Research ArticleGlia

Receptor-Selective Diffusion Barrier Enhances Sensitivity of Astrocytic Processes to Metabotropic Glutamate Receptor Stimulation

See allHide authors and affiliations

Science Signaling  03 Apr 2012:
Vol. 5, Issue 218, pp. ra27
DOI: 10.1126/scisignal.2002498

Abstract

Metabotropic glutamate receptor (mGluR)–dependent calcium ion (Ca2+) signaling in astrocytic processes regulates synaptic transmission and local blood flow essential for brain function. However, because of difficulties in imaging astrocytic processes, the subcellular spatial organization of mGluR-dependent Ca2+ signaling is not well characterized and its regulatory mechanism remains unclear. Using genetically encoded Ca2+ indicators, we showed that despite global stimulation by an mGluR agonist, astrocyte processes intrinsically exhibited a marked enrichment of Ca2+ responses. Immunocytochemistry indicated that these polarized Ca2+ responses could be attributed to increased density of surface mGluR5 on processes relative to the soma. Single-particle tracking of surface mGluR5 dynamics revealed a membrane barrier that blocked the movement of mGluR5 between the processes and the soma. Overexpression of mGluR or expression of its carboxyl terminus enabled diffusion of mGluR5 between the soma and the processes, disrupting the polarization of mGluR5 and of mGluR-dependent Ca2+ signaling. Together, our results demonstrate an mGluR5-selective diffusion barrier between processes and soma that compartmentalized mGluR Ca2+ signaling in astrocytes and may allow control of synaptic and vascular activity in specific subcellular domains.

Introduction

Astrocytes, named for the star-like shape conferred by their numerous radiating processes, are the most numerous form of glial cells in the central nervous system. Historically, they were thought to play solely supportive roles, such as maintaining extracellular ion homeostasis and providing nutrient support to neurons (1, 2). However, the discovery through Ca2+-imaging techniques of intracellular Ca2+ signals in astrocytes, together with the realization that astrocytes, like neurons, release chemical messengers, suggested that astrocytes might play a more active role in brain function (3). Various plasma membrane receptors trigger Ca2+ signaling in astrocytes (4), including the metabotropic glutamate receptor 5 [mGluR5, a Gq-linked heterotrimeric guanosine triphosphate–binding protein (G protein)–coupled receptor (GPCR)], activation of which stimulates Ca2+ release through inositol 1,4,5-trisphosphate receptors (IP3Rs) from intracellular stores in the endoplasmic reticulum (ER). mGluR5-dependent Ca2+ signaling mediates two of the major active roles of astrocytes: the regulation of synaptic transmission (5, 6) and that of local arteriolar dilation (7, 8). A single astrocyte with processes located close to neuronal synapses and cerebral arterioles could conceivably coordinate both of these functions; thus, the spatial compartmentation of Ca2+ signaling within a single astrocyte provides a potential mechanism for controlling the topological influence of astrocyte processes on the network of surrounding cells. In this scenario, a highly localized Ca2+ signal could affect partners in the neighborhood of a single process, whereas a Ca2+ signal that involves the entire astrocyte, including the soma, could have a more global influence (9). The importance of spatial regulation of astrocyte Ca2+ signaling is also implied by reports of altered spatial Ca2+ signaling—such as abnormal Ca2+ waves or enhanced somatic Ca2+ signals—in tissue from pathological brains (1014). However, relatively few studies have reported the detailed subcellular Ca2+ dynamics of astrocytes, and the spatial regulation of mGluR5-dependent Ca2+ signaling in astrocytes remains largely unexplored (1518).

Here, we expressed a genetically encoded Ca2+ indicator in neuron-astrocyte cocultures and in hippocampal slices and found that astrocyte processes are more sensitive than the soma to mGluR stimulation. Immunocytochemical analysis indicated that this polarized Ca2+ response derives from a greater surface mGluR5 density in the process relative to that in the soma. Quantum dot (QD)–based single-particle tracking (QD-SPT) of endogenous mGluR5s (19) revealed the existence of a membrane barrier that blocked the movement of mGluR5 between the soma and the processes. We demonstrated that this barrier represents the primary mechanism underlying compartmentalization of astrocyte mGluR distribution and thereby of polarized mGluR-dependent Ca2+ signaling.

Results

Astrocyte processes are more responsive than the soma to mGluR stimulation

Conventional Ca2+ reporter dyes are preferentially loaded in the soma; therefore, to investigate the spatial properties of mGluR-dependent Ca2+ signaling in single astrocytes, we used the genetically encoded Ca2+ indicator GCaMP2 (20), which diffused throughout the astrocyte, and reported Ca2+ dynamics in processes as clearly as those in the soma. Transfection of GCaMP2 into cocultures of astrocytes and neurons, in which the former developed processes reminiscent of those in in vivo astrocytes, enabled us to monitor detailed Ca2+ dynamics within entire astrocytes. Some astrocytes showed spontaneous increases in the intracellular Ca2+ concentration ([Ca2+]i), with individual processes of the astrocyte displaying independent patterns of Ca2+ transients (fig. S1 and movie S1) similar to those described in astrocytes in hippocampal slices (17). This indicates that astrocytes cocultured with neurons exhibit subcellular Ca2+ dynamics similar to those of astrocytes in brain slices and therefore provide a suitable system for assessing the properties of astrocytic Ca2+ signaling. When we bath-applied the group I mGluR agonist DHPG [(S)-3,5-dihydroxyphenylglycine, at 10 μM] onto GCaMP2-transfected astrocytes, an increase in [Ca2+]i was first observed in processes and then in the soma (Fig. 1, A to C). Of 59 DHPG-responsive cells, more than 20% of cells did not exhibit any increase in somatic [Ca2+]i, and no cell had Ca2+ transients in the soma alone. Of those cells that exhibited Ca2+ responses in both the processes and the soma, 98 ± 2.5% of the Ca2+ transients were initiated in processes followed by a [Ca2+]i increase in the soma, even though the entire astrocyte was stimulated simultaneously (Fig. 1D). Ca2+ transients were initiated independently in multiple processes in 89 ± 5.6% of the cells (Fig. 1E), indicating that individual processes can generate independent Ca2+ transients. These results indicate that astrocytic processes are more responsive than the soma to mGluR agonists.

