Activation of Pannexin-1 Hemichannels Augments Aberrant Bursting in the Hippocampus

See allHide authors and affiliations

Science  05 Dec 2008:
Vol. 322, Issue 5907, pp. 1555-1559
DOI: 10.1126/science.1165209


Pannexin-1 (Px1) is expressed at postsynaptic sites in pyramidal neurons, suggesting that these hemichannels contribute to dendritic signals associated with synaptic function. We found that, in pyramidal neurons, N-methyl-d-aspartate receptor (NMDAR) activation induced a secondary prolonged current and dye flux that were blocked with a specific inhibitory peptide against Px1 hemichannels; knockdown of Px1 by RNA interference blocked the current in cultured neurons. Enhancing endogenous NMDAR activation in brain slices by removing external magnesium ions (Mg2+) triggered epileptiform activity, which had decreased spike amplitude and prolonged interburst interval during application of the Px1 hemichannel blocking peptide. We conclude that Px1 hemichannel opening is triggered by NMDAR stimulation and can contribute to epileptiform seizure activity.

Hemichannels are formed by pannexin or connexin proteins and mediate large ionic currents and the passage of small molecules (<1 kD) across plasma membranes. Pannexin-1 (Px1) forms hemichannels in a number of cell types and can be opened by ischemic-like conditions in pyramidal neurons (1) or purinergic receptor stimulation in red blood cells (2). Px1 has been observed at the postsynaptic density by electron microscopy and colocalization with postsynaptic density protein 95 (PSD95) (3); therefore, we hypothesized that Px1 hemichannels may have an undiscovered function at postsynaptic sites. Glutamate mediates excitatory synaptic communication via activation of fast AMPA/kainate and slower N-methyl-d-aspartate receptors (NMDARs). We investigated the possibility that NMDAR activation opens Px1 hemichannels because of reports that NMDARs lead to a prolonged but unidentified secondary inward current (46).

Under conditions in which voltage-dependent ion channels were blocked, we recorded NMDAR secondary currents (I2nd) from acutely isolated hippocampal neurons with whole-cell patch clamp and activated them by either repeated (10-s duration at 1-min intervals) (Fig. 1A) or continuous (5 to 15 min) 100 μM NMDA with concomitant voltage commands from –80 to +80 mV (Figs. 1D and 2) (7). I2nd was evident as an increase in holding current (Fig. 1A, downward shift in middle trace) and was secondary to the NMDAR because it persisted after washout of the agonist (Fig. 1A). Furthermore, I2nd was blocked by 50 μM carbenoxolone (Cbx) (Fig. 1, A to C), an inhibitor of gap junctions and hemichannels (8). We tested a selective small peptide inhibitor of Px1, 10panx (100 μM; WRQAAFVDSY) (9, 10), that blocked I2nd [Fig. 1, B and C; P < 0.05, analysis of variance (ANOVA)], whereas a scrambled version, scpanx (FSVYWAQADR), was ineffective (Fig. 1C; Px1 group, P > 0.05, ANOVA). Block of the NMDAR ligand-gated currents with 1 mM kyenurinic acid prevented activation of I2nd (fig. S1). Furthermore, the NMDAR currents were not directly affected by Cbx or 10panx as determined by applying these blockers before activation of I2nd (Fig. 1C; P > 0.05, ANOVA). Similar to Px1 activation by ischemia (1), I2nd had a linear current-voltage relation (Fig. 1B). Although this differs from some Px1 expression systems (8, 9, 11), it is similar to the “large-conductance” mode of P2X7 in human embryonic kidney cells (12), which may be mediated by Px1 (9).

Fig. 1.

NMDAR activation opens Px1 hemichannels in hippocampal neurons. (A) Continuous voltage-clamp recording (Vm = –60mV) from a neuron exposed to repetitive 10 s applications of 100 μM NMDA (bars). I2nd (downward shift in middle trace) was blocked by carbenoxolone (Cbx). Each spike is an applied voltage ramp from –80 to +80 mV. (B) Current-voltage plots of applied voltage ramps. I2nd was inhibited by Cbx and a peptide inhibitor of Px1, 10panx. (C) Quantification of NMDAR current (INMDA) and I2nd recorded from isolated hippocampal neurons. Activation of I2nd was significantly (asterisk) larger than control and blocked (two asterisks) by Cbx or 10panx, but not scrambled 10panx (scpanx). INMDA (before I2nd activation) was not affected by Cbx or 10panx. Error bars indicate SEM. (D) Acutely isolated hippocampal neuron loaded with Fluo-4 and calcein red/orange. Calcium remained elevated and calcein red/orange efflux occurred after NMDA exposure. (E) Efflux was blocked by 10panx when applied concomitantly with NMDA (red) or once dye loss had begun (blue), as quantified in (F). Cbx, 10panx, and Mg2+ prevented efflux. In the right graph of (F), a value of 1.0 indicated no effect of the blocker.

