Hippocampal Short- and Long-Term Plasticity Are Not Modulated by Astrocyte Ca2+ Signaling

See allHide authors and affiliations

Science  05 Mar 2010:
Vol. 327, Issue 5970, pp. 1250-1254
DOI: 10.1126/science.1184821


The concept that astrocytes release neuroactive molecules (gliotransmitters) to affect synaptic transmission has been a paradigm shift in neuroscience research over the past decade. This concept suggests that astrocytes, together with pre- and postsynaptic neuronal elements, make up a functional synapse. Astrocyte release of gliotransmitters (for example, glutamate and adenosine triphosphate) is generally accepted to be a Ca2+-dependent process. We used two mouse lines to either selectively increase or obliterate astrocytic Gq G protein–coupled receptor Ca2+ signaling to further test the hypothesis that astrocytes release gliotransmitters in a Ca2+-dependent manner to affect synaptic transmission. Neither increasing nor obliterating astrocytic Ca2+ fluxes affects spontaneous and evoked excitatory synaptic transmission or synaptic plasticity. Our findings suggest that, at least in the hippocampus, the mechanisms of gliotransmission need to be reconsidered.

Calcium transients in astrocytes are physiologically driven by metabotropic Gq G protein–coupled receptors (Gq GPCRs), which can be activated after neurotransmitter release from presynaptic terminals (1, 2). At Schaffer collateral-CA1 (SC-CA1) synapses in acute hippocampal slices, astrocytes can modulate neuronal activity by elevations in Ca2+ that are evoked by the following: (i) uncaging IP3 or Ca2+ in individual astrocytes, (ii) repetitive depolarization of the astrocyte membrane, (iii) mechanical stimulation of an astrocyte, or (iv) bath application of endogenous Gq GPCR agonists. With these pharmacological approaches, astrocyte Ca2+ elevations have been reported to trigger gliotransmitter release from astrocytes, resulting in the modulation of synaptic transmission and plasticity through the activation of presynaptic [for example, group I metabotropic glutamate receptors (mGluRs) or adenosine A(1) receptors (A1Rs)] or postsynaptic receptors [N-methyl- d-aspartate receptors (NMDARs)] (311). To circumvent a number of caveats associated with the pharmacological approaches described above (1215), we have recently developed and characterized two genetically modified mice [the MrgA1+ and IP3R2 knockout (KO) mice] that enable either selective activation or inactivation of Gq GPCR Ca2+ signaling in astrocytes (13, 16, 17). Within the hippocampus, the stimulation of transgenic MrgA1 Gq GPCRs leads to astrocyte-specific Ca2+ responses that mimic the “Ca2+ fingerprint” response that is elicited by endogenous Gq GPCRs (13). In hippocampal slices derived from IP3R2 KO mice (17), Gq GPCR Ca2+ signaling is obliterated selectively in 100% of astrocytes without affecting neuronal Ca2+ responses (16).

We first tested the possibility that astrocytic Gq GPCR Ca2+ is involved in the modulation of spontaneous excitatory postsynaptic currents (sEPSCs). In these and the following experiments, a high percentage of astrocytes (~90 to 100%) were stimulated so that each CA1 neuron has the vast majority of its synapses embedded in astrocyte processes that elevate Ca2+ upon Gq GPCR agonist application. Control experiments showed that MrgA1R expression by itself in astrocytes does not affect basal neuronal activity in a nonspecific manner [supporting online material (SOM) text S1]. MrgA1R agonist Phe-Met-Arg-Phe-NH2 amide (FMRF, 15 μM) was applied to trigger Ca2+ elevations in ~90% of mature MrgA1+ passive astrocytes (13) in cell bodies as well as fine processes (Fig. 1, A and B, boxes/traces 1 to 5, SOM text S2, and movie S1). No significant effect of astrocyte Ca2+ elevations on sEPSC frequency and amplitude in CA1 neurons from MrgA1+ mice was found (Fig. 1C and SOM text S3, n = 7, P > 0.05). To test the possibility that this lack of effect might be caused by the stimulation of a transgenic Gq GPCR, we also stimulated endogenous astrocytic endothelin Gq GPCRs (ETRs), which were selected as optimal candidates because they evoke gliotransmitter release in vitro (18), they are thought to be very weakly expressed by neurons and heavily expressed by brain astrocytes at postnatal day 1 to 30 (19), and no direct effects on neuronal activity have been reported when stimulating ETRs (13). Astrocytic ETR-mediated Ca2+ increases in ~100% of astrocytes from wild-type (WT) hippocampal slices [endothelin 1 (ET1) and ET3, 10 nM each; SOM text S4, and fig. S1) had no effect on the frequency or amplitude of sEPSCs (Fig. 1, D to F, and SOM text S5, n = 5, P > 0.05).

