Excitotoxicity: How Glutamate Kills the Neurons It Excites

01The excitotoxicity paradox

Excitotoxicity runs on the same receptors and the same calcium as ordinary signaling

Excitotoxicity runs on the ordinary machinery of excitation. Glutamate is the principal excitatory transmitter of the central nervous system, released from the presynaptic terminal onto AMPA and NMDA receptors on the postsynaptic membrane. In measured amounts that traffic drives learning and plasticity. Held too long, the same traffic dismantles the cell. No foreign poison arrives and no new receptor is recruited.

Degree is only half of the account. In cultured neurons, stimulating synaptic NMDA receptors produced anti-apoptotic activity, while stimulating extrasynaptic NMDA receptors shut off CREB and collapsed mitochondrial membrane potential, an early marker of glutamate-induced damage Hardingham 2002. The protein is the same protein and the ion is the same ion. Position on the membrane decides which program runs.

The Unified Model of Tone reads excitotoxicity as a failure of organization. Tone is the organization of the whole state, and in the glutamate system that organization is the arrangement that decides where transmitter lands, how fast it is cleared, and how long the channel stays occupied. A nervous system holding its tone inside its healthy range can answer a strong input and then let it go, because the range is where the freedom to adapt lives. Tone drifted outside that range is what manifests as illness, and here it manifests as excitation the cell can no longer place, clear, or end.

Nothing about the excitotoxic synapse is exotic. It holds the same transmitter, the same receptors, and the same calcium the healthy synapse uses, and the difference is how long the channel stays occupied and which population of receptors carries the traffic. Sustained occupancy converts a signal the cell reads into a load the cell carries.

Excitotoxicity is a continuum, not a single switch

Excitation and inhibition exist in a living ratio, and excitotoxicity is what happens when that ratio tilts. The clinician reads a set of factors for and against, not a discrete event with a moment of onset. Glutamate arrives without restraint at a postsynaptic cell already primed to answer, and the cost accumulates across the synapse, the astrocyte, and the mitochondrion at once. Every step in the sequence that follows is a normal step running past its safe range.

02Findings

What the research shows

Measured values for calcium, clearance, and the excitatory load.

03Calcium as the trigger

Calcium, not excitation itself, is the pivot of the excitotoxic cascade

Calcium is what converts glutamate signaling into excitotoxic injury. It can carry that message because the neuron keeps its resting level near 0.1 micromolar. That is low enough to read as zero against the calcium-rich fluid outside. Roger Carpenter, in his Neurophysiology, likened the sudden appearance of free calcium inside a neuron to ringing the cell's doorbell. The steepness of the gradient is what makes the message legible.

NMDA receptor activation admits calcium directly. Excitatory amino acids acting at NMDA receptors on cultured spinal cord neurons raised intracellular calcium through the receptor channel itself, measured with the indicator dye arsenazo III MacDermott 1986. Kainate, acting at another receptor subtype, was much less effective. Calcium does not need voltage-gated channels opened by depolarization to reach the cytoplasm.

The injury separates into two components. Early glutamate toxicity depends on sodium and chloride and appears as cell swelling. Late toxicity depends on calcium and appears as gradual disintegration, and the calcium mechanism predominates at lower glutamate exposures. Two tests in cultured mouse neocortical neurons located the route of the lethal calcium. Raising extracellular magnesium to 20 mM, which blocks voltage-gated calcium channels, did not substantially block the toxicity. The toxicity was also largely preserved in sodium-free solution. Voltage-dependent calcium channels are therefore not required for the lethal entry Choi 1987.

NMDA channels kill at loads other channels survive

Calcium neurotoxicity depends on where the calcium comes in. Intracellular calcium was raised equally through NMDA receptor channels, non-NMDA receptors, and voltage-gated calcium channels in cultured mouse spinal neurons. The NMDA route killed far more of them than the other two Tymianski 1993. Equal load, different outcome. The Unified Model of Tone reads that result as input meeting tone: the reply belongs to the receiving organization, not to the size of the signal. The model extends the same principle to inputs delivered from outside the cell. A matched input reaches a system that a larger unmatched one misses.

Death arrives on a delay

Five minutes of 100 micromolar glutamate produced a three-phase sequence in cultured rat hippocampal neurons. Ninety-two percent of 64 neurons buffered the calcium load back to basal levels. After a variable delay, 44 percent of those recovered cells rose again to a sustained plateau and failed to recover, and that delayed overload correlated with cell death Randall 1992.

