (project Millennium W8072–35806 N) and DGAPA, UNAM (project IN206100). The authors wish to thank Dr Beatriz Jiménez for advice in the immunocytochemical technique, Miguel Angel Aguileta for his help in the surgical design, and Federico Jandete for cresyl violet staining procedure.