Background
Astrocytes exhibit dynamic Ca
2+ mobilization via an inositol 1,4,5-trisphosphate (IP
3)-induced Ca
2+ release (IICR)-dependent mechanism [
1,
2]. The role of astrocytic Ca
2+ dynamics, however, has been debated over the last decade [
3]. Some studies report intact hippocampal short-and long-term plasticity in situ in IP
3 receptor (IP
3R)-type 2 knockout mice, in which hippocampal astrocytes completely lack IICR [
4,
5], whereas others report that these mice show no cholinergic-induced long-term potentiation (LTP) in vivo [
6,
7]. In situ evidence also indicates that hippocampal LTP depends on D-serine release from astrocytes under the control of astrocytic Ca
2+ signaling [
8]. A number of reports describe spontaneous or stimulation-evoked Ca
2+ activity in astrocytes in vivo [
9‐
12], but still very little is known about the in vivo significance of astrocytic Ca
2+ mobilization [
6,
7], especially in higher brain functions such as learning and memory. In the present study, we generated a new mouse model in which astrocytic IICR is attenuated, and examined the role of astrocytic Ca
2+ signaling at the level of both the tripartite synapse and behavior.
Discussion
The IP
3 sponge, a novel recombinant hyperaffinity IP
3 absorbent [
13], inhibits IICR in cultured mammalian cells, starfish oocytes, and
Drosophila[
13,
29,
30]. In the present study, we generated DTg mice in which the GST-IP
3 sponge was expressed in astrocytes using the Tet-OFF system [
31]. This transgenic mouse system enabled us to induce the expression of the lacZ marker protein in approximately 90% of S100B-positive astrocytes in the hippocampal CA1 region of the DTg mice. Ca
2+ imaging of hippocampal slices from transgenic mice expressing G-CaMP2 confirmed that agonist-evoked IICR was attenuated in the processes of GST-IP
3 sponge-positive astrocytes. Thus, we obtained a mouse model in which astrocytic IICR was attenuated. Analyses of the DTg mice revealed 1) a significant reduction in the astrocyte coverage of asymmetric synapses in the hippocampal CA1 region, 2) enhanced glutamate spillover at CA1-CA3 synapses during high-frequency activity, and 3) impaired spatial reference memory and remote contextual fear memory.
IICR-dependent regulation of astrocytic ensheathment of synapses provides new insight into the controversy surrounding whether astrocytic Ca2+ signaling affects neuronal functions. These findings raise intriguing possibilities regarding neuron-astrocyte interactions: when the synapse releases glutamate on metabotropic glutamate receptors in the surrounding astrocyte, the level of Ca2+ activity in astrocytic processes is maintained, providing high astrocytic coverage of the synapse. An inactive synapse would then not have astrocytic coverage, allowing for neurotransmitter diffusion in and out of the synaptic cleft. Thus, our data provide a possible mechanism for activity-dependent regulation of astrocytic coverage in the tripartite synapse.
Indeed, astrocytic processes are considered very plastic [
32‐
37], as is astrocytic coverage of synapses [
38,
39]. The dynamic motility of astrocytic processes is essential for astrocytes to modulate synaptic activity, but its regulation in vivo remains largely unknown. Glutamate causes filopodia formation in cultured hippocampal astrocytes [
37], which is mediated by metabotropic glutamate receptors [
40]. Metabotropic glutamate receptor-activation in astrocytes in vitro [
41‐
44] and in situ [
45] leads to an increase in intracellular Ca
2+ concentrations through the release from intracellular stores. Our results provide a link between IICR and plasticity of the astrocytic processes in vivo. The downstream cascade leading to changes in the morphology of astrocytic processes, however, remains to be clarified, and might involve regulation of the organization and assembly of the actin cytoskeleton in astrocytes [
40]. IICR in the growth cone acts on downstream effectors to regulate microtubule assembly and promote neurite extension [
19]. Microtubule networks and actin networks might contribute to the motility of peripheral astrocytic processes in vivo. Moreover, the Ca
2+ concentration regulates the migration of astrocytoma cells by forming or disassembling focal adhesions [
44,
45]. Local concentrations of Ca
2+ are further suggested to be involved in filopodia formation by regulating focal adhesions [
46]. These mechanisms might cooperatively contribute to the structural integrity of tripartite synapses.
