Mediation of Neuronal Apoptosis by Enhancement of Outward Potassium Current

See allHide authors and affiliations

Science  03 Oct 1997:
Vol. 278, Issue 5335, pp. 114-117
DOI: 10.1126/science.278.5335.114


Apoptosis of mouse neocortical neurons induced by serum deprivation or by staurosporine was associated with an early enhancement of delayed rectifier (I K) current and loss of total intracellular K+. This I Kaugmentation was not seen in neurons undergoing excitotoxic necrosis or in older neurons resistant to staurosporine-induced apoptosis. Attenuating outward K+ current with tetraethylammonium or elevated extracellular K+, but not blockers of Ca2+, Cl, or other K+ channels, reduced apoptosis, even if associated increases in intracellular Ca2+ concentration were prevented. Furthermore, exposure to the K+ ionophore valinomycin or the K+-channel opener cromakalim induced apoptosis. Enhanced K+ efflux may mediate certain forms of neuronal apoptosis.

Neurons undergo apoptosis during normal development and in certain disease states (1). Elevated extracellular K+ interdicts this death (2,3), an effect attributed to increasing Ca2+ influx through voltage-gated Ca2+ channels, thus increasing intracellular Ca2+ concentration ([Ca2+]i) toward a set point that is optimal for survival (3). The antiapoptotic effect of high K+ concentration can be attenuated by removing extracellular Ca2+ or adding Ca2+-channel blockers (3, 4), and neuronal survival can be enhanced by opening Ca2+ channels with dihydropyridine agonists (4, 5) or inhibiting intracellular Ca2+sequestration with thapsigargin (6). However, although these data support an antiapoptotic effect of increasing [Ca2+]i, they do not exclude other possibilities. Furthermore, increases in [Ca2+]i can induce apoptosis under some conditions (7), and, in the absence of nerve growth factor, a high concentration of K+ promoted survival of sympathetic neurons without an increase in [Ca2+]i(8).

To test the idea that high extracellular K+ might also attenuate neuronal apoptosis by reducing K+ efflux, we first examined whether neurons undergoing apoptosis exhibited an up-regulation of outward K+ currents. Cultured mouse cortical neurons in mixed (neurons and glia) or near-pure neuronal cultures (9, 10) were studied with whole-cell recording between 7 and 12 days in vitro (DIV) (11). A persistent outward current, consistent with the delayed rectifier IK (Fig. 1A), and a transient outward current, consistent with IA(Fig. 1B) (12), were the two major voltage-gated K+ currents observed. IK was reduced by increasing extracellular K+ (Fig. 1C) and exhibited slow kinetics, outward rectification, and sensitivity to tetraethylammonium (TEA) (Fig. 1A) (13). IA exhibited rapid activation and inactivation as well as sensitivity to 4-aminopyridine (4AP) (Fig. 1B). A small inward rectifier current activated with hyperpolarization and a small outward current consistent with the ATP-sensitive K+ current (KATP) was also observed. Other major K+ currents, the M-current ( IM) and the Ca2+-dependent, high-conductance K+ current (BK current), were not detected.

Figure 1

Voltage-gated K+currents in cultured cortical neurons. (A) NormalI K recorded from neurons in mixed neuronal glial cultures. Current was activated by stepping from a holding potential of −70 to +40 mV for 300 to 600 ms, with leak current subtraction. The noninactivating outward current was dose dependently blocked by 1 to 40 mM bath-applied TEA (effects shown are 5 and 40 mM). (B) 4AP selectively blocked a transient outward K+ current, I A. Currents were activated by a voltage step from −100 to −10 mV. The initial transient outward current was blocked by 5 mM 4AP, consistent withI A. The 4AP resistant current was blocked by TEA, consistent with I K as shown in (A). (C) I K was sensitive to extracellular K+ concentration. When K+ was increased from 5 to 22.5 mM, the same voltage step activated a smaller outward current, similar to the effect of 5 mM TEA. The inhibitory effect of medium with high K+ concentration is expected from a reduced driving force for K+ efflux, predicted by the Nernst equation.

In seven to nine DIV neurons in neuronal cultures, the steady-state current of IK activated by voltage jump from −70 to +40 mV was 506 ± 46 pA (n = 25 cells, mean ± SEM). Six hours after serum withdrawal, IK at +40 mV was increased by 61% and the maximal K+ conductance more than doubled without change in voltage sensitivity or kinetics (Fig. 2, A, C, and D; Table1). Holding current at −70 mV shifted from −18 ± 6 pA at baseline to −2 ± 4 pA after 5 hours of serum deprivation (P < 0.05; n = 44 and 23, respectively), indicative of membrane hyperpolarization. Cell capacitance gradually decreased, consistent with progressive cell body shrinkage (Table 1), and total intracellular K+ dropped by 7 ± 3% and 13 ± 4%, respectively, after 5 and 9 hours in serum-free medium (both different from baseline at P < 0.05; n = 10) (14). The enhanced IK remained sensitive to block by TEA (Fig.2B); 5 mM TEA completely prevented loss of intracellular K+.

