Microglia–brain macrophages are immune-competent cells of the CNS and respond to pathologic events. Using bacterial lipopolysaccharide (LPS) as a tool to activate cultured mouse microglia, we studied alterations in the intracellular calcium concentration ([Ca 2+]i) and in the receptor-evoked generation of transient calcium signals. LPS treatment led to a chronic elevation of basal [Ca 2+]i along with a suppression of evoked calcium signaling, as indicated by reduced [Ca 2+]i transients during stimulation with UTP and complement factor 5a. Presence of the calcium chelator BAPTA prevented the activation-associated changes in [Ca 2+]i and restored much of the signaling efficacy. We also evaluated downstream consequences of a basal [Ca 2+]i lifting during microglial activation and found BAPTA to strongly attenuate the LPS-induced release of nitric oxide (NO) and certain cytokines and chemokines. Furthermore, microglial treatment with ionomycin, an ionophore elevating basal [Ca 2+]i, mimicked the activation-induced calcium signal suppression but failed to induce release activity on its own. Our findings suggest that chronic elevation of basal [Ca 2+]i attenuates receptor-triggered calcium signaling. Moreover, increased [Ca 2+]i is required, but by itself is not sufficient, for release of NO and certain cytokines and chemokines. Elevation of basal [Ca 2+]i could thus prove a central element in the regulation of executive functions in activated microglia.
Microglia constitutes the macrophage equivalent of the CNS. In response to pathological events, the normally “resting” microglia gradually transform into motile, secretory active and potentially cytotoxic phagocytes (Kreutzberg, 1996; Schwaiger et al., 1998; Raivich et al., 1999a; Streit et al., 2000; Hanisch, 2001). Activated microglial cells then participate in mechanisms of innate and immune defense, tissue repair, and neuroprotection. However, experimental and clinical evidence also supports the notion that excessive microglial activation exacerbates destructive cascades.
Microglia express a variety of receptors that allow for a monitoring of the surrounding tissue (Nörenberg et al., 1994; Kreutzberg, 1996; Streit et al., 2000). Receptors for neurotransmitters and cotransmitters could be important for sensing neuronal activity, and microglia express functional glutamatergic, adrenergic, and purinoreceptors (Noda et al., 2000; Prinz et al., 2001). Abnormal receptor signaling (excessive intensities or disruption) could indicate homeostatic disturbance and trigger microglial responses. Microglia also express receptors for immune system mediators, namely for cytokines, chemokines, and complement (Möller et al., 1997; Harrison et al., 1998; Raivich et al., 1999b; Hanisch, 2001). Some of these factors may play physiological roles in the developing and healthy adult CNS. Others appear only under pathological conditions (Hopkins and Rothwell, 1995; Merrill and Benveniste, 1996; Hanisch, 2002; Häusler et al., 2002). Some receptors are upregulated during activation (Raivich et al., 1998). In contrast, activated microglia can also exhibit decreased responsiveness to receptor stimulation. As demonstrated recently, purinergic ligands evoked only attenuated reactions in cells that were activated with bacterial lipopolysaccharide (LPS) (Möller et al., 2000b). LPS triggers activation of macrophage-like cells and is used as a tool to mimic Gram-negative infection (abd-el-Basset and Fedoroff, 1995; Ulevitch and Tobias, 1995).
The intracellular calcium concentration ([Ca 2+]i) influences multiple cellular functions, including enzyme or release activities. Transient [Ca 2+]i increases serve the intracellular signaling of numerous plasma membrane receptors (Verkhratsky et al., 1998). Also in microglia, several receptors are linked to the calcium signaling machinery, e.g., those for “classical” neurotransmitters as well as those for immune system mediators (Kreutzberg, 1996; Nolte et al., 1996; Verkhratsky and Kettenmann, 1996; Möller et al., 1997; Boddeke et al., 1999; Hide et al., 2000; Möller et al., 2000a,b). Calcium may serve as an integrator of their cytosolic consequences to control microglial behavior under resting and activated conditions.
