Astrocyte-Derived ATP Induces Vesicle Shedding and IL-1 Release from Microglia1
http://www.100md.com
免疫学杂志 2005年第11期
Abstract
ATP has been indicated as a primary factor in microglial response to brain injury and inflammation. By acting on different purinergic receptors 2, ATP is known to induce chemotaxis and stimulate the release of several cytokines from these cells. The activation of purinergic receptors 2 in microglia can be triggered either by ATP deriving from dying cells, at sites of brain injury or by ATP released from astrocytes, in the absence of cell damage. By the use of a biochemical approach integrated with video microscopy experiments, we investigated the functional consequences triggered in microglia by ATP released from mechanically stimulated astrocytes, in mixed glial cocultures. Astrocyte-derived ATP induced in nearby microglia the formation and the shedding of membrane vesicles. Vesicle formation was inhibited by the ATP-degrading enzyme apyrase or by P2X7R antagonists. Isolation of shed vesicles, followed by IL-1 evaluation by a specific ELISA revealed the presence of the cytokine inside the vesicular organelles and its subsequent efflux into the extracellular medium. IL-1 efflux from shed vesicles was enhanced by ATP stimulation and inhibited by pretreatment with the P2X7 antagonist oxidized ATP, thus indicating a crucial involvement of the pore-forming P2X7R in the release of the cytokine. Our data identify astrocyte-derived ATP as the endogenous factor responsible for microvesicle shedding in microglia and reveal the mechanisms by which astrocyte-derived ATP triggers IL-1 release from these cells.
Introduction
Microglial cells play a major role in the inflammation processes which take place in the nervous tissue by secretion of specific cytokines. Among agents reported to promote cytokine release, ATP, acting at high concentrations on the ionotropic P2X7 purinergic receptors, is one of the most powerful stimuli for the processing and release of IL-1, the key initiator of acute inflammatory response (1). The physiological source of ATP needed to trigger cytokine release is still debated. Large amounts of the purine, accumulated extracellularly at the site of lesion, can directly activate P2X7-mediated cytokine release (2). In contrast, cytokine secretion at sites far away from damaged cells could proceed through the spreading of ATP-mediated calcium signals among astrocytes (3, 4). In a previous study we reported that ATP released by astrocytes during calcium wave propagation can activate P2X7Rs on microglial cells (3). Therefore, in principle, astrocyte-released ATP could be relevant for the control of cytokine secretion from microglia.
The signaling mechanism by which P2X7 activation regulates IL-1 release in microglia is not completely defined yet, although it involves activation of caspase-1, which proteolytically cleaves the biological inactive precursor of the cytokine (pro-IL-1), generating the mature, active form (2). Caspase-1 activation is promoted both by calcium release from intracellular stores (5) and by K+ efflux through P2X7Rs, which ultimately triggers phospholipase A2 activation (6, 7). Release of IL-1 also requires the up-regulation of pro-IL-1 gene expression, which is triggered by cell exposure to bacterial LPS (8). However the molecular mechanism by which bioactive IL-1 is released is still debated. IL-1 lacks a conventional secretory sequence and is known not to be released through the classical endoplasmic reticulum (ER)3-Golgi pathway (9). Alternative pathways of secretion have been proposed in different cell types, including simple cell lysis (10), fusion of endolysosome-related vesicles (9), bleb formation, and shedding of membrane vesicles (11). The latter mechanism, which commonly occurs in hemopoietic and immune cells (12), represents a process whereby signal molecules are released via membrane-enclosed particles into the microenvironment. Based on the finding that IL-1 is detected inside shed vesicles shortly after P2X7 activation and becomes detectable in the extracellular medium only at later time, it has been suggested that formation and shedding of membrane vesicles could represent a pathway for the release of the cytokine (11). More recently it has been suggested in murine macrophages that IL-1 secretion can be dissociated from P2X7-dependent formation of blebs at the plasma membrane, although a role of ATP-induced shed microvesicle in mediating IL-1 secretion has not been excluded (13). Because microglia has been indicated as the major source of IL-1 in the brain, it is of crucial importance to define the molecular mechanisms which mediate the cytokine release from these cells. Aims of the present study were to investigate 1) whether IL-1 is released from microglia similarly to monocytes, through the shedding of membrane vesicles, and 2) whether astrocyte-derived ATP can control the cytokine production from microglial cells. Our results indicate that ATP released from astrocytes modulates the cytokine release, which proceeds in microglia through the shedding of plasma membrane vesicles, containing packages of IL-1. Our data also suggest a role for the pore-forming P2X7Rs in the processing and release of the cytokine from shed vesicles into the extracellular medium.
Materials and Methods
Data analysis
The data are presented as means ± SE. Statistical significance was evaluated by the Student’s test or one-way ANOVA. Differences were considered significant if p < 0.05 and are indicated by an asterisk in all figures, whereas those at p < 0.01 are indicated by double asterisks.
Results
Microglia, but not hippocampal astrocytes, release IL-1 upon ATP exposure
Several evidences have been reported demonstrating that large amounts of exogenous ATP induce IL-1 processing and release from microglia, the cell type which serves as a major source of the cytokine in the CNS upon injury or inflammation (18). Less clear is the scenario in astrocytes, given that most of the evidence for IL-1 release from these cells derives from studies using mixed primary cultures, containing both astrocytes and microglia (19). To verify, in our experimental model, the relative contribution of microglia and astrocytes to IL-1 production, either N9 microglia cells or cultures of pure hippocampal astrocytes were stimulated for 30 min with 1 mM ATP after 6 h of LPS priming. Although both cell types express P2X7Rs and respond with intracellular calcium elevations to 100 μM BzATP (20, 21), an agonist rather selective for this receptor subtype, the mature form of IL-1 (17 kDa) was detected, by Western blot analysis, in the supernatant of N9 cells but not in the supernatant of primary astrocytes (Fig. 1A). In contrast, the immature form of the cytokine (36 kDa) was clearly detectable in both astrocyte and N9 homogenates (Fig. 1A). In the absence of ATP treatment, the mature form of IL-1 fails to be detected in N9 supernatant (data not shown). These data indicated that, at least in our experimental model, hippocampal astrocytes are not able to process and release IL-1 upon purinergic receptor 2 (P2R) activation. The ELISA analysis of IL-1 levels in the cell supernatants of N9 cells, primary purified microglia and primary hippocampal astrocytes further confirmed that ATP stimulation induces IL-1 release essentially from microglia (Fig. 1B). In the absence of ATP stimulation, IL-1 levels were below the assay sensitivity. Western blot analysis for CSF-1R and GFAP, specific markers of microglia and astrocytes (3) ruled out the possible contamination of astrocytic cultures with primary microglia (Fig. 1C). The kinetic analysis of IL-1 release from microglial N9 cells indicated that significant levels of the cytokine are present extracellularly 15–20 min after ATP stimulation. In line with previous results, removal of calcium ions from the extracellular medium by addition of 500 μM EDTA strongly prevented the release of the cytokine, thus suggesting a calcium-dependent mechanism of secretion (22) (Fig. 1D).
Exposure to exogenous ATP induces vesicle formation in microglial cells
To investigate whether ATP stimulation induces vesicle formation in microglia, video microscopy experiments were conducted upon 1 mM ATP treatment both in primary hippocampal microglia and in N9 cells. Bright field time-lapse microscopy revealed the formation of plasma membrane structures (blebs) continuously extruding and retracting from the cell membrane (Fig. 2A), 180 s after 1 mM ATP addition. As already described for platelets and monocytes (11, 23), in microglia ATP-induced blebs have phosphatidylserine (PS) exposed on the outer leaflet of the plasma membrane and bind annexin V, a high affinity ligand for the phospholipid. Annexin V-positive vesicle formation was confined to discrete regions of the cell surface (Fig. 2B) or more evenly distributed to the entire surface of the cells (see an extreme case in Fig. 2C). Vesicle formation proceeded even in the absence of extracellular calcium ions, although with a latency of 60 s as compared with controls (Fig. 2D); calcium chelation by EDTA did not prevent annexin V binding to the forming vesicles (number of annexin-positive cells per field increases by 4.44 ± 0.18-fold upon ATP treatment, by 2.9 ± 0.086 upon ATP treatment in the presence of EDTA). Pretreatment of N9 cells with 100 μM oATP or 100 nM Brilliant Blue G, two antagonists rather selective for the P2X7Rs (24, 25), quite completely inhibited vesicle formation, thus suggesting the requirement of P2X7R activation for ATP-induced bleb formation (Fig. 2D). ATP-induced vesicle formation also occurred in microglial cells not primed with LPS (data not shown).
