|Home | About | Journals | Submit | Contact Us | Français|
Astrocytes are the most abundant glial cell type in the brain. Although not apposite for long-range rapid electrical communication, astrocytes share with neurons the capacity of chemical signaling via Ca2+-dependent transmitter exocytosis. Despite this recent finding, little is known about the specific properties of regulated secretion and vesicle recycling in astrocytes. Important differences may exist with the neuronal exocytosis, starting from the fact that stimulus-secretion coupling in astrocytes is voltage independent, mediated by G-protein-coupled receptors and the release of Ca2+ from internal stores. Elucidating the spatiotemporal properties of astrocytic exo-endocytosis is, therefore, of primary importance for understanding the mode of communication of these cells and their role in brain signaling. We here take advantage of fluorescent tools recently developed for studying recycling of glutamatergic vesicles at synapses (Voglmaier et al., 2006; Balaji and Ryan, 2007); we combine epifluorescence and total internal reflection fluorescence imaging to investigate with unprecedented temporal and spatial resolution, the stimulus-secretion coupling underlying exo-endocytosis of glutamatergic synaptic-like microvesicles (SLMVs) in astrocytes. Our main findings indicate that (1) exo-endocytosis in astrocytes proceeds with a time course on the millisecond time scale (τexocytosis = 0.24 ± 0.017 s; τendocytosis = 0.26 ± 0.03 s) and (2) exocytosis is controlled by local Ca2+ microdomains. We identified submicrometer cytosolic compartments delimited by endoplasmic reticulum tubuli reaching beneath the plasma membrane and containing SLMVs at which fast (time-to-peak, ~50 ms) Ca2+ events occurred in precise spatial-temporal correlation with exocytic fusion events. Overall, the above characteristics of transmitter exocytosis from astrocytes support a role of this process in fast synaptic modulation.
Increasing evidence indicates that astrocytes are competent for regulated exocytosis and play a role in intercellular communication in the brain [for review, see Volterra and Meldolesi (2005) and Haydon and Carmignoto (2006)]. By responding to environmental stimuli with intracellular Ca2+ ([Ca2+]i) elevations and release of chemical transmitters (gliotransmitters) (Bezzi and Volterra, 2001), astrocytes are in the position to exchange a large variety of regulatory signals with surrounding cells. These findings constitute a conceptual breakthrough, although more work is needed to establish the exact contribution of astrocyte signaling to specific brain processes. A major obstacle to achieving this goal is the lack of a detailed description of the basic cell biological properties of astrocytes. The secretory aspect is a typical example. There is accumulating evidence that astrocytes secrete numerous transmitters by Ca2+-dependent exocytosis (Bezzi et al., 2004; Zhang et al., 2004; Chen et al., 2005; Domercq et al., 2006; Bowser and Khakh 2007; Nadrigny et al., 2007). However, the available data are far from giving a complete and coherent description of the properties of the exo-endocytosis processes in astrocytes. Critical information is missing concerning the spatial-temporal characteristics of exocytosis, endocytosis recycling, and the underlying stimulus-secretion coupling mechanism. Moreover, further uncertainty is created by the emerging heterogeneity of the secretory pathways in astrocytes. Various types of organelles have been proposed to underlie exocytosis of gliotransmitters: from synaptic-like microvesicles (SLMVs) (Bezzi et al., 2004; Crippa et al., 2006; Jourdain et al., 2007), to dense-core granules (Coco et al., 2003; Striedinger et al., 2007), to lysosomes (Zhang et al., 2007), and to extra-large organelles with several μm long diameter (Xu et al., 2007). Past studies have not taken into account such heterogeneity. As a consequence, their descriptions of astrocytic secretion are probably not fully accurate, mixing contributions by more than one exocytic organelle population (Chen et al., 2005; Bowser and Khakh, 2007; Jaiswal et al., 2007; Nadrigny et al., 2007; Xu et al., 2007). Among the organelles putatively secreting gliotransmitters, glutamatergic SLMVs are the best and most extensively characterized. They have been identified both in adult brain tissue and in cell culture and studied at the ultrastructural as well as functional level (Bezzi et al., 2004; Crippa et al., 2006; Jourdain et al., 2007; Ni et al., 2007).
To define the specific characteristics of exo-endocytosis of glutamatergic SLMVs, we apply for the first time in astrocytes a strategy recently developed for studying the dynamics of glutamatergic synaptic vesicles (SVs) at synapses (Voglmaier et al., 2006). This consists of using the chimerical protein vesicular glutamate transporter 1 (VGLUT1)-pHluorin in combination with epifluorescence (EPI) and total internal reflection fluorescence (TIRF) illuminations (EPIi; TIRFi). VGLUT1-pHluorin combines the SLMV-specific targeting ensured by VGLUT1 to the properties of super-ecliptic pHluorin, a green fluorescent protein (GFP) with modified pH sensitivity (Miesenböck et al., 1998; Ryan and Reuter, 2001). When attached to the luminal domain of a vesicle resident protein, pHluorin allows a direct detection of fusion and retrieval of single vesicles (Sankaranarayanan and Ryan, 2000; Balaji and Ryan, 2007).
By using EPIi and TIRFi to monitor exo-endocytosis, respectively, at the whole-cell level and at the single-vesicle level, and dual-wavelength TIRF imaging for monitoring in parallel submembrane Ca2+ and SLMV dynamics, our study aims at (1) obtaining the first accurate kinetic description of exocytosis and endocytosis of astrocytic SLMVs and (2) defining the spatial-temporal characteristics of the underlying stimulus-secretion coupling mechanism.
