Inhibition of lipopolysaccharide-induced cyclooxygenase-2, tumor necrosis factor-α and [Ca2+]i responses in human microglia by the peripheral benzodiazepine receptor ligand PK11195


Address correspondence and reprint requests to Dr James G. McLarnon, Department of Pharmacology and Therapeutics, Faculty of Medicine, 2176 Health Sciences Mall, The University of British Columbia, Vancouver, BC Canada, V6T 1Z3. E-mail:


The anti-inflammatory actions of the mitochondrial peripheral benzodiazepine receptor (PBR) agonist PK11195 [1-(2-chloro- phenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinoline-carboxamide] were investigated in human microglia. Application of the microglial inflammatory stimulus lipopolysaccharide (LPS, at 100 ng/mL for 3 h), induced enhancement of the expressions of the inducible enzyme, cyclooxygenase-2 (COX-2) and the pro-inflammatory cytokine, tumor necrosis factor-α (TNF-α). PK11195 (at 50 µm) significantly inhibited the LPS-induced up-regulation of both inflammatory factors; at a lower concentration of PK11195 (2 µm) expression of TNF-α, but not COX-2, was reduced. Production of both factors, using immunocytochemistry for COX-2 and ELISA for TNF-α, was markedly reduced with 50 µm of PK11195 added to LPS solution. Acute application of LPS induced a transient increase in intracellular Ca2+[Ca2+]i exhibiting both a slow development and recovery in kinetic behavior. This increase in [Ca2+]i consisted primarily of a Ca2+ influx component accompanied by a smaller mobilization from intracellular Ca2+ stores. In the presence of PK11195, the amplitude of the [Ca2+]i response induced by LPS was reduced by 54%. Another mitochondrial agent cyclosporin A (CsA), which also acts at the permeability transition pore (PTP) of mitochondrial membrane but at a site different from the PBR, was ineffective in reducing either the LPS-induced expression of COX-2 and TNF-α or the endotoxin increase in [Ca2+]i. These results indicate that the mitochondrial effector PK11195 is a specific and effective agent for inhibiting LPS-induced microglial expressions of COX-2 and TNF-α and that modulation of Ca2+-mediated signaling pathways could be involved in the anti-inflammatory actions.

Abbreviations used

intracellular calcium concentration




cyclosporin A




Dulbecco's modified Eagle's medium


endoplasmic reticulum


glyceraldehyde-3-phosphate dehydrogenase




phosphate-buffered saline


peripheral benzodiazepine receptor




standard physiological solution


permeability transition pore


store-operated channel


tumor necrosis factor-α

Microglia are resident cells of the CNS that serve a functional role as scavenger cells similar to that of peripheral macrophages. These cells have been implicated as key mediators of inflammation in the CNS, during which microglia proliferate and contribute to the inflammatory process by phagocytosis and secretion of cytokines, eicosanoids and other agents (McGeer and McGeer 1995). Although inflammation of the CNS is intended as a homeostatic means to contain damage to neural tissue, such damage may occur from the inflammatory process itself (McGeer and McGeer 1995; Rogers and Griffin 1998). In this case microglial responses to inflammatory stimuli in the brain may actually exacerbate, rather than inhibit, neuronal damage.

Recent work has established that mitochondria regulate cellular signaling in a diversity of cell types (Simpson and Russell 1998). At present, however, the roles of mitochondria in modulating microglial responses to inflammatory stimuli are not well studied. One key component of the mitochondrial membrane is the permeability transition pore (PTP), which serves as a high conductance, non-selective channel. This pore is composed of an assembly of both inner and outer mitochondrial proteins, including cyclophilin D and the peripheral benzodiazepine receptor (PBR) (Halestrap et al. 1998). The PBR is an integral component of the PTP complex associated with the outer mitochondrial membrane. These receptors have been located in peripheral tissue such as monocytes (Canat et al. 1993) and in brain in human (Banati et al. 1997) and murine (Park et al. 1996) microglia. Agonists of the PBR have been reported to suppress immune responses through the modulation of monocyte proliferation and secretion of a variety of cytokines including IL-1β, TNF-α, and IL-6 (Zavala et al. 1990; Taupin et al. 1991). Certain benzodiazepine derivatives such as the isoquinolinecarboxamide compound PK11195 [1-(2-chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinoline-carboxamide] act as specific ligands of the PBR (Kinnally et al. 1993; Hirsch et al. 1998) and act to modulate immune responses in vitro and in vivo (Taupin et al. 1991; Vowinckel et al. 1997; Torres et al. 1999; Klegeris et al. 2000).

