Neuroprotective role of bradykinin because of the attenuation of pro-inflammatory cytokine release from activated microglia


Address correspondence and reprint requests to Mami Noda, Laboratory of Pathophysiology, Graduate School of Pharmaceutical Sciences, Kyushu University, Fukuoka 812-8582, Japan. E-mail:


Bradykinin (BK) has been reported to be a mediator of brain damage in acute insults. Receptors for BK have been identified on microglia, the pathologic sensors of the brain. Here, we report that BK attenuated lipopolysaccharide (LPS)-induced release of tumor necrosis factor-alpha (TNF-α) and interleukin-1β from microglial cells, thus acting as an anti-inflammatory mediator in the brain. This effect was mimicked by raising intracellular cAMP or stimulating the prostanoid receptors EP2 and EP4, while it was abolished by a cAMP antagonist, a prostanoid receptor antagonist, or by an inhibitor of the inducible cyclooxygenase (cyclooxygenase-2). BK also enhanced formation of prostaglandin E2 and expression of microsomal prostaglandin E synthase. Expression of BK receptors and EP2/EP4 receptors were also enhanced. Using physiological techniques, we identified functional BK receptors not only in culture, but also in microglia from acute brain slices. BK reduced LPS-induced neuronal death in neuron–microglia co-cultures. This was probably mediated via microglia as it did not affect TNF-α-induced neuronal death in pure neuronal cultures. Our data imply that BK has anti-inflammatory and neuroprotective effects in the central nervous system by modulating microglial function.

Abbreviations used

6-isoproxy-9-osoxanthene-2-carboxylic acid




central nervous system




dibutyryl cyclic AMP


Dulbecco’s modified Eagle’s medium


fetal calf serum


N-2-hydroxyethylpiperazine-N′-2-ethansulphonic acid








microsomal prostaglandin E synthase


adenosine 3′,5′-cyclic monophosphorothioate


tumor necrosis factor-alpha

Bradykinin (BK) is an endogenous peptide which is produced via the kallikrein–kinin system and is widespread in the brain (Scicli et al. 1984; Chao et al. 1987; Walker et al. 1995; Raidoo and Bhoola 1998). The effects of BK are mediated by two types of receptors, B1 and B2 (Hall 1992, 1997). Both BK receptors belong to the family of seven transmembrane receptors and are coupled to G-proteins (Gq/11 and Gi/o). BK receptors are not only present on neurons (Delmas et al. 2002), but also on all types of glial cells, astrocytes, and oligodendrocytes (Gimpl et al. 1992; Hosli et al. 1992; Lin and Chuang 1992; Stephens et al. 1993a,b) and, as recently demonstrated, also on microglial cells (Noda et al. 2003, 2004). In astrocytes, BK stimulates the release of arachidonic acid (Burch and Kniss 1988; Kaeser et al. 1988) and induces the calcium-dependent release of glutamate and IL-6 expression (Parpura et al. 1994; Schwaninger et al. 1999).

After brain trauma or stroke, BK production is up-regulated and the increased BK levels lead to an increase in blood–brain barrier permeability and an accumulation of leukocytes (Abbott 2000; Lehmberg et al. 2003). BK is thus thought to be involved in secondary brain damage. By blocking BK receptors with specific antagonists, post-ischemic brain swelling after focal cerebral ischemia is reduced in concert with improved functional neuronal recovery (Zausinger et al. 2002). In contrast, in an animal model of global cerebral ischemia, BK receptor antagonists increased mortality (Lehmberg et al. 2003).