Fig. 1

High responsiveness to mGluR stimulation exhibited by astrocyte processes. (A to C) An astrocyte transfected with GCaMP2 (A) and sequential images of [Ca2+]i changes (ΔF/F0) (B). Time represents the time after DHPG application. Arrowheads: processes; arrow: the soma. (C) Time-course plot of ΔF/F0 monitored in the regions of interest (ROIs) indicated in (A). DHPG was applied in the bath as indicated by the gray line. Arrows indicate Ca2+ transient peaks in the processes (filled) and the soma (open). (D) The average (±SEM) percentage of astrocytes that initiated a Ca2+ transient at the process (P), at the process and soma simultaneously (P + S), or at the soma (S). (E) The average (±SEM) percentage of astrocytes in which a Ca2+ transient was initiated from single or multiple processes. (F and G) An astrocyte transfected with GCaMP3 in hippocampal slice (F) and sequential images of [Ca2+]i changes (G). 0.0 s indicates the time of the initial Ca2+ transient induced by DHPG stimulation. Arrowheads: processes; arrow: the soma. (H) Time course of ΔF/F0 monitored in the ROIs indicated in (F). Gray line: the timing of DHPG application. Arrows indicate the first DHPG-evoked [Ca2+]i increase in the process (filled) and the soma (open). (I) The average (±SEM) percentage of astrocytes that initiated a Ca2+ transient at the process (P), at the process and the soma simultaneously (P + S), and at the soma (S). ***P < 0.005, Student’s t test. n = 10 cultures, 46 cells for (D) and (E); n = 9 animals, 58 cells for (I). Scale bars, 20 μm.

To confirm the high responsiveness to mGluR stimulation of astrocytic processes under more physiological conditions, we examined astrocytic Ca2+ responses to DHPG in hippocampal slices. We introduced the genetically encoded Ca2+ indicator GCaMP3 (21) into hippocampal slices with a gene gun, enabling us to observe Ca2+ dynamics in the entire astrocyte including its processes (Fig. 1, F and G). We identified GCaMP3-expressing cells as astrocytes according to their morphology and confirmed that GCaMP3-expressing cells classified as astrocytes were glial fibrillary acidic protein (GFAP)–immunoreactive (12 cells) (fig. S2). As found in cultured astrocytes, the initial Ca2+ transients observed in response to perfusion of 10 μM DHPG occurred in processes; these initial transients eventually led to [Ca2+]i signals in the soma in almost 90% of the astrocytes (Fig. 1, F to I). Thus, the relative responsiveness of astrocytic processes to mGluR agonists compared to that in the soma is present in intact astrocyte networks in situ, validating the physiological relevance of this property in astrocytes cocultured with neurons.

mGluR5 density is enhanced in astrocytic processes compared to that in the soma

To determine the mechanism underlying the greater responsiveness of processes, we first assessed the subcellular distribution of the Ca2+-releasing machinery. We found that the immunoreactivity of the ER marker calreticulin and that of the IP3R type 2 Ca2+ channel (22) were uniformly distributed throughout almost the entire astrocyte (Fig. 2, A and B), indicating that the Ca2+-releasing machinery is equally abundant in the soma and processes. We also confirmed that Ca2+ release from IP3R can be equally evoked in astrocyte somata (n = 56 events) and processes (n = 28 events) through localized photolysis of caged IP3 loaded into the cell (Fig. 2C). These results indicate that the Ca2+-release machinery is equally distributed in both the process and the soma. We then examined whether high responsiveness of the astrocytic process to mGluR agonist could be explained by the polarized distribution of cell surface mGluR5s. Immunocytochemical analysis of cell surface mGluR5 with an antibody recognizing the extracellular domain of mGluR5 on nonpermeabilized cells revealed regions of high mGluR5 in astrocyte processes and a weaker somatic signal (Fig. 2D). Indeed, the average fluorescence intensity measured in processes was about 1.5 times stronger than that of the soma (process: 1.05 ± 0.03, soma: 0.7 ± 0.05, average ± SEM, n = 12 cells; Fig. 2E), indicating that astrocytic processes are enriched in surface mGluR5 relative to the soma. Together, these findings indicate that distribution of surface mGluR5 rather than that of the Ca2+-releasing machinery underlies the increased responsiveness of processes to mGluR5 stimulation.

Fig. 2

Homogeneous distribution of Ca2+-releasing machinery and polarized distribution of mGluR5 in astrocytes. (A and B) Immunocytochemical visualization of the distribution of ER (A) and IP3Rs (B) in astrocytes with antibodies against calreticulin and IP3R2, respectively. The PM-CFP signal shows cell morphology. (C) Local Ca2+ transient evoked by flash photolysis of caged IP3. Top: GCaMP2 signal and the location of flash photolysis (red circle). Bottom: [Ca2+]i increase in the process (left) and soma (right) after flash photolysis stimulation was applied at 0.0 s. Pseudocolor represents the fluorescence intensity of GCaMP2 subtracted from the average fluorescence intensity of the 15 frames before uncaging (Fbase). Locations of photo uncaging are indicated by the arrowheads. A Ca2+ transient was evoked regardless of the site of stimulation. Ca2+ transient evoked in the soma propagated into other processes (arrow). n = 32 cells. (D) Immunocytochemistry of mGluR5s located on the plasma membrane of astrocytes. Cell morphology was visualized by tdTomato. More intense mGluR5 immunoreactivity was observed in the processes (arrowheads) than the soma (arrow). (E) Fluorescence intensities of mGluR5s in the processes and the soma normalized by the average fluorescence of entire astrocyte (average ± SEM, n = 12 cells). ***P < 0.005, Student’s t test. Scale bars, 20 μm.

