NMDA Receptor-Mediated K+ Efflux and Neuronal Apoptosis

See allHide authors and affiliations

Science  09 Apr 1999:
Vol. 284, Issue 5412, pp. 336-339
DOI: 10.1126/science.284.5412.336


Neuronal death induced by activatingN-methyl-d-aspartate (NMDA) receptors has been linked to Ca2+ and Na+ influx through associated channels. Whole-cell recording from cultured mouse cortical neurons revealed a NMDA-evoked outward current,I NMDA-K, carried by K+ efflux at membrane potentials positive to –86 millivolts. Cortical neurons exposed to NMDA in medium containing reduced Na+ and Ca2+ (as found in ischemic brain tissue) lost substantial intracellular K+ and underwent apoptosis. Both K+ loss and apoptosis were attenuated by increasing extracellular K+, even when voltage-gated Ca2+channels were blocked. Thus NMDA receptor–mediated K+efflux may contribute to neuronal apoptosis after brain ischemia.

N-methyl-d-aspartate (NMDA) receptor–gated channels are permeable to Ca2+, Na+, and K+ (1). NMDA receptor–mediated Na+ and Ca2+ influx participates in synaptic transmission (2, 3) and excitotoxicity (4). In contrast, NMDA receptor–mediated K+ efflux has received little scrutiny, and its functional significance in either normal or abnormal states has not been defined. Stimulating NMDA receptors can induce central neuronal apoptosis (5, 6), and loss of cellular K+ may be a key step in caspase activation (7) and programmed cell death (8,9). We set out to test the hypothesis that NMDA receptor–mediated K+ efflux might promote neuronal apoptosis.

To detect K+ efflux through NMDA receptor channels, the membrane current triggered by NMDA was recorded in mouse cortical neurons by means of patch clamp whole-cell recording in an extracellular solution where Ca2+ and Na+ were replaced by the impermeable cationN-methyl-d-glutamine (NMG) (10). At the membrane potential of –60 mV, where NMDA normally evokes inward currents in bathing solutions containing physiological concentrations of Na+ and Ca2+, application of 200 μM NMDA plus 10 μM glycine induced an outward current, designated here asI NMDA-K, of 33 ± 6 pA (mean ± SEM,n = 11 cells; Fig. 1). This outward current was enlarged at depolarized membrane potentials, reaching 315 ± 39 pA at 0 mV (n = 11). The current-voltage curve of I NMDA-K showed slight outward rectification, and the current reversed at –86 ± 4 mV (n = 8), near the calculated K+equilibrium potential of –93 mV. TheI NMDA-K reversal potential shifted toward more positive potentials when external K+ was increased from 3 mM to 25 mM (Fig. 1) or when internal K+ was decreased from 120 mM to 12 mM [an inward current of –29 ± 10 pA was then triggered by NMDA at –60 mV (n = 4)].

Figure 1

NMDA receptor channel–mediated outward currentI NMDA-K. Local application of 200 μM NMDA plus 10 μM glycine triggered outward whole-cell currents in a Na+/Ca2+-free medium (inset) (10). Raising external K+ from 3 mM (n = 11 cells) to 25 mM (n = 8) shifted the reversal potential of the current to the right, which is consistent with K+ as the charge carrier. Error bars indicate mean ± SEM.