In the brain, microglial cells are considered as the pathologic response element (Perry et al. 1993; Kreutzberg 1996; Kim and de Vellis 2005). Microglial cells are dispersed throughout the entire central nervous system (CNS) and exhibit a ramified morphology under normal conditions. As recently shown by Davalos et al. (2005) and Nimmerjahn et al. (2005), microglial processes are highly dynamic in the intact brain, suggesting that microglial cells scan the brain parenchyma with their processes and potentially shield it from injury. Under pathologic conditions, such as a lesion, stroke, neurodegenerative disorders or tumor invasion, activated microglia migrate rapidly to the affected sites of the CNS. At the same time, microglial activation is accompanied by the release of immunocompetent molecules such as cytokines or chemokines, and other substances such as growth factors (Hanisch 2002). In culture, their activation can be stimulated by LPS, a cell wall component of Gram-negative bacteria. Treatment with LPS or cytokines such as interleukin-1β (IL-1β) has been shown to increase the expression of the B1 receptor in rat or rabbit tissue (Regoli et al. 1981; Galizzi et al. 1994; Nicolau et al. 1996), although in the brain this has not yet been studied.

Here, we provide evidence that BK receptors in microglia mediate anti-inflammatory or neuroprotective effects, namely the inhibition of LPS-induced release of tumor necrosis factor-alpha (TNF-α) and IL-1β.

Materials and methods

Cell culture

The study was approved by the Animal Research Committee of Kyushu University and Max-Delbrück Center for Molecular Medicine, and carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Rat microglial cells were isolated from mixed cultures of cerebrocortical cells in postnatal day 3 Wistar rats (Kyudo, Kumamoto, Japan) as described previously (Sastradipura et al. 1998; Noda et al. 2000). In brief, for rat microglia, the cerebral cortex was minced, and treated twice with papain (90 U) and DNase (2000 U) at 37°C for 15 min. Dissociated cells were seeded into 300 cm2 plastic flasks at a density of 107 per 300 cm2 in Eagle’s medium (MEM) with 0.17% NaHCO3 and 10% fetal calf serum (FCS), and maintained at 37°C in 10% CO2/90% air. Medium was changed twice per week. After 10–14 days, floating cells and weakly attached cells on the mixed primary cultured cell layer were obtained by gently shaking for 5–8 min. The resulting cell suspension was seeded on plastic dishes, with or without glass coverslips, or 96-well plates and allowed to adhere for 30 min at 37°C. Microglial cells were then isolated as strongly adhering cells after unattached cells were removed.

Mouse microglial cells were isolated from NMRI mice (Tierzucht Schönwalde, Schönwalde, Germany) or C57BL/6 mice (Kyudo, Kumamoto, Japan), as described previously (Prinz et al. 1999). In brief, cortical tissue was trypsinized for 2 min, dissociated with a fire-polished pipette and washed twice. Mixed glial cells were cultured for 9–12 days in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS, 2 mmol/L l-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin, with medium changes every third day. Microglial cells were then separated from the underlying astrocytic layer by gentle shaking of the flask for 1 h at 37°C in a shaker–incubator (100 rpm). The cells were seeded on glass coverslips in 35 mm dish. Cells were used for experiments 1–5 days after plating. The purity of both rat and mouse microglia, as measured by binding of isolectin, was >99%.

Neuronal cell cultures were prepared from cerebral cortices of embryonic day 14–16 (E14–16) C57BL/6 mice brains. The cortices were trypsinized together with DNaseI and dissociated with a fire-polished pipette and washed twice. Cells were plated on 4-well culture slides (BD Falcon, Bedford, MA, USA) that had been previously coated overnight with poly-d-lysine (50 μg/mL, Sigma, St Louis, MO, USA) and incubated in culture medium (high-glucose DMEM; Invitrogen, Carsbad, CA, USA) supplemented with 110 μg/mL pyruvic acid, 3.7 mg/mL NaHCO3, 5 μg/mL insulin, 2 mmol/L l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, 2% B27 supplement (Invitrogen). Cells were used after 7 days. The density of neuronal cells was about 5 × 104 cell/well.

Mixed neuron–microglia cultures were obtained by seeding mouse microglia onto a neuronal cell culture at a density of 5 × 104 cell/well with 10% microglial medium.