Movement of mGluR5 between soma and processes is blocked by a membrane barrier

Various cell types show a polarized distribution of plasma membrane molecules and overcome the randomizing effects of the free diffusion predicted by the fluid mosaic model (23) through diffusion barriers (2429) or by restricting molecular diffusion within microdomains (30). To determine whether the regulation of mGluR5 lateral diffusion could contribute to the differing densities of mGluR5 in astrocytic processes and somata, we analyzed the diffusion of endogenous mGluR5 molecules with QD-SPT (19). We recorded the movement of surface mGluR5s targeted with QDs (mGluR5-QDs) over a 15.2-s sequence and reconstructed their trajectories (Fig. 3A), which demonstrated diffusive behavior of mGluR5-QDs on astrocytes, as reported for mGluR5s on hippocampal neurons (31, 32). The diffusion properties in the soma and the processes were analyzed by calculating diffusion parameters from the reconstructed trajectories (33, 34). The diffusion coefficient of mGluR5 in the soma was significantly smaller than that in the process (Fig. 3B and Table 1). Moreover, the percentage of mGluR5-QD exhibiting “confined diffusion,” that is, lateral diffusion limited to a small surface area (35), was 1.9 times larger in the somata than in the processes (Fig. 3C). These results suggest that mGluR5 diffusion properties differ in the morphologically distinct regions of the astrocyte, the soma and the processes: mGluR5 in the soma is less dynamic with a larger probability of confinement than in the processes.

Fig. 3

Regulation of mGluR5 diffusion by a soma-process diffusion barrier. (A) Representative trajectories of mGluR5-QD in processes (top) and the soma (bottom) recorded over 15.2 s. Scale bar, 1 μm. (B) Median diffusion coefficients of mGluR5-QD (arrowhead) plus (hatched) and minus (filled) interquartile range (IQR) at the process and the soma. Leftward shift of the diffusion coefficient in the soma indicates slower diffusion of mGluR5-QD. ***P < 0.005, Mann-Whitney U test. The number of QDs analyzed is shown in Table 1. (C) Percentage of mGluR5-QD presenting confined lateral diffusion (process: 31.4 ± 2.6%, soma: 60.9 ± 4.4%, average ± SEM, n = 27 cells). ***P < 0.005, paired t test. (D) Surface area explored by mGluR5-QDs visualized by maximum-intensity projection of time-lapse images of mGluR5-QD recorded over 10 min (green) on three representative astrocytes expressing PM-CFP (blue). (E) Trajectories of mGluR5-QDs on the astrocyte indicated in (D, left). (F) Magnified images of red trajectories indicated in (E). (a) Example of a trajectory that started in midprocess. (b to d) Examples of trajectories that started near the soma (*). Yellow squares represent the starting point of the trajectory. Scale bars, 10 μm [(D) and (E)] and 5 μm (F).

Table 1

Summary of median diffusion coefficients (×10−2 μm2/s). ***P < 0.005 in the process compared to the soma; NS: P > 0.05, Mann-Whitney U test. Numbers in parentheses: number of particles analyzed.

View this table:

Longer recordings of mGluR5-QD behavior revealed additional distinctions. Maximum projection of image sequences revealed that a large percentage of the astrocyte process area was traversed by mGluR5-QDs within a 10-min period (Fig. 3D). However, no mGluR5-QDs found in processes were observed to enter the soma, and similarly, somatic mGluR5-QDs did not travel into the processes (n = 44 cells from 11 independent batches). Indeed, the reconstructed trajectories of mGluR5-QDs located on processes around the soma emphasized the boundary between the soma and the processes (Fig. 3E). Freely diffusing mGluR5-QDs in midprocess exhibited uniform diffusion (Fig. 3Fa). However, freely diffusing mGluR5-QDs that were located near the soma at the beginning of the recording, and were therefore expected to diffuse into the somatic compartment, failed to do so (Fig. 3F, b to d). We estimated the likelihood of a soma-process transition by determining the number of events in which an mGluR5-QD crossed a line tangent to the soma during a 10-min recording (Eq. 4 in fig. S3A). The average number of such soma-process transitions was 0.18 ± 0.06/10 min/QD (n = 189 QDs); 1.4 ± 0.2/10 min/QD crossings (n = 227 QDs) were observed when a line was placed in midprocess (10 μm away from the p–s line, Eq. 5 in fig. S3A), suggesting that transition of a mGluR5-QD across the soma-process boundary is much less likely (13 ± 4.8%) than a line-crossing event in the process (fig. S3B). Together, these observations suggest that the diffusion of mGluR5-QDs is limited to either the processes or the soma and supports the existence of a diffusion barrier that inhibits the surface exchange of mGluR5 molecules between these two compartments.

Next, we asked whether the diffusion barrier blocking the transition of mGluR5 between the process and the soma also inhibited the movement of other molecules. We investigated the dynamics of the phospholipid DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine) and those of P2X7R (purinergic receptor P2X, ligand-gated ion channel 7), a transmembrane protein that is found in astrocytes (36). Both DOPE-QDs and P2X7R-QDs were observed to travel between the process and the soma (Fig. 4, A to F), indicating that the soma-process barrier is selective for mGluR5. The diffusion properties of mGluR5 within the two compartments were distinct: mGluR5 movement was less dynamic and more likely to be confined in the soma than was mGluR5 movement in processes (Fig. 3, A to C). In contrast, the diffusion coefficients of DOPE and P2X7R did not differ between the processes and the soma (Fig. 4, G and H, and Table 1), indicating that existence of compartment-specific diffusion properties was also peculiar to mGluR5.