In excitotoxicity the doorbell is held down. Sustained glutamate binding keeps NMDA channels open, and calcium enters faster than pumps and mitochondria remove it. Prolonged elevation recruits enzymes built for transient duty, including the nitric oxide synthase that seeds the next stage of injury.

Calcium recordings during the injury run biphasic in the same cultured preparations. A primary transient decays within minutes, and a secondary sustained rise arrives later and indicates imminent death Tymianski 1993. Mitochondria buffer the first rise and pay for it in membrane potential and ATP. The second rise arrives after buffering has already failed, which is why a neuron that looks recovered can still be committed to die.

04Nitric oxide and PARP-1

Calcium overload converts an electrical insult into chemical destruction

Calcium activates neuronal nitric oxide synthase, and that step turns the calcium load into chemical damage. The enzyme converts L-arginine to L-citrulline and releases nitric oxide. Nitric oxide synthase inhibitors prevented neurotoxicity from NMDA and related excitatory amino acids in primary cortical cell cultures, with an EC50 of 20 micromolar for nitro-L-arginine and 170 micromolar for monomethyl-L-arginine Dawson 1991. Depleting the culture medium of arginine with arginase abolished the toxicity, and L-arginine competitively reversed the inhibition.

Nitric oxide meets superoxide generated by struggling mitochondria, and the two react to form peroxynitrite. Measured in vitro, peroxynitrite has a pKa of 7.49 at 37 degrees. Once protonated it decomposes with a half-life of 1.9 seconds at pH 7.4, producing an oxidant with reactivity like the hydroxyl radical Beckman 1990. The species is at its most destructive at the pH the brain runs at, and it spends itself within seconds of forming, so the damage stays close to where the calcium came in.

Peroxynitrite inflicts oxidative and nitrosative injury on proteins, lipids, and DNA. The DNA damage is what carries the cascade to its end.

PARP-1 turns DNA repair into an energy collapse

DNA damage drives massive activation of poly ADP-ribose polymerase-1. Mice with the PARP gene disrupted were profoundly protected against glutamate and nitric-oxide-mediated ischemic insults in culture and showed major decreases in infarct volume after reversible middle cerebral artery occlusion Eliasson 1997. PARP consumes NAD and the ATP spent regenerating it. The repair response strips the cell of the energy it needs to survive the insult, which is where excitotoxicity becomes a metabolic problem.

Parthanatos is a defined route to death

PARP-1 activation drives translocation of apoptosis-inducing factor from the mitochondria to the nucleus in cultured cells, and that factor is required for the death Yu 2002. PARP inhibitors and PARP-1 knockout both prevented it, and the death proceeded without caspases. Parthanatos names that route, and its caspase independence is what separates it from classical apoptosis.

The death signal itself is a polymer. PAR made by PARP-1 is directly toxic to neurons, and PARP-1-dependent NMDA excitotoxicity in cultured cortical neurons is reduced by neutralizing antibodies to PAR and by overexpression of the degrading enzyme PARG Andrabi 2006. Transgenic mice overexpressing PARG had significantly smaller infarcts after focal ischemia, and mice with reduced PARG had significantly larger ones.

Each link in the chain has been removed experimentally, and removing any one of them protects the neuron. Nitric oxide synthase inhibition prevents the death Dawson 1991. PARP deletion cuts infarct volume after middle cerebral artery occlusion Eliasson 1997. Degrading the PAR polymer with PARG rescues the cell one step further along Andrabi 2006. The sequence is causal at every joint, and every stage of it is ordinary biochemistry recruited past its safe range, which is why excitotoxicity is a regulation failure and not an intoxication.

05Clearance and restraint

Clearance capacity decides whether excitation stays physiological

Astrocytes hold the excitatory range. Mice lacking the astrocytic glutamate transporter GLT-1 show lethal spontaneous seizures and increased susceptibility to acute cortical injury Tanaka 1997. In rats, antisense knockdown of the glial transporters GLAST or GLT-1 raised extracellular glutamate, produced neurodegeneration characteristic of excitotoxicity, and paralyzed the animals progressively Rothstein 1996. Knocking down the neuronal transporter EAAC1 left extracellular glutamate in the striatum unchanged and produced mild neurotoxicity and epilepsy instead.

