Glutamate is the major excitatory neurotransmitter in the mammalian central nervous system playing critical roles in basal synaptic transmission and mechanisms of learning and memory. Under normal conditions, glutamate is sequestered within synaptic vesicles (~100 mM) with extracellular glutamate concentrations being limited (<1 μM), via retrieval by plasma-membrane transporters on neuronal and glial cells. In the case of central nervous system trauma, stroke, epilepsy, and in certain neurodegenerative diseases, increased concentrations of extracellular glutamate (by vesicular release, cell lysis and/or decreased glutamate transporter uptake/reversal) stimulate the overactivation of local ionotropic glutamate receptors that trigger neuronal cell death (excitotoxicity). Other natural agonists, such as domoic acid, alcohol and auto-antibodies, have also been reported to induce excitotoxicity.
- dendritic bead
Immediate neuronal responses to insult include the rapid formation of neuritic swellings (termed varicosities or beads), mitochondrial dysfunction, decreased ATP concentration, microtubule collapse and neuronal inactivation. In the present review, we summarize current knowledge of the mechanisms that underlie these early excitotoxic responses. Moreover, we speculate on whether these responses are good (a neuroprotective strategy), bad (a pathological response) or just ugly (aesthetic changes only).
Neuronal function is dictated by a neuron's complex and highly polarized morphology. Neural signals are received at postsynaptic sites on the spines (excitatory signals) or shafts (inhibitory signals) of dendrites. The dendritic spine is a basic functional unit of neuronal circuit integration and a site of structural and functional synaptic plasticity (see the reviews by Amici et al.  on pp. 1359–1363 and Moult and Harvey  on pp. 1364–1368 of this issue). Neuronal information (input) is accumulated within the somatodendritic compartment and, if necessary, information is then transferred to the axon, where an action potential is initiated (courtesy of the concentrated voltage-gated Na+ channels at the axon initial segment)  and conducted along the axon to the presynaptic terminals, where it activates voltage-gated calcium channels, to stimulate neurotransmitter release. The directional flow of information is achieved by the spatial distribution of the neurotransmitter receptors and voltage-gated ion channels and the isolation of the somatodendritic compartment from the axon.
As the dendrites provide the major sites (spines) for excitatory input to a neuron, the high concentration of excitatory glutamate receptors renders them particularly vulnerable to excitotoxicity. Excitotoxic dendritic injury (also termed dendrotoxicity)  is characterized by the formation of focal swellings (termed beads) along the length of the dendritic arbour that are typically separated from each other by thin dendritic segments [4,5]. The presence of dendritic beads has been hailed as an early hallmark of neuronal toxicity and has been documented under several pathological conditions, including ischaemia , epilepsy , brain tumour  and aging . Dendritic beads are also observed in several neurodegenerative diseases, including Parkinson's disease  and Alzheimer's disease . In the latter condition, dendritic beads are associated with both extracellular amyloid plaques and intracellular neurofibrillary tangles [12,13]. The formation of dendritic beads has also been observed in cultured neurons after hypoxia [14–16] or exogenous glutamate [14–22].
The pathological and functional significance of focal dendritic beading, as well as the mechanisms underlying bead formation, remains unclear. In particular, it is not known why localized beads form in preference to a generalized swelling. Structural studies [16,17,20,22–25] have identified disrupted microtubules and microtubule-associated proteins within dendritic beads. Although the use of a microtubule stabilizing drug, taxol, is capable of blocking microtubule collapse, dendritic beading still appeared to form . We have now explored this directly and can confirm that dendritic beading is not blocked (Figure 1), suggesting that microtubule collapse is a secondary event.
It is now generally accepted that dendritic beads result from osmotic swelling due to the ionic imbalances initiated by glutamate receptor activation [16,18–20,26,27]. This is supported by observations that dendritic beading can be prevented in hyperosmotic saline [16,18,19] and can be triggered directly by hypo-osmotic saline . The fast morphological changes are induced by the influx of Na+ and Cl−, but not Ca2+ [16,18–20,26], while delayed responses to excitotoxicity involve mechanisms that are largely Ca2+-dependent [18,20,28]. However, it has been suggested that a Ca2+-dependent process may be involved in the cytoskeletal rearrangements that underlie bead morphogenesis .