In this work we have used LPS, a potent activator of microglia (Buttini et al. 1996), to induce expressions of two factors present in chronic brain inflammation, the enzyme cyclooxygenase-2 (COX-2) and the pro-inflammatory cytokine tumor necrosis factor-α (TNF-α). We then investigated the actions of PK11195 to modify LPS-evoked expressions of COX-2 and TNF-α; additionally immunocytochemistry (for COX-2) and ELISA (for TNF-α) assays have been applied to study production. We have also examined the actions of PK11195 on LPS-induced mobilization of Ca2+. The results show PK11195 inhibits LPS enhancement of microglial expression and production of both COX-2 and TNF-α and also reduces the amplitudes of LPS evoked [Ca2+]i responses. Another mitochondrial effector cyclosporin A (CsA), which acts at a different mitochondrial site from the PBR (Nicolli et al. 1996), was ineffective in blocking LPS-mediated increases in expressions of COX-2 and TNF-α and altering LPS-induced mobilization of calcium.

Materials and methods

Preparation and culture of human microglia

The use of embryonic human tissues was approved by the Clinical Screening Committee for Human Subjects of the University of British Columbia. The preparation of human microglia cultures has been described previously (Satoh and Kim 1994; Satoh et al. 1995). In brief, embryonic brain tissues of 12–18 weeks gestation were incubated in phosphate-buffered saline (PBS) containing 0.25% trypsin and DNase I (40 µg/mL) for 30 min at 37°C. Dissociation of the tissue into single cells was achieved by repeated pipetting. Dissociated cells were then cultured in T75 flasks in a medium consisting of Dulbecco's modified Eagle's medium (DMEM) containing 5% horse serum, 5 mg/mL glucose, 20 µg/mL gentamicin and 2.5 µg/mL amphotericin B. Following 7–10 days of growth in culture flasks, freely floating microglia were harvested from a medium of mixed cell cultures and subsequently plated at relatively low density on poly-d-lysine coated glass coverslips. Immunostaining with the cell specific markers CD11b or ricinus communis agglutinin-1 ensured that the purity of cultures was in excess of 98%.

Ca2+ spectrofluorometry

Some of the procedures used in Ca2+-sensitive fluorescence microscopy in this laboratory have been previously described (Wang et al. 2000; Khoo et al. 2001; McLarnon et al. 2001). Cultured human microglia were loaded for 25 min with the Ca2+ indicator Fura-2 acetoxymethylester (Fura-2/AM). The dye was applied with pluronic acid (both at concentrations of 1 µm) to standard physiological solution (PSS). Following a wash in dye-free solution (5 min), the coverslips were placed on the stage of a Zeiss Axiovert inverted microscope containing a 40 × objective lens. Alternating excitation wavelengths (340/380 nm) of UV light were applied at intervals of 8 s and fluorescence signals were obtained at 510 nm of emission light (bandwidth of 40 nm) from 8 to 20 cells in the field of view. Signals were acquired from a digital camera (DVC-1310, DVC Co. Austin, TX, USA) and were processed using an imaging system (Empix, Mississauga, ON, USA) to determine ratios of the 340 and 380 nm intensities which have been used as quantitative measures of fluorescence levels in this work. A change in F340/F380 of 0.1 would correspond to an approximate change in absolute [Ca2+]i of 90 nm. All studies were done at room temperature (20–22°C).

Treatment of cultured human microglia

Cultured human microglia were plated in six-well multiplates and bathed in DMEM containing 5% horse serum at 37°C. Serum-free medium was applied 48 h prior to application of agents. After two days in serum-free medium most cells showed a ramified morphology which may indicate a low level of activation. Pretreatment with PK11195 (50 µm) or CsA (5 µm) were for durations of 30 min; subsequent activation by LPS (100 ng/mL) exposure was carried out in the presence of the agents for a further 3 h.