Amplitude and kinetics of a calcium signal are determined by features of the ligand-receptor system and its cytosolic effector cascade. This includes Ca 2+ release from inositol-1,4,5-triphopshate or ryanodine-sensitive intracellular Ca 2+ stores, the status of calcium release-activated calcium channels, and the efficacy of the calcium-buffering system (Verkhratsky et al., 1998).
However, the calcium signal might also be influenced by the basal [Ca 2+]i. Its level is maintained by the activity of calcium pumps and the rate of constitutive Ca 2+ influx. We demonstrate that LPS activation of mouse microglia in vitro leads to elevated basal [Ca 2+]i along with attenuated Ca 2+ signaling in response to extracellular stimulation of selected purinoreceptors and complement receptors. Furthermore, we present evidence that the rise in [Ca 2+]i is necessary to allow for characteristic features of microglial activation, such as release of nitric oxide (NO) and certain cytokines and chemokines. We hypothesize that an activation-related lasting increase in the basal [Ca 2+]i level critically determines receptor signaling efficacy and executive behavior of macrophage-like cells.
Materials and Methods
Microglial cultures. Cells were prepared from newborn NMRI mice (Tierzucht Schönwalde, Schönwalde, Germany) and cultured in DMEM–10% fetal calf serum as described previously (Prinz et al., 1999). Cells were plated on glass coverslips (105 cells/cm 2, imaging and staining experiments) or in Petri dishes (2 × 106 cells, biochemical analyses). Ninety-eight percent of the cells were microglia, as revealed by staining (Prinz et al., 1999). Cells were activated with LPS (100 ng/ml, Escherichia coli K-235, 24 hr; Sigma, Deisenhofen, Germany). Incubations with BAPTA AM (10–50 μm) (Sigma) and ionomycin (1–1000 nm) (Sigma) were performed for the same time period.
Calcium measurements. Experiments were performed in HEPES buffer (in mm: 150 NaCl, 5.4 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES/NaOH, and 10 glucose, pH 7.35) at room temperature. [Ca 2+]i was monitored using the calcium-sensitive fluorescent indicator fura-2 AM. Cells were loaded with fura-2 AM (5 μm; stock solution in DMSO; Molecular Probes, Eugene, OR) by incubation in bathing solution (30 min). For fura-2 excitation, cells were illuminated with two alternating wavelengths, 340 ± 5 and 380 ± 5 nm. Excitation was performed by using a monochromator (Polychrome IV; T.I.L.L. Photonics, Martinsried, Germany). The emitted light was collected at 530 ± 10 nm by the long-term exposure CCD camera QuantiCam Long Exposure VGA-b/w (Phase, Luebeck, Germany) coupled to a Zeiss (Oberkochen, Germany) Axioskop equipped with a 20× water immersion Zeiss objective (numerical aperture of 0.5). The monochromator and CCD camera were controlled by Axon Imaging Workbench 2.2 software (Axon Instruments, Union City, CA), being also used for image processing. Data analysis was performed using standard personal computer software. The [Ca 2+]i was calculated from the ratio (R) of fluorescence recorded at 340 and 380 nm excitation wavelengths. Calibrations (conversion of R 340/380 values into molar calcium concentrations) were performed as described previously (Grynkiewicz et al., 1985).
Cells were superfused with HEPES buffer, and images were taken every 3 sec. Basal [Ca 2+]i was determined from the initial 10 images of each cell recording. For eliciting [Ca 2+]i transients, the superfusion medium was switched for 30 sec to a solution of UTP, recombinant human complement factor 5a (C5a), or ATP (as survival control after measurements) (100 μm, 2 nm, and 100 μm in HEPES buffer, respectively; Sigma). A [Ca 2+]i signal was defined as an increase in R 340/380 with clear time correlation to the agonist application. Amplitudes (ΔR) were calculated as difference between R 340/380 basal and peak values.