Fig. 4

Diffusion of DOPE and P2X7R independent of the soma-process boundary. (A) Example of surface exploration of DOPE-QDs (green) visualized by maximum-intensity projection of DOPE-QD recorded over 3.3 min in cells expressing GFP (blue). (B) Trajectory of a DOPE-QD migrating between the soma and the process. (C) Magnified image of the area indicated by the white box in (B). (D) Example of surface exploration of P2X7R-QDs (green) visualized by maximum-intensity projection of P2X7R-QD recorded over 10 min in cells expressing GFP (blue). (E) Trajectories of P2X7R-QDs migrating between the soma and the process. (F) Magnified images of the area indicated by white boxes in (E). Yellow squares in (B), (C), (E), and (F): the starting points of the trajectory; * in (C) and (F): soma. Scale bars, 10 μm [(A), (B), (D), and (E)] and 5 μm [(C) and (F)]. (G and H) Diffusion coefficients (median ± IQR) of DOPE-QD (G) and P2X7R-QD (H) at the process and the soma. NS, not significant, Mann-Whitney U test. The numbers of QDs analyzed in (G) and (H) are given in Table 1.

In summary, mGluR5 exhibited distinct diffusion properties in processes versus the soma, and the exchange of surface mGluR5 between soma and process is limited by an mGluR5-selective diffusion barrier at the soma-process border.

Dysfunction of the mGluR5 barrier results in altered mGluR5 distribution and disrupted Ca2+ signaling

To investigate the molecular mechanism underlying the mGluR5-selective diffusion barrier, we searched for conditions in which the barrier could be overcome. FRAP (fluorescence recovery after photobleaching) experiments showed that overexpressed mGluR5-pHluorin [mGluR5 tagged with the pH-sensitive green fluorescent protein (GFP)–based sensor pHluorin (37)], which reported the dynamics of mGluR5 on the cell surface, underwent lateral diffusion not only within astrocyte processes but also between the processes and the soma (fig. S4). This indicated that, when overexpressed, mGluR5 can overcome the mGluR5-selective diffusion barrier between the soma and the processes. This notion was confirmed by QD-SPT experiments on astrocytes overexpressing mGluR5-pHluorin, where QDs labeled both the endogenous mGluR5 and the overexpressed mGluR5-pHluorin. Under these conditions, mGluR5-QDs found on the processes were observed to migrate into the soma within 10 min (Fig. 5, A to C). The average number of crossings at the process-soma line was 0.47 ± 0.09/10 min/QD (n = 188 QDs), and that at the midprocess line was 0.43 ± 0.08/10 min/QD (n = 208 QDs) in mGluR5-pHluorin–overexpressing cells, indicating that the line crossing of mGluR5-QD at process-soma is as frequent as that in processes (fig. S3C). Moreover, the differences in mGluR5-QD diffusion coefficient between the soma and the processes (Fig. 3B) and in the percentage of GluR5-QDs displaying confined diffusion (Fig. 3C) disappeared with mGluR5-pHluorin overexpression (Fig. 5, D and E, and Table 1). Thus, overexpression of mGluR5 in astrocytes resulted in altered mGluR5 diffusion: mGluR5 overcame the soma-process diffusion barrier, and compartment-specific differences in mGluR5 diffusion properties were lost. We hypothesized that the mGluR5-selective diffusion barrier depended on mGluR5-interacting proteins and that the abundance of these proteins was not sufficient to maintain the function of the mGluR5-selective diffusion barrier in mGluR5-overexpressing cells. mGluR5 interacts with various cytosolic proteins by means of its intracellular C terminus (38). Therefore, to investigate this possibility, we tried to inhibit the interaction of mGluR5 and mGluR5-interacting proteins by expressing an mGluR5 fragment corresponding to this region (C terminus). As seen in mGluR5-overexpressing astrocytes, we observed free transition of endogenous mGluR5 between the processes and the soma (Fig. 5, F to H) in astrocytes expressing the C-terminal fragment. Quantification revealed that the likelihood of mGluR5-QD crossing the process-soma line increased to that of line crossing in processes (fig. S3D). The differences between mGluR5-QD diffusion coefficient and confinement in the soma and the processes also disappeared in C-terminal–expressing cells (Fig. 5, I and J, and Table 1). These results indicate that C-terminal–interacting proteins contribute to the molecular mechanism(s) responsible for compartment-specific diffusion of mGluR5 and for preventing mGluR5 from crossing the mGluR5-selective soma-process diffusion barrier. In addition, although movement of mGluR5-QDs between astrocyte somata and processes was not enabled by the expression of individual fragments of the mGluR5 C terminus, the expression of all three fragments that together composed the entire C terminus enabled the free transition of mGluR5-QDs between somata and the processes (fig. S5). This suggests that multiple intracellular mGluR5 interacting proteins are required to make the barrier effective.

Fig. 5

Dysfunction of the barrier caused by overexpression of mGluR5 or expression of mGluR5 C terminus. (A) Surface exploration of mGluR5-QDs (green) over 10 min in an astrocyte overexpressing mGluR5 (blue). (B) Trajectories of two mGluR5-QDs that crossed between the soma and a process. (C) Magnified image of the area delineated by the white box in (B). (D and E) Diffusion coefficient of mGluR5-QDs (median ± IQR) (D) and percentage of mGluR5 showing confined diffusion (E) in astrocytes with only endogenous mGluR5 (Endo) or those overexpressing mGluR5 (Endo + OE). (F) Area explored in 10 min by endogenous mGluR5s tagged with QDs (green) in an astrocyte expressing the mGluR5 C-terminal fragment (blue). (G) Trajectories of mGluR5-QDs that cross the soma-process boundary. (H) Magnification of the areas indicated by white boxes in (G). Yellow squares in (B), (C), (G), and (H): starting points; * in (C) and (H): soma. Scale bars, 10 μm [(A), (B), (F), and (G)] and 5 μm [(C) and (H)]. (I and J) Diffusion coefficient of mGluR5-QDs (median ± IQR) (I) and proportion of confined mGluR5-QDs (J) in control and mGluR5 C-terminal–expressing (+Cterm) astrocytes. Numbers of QDs analyzed in (D) and (I) are given in Table 1. Numbers in parentheses in (E) and (J): numbers of cells. ***P < 0.005; NS, not significant, U test for (D) and (I); paired t test for (E) and (J).

