Regulation of Store-Operated Calcium Entry by Calcium-Independent Phospholipase A₂ in Rat Cerebellar Astrocytes

Identification of cell types in acute rat brain slices

We used acute slices of the rat cerebellum to monitor cytosolic Ca²⁺ in somata of Bergmann glia and of astrocytes in the GCL in situ. Bergmann glial cells are a special kind of cerebellar astrocytes with their cell bodies grouped around a Purkinje neuron and extend radial or Bergmann fibers toward the pial layer (Fig. 1A,B). In mammals, most of the radial glial cells disappear during development; only a vestige persists in some specialized areas, such as Bergmann glia in cerebellar cortex, in which they adapt to perform the local functional requirements (Rakic, 2003). Although Bergmann glial cells are classified as belonging to the astrocyte lineage (Campbell and Götz, 2002), we treated these cells as a distinct glial cell category because of their functional importance in the cerebellum. Bergmann glial cells were readily identified by their location, typical radial morphology, and their ability to generate Ca²⁺ transients during application of ATP or ADP (Kirischuk et al., 1995; Hammer, 2006). After dye loading, the cell body and the radial dendrites of Bergmann glia were clearly visible during their response to ATP (20 μm for 30 s) (Fig. 1B,D,E), whereas Purkinje neurons were hardly stained by the dye. Ca²⁺ responses to ATP indicated the functional status of the intracellular Ca²⁺ stores. Bergmann glial cells, however, often lacked the Ca²⁺ rise in response to K⁺-free saline (0 mm K⁺) typical for astrocytes, which was used to identify astrocytes in the GCL (Fig. 1C,F,G) (Dallwig and Deitmer, 2002; Beck et al., 2004).

Application of CPA (20 μm), an inhibitor of the sarcoendoplasmic Ca²⁺-ATPase, in the absence of external Ca²⁺ evoked a transient cytosolic Ca²⁺ rise (Fig. 1E,G), indicative of Ca²⁺ leak from the intracellular Ca²⁺ stores. Readdition of Ca²⁺ in the presence of CPA resulted in another cytosolic Ca²⁺ rise (Fig. 1E,G), indicative of SOCE after the depletion of the intracellular Ca²⁺ stores. These Ca²⁺ responses were recorded in both Bergmann glial cells and astrocytes in the GCL.

Store-operated Ca²⁺ entry after Ca²⁺ store depletion by cyclopiazonic acid

First, we studied SOCE as induced by CPA in Bergmann glia and in astrocytes of the GCL with respect to its sensitivity to different drugs (Fig. 2). In Bergmann glia, CPA-induced depletion of Ca²⁺ stores resulted in Ca²⁺ transients during readdition of Ca²⁺ (SOCE), with a mean amplitude of 48 ± 8% ΔF (n = 92 cells/8 slices). This Ca²⁺ transient was significantly reduced to 21 ± 2% ΔF (p < 0.05; n = 42/3) by 2-APB (100 μm), an inhibitor of SOCE channels (Ma et al., 2000; Bakowski et al., 2001; Prakriya and Lewis, 2001; Voets et al., 2001). 2-APB was applied 5 min before the readdition of Ca²⁺ in the presence of CPA (Fig. 2A,B). Comparable results were obtained with BTP2 (20 μm) (Ishikawa et al., 2003; Zitt et al., 2004), a structurally distinct blocker of Ca²⁺ channels that are activated by store depletion. BTP2, preincubated for 15–20 min, significantly reduced SOCE to 12 ± 1.3% ΔF (p < 0.01; n = 46/3) (Fig. 2A,B). To test whether iPLA₂ plays a role in SOCE in Bergmann glia, the specific, irreversible inhibitor of iPLA₂, BEL (25 μm) (Ackermann et al., 1995; Winstead et al., 2000), was used. BEL requires the basal activity of iPLA₂ and thus inhibits the enzyme more effectively at physiological temperature (37–40°) than at room temperature (20–22°), when the enzyme activity is low. Moreover, it permeates only slowly into intact cells and hence requires relatively long incubation times (Balsinde et al., 1995). Because the efficacy of BEL to block iPLA₂ is temperature and time dependent, acute brain slices were preincubated in BEL for at least 30 min at 37°C. BEL was removed by washing for 15–20 min before starting the protocol of measuring cytosolic Ca²⁺ changes. In BEL-treated Bergmann glial cells, the mean amplitude of the SOCE was reduced to 4.5 ± 0.2% ΔF (n = 62/6; p < 0.01) (Fig. 2A,B). In addition to iPLA₂ inhibition, BEL has been shown to inhibit another cellular phospholipase, the Mg²⁺-dependent phosphatidate phosphohydrolase (PAP-1) (Balsinde and Dennis, 1996; Fuentes et al., 2003), which is known to be involved in the synthesis of diacylglycerol. To check the possible role of PAP-1 in BEL-induced inhibition of SOCE, we used propranolol (50–100 μm) to inhibit PAP-1 (Fuentes et al., 2003). The acute slices were pretreated with propranolol for 30 min at room temperature. Propranolol did not inhibit the CPA-induced Ca²⁺ influx in either Bergmann glial cells or astrocytes of the GCL (data not shown), which indicates that PAP-1 is not involved in SOCE regulation in cerebellar astrocytes.