Assay of TNF-α, IL-1β, IL-6, and IL-10

Microglial cells isolated from Wistar rats were seeded in a 96-well plate at a density of 4 × 104 cells/well and were treated with or without BK and/or LPS for 4, 6, and 18 h for measuring TNF-α, IL-1β, IL-6, and IL-10, respectively, according to the time-dependent release of each cytokine. The amount of each cytokine released into the culture medium was measured using an ELISA Kit following the manufacturer’s protocol (Biosource, Camarillo, CA, USA). The absorbence at 450 nm was measured with a Microplate Reader (ImmunoMini NJ-2300, Nalge Nunc International, KK, Tokyo, Japan).

Microglial cell counting

The number of live microglial cells with or without drug application was checked by using Cell Counting Kit-8 (Dojindo, Kumamoto, Japan), according to the manufacturer’s instructions.

Assay of PGE2

Microglial cells isolated from postnatal day 3 Wistar rats were seeded in a 96-well plate at a density of 5 × 104 cell/well and were treated with or without BK and/or LPS for 24 h. The serum was reduced to 0.1%. Amount of PGE2 in the supernatant fluid was measured with a PGE2 EIA kit from Amersham (Piscataway, NJ, USA), according to the manufacturer’s instructions.


Rat primary cultured microglia were treated with or without 100–300 nmol/L BK or 100 ng/mL LPS, and fixed with 4% paraformaldehyde. The fixed cells were incubated with rabbit anti-microsomal prostaglandin E synthase (mPGES) polyclonal antibody (1/500, Cayman Chemical, Ann Arbor, MI, USA), goat polyclonal anti-B1 receptor (1/200, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA), or mouse monoclonal anti-B2 receptor (1/500, Research Diagnostics Inc., Flanders, NJ, USA) together with goat serum (1/10 volume, Cedarlane, Ontario, Canada) overnight at 4°C. After four washes, the cells were incubated with the fluorescence-conjugated secondary antibody together with goat serum (10% in volume) at 20–25°C for 1 h. Control cells were incubated with nonimmune rat IgG. For the identification of microglia in the staining of mPGES, the same cells were treated with FITC-labeled isolectin-B4 (1/200, Santa Cruz). For identification of neurons, monoclonal anti-MAP2 (1/1000, Sigma) was used. Each treatment was followed by the washing with PBS for a couple of times. Then the cells were mounted in the anti-fading medium Vectashield (Vector Laboratories, Burlingame, CA, USA). For mPGES, the cells were examined with a digital, high-resolution camera system (AxioCam, Carl Zeiss, Obenkochen, Germany) mounted to a fluorescence microscope (Axioscope2 plus, Carl Zeiss). For B1 and B2 receptors, series of images were sequentially acquired to avoid signal cross-over, and were examined with a confocal laser scanning microscope (LSM510META, Carl Zeiss).