CD88 immunocytochemistry. Cells were fixed (4% paraformaldehyde in PBS, 20 mm Na2HPO4/NaH2PO4, pH 7.4, and 140 mm NaCl, 15 min) and incubated in block solution [PBS, 2% bovine serum albumin (BSA), and 2% normal rabbit serum (NRS), 1 hr], with PBS rinses in between (two times for 15 min each), incubated with goat anti-mouse CD88 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) [2 μg/ml in PBS, 0.1% BSA, and 0.1% NRS (PBS-B), 1 hr], rinsed (PBS-B, three times for 15 min each), and incubated with rabbit anti-goat-IgG F(ab′)2-Cy3 (1: 200 in PBS-B, 1 hr; Jackson ImmunoResearch, West Grove, PA), and rinsed in PBS-B (three times for 15 min each) and PBS (15 min). Cover-slips were mounted under moviol (Sigma).
CD88 immunodetection. Cells were surface labeled with EZ-Link sulfo–N-hydroxysuccinimidyl–biotin (1 mg/ml of PBS for 30 min at room temperature; Pierce, Rockford, IL). After rinses with PBS, cells were lysed in extraction buffer [20 mm Tris/HCl, pH 7.3, 140 mm NaCl, 0.5% Triton X-100, 150 U of DNase I (Worthington, Lakewood, NJ), and one tablet per 50 ml of Complete protease inhibitors (Boehringer Mannheim, Mannheim, Germany) for 30 min] (Hanisch et al., 2001). After freezing–thawing, anti-CD88 antibody (6 μg/ml) was added for 75 min, followed by 10 μl/ml protein A/G resin (ImmunoCatcher; CytoSignal, Irvine, CA). The mixture was rocked for another 30 min and transferred to spin filters in microcentrifuge tubes equilibrated with mild lysis solution (ImmunoCatcher). After rinses, tubes were centrifuged (16,000 × g, 1 min), and the resin was rinsed again, followed by centrifugation and incubation with SDS-PAGE sample buffer. After centrifugation, the filtrates were collected, and lysate aliquots (protein measured by MicroBCA assay; Pierce) were separated on a 10% SDS-PAGE gel and transferred to polyvinylidene difluoride membrane (Pierce). Membranes were blocked with BSA (5%, 20 mm Tris/HCl buffer, pH 7.3, containing 150 mm NaCl, and TBS). Biotinylated CD88 was visualized with ExtrAvidin peroxidase (at 1:100,000, TBS, containing 0.05% Tween 20 for 90 min; Sigma), followed by ECL (Super Signal Ultra; Pierce). Another sample set underwent analysis for total CD88 protein. After blocking, membranes were incubated with anti-CD88 antibody (1 μg/ml in TBS, 0.05% Tween 20, and 10 mg/ml BSA for 90 min), rinsed, incubated with rabbit anti-goat IgG antibody peroxidase (1:50,000 in TBS, containing Tween 20 and BSA for 60 min; Sigma), washed, and developed by ECL.
Nitric oxid release. NO production was studied in cultures (24-well plates with 1 × 105 cells per well) with and without stimulation by LPS (100 ng/ml), BAPTA (10–50 μm), BAPTA and LPS, and ionomycin (1–1000 nm). NO was determined in the supernatants after 24 hr by the Griess reaction as accumulated nitrite, a breakdown product of NO, using sodium nitrite as a standard. Each supernatant (100 μl) was mixed with 100 μl of Griess reagent and incubated at room temperature for 10 min. Optical density was determined in a microplate reader (1420 Victor; Wallac Oy, Turku, Finland) at 540 nm wavelength.
Cytokine and chemokine release. Microglial cytokine and chemokine release was assayed in 96-well plates with 5 × 104 cells per well after 6 and 24 hr of incubation in the absence or presence of LPS (100 ng/ml), BAPTA (50 μm), BAPTA and LPS, or ionomycin (1–1000 nm). Supernatant samples were analyzed for mouse tumor necrosis factor α (TNFα), interleukin-6 (IL-6), total interleukin-12 (IL-12) (collecting the IL-12 forms p70, p40, and p402), macrophage inflammatory protein-1α (MIP-1α), and KC [the mouse equivalent of growth-related oncogene (GROα)] in sandwich ELISA based on mouse- and factor-specific antibody pairs (R & D Systems, Wiesbaden, Germany) following the procedure of the manufacturer. The color reaction was analyzed in a microplate reader (SLT, Spectra; LabInstruments Deutschland, Crailsheim, Germany). Total protein was determined using the MicroBCA protein assay (Pierce).