Finally, we examined whether dysfunction of the mGluR5-selective diffusion barrier in astrocytes overexpressing mGluR5-mRFP (monomeric red fluorescent protein) or expressing the C terminus affected the polarized distribution of mGluR5 and the spatial pattern of mGluR-dependent Ca2+ signaling. In marked contrast with the polarized distribution of mGluR5 observed in control astrocytes (Fig. 2, D and E), the density of mGluR5 in the somata and processes of mGluR5-mRFP–overexpressing or C-terminal–expressing astrocytes was equivalent (Fig. 6, A to D), indicating that the mGluR5-selective diffusion barrier maintains the polarized distribution of mGluR5. Finally, Ca2+ transients in response to application of an mGluR5 agonist were monitored in cells overexpressing mGluR5 or expressing the C-terminal fragment. Astrocytes overexpressing mGluR5-mRFP or expressing its C terminus gave rise to simultaneous increases in [Ca2+]i in their somata and processes in response to DHPG, an occurrence that was seldom observed in control astrocytes (Fig. 6, E and F). Together, these results indicate that restriction of endogenous mGluR5 diffusion is responsible for mGluR5 polarity and polarized mGluR-dependent Ca2+ signaling in astrocytes.

Fig. 6

Loss of polarity of mGluR5 distribution and mGluR-dependent Ca2+ signaling elicited by barrier dysfunction. (A) Distribution of surface mGluR5 in an astrocyte overexpressing mGluR5. (B) Mean (±SEM) fluorescence intensity for mGluR5 immunoreactivity in the processes or the soma of mGluR5-overexpressing astrocytes, normalized by the average intensity of the entire astrocyte. NS, not significant, paired t test, n = 21 cells. (C) Distribution of endogenous mGluR5 in an astrocyte expressing the mGluR5 C-terminal fragment. (D) Mean (±SEM) fluorescence intensity for mGluR5 immunoreactivity in the processes or soma of mGluR5 C-terminal–expressing astrocytes, normalized by the average intensity of the entire astrocyte. NS, not significant, paired t test, n = 22 cells. Cell morphology in (A) and (C) was reported by coexpressed mRFP and GFP, respectively. Scale bars, 20 μm. Arrows, soma. (E and F) Changes in the percentage of astrocytes that initiated the Ca2+ transient at the process (P), at the process and soma simultaneously (P + S), or at the soma (S) in astrocytes overexpressing mGluR5 [(E), +mGluR5)] or expressing its C-terminal fragment [(F), +Cterm]. Numbers in parentheses represent the number of cells.

Discussion

Here, we identify a mGluR5-selective barrier that blocks the diffusion of mGluR5 between astrocyte processes and their somata. This barrier maintained a higher density of mGluR5, and thereby an enhanced Ca2+ response to mGluR agonist stimulation, in the processes compared to the soma.

In studies using fluorescent Ca2+ indicator dyes, which mainly report somatic Ca2+ signaling (18), the existence of Ca2+ transients in processes preceding increases in somatic [Ca2+]i could well be missed, and cells responding only within their processes may be completely neglected. Thus, to monitor the details of the Ca2+ response in astrocytes, it is crucial to be able to detect increased [Ca2+]i in whole processes with high sensitivity. Ca2+ imaging of astrocytes in hippocampal slices with GCaMP3 showed that GCaMP3, like GCaMP2, diffused evenly throughout the entire astrocyte and reported Ca2+ signaling in astrocytic processes in situ with improved brightness and dynamic range compared to GCaMP2. Further improvements to genetically encoded Ca2+ indicators will enable the elaborate monitoring of the spatiotemporal dynamics of subcellular Ca2+ signaling in astrocytes in vivo.

Using genetically encoded Ca2+ indicators, we have identified factors involved in the spatial regulation of subcellular Ca2+ signaling in astrocytes. Ca2+ transients induced by synaptic activation are more likely to initiate at the processes and then propagate into the soma both in situ and in vitro (39, 40). This phenomenon can be explained by the fact that processes abut the synapse, where they are likely to receive excitatory inputs (9). Our present results demonstrating that astrocytes initiate Ca2+ transient at the process even with global stimulation indicate that astrocytes have intrinsically polarized Ca2+ signaling. Studies using neuroblastoma or Bergmann glial cells have identified cell geometry, especially a larger surface-to-volume ratio of fine processes relative to the soma, as a critical factor in polarized Ca2+ signaling (16, 41). It is possible that cell geometry also plays a substantial role in determining the patterns of astrocyte Ca2+ signaling. However, our finding that altered distribution of endogenous mGluR5, induced by expression of the mGluR5 C terminus, altered mGluR-dependent Ca2+ signaling, indicating that the localization of mGluR5 in processes is also a key determinant of the spatial patterns of astrocyte Ca2+ signaling.

We found that the lateral diffusion of mGluR5 in astrocytes was regulated through two mechanisms: (i) the confinement of diffusion and (ii) the mGluR5-selective diffusion barrier at the process-soma boarder. A considerable proportion of endogenous astrocyte mGluR5s showed confined diffusion, with the confined mGluR5 fraction about 1.9 times larger in the soma than in the processes. Confinement of mGluR5 diffusion may contribute to mGluR5 retention in a particular compartment (soma or processes). However, the predominant mechanism for maintaining the polarization of mGluR5 density on the astrocyte surface is the barrier against mGluR5s, which blocks the transition of even freely diffusing mGluR5s from one compartment to the other. This barrier failed to block diffusion of molecules other than mGluR5 and could be overcome by mGluR5 overexpression or by disrupting interactions between mGluR5 and its cytosolic partners. The use of a membrane barrier to control the distribution of molecules has been reported in other cell types, including neurons (24, 27, 28) and sperm (25). The diffusion barriers identified in these cell types block the trans-barrier movement of multiple membrane molecules, including phospholipids, compartmentalizing different signaling functions to different regions of the cell. However, the mGluR5-selective diffusion barrier we identified in astrocytes appears to be selective to a particular protein. It contributes to maintenance of a polarized distribution of mGluR5 and the generation of a polarized Ca2+ signaling pattern.