Temporal and spatial analyses of these events revealed that dendritic beading is coincident with mitochondrial depolarization and a decrease in neuronal ATP levels [16,22,29]. Interestingly, glutamate-induced dendritic beads contain dysfunctional mitochondria . Might mitochondrial dysfunction cause dendritic beading? Mitochondria are capable of consuming ATP if Δψm (mitochondrial membrane potential) is depolarized sufficiently; however, both mitochondrial dysfunction and depleted cellular ATP levels themselves are insufficient to trigger dendritic beading . Moreover, in response to glutamate, mitochondrial depolarization is only partial and the depletion of cellular ATP observed during dendritic beading is mediated solely by the plasma-membrane Na+/K+-ATPase in its futile attempt to counteract the large Na+ influx [16,27]. Thus dendritic beads arise because of an increase in intracellular Na+ and Cl− (and thus osmotically driven solute) after the exhaustion of the cellular ATP supply and failure of Na+ efflux. Nevertheless, mitochondrial dysfunction may render neurons more vulnerable to subsequent insults [16,30]. Under these conditions, the neuron may have insufficient ATP to fuel Na+ export and thus becomes vulnerable to normally innocuous ionic challenges.
Although dendritic beading has been considered an early feature of neuronal dysfunction that precedes death , it is a poor indicator of neuronal cell fate. Dendritic beading may be initiated by innocuous (low-temperature) conditions [31,32] and subtoxic insults [20,21]. Indeed, dendritic beads can actually disappear after the removal of a lethal (100 μM glutamate, 20 min) insult . An alternative hypothesis exists, in that dendritic beading and/or associated events may actually represent a neuroprotective strategy [20,27]. These questions remain to be resolved.
In contrast with the simultaneous occurrence of the events during global exposure of cultured hippocampal neurons to glutamate, during in vitro ischaemic conditions (oxygen–glucose deprivation) the events occur sequentially (mitochondrial depolarization→mitochondrial collapse→dendritic beading) and travel as a wave of events from distal dendrites towards the cell body . A similar propagating ‘delayed calcium deregulation’ response (towards the cell body) has been reported recently , suggesting that both acute and delayed neuronal responses to an excitotoxic challenge follow the same pattern. The mechanism(s) of this propagation is unknown, but may be related to ATP deficiencies [27,30,33]. If the insult were localized and the neuronal responses restricted spatially, localized Na+ and Ca2+ influx might lead to localized Na+/K+-ATPase overactivity and mitochondrial depolarization, together leading to a local ATP depletion and solute influx. This local crisis might spread to neighbouring regions of the dendrites, leading to a propagation of the responses. Given that dendritic beading is a poor predictor of cell fate, it is possible that neuronal cell fate might not be determined until the propagating wave of bioenergetic insufficiency and downstream failure of plasma-membrane ion pumps reaches the cell body [16,21,30,34,35].
Along with Ca2+-dependent depolarization of their membrane potential (; and see also the review by Nicholls  on pp. 1385–1388 of this issue), mitochondria also exhibit changes in their trafficking and morphology during excitotoxicity (Figure 2). Under basal conditions, mitochondria within dendrites appear elongated and undergo extensive directional and lateral movement. However, after increased synaptic activity  or an excitotoxic exposure to glutamate , mitochondrial movement is inhibited and mitochondria exhibit a rounded/swollen morphology [37–41]. Although blocking mitochondrial ATP production is sufficient to inhibit mitochondrial transport , the collapse of mitochondrial structure has been reported to require either extracellular Ca2+ influx via NMDA (N-methyl-D-aspartate) receptors [38,40,41] or solute influx, as a consequence of Na+ accumulation [16,27].
Although the structural collapse of somatodendritic mitochondria has been well documented during excitotoxicity, the consequence of such a change in mitochondrial morphology is far from clear. It is not known whether the ability of mitochondria to generate Δp (protonmotive force) (and thus Δψm) is influenced by their morphology. However, we have demonstrated recently that the collapse of neuronal mitochondria in the absence of Ca2+ does not trigger an apparent reduction in Δψm . Therefore mitochondrial depolarization is not a prerequisite for beading. One known detrimental consequence of prolonged mitochondrial swelling is the rupture of the outer mitochondrial membrane and consequent release of apoptogenic proteins, including cytochrome c .
An important consequence of glutamate exposure is spine loss. Concomitant with bead formation is a decrease in dendritic spine density (Figure 2, top panels) [14,15,20,21]. Although spine loss has been reported to be greatest around dendritic beads, such beads do not appear to form preferentially at sites of pre-existing spines (they are ubiquitous). It is possible that the surrounding cytoplasm is drawn into beads as they form, at the expense of local synapses. Interestingly, dendritic beading and the loss of dendritic spines are coincident with a cessation of synaptic transmission [20,24,31,42]. This may result from a rapid induction of long-term depression of synaptic transmission that is associated with AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid) receptor internalization  or decreases in neurotransmitter release (or both). Like beading, spine loss is a poor indicator of neuronal fate and they re-emerge at their original sites upon the removal of the insult [15,21,31]. In fact, the loss of spine structures may not necessarily be associated with the loss of synaptic contacts [15,31].