BEL has also been reported to induce cell death by inhibiting endoplasmic iPLA₂ (Cummings et al., 2002) or PAP-1 (Fuentes et al., 2003). After BEL treatment, Bergmann glial cells still generated Ca²⁺ transients in response to ATP (20 μm, 30 s), which evokes IP₃-mediated Ca²⁺ release from ER stores (Shao and McCarthy, 1995) (Fig. 2B, inset). This indicates the functional status of the Ca²⁺ stores, ruling out the possibility that BEL, at the concentration used here, caused cell death in Bergmann glia. We also used a second iPLA₂ blocker, PACOCF₃ (10 μm), which belongs to a different chemical family than BEL and is known to inhibit iPLA₂ with an IC₅₀ value of 3.8 μm (Ackermann et al., 1995), to further confirm the involvement of iPLA₂ in regulating SOCE. The Ca²⁺ transient during Ca²⁺ readdition in the presence of CPA was significantly reduced to 5.0 ± 0.8% ΔF (n = 73/6; p < 0.01) (Fig. 2A,B) by PACOCF₃ (Fig. 2A,B). In the presence of BEL or PACOCF₃, the Ca²⁺ transient during Ca²⁺ readdition was even smaller than in control cells (the mean amplitude being 10.2 ± 1% ΔF; p < 0.05; n = 52/5), which were not exposed to CPA but left for the same time period in Ca²⁺-free solution. This suggests that some iPLA₂ activity is present even when Ca²⁺ stores are not depleted by CPA (Fig. 2B).

Like in Bergmann glial cells, readdition of Ca²⁺ induced an influx of Ca²⁺ in GCL astrocytes in the presence of CPA (Fig. 2C,D). During readdition of Ca²⁺ to the extracellular solution, cytosolic Ca²⁺ increased by 70 ± 8% ΔF (n = 62/8). This SOCE was significantly reduced by 2-APB (100 μm) to 32 ± 2.5% ΔF (p < 0.01; n = 31/4) and by BTP2 (20 μm) to 13 ± 1% ΔF (p < 0.005; n = 22/3). Preincubation with BEL or PACOCF₃ strongly inhibited SOCE to 5.0 ± 0.5% ΔF (p < 0.005; n = 57/6) and to 8 ± 1% ΔF (p < 0.005; n = 53/6), respectively, indicating that iPLA₂ is essential for the activation of SOCE (Fig. 2C,D). The PAP-1 inhibitor propranolol (50–100 μm) did not affect SOCE induced in the presence of CPA (data not shown). GCL astrocytes responded to ATP (20 μm, 30 s) with a Ca²⁺ transient after BEL treatment, which indicated the functional Ca²⁺ stores in the cells (Fig. 2D, inset). In control cells, readdition of external Ca²⁺ induced a Ca²⁺ transient of 22 ± 1.3% ΔF (p < 0.005; n = 59/5), which was still larger than in cells treated with BEL or PACOCF₃ (p < 0.01) (Fig. 2C,D). These results suggest that iPLA₂ activation by depletion of Ca²⁺ stores in both Bergmann glia and GCL astrocytes is required for the SOCE in these cells. Thus, similar results were obtained in experiments using either BEL or PACOCF₃ to inhibit iPLA₂ activity, and, for the following experiments, BEL was chosen as potent iPLA₂ inhibitor.

Ca²⁺ influx induced by calmidazolium

After obtaining evidence for a role of iPLA₂ in CPA-induced Ca²⁺ influx, we wanted to know whether the activation of iPLA₂ itself, without previous Ca²⁺ store depletion, induces influx of Ca²⁺ in Bergmann glial cells and GCL astrocytes. The activity of iPLA₂ is regulated by CaM, which binds to the C terminal of the enzyme and thereby inhibits its activity (Wolf and Gross, 1996a,b; Wolf et al., 1997). We used the CaM antagonist CMZ at a concentration of 1 μm to dissociate CaM from iPLA₂. At this concentration, CMZ does not cause release of Ca²⁺ from stores itself (Smani et al., 2004), which was confirmed in the present study by comparing the cytosolic Ca²⁺ response induced by ADP (20 μm) in untreated cells and cells treated with 1 μm CMZ. No significant difference in the amplitude of the Ca²⁺ transients evoked by ADP was found in untreated (n = 24/3) and CMZ-treated (n = 29/3) cells (data not shown), indicating that CMZ does not affect the filling state of the Ca²⁺ stores.