A role for regulation of mGluR5 diffusion in mGluR-dependent Ca2+ signaling has also been reported in neurons. A reduction in the lateral diffusion of synaptic mGluR5s induced by the accumulation of amyloid β oligomers results in an increase in [Ca2+]i that leads to synapse deterioration in hippocampal neurons (32). This report on neurons—together with our findings in astrocytes—indicates that lateral diffusion of mGluR5 on the cell surface may play a crucial role in the spatial regulation of Ca2+ signaling in the brain.

Increasing evidence for astrocyte modulation of neuronal function has identified astrocyte dysfunction as a contributing factor in various neurological disorders (42). Several studies have unearthed a relationship between spatially and temporally abnormal mGluR5-dependent Ca2+ signaling in astrocytes and some neurological disorders, including Alzheimer’s disease (AD) and epilepsy. In experimental models of these diseases, astrocytes show more frequent somatic Ca2+ transients (1013) and increased mGluR5 abundance (14, 43). Our finding that mGluR5 overexpression enables mGluR5 to overcome the mGluR5-selective diffusion barrier, resulting in increased somatic mGluR5 density, provides a mechanism that could explain how increased mGluR5 abundance could lead to enhanced Ca2+ signaling in the soma. It is conceivable that loss of polarity of mGluR5 distribution, induced by altered mGluR5 lateral diffusion, contributes to abnormal astrocytic Ca2+ signaling in epilepsy and AD. Investigating the molecular mechanisms that control mGluR5 diffusion and spatial patterns of astrocytic Ca2+ signaling may thus lead to a better understanding of these neurological disorders.

In summary, we propose that the existence of a membrane barrier against astrocyte mGluR5 diffusion may have major implications for brain function. In normal astrocytes, the mGluR5-selective diffusion barrier could, by compartmentalization of Ca2+ signaling, allow each process to regulate its contacting partner (synapses or blood vessels) independently, preventing somatic Ca2+ signal, which could be global, metaplastic, and toxic under certain disease conditions. In analogy to the spatiotemporal control of Ca2+ signaling in neurons, we hypothesize that compartmentalized Ca2+ signaling in astrocytes could provide the glial computational equivalent of the neuronal dendrite spine or branch, with substantial implications for coordinated signaling by clustered units of synapses, astrocyte processes, and blood vessels in healthy and diseased brain.

Materials and Methods

Plasmids

For Ca2+ imaging, we used GCaMP2 or GCaMP3. To visualize cell morphology, we used plasmids encoding tdTomato or plasma membrane–targeted cyan fluorescent protein (PM-CFP). PM-CFP was constructed by fusing the N-terminal 13 amino acids of lyn to the N terminal end of enhanced CFP subcloned into pcDNA3.1 (zeo) vector (Invitrogen). For overexpression of mGluR5, mGluR5-pHluorin or mGluR5-mRFP was used. mGluR5-pHluorin was constructed by introducing a fragment coding superecliptic pHluorin with Mlu I cohesive ends into the artificial N-terminal Mlu I site (between Ser22 and Ser23 codons) of mGluR5a (31, 32) in pME18s-mGluR5a plasmid (44). To construct mGluR5-mRFP, we inserted mRFP in front of the first Met of mGluR5a with an intervening sequence of Asn-Ser-Phe-Pro-Lys (NSFPK) in pME18s-mGluR5a plasmid. To express the mGluR5 C terminus, we used mGluR5 C terminus–internal ribosomal entry site (IRES)–GFP and mGluR5 C terminus–IRES–tdTomato. To construct mGluR5 C terminus–IRES–GFP, we inserted mGluR5 C-terminal (amino acids 840 to 1171) fragment into pIRES2-AcGFP1 (Clontech). mGluR5 C terminus–IRES–tdTomato was constructed by replacing GFP of the mGluR5 C terminus–IRES–GFP with tdTomato. GCaMP3 was purchased from Addgene. GCaMP2 was a gift from J. Nakai (Saitama University). pME18s-mGluR5a was a gift from S. Nakanishi (Osaka Bioscience Institute). tdTomato and mRFP were a gift from R. Tsien (University of California, San Diego).

Primary hippocampal culture and transfection

Primary cultures of hippocampal astrocytes cocultured with neurons were prepared from E18 (embryonic day 18) to E20 Wistar rats (Japan SLC Inc.) as previously described (45) with some modifications. Dissociated cells in plating medium were plated at a density of 1.4 × 105 cells/ml onto 18-mm-diameter glass coverslips precoated with 0.04% polyethyleneimine (Sigma). Three days after plating, culture medium was replaced by maintenance medium consisting of Neurobasal A medium (Invitrogen) supplemented with B27, l-glutamine (2 mM), and antibiotics. At least three independent batches of cells were used for each experiment. For transfection, a plated coverslip in 1 ml of culture medium was treated with transfection mixture containing 100 μl of Opti-MEM (Invitrogen), 0.5 μg of DNA, and 1 μl of Lipofectamine 2000 (Invitrogen) and incubated for 1 day for GCaMP2 and GFP, or for 2 days for other DNAs.

Slice preparation and gene transfection

Sagittal hippocampal slices (300 μm thick) were prepared from 11-day-old Wistar rats (Japan SLC Inc.), mounted on cell culture inserts (Millicell, Millipore), and incubated at 33°C under 5% (w/w) CO2 in medium as previously described (46). After incubation for 1 hour, slices were transfected with GCaMP3 plasmid by means of a biolistic particle delivery system (Helios gene gun system, Bio-Rad Laboratories) and then further incubated for 1 day before imaging.