We next examined whether NMDA receptor–mediated K+ efflux could affect neuron viability in mixed cortical cultures containing neurons and glia (12 to 13 days in vitro) (11,12). In medium containing physiological concentrations of Na+ and Ca2+, overstimulation of NMDA receptors in our cortical cultures triggered neuronal necrosis characterized by prominent, acute, cell body swelling, little or no DNA laddering, and insensitivity to protein synthesis inhibition (4,13). However, if NMDA-induced inward cation influx was attenuated by lowering extracellular Na+ and Ca2+ to 30 mM and 0.1 mM, respectively (a maneuver that in itself was not toxic), adding 100 μM NMDA plus 10 μM glycine for 1 hour (followed by returning to control medium plus 10 μM MK-801) produced widespread neuronal apoptosis over the next 23 hours. This NMDA-induced apoptosis was characterized by cell body shrinkage, nuclear condensation, internucleosomal DNA fragmentation, and sensitivity to protein synthesis inhibition with cycloheximide (CHX) or caspase inhibition with Z-Val-Ala-Asp-fluoromethylketone (Z-VAD) (Fig. 2, A and B, and Fig. 3A). Neuronal cell body shrinkage preceded neuronal death (Fig 2C); neuronal cell cross-sectional area decreased from 223.0 ± 4.3 μm2 to 146.9 ± 3.4 μm2 at the end of the 1-hour NMDA exposure (a 34% reduction, different at P < 0.01; n = 206 cells per group) and was accompanied by substantial loss of intracellular K+ (Fig. 2D) (12). Consistent with the hypothesis that the observed cellular K+ loss and neuronal apoptosis were mediated by I NMDA-K, they were blocked by the NMDA antagonist, 1 μM MK-801 (Fig. 3A) (cellular K+ content was 111 ± 7% of the control at the end of the 1-hour NMDA exposure, n = 3 assays). Increasing external K+ from 5 to 25 mM, which reduced K+ efflux, attenuated the NMDA-induced neuronal death. High-K+ medium may increase intracellular Ca2+because of the opening of voltage-gated Ca2+ channels upon membrane depolarization; however, the neuroprotection by 25 mM K+ was not affected when voltage-gated Ca2+channels were blocked by addition of 5 or 10 μM gadolinium [Gd3+ (Fig. 3A)] (9). Increasing extracellular K+ also blocked cellular K+ loss (K+ content at the end of the NMDA treatment was 105 ± 4% of the control, n = 3) and attenuated cell body shrinkage (the cell cross-sectional areas at the end of the NMDA exposure were 66 ± 2% and 80 ± 3% of controls in 5 and 25 mM K+ medium, respectively; P < 0.01,n ≥ 140 cells for each group). No additive protection was observed with the co-application of Z-VAD and 25 mM K+(Fig. 3A).

Figure 2

NMDA-induced neuronal apoptosis, cell body shrinkage, and cellular K+ loss. (A) Transmission electron micrographs (9) of neurons 3 hours after sham wash or after the 100 μM NMDA plus 10 μM glycine exposure in the 30 mM Na+ medium (11), showing cell body shrinkage, nuclear condensation, and apoptotic bodies. Scale bar, 4 μm. (B) Agarose gel electrophoresis of DNA extracted from cultures (9) after sham wash or 3 hours after NMDA exposure in the 30 mM Na+ medium, showing characterized DNA fragmentation (laddering). (C) NMDA-induced cell body shrinkage. Mixed cortical cultures were exposed for 1 hour in the 30 mM Na+ medium, which itself was neither toxic nor changed the morphology of cells (sham, representative of eight cultures), whereas adding 100 μM NMDA plus 10 μM glycine caused marked cell body shrinkage after 1 hour of exposure (NMDA, representative of 24 cultures). Photos were focused on the maximum cell diameter (12). Scale bar, 50 μm. (D) The cell body shrinkage was accompanied by cellular K+ loss during and after the NMDA exposure [n = 4 assays for sham control (solid circles) and NMDA groups (open circles), respectively;n = 7 for the NMDA group at 1 hour of exposure]. After the 1-hour NMDA exposure, cells were returned to control medium containing 10 μM MK-801. No cell death was found at this time (n = 8 cultures) by either propidium iodide staining or LDH release (12), although 3 hours later, about 30% cell death was detected. Co-applied MK-801 (1 μM) or raising the extracellular K+ concentration to 25 mM prevented the K+ loss. Asterisks indicate difference from controls at each time point or before treatment [P < 0.01, one-way analysis of variance (ANOVA)].