All agents were from Sigma-Aldrich, unless otherwise indicated. Plasmids containing the VGLUT1-pHluorin and VGLUT1-mCherry constructs were prepared as described previously (Voglmaier et al., 2006); the endoplasmic reticulum (ER)-GFP plasmid was kindly provided by N. Demaurex (University of Geneva, Geneva, Switzerland) (Demaurex and Frieden, 2003). Each of these plasmids (0.5 μg for single transfection experiments or 0.25 μg for double transfection experiments) was transfected into primary rat cortical astrocytes cultures with FuGene6 (3 μl; Roche Diagnostics).
Astrocyte cultures containing >99% GFAP-positive cells (≤8% of which positive for the neural precursor marker LeX) were obtained from newborn rats. They were prepared as described previously (Bezzi et al., 2001), plated (2.5 × 104 cells) on glass coverslip and transfected 6–8 d later with VGLUT1-pHluorin, VGLUT1-mCherry or ER-GFP constructs. The double transfection protocol with VGLUT1-pHluorin and VGLUT1-mCherry resulted in an almost complete colocalization of the two fluorescent proteins (85 ± 11%). After transfection (2–5 d), coverslips were mounted in the open laminar flow perfusion incubator at 37°C (Harvard Apparatus) on the stage of a Zeiss Axiovert 200 fluorescence inverted microscope modified for TIRF and EPI experiments (Visitron System). The experimental chamber (250 μl volume) was perfused at a rate of 1–1.5 ml/min. The stimulus [(RS)-3,5-dihydrophenylglycine (DHPG); Tocris Bioscience] was applied rapidly (2 s) via a software-controlled microperfusion fast-step device (100 μl/min; Warner Instruments). Cells were perfused at 37°C in a HEPES-Krebs-Ringer HEPES (KRH) buffer containing (in mm) 120 NaCl, 3.1 KCl, 2 MgCl2, 1.8 CaCl2, 1.25 NaH2PO4, 25 HEPES-Na (buffered to pH 7.4), glucose 4. Bafilomycin A1 (Baf A1) (Calbiochem) was used at a final concentration of 5 μm (diluted with HEPES-KRH) and incubated for 7–10 min before the stimulus. In experiments with thapsigargin (thapsi) or cyclopiazonic acid (CPA) (Calbiochem), the drugs were diluted in HEPES-KRH, incubated for 15–20 min, and the stimulus was subsequently applied when the Ca2+ signal recovered to basal level. SLMVs expressing VGLUT1-pHluorin were labeled by DHPG (2 s, 100 μm) in saline containing 15 μm FM 4–64 (Invitrogen).
For Ca2+ imaging experiments, cells (transfected or not transfected) were loaded with 5 μm of Fluo4 AM (Invitrogen) for 15 min or with 2.5 μg/ml of x-rhod-1 AM (Invitrogen) for 10 min in the presence of 0.02% pluronic F-127 (Invitrogen) at room temperature in the dark in a HEPES-KRH buffer, pH 7.4, and then de-esterified for 10–15 min before imaging.
EPIi or TIRFi were alternatively used for our experiments. EPIi and TIRFi were alternated in real time for whole-cell pHluorin fluorescence (pHF) signal measurements in individual cells, in basal conditions and upon stimulation with DHPG. By imaging with excitation light at 488 nm generated by a 488 nm laser (20 mW; Laserphysics) and by a polychromator illumination system (Visichrome), the pHF signal was recorded at 10 Hz through a 100× objective lens (Zeiss; α-plan FLUAR 100X, 1.45 NA) and filtered with Zeiss filter set 10 (Zeiss).
TIRFi was used for all of the remaining experiments. The expanded beam of a 488/568 nm argon/krypton multiline laser (20 mW; Laserphysics) passed through an acousto-optic tunable filter laser wavelength selector (VisiTech International) synchronized with a soluble N-ethylmaleimide-sensitive factor attached protein-HQ charge-coupled device camera (Roper Scientific) under Metafluor software (Universal Imaging) control and was introduced to the coverslip from the high numerical aperture objective lens (α-plan FLUAR 100X; Zeiss). Light entered the coverslip and underwent total internal reflection at the glass-cell interface. In our experimental conditions, penetration depth of TIRFi was calculated by following, with some modifications, the method setup by Mattheyses and Axelrod (2006) to be ~92.12 nm. In single-wavelength TIRFi experiments (488 nm), the laser beam was filtered via the Zeiss filter set 10, and images were acquired at 20–40 Hz (Zeiss). In dual-wavelength TIRF illumination (488/568 nm), laser beams were combined by a dichroic mirror from the Zeiss filter 24 at 20–40 Hz. In TIRFi experiments, at which we monitored [Ca2+]i events as fast as possible (40 Hz), we restricted analysis to 10 regions of interest (ROIs) per cell located at different subdomains (i.e., in the processes and cell body). The pixel size was 126 nm (at binning 2).