Fig. 1

Stimulation of astrocytic Gq GPCRs does not affect sEPSCs in CA1 pyramidal neurons. (A) Typical MrgA1+ astrocyte in stratum radiatum (s.r.), filled with Oregon-Green-BAPTA-1 (OGB1), before (left) and after (right) stimulation of MrgA1Rs with FMRF (15 μM). Astrocyte Ca2+ responses are included as representative responses of the ~90% of responsive astrocytes within the slice and were used to monitor the beginning of Ca2+ elevations relative to neuronal activity. (B and C) sEPSCs frequency and amplitude remain unchanged [(B), upper] during astrocyte Ca2+ increases [(B), traces 1 to 5]. To the right are the averaged sEPSCs from the trace in (B). (D) Astrocytes (1 to 9) bulk-loaded with Ca2+ green 1-AM in s.r. from WT slices. (E and F) sEPSCs frequency and amplitude were unchanged during astrocyte ETR-mediated Ca2+ increases (ET1 and ET3, 10 nM each). Due to long-tailing ETR Ca2+ increases, sEPSCs were analyzed only during the first 120 s of astrocyte Ca2+ increases. Scale bars in (A) and (D) indicate 10 and 20 μm, respectively. Error bars indicate SEM.

Previous studies using conventional pharmacological approaches have suggested that postsynaptic NMDARs might be preferential targets for glutamate release from astrocytes (37, 9, 10), prompting us to examine the possibility that astrocytic Gq GPCR Ca2+ elevations modulate the NMDAR-mediated component of evoked whole-cell EPSCs (eEPSCs). FMRF does not produce a nonspecific effect on NMDA eEPSCs (Fig. 2, A and A1, and SOM text S6). FMRF or ETs were applied to MrgA1+ or WT slices, respectively, and the amplitude of NMDA eEPSCs was unaffected during agonist-mediated Ca2+ increases in astrocytes (Fig. 2, B and B1, and SOM text S7, MrgA1+, n = 11, P > 0.05; Fig. 2, C and C1, and SOM text S7, WT, n = 7, P > 0.05).

Fig. 2

Astrocytic Gq GPCR Ca2+ signaling does not modulate the NMDA eEPSC peak amplitude or PPF in CA1 pyramidal neurons. (A and A1) WT astrocytes were not activated by FMRF [(A) lower Ca2+ traces from three astrocytes surrounding recorded neuron]. FMRF did not nonspecifically affect NMDA eEPSC peak amplitude in WT mice. Representative example (A) and pooled data (A1). (B, B1, C, and C1) Activation of MrgA1+ (B and B1) or WT astrocytes (C and C1) by FMRF or ETs, respectively, did not affect NMDA eEPSC peak amplitude. (D) MrgA1R expression in astrocytes of MrgA1+ mice did not have a nonspecific effect on PPF ratio compared with WT littermate mice. (E and F) PPF ratios in MrgA1+ slices before (solid squares) and during (open squares) FMRF-evoked astrocyte Ca2+ elevations (E) or in WT slices before (solid circles) and during (open circles) ET-evoked astrocyte Ca2+ increases (F) were not significantly different. (G) PPF is not altered in IP3R2 KO mice compared with WT littermate control mice.

Uncaging IP3 or Ca2+ in astrocytes produces a transient enhancement of the probability of neurotransmitter release at a fraction of the SC terminals (10, 20). We directly tested whether activating or inactivating astrocytic Gq GPCR Ca2+ signaling affects presynaptic release probability and short-term plasticity by measuring the paired-pulse facilitation (PPF) index of evoked field potentials (fEPSPs). Astrocytic MrgA1R expression by itself did not have a nonspecific effect on PPF (Fig. 2D and SOM text S8). No overall PPF profile changes were observed in association with astrocyte MrgA1R- or ETR-mediated Ca2+ elevations in MrgA1+ or WT slices, respectively (Fig. 2E, MrgA1+, n = 7, P > 0.05; Fig. 2F, WT, n = 6, P > 0.05). We reasoned that if astrocytic Ca2+ elevations regulate gliotransmitter release, then the removal of astrocytic Gq GPCR Ca2+ signaling should affect tonic and activity-induced gliotransmitter release and, consequently, PPF. Therefore, PPF was measured in IP3R2 KO mice versus WT littermate controls. Again, no changes in PPF profiles were observed between the two groups (Fig. 2G, IP3R2 KO, n = 9; WT, n = 9; P = 0.73).