Fig. 2.

NMDAR I2nd block by RNA interference of Px1. (A) Cultured hippocampal neurons infected with GFP and shRNA against Px1 had significant (P < 0.05, ANOVA) knockdown of the hemichannel, as assayed by Western blot. siRNA, small interfering RNA. Error bars indicate SEM. (B) Time-dependent activation of the I2nd on exposure of cultures to NMDA (arrow) for the time indicated. (C) 10panx blocked the I2nd activated by 5-min NMDA. Knockdown of Px1 with shRNA, but not a scrambled shRNA (sc), blocked the I2nd recorded from GFP-positive neurons. (D) Quantification of I2nd (at –60mV) of cultured hippocampal neurons exposed to NMDA. All statistical comparisons were made to the current amplitude for a 5-min exposure to NMDA (ANOVA; P < 0.05). The asterisk indicates a significant difference from the 5-min NMDA application.

The pores of hemichannels are large enough to permit flux of large molecules (13), making dye flux a powerful tool for identifying the involvement of Px1. If Px1 mediates I2nd, then NMDAR activation should evoke efflux of calcein (a non-reactive fluorescent indicator) from hippocampal neurons. We loaded acutely isolated hippocampal neurons with calcein red/orange and with a calcium indicator (Fluo-4 or Fluo-4FF) to monitor activation of the NMDAR. NMDA (100 μM), applied for 5 to 10 min to acutely isolated hippocampal neurons in 0 Mg2+ solution to enhance NMDA currents without requiring simultaneous membrane depolarization, evoked rapid rises in intracellular calcium concentration ([Ca]i) that persisted after washout of the agonist. Calcein red/orange efflux from single neurons occurred with a delay (average 7.9 ± 1.6 min; n = 7 number of cells) after the NMDA-induced [Ca]i rise, (Fig. 1, D to F). Dye efflux was blocked when 100 μM 10panx was present (Fig. 1, E and F) (1, 14).

By using RNA interference of Px1 in cultured hippocampal neurons, we next confirmed that the NMDAR-evoked I2nd was due to Px1 hemichannels. We first demonstrated that short hairpin RNA (shRNA) delivered to cultured hippocampal neurons via lentivirus (see supporting online material) reduced Px1 levels. Infection of cultured neurons with green fluorescent protein (GFP) and the shRNA vector was achieved at >80% efficiency and with minimal infection of non-neuronal cells. This resulted in knockdown of the Px1 protein in the total culture to 43 ± 10% of control (P < 0.05; ANOVA; n = 4), as determined by Western blot analysis with an antibody against a C-terminal region of Px1 (Fig. 2A) (15). The remaining 43% may be due to expression of Px1 in other cell types in the culture, such as astrocytes (16), and incomplete efficacy of the shRNA. We then used sister cultures to test whether the NMDAR I2nd was reduced after shRNA knockdown of Px1. Intensely GFP-positive (that is, shRNA-expressing) neurons were patch-clamped, and NMDA (100 μM) was applied for 2 to 10 min, which activated I2nd (Fig. 2, B and D). Maximal activation was achieved after 5 min of agonist application (Fig. 2, B and D). The Px1 inhibitor, 10panx (100 μM) blocked activation of I2nd (Fig. 2, C and D), and shRNA-expressing neurons had significantly reduced I2nd (by >70%), but those infected with a scrambled shRNA were not different from control (Fig. 2, A, C, and D). The residual currents that remained (<30% of control) in the shRNA-infected individual neurons (Fig. 2, C and D) were probably due to incomplete block of Px1 expression by shRNA vectors, but the possibility that other cation channels make a minor contribution cannot be excluded.

The activation of Px1 hemichannels by NMDAR stimulation and the reported expression of these hemichannels at postsynaptic sites (3) raised the interesting possibility that Px1 can be opened by synaptically released glutamate. Hippocampal slices were thus exposed to low Mg2+ concentrations plus 5 mM KCl to potentiate NMDAR currents in intact tissue. These slices displayed a pattern of rhythmic epileptiform-like bursting called interictal spiking—a correlate to repetitive bursting electroencephalography rhythms observed in epilepsy patients (17). We confirmed that the initiation of interictal bursts in brain slices was triggered by synaptic NMDAR activation, as previously reported (18), by blocking the induction of bursting with NMDAR antagonists. We tested whether Px1 hemichannels are opened by synaptic activity (during 0 Mg2+ triggered bursting) via measuring dye uptake by neurons through Px1 and the electrophysiological contribution of Px1 to interictal discharges.

Exposure of 400-μm-thick rodent brain slices to nominally Mg2+-free solutions induced uptake of a fluorescent dye, sulforhodamine 101 (SR101), into neurons in the CA1 region of the hippocampus (Fig. 3). SR101 is normally a selective marker of astrocytes and is excluded from neurons (Fig. 3A) (19). However, in our previous work, SR101 influx into cortical neurons occurred during ischemia and was indicative of Px1 opening (1). In 0 Mg2+ exposed hippocampal slices, SR101 influx into CA1 hippocampal neurons was observed (Fig. 3A), and dye influx was blocked by 10panx or pre-treatment with 50 μM APV [(2R)-amino-5-phosphonovaleric acid], an NMDAR antagonist (Fig. 3).