To study exo-endocytosis of astrocytic glutamatergic SLMVs, we transiently expressed VGLUT1-pHluorin in cultured astrocytes. Confocal analysis of the intracellular distribution of the VGLUT1-pHluorin showed that the protein mainly colocalized with endogenous cellubrevin (>90%) (supplemental Fig. 1 SI, available at www.jneurosci.org as supplemental material), endogenous glutamate (supplemental Fig. 3 SI, available at www.jneurosci.org as supplemental material), and partially overlapped with early or recycling endosomes (~20 and 30%, respectively), trans-Golgi network (~8%), and late endosomes/lysosomes (~6%) (Fig. 2 SI, available at www.jneurosci.org as supplemental material). By combining EPIi and TIRFi (10 Hz), we monitored pHF changes under basal conditions and upon stimulation of group I metabotropic glutamate receptors (mGluR) via local application of the specific agonist DHPG. The use of EPIi in real time with TIRFi was aimed at providing a comparative analysis of pHF changes in the whole cell and in its submembrane compartment. The sequence of images in Figure 1A shows that, under basal conditions, the pHF signal is very low both under TIRFi and EPIi; only a few fluorescent spots are visible, representing vesicles undergoing spontaneous discharge. Application of DHPG (100 μm, 2 s), which is known to trigger glutamate release from astrocytes (Bezzi et al., 1998, 2004), causes a rapid and transient increase in pHF consistent with exposure of the pHluorin epitope to the external pH (~7.4) in coincidence with vesicular fusions with the plasma membrane (Fig. 1 Aa–c). By placing a ROI on the whole-cell perimeter, we measured the kinetics of the DHPG-evoked pHF signal at the whole-cell level. Figure 1B compares the kinetics of the total pHF signal induced by DHPG under TIRFi and EPIi. In both cases, the rising phase of the signal is faster than its decay phase. However, under TIRFi, rise and decay time are significantly faster than under EPIi. To interpret this observation one has to take into account that the curve of the DHPG-evoked pHF signal represents at all times the net balance of exocytosis and endocytosis/reacidification (Sankaranarayanan and Ryan, 2000; Atluri and Ryan, 2006; Voglmaier et al., 2006). The rising phase indicates prevalent exocytosis and the decaying phase prevalent endocytosis and reacidification (Sankaranarayanan and Ryan, 2000; Schweizer and Ryan, 2006). On this basis, one can hypothesize that (1) the rise time of the DHPG-evoked pHF signal under TIRFi is faster than under EPIi because EPIi monitors vesicle reacidification processes occurring in cytosolic layers not accessible to TIRFi [penetration of the evanescent field (EW) in our apparatus is ~90 nm], and (2) the decay time under TIRFi is faster than under EPIi because endocytosed vesicles move out of the EW field before undergoing reacidification. To test these hypotheses, we performed “alkaline trapping” experiments to fully separate the three processes and determine their respective kinetics by using the V-type ATPase inhibitor Baf A1 (Sankaranarayanan and Ryan, 2000). Taken up by SVs during exocytosis, Baf A1 blocks the proton pump and prevents reacidification, trapping VGLUT1-pHluorin-expressing vesicles in the fluorescent state. Addition of Baf A1 thereby eliminates all changes of the pHF signal attributable to the reacidification component of recycling to reveal pure exocytosis. Astrocytes were exposed to 5 μm Baf A1, a dose known to completely inhibit extinction of the pHF signal in synaptic terminals (Sankaranarayanan and Ryan, 2001). As shown in Figure 1C, in the presence of Baf A1, under EPIi, DHPG evoked a nondecaying pHF signal that reached plateau in 1.6 ± 0.33 s. Given that its rising phase corresponds to pure exocytosis, the slope of the linear fit represents the rate of the exocytic process (0.77 ± 0.056 A.U./s). By calculating the first derivative of the curve, we found that exocytosis reaches maximal speed at 800 ms (Fig. 1C, inset). This time represents the inflection point of the curve that indeed tends to have a flattened “S” shape, suggesting that exocytosis proceeds with an exponential time course for 800 ms and then gradually slows (τriseEPIBaf 0–800 ms = 0.24 ± 0.017 s). Interestingly, in the presence of Baf A1, the TIRFi curve shows a rise time almost identical to the one of EPIi and comparable with those of TIRFi curves obtained in the absence of Baf A1 (τriseTIRFBaf = 0.25 ± 0.092 s vs τriseTIRF = 0.3 ± 0.088 s). These results confirm our hypothesis and indicates that (1) the difference between EPIi and TIRFi observed in the absence of the drug is attributable to reacidification occurring in the EPIi field but not in the TIRFi field and that (2) Baf A1 has no effect on both the rise and decay phases of the TIRFi (τdecayTIRF Baf = 1.59 ± 0.09 s vs τdecayTIRF = 1.63 ± 0.1 s) curve because vesicles escape the EW field before reacidification. Moreover, the identical slope observed in TIRFi and EPIi curves in the first 800 ms indicates the absence of vesicle movement out of the EW field (endocytosis) in this initial period. All of the parameters we calculated are summarized in supplemental Table 1, available at www.jneurosci.org as supplemental material.