A recent study has shown that temporal coincidence of astrocyte Ca2+ elevations (evoked by Ca2+ uncaging) and transient depolarization of CA1 neurons can induce a presynaptic form of long-term potentiation (LTP) in SC-CA1 synapses (10). Therefore, stimulating astrocytic Gq GPCR Ca2+ signaling simultaneously with depolarization of large ensembles of CA1 neurons should either directly induce LTP or at least modulate the baseline slope of fEPSPs through gliotransmitter activation of presynaptic group I mGluRs, or, alternatively, A1Rs (10, 21). Our data do not support these predictions (fig. S2, B, B1, C, C1, and SOM text S9).

We also directly tested whether astrocytic Gq GPCRs regulate LTP magnitude as well as post-tetanic potentiation (PTP). Similar to PPF, PTP is a form of short-term plasticity (22). First, a battery of control experiments clearly demonstrated that neither the selective astrocytic expression of MrgA1Rs nor the application of FMRF to WT slices affected input-output (I/O) curves, PTP, or LTP (Fig. 3, A and D, and SOM text S10). A prerequisite for the involvement of astrocytes in LTP is that their activation precedes the induction of LTP. After establishing a 15-min baseline recording of fEPSPs, Ca2+ elevations in astrocytes were induced by bath application of either FMRF to MrgA1+ slices or ETs to WT slices. LTP was induced ~2.5 min after agonist application, when the peaks of Ca2+ responses in astrocytes reached their maximum (Fig. 3, B and C). LTP magnitudes obtained from MrgA1+ slices stimulated with FMRF (137.97 ± 5.58%, n = 14 slices) or WT slices stimulated with ETs (133.13 ± 7.62%, n = 9), respectively, were no different from LTP magnitudes obtained in matching controls (Fig. 3, E and F, and SOM text S11, P > 0.05). Furthermore, PTP was also not affected (Fig. 3, E and F, and SOM text S11, P > 0.05).

Fig. 3

Astrocytic Gq GPCR Ca2+ signaling does not modulate the induction or magnitude of LTP in CA1 hippocampus. (A) There was no difference between I/O curves in MrgA1R+ versus WT littermate control slices. (B and C) LTP was induced when the peaks of FMRF- or ET-evoked astrocyte Ca2+ increases (lower and expanded traces) in MrgA1+ (B) or WT (C) slices, respectively, reached their maximum. (D) Astrocytic MrgA1R expression did not nonspecifically affect LTP. (E and F) Neither FMRF- nor ET-evoked astrocyte Ca2+ increases in MrgA1+ (E) or WT (F) slices, respectively, affected LTP compared with two sets of control slices [(E), MrgA1+ slices without FMRF application, WT littermate slices with FMRF application; (F), WT slices without ET application]. (G and H) I/O curves (G) and LTP (H) in IP3R2 KO mice are not affected compared with WT littermate control mice. Arrows indicate LTP induction (2 × 100 Hz).

Next, we examined whether the obliteration of astrocytic Gq GPCR Ca2+ signaling would affect basal synaptic transmission as well as PTP and LTP. No significant difference was found in I/O curves performed in IP3R2 KO versus WT littermate control mice (Fig. 3G, IP3R2 KO, n = 21; WT, n = 18; P = 0.94). These results indicate that both the pre- and postsynaptic responses, and thus the basal SC-evoked synaptic transmission, are intact in IP3R2 KO mice, even though astrocytes are completely incapable of producing Gq GPCR Ca2+ elevations. No significant alteration in PTP and LTP was detected between IP3R2 KO and control mice, demonstrating that astrocytic Gq GPCR–mediated Ca2+ signaling does not account for a tonic form nor an activity-induced form of short- and long-term synaptic plasticity (Fig. 3H and SOM text S12, IP3R2 KO, n = 10; WT, n = 8, P > 0.05). To validate these IP3R2 KO data, we showed that the LTP stimulation protocol used is sufficient to induce Ca2+ increases in astrocytes from WT slices (fig. S3). Finally, we also found that stimulating or removing astrocytic Gq GPCR Ca2+ signaling in MrgA1+ or IP3R2 KO mice, respectively, does not significantly alter LTP and PTP induced by theta-burst stimulation (SOM text S13).