Fig. 3.

NMDAR potentiation in hippocampal brain slices opens Px1. (A) CA1 region of the hippocampus is shown 75 μm deep into 400-μm slices. The red fluorescent dye SR101 is excluded from neurons (control). A 1-hour exposure to Mg2+-free (+5 mM KCl) bathing solution evoked dye uptake that was blocked by 10panx or APV. (B) Dye uptake, expressed as fluorescence intensity after treatment with 0 Mg2+, 0 Mg2+ + 10panx, or 0 Mg2+ + APV as the ratio of SR101 intensity to that in control slices. Error bars indicate SEM.

To further test the involvement of Px1 in hippocampal synaptic function, we recorded the extracellular field potentials that spontaneously occur after a ∼1-hr exposure of hippocampal slices to 0 Mg2+ plus 5 mM KCl bathing solutions. These interictal bursts (Fig. 4A) in the CA1 region occurred at 0.6 ± 0.2 Hz (n = 5 slices), and the addition of 10panx reduced the interburst frequency by 24 ± 6%, which recovered to 103 ± 8% of control upon washout of the blocking peptide (Fig. 4, A, B, and D). Block of Px1 during interictal bursting also decreased the mean amplitude of spikes within a burst by 34 ± 6% (P < 0.05 by ANOVA), with recovery to 73 ± 12% of control (n = 5) (Fig. 4, A to C). When 10panx was coapplied initially with the 0 Mg2+ solutions, interictal bursting still occurred, but the amplitude and frequency increased after washout of the peptide (fig. S2). The effect of Px1 block on spike properties and bursting in the epileptic-like hippocampus was not attributable to nonspecific inhibition of fast (AMPA receptor–mediated) synaptic transmission because 10panx did not affect field potentials evoked by stimulation of the Schaeffer collateral pathway in the presence of 2 mM extracellular Mg2+ (Fig. 4D).

Fig. 4.

Px1 block alters bursting properties of hippocampal neurons during seizure-like events. (A) Interictal spiking, induced by low Mg2+ concentration plus 5 mM KCl, was affected by 10panx. Decreases in burst frequency and amplitude are evident. (B) Expanded bursts demonstrated significant (ANOVA; P < 0.05) alteration of electrophysiological properties by Px1 inhibition. (C) Cumulative frequency plots of spike amplitude and interburst interval showing decreased spike amplitude and a longer interval between bursts when Px1 was blocked. (D) Evoked potentials by stimulation of Schaeffer collaterals in the presence of 2 mM extracellular Mg2+ to block NMDARs, demonstrating that 10panx did not directly block fast synaptic transmission via AMPA receptors.

The data presented here demonstrate that Px1 hemichannels are activated by NMDARs and contribute to postsynaptic responses in the hippocampus during seizure-like activity. This result is consistent with Px1 providing a tonic depolarizing current in dendrites during intense NMDAR activity. This could be initiated either synaptically or extra-synaptically, but the presence of Px1 in the postsynaptic density suggests that synaptic activation of Px1 is probably involved. We have determined that Px1 hemichannels constitute a major portion of the NMDAR-evoked I2nd, which can lead to Ca2+ deregulation in hippocampal pyramidal neurons and possibly to the establishment of acquired epilepsy (46, 20). It has been suggested that interneuronal gap junctions contribute to epileptogenesis in the hippocampus (21). It will be interesting to determine the relation between these junctions and Px1 hemichannel activation during seizures.

What is the mechanism that links NMDAR to Px1 activation? Surprisingly, it does not appear to involve increased intracellular Ca2+, because altering neuronal Ca2+ buffering did not affect Px1 opening (fig. S3). The mechanism may, however, involve intracellular adenosine triphosphate (ATP) depletion during NMDAR activation (22), because intracellular perfusion with 4 mM MgATP significantly protected against Px1 activation; current density was –30.2 ± 8.6 pA/pF (n = 3) with 1 mM MgATP and –9.4 ± 3.7 pA/pF (n = 5) with 4 mM MgATP. We recently reported that Px1 is opened by ischemia (1) and that the timing of this opening appears to follow the anoxic depolarization (23) in a manner analogous to Px1 activation after NMDAR stimulation. Therefore, Px1 not only appears to be involved in neuronal dysfunction during ischemia but also plays a role in the potentiation of seizure-like activity. These unique ion channels should therefore be considered important targets for the treatment of neurological disorders such as epilepsy and stroke.

Supporting Online Material

Materials and Methods

Figs. S1 to S3


References and Notes

View Abstract

Navigate This Article