We subsequently estimated reacidification (Fig. 1D; supplemental Fig. 2 SI, available at www.jneurosci.org as supplemental material) by subtracting the normalized EPIi curve in the absence of Baf A1 (Fig. 1B) from the one in the presence of the drug (Fig. 1C). During the stimulus, similarly to exocytosis, the reacidification proceeds with an exponential time course for 800 ms and then gradually slows (the curve shows a sigmoidal shape centered at 800 ms). The exponential fits allowed us to estimate the time constant of reacidification processes (τrise 0–800 ms = 0.22 ± 0.02 s) that, based on the previous results, are detected only by EPIi and involved pool of vesicles located deep in the cytoplasm and out of the EW field. The poststimulus reacidification instead displays a linear increase reaching a plateau in ~10 s (0.052 ± 0.0008 A.U./s). We finally estimated endocytosis (Fig. 1E; supplemental Fig. 2 SI, available at www.jneurosci.org as supplemental material) by subtracting the normalized TIRFi curve from the normalized EPIi curve, both obtained in the presence of Baf A1 (Fig. 1C). The endocytosis starts 600 ÷ 800 ms after the stimulus, proceeds with an exponential time course for 400 ÷ 600 ms, to then gradually decline (τrise 800–1200 ms = 0.26 ± 0.03 s). Interestingly, by calculating the first derivative (Fig. 1E, inset), we found that the endocytosis curve contains two inflection points at 1.2 s and at 2.8 s (τrise 2000–2800 ms = 1 ± 0.25 s), suggesting the presence of two distinct phases of endocytosis. The rates of endocytosis during and after the stimulus were estimated by linear fitting (0.8–2 s, 0.66 ± 0.025 A.U./s; >2 s, 0.16 ± 0.005 A.U./s) (see Fig. 6).
In the subsequent set of experiments, we confirmed that exo-endocytosis is controlled by the intensity of G-protein-coupled receptor (GPCR) stimulation as the peak value of the pHF signal varied with the dose of agonist (Fig. 1F). Moreover, we found that the process requires release of Ca2+ from internal stores. Indeed, the DHPG-evoked pHF signal was abolished by inhibition of the ER Ca2+-ATPase by either thapsi [−91.7 ± 2.7%; 50 nm, 20 min; n = 10 cells, 5 experiments (exps)] or CPA (−80.6 ± 6.6%; 10 μm, 15 min; n = 8 cells, 4 exps).
To study the kinetics of exo-endocytosis at the level of individual SLMVs, we performed experiments under TIRFi that, producing an EW, in our set-up excites fluorophores within 80–90 nm from the plasma membrane (Domercq et al., 2006). Astrocytes were transfected with two different constructs: VGLUT1 coupled to mCherry (red, VGLUT1-mCherry), labeling the SLMV membrane, and VGLUT1-pHluorin (green), signaling exocytic fusion of the SLMV (Tsuboi and Rutter, 2003).
By using a dual wavelength TIRF (568/488 nm), we then monitored (at 20 Hz) individual double-fluorescent SLMVs and evaluated their emission intensities in both a small circle around the vesicle and a concentric annulus around the circle (Bezzi et al., 2004). Under basal conditions, the vast majority (97.4%) of the observed fusion events fell into two well-defined and distinct categories (subtype 1 and subtype 2 events). Subtype 1 events accounted for 63% of the events (n = 26 cells). They were characterized by the abrupt appearance in the EW field of the pHF signal that remained constant for 400 ms to then disappear without any significant lateral diffusion (Fig. 2A). VGLUT1-mCherry fluorescence (mCF), already present in the EW field at the appearance of the pHF signal, remained constant (approximately −0.01% of ΔF/F0) during the 400 ms of the fusion event, and then disappeared together with the pHF signal. We interpret the simultaneous disappearance of the two fluorescent markers as a movement of the SLMV out of the EW field. Moreover, the fact that, during the fusion event, both markers were retained in the vesicle membrane strongly suggests a “kiss-and-run” mode of exo-endocytosis (Fesce et al., 1994) (Fig. 2A, diagram). SLMVs (8.7%) showing a subtype 1 fusion displayed a much longer increase of fluorescence (up to 7.5 s) before returning to basal level.
In the case of a subtype 2 event, fusion lasted longer and proceeded differently. The pHF signal, once it appeared, started to diffuse in the external annulus and was diluted out in ~700 ms; in parallel, the mCF signal progressively decreased in intensity (−86 ± 0.7% of ΔF/F0 in 400 ms; n = 475; p < 0.05, unpaired t test) (Fig. 2B). These two combined events suggest that the SLMV loses its identity by mixing with the plasma membrane and undergoes a classical “full-collapse” type of fusion (Fig. 2B, diagram).
To confirm that increases in pHF signal depended on exocytic fusion of SLMVs with the plasma membrane, we performed a parallel set of experiments using VGLUT1-pHluorin-transfected astrocytes bathed in external medium containing FM 4–64, an amphipathic dye that loads into vesicles only when they fuse with the plasma membrane (Gaffield and Betz, 2006). Indeed, in all of the cases when a pHF signal appeared, the FM dye immediately started to accumulate in the SLMV membrane (Fig. 2A,B, bottom). Its fluorescence remained visible throughout the duration of the pHF signal and eventually disappeared with it, indicating that FM 4–64 was taken up by the fused vesicle and followed its fate.
Subsequently, we studied the time distribution of the individual fusion events evoked by GPCR stimulation in the doubly transfected cells. Figure 2C shows that application of DHPG triggered a burst of exocytosis comprising two phases: a rapid one, peaking at 200–300 ms, and a slower one, peaking at 400–600 ms. We used the mCF signal to subdivide SLMVs into “residents” (vesicles present in the EW field before the stimulus) and “newcomers” (vesicles whose mCF signal appeared in the EW after the stimulus) (Zenisek et al., 2000). We found that SLMVs undergoing exocytosis during the rapid phase (0–400 ms) were mostly residents (83%, corresponding to 42% of the total number of DHPG-evoked fusion events; n = 8400), whereas vesicles undergoing exocytosis during the slow phase (500 ms–1.6 s) were almost all newcomers (88%), possibly representing recycling vesicles (Fig. 2D). Interestingly, the majority of resident SLMVs that fused underwent subtype 1 (92.5%), whereas the majority of newcomer SLMVs that fused underwent subtype 2 (84.6%).