Figure 2

Enhancement ofI K by serum deprivation. (A) (Left) Recordings from DIV 9 neurons in pure neuronal cultures showing theI-V relationship of I K as revealed by rectangular and ramp voltage steps. (Right) After 6 hours of serum deprivation, the same voltage steps triggered much larger currents and outward rectification. (B) The I Kenhanced by serum deprivation was still sensitive to TEA. (C) Change in I K with time in control cells (•) and in cells undergoing apoptosis induced by serum deprivation (▪) (n = 5 to 15 cells at each point). (D) Increased K+ conductance byI K channels after 6 hours in serum-free medium. Asterisk indicates difference from control at P < 0.05 byt-test.

Table 1

Effects of apoptotic and necrotic insults on I K. I K, steady-state current activated by voltage step from −70 to +40 mV;G max, maximum conductance;V threshold, voltage threshold forI K activation; C M, membrane capacitance;V 2/1, voltage forI K half activation; τopening, time constant for I K opening at +40 mV measured in the presence of 5 mM 4AP by single exponential curve fitting. These parameters were examined before (control) and at plateau levels after insults: 11 hours in staurosporine (0.1 μM), 6 hours in serum deprivation, and 11 hours in 15 μM NMDA (plus 10 μM glycine; DIV 10 to 12 mixed cultures). Cycloheximide concentration was 1 μg/ml. Serum deprivation and staurosporine exposure were performed in pure neuron cultures and mixed cultures, respectively (9,10).

View this table:

No change in IK was seen in neurons exposed to a sham wash (Fig. 2C). The enhancement in IK induced by serum deprivation was not blocked by cycloheximide (1 μg/ml) (Table 1), suggesting that the enhancement did not require new protein synthesis.

Neuronal apoptosis induced in mixed cultures (DIV 10 to 12) by 0.1 μM staurosporine was also associated with an enhancement in IK. Despite an initial nonsignificant trend toward reduced IK after 30 min, after 9 to 11 hours IK was increased by 90% and maximum K+ conductance was doubled (Table 1). In contrast, IK did not change in neurons that underwent N -methyl-d-aspartate (NMDA)–induced excitotoxic necrosis (Table 1), although by 11 hours neurons exhibited substantial cell swelling. Furthermore, exposure to staurosporine did not alter IK in older neurons (DIV 17; n = 5), which survive such exposure (15).

Serum deprivation and staurosporine exposure also altered the transient current IA, although with opposite effects. At times of maximal change, serum deprivation (6 hours) increased IA by 28% (P < 0.05; n = 20), whereas staurosporine (0.1 μM; 9 hours) decreased IA by 75% (P < 0.05; n = 8). Exposure to NMDA did not alter IA. Because activity of voltage-gated Ca2+ channels may affect apoptosis, we monitored high-voltage–activated (HVA) Ca2+ currents during serum deprivation and found no significant change (232 ± 35 pA in control cells, 192 ± 33 and 147 ± 51 pA after 6 and 9 hours; P = 0.43 and 0.22, respectively; n = 5 per condition). Tested after 5 to 9 hours in 0.1 μM staurosporine, the IM and BK currents remained undetectable; the inward rectifier and KATPcurrent were not altered.

The selective enhancement of IK induced by either serum deprivation or staurosporine exposure, as well as the increase in IA induced by the former, occurred before development of neuronal apoptotic death (16), consistent with a critical early role. In support of this idea, adding 1 to 5 mM TEA or increasing extracellular K+ from 5 to 25 mM reduced both forms of apoptosis (Fig. 3, A and B). In contrast, neither the TEA analog tetramethylammonium (5 mM; inactive on IK) nor 4AP (5 mM; antagonist for the slowly inactivating K+current ID as well as IA) was effective (Fig. 3B). Other antagonists that targeted KATP (tolbutamide; 100 to 500 μM) or the SK channel (apamin; 1 μM) also lacked protective effect. TEA may inhibit Cl currents in cortical neurons (17); however, no neuroprotection was observed with the Cl-channel antagonist anthracene-9-carboxylic acid (500 μM) (data not shown).