CMZ was applied in the absence of external Ca²⁺, and a Ca²⁺ transient was recorded in Bergmann glial cells, when Ca²⁺ was readded to the extracellular solution (Fig. 3A). This Ca²⁺ transient was similar to that induced by the depletion of Ca²⁺ stores by CPA and had an amplitude of 39 ± 4% ΔF (n = 61/7) (Fig. 3B). To determine the mechanism by which CMZ activates Ca²⁺ influx in these cells, we examined the effect of SOCE channel inhibitors 2-APB (100 μm) and BTP2 (20 μm) on the Ca²⁺ rise in the presence of CMZ. In Bergmann glia, this Ca²⁺ rise was reduced to 13 ± 1.3% ΔF in the presence of 2-APB (p < 0.005; n = 40/3), and pretreatment of the slices with BTP2 for 15 min decreased this Ca²⁺ rise to 11 ± 1.2% ΔF (p < 0.005; n = 25/3). The iPLA₂ inhibitor BEL reduced the Ca²⁺ rise in CMZ even to 1.0 ± 0.1% ΔF (p < 0.005; n = 39/3) compared with the untreated control slices, in which the Ca²⁺ rise was 10 ± 1% ΔF (p < 0.005; n = 52/5) (Fig. 3A,B).

CMZ has been shown to inhibit the activity of adenylyl cyclase (AC), albeit at greater concentrations than those used in the present study (Haunso et al., 2003). We tested whether the CMZ-induced augmentation of the Ca²⁺ transient during Ca²⁺ readdition can be mimicked by the inhibition of AC activity. We used the AC inhibitor SQ 22536 (200 μm) in the absence of CMZ for 15 min before reading Ca²⁺. Ca²⁺ readdition resulted in a Ca²⁺ influx with a mean amplitude of 13 ± 0.9% ΔF (n = 42/5; data not shown), which was not significantly different from control slices (same treatment except without SQ 22536). Hence, inhibition of AC does not increase the Ca²⁺ rise after readdition of external Ca²⁺, as CMZ does.

In GCL astrocytes, the Ca²⁺ rise in the presence of CMZ decreased from 78 ± 8% ΔF (n = 61/5) to 35 ± 6.5% ΔF in the presence of 2-APB (p < 0.01; n = 36/4) and to 25 ± 2% ΔF in the presence of BTP2 (p < 0.005; n = 25/3). Preincubation of BEL almost completely blocked the Ca²⁺ rise in CMZ to a mean amplitude of 2.5 ± 0.4% ΔF (p < 0.005; n = 27/3). This was significantly smaller than in control cells (p < 0.005), in which the readdition of Ca²⁺ induced a Ca²⁺ rise of 22 ± 1.3% ΔF (p < 0.005; n = 59/5) (Fig. 3C,D). When the adenylate cyclase inhibitor SQ 22536 was present, Ca²⁺ readdition evoked a Ca²⁺ influx with a mean amplitude of 26 ± 1% ΔF (n = 44/5; data not shown), which was not significantly different from control cells. Our experiments show that inhibiting calmodulin with CMZ induces a Ca²⁺ influx that is dependent on the activation of iPLA₂ and SOCE channels. Furthermore, these results indicate that CMZ activates a Ca²⁺ influx even when intracellular Ca²⁺ stores are not depleted.

Ca²⁺ influx induced by lysophospholipids

Because the activation of iPLA₂ releases lysophospholipids and a free fatty acid (AA), we tested whether these products can lead to Ca²⁺ influx into astrocytes. We used three different lysophospholipids: LPI (250 nm), LPC (250 nm), and LPA (250 nm). In Ca²⁺-free saline, all three lysophospholipids were added 3 min before readdition of Ca²⁺ to the extracellular solution. LPC and LPI induced a larger Ca²⁺ rise during Ca²⁺ readdition in Bergmann glia (Fig. 4) and GCL astrocytes (Fig. 5), whereas Ca²⁺ rises in the presence of LPA were not significantly changed compared with those in control experiments in the absence of lysophospholipids (Figs. 4D, 5D). Notably, both LPI and LPC still induced augmented Ca²⁺ rises after incubation of the slices with BEL (Figs. 4A–C, 5A–C). Because LPI and LPC are products of iPLA₂ activity and appear downstream of iPLA₂ activity, this indicates that BEL does not suppress Ca²⁺ influx per se. Moreover, this suggests that iPLA₂, by producing lysophospholipids, promotes Ca²⁺ influx.