Reverse-transcriptase polymerase chain reaction (RT-PCR)

COX-2 and TNF-α mRNAs were detected using RT-PCR. Total RNAs were isolated using TRIzol (Gibco-BRL, Gaithersburg, MD, USA) and complimentary DNA (cDNA) synthesis was performed using M-MLV reverse transcriptase (Gibco-BRL). Human specific primers for COX-2 and TNF-α were used and their corresponding sequences were as follows: COX-2 sense primer 5′-TTCAAATGAGATTGTGGGAAAATTGCT-3′, antisense primer 5′-AGATCATCTCTGCCTGAGTATCTT-3′, TNF-α sense primer 5′-CAAAGTAGACCTGCCCAGAC-3′, antisense primer 5′-GACCTCTCTCTAATCAGCCC-3′. PCR measurements were carried out using the GeneAmp thermal cycler (Applied Biosystems, Foster City, CA, USA). PCR products sizes were 304 bp for COX-2 and 490 bp for TNF-α, respectively. PCR conditions were as follows: initial denaturation at 95°C for 8 min followed by 38 and 34 cycles (COX-2 and TNF-α, respectively) of denaturation at 95°C for 35 s, annealing at 55°C for 1 min and extension at 72°C for 1 min. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a reaction standard. The amplified DNAs were identified using 1.5% agarose gels containing ethidium bromide and visualized under UV light.

Immunocytochemistry for COX-2

The production of COX-2 was measured from human microglia exposed to control, LPS alone (100 ng/mL), LPS plus PK11195 (50 µm) and PK11195 alone. The same procedures and concentrations were used as for the RT-PCR assay except the treatment time was 12 h. After treatment, cells were fixed in 4% paraformaldehyde in 0.1 m PBS, washed in PBS and permeabilized in 0.2% Triton X-100 in 0.1 m PBS for 20 min. The cells were then incubated in rabbit anti-human COX-2 (1 : 200 dilution) at 4°C for 48 h. Following washing in PBS, cells were incubated in Alexa Fluor 488 anti-rabbit IgG secondary antibody (1 : 100, Molecular Probes, Eugene, OR, USA) at room temperature for 1 h in the dark. After a wash in PBS, cells were incubated in 4′,6‘-diaminodino-2-phenylindole (DAPI, Molecular Probes) at 1 µg/mL in PBS to visualize nuclei and to determine total cell numbers. After rinsing in PBS, coverslips were mounted onto glass slides using gelvatol, examined under a Zeiss fluorescence microscope and photographed using a cooled CCD camera. Stained cells were counted in a minimum of four parallel fields and the ratio of COX-2 positive cells to total cell number per field (at × 200 magnification) was calculated.

ELISA assay for TNF-α

An ELISA kit specific for human TNF-α (R & D Systems, Minneapolis, MN, USA) was used to measure production of the cytokine. These kits are capable of detecting low levels of TNF-α (to 4.4 pg/mL). Treatments were for 6 h and included control, LPS alone (at 100 ng/mL), LPS + PK11195 (at 50 µm) and PK11195 alone. Supernatants were then collected for immediate assay.

Cell viability

In all experimental procedures used in this work, we observed no evident effects on viability of microglia. This point was further assessed using DAPI staining as described above (see Immunocytochemistry for COX-2). Cells with round and intact nuclei were considered as normal cells. However, cells with condensed and fragmented nuclei were considered as damaged cells. Using a Zeiss Axiophot inverted fluorescence microscope, no nuclear condensation and fragmentation were observed indicating no loss of cell viability under any of the experimental conditions.

Solutions and reagents

For spectrofluorescence studies, the normal physiological saline solution (PSS) contained (in mm): NaCl (126), KCl (5), MgCl2 (1.2), d-glucose (10), HEPES (10) and CaCl2 (1) adjusted to a pH of 7.4. In some experiments, Ca2+-free PSS was utilized where EGTA at 1 mm was added to normal PSS in the absence of any CaCl2. To depolarize cell membranes, a modified solution containing a low concentration of Cl was used in some experiments; this solution consisted of (in mm): Na+ gluconate (126), K2SO4 (5), MgSO4 (1.2), HEPES (10), d-glucose (10), CaCl2 (1) adjusted to a pH of 7.4. The fluorescent indicator fura-2AM and pluronic acid were purchased from Molecular Probes and the agents CsA, PK11195, SKF96365, LPS and ionomycin were obtained from Sigma (St. Louis, MO, USA).


Values obtained from multiple cells are reported as mean ± standard error of mean (SEM) calculated from a selected number (‘n’) of cells. In references to representative experiments, values for mean and ‘n’ refer only to cells included in the representative experiment. Statistical significance (p≤0.05) was determined using the Student's t-test.