Figure 3

Prevention of apoptosis byI K blocker TEA or by raising extracellular K+ concentration. Pure neuronal culture (DIV 7 to 9) in 24-well plates was used for serum deprivation because the presence of glia prevents neuronal degeneration after serum deprivation. Mixed culture containing neurons and a glia bed (DIV 10 to 12) was used for exposure to staurosporine (0.1 μM). NMDA receptor antagonist 1 μM dizocilpine maleate {(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine (MK-801)} and non-NMDA receptor blocker 2,3-dihydro-6-nitro-7-sulfamoylbenzo(f)quinoxaline (5 μM) were added to block glutamate excitotoxicity in serum-deprivation experiments. Neuronal death was detected 24 and 48 hours after apoptotic insult. (A) (Left) Phase-contrast micrograph of a pure neuronal culture 48 hours after onset of serum deprivation, showing widespread neuronal apoptosis. (Right) Preservation of neurons in serum-free medium in the presence of 5 mM TEA. Bar = 50 μm. (B) Neuronal apoptosis, expressed as a fraction of the total number of neurons, induced by 48 hours of exposure to serum deprivation (left) or 0.1 μM staurosporine (right), either alone or in the presence of the indicated bath-applied drug (mean ± SEM;n = 4 to 16 cultures per condition). Serum-deprivation–induced cell death was assessed by cell counts after staining with 0.4% trypan blue dye. For staurosporine experiments, cell death was measured by lactate dehydrogenase released into the medium (30). Cell deaths by both insults were assayed alternatively by two methods and similar results were obtained. Higher concentrations of TEA (20 mM) were toxic. Asterisk indicates significant difference from the control at P < 0.05 byt-test with Bonferroni correction for four comparisons.

We considered the possibility that the protective effect of TEA was mediated by an increase in [Ca2+]i. Neuronal [Ca2+]i measured with fura-2 (18) was about 100 nM at rest and increased to a plateau value of about 200 nM after 1 to 2 hours of exposure to 5 mM TEA or 25 mM K+. Gadolinium (2 to 10 μM), which completely blocked the HVA Ca2+ current (Fig. 4A), kept [Ca2+]i at resting level during 2 to 16 hours of exposure to TEA or 25 mM K+ (Fig. 4B); neither gadolinium (Fig. 4C) nor the L-type Ca2+-channel antagonist nifedipine (5 μM; Fig. 4D) blocked the antiapoptotic effects. Neither gadolinium nor nifedipine showed neuroprotection when applied alone (Fig. 4, B and C).

Figure 4

Protective effects of TEA and 25 mM K+ were not dependent on an increase in [Ca2+]i. (A) HVA Ca2+currents, activated by a voltage step from −70 to +10 mV, were blocked completely by bath-added 2 μM gadolinium (Gd3+). The same results were obtained from three additional experiments. (B) Before break, bath application of 5 mM TEA (▪) triggered an initial sharp increase in [Ca2+]i as measured by fura-2 video microscopy, followed by relaxation to a lower plateau value (mean ± SD; n > 30 neurons for each time point). Results were similar in two additional experiments. TEA plus staurosporine produced a pattern of [Ca2+]i increase similar to that produced by TEA alone; exposure to 0.1 μM staurosporine alone for 2 hours did not alter baseline neuronal [Ca2+]i(data not shown). Gd3+ (5 μM) completely blocked the TEA-induced increase in [Ca2+]i in the presence of 0.1 μM staurosporine. After break, in the presence of TEA plus Gd3+ (▴), [Ca2+]i remained at resting levels for up to 16 hours (mean ± SD; n ≥ 200 cells from four experiments). Similarly, 2 μM Gd3+completely prevented the 25 mM K+-induced [Ca2+]i increase (data not shown). •, wash control. (C) Ability of 5 mM TEA or 25 mM K+ to reduce neuronal apoptosis induced by serum deprivation (SD) (left) or staurosporine exposure (STP) (right) was not blocked by coapplication of 2 or 10 μM Gd3+ (n = 20 to 32 cultures per condition except n = 4 for staurosporine plus Gd3+ condition). SW, sham wash. (D) Neuroprotective effects of 5 mM TEA or 25 mM K+ against apoptosis induced by either serum deprivation SD (left) or staurosporine STP (right) were not affected by coapplication of 5 μM nifedipine (n = 12 for serum deprivation and n= 4 for staurosporine experiments). Cell death was measured as described in Fig. 3. Asterisk indicates significant difference from control at P < 0.05 by t-test with Bonferroni correction for two or three comparisons.