After a brief ATP exposure (4–5 min) both N9 and primary microglia were able to recover the cell shape, indicating that bleb formation is not necessarily linked to an apoptotic fate of the cell. Accordingly, a negligible increase of LDH concentration was detected in medium collected from N9 cells exposed to ATP up to 20 min (increase of LDH concentration: 4.83 ± 0.89%). Furthermore, in line with previous results indicating that blebbing induced by ATP does not require activity of the apoptotic marker caspase-3 (13), no significant activation of caspase-3 was observed in N9 cells exposed to ATP up to 30 min. Caspase-3 activity became detectable in N9 cells only after 60 min of chronic ATP treatment (Fig. 2E), thus excluding that ATP-induced bleb formation, which takes place within 4–5 min from ATP addition, is necessarily linked to apoptotic cell death.
To better follow the dynamics of bleb formation in ATP-stimulated microglia we conducted fluorescence video microscopy on cells incubated with the fluorescent styryl dye FM1–43, which labels lipid bilayers. By this approach, we clearly observed not only the formation but also the shedding from the plasma membrane of vesicular structures of variable size (Fig. 3, A and B) (see also supplemental movie).4 To visualize vesicles shed into the extracellular medium, the supernatant collected from FM1–43-labeled cells exposed to ATP was deposed on glass coverslips. As shown in Fig. 3C collected supernatants contained a heterogeneous population of FM1–43-labeled vesicles (mean diameter value 980 nm, n = 780) (Fig. 3D). Pretreatment with 100 μM oATP or removal of calcium ions from the extracellular saline significantly reduced the number of shed vesicles in the supernatants (Fig. 3C). Accordingly the amount of fluorescence in N9 collected supernatants, due to the presence of FM1–43-labeled vesicles, was significantly reduced when cells were exposed to ATP in the absence of calcium ions and when cells were pretreated with oATP (Fig. 3E). Similar experiments were conducted on a clone of N9 cells, which stably express high levels of GFP in the cytosol. GFP-expressing N9 cells shed GFP-vesicles upon ATP stimulation. An example of GFP-vesicles are shown in Fig. 3F. Quantitative analysis of total green fluorescence of supernatants containing GFP-vesicles confirmed that vesicle shedding is markedly inhibited by calcium chelation (fluorescence intensity, arbitrary units: 18.83 ± 0.4 controls, 105.83 ± 2.35 ATP, 3.67 ± 0.2 ATP + EDTA, n = 3, p = 0.004, one-way ANOVA, post hoc Dunnett’s method) differently from ATP-induced bleb formation which appeared to be only slightly delayed by EDTA (Fig. 2D).
Shed vesicles contain IL-1
To verify whether vesicle formation could mediate IL-1 release from microglia, shed vesicles were isolated from the supernatant of LPS-primed N9 cells, 8–10 min after ATP stimulation, when low levels of IL-1 were detectable in the extracellular medium (Fig. 1D). Shed vesicles were isolated using annexin V-coated beads, permeabilized with 2% Triton X-100 to release vesicle content and assayed for the presence of IL-1. High levels of IL-1 were detected in shed vesicles (14.66 ± 3.5 pg/ml) while minimum amounts of the cytokine were observed in the vesicle-depleted medium (1.08 ± 0.14 pg/ml, n = 3, p = 0.008, Fig. 4A). The amount of IL-1 measured in Triton X-100-treated supernatant of N9 cells (i.e., total extracellular IL-1) at 10 min was 27% of the total cellular content, in agreement with data from Sanz and Di Virgilio (2) who reported a releasable fraction of 25%. To investigate whether the cytokine could be released from shed vesicles into the extracellular medium, isolated vesicles were maintained in KRH at 37°C for different time periods. The evaluation of IL-1 content of both vesicular and vesicle-free supernatant fractions indicated a progressive flow of the cytokine from shed vesicles into the extracellular medium (IL-1 in vesicle-free supernatant: 10 min = 1.25 ± 0.35 pg/ml corresponding to 9.36 ± 1.97% of total IL-1 present in the extracellular medium; 20 min = 11.5 ± 3.5 pg/ml corresponding to 27.37 ± 9.3% of total IL-1 30 min = 22.5 ± 2.5 pg/ml corresponding to 41.6 ± 9.5% of total IL-1, n = 3, Fig. 4B). Cytokine release from shed vesicles was enhanced when vesicles were stimulated with 1 mM ATP (IL-1 in vesicle-free medium: control = 22.5 ± 2.5 pg/ml, ATP = 42.99 ± 2.1 pg/ml) while it was significantly inhibited when vesicles were pretreated with 100 μM oATP (4.56 ± 0.56 pg/ml) or maintained in a calcium-free medium (6.36 ± 1.16 pg/ml), thus indicating the involvement of the pore-forming P2X7R in the release of the cytokine (n = 3, p = 0.016, one-way ANOVA, post hoc Dunnett’s method, Fig. 4C). Interestingly, IL-1 concentration in the extracellular medium was reduced even below control level after vesicle exposure to oATP. This may suggest a partial activation of P2X7R on shed vesicles, due to the previous ATP treatment needed to induce vesicle shedding. The gradual increase of IL-1 in the extracellular medium does not appear to result from progressive vesicle disruption, as indicated by the time course analysis of fluorescence in GFP-vesicle and GFP-vesicle-free medium, collected from N9 cells stably expressing cytosolic GFP (Fig. 4B). If GFP-vesicles broke with time, GFP would gradually flow into the vesicle-free supernatant. A similar percentage of GFP fluorescence was instead detected in vesicle-free fractions at all analyzed time points (16.8 ± 5.0, 10 min; 10 ± 5.3, 20 min; 12.8 ± 7.22, 30 min; Fig. 4B). Accordingly, no significant reduction of GFP fluorescence was detected at different time points in isolated vesicles (65.5 ± 4.1, 10 min; 67.5 ± 6.5, 20 min; 62 ± 5.16, 30 min). Western blot analysis revealed the presence of pro-IL-1 (Fig. 4E), procaspase-1 (Fig. 4F), and P2X7Rs (Fig. 4D) in vesicle fractions, solubilized immediately after isolation. Isolated vesicles were also stained for the ER marker ribophorin, for the plasma membrane protein SNAP-23 and for actin (Fig. 4D), in line with previous reports indicating that microvesicles and apoptotic blebs could carry a number of ER-derived proteins (26) and could contain cytoskeletal elements. Isolated vesicles were not immunoreactive for the endosomal (45 kDa) and lysosomal (31 kDa) forms of cathepsin D (Fig. 4D), for the endolysosomal markers rab-7 (Fig. 4D) and LAMP-1 (data not shown). Lack of cathepsin D and rab-7 staining in isolated vesicles but not in N9 homogenate, aside from indicating that isolated vesicles do not derive from lysosomal organelles, ruled out the possible contamination of isolated vesicles with apoptotic cell fragments (Fig. 4D). Interestingly, after 30 min of in vitro exposure to ATP, a processed band of both IL-1 and caspase-1 appeared in isolated vesicles (Fig. 4, E and F), suggesting that both precursors can be locally activated upon ATP stimulation (active-/pro-IL-1 ratio: 0.39 after 30 min of stimulation with 1 mM ATP, n = 3). In line with this finding, quantitative analysis of active caspase-1 by FLICA assay showed an increase in the number of active caspase-1 enzymes upon ATP stimulation in isolated vesicles as compared with vesicles incubated in vitro in the presence of oATP, after isolation (fluorescence arbitrary units: 601.85 ± 105.95, ATP; 147.1 ± 76.0, ATP + oATP, p = 0.01228, n = 2).