SYBR green-based real-time quantitative RT-PCR

Total RNAs were prepared from 2 × 106 rat microglial cells and a rat brain with RNeasy™ RNA purification kit (QIAGEN, Valencia, CA, USA), according to the manufacturer’s protocol. First-strand cDNA synthesized from 1 μg total RNA with random hexamer primers was used as template for each reaction. SYBR Green-based real-time quantitative RT-PCR was performed as described (Aoki et al. 2002). 7700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) was used for the signal detection and the PCR was performed in 1× SYBR Green Master mix (Applied Biosystems) and 50 nmol/L of each primer. For standardization and quantification, rat β-actin was amplified simultaneously. Primer sequences were designed with Primer Express™ Software (Applied Biosystems). The following primer pairs were employed: 5′- CTGTGCCAAGTCAATCAAGC-3′ (upper, 387–407) and 5′-TCCCTTTCAGATCCCACTTCA-3′ (lower, 487–467) for amplification of rat EP1 receptor (GenBank accession No. NM_013100): 5′- TTGCTCTTCTGTTCTCTGCCG-3′ (upper, 538–558) and 5′-CAGCTGAAGGTATGCGGTCC-3′ (lower, 642–623) for amplification of rat EP2 receptor (GenBank accession No. NM_031088): 5′-ACTCGGTTGAAGCGCACAG-3′ (upper, 111–129) and 5′-CGACACAAGCAACATGGCC-3′ (lower, 238–220) for amplification of rat EP3 receptor (GenBank accession No. NM_012704): 5′-GTGACCATTCCCGCAGTGA-3′ (upper, 61–79) and 5′-ACAGCCAGCCCACATACCA-3′ (lower, 188–170) for amplification of rat EP4 receptor (GenBank accession No. NM_032076): 5′-ATCGCTGACAGGATGCAGAAG-3′ (upper, 925-945) 5′- AGAGCCACCAATCCACACAGA-3′ (lower, 1032–1012) for amplification of rat β-actin. PCR conditions were: 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. The threshold cycle of each gene was determined as the PCR cycle at which an increase in fluorescence was observed above the baseline signal in an amplification plot (Wada et al. 2000). The ‘normalized expression level of target’ (dCt) was calculated as the difference in threshold cycles for target and reference (β-actin). Subtraction of dCt for microglial cells from dCt for the rat brain provided the ddCt value. The formula, 2-ddCt, was used to calculate relative expression levels for microglial cells compared with the rat brain. To reduce possible error, RT-PCR reaction was performed three times and averaged 2-ddCt values were obtained.

Electrophysiological measurements

Whole-cell slice patch recordings were made as reported previously (Brockhaus et al. 1993, 1996). The forebrain hemispheres of young NMRI mice (aged postnatal day 6–9) were cut into 150 μm thin coronal slices using a vibratome (LEICA VT1000S, Leica, Nussloch, Germany). Slices were stored in 5% CO2 and 95% O2-gassed solution at 20–25°C prior to use. For the patch-clamp experiments, a slice was placed into a recording chamber mounted on the stage of a Zeiss upright microscope; the slice was held in place with a grid of nylon threads framed by a U-shaped platinum wire. The chamber was continuously perfused with external solution (in mmol/L: NaCl, 150; KCl, 5.4; CaCl2, 2; MgCl2, 1; N-2-hydroxyethylpiperazine-N′-2-ethansulphonic acid (HEPES), 5; and glucose, 10, pH 7.4). Substances were introduced by changing the perfusate. Cells in the corpus callosum near to or on the surface of the slice were viewed with 60× water immersion bright-field optics. While approaching a cell with the patch pipette, positive pressure was applied to the pipette, thereby blowing aside any material between the slice surface and the cell. The composition of patch-pipette was (in mmol/L): KCl, 130; MgCl2, 2; CaCl2, 0.5; ethyleneglycol-bis-N, N, N’, N’-tetraacetic acid, 4; HEPES, 11, Mg2ATP3, 2, pH 7.3.

Membrane currents were measured with the patch-clamp technique in whole-cell recording configuration (Hamill et al. 1981). Current signals were amplified with conventional electronics (EPC-9 amplifier, HEKA electronics, Lambrecht/Pfalz, Germany), and filtered at 0.1 kHz during recording. Measured data was filtered at 12 kHz by an interface connected to an AT-compatible computer system, which also served as stimulus generator.

Assessment of neuronal death

To induce inflammatory toxicity, LPS (10 μg/mL) or TNF-α (1 ng/mL) was applied for 48 h to induce neuronal toxicity in neuron–microglia cultures or neuron. BK at concentrations of 100 and 300 nmol/L were added to cells for 30 min before treatment of neuron–microglia cultures with LPS. Cells were fixed and stained with MAP2 antibodies as described above. The average number of MAP2-positive cells/mm2 was calculated by counting cells at a 200-fold magnification in four visual fields in each well, using a fluorescence microscope (Axioscope2 plus; Carl Zeiss) equipped with a digital camera system (AxioCam; Carl Zeiss).