Glutamate is abundant and kept. Plasma runs at 50 to 100 micromolar and whole brain at 10,000 to 12,000 micromolar, while brain extracellular fluid holds 0.5 to 2 micromolar Hawkins 2009. Sodium-dependent EAAT transporters sit exclusively on the abluminal membrane of the blood-brain barrier. In bovine cerebral capillaries all three were detected on abluminal membranes and none on the luminal side. The apparent Km was 14 micromolar, with relative activities near 1:3:6 for EAAT1, EAAT2, and EAAT3 O'Kane 1999. They move glutamate from extracellular fluid toward the blood.

That arrangement allows no net entry of glutamate into the brain, and the barrier stays impermeable even at high plasma concentrations apart from the circumventricular organs Hawkins 2009. Glutamic acid is abundant in high-protein foods and in monosodium glutamate and related salts. The barrier pumps that glutamate outward toward the blood, so dietary and added glutamate does not drive central excitotoxicity.

The synapse requires a second permission signal

Glutamate alone does not open the NMDA receptor. D-serine is an endogenous ligand at the receptor's glycine site and fully occupies that site at some functional synapses. In rodent hippocampal recordings, selectively destroying D-serine with D-amino acid oxidase greatly attenuated NMDA receptor transmission and the calcium flux that follows it, and exogenous D-serine fully restored both Mothet 2000. The D-serine comes from the astrocyte, so permission to open the channel is held outside the neuron. The neuroglia lesson carries the transporter densities and the astrocytic wrap that put it there.

Zinc is a standing brake on the receptor

Synaptic zinc restrains the NMDA receptor at nanomolar concentrations. In recombinant receptors expressed in HEK 293 cells and Xenopus oocytes, voltage-independent zinc inhibition of NR1a-NR2A receptors had an IC50 in the nanomolar range, against the micromolar range for NR1a-NR2B Paoletti 1997. Trace zinc contaminating standard solutions tonically inhibits NR1a-NR2A, so chelating it potentiates the response. Low synaptic zinc disinhibits NMDA signaling and prolongs calcium elevation, and zinc in surplus is toxic on its own.

L-arginine promotes the cascade by supplying substrate for nitric oxide synthase. Inhibitory tone from GABA, adequate zinc, and intact mitochondrial buffering weigh the other way. Excitotoxicity is a tilted equilibrium held by clearance, inhibition, and permission, and the model reads that equilibrium as tone doing its work in the glutamate system. Health here is a defended range and not one perfect concentration. A system can lose some clearance capacity and still hold the range. A system that has spent that margin turns an ordinary burst of glutamate into an injury.

06Location and the failed trials

The same NMDA receptor builds protection at the synapse and death outside it

Stimulating synaptic NMDA receptors acts primarily through nuclear calcium signaling and builds up a neuroprotective shield, while stimulating extrasynaptic NMDA receptors promotes cell death Hardingham 2010. Ischemia and Huntington's disease are read in that literature as a perturbed balance between the two populations. The pathology sits in the ratio, and the ratio is a property of the whole synapse.

The two populations run opposite programs on the same molecules. In cultured neurons, extrasynaptic stimulation shut off CREB and collapsed mitochondrial membrane potential, while synaptic stimulation induced CREB activity and BDNF expression and was anti-apoptotic Hardingham 2002. One receptor type carries both the survival signal and the death signal, sorted by a few micrometers of membrane.

Blocking the receptor did not rescue the brain

Selfotel, aptiganel, eliprodil, licostinel, and gavestinel all failed in clinical trials of stroke and traumatic brain injury Ikonomidou 2002. The proposed reason is timing. Glutamate is destructive during a brief acute phase after the insult and afterwards resumes its normal survival-promoting role, so blockade of NMDA transmission hinders neuronal survival. A drug given on a clinical timescale meets a receptor doing its ordinary job, and the blockade subtracts the shield that synaptic activity was building Hardingham 2010.

Selfotel went past failing. In the acute ischemic stroke trials, 90-day mortality did not differ between the arms. Day-30 mortality ran higher on the drug, 54 of 280 against 37 of 286, and the trials were suspended Davis 2000. Removing the whole receptor population removed the survival signaling along with the injury signaling. The full mortality tables and the stroke glutamate concentrations are read as clinical numbers on brain injury.