Altogether, our results indicate that shed vesicles, isolated from ATP-stimulated microglia cells, contain pro-IL-1, which is converted in the active cytokine upon ATP exposure and is progressively released into the external medium via a calcium and P2X7-dependent mechanism.
Astrocyte-derived ATP stimulates IL-1 release from microglial cells
To investigate whether ATP derived from astrocytes is sufficient to induce IL-1 release from adjacent microglial cells, we evaluated the presence of the cytokine in the supernatant of astrocyte-N9 cocultures. It has been previously shown that release of ATP from astrocytes is enhanced by mechanical stimulation (17, 27) and that astrocyte-derived ATP may activate P2Rs present in adjacent microglial cells (3, 4). To evaluate whether ATP release, induced by mechanical stimulation of astrocytes, triggers IL-1 secretion from adjacent microglia cells, astrocyte-N9 cocultures were mechanically stimulated on an orbital shaker, as previously described (17). ELISA analysis of collected media indicated a 3- to 4-fold increase of IL-1 concentration in the supernatant of mechanically stimulated cocultures (27.5 ± 0.5 pg/ml), as compared with supernatants from control cultures maintained under static conditions (7.25 ± 0.75 pg/ml) (Fig. 5A). Interestingly, IL-1 decreased even below control levels (0.5 ± 0.05 pg/ml) when mechanical stimulation was performed in the presence of the ATP-degrading enzyme apyrase (30 U/ml), which is known to significantly prevent the ATP-mediated astrocyte-to-microglia calcium signaling (3). IL-1 release from primary astrocyte-microglia cocultures was also markedly reduced when mechanical stimulation was performed in high extracellular concentration of K+ (IL-1, 27.5 ± 0.5 pg/ml, control medium; 4.5 ± 0.375 pg/ml, high K+ medium), thereby indicating that inhibition of P2X7-induced K+ efflux (28) by astrocyte-derived ATP does repress the paracrine activation of caspase-1 and of IL-1 release from microglia. IL-1 concentration in the supernatant of N9 cultures in the absence of astrocytes was significantly lower as compared with cocultures (1.00 ± 0.3 pg/ml) and the mechanical stimulation of pure N9 cultures only slightly increased IL-1 concentration in the supernatant (1.92 ± 0.5 pg/ml) (Fig. 5A). These data suggested that astrocyte-derived ATP, released either spontaneously or upon mechanical stimulation, induces IL-1 secretion from N9 cells (Fig. 5A). An almost negligible contribution to cytokine release seems to be due to ATP derived from microglia (Fig. 5A).
Astrocyte-derived ATP induces bleb formation in microglial cells
To evaluate whether ATP released from astrocytes could be sufficient to induce vesicle formation in microglial cells, bright field video microscopy experiments were conducted on cocultures of astrocytes and either primary microglia or N9 cells. Gentle mechanical contact between a patch pipette and the surface of a single astrocyte induced in adjacent microglia the formation of vesicular structures, which bound Annexin VFITC (Fig. 5B). The quantitative analysis of annexin V-positive cells was conducted in astrocyte-N9 cocultures, mechanically stimulated on an orbital shaker. Fig. 5C showed a 3-fold increase in the number of annexin-positive N9 cells upon mechanical stimulation. Binding of annexin V was not significantly enhanced in pure N9 cultures, upon mechanical stimulus, thus excluding that the mechanical stress per se or ATP secretion from microglial cells play a relevant role in vesicle formation (number of annexin V-positive cells per field, normalized to static coculture condition: 3.84 ± 0.28 cocultures, mechanical stimulation; 1.05 ± 0.28, cocultures, mechanical stimulation plus apyrase; 0.67 ± 0.06, N9 cells, static conditions; 0.82 ± 0.08, N9 cells, mechanical stimulation; one-way ANOVA, post hoc Dunn’s method, p = 0.006, n = 3 Fig. 5C). Note that the presence of apyrase during mechanical stimulation almost completely inhibited bleb formation (Fig. 5C), further supporting a key role of astrocyte-derived ATP in the observed blebbing phenomenon.
Discussion
It is currently assumed that, in the CNS, IL-1 is mainly expressed and released from microglia, although there is evidence that also astrocytes and neurons may contribute to IL-1 production, particularly in the late phase after excitotoxicity (18, 29).
In the present study we directly evaluated the specific contribution of astrocytes and microglia to the cytokine release. We demonstrate that 1) in hippocampal primary astrocyte-microglia cocultures, microglia is indeed the cell type which mainly releases IL-1, 2) ATP released from astrocytes is the endogenous stimulus which leads to the cytokine secretion from microglial cells, and 3) astrocyte-derived ATP causes IL-1 secretion by inducing the shedding from microglial surface of membrane vesicles which contain all the machinery necessary for the cytokine processing, including P2X7Rs.
Pro-IL-1 is also expressed by primary hippocampal astrocytes, primed with LPS, although these cells fail to release the cytokine upon ATP stimulation. Because both astrocytes and microglia bear P2X7Rs (30, 31), which are essential for the cytokine release (2) further studies will be required to elucidate the molecular mechanisms underlying the different behavior of hippocampal astrocytes and microglia regarding IL-1 secretion. Lack of IL-1 release from astrocytes might result from differences in caspase-1 activation and/or cytokine processing, possibly due to insufficient intracellular Ca2+ concentration increases and K+ drop, following P2X7 activation (F. Bianco and C. Verderio, unpublished observations).
The main finding of this study is that the release of at least a fraction of IL-1 from microglia takes place, upon ATP stimulation, through the shedding of vesicles from the plasma membrane. Shed vesicles appear to be the site of IL-1 processing.
Different populations of membrane vesicles which can be released from cells in the extracellular environment have been described so far, including exosomes, microvesicles, and apoptotic vesicles. Exosomes originate from exocytosis of endolysosome-related multivesicular bodies and represent a population of vesicles homogenous in size and shape (40–80 nm) (32). Microvesicles and apoptotic vesicles represent instead a heterogeneous population of vesicles relatively larger than exosomes (100 nm to 1 μm) which bud directly from the plasma membrane of either viable cells or cells undergoing apoptosis (26). Exosomes and microvesicles/apoptotic vesicles are biochemically distinct; the latter population have PS exposed to the outer cell membrane (11, 26) carry some cytoskeletal elements (26) and can bear both nuclear and ER proteins (33, 34). Exosomes instead do not have PS exposed to the outer membrane leaflet and carry a selected subset of protein including rab-7 (26, 35).
Our video microscopy data indicate that ATP stimulation induces in microglia the shedding of membrane vesicles of heterogeneous diameter, which detach from the membrane in the presence of extracellular calcium. Because calcium chelation slightly delayed formation of ATP-induced membrane blebs, but markedly inhibited vesicle shedding, our results seem to dissociate the P2X7-dependent formation of membrane blebs from the release of vesicles. Bleb formation and vesicle shedding appear to be distinct parallel pathways both initiated by P2X7 activation. Shed vesicles carry the plasma membrane soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor (SNARE) protein SNAP-23, have PS exposed to the outer cell membrane, do not express lysosomal markers, including rab-7, lamp-1, and cathepsin D, but bear the ER marker ribophorin and some actin. Based on biochemical and morphological analysis, vesicles isolated from the supernatant of ATP-stimulated microglia have all the features of microvesicles/apoptotic vesicles. Given that microglia are able to recover from ATP challenge and do not necessarily undergo apoptosis, and because no caspase-3 activity is detectable inside isolated vesicles (F. Bianco and C. Verderio, unpublished observations), we exclude these formations to be apoptotic vesicles. Accordingly, it has been reported that caspase-3 activation is not a necessary prerequisite for vesicle formation induced by ATP, although it plays a necessary role in blebbing induced by apoptosis (13).