Statistical analysis

Data are presented as mean ± SEM of 3–8 experiments. Statistical significance was determined by Student’s t-test or anova using origin program (Microcal, Northampton, MA, USA).

Drugs and reagents

Papain and DNase were purchased from Worthington Biochemical (Freehold, NJ, USA). DMEM was obtained from Nissui (Tokyo, Japan). FCS was from Hyclone Laboratories Inc. (Logan, UT, USA). Isolectin-B4, LPS, dibutyryl cyclic AMP (dbcAMP), cholera toxin, and 6-isoproxy-9-osoxanthene-2-carboxylic acid (AH6809) were purchased from Sigma. Bradykinin, and Des-Arg9-Bradykinin (bradykinin B1 receptor agonist) were from PEPTIDE Institute Inc. (Osaka, Japan). The cyclooxygenase-2 (COX-2) inhibitor, N-(2-Cyclohexyloxy-4-nitrophenyl)-Methanesulfonamide (NS398), was from Biomol Research Labs (Plymouth Meeting, PA, USA). The membrane-permeable cAMP antagonist adenosine 3′,5′-cyclic monophosphorothioate (RpcAMP) was from Calbiochem (San Diego, CA, USA). Specific EP2 and EP4 agonists (ONO-AE1-259-01 and ONO-AE1-329) were generous gifts from Ono Pharmaceutical Inc. (Osaka, Japan).


Bradykinin inhibits lipopolysaccharide-induced TNF-α release in cultured microglia

Activation of microglia in pathologic conditions stimulates the release of TNF-α (Brosnan et al. 1988; Fillit et al. 1991). When microglial cells were treated with LPS (100 ng/mL) for 24 h, we could measure TNF-α levels in the supernatant ranging between 1003 and 3241 pg/mL, while in untreated controls levels were very low (9 pg/mL). When BK (100 nmol/L to 1 μmol/L) was co-applied with LPS, TNF-α release was reduced in a dose-dependent manner (Fig. 1a). The relative amount of LPS-induced TNF-α release decreased to 60.4 ± 9.7% (n = 3) and 48.0 ± 11.1% (n = 3) by addition of 300 nmol/L and 1 μmol/L BK, respectively. The inhibition of LPS-induced TNF-α release was also observed after addition of a bradykinin B1 receptor agonist (Des-Arg9-Bradykinin: 1 μmol/L) leading to a reduction of TNF-α release to 53.5% ± 7.9% (n = 3). Application of BK alone did not significantly increase TNF-α levels in the supernatant even at concentrations up to 1 μmol/L (15–48 pg/mL).

Figure 1.

 Inhibition of lipopolysaccharide (LPS)-induced TNF-α release by bradykinin (BK). (a) The relative inhibition rates of LPS-induced TNF-α release by addition of 100, 300, and 1000 nmol/L BK or bradykinin B1 receptor agonist (B1 ago). *p < 0.05, **p < 0.01 (anova). (b) The number of live cells was not changed by the treatment of LPS with or without BK or B1 agonist. Drugs were added for 4 h.

Inhibition of LPS-induced TNF-α release by BK was not because of a decrease in the cell number, because the cell viability did not change by treatment of cells with both LPS and BK or B1 agonist (Fig. 1b).

Inhibition of LPS-induced TNF-α release by bradykinin was mimicked by an increase in intracellular cAMP and stimulation of prostanoid receptors