The spatial-temporal characteristics of the [Ca2+]i signals triggering exocytosis in astrocytes have not yet been investigated. According to previous study, somatic [Ca2+]i elevations in response to GPCR activation peak within 1–2.5 s from stimulation (Fiacco and McCarthy, 2006). Therefore, these signals are too slow to underlie the exo-endocytosis processes reported here. Considering that SLMV exocytosis takes place on a millisecond time-scale and depends on Ca2+ release from internal stores, one would predict the existence of localized ER-dependent [Ca2+]i events near the fusion sites. As a consequence, we predicted that ER tubuli lie close to SLMVs and in proximity of the cell plasma membrane. To directly verify this possibility, we labeled the ER tubuli by transfecting astrocytes with ER-GFP, a construct that targets GFP expression to the ER (Demaurex and Frieden, 2003). Figure 3A shows that the ER in astrocytes is a continuous network of tubuli and cisternae. Under TIRFi, the fluorescent structures appear discrete, consisting of tubular tips approaching the plasma membrane (Fig. 3B). Cotransfection of ER-GFP and VGLUT1-mCherry revealed an extensive overlay of ER and SLMVs, demonstrated by the close proximity of most of the red dots (~87%) to the green ER structures (Fig. 3B). By means of a cross-correlation analysis performed at the all-cell level, we found that ER and SLMVs are spatially correlated (supplemental Fig. 8 SI, available at www.jneurosci.org as supplemental material). For a detailed analysis, we then focused on red dots corresponding to individual SLMVs (supplemental Fig. 6 SI, available at www.jneurosci.org as supplemental material). A line of fluorescence intensity profile was traced across any given red dot and its closest green structure to analyze their fluorescence distribution and to extrapolate the interdistance between ER and SLMV (supplemental Fig. 7 SI, available at www.jneurosci.org as supplemental material, and image analysis SI, available at www.jneurosci.org). The most frequent result (47 ± 7.5% of cases; n = 6 cells) was an SLMV and ER tubuli located side by side (Fig. 3B, case 1). Their fluorescence curves typically overlapped, indicating that ER and SLMV were located within a distance smaller than the point spread function of our microscope. Their average interdistance interval was 375 ± 185 nm. Another frequent situation (31.4 ± 5.9% of cases) was the alignment of SLMV and ER on the z-axis, as shown by the complete overlap of the two intensity profiles despite a difference in intensity between corresponding peaks (Fig. 3B, case 2). In a small number of cases (8.4 ± 7.4%), the ER-GFP profile had a “donut-shaped” distribution with a central dimple occupied by the VGLUT1-mCherry-expressing vesicle (Fig. 3B, case 3). Taking into account that our study was performed in the limited submembrane field illuminated by TIRF, we conclude that ER tubuli and SLMVs, together with the plasma membrane, delimit submicrometer cytosolic compartments.
Given the presence of ER tubuli together with SLMVs just beneath the plasma membrane, we subsequently used TIRFi imaging (488 nm, 20 Hz) to selectively investigate submembrane Ca2+ signals occurring in the EW field (Steyer and Almers, 2001). Experiments were performed on nontransfected astrocytes loaded with a green [Ca2+]i indicator, Fluo4-AM, and Ca2+ signals studied both under basal conditions and upon GPCR stimulation with DHPG. For analysis, we divided the cell profile appearing under TIRFi in small equal-sized (4 μm2) squared regions, each of which could contain one SLMV and its surrounding ER structure (supplemental Fig. 9A SI, available at www.jneurosci.org as supplemental material). At first, we compared the average submembrane fluorescence signal (F) in 1 s periods taken before, during, and after stimulation with DHPG. F corresponds to the mean fluorescence value given by the sum of the signals from individual regions divided by the number of regions. In the presence of DHPG, F was significantly higher than in the periods immediately preceding and after the stimulation, indicating that mGluR stimulation triggers elevation of submembrane Ca2+ (supplemental Fig. 9B SI, available at www.jneurosci.org as supplemental material) (DHPG, 234.32 ± 2.92 F; before, 203.02 ± 0.86 F; after, 218.33 ± 0.82 F; n = 1275 traces in 15 cells; p < 0.05, multiple-comparison ANOVA). We then proceeded to define the spatial-temporal characteristics of individual submembrane Ca2+ events. By means of a retrospective analysis, we centered the ROIs on each visualized Ca2+ signal and analyzed the fluorescence traces. The submembrane Ca2+ signals have been expressed has ΔF (F−F0) divided by F0, in which the F0 is the mean of ~1 s of the basal level in each respective ROI before the stimulus (supplemental Fig. 10 SI, available at www.jneurosci.org as supplemental material; Image analysis SI, available at www.jneurosci.org). As shown in Figure 4A, very rapid, spike-like Ca2+ events were easily distinguishable. By acquiring images at 40 Hz, we highlighted the morphology of such spiking Ca2+ events and established that most of them last ~150 ms and peak at 50 ms (n = 583 traces from five cells) (Fig. 4A, insert). Under basal conditions, the frequency of the rapid submembrane Ca2+ events was very low (0.87 ± 0.34/s/ROI; n = 1005 events in 25 cells). Application of DHPG induced within 300 ms an ~4-fold increase of their mean frequency (3.45 ± 1.13/s/ROI; n = 6706 events in 25 cells; p < 0.05, unpaired t test). It also increased their mean amplitude by ~1.7-fold (1.62 ± 0.72 basal vs 2.8 ± 0.7 stimulated ΔF/F0; n = 6706 events in 25 cells; p < 0.05, unpaired t test) but did not modify their temporal profile. In parallel, we evaluated the propagation of the spiking Ca2+ signals in the x–y plane by following the lateral spread of fluorescence in three concentric ROIs (diameter: 2, 4, and 6 μm, respectively) centered on each Ca2+ signal. The diameter of most of the Ca2+ events was ≤3 μm. The average diameter (1.75 ± 0.18 μm) was increased by DHPG application (2.5 ± 0.34 μm; p < 0.05, unpaired t test). Interestingly, we noticed that DHPG induced a small fraction (~7.5%) of the Ca2+ events to propagate at long distance (up to 14 μm) from the ROI in which they originate, causing a submembrane Ca2+ wave (data not shown). Our results indicate that submembrane Ca2+ events are generally short lived and spatially confined. Their specific characteristics in terms of amplitude (ΔF/F0 maximum), kinetics, and lateral propagation are summarized in Figure 4B. The temporal distribution of the submembrane Ca2+ events evoked by DHPG application is shown in Figure 4C. Such a distribution corresponds to a biphasic burst in all similar to the one observed for the exocytic events (Figs. 2C; ;4C,4C, ghost). Indeed, DHPG-evoked Ca2+ and exocytic events were found to be strictly temporally correlated; in particular, each one of the two peaks of Ca2+ events preceded the corresponding peak of exocytic events by an interval of ~100 ms.