In Bergmann glia, the Ca²⁺ rise during Ca²⁺ readdition had an amplitude of 42 ± 6% ΔF (n = 54/5) in the presence of added LPI. This Ca²⁺ rise was reduced to 16 ± 2.7% ΔF (n = 51/4) and 13 ± 2.3% ΔF (n = 28/3) in the presence of 100 μm 2-APB and after preincubation of 20 μm BTP2, respectively (p < 0.05) (Fig. 4A,B). The mean amplitude of the Ca²⁺ rise in the presence of added LPC was 36 ± 3.7% ΔF (n = 89/7), which was reduced to 11 ± 1.3% ΔF (n = 72/4) and 12 ± 1.3% ΔF (n = 32/3) by 2-APB and BTP2, respectively (p < 0.01) (Fig. 4C). The Ca²⁺ rise in the presence of LPA was 15 ± 1.6% ΔF (n = 67/5) and was not significantly different from control cells without any added lysophospholipids, under which condition the Ca²⁺ rose by 9 ± 1% ΔF (n = 32/5) (Fig. 4D).

In GCL astrocytes, the Ca²⁺ rise during Ca²⁺ readdition in the presence of LPI had an amplitude of 58 ± 6.2% ΔF (n = 61/5). This Ca²⁺ rise was inhibited to 27 ± 4.4% ΔF by 2-APB (p < 0.05; n = 48/4) and 25 ± 2.1% ΔF after preincubation with BTP2 (p < 0.05; n = 31/3) (Fig. 5A,B). The mean amplitude of this Ca²⁺ rise in the presence of LPC was 53 ± 7% ΔF (n = 89/7) and was reduced to 26 ± 4 and 22 ± 2.5% ΔF by 2-APB (n = 67/4) and BTP2 (n = 35/3), respectively (p < 0.05) (Fig. 5C). In the presence of LPA, the Ca²⁺ rise had an amplitude of 24 ± 1.7% ΔF (n = 67/6), which was not significantly different from that in control cells, which was 23 ± 2.5% ΔF (n = 43/5) (Fig. 5D). These results show that LPI and LPC, but not LPA, induce a Ca²⁺ influx mediated by 2-APB- and BTP2-sensitive ion channels.

We then asked whether the Ca²⁺ influx after Ca²⁺ readdition uses the same pathway after depleting the stores with CPA or with added LPI and LPC. Therefore, we tested whether LPI and LPC led to an additional Ca²⁺ influx after inducing SOCE in CPA. After depletion of Ca²⁺ stores by CPA, readdition of external Ca²⁺ evoked a Ca²⁺ increase (Fig. 6A–D), as described above (Fig. 2). In the continued presence of CPA and external Ca²⁺, LPI and LPC, both at 250 nm, were added (Fig. 6A,C, arrows). We observed no additional Ca²⁺ rise by either lysophospholipid in Bergmann glia (Fig. 6A,B) or in GCL astrocytes (Fig. 6C,D), suggesting that CPA- and lysophospholipid-induced Ca²⁺ influx are not additive.

Ca²⁺ influx induced by arachidonic acid

Because iPLA₂-mediated hydrolysis of membrane phospholipids also generates AA, we also examined the effect of AA, which has been demonstrated to activate Ca²⁺ influx through arachidonate-regulated channels in a variety of different cell types, which is clearly distinct from SOCE (Shuttleworth and Thompson, 1999). Acute slices were exposed to AA (10 μm) in Ca²⁺-free solution 3 min before external Ca²⁺ was readded (Fig. 7A). In GCL astrocytes, AA induced a significant increase of the Ca²⁺ rise after readdition of Ca²⁺, which was often oscillatory and had a mean maximal amplitude of 48 ± 5% ΔF (p < 0.05; n = 53/6). In comparison, the Ca²⁺ influx in control cells without added AA had an amplitude of 23.5 ± 2% ΔF (n = 38/5). The larger Ca²⁺ rise in the presence of AA was not significantly changed in the presence of 2-APB (n = 47/6) (Fig. 7A,B), indicating that AA activated Ca²⁺ channels, which were insensitive to 2-APB in contrast to those activated in the presence of CPA, CMZ, and lysophospholipids. The Ca²⁺ rise in Bergmann glia in the presence of AA was not significantly larger compared with control experiments without AA (Fig. 7C), although their mean amplitude was 17 ± 3.3% ΔF (n = 36/4) compared with 8 ± 1.2% ΔF (n = 23/3) in control cells. 2-APB had no effect on the Ca²⁺ rise in the presence of AA.