Axonal morphological responses
Dysfunction in fast axonal transport has been implicated in the pathogenesis of neurodegenerative diseases . It is known that long-range (calcium-sensitive) mitochondrial transport is driven by molecular motors. These motors utilize ATP hydrolysis to move cargo along microtubule tracks. Anterograde transport of mitochondria has also been reported to be dependent on mitochondrial membrane potential [44–46]. It is therefore notable that the fast axonal transport of mitochondria in mature neurons [DIV (days in vitro) >12], but not immature (DIV 7) neurons [47,48], is not blocked by glutamate exposure (see Supplementary Movie S2 at http://www.biochemsoctrans.org/bst/037/bst0371389add.htm), compared with untreated neurons (see Supplementary Movie S1 at http://www.biochemsoctrans.org/bst/037/bst0371389add.htm). Likewise, axonal microtubule tracks appear to be resistant to glutamate exposure (Figure 2). Therefore Na+, Ca2+ and solute influx, microtubule collapse, mitochondrial depolarization and a decrease in ATP concentration, which occur in dendrites after glutamate exposure, do not appear to invade the axon, presumably due to the diffusion barrier at the axon initial segment in fully polarized neurons. In keeping with this, veratridine-induced Na+ entry into axons, via voltage-gated Na+ channels, triggers axonal beading [18–20] and blocks axonal mitochondrial transport (see Supplementary Movie S3 at http://www.biochemsoctrans.org/bst/037/bst0371389add.htm).
To date, the physiological role of dendritic beads remains unknown. Consistent with their classification as an early hallmark of toxicity, it has been suggested that the formation of dendritic beads is merely an early feature of neuronal dysfunction that precedes cell death. However, the ability of dendritic morphology to recover suggests that dendritic beading does not predict cell death [14–20]. Furthermore, cooling of acute brain slices results in the Na+-dependent formation of dendritic beads that resolve on re-warming, with no apparent neuropathological consequences [31,32]. At the other end of the spectrum, dendritic beading and/or the loss of synaptic sites has been proposed to be a cellular defence against excitotoxicity, as conditions that prevent these events exacerbate glutamate receptor-activated cell death [20,49]. The formation of dendritic beads may therefore reflect an attempt to isolate the neuronal cell body from potentially toxic ionic loads. Interestingly, it has been reported that neurons can survive the formation of distal beads, but not if these spread to the proximal dendrites .
Thus rapid morphological responses to excitotoxic insults may be poor predictors of cell mortality. Cell death is more likely to be predicted by the intracellular Ca2+ influx experienced during the insult. It should be considered that large insults would be expected to mask any protective effects of beading. Therefore it remains possible that neuroprotective effects of dendritic beading may limit any damaging consequences of intense physiological (sublethal) glutamate exposures in which Ca2+ influx is relatively small.
Interestingly, synaptic transmission is immediately decreased after NMDA exposure [20,24,31,42], yet recovers fully by 48 h post-treatment , presumably due to the resolution of dendritic beads. However, no direct evidence for a role of beads in neuronal inactivation has been shown. It should be noted that these conditions also result in the internalization of AMPA receptors, which may underlie the long-term decrease of synaptic transmission observed in response to NMDA application . However, contrary to a putative neuroprotective role of this block in synaptic transmission, it is well established that synaptic activity is neuroprotective (; and see also the Colworth Medal Lecture article by Hardingham  on pp. 1147–1160 of this issue). Perhaps neuroprotection is exerted, not on the neurons insulted, but on the surrounding (bystander) neurons with which they communicate?
This work was supported by the Biotechnology and Biological Sciences Research Council [grant number BBS/S/M/2005/12410] and Tenovus Scotland.
Neuronal Glutamate and GABAA Receptor Function in Health and Disease: Biochemical Society Focused Meeting held at University of St Andrews, St Andrews, U.K., 21–24 July 2009. Organized and Edited by Chris Connolly and Jenni Harvey (Dundee, U.K.).
Abbreviations: Δψm, mitochondrial membrane potential; AMPA, α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid; DIV, days in vitro; NMDA, N-methyl-D-aspartate
- © The Authors Journal compilation © 2009 Biochemical Society