Contribution of Astroglial Cx43 Hemichannels to the Modulation of Glutamatergic Currents by D-Serine in the Mouse Prefrontal Cortex

Abstract

Astrocytes interact dynamically with neurons by modifying synaptic activity and plasticity. This interplay occurs through a process named gliotransmission, meaning that neuroactive molecules are released by astrocytes. Acting as a gliotransmitter, D-serine, a co-agonist of the NMDA receptor at the glycine-binding site, can be released by astrocytes in a calcium [Ca²⁺]_i-dependent manner. A typical feature of astrocytes is their high expression level of connexin43 (Cx43), a protein forming gap junction channels and hemichannels associated with dynamic neuroglial interactions. Pharmacological and genetic inhibition of Cx43 hemichannel activity reduced the amplitude of NMDA EPSCs in mouse layer 5 prefrontal cortex pyramidal neurons without affecting AMPA EPSC currents. This reduction of NMDA EPSCs was rescued by addition of D-serine in the extracellular medium. LTP of NMDA and AMPA EPSCs after high-frequency stimulation was reduced by prior inhibition of Cx43 hemichannel activity. Inactivation of D-serine synthesis within the astroglial network resulted in the reduction of NMDA EPSCs, which was rescued by adding extracellular D-serine. We showed that the activity of Cx43 hemichannels recorded in cultured astrocytes was [Ca²⁺]_I dependent. Accordingly, in acute cortical slices, clamping [Ca²⁺]_i at a low level in astroglial network resulted in an inhibition of NMDA EPSC potentiation that was rescued by adding extracellular D-serine. This work demonstrates that astroglial Cx43 hemichannel activity is associated with D-serine release. This process, occurring by direct permeation of D-serine through hemichannels or indirectly by Ca²⁺ entry and activation of other [Ca²⁺]_i-dependent mechanisms results in the modulation of synaptic activity and plasticity.SIGNIFICANCE STATEMENT We recorded neuronal glutamatergic (NMDA and AMPA) responses in prefrontal cortex (PFC) neurons and used pharmacological and genetic interventions to block connexin-mediated hemichannel activity specifically in a glial cell population. For the first time in astrocytes, we demonstrated that hemichannel activity depends on the intracellular calcium concentration and is associated with D-serine release. Blocking hemichannel activity reduced the LTP of these excitatory synaptic currents triggered by high-frequency stimulation. These observations may be particularly relevant in the PFC, where D-serine and its converting enzyme are highly expressed.

Keywords: astrocyte; connexin; gliotransmitter; hemichannel; neuroglial interaction; prefrontal cortex.

Figure 1.

A, Illustration of experimental arrangement and astrocyte properties in PFC slices from hGFAP-eGFP mice. One pyramidal cell in L5 was loaded with biocytin (red; 1 mg/ml) during whole-cell recording in PFC acute slices, whereas several astrocytes expressed eGFP (green). Stimulation electrode was placed in L2/3 and dual recording of an astrocyte and a pyramidal cell was performed in L5. The average distance between the stimulating electrode and the soma of the recorded L5PC was ∼300 μm. B, Confocal image of a representative eGFP-positive astrocyte in L5 of the PFC. Scale bars, 100, 100, and 25 μm, respectively. C, Dye coupling in L5 of the PFC. An eGFP-positive astrocyte was recorded for 10 min in whole-cell configuration with a patch pipette containing sulforhodamide B (1 mg/ml) that diffused in neighboring cells through gap junction channels.

Figure 2.