We considered the possibility that membrane depolarization might mediate the antiapoptotic effects of TEA or high K+ concentration. However, the Na+-channel opener veratridine, which depolarized the membrane from control −51 ± 2 mV (n = 45) to −30 ± 2 mV (n = 21), similar to the membrane depolarization by 5 mM TEA (−34 ± 4 mV; n = 15), increased staurosporine-induced neuronal cell death (Fig. 3B).

In support of the hypothesis that increased K+ efflux might be a primary step leading to apoptosis, application of the selective K+ ionophore valinomycin (19) triggers apoptosis in thymocytes, lymphocytes, and tumor cells (20). Exposure to 20 nM valinomycin for 24 to 48 hours induced typical neuronal apoptosis in cortical cultures, characterized by chromatin condensation, cell body shrinkage, internucleosomal DNA fragmentation, and sensitivity to cycloheximide (1 μg/ml) or the caspase inhibitor Z-Val-Ala-Asp-fluoromethylketone (100 μM) (zVAD; Fig.5). Furthermore, 24 to 48 hours of exposure to the K+-channel opener cromakalim, which activates KATP channels as well as IK-like currents in mammalian central neurons (21), also induced typical neuronal apoptosis (Fig. 5).

Figure 5

Valinomycin- and cromakalim-induced neuronal apoptosis. (A) Electron micrographs of a control neuron (left) and apoptotic neurons (arrows) induced by 24 hours of exposure to the selective K+ ionophore valinomycin (20 nM) (middle) or by the K+-channel opener cromakalim (500 μM) (right). Bar = 3 μm. (B) DNA laddering on agarose gels after 24 hours of exposure to valinomycin (lane 3) or cromakalim (lane 6). The marker columns (lanes 1 and 4) show Hind III-digested λ DNA. Lanes 2 and 5, controls. (C) Valinomycin-induced neuronal death, assayed by staining with 0.4% trypan blue dye, was attenuated by addition of cycloheximide (1 μg/ml) (CHX; n= 12) or 100 μM zVAD (n = 12). Asterisk indicates significant difference from control at P < 0.05 byt-test with Bonferroni correction for two comparisons. SW, sham wash.

In summary, four arguments suggest that a long-lasting enhancement of outward K+ current is a key mediator of the forms of cortical neuronal apoptosis studied here. First, augmentation in IK and loss of neuronal cell K+occurred early in the course of neuronal apoptosis, well before the commitment point (16). The magnitude of this increase in K+ current is comparable with that associated with mitogenesis (22) and proliferation (23) in several cell types (∼0.5- to 3-fold increases of peak outward K+ current). Second, this IKaugmentation was specific to apoptosis but not triggered by necrotic insult or in older cells resistant to apoptosis. Third, blocking of this IK enhancement and cellular K+depletion by TEA or by increasing extracellular K+prevented apoptosis. Finally, increasing membrane K+permeability by adding either the ionophore valinomycin or the endogenous K+-channel opener cromakalim sufficed to induce neuronal apoptosis.

In isolation, the third argument stated above is unsurprising, because both TEA and raising extracellular K+ increase neuronal excitation and [Ca2+]i, and thus reduced apoptosis would be predicted by the Ca2+ set-point hypothesis. However, neuroprotection was maintained even when associated increases in [Ca2+]i were completely blocked. Some membrane-associated signaling proteins such as adenylyl cyclase may be voltage sensitive (24), but comparable membrane depolarization induced by veratridine was not neuroprotective.

Involvement of K+ efflux in apoptosis has intuitive appeal, because loss of cell volume is a cardinal feature of apoptosis, and K+ extrusion is a plausible mechanism to achieve this loss (25). The IA blocker 4AP recently has been reported to inhibit the shrinkage of human eosinophils undergoing cytokine deprivation-induced apoptosis (26). Additional study is needed to delineate the exact mechanisms by which activation of IK might promote neuronal apoptosis. One possible arena for linkage between these events is in cell cycle control, because K+ channels and a decrease in intracellular K+ have been implicated in initiation of mitosis (27), and apoptosis has been postulated to reflect an “abortive mitosis” (28). Furthermore, agents that reduce intracellular K+ may activate caspases in macrophages or monocytes (29). We suggest that interventions directed at blocking excessive K+ efflux, in particular by the noninactivating delayed rectifier K+ channel, be explored as a strategy for attenuating neuronal apoptosis in disease states.

  • * To whom correspondence should be addressed at the Department of Neurology, Box 8111, Washington University School of Medicine, 660 South Euclid Avenue, St. Louis, MO 63110, USA. E-mail: choid{at}


View Abstract

Stay Connected to Science

Navigate This Article