Effects of PK11195 on microglial expression of COX-2 and TNF-α

The initial studies investigated the effects of PK11195 on the expression of COX-2 and TNF-α by human microglia. Both of these agents have been documented as pro-inflammatory factors in the brain and are up-regulated in neurodegenerative diseases (McGeer and McGeer 1995; Rogers and Griffin 1998). Lipopolysaccharide (LPS), a potent activator of microglia, was applied (at 100 ng/mL for a period of 3 h) in three separate experiments. The results from one of these experiments are presented in Fig. 1 and the RT-PCR analysis shows that LPS induced expressions of both COX-2 and TNF-α (lane 2, Fig. 1). Unstimulated control cells showed no evident expression of either COX-2 or TNF-α (lane 1, Fig. 1) indicating no constitutive expression of either inflammatory factor. In order to assess possible anti-inflammatory actions of PK11195, the compound was applied (50 µm for 30 min) prior to and including the 3 h treatment with LPS. As shown in Fig. 1 (lane 4) the cells pre-treated with PK11195 showed marked reductions in levels of both COX-2 and TNF-α. PK11195 applied alone (lane 3) had no effect to increase either of the inflammatory factors. The same results, that are complete inhibition by PK11195 of the expressions of COX-2 and TNF-α, were obtained from the two additional studies.

Figure 1.

(a) Lipopolysaccharide (LPS)-induced expressions of cyclooxygenase-2 (COX-2) and tumor necrosis factor-α (TNF-α) in human microglia. LPS was applied (100 ng/mL) alone or with 30 min pre-treatment using either PK11195 (50 µm) or cyclosporin A (CsA; 5 µm). Lane 1: control; lane 2: LPS; lane 3: PK11195; lane 4: PK11195 + LPS, lane 5: CsA; lane 6: CsA + LPS. (b) Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a reaction standard.

In one experiment the effects of lower concentrations of PK11195 on microglial expression of the inflammatory factors was also studied. The results with PK11195 treatment (concentrations of 2, 10 and 50 µm) are shown in Fig. 2 for expressions of TNF-α and COX-2. PK11195 inhibited expressions of TNF-α in a dose-dependent manner with effects evident even at 2 µm. A concentration of 10 µm PK11195 appeared to be close to threshold for inhibition of COX-2 with no evident block at the lowest concentration of 2 µm.

Figure 2.

(a) Lipopolysaccharide (LPS)-induced cyclooxygenase-2 (COX-2) and tumor necrosis factor-α (TNF-α) expressions with different concentrations of PK11195. Three different concentrations of PK11195 were applied in this study; 2, 10 and 50 µm lane 1: Control; lane 2: LPS; lane 3: PK11195 at 50 µm; lane 4: PK11195 (at 50 µm) + LPS; lane 5: PK11195 (at 10 µm) + LPS; lane 6: PK11195 (at 2 µm) + LPS. (b) Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a reaction standard.

In summary, the results of the RT-PCR study show that activation of the mitochondrial peripheral benzodiazepine receptor by PK11195 markedly inhibits microglial expressions of both COX-2 and TNF-α.

Immunocytochemistry: COX-2 expression is inhibited by PK11195

We also examined the effects of PK11195 on LPS-induced production of COX-2 protein. Human microglia, incubated in serum-free medium were used as control. As shown in Fig. 3(a), no stained microglia were evident in control; overall only 0.2 ± 0.5% of COX-2 positive microglia were found in control (Fig. 3e). The majority of microglia had a morphology characterized by thin, long processes and relatively small cell bodies (see inset of Fig. 3a). Treatment of cells with LPS (100 ng/mL for 12 h) resulted in an intense expression of COX-2 (Fig. 3b) and over one-half of total cells were positive for COX-2 (54.2 ± 3.2%, Fig. 3e). COX-2 immunoreactivity was observed in both the perinuclear and cytoplasmic regions. Immunoreactive microglia showed short processes and cytoplasmic swelling, a characteristic morphology of activated cells (see inset of Fig. 3b). In microglia treated with LPS plus PK11195 there was no evident expression of COX-2 protein (Fig. 3c). Overall, when PK11195 was added to the LPS solution only 3.3 ± 1.2% of total cells stained positive for COX-2 (Fig. 3e). Furthermore, cells showed a similar morphology (inset of Fig. 3c) to those in control (inset of Fig. 3a). PK11195, applied in the absence of LPS, had no effect to induce COX-2 expression (Figs 3d and e) and exhibited characteristic morphology (inset of Fig. 3d) of control microglia. Under all experimental conditions microglia displayed intact nuclei as demonstrated using DAPI staining.