Ca2+ imaging

Ca2+ imaging of cultured astrocytes transfected with GCaMP2 was performed in balanced salt solution (BSS: 20 mM Hepes, 115 mM NaCl, 5.4 mM KCl, 10 mM glucose, 2 mM CaCl2, 1 mM MgCl2, pH 7.4) at 24° to 26°C. GCaMP2 signal was detected with an inverted microscope (IX70, Olympus) equipped with a Plan Apo 60× objective lens [NA (numerical aperture) 1.42] (Olympus), a cooled charge-coupled device (CCD) camera (ORCA II-ER, Hamamatsu Photonics), and appropriate filter sets (excitation: 470 to 490 nm, emission: 515 to 550 nm). Images were acquired at 2 Hz with 100-ms exposure time. For measurement of DHPG-induced Ca2+ transients in Fig. 1, A to C, GCaMP2 fluorescence was excited with a xenon lamp. DHPG (10 μM; Tocris) was bath-applied during the recording. All other Ca2+ imaging experiments in cultured astrocytes, including measurements of spontaneous Ca2+ transients, were recorded with a 490-nm light-emitting diode (LED) illumination system (450 to 550 nm, precisExcite, CoolLED) as a light source.

Ca2+ imaging of astrocytes in hippocampal slices transfected with GCaMP3 was performed in artificial cerebrospinal fluid (ACSF) (125 mM NaCl, 2.5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 1.25 mM NaH2PO4, 25 mM NaHCO3, and 25 mM d-glucose) at 24° to 26°C. The GCaMP3 signal was detected with an upright microscope (BX51WI, Olympus) equipped with a water-immersion objective lens (LUMFL 60×, NA 1.10; Olympus), a CCD camera (ORCA ER, Hamamatsu Photonics), and an appropriate mirror unit (U-MINIBA2, Olympus). Images were acquired at 2 Hz with 200-ms exposure time. GCaMP3 fluorescence was excited with a xenon lamp (Lambda DG-4, Sutter Instrument). DHPG (10 μM; Tocris) was perfused for 1 min during the recording.

Immunocytochemistry

For immunolabeling of calreticulin, astrocytes were fixed for 15 min with 4% (w/v) paraformaldehyde (PFA) in phosphate-buffered saline (PBS). Cells were incubated for 60 min with calreticulin antibody (1 μg/ml; ABR) followed by permeabilization with 0.1% (v/v) Triton X-100 and blocking. For IP3R2, astrocytes were fixed with ice-cold methanol for 5 min and then incubated with IP3R2 antibody (1:250; Chemicon). For immunostaining of mGluR5 on the cell surface, cells were fixed with 4% PFA and incubated with mGluR5 antibody (3.8 μg/ml; Alomone Labs) after blocking without permeabilization. Alexa Fluor 488– or 594–conjugated goat anti-rabbit immunoglobulin G (IgG) (Invitrogen) was used for the secondary antibody. Organotypic slices expressing GCaMP3 were fixed with 4% PFA in PBS for 15 min and then incubated with 0.3% Triton X-100 containing 5% skim milk for 30 min. The slices were then incubated with anti-GFAP antibody (Sigma) followed by incubation with Alexa Fluor 594–conjugated anti-mouse IgG (Invitrogen). Coverslips and slices were mounted on slides with Vectashield (Vector Laboratories).

Immunofluorescence signals were observed with a confocal microscope (FV-1000, Olympus) equipped with a 60× objective (NA 1.42, UPlanApo, Olympus) and a laser diode (LD) of 473 and 559 nm. A series of images was acquired along the z axis at 1-μm intervals (z stack), and maximum projection of z stack was generated with the “stack arithmetic” function in MetaMorph software (Molecular Devices).

Quantitative analysis of mGluR5 immunoreactivity was performed with MetaMorph software (Molecular Devices). Background signal was first subtracted from the maximum projection of z stack image for mGluR5. The average intensity of the mGluR5 signal in the region corresponding to a whole process or the whole soma was measured and then normalized by that of the entire astrocyte. Only mGluR5 signals within the area defined as an astrocyte were included. To define the morphology of an astrocyte, we processed the maximum projection of the corresponding tdTomato z stack image by “nearest-neighbor” function, followed by manual thresholding.

Photolysis of caged IP3

Astrocytes transfected with GCaMP2 were loaded with a mixture of 10 μM membrane-permeable caged IP3 [iso-Ins(1,4,5)P3/PM (caged), Enzo Life Sciences] and 1× PowerLoad (Invitrogen) in BSS for 30 min, washed, and incubated for another 30 min at room temperature in BSS. GCaMP2 signal was recorded with a confocal scanning microscope (FV1000-D, Olympus) with a 60× objective (NA 1.42, UPlanApo, Olympus) and excited with a Multi Ar laser (488 nm). Images were acquired every 250 ms for 10 s. The uncaging stimulation was performed with an LD (405 nm) within an area of 9 μm2 for 1 s during the recording.

Fluorescence recovery after photobleaching

Astrocytes transfected with mGluR5-pHluorin were imaged in BSS on a confocal microscope (FV-1000) with an excitation of 473 nm. After one z stack (15 slices at 1-μm intervals) was acquired with 1% laser power, either the entire soma or a part of a process with an area similar to that of the soma was bleached out with continuous 100% laser power. Laser power was then returned to the initial settings (1%), and the fluorescence recovery in the photobleached area was monitored by acquiring a z stack every 2 min for 58 min. For the analysis of FRAP data, a maximum projection was generated with MetaMorph from the z stack at each time point.

Single-particle tracking

For imaging mGluR5 and P2X7R dynamics, mGluR5s or P2X7R on the cell surface was targeted with streptavidin-conjugated QDs (Qdot605 or 625, Invitrogen) through antibodies recognizing the extracellular domain of mGluR5 or P2X7R (1.9 μg/ml for endogenous mGluR5, 0.38 μg/ml for overexpressed mGluR5, 8 μg/ml for P2X7R; Alomone Labs) and biotinylated Fab fragment as described previously (19). For imaging DOPE dynamics, cells were incubated for 10 min with 10 μM biotinylated DOPE (Avanti) and then incubated with streptavidin-conjugated QDs. Incubation with antibodies or DOPE, washes, and imaging were performed in minimum essential medium without phenol red (Invitrogen) supplemented with 20 mM Hepes, 33 mM glucose, 2 mM glutamine, and 1 mM sodium pyruvate at 37°C for QD labeling and at room temperature for imaging.