BEL-sensitive SOCE and iPLA₂ activity in cultured astrocytes

To confirm the involvement of iPLA₂ in SOCE in cerebellar astrocytes, we wanted to measure the activity of iPLA₂ under store-depleted condition and after inhibition of CaM, in astrocyte culture. First, we established that the Ca²⁺ rises during readdition of external Ca²⁺ in the presence of CPA and CMZ were sensitive to the iPLA₂ inhibitor BEL (Fig. 8A–C). Addition of CPA (10 μm) in Ca²⁺-free saline evoked a Ca²⁺ transient in all cells. During readdition of external Ca²⁺ after store depletion by CPA, a Ca²⁺ rise, indicative of SOCE, was measured with a mean amplitude of 75 ± 4% ΔF (n = 150/5). This Ca²⁺ rise was much larger than in control cells (12 ± 0.5% ΔF; p < 0.005; n = 163/5), in which the cells were not exposed to CPA. In astrocyte cultures pretreated with 25 μm BEL, the SOCE induced in the presence of CPA was reduced to 4 ± 0.9% ΔF (p < 0.005; n = 141/5) (Fig. 8A,C). In the presence of 1 μm CMZ, the cytosolic Ca²⁺ rose to 87 ± 7% ΔF (n = 132/5), which was reduced to 4 ± 0.8% ΔF (p < 0.005; n = 146/5) after preincubation with BEL (Fig. 8B,C). Again, the Ca²⁺ rise induced by CMZ was significantly smaller after BEL treatment than the Ca²⁺ rise in control cells without CMZ (p < 0.005).

We determined BEL-sensitive iPLA₂ activity in cerebellar astrocyte cultures by absorbance measurements using an iPLA₂ activity assay (see Materials and Methods). Maximal iPLA₂ activity (100%) was measured in the presence of 10 mm EGTA to suppress all Ca²⁺/calmodulin-dependent iPLA₂ inhibition. Basal iPLA₂ activity was 4 ± 1.4% of the maximal iPLA₂ activity as measured in cell homogenates from untreated cerebellar astrocytes (Fig. 8D). In astrocytes treated with 10 μm CPA (10 min at 37°C) or 1 μm CMZ (10 min at 37°C), the iPLA₂ activity increased to 62 ± 1.2 and 55 ± 3.0% of the maximal iPLA₂ activity, respectively. These results suggest that iPLA₂ activity is increased under conditions when SOCE is activated in cerebellar astrocytes.

Antisense inhibition of iPLA₂

In another approach to examine the role of iPLA₂ in SOCE, we used antisense technology to knock down the expression of iPLA₂. Cerebellar astrocyte cultures were transfected with S-ODN and AS-ODN of iPLA₂-A isoforms using a Lipofectamine-assisted transfection protocol (Balsinde et al., 1997; Smani et al., 2003). Control cells were exposed to the transfection protocol without ODNs to check whether LF alone had any nonspecific effects. Cultured astrocytes were loaded with fura-2, and the fluorescence ratio at 340 and 380 nm excitation was measured, indicative of the Ca²⁺ concentration in the cells. First, we used ATP (20 μm, 20 s) to check the functional status of the Ca²⁺ stores. ATP induced a similar Ca²⁺ rise in LF-treated and S-ODN-transfected cells, with mean amplitudes of 0.55 ± 0.13 ratio units (n = 176/6) and 0.59 ± 0.12 ratio units (n = 142/8) (Fig. 9A,C), respectively. In AS-ODN-transfected cells, the Ca²⁺ transient induced by ATP was strongly reduced to 0.17 ± 0.08 ratio units (p < 0.001; n = 153/8) (Fig. 9B,C), possibly reflecting the impaired filling status of the intracellular Ca²⁺ stores, when iPLA₂ expression was inhibited. CPA (10 μm) was then applied in the absence of extracellular Ca²⁺ for 7 min, which led to a slow transient Ca²⁺ rise in all cells. The absolute mean amplitude of Ca²⁺ transient after the addition of CPA in LF-treated cells was 0.16 ± 0.03 ratio units (n = 176/6), which was similar to that of S-ODN-treated cells, being 0.15 ± 0.03 ratio units (n = 142/8). The AS-ODN-transfected cells showed a reduced Ca²⁺ transient during CPA addition of 0.08 ± 0.04 ratio units (p < 0.001; n = 153/8) (Fig. 9B,C), suggesting reduced Ca²⁺ release from the stores, possibly attributable to an incomplete filling status of the Ca²⁺ stores after interference with iPLA₂ expression.