Figure 3.

Cyclooxygenase-2 (COX-2). (a) Human microglia were incubated in serum-free medium which served as control. The inset shows a morphology typical of ramified microglia. (b) LPS treatment of cells induced an intense expression of COX-2. The inset shows treated cells with ameboid morphology. (c) Lipopolysaccharide (LPS) plus PK11195 treatment inhibited production of COX-2. (d) Application of PK11195 in the absence of LPS. The insets for both (c) and (d) show cell morphology similar to control. The green and blue indicate staining for COX-2 and DAPI positive nuclei, respectively. (e) The percentage of COX-2 positive microglia relative to total cells are shown for (a–d). Insets show the phase contrast photomicrograph of each panel. Data are means ± SEM from four independent experiments. *Denotes significant difference compared with control and **denotes significant difference compared with LPS. Scale bar is 50 µm in (a–d) and 25 µm in insets.

ELISA assay for effects of PK11195 on LPS-induced production of TNF-α

A single study was undertaken in order to determine if PK11195 was effective in diminishing the production of TNF-α induced by LPS. As shown in Fig. 4, the incubation of human microglia with LPS (100 ng/mL for 6 h) caused a large increase in production of TNF-α relative to control. The small value for control would indicate a low constitutive microglial production of the cytokine. In the presence of PK11195, added with LPS, a strong inhibitory effect on production of TNF-α was evident. Treatment with PK11195 alone had no effect to alter production of the cytokine.

Figure 4.

Lipopolysaccharide (LPS)-induced microglial production of tumor necrosis factor-α (TNF-α). LPS (at 100 ng/mL) applied alone yielded a high production of TNF-α relative to control. Addition of PK11195 with LPS considerably inhibited the production of the cytokine. PK11195 alone did not increase the production of TNF-α.

Effects of cyclosporin A on microglial expression of COX-2 and TNF-α

It was of interest to determine if cyclosporin A (CsA), an immunosuppressive compound with actions on mitochondria, could also inhibit LPS-induced microglial expressions of COX-2 and TNF-α. This agent acts on the permeability transition pore of mitochondria but at a different site from that of the PBR binding site of PK11195 (Nicolli et al. 1996). A concentration of CsA of 5 µm was generally employed in these studies; a lower concentration of 1 µm was used in several experiments. It should be noted that CsA exhibits non-selective actions on tissue including inhibition of calcineurin in T lymphocytes (Fruman et al. 1992). The results from one study are presented in Fig. 1 and show that CsA pre-treatment was totally ineffectual in blocking the expressions of COX-2 or TNF-α (lane 6) induced by a treatment for 3 h with LPS. In two additional studies, CsA also showed no effects to inhibit microglial expressions of either of the two inflammatory factors. Also, no change in cell viability was evident with treatment of CsA applied alone, or together with LPS. Thus the RT-PCR experiments yield the interesting result that although PK11195 and CsA act at sites associated with the permeability transition pore, only one of these, PK11195, shows evident anti-inflammatory activity.

Effects of acute applications of LPS on intracellular calcium

The following studies investigated if PK11195 inhibition of LPS-induced microglial expression of TNF-α and COX-2 could be due to actions directed at the endotoxin-mediated mobilization of Ca2+. The initial aim was to characterize the specific Ca2+-mediated pathways involved in LPS activation of microglia; subsequent experiments then examined PK11195 effects on the LPS mobilization of Ca2+.

The effects of acute applications of LPS (100 ng/mL) on [Ca2+]i were studied over periods up to 20 min. A typical response is shown in Fig. 5(a) where LPS elicited a slowly developing, transient rise in [Ca2+]i. Once the peak of the response was reached, [Ca2+]i declined gradually back to baseline levels over a course of several minutes. The amplitude of this response was 0.25 ratio units (n = 23 cells); overall the mean amplitude of the peak [Ca2+]i induced by LPS was 0.24 ± 0.01 ratio units (n = 337 cells).

Figure 5.