Viability tests. Microglial metabolic activity was assayed in 96-well plates with 5 × 104 cells per well using WST-1 reagent (4-[3-(4-iodophenyl)-2-(4-nitrophenyl)-2 H-5-tetrazolio]-1,3-benzene disulfonate) (Roche Diagnostics, Mannheim, Germany) according to the instructions of the manufacturer. The assay is based on the enzymatic cleavage of WST tetrazolium salt to formazan by the succinate–tetrazolium reductase system of the respiratory chain of intact mitochondria. The color reaction was measured in a microplate reader (1420 Victor; Wallac Oy) at 540 nm wavelength after 24 hr of incubation without and with LPS (100 ng/ml), BAPTA (50 μm), BAPTA and LPS, and ionomycin (1–1000 nm).
In addition, cells were labeled by nuclear dyes to determine the amount of surviving cells after an incubation. Staining with DAPI (4,6-diamidino-2-phenylindoldihydro-chlorid) (Boehringer Mannheim) showed the total amount of cells, whereas staining with ethidium bromide (Molecular Probes) revealed the number of dead microglia. Cells were counted in four observation fields using a fluorescence microscope (Zeiss Axioplan) with a 40× magnification (numerical aperture of 0.75).
Statistical analysis. The effect of treatments between groups was tested for statistical significance performing Mann–Whitney U test and χ2 test with SPSS software (SPSS, Chicago, IL). p < 0.05 denoted statistical significance.
Activated microglial cells reveal attenuated calcium signaling
To study the impact of microglial activation on receptor-mediated Ca 2+ signaling, we compared untreated microglial cells with those incubated with LPS (100 ng/ml for 24 hr). LPS is commonly used as a model agent in studies on inducible microglia–macrophage functions and bacterial CNS infections (Hanisch, 2001). The LPS dose was chosen because it also relates to a clinically relevant concentration range (Hanisch et al., 2001). For eliciting transient increases in the [Ca 2+]i, UTP and C5a were applied. Both have functional receptors on microglia, which are linked to calcium signaling, namely metabotropic purinergic receptors and CD88, respectively. In addition, initial experiments also used ATP as another and previously studied purinoreceptor ligand.
Addition of UTP (100 μm) or C5a (2 nM) for 30 sec to the bath solution induced reproducible transient increases in the fura-2-based calcium signal. An example of a trace (R 340/380) as recorded from an individual microglial cell under control conditions is given in Figure 1A. The signal intensity and the shape of the respective [Ca 2+]i transients compared with previous reports (Möller et al., 2000b). However, LPS-activated microglia generated much smaller [Ca 2+]i signals for both of the ligands (Figs. 1B, 2A).
Averaging the recorded R 340/380 traces for a large number of control (n = 799 from nine independent experiments) (Fig. 1C) and LPS-activated (n = 754 from 10 experiments) cells (Fig. 1D) revealed a significant suppression of the receptor-evoked calcium signals, as illustrated for the response to UTP (for statistics, see Table 1). For direct comparison, an overlay of the two graphs is illustrated in Figure 1E. Figure 1, G and H, is amplitude histograms of the Ca 2+ responses of untreated and LPS-activated cells. For each cell, the signal amplitude was calculated as the difference between the basal R 340/380 signal (average of the first 10 recorded image frames) and the respective peak level of the UTP-induced transient. According to their size, the amplitudes were then classified into ranges. As reflected by the shift of the amplitude distribution, activated microglia exhibited smaller UTP-evoked increases in the [Ca 2+]i (Table 1).
Microglial activation is accompanied by an increased resting [Ca 2+]i
Interestingly, a comparison of the [Ca 2+]i traces in Figure 1E revealed a marked increase in the basal level for the activated cells. Figure 1F shows the values of basal [Ca 2+]i in untreated and LPS-activated cells as averaged from the first 10 image frames of these traces (Table 1). Apparently, the reduction in receptor signaling efficacy was paired with a general elevation in the basal [Ca 2+]i. The smaller response amplitudes superimposed on the increased resting level in LPS-treated cells were, however, not attributable to technical limitations in recording higher R 340/380 values. The cells did not reach a ceiling level of fluorescence, because amplitudes with higher maximal ratios were recorded in control cells.