Because the Ca²⁺ store content appeared to be reduced by AS-ODN, we wanted to check whether this reduction might be attributable to dysregulated ER-Ca²⁺ homeostasis. The decay time constant τ of the Ca²⁺ transients evoked by added ATP was analyzed in the S-ODN- and AS-ODN-treated cells. In S-ODN-treated astrocytes, τ was 10 ± 3.8 s (n = 62), which was not significantly different from that in AS-ODN-treated cells, with a τ of 11 ± 5.7 s (n = 51; data not shown); this finding is in line with an unimpaired ER homeostasis in AS-ODN-treated cells. We also measured the basal Ca²⁺-dependent fura-2 ratio level in S-ODN- and AS-ODN-transfected cells to get an indication of a change in resting cytosolic Ca²⁺ level after inhibiting iPLA₂. The basal ratio was 0.48 ± 0.03 (n = 142/8) in S-ODN transfected cells, which was indeed significantly different from the AS-ODN-transfected cells, with a ratio of 0.41 ± 0.03 (p < 0.001; n = 153/8; data not shown), suggesting that the reduced cytosolic Ca²⁺ level in the AS-ODN-transfected cells might be attributable to the impaired iPLA₂ activity.

During readdition of Ca²⁺, the Ca²⁺ rise was similar in LF-treated cells and cells transfected with the S-ODN, with a mean amplitude of 0.30 ± 0.06 ratio units (n = 176/6) and 0.33 ± 0.09 ratio units (n = 142/8), respectively. There was a significant reduction, however, in the Ca²⁺ transient evoked by Ca²⁺ readdition in AS-ODN-transfected cells to 0.16 ± 0.07 ratio units (p < 0.001; n = 153/8) (Fig. 9B,C). Hence, only transfection with iPLA₂ antisense-ODN, but not sense-ODN or LF alone, affected SOCE in cultured astrocytes, indicating a crucial role of iPLA₂ for generating SOCE.

To ascertain the effective knockdown of iPLA₂ expression in AS-ODN-transfected cells, we labeled the cells with anti-iPLA₂ antibody and measured the relative fluorescence. Antibody staining of all transfected cells (LF, S-ODN, and AS-ODN) was performed 48 h after transfection. All cells were processed in parallel with the same protocol, and the anti-iPLA₂ immunoreactivity was assessed under the same settings of the confocal microscope. The relative fluorescence intensity of iPLA₂ expression was measured only in cells labeled with fluorescein, indicating successful transfection, and is given in arbitrary units (AU). In LF-treated and S-ODN-transfected cells, the mean fluorescence was 110 ± 1.5 AU (n = 324) (Fig. 9D,G) and 130 ± 1.6 AU (n = 330) (Fig. 9E,G), respectively. In AS-ODN-treated cells, the fluorescence intensity was significantly reduced to 69 ± 1.5 AU (p < 0.001; n = 509) (Fig. 9F,G), consistent with a reduction of iPLA₂ expression in AS-ODN treated cells. The background fluorescence was measured to be 17 ± 0.8 AU (p < 0.001; n = 290) (Fig. 9G) in control cells, which were stained only with the secondary antibody.

Spontaneous Ca²⁺ oscillations in rat cerebellum

If SOCE was indeed instrumental for the refilling of intracellular Ca²⁺ stores, suppression of SOCE would be expected to affect spontaneous Ca²⁺ oscillations (Nett et al., 2002; Zur Nieden and Deitmer, 2006). We therefore examined the effects of two substances, known to inhibit SOCE, on spontaneous Ca²⁺ oscillations in astrocytes in situ. Bergmann glia and GCL astrocytes showed spontaneous activity in cerebellar brain slices (Fig. 10A,B), although Bergmann glial cells generated Ca²⁺ oscillations less frequently than GCL astrocytes. We analyzed the relative number of cells with spontaneous Ca²⁺ oscillations and the frequency of the Ca²⁺ oscillations. Over 20 min, spontaneous Ca²⁺ signals were recorded in the standard solution as control. The experimental protocol continued for another 20 min after the SOCE channel blocker 2-APB was applied. Ca²⁺ signals were analyzed for at least 15 min, leaving the first 5 min to allow the drugs to work. Spontaneous Ca²⁺ responses could be recorded in 28% (n = 87 of 313/31) of the cells identified as Bergmann glia and in 47% (n = 107 of 227/27) of the GCL astrocytes. The overall frequency of spontaneous Ca²⁺ transients averaged 0.17 ± 0.06 transients/min (n = 87/31) in Bergmann glial cells and 0.40 ± 0.07 transients/min (n = 107/27) in GCL astrocytes.

Inhibiting iPLA₂ by preincubation with BEL for 30–40 min, as described above, decreased the frequency of the Ca²⁺ signals in GCL astrocytes (Fig. 10A) on average to 40% of the control, whereas the number of spontaneously active cells decreased to 46% (Fig. 10B,C). In Bergmann glial cells, the number of active cells decreased to 28%, and the frequency of Ca²⁺ signals decreased to 39% of the control.