LPS-induced changes in [Ca2+]i. (a) Response to LPS in Ca2+-containing PSS (n = 23). (b) Response to LPS in Ca2+-free PSS (n = 22).

In order to determine differential contributions between influx and intracellular sources for the [Ca2+]i rise induced by LPS, studies were carried out with extracellular Ca2+ removed (Ca2+-free PSS) during the application of LPS. As shown in the representative trace (mean response from 22 cells, Fig. 5b), LPS was still able to induce a transient increase in [Ca2+]i in the absence of extracellular Ca2+. However, the amplitude of the LPS response recorded in Ca2+-free PSS (0.12 ratio units) was reduced from the response elicited in the presence of extracellular Ca2+ (Fig. 5a). Overall, the mean from all experiments was 0.10 ± 0.02 ratio units (n = 161 cells) which was significantly lower (p < 0.05) than the mean amplitude in Ca2+-PSS. These results indicate that the primary contribution to the LPS response was due to entry of Ca2+ with a smaller component attributable to actions mediated by some intracellular source; the latter would account for approximately 40% of the total LPS response. This point was also studied with LPS application, in both Ca2+-containing and Ca2+-free PSS, to the same cells. The protocol was to initially attain a constant increase of [Ca2+]i with LPS applied in Ca2+-free PSS then exchange solution to LPS in standard PSS. As shown in Fig. 6 (mean response from 24 cells), the introduction of standard PSS caused a rapid increase in the slope of the change in [Ca2+]i with time. To substantiate that the increase in slope was due to influx, low Cl solution was applied near the peak of the LPS response. Previous work has established that this maneuver depolarizes microglia since anion channels in these cells regulate membrane potential (Wang et al. 2000). In essence, it was suggested that cellular depolarization reduces the driving force for entry of Ca2+. As shown in Fig. 6, the introduction of low Cl PSS caused a rapid decrease in [Ca2+]i indicating inhibition of Ca2+ influx; a similar result was obtained with application of Ca2+-free PSS (data not shown).

Figure 6.

Lipopolysaccharide (LPS) applied to same cells in Ca2+-free and Ca2+ standard physiological solution (PSS). Following attainment of a constant rate of increase in [Ca2+]i in Ca2+-free PSS, bath solution was exchanged to Ca2+-PSS. The dashed line indicates extrapolation of the initial slope in Ca2+-free PSS. The response shown is mean from 24 cells.

One possible source for the intracellular component of the [Ca2+]i response could be depletion from endoplasmic reticulum (ER) stores. This point was investigated by initially depleting ER stores with ATP (at 100 µm). The protocol used Ca2+-free PSS in order to prevent replenishing of stores after ATP stimulation. The results from a representative experiment are shown in Fig. 7(a) (mean response from 25 cells). Following recovery from the ATP response, application of LPS elicited a slow, progressive increase in [Ca2+]i with an amplitude of 0.06 ratio units. Control experiments, with no application of LPS following ATP, showed steady levels of [Ca2+]i over similar durations of recording (Fig. 7b, mean response from 33 cells). Overall (n = 126), the mean amplitude of the [Ca2+]i response induced by LPS following ATP stimulation was 0.06 ± 0.02. This amplitude was smaller than recorded for the effects of LPS in Ca2+-free PSS with no ATP pre-treatment, however, the difference was not significant (p ≥ 0.05). These results indicate that depletion of ER had little or no contribution to the increase in [Ca2+]i evoked by LPS in Ca2+-free solution (see Discussion).

Figure 7.

Lipopolysaccharide (LPS) responses after ATP and with SKF96365. (a) An ATP response was elicited in Ca2+-free PSS to deplete endoplasmic reticulum (ER) stores and following recovery LPS was then applied (mean response from 25 cells). (b) Control with no application of LPS after stimulation with ATP (mean response from 33 cells). (c) LPS was applied in Ca2+ PSS and SKF96365 was added (at 50 µm) near the peak of the response (mean from 36 cells).