Surface expression of C5a receptors is not affected during microglial activation
Attenuated signal intensity could be attributable to a downregulation of the respective receptor. We thus determined whether receptor expression was altered during LPS treatment. We focused on the receptor for C5a because it consists of a single protein, CD88, allowing for a convenient expression analysis. In contrast, UTP binds to a heterogeneous population of metabotropic purinoreceptors (Ralevic and Burnstock, 1998). Immunocytochemical staining for CD88 revealed a punctuated distribution on microglia (Fig. 2B). Combination of biotin labeling of cell surface molecules with subsequent anti-CD88 immunoprecipitation revealed that neither the total nor the membrane-inserted amounts of CD88 were affected by the LPS treatment (Fig. 2C).
LPS-induced changes in calcium signaling and basal [Ca 2+]i can be reversed by treatment with BAPTA
To test for a correlation between the elevated basal [Ca 2+]i and the impaired calcium signaling, we treated control and LPS-activated cells with the intracellular calcium chelator BAPTA. We used 50 μm, a concentration efficient to affect LPS-induced release activity (shown below). Averaging the recorded R 340/380 traces of cells loaded with BAPTA only (n = 705 from eight independent experiments) (Fig. 3A) revealed no significant difference from untreated controls (Fig. 1C). The UTP response amplitude as an indicator of the signaling capacity was not altered (Fig. 3E, Table 1). This could be confirmed by also comparing the amplitude histograms of UTP responses in the presence and absence of BAPTA (compare with Fig. 3G). Thus, BAPTA did not impair calcium signaling.
In contrast, the averaged trace of cells treated with LPS in the presence of BAPTA (n = 591, 10 experiments) (Fig. 3B) showed a marked difference compared with cells exposed to LPS alone (Fig. 3C, Table 1). Cells treated with LPS in the presence of BAPTA did not show the otherwise LPS-inducible rise in basal [Ca 2+]i (Fig. 3D). The basal [Ca 2+]i equaled control conditions. Moreover, the LPS-induced shift toward smaller amplitudes (Fig. 3H) was much less prominent in the presence of BAPTA (Fig. 3F), indicating partially restored signaling ability. These results suggest that moderate chelation of intracellular calcium could, indeed, revert the activation-induced decrease in signaling efficacy.
Treatment with ionomycin can mimic the LPS-induced changes in calcium signaling and basal [Ca 2+]i
To test our hypothesis that the attenuated signaling ability in activated microglia was attributable to the concomitant rise in basal [Ca 2+]i, we induced an elevation of the basal [Ca 2+]i by applying the calcium ionophor ionomycin. We postulated that an ionomycin-mediated influx of calcium into the cytosol should then hamper the signaling capacity. UTP stimulation of ionomycin-treated cells should yield signaling amplitudes comparable with those of LPS-treated cells. Indeed, ionomycin loading did not only cause the expected increase in basal [Ca 2+]i (Fig. 4C, Table 1), as similarly seen with LPS treatment. Moreover, the averaged trace of the UTP-evoked [Ca 2+]i signal (n = 515, six experiments) (Fig. 4A) demonstrated a clear reduction of the response peak when compared with untreated controls (Fig. 4B, Table 1). Finally, histograms revealed a similar shift toward smaller signaling amplitudes, as it was observed with LPS-activated cells (Fig. 4D). Together, these results confirmed our assumption that elevated basal [Ca 2+]i as associating with microglial activation may lead to attenuated receptor-evoked calcium signaling.
LPS-induced NO and cytokine–chemokine release activities are reduced in the presence of BAPTA
Chronic elevation of the basal [Ca 2+]i after LPS activation could be a central event in the microglial activation process. Therefore, we tested whether microglial release activities in response to LPS challenge were influenced by the basal [Ca 2+]i level. We prevented the LPS-inducible elevation of basal [Ca 2+]i by BAPTA administration and determined the outcomes for the production of NO, as well as the release of certain cytokines and chemokines, i.e., characteristic executive features of activated microglia.