The Unified Model of Tone separates inputs by what they do to the state receiving them, and a receptor blocker sits on the wrong side of that line. Selfotel subtracted the same NMDA current from every patient, from brains starved of the synaptic traffic that builds the protective program and from brains flooded with extrasynaptic traffic alike. Lowering excitation in everyone rescued no one, because the quantity of excitation was never the disorder. An input that helps has to match the clearance, the occupancy, and the location a given nervous system is actually holding, and no dose of a channel blocker carries that information.

07From sensitization to death

One receptor carries the whole range from central sensitization to cell death

Cerebrospinal fluid and plasma glutamate measured higher in 128 patients with ischemic stroke than in 43 controls, and both scaled with infarct size and with the severity of the neurological deficit Castillo 1996. The concentrations themselves, and what a rising value predicts, are read on brain injury.

Peripheral tissue input sets central excitatory tone through the same receptor. Twenty seconds of 1 Hz C-fiber strength stimulation of the sural nerve produced prolonged facilitation of the flexion reflex in the decerebrate rat. Both MK-801 and D-CPP reduced that facilitation. Given after a facilitation had already been established with cutaneous mustard oil, both returned the reflex to its pretreatment level Woolf 1991. Twenty seconds of afferent traffic rewrote spinal excitability. The induction and the maintenance of central sensitization both depend on NMDA receptor activation, and both are reversible.

Those two experiments set the boundaries of one continuum. Stroke sits at one end, where glutamate rises far enough to kill tissue outright and the concentration tracks the size of the infarct Castillo 1996. Central sensitization sits at the other, where the same receptor changes how a spinal circuit answers ordinary input and nothing dies Woolf 1991. One chemistry, read at two magnitudes.

Between those ends, the same glutamate, calcium, and nitric oxide chemistry sets how readily networks fire and fatigue, how a system handles metabolic load, and whether plasticity is gained or lost. The stroke and migraine spectrum and the peripheral and central sensitization of pain sit on that span.

Signs of excessive excitation an examination can document

Excitotoxicity is physiology to be read, not a condition treated by adjustment. The examination asks whether a presentation is suggestive of excessive excitation. It records signs an examiner can observe: chronic pain that has stopped responding to peripheral treatment, and gait changes that appear only under a dual-task load. These are correlates, not a diagnosis, and they describe the surface of a nervous system whose excitatory and inhibitory balance has shifted.

Where chiropractic sits in this chemistry

We are the practice organized around the nervous system, and the input we deliver is graded mechanical and sensory information carried by peripheral afferents into the cord. Woolf's twenty seconds is the reason that matters here. Afferent traffic sets NMDA-dependent excitability in the dorsal horn, and that setting can be moved in both directions. The Unified Model of Tone holds that changing what a segment receives changes what its dorsal horn amplifies. The excitatory balance follows the input a segment gets, not the map of where the pain is reported.

That prediction is testable in this lesson's own quantities. Transporter capacity, dorsal-horn NMDA excitability, and infarct size after ischemia moving together is what establishes the unification the prediction rests on. The general form of that commitment is set out on tone.

08Tone

How this system expresses tone

Every part of the nervous system expresses all of tone. In excitotoxicity three foundations carry the signature.

The remaining foundations run through the same calcium. Set point: resting intracellular calcium is defended near 0.1 micromolar, which is what makes any rise legible as a message. Gain: chelating the trace zinc that contaminates standard solutions potentiates NR1a-NR2A responses, so the answer to identical glutamate is set by a metal in the background. Coupling: neuron, astrocyte, and capillary endothelium run as one clearance system, and the barrier's glutamate transporters face inward, on the abluminal membrane only. Oscillation: glutamate normally arrives in brief bursts, and sustained occupancy of the NMDA channel is the condition that turns signaling toxic. Prediction: synaptic NMDA activity builds a protective shield in advance, so the cell is defended by traffic that arrived before the insult. Load: PARP-1 consumes NAD and ATP as it works, which is how a repair response becomes an energy crisis.

09Across the library

How this page relates to the rest of the library

Where glutamate, calcium, and clearance are read across the rest of the library.

10Frequently asked

Questions about this topic

11The sources

References

Sources: primary literature, linked inline.