The experiments described in this study indicate that shed vesicles contain pro-IL-1 which is cleaved in the active form and released into the extracellular environment in an ATP- and P2X7-dependent manner. These findings strongly suggest that IL-1 processing could proceed in shed vesicles, which indeed contain caspase-1 both as proenzyme and as active form. The presence of caspase-1 activity in shed vesicles is in agreement with previous reports showing that these vesicles act as carriers of proteases (36, 37), which can locally process their substrates (38). Our results point to a crucial involvement of the P2X7Rs in the mechanism mediating the cytokine processing and release from shed vesicles, although the mechanism underlying the efflux of the cytokine from the vesicles remains to be defined. Vesicle disruption does not significantly contribute to the release of the cytokine, differently from what is described in fibroblasts where vesicle lysis mediates the release of fibroblast growth factor-2, another protein lacking a conventional secretory peptide which is released by vesicle shedding (32).
Globally, our data support and complement results obtained in monocytes by MacKenzie et al. (11), which indicated the existence of a vesicular pathway for IL-1 release by vesicle shedding. However, the observation that vesicle shedding can occur in resting microglia, which does not release IL-1, supports the idea that vesicle shedding does not represent a secretory mechanism selective for the cytokine, but rather a secretory pathway possibly shared by other releasable factors, which is used by IL-1 when produced in activated microglia. Given that P2X7R activation in microglia triggers the synthesis and release of other cytokines like TNF (39) and proteases as well as their precursors, including plasminogen (40), it would be of interest to investigate whether ATP-induced shed vesicles can also mediate the release of such factors. Furthermore, vesicle shedding, through the release of cell surface components, including membrane receptors (41, 42, 43), can serve functions other than secretion of soluble factors. In this regard, we clearly demonstrate the presence of P2X7Rs on isolated vesicles, thus indicating that vesicles shedding might represent a mechanism to remove P2X7Rs from the cell surface in response to astrocyte-derived ATP signaling. From this point of view the shedding process could represent a defense strategy of microglia against apoptotic insults, produced by excessive or repetitive stimulation by astrocyte-derived ATP. Removal from the cell surface of functional P2X7Rs could facilitate microglial survival and avoid P2X7-mediated microglial apoptosis (3).
Based on the amount of IL-1 detected in Triton X-100-treated supernatant at 10-min stimulation, when negligible levels of the cytokine are present in vesicle-free supernatant, we estimated the fraction releasable by vesicle shedding to be 27% of total cellular content. This estimation is in agreement with results from a previous study (2), which reported a releasable IL-1 fraction of 25% in these cells. Based on these findings, one could speculate that vesicle shedding may be the predominant mechanism of IL-1 secretion in microglia. However, different pathways of IL-1 release might coexist in these cells (Fig. 6). This possibility has been already described for monocytes, where vesicle shedding (11) and exocytosis of endolysosome-related vesicles (6) have been shown to account for 12 and 30% of total IL-1 cellular content, respectively. Further studies are needed to determine the possible contribution of the endolysosome pathway to IL-1 secretion in microglia, although the lack of inhibition of IL-1 release by AAC0CF3 (F. Bianco and C. Verderio, unpublished observations), a PLA2 blocker which strongly inhibits the exocytosis of secretory lysosomes (6), plays against a major contribution of the endolysosomal pathway in microglia.
Although vesicle shedding is most likely a nonspecific and nonunique pathway for IL-1 release, it is extremely important to understand the physiological relevance of IL-1 release from microglial cells by this mechanism. Shed vesicles, containing packages of IL-1, can deliver the cytokine in possible proximity to IL-1Rs present on target cells. Furthermore, given that IL-1 release is enhanced by ATP, the cytokine might be released from shed vesicles when they impact local high concentration of the purine, presumably in close proximity of releasing astrocytes, thus avoiding dilution of the cytokine in the extracellular environment. This might also turn out to be of crucial importance for the onset of a detrimental effect of IL-1 toward degenerating cells, which are known to release large amounts of ATP. Therefore, IL-1 release from microglia can represent a backward signal sent by microglial cells in response to ATP-mediated astrocyte-to-microglia signaling.
Acknowledgments
We thank Dr. Paola Castagnoli-Ricciardi (University of Milano-Bicocca, Milan, Italy) for generously providing N9 microglial cells. We thank Dr. Elisabetta Menna (Consiglio Nazionale delle Ricerche-Institute of Neuroscience, Milan, Italy) for helpful comments and discussion.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This research was supported by Fondo per gli Investimenti della Ricerca di Base Grants RBAUO19 ZEN and RBNE01ERXR_007, European Community Grant QLG 3-2000-01343, Ministero dell’Istruzione dell’Università e della Ricerca-Programmi di Ricerca Scientifica di Rilevante Interesse Nazionale 2002, and Fondazione Italiana Sclerosi Multipla Grant 2003/R/35.
2 Address correspondence and reprint requests to Dr. Claudia Verderio, Consiglio Nazionale delle Ricerche-Institute of Neuroscience, Via Vanvitelli 32, 20129 Milano, Italy. E-mail address: c.verderio{at}in.cnr.it
3 Abbreviations used in this paper: ER, endoplasmic reticulum; GFAP, glial fibrillar acidic protein; LDH, lactate dehydrogenase; oATP, oxidized ATP; PS, phosphatidylserine; P2Rs, purinergic receptor 2; FLICA, fluorochrome inhibitor of caspases; SNAP-23, synaptosomal-associated protein of 23 kDa.
4 The online version of this article contains supplemental material.
Received for publication August 2, 2004. Accepted for publication March 15, 2005.
References
Solle, M., J. Labasi, D. G. Perregaux, E. Stam, N. Petrushova, B. H. Koller, R. J. Griffiths, C. A. Gabel. 2001. Altered cytokine production in mice lacking P2X7 receptors. J. Biol. Chem. 276: 125-132.
Sanz, J. M., F. Di Virgilio. 2000. Kinetics and mechanism of ATP-dependent IL-1 release from microglial cells. J. Immunol. 164: 4893-4898.
Verderio, C., M. Matteoli. 2001. ATP mediates calcium signaling between astrocytes and microglial cells: modulation by IFN-. J. Immunol. 166: 6383-6391.
Schipke, C. G., C. Boucsein, C. Ohlemeyer, F. Kirchhoff, H. Kettenmann. 2002. Astrocyte Ca2+ waves trigger responses in microglial cells in brain slices. FASEB J. 16: 255-257.
Brough, D., R. A. Le Feuvre, R. D. Wheeler, N. Solovyova, S. Hilfiker, N. J. Rothwell, A. Verkhratsky. 2003. Ca2+ stores and Ca2+ entry differentially contribute to the release of IL-1 and IL-1 from murine macrophages. J. Immunol. 170: 3029-3036.
Andrei, C., P. Margiocco, A. Poggi, L. V. Lotti, M. R. Torrisi, A. Rubartelli. 2004. Phospholipases C and A2 control lysosome-mediated IL-1 secretion: implications for inflammatory processes. Proc. Natl. Acad. Sci. USA 101: 9745-9750.
Walev, I., J. Klein, M. Husmann, A. Valeva, S. Strauch, H. Wirtz, O. Weichel, S. Bhakdi. 2000. Potassium regulates IL-1 processing via calcium-independent phospholipase A2. J. Immunol. 164: 5120-5124.