Stimulation with LPS led to the expected strong increase in NO production compared with untreated controls (Fig. 5A). BAPTA treatment by itself did not have any effect. However, in the presence of BAPTA, the LPS-induced NO production was significantly reduced in a concentration-dependent manner as of 10 μm (Fig. 5A). Indeed, the release under 50 μm BAPTA declined to control levels (Table 2).
Similarly, BAPTA affected the release of certain cytokines and chemokines (Fig. 5B). LPS caused the extracellular accumulation of the cytokines TNFα, IL-6, and IL-12, as well as of the chemokines KC (the rodent equivalent of GROα) and MIP-1α. Supernatants of untreated cultures were virtually devoid of any of these cytokines and chemokines. In the presence of BAPTA, the release of some factors was strongly suppressed, whereas that of others was only slightly attenuated. IL-12 appeared to be most and MIP-1α least affected by the BAPTA-mediated prevention of LPS-induced [Ca 2+]i elevation. The individual release reduction was similar at the two time points studied during an ongoing LPS–BAPTA treatment, i.e., 6 hr and 24 hr.
To exclude toxic BAPTA effects, metabolic activity was monitored using the WST-based viability assay (Fig. 5C). BAPTA, even at the highest concentration used for the calcium imaging (50 μm), did not show any change in the mitochondrial WST cleavage activity (Fig. 5C). Similarly, life–death staining of BAPTA AM-exposed microglia did not reveal impaired survival (Fig. 5D). In addition, examination of the microglial gross morphology did not indicate any BAPTA-induced alteration when compared with the corresponding control situations (Fig. 6). Thus, the observed decrease in release activities was unlikely to be related to a reduced amount or vitality of cells.
These results indicate that a rise in basal [Ca 2+]i during microglial activation is necessary for effective induction of NO, as well as cytokines and chemokines. Nevertheless, it is not a mandatory prerequisite for these features per se, because certain cytokines and chemokines were less affected by BAPTA than others. In addition, the fact that both soluble mediator “systems,” i.e., NO (as a diffusible gas) and cytokines–chemokines (as polypetides–proteins), were sensitive to BAPTA further points to some implication of [Ca 2+]i in the proximal steps of the induction cascades rather than an interference with later steps of release performance.
Treatment with ionomycin cannot induce microglial release
The ionomycin-induced rise in basal [Ca 2+]i had caused an LPS-like effect on the UTP signaling, i.e., it attenuated Ca 2+ responses (Fig. 4). Assuming that an elevation of the basal [Ca 2+]i is a crucial step toward microglial release activities, we also tested whether ionomycin treatment was sufficient to induce NO and cytokine–chemokine production. However, none of the tested concentrations was able to induce any release (Fig. 7A,B). This failure was not attributable to toxic effects, because mitochondrial WST cleavage remained unchanged (Fig. 7C). In addition, the survival rates of microglia in culture were found to be unaffected (Fig. 7D). Finally, microscopic inspection of the cells did not reveal signs of damage, although the morphology resembled the appearance of LPS-challenged microglia (Fig. 6). The results thus render unlikely that a lifting of the basal [Ca 2+]i alone could cause the production of NO or the initiation of cytokine–chemokine induction pathways.
Microglial activation is associated with an increase in the resting [Ca 2+]i
We found that exposure of microglia to LPS caused a significant increase in the resting [Ca 2+]i. Chronic elevations in the basal [Ca 2+]i level of up to 1 d have not yet been reported for microglia but have been described for other CNS cell types. In cortical neurons, amyloid β treatment leads to increased [Ca 2+]i and thereby to enhanced excitotoxicity (Mattson et al., 1992). Astrocytes from trisomy 16 mice, a model of human trisomy 21 (Down syndrome), show abnormal calcium homeostasis with drastically increased [Ca 2+]i at rest (Bambrick et al., 1997). A finding similar to our observation was reported for rat mesenteric arteries. Basal [Ca 2+]i was higher in vessels from LPS-treated animals compared with controls (Martinez et al., 1996). In contrast, treatment of the macrophage-like cell line P338D.1 with LPS for 8 hr rather led to a reduction in the resting [Ca 2+]i, whereas shorter activation periods had no consequence (Lowry et al., 1999). Moreover, LPS treatment of osteoblasts did not alter basal [Ca 2+]i (Tam et al., 1998). Apparently, the consequences of LPS signaling for the intracellular calcium are not uniform and may depend on the cell type.