The data presented in Fig. 5 indicate that the LPS-induced increase in Ca2+ mobilization is primarily due to influx. An important pathway mediating Ca2+ entry in human (McLarnon et al. 2000; Wang et al. 2000) and rodent (Toescu et al. 1998) microglia is store-operated channels (SOC); voltage-dependent Ca2+ channels seem minimally expressed in these cells (Eder 1998). Although specific blockers of SOC have not been identified, the compound SKF96365 would be expected to inhibit the increase in [Ca2+]i if entry was mediated by SOC (Toescu et al. 1998). As shown in the representative trace (n = 36) of Fig. 7(c), SKF96365, applied at the peak of the LPS response, had no effect to reduce [Ca2+]i. A similar result was obtained from two additional experiments (overall, n = 85) indicating that LPS did not activate SOC-mediated influx of Ca2+.

Effects of PK11195 pre-treatment on LPS-induced Ca2+ responses

The next set of experiments was designed to test if PK11195 altered LPS mediated increases in [Ca2+]i and possibly correlate modulation of Ca2+ mobilization with effects of the PBR ligand on cellular expressions of COX-2 and TNF-α. The procedure was to incubate human microglia with PK11195 for 12 min (pre-treatment) prior to application of LPS (with maintained PK11195). Each pre-treatment study was accompanied by a corresponding control experiment, performed on the same day, where LPS was applied to cells in the absence of PK11195. A concentration of 1 µm of PK11195 was employed in these experiments since higher levels of the compound (≥ 5 µm) had small effects to increase levels of [Ca2+]i which could interfere with the measurement of the subsequent response to LPS.

The results from a representative pair of experiments are presented in Fig. 8. As shown in Fig. 8(b), cells pre-treated and maintained in PK11195 solution exhibited a [Ca2+]i response (n = 27) to LPS with an amplitude approximately 50% of that elicited in control (Fig. 8a, n = 12). Overall, the mean amplitude of LPS responses generated from cells pre-treated with PK11195 was significantly reduced (p < 0.05) to 46 ± 8% of the mean amplitude seen in control cells (n = 82 cells in control; n = 57 cells with PK11195 pre-treatment).

Figure 8.

PK11195 effects on lipopolysaccharide (LPS)-induced responses. (a) Increase in [Ca2+]i with LPS in control. Response is mean from 12 cells. (b) Increase in [Ca2+]i with LPS in cells pre-treated with PK11195. Response is mean from 27 cells.

Effects of cyclosporin A pre-treatment on LPS-induced Ca2+ responses

Since CsA was found ineffective in reducing LPS-mediated increases in expressions of COX-2 and TNF-α (Fig. 1), it was of interest to determine if this immunosuppressive agent had any effect on the LPS-induced rise in [Ca2+]i. Preliminary experiments showed CsA, applied at 1 µm as pre-treatment and maintained for the duration of the experiment, had no effect to alter the amplitude of [Ca2+]i induced by LPS in the absence of CsA (data not shown). All subsequent studies then used CsA at a concentration of 5 µm. As shown in the representative pair of experiments, the amplitude of the LPS-induced increase in [Ca2+]i evoked in cells pre-treated with CsA (n = 19, Fig. 9b) was unchanged from that recorded in control (LPS, in the absence of CsA, n = 19, Fig. 9a). Overall, there was no significant difference (p > 0.05) in amplitudes of LPS responses elicited in control (n = 62) and after CsA pre-treatment (n = 55); the ratio of amplitudes (CsA pre-treatment/control) was 98 ± 4%.

Figure 9.

Cyclosporin A (CsA; at 5 µm) effects on lipopolysaccharide (LPS)-induced responses. (a) Increase in [Ca2+]i with LPS in control. Response is mean from 19 cells. (b) Increase in [Ca2+]i with LPS in cells pre-treated with CsA. Response is mean from 19 cells.


The principal finding in this work is that PK11195 causes a significant reduction in the LPS-induced expressions of COX-2 and TNF-α in human microglia. At all concentrations employed, from 2 to 50 µm, PK11195 reduced microglial expression of TNF-α. Concentrations in excess of 10 µm of PK11195 were effective in inhibiting the expression of COX-2. No constitutive expression of either TNF-α or COX-2 was evident in the absence of LPS treatment suggesting that the human microglia were in a relatively inactive functional state.

PK11195, applied at 50 µm, was also effective in inhibiting the LPS induced production of both COX-2 and TNF-α. In the former case, immunocytochemistry assay provided a quantitative measurement showing significant PK11195 block of COX-2 production (Fig. 3). At present, effects of PK11195 to suppress processes mediated by COX-2 enzymatic activity, such as production of prostaglandins, require study.