Chauvet, N., K. Palin, D. Verrier, S. Poole, R. Dantzer, J. Lestage. 2001. Rat microglial cells secrete predominantly the precursor of interleukin-1 in response to lipopolysaccharide. Eur. J. Neurosci. 14: 609-617.(Fabio Bianco*,, Elena Pra)
ATP has been indicated as a primary factor in microglial response to brain injury and inflammation. By acting on different purinergic receptors 2, ATP is known to induce chemotaxis and stimulate the release of several cytokines from these cells. The activation of purinergic receptors 2 in microglia can be triggered either by ATP deriving from dying cells, at sites of brain injury or by ATP released from astrocytes, in the absence of cell damage. By the use of a biochemical approach integrated with video microscopy experiments, we investigated the functional consequences triggered in microglia by ATP released from mechanically stimulated astrocytes, in mixed glial cocultures. Astrocyte-derived ATP induced in nearby microglia the formation and the shedding of membrane vesicles. Vesicle formation was inhibited by the ATP-degrading enzyme apyrase or by P2X7R antagonists. Isolation of shed vesicles, followed by IL-1 evaluation by a specific ELISA revealed the presence of the cytokine inside the vesicular organelles and its subsequent efflux into the extracellular medium. IL-1 efflux from shed vesicles was enhanced by ATP stimulation and inhibited by pretreatment with the P2X7 antagonist oxidized ATP, thus indicating a crucial involvement of the pore-forming P2X7R in the release of the cytokine. Our data identify astrocyte-derived ATP as the endogenous factor responsible for microvesicle shedding in microglia and reveal the mechanisms by which astrocyte-derived ATP triggers IL-1 release from these cells.
Introduction
Microglial cells play a major role in the inflammation processes which take place in the nervous tissue by secretion of specific cytokines. Among agents reported to promote cytokine release, ATP, acting at high concentrations on the ionotropic P2X7 purinergic receptors, is one of the most powerful stimuli for the processing and release of IL-1, the key initiator of acute inflammatory response (1). The physiological source of ATP needed to trigger cytokine release is still debated. Large amounts of the purine, accumulated extracellularly at the site of lesion, can directly activate P2X7-mediated cytokine release (2). In contrast, cytokine secretion at sites far away from damaged cells could proceed through the spreading of ATP-mediated calcium signals among astrocytes (3, 4). In a previous study we reported that ATP released by astrocytes during calcium wave propagation can activate P2X7Rs on microglial cells (3). Therefore, in principle, astrocyte-released ATP could be relevant for the control of cytokine secretion from microglia.
The signaling mechanism by which P2X7 activation regulates IL-1 release in microglia is not completely defined yet, although it involves activation of caspase-1, which proteolytically cleaves the biological inactive precursor of the cytokine (pro-IL-1), generating the mature, active form (2). Caspase-1 activation is promoted both by calcium release from intracellular stores (5) and by K+ efflux through P2X7Rs, which ultimately triggers phospholipase A2 activation (6, 7). Release of IL-1 also requires the up-regulation of pro-IL-1 gene expression, which is triggered by cell exposure to bacterial LPS (8). However the molecular mechanism by which bioactive IL-1 is released is still debated. IL-1 lacks a conventional secretory sequence and is known not to be released through the classical endoplasmic reticulum (ER)3-Golgi pathway (9). Alternative pathways of secretion have been proposed in different cell types, including simple cell lysis (10), fusion of endolysosome-related vesicles (9), bleb formation, and shedding of membrane vesicles (11). The latter mechanism, which commonly occurs in hemopoietic and immune cells (12), represents a process whereby signal molecules are released via membrane-enclosed particles into the microenvironment. Based on the finding that IL-1 is detected inside shed vesicles shortly after P2X7 activation and becomes detectable in the extracellular medium only at later time, it has been suggested that formation and shedding of membrane vesicles could represent a pathway for the release of the cytokine (11). More recently it has been suggested in murine macrophages that IL-1 secretion can be dissociated from P2X7-dependent formation of blebs at the plasma membrane, although a role of ATP-induced shed microvesicle in mediating IL-1 secretion has not been excluded (13). Because microglia has been indicated as the major source of IL-1 in the brain, it is of crucial importance to define the molecular mechanisms which mediate the cytokine release from these cells. Aims of the present study were to investigate 1) whether IL-1 is released from microglia similarly to monocytes, through the shedding of membrane vesicles, and 2) whether astrocyte-derived ATP can control the cytokine production from microglial cells. Our results indicate that ATP released from astrocytes modulates the cytokine release, which proceeds in microglia through the shedding of plasma membrane vesicles, containing packages of IL-1. Our data also suggest a role for the pore-forming P2X7Rs in the processing and release of the cytokine from shed vesicles into the extracellular medium.
Materials and Methods
Data analysis
The data are presented as means ± SE. Statistical significance was evaluated by the Student’s test or one-way ANOVA. Differences were considered significant if p < 0.05 and are indicated by an asterisk in all figures, whereas those at p < 0.01 are indicated by double asterisks.
Results
Microglia, but not hippocampal astrocytes, release IL-1 upon ATP exposure
Several evidences have been reported demonstrating that large amounts of exogenous ATP induce IL-1 processing and release from microglia, the cell type which serves as a major source of the cytokine in the CNS upon injury or inflammation (18). Less clear is the scenario in astrocytes, given that most of the evidence for IL-1 release from these cells derives from studies using mixed primary cultures, containing both astrocytes and microglia (19). To verify, in our experimental model, the relative contribution of microglia and astrocytes to IL-1 production, either N9 microglia cells or cultures of pure hippocampal astrocytes were stimulated for 30 min with 1 mM ATP after 6 h of LPS priming. Although both cell types express P2X7Rs and respond with intracellular calcium elevations to 100 μM BzATP (20, 21), an agonist rather selective for this receptor subtype, the mature form of IL-1 (17 kDa) was detected, by Western blot analysis, in the supernatant of N9 cells but not in the supernatant of primary astrocytes (Fig. 1A). In contrast, the immature form of the cytokine (36 kDa) was clearly detectable in both astrocyte and N9 homogenates (Fig. 1A). In the absence of ATP treatment, the mature form of IL-1 fails to be detected in N9 supernatant (data not shown). These data indicated that, at least in our experimental model, hippocampal astrocytes are not able to process and release IL-1 upon purinergic receptor 2 (P2R) activation. The ELISA analysis of IL-1 levels in the cell supernatants of N9 cells, primary purified microglia and primary hippocampal astrocytes further confirmed that ATP stimulation induces IL-1 release essentially from microglia (Fig. 1B). In the absence of ATP stimulation, IL-1 levels were below the assay sensitivity. Western blot analysis for CSF-1R and GFAP, specific markers of microglia and astrocytes (3) ruled out the possible contamination of astrocytic cultures with primary microglia (Fig. 1C). The kinetic analysis of IL-1 release from microglial N9 cells indicated that significant levels of the cytokine are present extracellularly 15–20 min after ATP stimulation. In line with previous results, removal of calcium ions from the extracellular medium by addition of 500 μM EDTA strongly prevented the release of the cytokine, thus suggesting a calcium-dependent mechanism of secretion (22) (Fig. 1D).
Exposure to exogenous ATP induces vesicle formation in microglial cells
To investigate whether ATP stimulation induces vesicle formation in microglia, video microscopy experiments were conducted upon 1 mM ATP treatment both in primary hippocampal microglia and in N9 cells. Bright field time-lapse microscopy revealed the formation of plasma membrane structures (blebs) continuously extruding and retracting from the cell membrane (Fig. 2A), 180 s after 1 mM ATP addition. As already described for platelets and monocytes (11, 23), in microglia ATP-induced blebs have phosphatidylserine (PS) exposed on the outer leaflet of the plasma membrane and bind annexin V, a high affinity ligand for the phospholipid. Annexin V-positive vesicle formation was confined to discrete regions of the cell surface (Fig. 2B) or more evenly distributed to the entire surface of the cells (see an extreme case in Fig. 2C). Vesicle formation proceeded even in the absence of extracellular calcium ions, although with a latency of 60 s as compared with controls (Fig. 2D); calcium chelation by EDTA did not prevent annexin V binding to the forming vesicles (number of annexin-positive cells per field increases by 4.44 ± 0.18-fold upon ATP treatment, by 2.9 ± 0.086 upon ATP treatment in the presence of EDTA). Pretreatment of N9 cells with 100 μM oATP or 100 nM Brilliant Blue G, two antagonists rather selective for the P2X7Rs (24, 25), quite completely inhibited vesicle formation, thus suggesting the requirement of P2X7R activation for ATP-induced bleb formation (Fig. 2D). ATP-induced vesicle formation also occurred in microglial cells not primed with LPS (data not shown).