Figure 3

NMDA- and endogenous glutamate–induced neuronal apoptosis. (A) Neuronal apoptosis triggered by exogenously added NMDA. Mixed cortical cultures were exposed for 1 hour to 100 μM NMDA in the low-Na+ (30 mM), low-Ca2+ (0.1 mM) medium and then returned to control medium containing 10 μM MK-801. Cell death was assayed by LDH release and cell counts 24 hours after the onset of insult (12). Exposure to the 30 mM Na+ solution alone for 1 hour caused no perceptible cell death. NMDA-triggered neuronal death was attenuated by co-applied CHX (1 μg/ml) or 100 μM Z-VAD or by raising extracellular K+ to 25 mM. The 25 mM K+ effect was not affected by adding the Ca2+ channel antagonist Gd3+ (5 or 10 μM); Gd3+ alone at these concentrations had no effect on NMDA-induced apoptosis [n = 4 cultures (23)]. No synergy was seen when both Z-VAD and 25 mM K+ were applied. Similar neuronal apoptotic death was induced when extracellular Mg2+ was reduced from 2 to 1 mM (n = 12 cultures), a concentration that precluded a NMDA receptor–mediated, Mg2+ influx–induced neuronal necrosis (24). The CHX- and Z-VAD–insensitive residual cell death was likely due to necrosis. (B) Neuronal apoptosis triggered by endogenous glutamate. Exposure for 2 hours to the bathing medium containing 3 mM Na+ and 0.1 mM Ca2+, followed by return to normal medium plus MK-801 (10 μM) and 6-cyano-7-nitroquinoxaline-2,3-dione (100 μM), triggered substantial neuronal cell death 22 hours later (leftmost bar) (11). This death was largely blocked if the NMDA antagonists MK-801 (1 μM) ord-3-(2-carboxypiperazin-4-yl-)-propyl-1-phosphonic acid (D-CPP) (100 μM) or the glycine antagonist 7-chlorokynurenate (7-CK) (100 μM) was included during the 2-hour exposure. CHX (1 μg/ml), 100 μM Z-VAD, or 25 mM K+, but not TEA (5 mM), also attenuated neuronal death. The Ca2+ channel antagonist Gd3+ (5 or 10 μM) did not occlude the protective effect by 25 mM K+. Twelve cultures were used for each test in (A) and (B). Asterisks indicate difference from the 3 mM Na+medium alone (P < 0.01, one-way ANOVA).

In cultured rat hippocampal slices, lowering extracellular Na+ to 3.6 mM for 30 min induced a 20- to 40-fold increase in extracellular glutamate (14), probably because of a block or reversal of the Na+-dependent high-affinity glutamate transporter on neurons and astrocytes (15). We used this protocol to determine whether endogenous glutamate release could induce NMDA receptor–mediated K+ efflux and consequent neuronal apoptosis. Mixed cultures exposed for 2 hours to a medium containing 3 mM Na+ and 0.1 mM Ca2+induced shrinkage of neuronal cell bodies, followed by widespread neuronal apoptosis 22 hours later (Fig. 3B), as evidenced by morphology under electron microscopic examination and the presence of DNA laddering on agarose gel. The cell body shrinkage and cell death were attenuated by addition of one of several different NMDA antagonists or by the raising of external K+ to 25 mM, but not by addition of the K+ channel blocker tetraethylammonium (TEA) (5 mM) (Fig. 3B). Endogenous glutamate-induced apoptosis, sensitive to CHX, MK-801, or Z-VAD, was also observed in nearly pure neuronal cultures (12 days in vitro; n = 8 cultures for each test), although longer exposure (24 hours) to the 3 mM Na+ medium was required, possibly reflecting the absence of glutamate export from glial cells.

We have thus electrophysiologically isolated the outward NMDA receptor–mediated K+ currentI NMDA-K and demonstrated its ability to promote neuronal apoptosis. The ionic basis (the K+ efflux) of NMDA receptor–mediated apoptosis is distinct from the ionic basis (the Ca2+ and Na+ influx) of NMDA receptor–mediated necrosis. Although the latter may predominate at normal physiological levels of extracellular Na+ and Ca2+, the demonstration here of NMDA-induced apoptosis in extracellular medium containing 30 mM Na+ and 0.1 mM Ca2+ may be relevant to the neuronal apoptosis observed after brain ischemia in vivo (16), where comparable concentrations of these ions have been measured (17). NMDA receptor–mediated K+ efflux may also contribute to neuronal apoptosis under other conditions, as it occurs even in the presence of normal concentrations of extracellular Ca2+ and Na+ (1).

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: choid{at}


View Abstract

Navigate This Article