Limitations in amounts of human tissue precluded extensive measurements on microglial production of TNF-α. However, the results from an ELISA assay showed PK11195 had a marked effect to reduce cellular production of the cytokine (Fig. 4) in agreement with the RT-PCR data on expression. The present results differ from those reported using LPS stimulation of the human monocytic THP-1 cell line (Klegeris et al. 2000). In THP-1 cells PK11195, applied at concentrations up to 50 µm, had no effect to alter LPS mediated secretion of TNF-α. The contrasting results may reflect differences in signal transduction pathways between human microglia and the THP cell line. Another possibility is that altered treatment times with PK11195 could lead to differences in efficacy of inhibition. Interestingly, recent work has shown that production of TNF-α accompanies expression of the cytokine in LPS stimulated HMO6 cells, a recently developed microglial human cell line (Nagai et al. 2001).

The present results suggest that ligand binding to the PBR of the mitochondrial permeability transition pore inhibits LPS-mediated activation of microglia. The specificity of this action was evident in the RT-PCR study in the lack of effect of CsA to reduce microglial expressions of either COX-2 or TNF-α. CsA is an immunosuppressive compound which binds to the PTP but at a site on the inner membrane of the pore (Nicolli et al. 1996). However, this compound acts non-selectively in the CNS including binding to the Ca2+-dependent phosphatase calcineurin (Liu et al. 1991). In any event, interactions of CsA with mitochondria, or non-selective binding to other CNS components, had no evident effect on LPS stimulation of microglia.

The results from the imaging experiments show that PK11195 pre-treatment (at 1 µm) inhibited LPS-mediated responses in [Ca2+]i. This finding, that a low concentration of PK11195 reduced Ca2+-dependent LPS responses, could indicate that Ca2+-mediated signaling pathways were involved in microglial enhancement in expressions of TNF-α or COX-2. This conclusion was supported by the lack of effect of CsA to alter [Ca2+]i in addition to showing no inhibition of the LPS-induced inflammatory factors. However, it is likely that Ca2+-independent intracellular signaling pathways also contribute to the LPS responses in microglia (Buttini et al. 1996).

The effects of LPS to enhance levels of [Ca2+]i were due to a primary action on influx of Ca2+ (Fig. 5). A smaller component was mediated by some intracellular source since LPS also increased [Ca2+]i in Ca2+-free PSS. At present, specific identification of either the influx pathway or the intracellular source is not possible. In the former case, entry of Ca2+ was inhibited with low Cl PSS likely due to effects mediated by cellular depolarization. However, SKF96365 had no effects in blocking the LPS-induced entry of Ca2+ indicating that SOC were not involved. Although SKF96365 shows non-selective actions in tissue (Merritt et al. 1990), this compound (at 50 µm) inhibits SOC in a variety of cells including microglia. In terms of the intracellular component of the LPS response, the endotoxin, added following application of ATP in Ca2+-free solution, was still able to elicit an increase in [Ca2+]i. This result could suggest that depletion from ER stores was not involved but this conclusion would require the proviso that ATP to have fully depleted ER stores.

LPS responses have previously been recorded from rat microglia using a protocol with two successive applications of the endotoxin (Bader et al. 1994). The initial response of a rapid, transient change in [Ca2+]i was suggested to arise from a caffeine-sensitive intracellular store and the second response due to a La3+-sensitive influx pathway. The identification of the former pathway was based solely on block of LPS responses with cell preincubation with caffeine and not direct effects of caffeine to alter [Ca2+]i; such stores seem not generally reported in rodent cells and human microglia do not show transient changes in [Ca2+]i with application of caffeine (unpublished data). It should also be noted that LPS was applied to rat microglia at 50 µg/mL (Bader et al. 1994) which was 500 times the level employed in the present study.

The results from this work, together with results derived from other studies (Banati et al. 1997; Vowinckel et al. 1997), suggest that binding of agonists to the PBR site on mitochondrial outer membrane of microglia could serve an anti-inflammatory function in the CNS. Although the present data suggest the involvement of Ca2+ mobilization in these actions, future studies will be required to elucidate signaling pathways linking the PBR with microglial production of inflammatory factors. In addition, detailed work is necessary to define the physiological role of the PBR and characterize putative endogenous ligands for the PBR in brain.


This work was supported by grants from the Heart and Stroke Foundation of British Columbia and Yukon (JGM) and from KOSEF/BDRC, Ajou University (SUK).