The diffusive behavior of mGluR5-QDs and P2X7R-QDs was recorded with the same imaging system as described in the “Ca2+ imaging” section. QDs were excited by 440-nm LED (420 to 470 nm, CoolLED), and emitted light (595 to 615 nm) was captured. DOPE-QD signal was detected with an inverted microscope (IX71, Olympus) equipped with Plan Apo 60× NA 1.45 (Olympus), an EM-CCD camera (Hamamatsu Photonics), appropriate filter sets (excitation: 420 to 490 nm, emission: 595 to 615 nm), and a xenon lamp (Olympus). For the calculation of diffusion parameters, mGluR-QD and P2X7R-QD images were recorded with an integration time of 76 ms for 200 consecutive frames (15.2 s), and DOPE-QDs were recorded with an integration time of 20 ms for 200 consecutive frames (4 s). The trajectory of each particle was obtained by cross-correlating the image with a Gaussian model of the point spread function (47), with TI workbench software written by T. Inoue (Waseda University) as described previously (48). Only single QDs identified by intermittent fluorescence (that is, blinking) were analyzed.

Diffusion parameters were calculated from the trajectories reconstructed from sequential recordings of mGluR5-QD, P2X7R-QD, and DOPE-QD. Values of the mean square displacement (MSD) plot versus time were calculated for each trajectory of mGluR5-QD by applying the relation:MSD(nτ)=1Nni=1Nn[(x((i+n)τ)x(iτ))2+(y((i+n)τ)y(iτ))2](1)where τ is the acquisition time, N is the total number of frames, and n and i are positive integers, with n determining the time increment (49). Diffusion coefficients (D) were calculated by fitting the first four points of the MSD-versus-time curves with the equation:MSD(nτ)=4Dnτ+b(2)mGluR5-QD and P2X7R with diffusion coefficient (D) <0.0002 μm2/s or DOPE-QDs with D <0.016 μm2/s were defined as immobile and not analyzed further. To define the diffusion mode of mGluR5-QDs, we further fitted MSD-nτ plots with the following equation:MSD(nτ)=L23(1exp(12DnτL2))+4Dmacnτ(3)where L2 is the confined area in which diffusion is restricted, and Dmac is the diffusion coefficient on a long time scale (33). The diffusion of mGluR-QD with MSD-nτ plot that does not apply |DDmac| <0.1 × D or L <0.001 was defined as confined diffusion (35). For the comparison of the proportion of mGluR5-QD exhibiting confined diffusion, each pair of two values, that is, one from the process and the other from the soma, was acquired from the same cell.

For long-term tracking, mGluR5-QD and P2X7R-QD signal was recorded at 2 Hz for 10 min with 76-ms exposure time, and DOPE-QD signal was recorded with an integration time of 100 ms for 2000 consecutive frames (200 s). The surface area traveled by mGluR5-QD, P2X7R-QD, or DOPE-QD within this recording period was visualized by the maximum projections over a stack of time series with the stack arithmetic function in MetaMorph.

The likelihood of mGluR5-QD transition between the process and the soma was estimated by calculating the number of events in which one mGluR5-QD crossed a standard line representing the soma-process boundary (p–s line) in 10 min (Eq. 4, fig. S3A). The straight line facing the soma at the base of a process was defined as the p–s line. A parallel line in the process (10 μm from the p–s line) was defined as a process-process standard line (p–p line), and was used to quantify the transition of mGluR5-QD within the process (Eq.5, fig. S3A). Two 10 μm × 10 μm grids were placed on both sides of the p–s line or the p–p line (fig. S3A), and only mGluR5-QDs within these grids were included in the analysis, because this was about the maximum area within which mGluR5-QD could travel to the p–s or p–p lines over the 10-min recording time. The transition number was normalized by the average of control transition (transition over the p–p line) within the process (fig. S3, B and C).

Statistical analysis

All data were acquired from at least three independent batches of cells.

For the calculation of diffusion coefficient, Mann-Whitney U test was used, and for all other statistics, Student’s t test was used. Paired t test was used when two groups were compared within the same cell. P < 0.05 was regarded as significant.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/5/218/ra27/DC1

Fig. S1. Spontaneous Ca2+ transients in astrocytic processes.

Fig. S2. GFAP expression in a GCaMP3-transfected cell.

Fig. S3. Quantification of the likelihood of mGluR5-QD soma-process transition.

Fig. S4. Visualization of lateral diffusion of overexpressed mGluR5 on the cell surface using FRAP.

Fig. S5. Effect of mGluR5 C-terminal partial fragment expression on the mGluR5-selective diffusion barrier.

Movie S1. Spontaneous Ca2+ transients in astrocytes.

References and Notes

Acknowledgments: We are grateful to J. Nakai for GCaMP2; S. Nakanishi for pMe18S-mGluR5; L. Looger for GCaMP3; R. Tsien for tdTomato and mRFP; C. Nakada, K. Suzuki, and A. Kusumi for DOPE labeling protocol; and T. Inoue, M. Dahan, and A. Triller for the analysis program for QD-SPT. We also thank RIKEN BSI-Olympus collaboration center for providing their experimental instruments and C. Yokoyama for his valuable comments on this manuscript. We thank three anonymous reviewers for their insightful and constructive comments. Funding: Supported by RIKEN, Grants-in-Aid for Scientific Research (KAKENHI) (20220007 to K.M.), the Uehara Memorial Foundation, and Toray Science Foundation (to H.B.). Author contributions: M.A., H.B., K.N., T.M., A.M., M.W.S., and T.N. performed the experiments; M.A., H.B., F.N., and M.E. analyzed the data; M.A., H.B., T.N., and K.M. designed the project; and M.A., H.B., and K.M. wrote the paper. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article