In the presence of 2-APB, Ca²⁺ signals were observed in only 58% of the Bergmann glial cells and 62% of the GCL astrocytes that exhibited spontaneous Ca²⁺ oscillations under control conditions (Fig. 10B). The frequency of the Ca²⁺ oscillations was reduced to 49% (n = 17/11) in Bergmann glia and to 55% (n = 24/11) in GCL astrocytes in the presence of 2-APB (Fig. 10C). These results suggest that the functional activity of iPLA₂ and 2-APB-sensitive SOCE channels are required for maintaining spontaneous Ca²⁺ oscillations in Bergmann glia and GCL astrocytes at their normal level.

CPA-induced SOCE requires the functional activity of iPLA₂ in astrocytes

The possible link of iPLA₂ activation to store depletion was first reported in rat smooth muscle cells (Wolf et al., 1997) and pancreatic islets (Nowatzke et al., 1998), in which the depletion of intracellular Ca²⁺ stores resulted in an increased iPLA₂-mediated phospholipid hydrolysis and accumulation of free fatty acids. According to the proposed molecular mechanism for cultured smooth muscle cells, the generated lysophospholipids activate SOCE channels and capacitative Ca²⁺ influx, leading to the refilling of Ca²⁺ stores (Smani et al., 2004).

In rat brain, >70% of the PLA₂ activity could be attributed to iPLA₂ (Yang et al., 1999). iPLA₂ was detected in the cerebellar cortex in all cell layers by means of in situ hybridization and antibody staining (Shirai and Ito, 2004). In our study, the activation of SOCE by store depletion appeared to be attributable to stimulation of iPLA₂, because iPLA₂ impairment by the iPLA₂ inhibitors BEL and PACOCF₃ completely prevented Ca²⁺ influx in response to CPA in acute brain slices and in culture (BEL), as outlined in the hypothetical diagram in supplemental Figure 1 (available at www.jneurosci.org as supplemental material). Moreover, iPLA₂ activity increased when ER stores had been depleted in astrocyte culture. Importantly, the specific iPLA₂ AS-ODN impaired SOCE, presumably by a reduction of iPLA₂ expression in the cells. Notably, the CPA-mediated Ca²⁺ influx in the presence of BEL or PACOCF₃ was even smaller than the Ca²⁺ influx in control cells, in which Ca²⁺ stores were not depleted by CPA. This suggests some constitutive activity of iPLA₂, which might be independent of the filling state of intracellular Ca²⁺ stores. The Ca²⁺ influx evoked by lysophospholipids in BEL-treated brain slices indicated that BEL did not suppress Ca²⁺ influx per se and presumably did not affect SOCE directly. This drug has also been reported to inhibit PAP-1, but the CPA-mediated Ca²⁺ entry was not prevented by the PAP-1 inhibitor propranolol in our study, indicating that the BEL-induced inhibition of SOCE was not attributable to suppressing PAP-1 activity. Furthermore, ATP-induced Ca²⁺ transients in BEL-pretreated astrocytes demonstrated intact Ca²⁺ stores, suggesting that BEL did not induce conditions with impaired ER integrity or apoptosis, under our experimental conditions.

In cells not challenged with thapsigargin or CPA, CaM binds to iPLA₂ and thereby suppresses the activity of this enzyme. After the dissociation of CaM, iPLA₂ becomes functionally active (Wolf and Gross, 1996a,b). In cultured smooth muscle cells, the CaM inhibitor CMZ induces Ca²⁺ release as well as 2-APB-insensitive Ca²⁺ influx at concentrations larger than 1 μm, whereas lower doses of CMZ (≤1 μm) only evoke 2-APB-sensitive Ca²⁺ entry (Smani et al., 2004). In HeLa cells, CMZ at concentrations larger than 1 μm induced Ca²⁺ entry that involves a non-SOCE pathway and was sensitive to AA (Peppiatt et al., 2004). In the present study, we used 1 μm CMZ to evoke Ca²⁺ influx in acute brain slices and found that CMZ induced a Ca²⁺ influx without depleting the Ca²⁺ stores. This Ca²⁺ influx is sensitive to 2-APB and BTP2 and is suppressed after inhibiting iPLA₂. Because the Ca²⁺ signals induced by CPA-mediated store depletion and by CMZ-mediated CaM inhibition exhibit the same pharmacological profile, we suggest a similar signaling pathway in glial cells as reported for smooth muscle cells (Smani et al., 2004) (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). Hence, store depletion as well as CMZ would release the inhibitory interaction of CaM on iPLA₂ and then allow SOCE. This is supported by the enhanced enzyme activity of iPLA₂ in cerebellar astrocyte culture after store depletion by CPA and after inhibiting CaM with CMZ.