Inhibition of Cx43 hemichannel activity in astrocytes affects NMDA EPSCs in the PFC. A1, A2, EPSCs (averaged from five consecutive traces) generated by stimulation in L2/3 were recorded, in the presence of picrotoxin (100 μm) in pyramidal cells from L5 held at + 40 mV or −70 mV to isolate NMDA and AMPA EPSCs, respectively. Scale bars, 100 pA and 100 ms. A1, Typical NMDA and AMPA EPSCs in control condition. A2, Typical NMDA and AMPA EPSCs in presence of Gap26. B1, B2, Differential effect of Gap26 (200 μm) on the amplitude of AMPA (B1) and NMDA (B2) EPSCs. Note that there was no significant difference between control and Gap26 for AMPA currents (p > 0.05, t test, n = 7), whereas the amplitude of NMDA currents was significantly reduced (p < 0.05, t test, n = 7). C, Mean of NMDA EPSC amplitude in control conditions and in presence of Gap26. D, Quantification of AMPA/NMDA ratio under the indicated conditions. The AMPA/NMDA ratio was significantly increased in Cx43 KO mice (p < 0.05, t test, n = 7) and in presence of Gap26 (p < 0.05, t test, n = 7), whereas the Gap26 scramble peptide had no effect (p > 0.05, t test, n = 7).

Figure 3.

Modulation of NMDA EPSCs in pyramidal cells by D-serine involves astrocyte hemichannel activity. A1, A2, Typical recording of NMDA EPSCs at +40 mV (scale bars, 100 pA and 100 ms) in presence of picrotoxin (100 μm) and NBQX (10 μm). A1, NMDA EPSCs (averaged from 5 consecutive traces) were abolished by the NMDAR blocker D/L APV (p < 0.001, t test, n = 7). A2, NMDA EPSCs were significantly decreased in presence of Gap26 (p < 0.05, t test, n = 9) and this effect was rescued by adding D-serine (100 μm) to the extracellular solution. B, Graph summarizing the relative NMDA EPSC amplitude of the samples shown in A and B and for several concentration of D-serine in the external solution. Note that the rescue of Gap26 inhibitory effect became statistically significant for 10 and 100 μm D-serine (p < 0.05, t test, n = 7).

Figure 4.

Activation of Cx43 hemichannel in cultured cortical astrocytes is Ca²⁺ dependent. A1–A3, Fluorescent images of EtBr uptake in cultured astrocytes in control condition (CTL), after treatment with ionomycin (1 μm), and after treatment with ionomycin plus CBX (50 μm). Scale bar, 5 μm. Note that, whereas in control conditions, the uptake of EtBr was not observed (A1), it became detected in the presence of the Ca²⁺ ionophore ionomycin (1 μm; A2) and was prevented by adding CBX to ionomycin (A3). B, Time-dependent increase in resting [Ca²⁺]_i recorded in astrocytes loaded with Fluo-4/AM and treated with ionomycin. Error bars indicate mean ± SD, n > 100 astrocytes for the same experiment. C, Histogram showing EtBr uptake expressed as the mean fluorescence intensity per astrocyte nucleus in control and after treatment with ionomycin. The increase in EtBr uptake triggered by ionomycin was blocked by either CBX (50 μm) or Gap26 (100 μm; p < 0.05, Student's t test, n = 3 independent cultures).

Figure 5.

Patch-clamp experiments demonstrating Ca²⁺ activation of Cx43 hemichannels in cortical astrocytes. A, Example traces of unitary current recordings at 50 (left) and 200 nm (middle) [Ca²⁺]_i with voltage steps from −80 mV to the potentials indicated at left (15 s voltage steps). Current activity is apparent at +50 mV but also in the −70 to −30 mV range (activities shown at larger scale for −50 and −70 mV). Gap26 (200 μm) clearly inhibited current activities (right). B, I–V plot demonstrating macroscopic current traces obtained from voltage-ramp experiments (−70 to +70 mV, 10 s). C, I–V plot of unitary current amplitudes from voltage-step experiments as shown in A, demonstrating a single-channel slope conductance of 230 pS (n = 5). D, All point histograms constructed from unitary currents at negative (−70 to −30 mV), positive (+30 to +50 mV), and combined negative/positive voltages. Single channel conductances are indicated above each peak of the distributions and were determined from the fitted Gaussian curves shown in yellow. Gap26 (red) suppressed all unitary opening events and shifted the activity distribution to the closed state (data from five different experiments). E, F, Average Q_m data for voltage steps to +50 mV (n = 5). Asterisks indicate significant differences compared with the 50 nm [Ca²⁺]_i condition; hashtags indicate significant differences compared with the 200 nm [Ca²⁺]_i condition (one symbol p < 0.05; two symbols p < 0.01). All recordings were obtained from isolated replated primary culture of astrocytes (see Materials and Methods).