After a brief ATP exposure (4–5 min) both N9 and primary microglia were able to recover the cell shape, indicating that bleb formation is not necessarily linked to an apoptotic fate of the cell. Accordingly, a negligible increase of LDH concentration was detected in medium collected from N9 cells exposed to ATP up to 20 min (increase of LDH concentration: 4.83 ± 0.89%). Furthermore, in line with previous results indicating that blebbing induced by ATP does not require activity of the apoptotic marker caspase-3 (13), no significant activation of caspase-3 was observed in N9 cells exposed to ATP up to 30 min. Caspase-3 activity became detectable in N9 cells only after 60 min of chronic ATP treatment (Fig. 2E), thus excluding that ATP-induced bleb formation, which takes place within 4–5 min from ATP addition, is necessarily linked to apoptotic cell death.
To better follow the dynamics of bleb formation in ATP-stimulated microglia we conducted fluorescence video microscopy on cells incubated with the fluorescent styryl dye FM1–43, which labels lipid bilayers. By this approach, we clearly observed not only the formation but also the shedding from the plasma membrane of vesicular structures of variable size (Fig. 3, A and B) (see also supplemental movie).4 To visualize vesicles shed into the extracellular medium, the supernatant collected from FM1–43-labeled cells exposed to ATP was deposed on glass coverslips. As shown in Fig. 3C collected supernatants contained a heterogeneous population of FM1–43-labeled vesicles (mean diameter value 980 nm, n = 780) (Fig. 3D). Pretreatment with 100 μM oATP or removal of calcium ions from the extracellular saline significantly reduced the number of shed vesicles in the supernatants (Fig. 3C). Accordingly the amount of fluorescence in N9 collected supernatants, due to the presence of FM1–43-labeled vesicles, was significantly reduced when cells were exposed to ATP in the absence of calcium ions and when cells were pretreated with oATP (Fig. 3E). Similar experiments were conducted on a clone of N9 cells, which stably express high levels of GFP in the cytosol. GFP-expressing N9 cells shed GFP-vesicles upon ATP stimulation. An example of GFP-vesicles are shown in Fig. 3F. Quantitative analysis of total green fluorescence of supernatants containing GFP-vesicles confirmed that vesicle shedding is markedly inhibited by calcium chelation (fluorescence intensity, arbitrary units: 18.83 ± 0.4 controls, 105.83 ± 2.35 ATP, 3.67 ± 0.2 ATP + EDTA, n = 3, p = 0.004, one-way ANOVA, post hoc Dunnett’s method) differently from ATP-induced bleb formation which appeared to be only slightly delayed by EDTA (Fig. 2D).
Shed vesicles contain IL-1
To verify whether vesicle formation could mediate IL-1 release from microglia, shed vesicles were isolated from the supernatant of LPS-primed N9 cells, 8–10 min after ATP stimulation, when low levels of IL-1 were detectable in the extracellular medium (Fig. 1D). Shed vesicles were isolated using annexin V-coated beads, permeabilized with 2% Triton X-100 to release vesicle content and assayed for the presence of IL-1. High levels of IL-1 were detected in shed vesicles (14.66 ± 3.5 pg/ml) while minimum amounts of the cytokine were observed in the vesicle-depleted medium (1.08 ± 0.14 pg/ml, n = 3, p = 0.008, Fig. 4A). The amount of IL-1 measured in Triton X-100-treated supernatant of N9 cells (i.e., total extracellular IL-1) at 10 min was 27% of the total cellular content, in agreement with data from Sanz and Di Virgilio (2) who reported a releasable fraction of 25%. To investigate whether the cytokine could be released from shed vesicles into the extracellular medium, isolated vesicles were maintained in KRH at 37°C for different time periods. The evaluation of IL-1 content of both vesicular and vesicle-free supernatant fractions indicated a progressive flow of the cytokine from shed vesicles into the extracellular medium (IL-1 in vesicle-free supernatant: 10 min = 1.25 ± 0.35 pg/ml corresponding to 9.36 ± 1.97% of total IL-1 present in the extracellular medium; 20 min = 11.5 ± 3.5 pg/ml corresponding to 27.37 ± 9.3% of total IL-1 30 min = 22.5 ± 2.5 pg/ml corresponding to 41.6 ± 9.5% of total IL-1, n = 3, Fig. 4B). Cytokine release from shed vesicles was enhanced when vesicles were stimulated with 1 mM ATP (IL-1 in vesicle-free medium: control = 22.5 ± 2.5 pg/ml, ATP = 42.99 ± 2.1 pg/ml) while it was significantly inhibited when vesicles were pretreated with 100 μM oATP (4.56 ± 0.56 pg/ml) or maintained in a calcium-free medium (6.36 ± 1.16 pg/ml), thus indicating the involvement of the pore-forming P2X7R in the release of the cytokine (n = 3, p = 0.016, one-way ANOVA, post hoc Dunnett’s method, Fig. 4C). Interestingly, IL-1 concentration in the extracellular medium was reduced even below control level after vesicle exposure to oATP. This may suggest a partial activation of P2X7R on shed vesicles, due to the previous ATP treatment needed to induce vesicle shedding. The gradual increase of IL-1 in the extracellular medium does not appear to result from progressive vesicle disruption, as indicated by the time course analysis of fluorescence in GFP-vesicle and GFP-vesicle-free medium, collected from N9 cells stably expressing cytosolic GFP (Fig. 4B). If GFP-vesicles broke with time, GFP would gradually flow into the vesicle-free supernatant. A similar percentage of GFP fluorescence was instead detected in vesicle-free fractions at all analyzed time points (16.8 ± 5.0, 10 min; 10 ± 5.3, 20 min; 12.8 ± 7.22, 30 min; Fig. 4B). Accordingly, no significant reduction of GFP fluorescence was detected at different time points in isolated vesicles (65.5 ± 4.1, 10 min; 67.5 ± 6.5, 20 min; 62 ± 5.16, 30 min). Western blot analysis revealed the presence of pro-IL-1 (Fig. 4E), procaspase-1 (Fig. 4F), and P2X7Rs (Fig. 4D) in vesicle fractions, solubilized immediately after isolation. Isolated vesicles were also stained for the ER marker ribophorin, for the plasma membrane protein SNAP-23 and for actin (Fig. 4D), in line with previous reports indicating that microvesicles and apoptotic blebs could carry a number of ER-derived proteins (26) and could contain cytoskeletal elements. Isolated vesicles were not immunoreactive for the endosomal (45 kDa) and lysosomal (31 kDa) forms of cathepsin D (Fig. 4D), for the endolysosomal markers rab-7 (Fig. 4D) and LAMP-1 (data not shown). Lack of cathepsin D and rab-7 staining in isolated vesicles but not in N9 homogenate, aside from indicating that isolated vesicles do not derive from lysosomal organelles, ruled out the possible contamination of isolated vesicles with apoptotic cell fragments (Fig. 4D). Interestingly, after 30 min of in vitro exposure to ATP, a processed band of both IL-1 and caspase-1 appeared in isolated vesicles (Fig. 4, E and F), suggesting that both precursors can be locally activated upon ATP stimulation (active-/pro-IL-1 ratio: 0.39 after 30 min of stimulation with 1 mM ATP, n = 3). In line with this finding, quantitative analysis of active caspase-1 by FLICA assay showed an increase in the number of active caspase-1 enzymes upon ATP stimulation in isolated vesicles as compared with vesicles incubated in vitro in the presence of oATP, after isolation (fluorescence arbitrary units: 601.85 ± 105.95, ATP; 147.1 ± 76.0, ATP + oATP, p = 0.01228, n = 2).