Finally, with pharmacological experiments shown in Figure 4D, we confirmed that submembrane Ca2+ events depend exclusively on Ca2+ released from internal stores. Thus, the number of events remained unchanged when we removed Ca2+ from the incubation medium (0 Ca2+ with 5 mm EGTA) or added nifedipine (3 μm), a blocker of L-type voltage-gated Ca2+ channels. In contrast, the number of events increased during the first 15 s of incubation with CPA (approximately by 3.8-fold) (supplemental Fig. 11 SI, available at www.jneurosci.org as supplemental material) and returned to a baseline value between 1 and 7 min (0.76 ± 0.25/s/ROI) and eventually almost completely disappeared in ~10 min (0.095 ± 0.014/s/ROI). After 15 min of incubation with CPA, application of DHPG failed to induce submembrane Ca2+ events (Fig. 4D). These results constitute the first evidence of spatially confined, ER-mediated submembrane Ca2+ events occurring in astrocytes on a millisecond time scale.
To directly prove that fast submembrane Ca2+ elevations are the events triggering exocytosis of SLMVs, we monitored the two types of event simultaneously and defined their spatial and temporal correlation. First, we focused on the spatial aspect. Astrocytes were transfected with VGLUT1-mCherry, loaded with Fluo4-AM, and imaged under dual wavelength TIRFi (488/568 nm, 40 Hz). To monitor Ca2+ events at the maximal possible speed, we restricted sampling to 10 ROIs randomly selected at different cell locations, in both the cell body and the processes. ROIs (6 μm2) were centered on an individual SLMV, identified by its mCF signal (supplemental Fig. 1 SI, available at www.jneurosci.org as supplemental material). When we stimulated with DHPG, spike-like submembrane Ca2+ events (1.6 ± 0.5/s/ROI; n = 250 ROI in 25 cells) identical to those observed previously in nontransfected cells appeared in the ROIs (average time to peak, 58.6 ± 8.2 ms; diameter, 2.1 ± 1.6 μm; n = 1340 traces from 10 cells) (Fig. 5A). Interestingly, by means of cross-correlation analysis, we found that the submembrane Ca2+ events, similarly to the ER structures, occurs in strict spatial correlation with submembrane sites SLMVs (those most probably representing the ready to fuse vesicles) (supplemental Fig. 8 SI, available at www.jneurosci.org as supplemental material). For detailed analysis, we then traced a line of fluorescence intensity profile over the mCF signal and the closest Fluo4 fluorescence signal. Practically, in all cases that we analyzed (96.4 ± 2.4%; n = 250 pairs in 25 cells) (Fig. 5A), the two curves overlapped for >75% and their peaks had an average interdistance of 245 ± 8.3 nm. Interestingly, in some cases (12.6% of pairs), the [Ca2+]i event showed a “donut-shape distribution” (reminiscent of the ER distribution around SLMV) in which the fluorescence intensity profiles of the mCF were bell shaped, whereas those of Fluo-4 showed humps on both sides of the SLMV profile, suggesting that [Ca2+]i rises to highest levels around the vesicular organelle.
Subsequently, we considered the temporal aspect. Individual Ca2+ events and single SLMV fusions were monitored in parallel, the latter revealed by VGLUT1-pHluorin, the former by the red calcium indicator x-rhod-1-AM (488/568 nm, 20 Hz). To compensate for the fact that x-rhod gives a smaller ΔF than Fluo4 for the same Δ[Ca2+], we restricted our analysis to ROIs selected in the cell regions with highest signal-to-noise ratio (i.e., the cell processes and the border of the cell body). In 28 cells stimulated with DHPG, we most often observed (65% of 225 cases) a Ca2+ event rapidly followed (≤3 frames) by a fusion event (Fig. 5B). In only 20% of cases, the Ca2+ event was not followed by a detectable fusion event in the same ROI, and in 15% of cases, the fusion event was not preceded by a Ca2+ event. The averaged kinetics of Ca2+ and fusion events in Figure 5C show a strict temporal correlation, with the Ca2+ peak preceding the pHF peak by 57.9 ± 4.5 ms (n = 35 from 10 cells). Importantly, when we depleted the ER Ca2+ stores with CPA, we observed a dramatic reduction of both submembrane Ca2+ events and SLMV fusion events (−85 ± 14% and −89 ± 12%, respectively; n = 102 ROIs in 12 cells), implying that the two types of event are causally related. Overall, our observations indicate that exocytosis of SLMVs in response to GPCR stimulation is sustained by fast, local Ca2+ elevations mediated by Ca2+ release from the ER and suggest for the first time that the stimulus-secretion coupling in astrocytes occurs at the level of functional microdomains.