Several reports suggest Ca
2+-dependent release of glutamate and D-serine from astrocytes [
6‐
8,
47‐
49]. Both are co-agonists of NMDA receptors. Therefore, reduced Ca
2+ activity in astrocytes might decrease the activation of NMDA receptors by those gliotransmitters during synaptic transmission. In contrast, we found that high-frequency stimulation in DTg mice increases the NMDA receptor current in CA1 pyramidal neurons, which correlates with enhanced glutamate spillover. Recent reports suggest that NMDA receptors located at synaptic and extrasynaptic sites have distinct functional roles [
24,
50]. Spillover can actually increase the proportion of extrasynaptic NMDA receptors activated during burst firing, affecting hippocampal-based learning. In addition, glutamate spillover leads to an increase in intersynaptic crosstalk. The increased intersynaptic crosstalk could impair the independent operation of hippocampal synapses and thus affect synaptic characteristics, such as the spatial precision of synaptic inputs [
51], which might also be important for learning and memory. We cannot exclude the possibility that glutamate spillover during burst firing increases release probability by activating presynaptic NMDA receptors, which leaves open the possibility that reduced astrocytic coverage has presynaptic effects as well as postsynaptic effects. In the present study, however, it is unlikely that presynaptic mGluRs were involved in the enhanced NMDA receptor currents. NMDA receptor currents were recorded in the presence of S-MCPG (200 μM), which blocks Type I and Type II mGluRs. Activation of Type III mGluRs by enhanced glutamate spillover decreases the release probability at CA3-CA1 synapses with Shaffer collaterals [
52]. If this were the case, the observed potentiation of the NMDA receptor current could be underestimated in GST-IP
3 sponge-expressing animals. These and other effects of enhanced spillover on the hippocampal network should be addressed in future studies.
Concerning changes in long-term plasticity in DTg mice, we found that there were no significant differences in long-term potentiation (LTP), long-term depression (LTD), and depotentiation induced by conventional protocols in situ (i.e., 2 tetanic stimulations [
5], 5 theta-burst stimulations (TBS) [
53], and 3 TBS for LTP [
53]; low-frequency stimulation (LFS) [
54] and paired pulse-LFS [
55] for LTD; and 2 tetanic stimulations followed by LFS for depotentiation [
54]; data not shown). Despite these findings, our results indicate that enhanced glutamate spillover in the hippocampus correlates with impaired learning and memory in DTg mice, very likely by perturbing the independent operation of hippocampal synapses [
51]. Indeed, if LTP can be effectively triggered, but in the ‘wrong’ population of synapses, the memory still can be potentially affected.
The normal recent context fear memory in DTg mice contrasts with the impaired spatial reference memory in these mice. This discrepancy can be reconciled by the fact that spatial reference memory evaluated in the water maze is not associated with contextual fear memory [
56‐
58] and each of these distinct types of memories have different underlying mechanisms [
59]. Although the impairment of remote contextual fear memory in DTg mice indicates miscommunication between the hippocampus and cortical areas [
26‐
28], our result does not exclude the possibility that the anterior cingulate cortex, in which GST-IP
3 sponge expression in the astrocytes was also observed (Figure
1C and data not shown), is solely responsible for the phenotype [
27,
60,
61].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MT, PYS, AS, and SI conceived the study. MT, PYS, HG, TY, AS, and SI wrote the paper. MT, PYS, HG, JN, and RA performed the experiments. TY analyzed the calcium imaging data. TF and KM contributed to creation and characterization of GST-IP3 sponge. All authors have read and approved the manuscript.