Little is known as to the mechanism linking proximal receptor events of the LPS signaling to the regulation of calcium homeostasis. Conceivably, an elevated basal [Ca 2+]i could result from enhanced constitutive Ca 2+ influx, from a reduced activity of Ca 2+ extrusion pumps, or from a reduction in Ca 2+ buffer activity. Indeed, a chronically increased Ca 2+ influx can be triggered in microglia during massive depletion of cytoplasmic stores (Toescu et al., 1998). LPS in conjunction with LPS binding protein binds to the CD14–Toll-like receptor complex to induce cytosolic signaling and nuclear events (Schumann et al., 1990; Ulevitch and Tobias, 1995; Kirschning et al., 1998; Guha and Mackman, 2001). Several kinases and phosphorylation cascades have been shown to be recruited (Ulevitch and Tobias, 1995; Guha and Mackman, 2001; Hanisch et al., 2001). However, there is yet no evidence for direct LPS effects on Ca 2+ pumps.
Activated microglial cells show an attenuated calcium signaling
Whereas LPS treatment resulted in an increased basal [Ca 2+]i, the activated cells showed a decreased responsiveness to external stimulation with UTP. We described previously that microglial activation resulted in a reduced diversity of purinoreceptor signaling (Möller et al., 2000b). Now, we could correlate the lifting in basal [Ca 2+]i to a suppressed purinoreceptor signaling (Fig. 8).
Moreover, we also studied another microglial ligand-receptor system, i.e., C5a signaling through its receptor CD88 (Gasque et al., 1997; Möller et al., 1997; Nataf et al., 1999; Van Beek et al., 2000). Similar to purinergic signaling, C5a responses revealed the very same principle of regulation with elevated basal [Ca 2+]i and reduced amplitudes after treatment with LPS. In addition, CD88 as the sole receptor for C5a allowed for the convenient demonstration that the amount of membrane-expressed receptor protein was not altered by LPS treatment, suggesting that attenuated C5a responsiveness was not attributable to a decrease in receptors.
Adaptation of the calcium signaling behavior appears to be a widespread phenomenon. LPS treatment of C6 glioma cells led to an inhibition of serotonin- and thrombin-evoked signaling (Tawara et al., 1998). In osteoblasts, LPS activation altered parathyroid hormone- and prostaglandin E2-induced Ca 2+ influx and affected Ca 2+ release from stores, although these changes were not accompanied by a concomitant alteration of the resting calcium level (Kugaya et al., 1997; Tam et al., 1998).
Elevation of basal [Ca 2+]i as a central regulatory event in microglial activation
Could an elevation of the basal [Ca 2+]i be causally linked to the reduced ability of activated microglia to respond to an external stimulus with a [Ca 2+]i transient? We addressed this question in two experimental approaches. First, chelation of the free cytosolic calcium in LPS-activated microglia by intracellular presence of BAPTA did not only bring the [Ca 2+]i back to control level but also restored much of the purinoreceptor signaling capacity. Second, the opposite setting, namely an ionophore-imposed rise in [Ca 2+]i in naive cultures (no LPS treatment), impaired the ability for receptor-evoked calcium signaling, much in the same way as seen under LPS. In our model, calcium thus serves a “gating” function in the conversion of extracellular stimulations to cytosolic consequences, low basal [Ca 2+]i being permissive and increased levels elevating the threshold for a response to build up.
This link between calcium homeostasis and calcium signaling as seen in brain macrophages may even stand for a more general principle. Indeed, such an apparent coincidence was also described in B and T lymphocytes. Elevated basal [Ca 2+]i as seen in chronically activated B cells correlated with a reduced responsiveness to B cell receptor stimulation (Healy et al., 1997). It was further suggested that the amplitude and duration of [Ca 2+]i changes determine the differential recruitment of downstream signaling elements in these cells (Dolmetsch et al., 1997). Similarly, elevated basal [Ca 2+]i was reported in B cells of Igtransgenic mice along with refractory B cell receptor-mediated calcium mobilization and protein kinase activity (Benschopp et al., 2001). A causal relationship between elevated cytosolic free calcium and a suppressed calcium-mediated signal transduction has also been published recently for Jurkat (T) cells (Boldizsar et al., 2002).