Altogether, our results indicate that shed vesicles, isolated from ATP-stimulated microglia cells, contain pro-IL-1, which is converted in the active cytokine upon ATP exposure and is progressively released into the external medium via a calcium and P2X7-dependent mechanism.
Astrocyte-derived ATP stimulates IL-1 release from microglial cells
To investigate whether ATP derived from astrocytes is sufficient to induce IL-1 release from adjacent microglial cells, we evaluated the presence of the cytokine in the supernatant of astrocyte-N9 cocultures. It has been previously shown that release of ATP from astrocytes is enhanced by mechanical stimulation (17, 27) and that astrocyte-derived ATP may activate P2Rs present in adjacent microglial cells (3, 4). To evaluate whether ATP release, induced by mechanical stimulation of astrocytes, triggers IL-1 secretion from adjacent microglia cells, astrocyte-N9 cocultures were mechanically stimulated on an orbital shaker, as previously described (17). ELISA analysis of collected media indicated a 3- to 4-fold increase of IL-1 concentration in the supernatant of mechanically stimulated cocultures (27.5 ± 0.5 pg/ml), as compared with supernatants from control cultures maintained under static conditions (7.25 ± 0.75 pg/ml) (Fig. 5A). Interestingly, IL-1 decreased even below control levels (0.5 ± 0.05 pg/ml) when mechanical stimulation was performed in the presence of the ATP-degrading enzyme apyrase (30 U/ml), which is known to significantly prevent the ATP-mediated astrocyte-to-microglia calcium signaling (3). IL-1 release from primary astrocyte-microglia cocultures was also markedly reduced when mechanical stimulation was performed in high extracellular concentration of K+ (IL-1, 27.5 ± 0.5 pg/ml, control medium; 4.5 ± 0.375 pg/ml, high K+ medium), thereby indicating that inhibition of P2X7-induced K+ efflux (28) by astrocyte-derived ATP does repress the paracrine activation of caspase-1 and of IL-1 release from microglia. IL-1 concentration in the supernatant of N9 cultures in the absence of astrocytes was significantly lower as compared with cocultures (1.00 ± 0.3 pg/ml) and the mechanical stimulation of pure N9 cultures only slightly increased IL-1 concentration in the supernatant (1.92 ± 0.5 pg/ml) (Fig. 5A). These data suggested that astrocyte-derived ATP, released either spontaneously or upon mechanical stimulation, induces IL-1 secretion from N9 cells (Fig. 5A). An almost negligible contribution to cytokine release seems to be due to ATP derived from microglia (Fig. 5A).
Astrocyte-derived ATP induces bleb formation in microglial cells
To evaluate whether ATP released from astrocytes could be sufficient to induce vesicle formation in microglial cells, bright field video microscopy experiments were conducted on cocultures of astrocytes and either primary microglia or N9 cells. Gentle mechanical contact between a patch pipette and the surface of a single astrocyte induced in adjacent microglia the formation of vesicular structures, which bound Annexin VFITC (Fig. 5B). The quantitative analysis of annexin V-positive cells was conducted in astrocyte-N9 cocultures, mechanically stimulated on an orbital shaker. Fig. 5C showed a 3-fold increase in the number of annexin-positive N9 cells upon mechanical stimulation. Binding of annexin V was not significantly enhanced in pure N9 cultures, upon mechanical stimulus, thus excluding that the mechanical stress per se or ATP secretion from microglial cells play a relevant role in vesicle formation (number of annexin V-positive cells per field, normalized to static coculture condition: 3.84 ± 0.28 cocultures, mechanical stimulation; 1.05 ± 0.28, cocultures, mechanical stimulation plus apyrase; 0.67 ± 0.06, N9 cells, static conditions; 0.82 ± 0.08, N9 cells, mechanical stimulation; one-way ANOVA, post hoc Dunn’s method, p = 0.006, n = 3 Fig. 5C). Note that the presence of apyrase during mechanical stimulation almost completely inhibited bleb formation (Fig. 5C), further supporting a key role of astrocyte-derived ATP in the observed blebbing phenomenon.
Discussion
It is currently assumed that, in the CNS, IL-1 is mainly expressed and released from microglia, although there is evidence that also astrocytes and neurons may contribute to IL-1 production, particularly in the late phase after excitotoxicity (18, 29).
In the present study we directly evaluated the specific contribution of astrocytes and microglia to the cytokine release. We demonstrate that 1) in hippocampal primary astrocyte-microglia cocultures, microglia is indeed the cell type which mainly releases IL-1, 2) ATP released from astrocytes is the endogenous stimulus which leads to the cytokine secretion from microglial cells, and 3) astrocyte-derived ATP causes IL-1 secretion by inducing the shedding from microglial surface of membrane vesicles which contain all the machinery necessary for the cytokine processing, including P2X7Rs.
Pro-IL-1 is also expressed by primary hippocampal astrocytes, primed with LPS, although these cells fail to release the cytokine upon ATP stimulation. Because both astrocytes and microglia bear P2X7Rs (30, 31), which are essential for the cytokine release (2) further studies will be required to elucidate the molecular mechanisms underlying the different behavior of hippocampal astrocytes and microglia regarding IL-1 secretion. Lack of IL-1 release from astrocytes might result from differences in caspase-1 activation and/or cytokine processing, possibly due to insufficient intracellular Ca2+ concentration increases and K+ drop, following P2X7 activation (F. Bianco and C. Verderio, unpublished observations).
The main finding of this study is that the release of at least a fraction of IL-1 from microglia takes place, upon ATP stimulation, through the shedding of vesicles from the plasma membrane. Shed vesicles appear to be the site of IL-1 processing.
Different populations of membrane vesicles which can be released from cells in the extracellular environment have been described so far, including exosomes, microvesicles, and apoptotic vesicles. Exosomes originate from exocytosis of endolysosome-related multivesicular bodies and represent a population of vesicles homogenous in size and shape (40–80 nm) (32). Microvesicles and apoptotic vesicles represent instead a heterogeneous population of vesicles relatively larger than exosomes (100 nm to 1 μm) which bud directly from the plasma membrane of either viable cells or cells undergoing apoptosis (26). Exosomes and microvesicles/apoptotic vesicles are biochemically distinct; the latter population have PS exposed to the outer cell membrane (11, 26) carry some cytoskeletal elements (26) and can bear both nuclear and ER proteins (33, 34). Exosomes instead do not have PS exposed to the outer membrane leaflet and carry a selected subset of protein including rab-7 (26, 35).
Our video microscopy data indicate that ATP stimulation induces in microglia the shedding of membrane vesicles of heterogeneous diameter, which detach from the membrane in the presence of extracellular calcium. Because calcium chelation slightly delayed formation of ATP-induced membrane blebs, but markedly inhibited vesicle shedding, our results seem to dissociate the P2X7-dependent formation of membrane blebs from the release of vesicles. Bleb formation and vesicle shedding appear to be distinct parallel pathways both initiated by P2X7 activation. Shed vesicles carry the plasma membrane soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor (SNARE) protein SNAP-23, have PS exposed to the outer cell membrane, do not express lysosomal markers, including rab-7, lamp-1, and cathepsin D, but bear the ER marker ribophorin and some actin. Based on biochemical and morphological analysis, vesicles isolated from the supernatant of ATP-stimulated microglia have all the features of microvesicles/apoptotic vesicles. Given that microglia are able to recover from ATP challenge and do not necessarily undergo apoptosis, and because no caspase-3 activity is detectable inside isolated vesicles (F. Bianco and C. Verderio, unpublished observations), we exclude these formations to be apoptotic vesicles. Accordingly, it has been reported that caspase-3 activation is not a necessary prerequisite for vesicle formation induced by ATP, although it plays a necessary role in blebbing induced by apoptosis (13).