Although it is increasingly clear that astrocytes express multiple populations of secretory organelles, imaging studies performed to date used generic fluorescent markers of exocytosis that do not distinguish among populations. The simultaneous recording of fusion signals from distinct populations of secretory organelles may lead to significant discrepancies in the kinetic values of exocytosis (Bezzi et al., 2004; Chen et al., 2005; Bowser and Khakh, 2007; Jaiswal et al., 2007). In the present study, we took advantage of VGLUT1-pHluorin to overcome limitations of previous works. Indeed, the intracellular distribution of VGLUT1-pHluorin showed that the vast majority of the GFP-labeled vesicles were positive for a SNARE (soluble N-ethylmaleimide-sensitive factor attachment receptor) protein cellubrevin (Bezzi et al., 2004; Crippa et al., 2006), displayed a diameter compatible with SLMVs in other secretory cells and release glutamate (supplemental Figs. 1 SI, 2 SI, 3 SI, 5 SI, available at www.jneurosci.org as supplemental material). We performed several sets of experiments to investigate spatial and temporal relations between [Ca2+]i and exocytosis processes in astrocytes.
At first, we defined the kinetics of exocytosis, endocytosis, and reacidification at the whole-cell level (Fig. 1). A major challenge consisted in separating the exocytosis and the two components of the endocytic process: the movement of the recently fused vesicles from plasma membrane to cytoplasm (out of the EW field) and the reacidification process. This was achieved with alkaline trapping experiments in that we isolated the exocytosis signal by blocking reacidification with the proton pump inhibitor Baf A1 (Sankaranarayanan and Ryan, 2001). In the presence of Baf A1, the rising phase under EPIi represented a pure measure of exocytosis (Fig. 1C). Comparison with the EPIi signal in the absence of Baf A1 enabled us to calculate the cumulative amount of reacidification by calculating the difference between the two (Fig. 1D); the resulting “difference curve” provided the complete kinetic of reacidification during and after the stimulus. Interestingly, in the first 800 ms, we calculated a fast rate of reacidification (τ = 0.22 ± 0.02) (Fig. 1D) that occurs in pool of vesicles located out of the EW field. The fast reacidification proceeds with a time course similar to exocytosis and most likely represents a mechanism responsible for uptake glutamate into the cytosolic pool of SLMVs.
By comparing the normalized EPIi and TIRFi curves in the presence of Baf A1, we then calculated the kinetic of the movement of SLMVs out of EW field (the first step of the endocytosis) (Fig. 1E). The first derivative of the endocytosis curve shows the evolution of the speed of endocytic events (Fig. 1E, inset); they start at ~600 ms, reach two maxima at 1.2 s, and at 2.8 s to then end after ~6 s. The bimodal distribution indicates the existence of two distinct phases of endocytosis. By directly comparing the kinetics of endocytosis and exocytosis (Fig. 6), we found that during the stimulus, the rate of endocytosis was more than half that of exocytosis (Endo/Exo, 0.85), indicating that SLMV retrieval proceeds fast enough to compensate for exocytosis (Fernández-Alfonso and Ryan, 2004).
In the second set of experiments we performed the analysis at the single vesicle level with TIRFi (Fig. 2). The fast imaging protocol applied in these sets of experiments provided new informations on the kinetic and modalities of exocytosis and recycling.
We found that the exocytic burst, similar to endocytosis, displayed a bimodal distribution (Fig. 2C), strongly suggesting the existence of two modes of exo-endocytosis. We confirmed this hypothesis with a double transfection approach in that we could directly visualize the fate of SLMVs during exocytosis and retrieval (Fig. 2A,B,D). Indeed, we noticed heterogeneity both in the origin of the SLMVs that underwent exocytosis (Fig. 2D, residents vs newcomers) and in their mode of fusion (kiss-and-run vs full-collapse type of fusion). The rapid phase of the exocytic burst (0–400 ms) was sustained almost exclusively by resident vesicles undergoing kiss-and-run fusion, whereas the slow phase (500 ms–1.6 s) mainly by newcomer vesicles undergoing full-collapse fusion. This duality is reminiscent of observations in neurons in that only readily releasable SVs are rapidly recycled and reused (Harata et al., 2006). Our experiments do not clarify whether newcomers and resident vesicles represent distinct population of SLMVs. The newcomers may indeed represent the same population of resident vesicles that undergo a second round of exocytosis upon rapid recycling. In previous studies in which we used acridine orange (AO) as a fluorescent reporter of fusion events, DHPG stimulation induced a single monophasic burst of exocytosis, much shorter (~500 ms) than the one observed here (Bezzi et al., 2004; Domercq et al., 2006). In light of the present observations, it is possible that the release of AO in the external medium upon vesicle fusion prevented us from detecting exocytic events of recycled vesicles.