Previous studies have demonstrated that activation of synaptic group I mGluRs, A1Rs, or NMDARs depotentiates LTP at SC-CA1 synapses (2327). These three receptors have all been reported to be the targets of astrocytic Ca2+-dependent sources of glutamate or adenosine triphosphate (ATP)/adenosine under certain conditions (12). To further test whether astrocytic Gq GPCR Ca2+ signaling is sufficient to induce gliotransmitter release to affect synaptic transmission through the activation of synaptic mGluRs, A1Rs, or NMDARs, we investigated the role of astrocyte Ca2+ signaling in the maintenance of LTP. Fifty minutes after LTP induction, astrocyte Ca2+ increases were elicited by applications of FMRF to MrgA1+ slices or of ETs to WT slices. This did not lead to a significant change in the slope of fEPSPs (Fig. 4, A, B, and D, and SOM text S14, n = 9 MrgA1+ slices, n = 16 WT slices, P > 0.05). As a positive control for agonist-induced depotentiation, we applied (RS)-3,5-dihydroxyphenylglycine (DHPG, 50 μM), which exerted a significant depotentiation of the slope in all slices tested (Fig. 4, A and D; MrgA1+, 74.82 ± 3.45%, n = 12, P < 0.0001; Fig. 4, B and D; WT, 79.54 ± 2.03%, n = 11, P < 0.0001). To address the possibility that the DHPG-mediated depotentiation could be due in part to astrocyte Ca2+, we performed the same experiments using IP3R2 KO mice. The magnitude of DHPG-induced depotentiation in IP3R2 KO slices was not only significant (P < 0.0001), it was also similar to the magnitude of depotentiation that was recorded in WT littermate slices (P = 0.43), indicating that depotentiation does not rely, even in part, on Ca2+-dependent gliotransmitter release from astrocytes (Fig. 4, C and D; IP3R2 KO, 66.95 ± 2.79%, n = 6; WT, 72.70 ± 5.72%, n = 8). These results demonstrate and confirm previous data that DHPG-induced modulation of neuronal activity (28, 29), such as depotentiation (26), is due to the direct action of DHPG on neuronal group I mGluRs (26), and not to astrocytic group I mGluR-mediated Ca2+ elevations and putative gliotransmitter release.

Fig. 4

Astrocytic Gq GPCR Ca2+ signaling does not modulate the maintenance of LTP (2 × 100 Hz) in CA1 hippocampus. (A and B) Neither FMRF- nor ET-evoked astrocyte Ca2+ increases (lower and expanded traces) in MrgA1+ slices (A) or WT slices (B), respectively, modulated the maintenance of LTP. As a control for modulation of LTP maintenance, DHPG (50 μM) was applied. (C) ETs and DHPG did not evoke any astrocyte Ca2+ increases in IP3R2 KO slices. Whereas ETs did not modulate the maintenance of LTP, DHPG induced a clear depotentiation independent of astrocyte Ca2+ elevations. (D) Summary histogram showing that DHPG, but not FMRF or ETs, affected the maintenance of LTP in MrgA1+, WT, or IP3R2 KO slices. Asterisks indicate statistical significance (P < 0.0001).

We provide here strong evidence that Gq GPCR Ca2+ signaling in astrocytes does not affect spontaneous and evoked excitatory action potential (AP)-mediated synaptic transmission or short- and long-term plasticity at the SC-CA1 synapse. We used two molecular tools (the MrgA1+ and IP3R2 KO mouse models), as well as the activation of endogenous astrocytic Gq GPCRs, to manipulate Ca2+ in astrocytes. A battery of eight electrophysiological protocols (sEPSCs, NMDA eEPSCs, evoked AMPA fEPSPs, I/O curves, PPF, PTP, and two forms of LTP) were studied, all of which point to a lack of modulation of excitatory AP-mediated synaptic transmission by astrocytic Gq GPCR Ca2+ signaling. The most logical conclusion from the present analysis is that astrocytic Gq GPCRs and Ca2+ signaling activity are not tied to the release of gliotransmitters affecting synaptic transmission or short and long-term plasticity. Therefore, our results suggest that gliotransmission reflects the pharmacological approaches that were used in previous studies (310, 12) and, at least within the hippocampus, does not occur when the endogenous regulators of astrocyte Ca2+, the Gq GPCRs, or the IP3R2 themselves are stimulated or inactivated in a cellular-selective manner. These findings suggest that the mechanisms of gliotransmitter release should be reconsidered. These results have profound implications for our understanding of synaptic transmission and should affect the interpretation of a broad range of findings. Thus, the purpose of neuron-to-astrocyte Gq GPCR Ca2+ signaling in neurophysiology remains an open question.

Supporting Online Material

Materials and Methods

SOM Text S1 to S14

Figs. S1 to S3


Movie S1

References and Notes

  1. We thank K. Casper for making MrgA1+ mice; J. Chen for providing IP3R2 KO mice; and B. Djukic, B. Philpot, A. Roberts, and J. de Marchena for valuable help and discussions. This work was supported by NIH grants NS033938 and NS020212.
View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article