The experiments described in this study indicate that shed vesicles contain pro-IL-1 which is cleaved in the active form and released into the extracellular environment in an ATP- and P2X7-dependent manner. These findings strongly suggest that IL-1 processing could proceed in shed vesicles, which indeed contain caspase-1 both as proenzyme and as active form. The presence of caspase-1 activity in shed vesicles is in agreement with previous reports showing that these vesicles act as carriers of proteases (36, 37), which can locally process their substrates (38). Our results point to a crucial involvement of the P2X7Rs in the mechanism mediating the cytokine processing and release from shed vesicles, although the mechanism underlying the efflux of the cytokine from the vesicles remains to be defined. Vesicle disruption does not significantly contribute to the release of the cytokine, differently from what is described in fibroblasts where vesicle lysis mediates the release of fibroblast growth factor-2, another protein lacking a conventional secretory peptide which is released by vesicle shedding (32).
Globally, our data support and complement results obtained in monocytes by MacKenzie et al. (11), which indicated the existence of a vesicular pathway for IL-1 release by vesicle shedding. However, the observation that vesicle shedding can occur in resting microglia, which does not release IL-1, supports the idea that vesicle shedding does not represent a secretory mechanism selective for the cytokine, but rather a secretory pathway possibly shared by other releasable factors, which is used by IL-1 when produced in activated microglia. Given that P2X7R activation in microglia triggers the synthesis and release of other cytokines like TNF (39) and proteases as well as their precursors, including plasminogen (40), it would be of interest to investigate whether ATP-induced shed vesicles can also mediate the release of such factors. Furthermore, vesicle shedding, through the release of cell surface components, including membrane receptors (41, 42, 43), can serve functions other than secretion of soluble factors. In this regard, we clearly demonstrate the presence of P2X7Rs on isolated vesicles, thus indicating that vesicles shedding might represent a mechanism to remove P2X7Rs from the cell surface in response to astrocyte-derived ATP signaling. From this point of view the shedding process could represent a defense strategy of microglia against apoptotic insults, produced by excessive or repetitive stimulation by astrocyte-derived ATP. Removal from the cell surface of functional P2X7Rs could facilitate microglial survival and avoid P2X7-mediated microglial apoptosis (3).
Based on the amount of IL-1 detected in Triton X-100-treated supernatant at 10-min stimulation, when negligible levels of the cytokine are present in vesicle-free supernatant, we estimated the fraction releasable by vesicle shedding to be 27% of total cellular content. This estimation is in agreement with results from a previous study (2), which reported a releasable IL-1 fraction of 25% in these cells. Based on these findings, one could speculate that vesicle shedding may be the predominant mechanism of IL-1 secretion in microglia. However, different pathways of IL-1 release might coexist in these cells (Fig. 6). This possibility has been already described for monocytes, where vesicle shedding (11) and exocytosis of endolysosome-related vesicles (6) have been shown to account for 12 and 30% of total IL-1 cellular content, respectively. Further studies are needed to determine the possible contribution of the endolysosome pathway to IL-1 secretion in microglia, although the lack of inhibition of IL-1 release by AAC0CF3 (F. Bianco and C. Verderio, unpublished observations), a PLA2 blocker which strongly inhibits the exocytosis of secretory lysosomes (6), plays against a major contribution of the endolysosomal pathway in microglia.
Although vesicle shedding is most likely a nonspecific and nonunique pathway for IL-1 release, it is extremely important to understand the physiological relevance of IL-1 release from microglial cells by this mechanism. Shed vesicles, containing packages of IL-1, can deliver the cytokine in possible proximity to IL-1Rs present on target cells. Furthermore, given that IL-1 release is enhanced by ATP, the cytokine might be released from shed vesicles when they impact local high concentration of the purine, presumably in close proximity of releasing astrocytes, thus avoiding dilution of the cytokine in the extracellular environment. This might also turn out to be of crucial importance for the onset of a detrimental effect of IL-1 toward degenerating cells, which are known to release large amounts of ATP. Therefore, IL-1 release from microglia can represent a backward signal sent by microglial cells in response to ATP-mediated astrocyte-to-microglia signaling.
Acknowledgments
We thank Dr. Paola Castagnoli-Ricciardi (University of Milano-Bicocca, Milan, Italy) for generously providing N9 microglial cells. We thank Dr. Elisabetta Menna (Consiglio Nazionale delle Ricerche-Institute of Neuroscience, Milan, Italy) for helpful comments and discussion.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This research was supported by Fondo per gli Investimenti della Ricerca di Base Grants RBAUO19 ZEN and RBNE01ERXR_007, European Community Grant QLG 3-2000-01343, Ministero dell’Istruzione dell’Università e della Ricerca-Programmi di Ricerca Scientifica di Rilevante Interesse Nazionale 2002, and Fondazione Italiana Sclerosi Multipla Grant 2003/R/35.
2 Address correspondence and reprint requests to Dr. Claudia Verderio, Consiglio Nazionale delle Ricerche-Institute of Neuroscience, Via Vanvitelli 32, 20129 Milano, Italy. E-mail address: c.verderio{at}in.cnr.it
3 Abbreviations used in this paper: ER, endoplasmic reticulum; GFAP, glial fibrillar acidic protein; LDH, lactate dehydrogenase; oATP, oxidized ATP; PS, phosphatidylserine; P2Rs, purinergic receptor 2; FLICA, fluorochrome inhibitor of caspases; SNAP-23, synaptosomal-associated protein of 23 kDa.
4 The online version of this article contains supplemental material.
Received for publication August 2, 2004. Accepted for publication March 15, 2005.
References
Solle, M., J. Labasi, D. G. Perregaux, E. Stam, N. Petrushova, B. H. Koller, R. J. Griffiths, C. A. Gabel. 2001. Altered cytokine production in mice lacking P2X7 receptors. J. Biol. Chem. 276: 125-132.
Sanz, J. M., F. Di Virgilio. 2000. Kinetics and mechanism of ATP-dependent IL-1 release from microglial cells. J. Immunol. 164: 4893-4898.
Verderio, C., M. Matteoli. 2001. ATP mediates calcium signaling between astrocytes and microglial cells: modulation by IFN-. J. Immunol. 166: 6383-6391.
Schipke, C. G., C. Boucsein, C. Ohlemeyer, F. Kirchhoff, H. Kettenmann. 2002. Astrocyte Ca2+ waves trigger responses in microglial cells in brain slices. FASEB J. 16: 255-257.
Brough, D., R. A. Le Feuvre, R. D. Wheeler, N. Solovyova, S. Hilfiker, N. J. Rothwell, A. Verkhratsky. 2003. Ca2+ stores and Ca2+ entry differentially contribute to the release of IL-1 and IL-1 from murine macrophages. J. Immunol. 170: 3029-3036.
Andrei, C., P. Margiocco, A. Poggi, L. V. Lotti, M. R. Torrisi, A. Rubartelli. 2004. Phospholipases C and A2 control lysosome-mediated IL-1 secretion: implications for inflammatory processes. Proc. Natl. Acad. Sci. USA 101: 9745-9750.
Walev, I., J. Klein, M. Husmann, A. Valeva, S. Strauch, H. Wirtz, O. Weichel, S. Bhakdi. 2000. Potassium regulates IL-1 processing via calcium-independent phospholipase A2. J. Immunol. 164: 5120-5124.
Chauvet, N., K. Palin, D. Verrier, S. Poole, R. Dantzer, J. Lestage. 2001. Rat microglial cells secrete predominantly the precursor of interleukin-1 in response to lipopolysaccharide. Eur. J. Neurosci. 14: 609-617.(Fabio Bianco*,, Elena Pra)