Although the increase in basal [Ca 2+]i obviously affected the sensory limb of microglial functions as it relies on a receptor-triggered calcium signal generation, we also obtained evidence for outcomes on the executive properties of these cells.
Consequences of activation-associated changes in the [Ca 2+]i regulation
Calcium as a second-messenger system influences enzymes, ion channels, and cytoskeletal elements. It also exerts a control on gene transcription (Berridge, 2001). A significant portion of the executive features of activated microglia, i.e., the production of immunoregulatory and neuroregulatory cytokines and chemokines, depends on the de novo synthesis of proteins. We provide evidence for a differential regulation of cytokine–chemokine induction by [Ca 2+]i. Calcium buffering by BAPTA had an attenuating influence on five studied proteins, the outcome ranging from strong to moderate suppression. Similarly, the production of the diffusible mediator, NO was suppressed under BAPTA. Still, this pattern of individual downregulation indicated that the calcium chelation was not generally affecting the synthesis and release machineries (Fig. 8).
On the contrary, calcium did not dictate cytokine–chemokine induction by itself. Bypassing the LPS receptor system and just lifting [Ca 2+]i through an ionophore failed to trigger their release. For cytokines–chemokines, the signaling cascades from the LPS receptor system to their induction thus likely require more than a single route (Marriott et al., 1998; Hanisch et al., 2001). Our data indicate that calcium is not the only control element for the induction of cytokines–chemokines but rather assists in (or facilitates) this process. Ionomycin also failed in driving NO production. Apparently, the activities of NO-synthesizing enzymes are under a similar type of complex control. Calcium is needed for developing full activity but is unable to trigger any synthesis on its own. This assumption is supported by the fact that the “calcium-dependent” (calcium-inducible) version of NO synthases is, indeed, absent in microglia (Agullo et al., 1995).
However, increased basal [Ca 2+]i levels per se may not necessarily translate into always the same sensory or functional adjustments. In peritoneal macrophages, an experimental elevation of the resting [Ca 2+]i by ionophores did induce NO synthase and thus mimicked LPS (Raddassi et al., 1994). On the contrary, [Ca 2+]i elevation in articular chondrocytes led to an almost opposite effect. Ionophores did not trigger NO release but even inhibited an IL-1-induced NO production (Geng and Lotz, 1995).
Decreased ability to translate receptor stimulation into [Ca 2+]i transients could represent an intrinsic mechanism of activated cells to shut off sensitivity toward external influences. It could thereby indirectly attenuate additional executive reactions to extracellular signals, as far as they are under the control of [Ca 2+]i. In addition, downregulation of stimulus-evoked [Ca 2+]i increases may serve some protective function because an excessive [Ca 2+]i elevation may result in detrimental consequences for microglial cells and their primarily protective action.
This work was supported by German Research Foundation Grants Ho2205/1 (A.H.), GRK 238 (O.K.), and SFB 507 (U.-K.H. and H.K.). We thank Gerda Müller, Silke Fleischhauer, Jörg Scheffel, and Madlen Driesner for excellent technical assistance.
A.H. and O.K. contributed equally to this work.
U.-K.H. and H.K. contributed equally to this work.
Correspondence should be addressed to Dr. Helmut Kettenmann, Max Delbrück Center for Molecular Medicine, Cellular Neurosciences, Robert-Rössle-Strasse 10, D-13092 Berlin, Germany. E-mail:.
O. Kann's present address: Institute for Physiology, Humboldt University Berlin, Tucholskystrasse 2, D-10117 Berlin, Germany.
U.-K. Hanisch's present address: University of Applied Sciences Lausitz, Grossenhainer Strasse 57, D-01968 Senftenberg, Germany.
Copyright © 2003 Society for Neuroscience 0270-6474/03/234410-10$15.00/0