What can be the functional significance of the two modalities of secretion detected in astrocytes?
Studies in synaptic terminals indicate that full-collapse fusion (Heuser and Reese, 1973) and kiss-and-run (Fesce et al., 1994) coexist and play active roles in exocytic events. In particular, in small nerve terminals, kiss-and-run has an additional role of enabling nerve terminals to respond to high-frequency inputs with a scarce number of vesicles. In astrocytes, it is possible that, such as at nerve terminals, the two modalities of fusion can each contribute to maintain cellular communication over a wide range of different stimuli. Further studies are necessary to address this point.
The present study addresses for the first time whether localized microdomains of Ca2+ control fusion events in astrocytes. Previous studies that monitored Ca2+ responses to GPCR activation focused on bulk [Ca2+]i changes, in general limiting their observations to the somatic cell region (Pasti et al., 1997; Bezzi et al., 1998; Pasti et al., 2001; Fiacco and McCarthy, 2006). Such Ca2+ elevations normally peaked within 1–2 s from stimulation and thus are too slow to underlie the rapid burst of exocytosis described here.
Here, by using TIRFi, we observed that localized domains of Ca2+ exist at or near the sites of exocytosis. ER tubules and cisterns come in tight apposition of the plasma membrane, with a complex spatial arrangement that delimits tiny structural domains of submicrometer space (femtoliter). Such a spatial organization most likely limits diffusion of signaling molecules, including Ca2+. Moreover, ER structures and SLMVs lie in the submembrane compartments in tight spatial proximity, with an average distance of 300–500 nm. These results are in line with recent observations in that ER tubules have been considered to be distinct compartments, structurally and functionally coupled to the plasma membrane via cytoskeletal scaffold proteins (Tse et al., 1997; Blaustein and Golovina, 2001; Sala et al., 2005; Wu et al., 2006). In this respect, the organization of cytoskeleton in astrocytes could be similar to those at glutamatergic synapses, in which the scaffold protein Homer provides a molecular link between IP3 receptors at the tip of the ER cisternae and mGluRs on the plasma membrane (Tu et al., 1998; Xiao et al., 1998; Sala et al., 2005).
What is the functional implication of the subplasma membrane ER domains? We found that stimulation of group I mGluRs generates store-dependent submicrometer Ca2+ events characterized by fast kinetics (50–150 ms) and spatial segregation (≤3 μm).
Importantly, in most cases Ca2+ events occurred at or near sites in which SLMVs underwent exocytosis (≤280 nm) and were in strict temporal and spatial correlation with fusion events (supplemental Fig. 8 SI, available at www.jneurosci.org as supplemental material). Indeed, Ca2+ and exocytosis displayed similar, biphasic distribution, with the Ca2+ peaks preceding the corresponding exocytosis peaks of ~100 ms.
The typology of the subplasma membrane Ca2+ events observed in astrocytes is strongly reminiscent of the so-called “Ca2+ puffs” (Thomas et al., 2000; Bootman et al., 2001) that typically consist of a fast elevation of the [Ca2+]i and limited spatial spread (Rizzuto and Pozzan, 2006). As for the astrocytic signals, we do not know whether their spatial characteristics reflect the clustering of IP3Rs along the ER structures or the clustering of receptors along the plasma membrane, or both. The cumulative recruitment of Ca2+ puffs can lead to the initiation and propagation of a global cytosolic Ca2+ signal (Tovey et al., 2001). This may also occur in astrocytes; in limited cases, we observed long-distance (several μm) propagation of the Ca2+ events from the site of origin along the subplasma membrane, in the form of a submembrane Ca2+ wave. Moreover, in view of the limited depth of the TIRFi field, we cannot exclude that Ca2+ events propagate also in the z-axis.
How do subplasma membrane Ca2+ events control SLMV exo-endocytosis in astrocytes? In resting conditions, Ca2+ events display spontaneous oscillatory patterns that upon stimulation with DHPG increase in both the amplitude and the frequency (~2- and 4-fold). However, at present, we cannot say whether one of the two plays a predominant role in triggering exocytosis.
Are these findings relevant to understanding the functional role of transmitter release from astrocytes in brain function? A few studies performed in vivo to date demonstrate that astrocytes respond to neuronal activity with slow and long-lasting [Ca2+]i elevations, peaking within seconds from neuronal stimulation (Hirase et al., 2004; Nimmerjahn et al., 2004; Wang et al., 2006). A recent study, however, shows for the first time the occurrence in vivo of fast astrocytic Ca2+ events in response to neuronal activity, peaking in the millisecond time scale from sensory stimulation (Winship et al., 2007). This finding suggests that the fast Ca2+ events here identified are not a peculiarity of astrocytes in cell culture but may correspond to events taking place in astrocytes of the living brain. Establishing whether fast Ca2+ events in vivo are associated to transmitter release via SLMV exocytosis becomes, therefore, of the outmost importance to define the type of modulatory influence exerted by astrocytes on neighboring neuronal circuits.
This work was supported by a Faculty of Biology and Medicine 2006 grant, University of Lausanne, Lausanne, Switzerland (P.B.). We thank R. Tsien for providing the mCherry construct and N. Demaurex for providing the ER-GFP; O. Mirante and P. Spagnuolo for experimental support; M. Giugliano, S. Vesce, V. Schubert, and M. Santello for scientific discussions; and J. Meldolesi, O. Kochubey, N. Korogod, and A. Volterra for comments and revisions on this manuscript.