Figure 6.

LTP of NMDA synaptic currents in pyramidal cells depends on Cx43 hemichannel activation and [Ca²⁺]_i in astrocytes. LTP of NMDA synaptic currents in L5 was induced by HFS applied in L2/3 (see Materials and Methods and Fig. 1). A1–A3, Typical average (n = 5 consecutive traces) of NMDA EPSCs recorded at +40 mV in presence of picrotoxin (100 μm) and NBQX (10 μm). Scale bars, 100 pA and 100 ms. A1, Representative NMDA EPSCs recorded before (Control) and 30 min after HFS protocol. A2, NMDA EPSCs recorded 30 min after the HFS protocol and in presence of Gap26 (200 μm). A3, NMDA EPSCs recorded 30 min after the HFS protocol and after establishing a Ca²⁺ clamp in the astroglial network either without or in the presence of Gap26 (200 μm) in Ca²⁺-clamp condition. B, Diagram showing the change over time in the amplitude of NMDA EPSCs measured before and after HFS protocol in control (black), in the presence of a Ca²⁺ clamp of the astroglial network (red), and in the presence of Gap26 (blue). C, Histogram showing the relative amplitude of NMDA EPSCs measured 30 min after the HFS protocol (p < 0.05, t test, n = 7). The LTP was significantly reduced by either Ca²⁺ clamp of the astroglial network or in the presence of Gap26 (p > 0.05, t test, n = 8). Note that combining Ca²⁺ clamp and the application of Gap26 has no additive effect (p > 0.05, t test, n = 7).

Figure 7.

LTP of AMPA synaptic currents in pyramidal cells depends on Cx43 hemichannel activation in astrocytes. LTP of AMPA synaptic currents in L5 was induced by HFS applied in L2/3 (see Materials and Methods). A1, A2, Typical average (n = 5 consecutive traces) of AMPA EPSCs recorded at −70 mV in presence of picrotoxin (100 μm). Scale bars, 100 pA and 100 ms. A1, Representative AMPA EPSCs recorded before (Control) and 30 min after HFS protocol. A2, AMPA EPSCs recorded 30 min after the HFS protocol and in presence of Gap26 (200 μm). B, Diagram showing the change over time in the amplitude of AMPA EPSCs measured before and after HFS protocol in control (black), in the presence of Gap26 (blue), and in the presence of the scrambled peptide (green). C, Histogram showing the relative amplitude of AMPA EPSCs 30 min after HFS protocol (p < 0.05, t test, n = 7). The LTP was significantly reduced by in presence of Gap26 (p > 0.05, t test, n = 8), whereas the scrambled peptide had no significant effect (p < 0.05, t test, n = 7).

Figure 8.

D-serine release by astrocytes requires Cx43 hemichannel activation to induce LTP of NMDA EPSCs in PFC pyramidal neurons. A1–A3, Typical average (n = 5 consecutive traces) NMDA EPSCs recorded at +40 mV in the presence of picrotoxin (100 μm) and NBQX (10 μm). Scale bars, 100 pA and 100 ms. A1, Representative NMDA EPSCs recorded before (Control) and 30 min after the HFS protocol. A2, NMDA EPSCs recorded 30 min after the HFS protocol and after infusion of a serine-racemase inhibitor (HOAsp, 400 μm) in the astroglial network. A3, NMDA EPSCs recorded 30 min after the HFS protocol and after infusion of HOAsp in the astroglial network plus D-serine (100 μm) in the extracellular solution. C, Summary diagram showing a significant increase of NMDA EPSC amplitude 30 min after the HFS protocol (p < 0.01, t test, n = 7) that is prevented after infusion of HOAsp in the astroglial network (p > 0.05, t test, n = 7) and partially rescued by adding D-serine (100 μm) in the bath solution (vs HFS p < 0.05, t test, n = 7).