Activation of SOCE channel appears to be mediated by the products of iPLA₂ because both lysophospholipids LPC and LPI activated a SOCE-like Ca²⁺ influx. LPA did not generate a Ca²⁺ influx, which may be attributable to the lack of a large head group in LPA, in contrast to LPI and LPC. Suppression of iPLA₂ activity by BEL did not affect the LPI- and LPC-induced Ca²⁺ influx, in line with the hypothesis that the lysophospholipid-mediated activation of Ca²⁺ entry is downstream to iPLA₂ activation (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). Moreover, our results suggest that LPI- and LPC-induced Ca²⁺ influx passes through 2-APB- and BTP2-sensitive channels. Lysophospholipid-activated Ca²⁺ entry in Bergmann glial cells and GCL astrocytes apparently uses a pathway similar, if not identical, to that triggered by store depletion. It is yet unknown, however, how the generation of lysophospholipids activates SOCE. Lysophospholipids may open the SOCE channels either indirectly by affecting the lipid environment of the channels or directly by interacting with the membrane channel protein. Notably, a direct interaction of lysophosphatidylcholine with the cation channel TRPC5 was recently suggested in transfected HEK293 cells and arterial smooth muscle cells (Flemming et al., 2006).

Ca²⁺ oscillations rely on SOCE in cerebellar astrocytes

Spontaneous Ca²⁺ oscillations have been reported in astrocytes in culture (Fatatis and Russell, 1992; Harris-White et al., 1998) and in situ (Parri et al., 2001; Aguado et al., 2002; Nett et al., 2002; Zur Nieden and Deitmer, 2006). Ca²⁺ signaling in the form of repetitive oscillations are associated with regulating a variety of physiological processes, including gene expression (Dolmetsch et al., 1998), K⁺ conductance (Chen et al., 1997), and secretion (Thomas et al., 1996). The number of spontaneously active astrocytes and the number of Ca²⁺ transients exhibited by each astrocyte increased by increasing the external Ca²⁺ (Parri and Crunelli, 2003) and was reduced by inhibiting metabotropic glutamate receptors of groups I and II (Zur Nieden and Deitmer, 2006). SOCE and Ca²⁺ store refilling are required to maintain both spontaneous Ca²⁺ oscillations (Nett et al., 2002; Zur Nieden and Deitmer, 2006) and Ca²⁺ oscillations evoked by stimulation of metabotropic glutamate receptors in type I cortical astrocytes (Pizzo et al., 2001). It has been shown in rat hippocampal astrocytes that Ca²⁺ oscillations were attributable to phospholipase C-mediated Ca²⁺ release from intracellular stores (Zur Nieden and Deitmer, 2006). 2-APB effectively inhibits SOCE channels in many cell types (Bakowski et al., 2001; Braun et al., 2001; Broad et al., 2001; Gregory et al., 2001; Kukkonen et al., 2001; Prakriya and Lewis, 2001) but can also be a membrane-permeable inhibitor of IP₃ receptors (Wu et al., 2000). We observed that the number of spontaneously active astrocytes as well as the frequency of oscillations were reduced by blocking SOCE channels and by inhibiting iPLA₂ activity. Our results provide evidence that iPLA₂-mediated Ca²⁺ influx through SOCE channels is needed for the maintenance of Ca²⁺ signaling in astrocytes in situ, which requires continuous refilling of Ca²⁺ stores.

In summary, our results suggest that iPLA₂ is a major regulator of SOCE in cerebellar astrocytes; during depletion of Ca²⁺ stores, iPLA₂ could be activated to open the channels in plasma membrane by the formation of lysophospholipids (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). A CIF would transfer the signal from depleted stores to relieve the block of iPLA₂ by CaM in the plasma membrane, consistent with studies of iPLA₂-mediated SOCE in other cell types evoked by emptying stores by either inhibiting SERCA pumps (Smani et al., 2004; Van den Abeele et al., 2004) or activating translocon pores (Flourakis et al., 2006). CIF has been suggested to act through the activation of iPLA₂ (Smani et al., 2004; Van den Abeele et al., 2004); however, it remains unresolved how the formation of CIF is triggered in response to depletion of Ca²⁺ stores. The STIM1 is likely the sensor (CIF) of depleted Ca²⁺ inside the ER Ca²⁺ stores, which translocates in response to store depletion from the ER to the plasma membrane (Roos et al., 2005; Zhang et al., 2005) and also controls the operation of the SOCE channels within the plasma membrane (Spassova et al., 2006). Recently, the plasma membrane four-transmembrane-spanning protein Orai1 or CRACM1 (Ca²⁺ release-activated channel blocker) has been reported to be necessary for SOCE and CRAC channel activity (Feske et al., 2006; Vig et al., 2006). Orail and STIM1 are likely to be functionally coupled and sufficient to mediate SOCE (Peinelt et al., 2006; Soboloff et al., 2006). Whether STIM1 is the first step in the molecular cascade to activate iPLA₂- and LPL-induced Ca²⁺ entry to refill intracellular Ca²⁺ stores in astrocytes still needs experimental evidence.