We investigated whether an experimentally induced increase in cAMP would mimic the effect of BK on the LPS-induced TNF-α release. We co-applied the membrane-permeable cAMP analogue, dbcAMP in concert with LPS. When microglial cells were treated with 100 ng/mL LPS, the release of TNF-α was increased from 30.5 ± 16.4 to 373.9 ± 18.9 pg/mL. Co-treatment of the cells with LPS and dbcAMP (1 mmol/L) decreased the LPS-induced TNF-α release to 55.7 ± 6.3 pg/mL (Fig. 2a), thus leading to a reduction of 85%. Co-application of cholera toxin, which has been reported to stimulate cAMP accumulation and PGE2 synthesis in murine macrophages (Burch et al. 1988), also dose-dependently inhibited the LPS-induced TNF-α release. TNF-α release induced by 100 ng/mL LPS was reduced from 980.5 ± 15.2 (n = 3) to 280 ± 19.2 (n = 3) and 194 ± 39.5 pg/mL (n = 3) in the presence of 100 ng/mL and 1 μg/mL cholera toxin, respectively (Fig. 2b). Conversely, the inhibition of LPS-induced TNF-α release by BK was prevented by 50 μmol/L RpcAMP, a membrane-permeable cAMP antagonist (Fig. 2c). TNF-α release induced by 100 ng/mL LPS was also inhibited by specific agonists for Gs-coupling prostanoid receptors, EP-2 and EP-4 (Fig. 2d), suggesting the activation of microglia by prostaglandins. The amount of TNF-α release was reduced from 2910 ± 12.6 to 263 ± 34 and to 2019 ± 73 pg/mL (n = 3) by EP-2 agonist (ONO-AE1-259-01, 1 μmol/L) and EP-4 agonist (ONO-AE1-329, 10 μmol/L), respectively. The involvement of prostaglandins and their receptors were confirmed by the facts that both COX-2 inhibitor (NS398, 10 μmol/L) and PGE2 antagonist (AH6809, 10 μmol/L) (Capehart and Biddulph 1991) antagonized the effect of BK on LPS-induced TNF-α release (Fig. 2c). The non-specific COX inhibitor indomethacin (10 μmol/L) had an effect similar to, but no greater than, that of NS398 (data not shown).

Figure 2.

 Inhibition of lipopolysaccharide (LPS)-induced TNF-α release by raising intracellular cAMP and via formation of prostaglandin and prostanoid receptors. (a) Co-treatment of microglial cells with LPS and dibutyryl cyclic AMP (dbcAMP), a membrane-permeable cAMP analogue, attenuated the LPS-induced TNF-α release. (b) Co-treatment of microglial cells with LPS and cholera toxin, which accumulates intracellular cAMP, also attenuated the LPS-induced TNF-α release. ***p < 0.005 (Student’s t-test). (c) BK-induced attenuation of LPS-induced TNF-α release was cancelled either by membrane-permeable cAMP-inhibitor (RpcAMP, 50 μmol/L), COX-2 inhibitor (NS398, 10 μmpl/L), or prostanoid receptor antagonist (AH6809, 10 μmol/L). Each experiment was performed individually, therefore each block of TNF-α release was compared directly with controls and expressed as % of control. *p < 0.05 (anova). (d) Specific agonists for Gs-coupling prostanoid receptors, EP2 (ONO-AE1-259-01; 1 μmol/L) and EP4 (ONO-AE1-329; 10 μmol/L), significantly inhibited the LPS-induced TNF-α release. ***p < 0.005 (Student’s t-test).

Bradykinin induced expression of microsomal PGE synthase and the release of PGE2

Prostaglandins are produced by activated microglia which express the mPGES (Ajmone-Cat et al. 2003; Ikeda-Matsuo et al. 2005), a terminal enzyme for PGE2 biosynthesis. LPS and BK have been reported to trigger the release of PGE2 (Minghetti and Levi 1995; Jenkins et al. 2003). BK was reported to stimulate phospholipase A2 to generate arachidonic acid (Welsh et al. 1988; Yanaga et al. 1991) and other cytokines (Rang et al. 1991), inducing long-term effects. To investigate whether BK induces the production of prostaglandin E2 (PGE2) in cultured microglia, we studied the level of mPGES expression and the release of PGE2.

Immunostaining of microglia using a specific antibody against mPGES showed strong expression of mPGES after application of 300 nmol/L BK and 100 ng/mL LPS for 24 h (not shown) or 42 h (Fig. 3a). The strong staining of mPGES was not observed when the cells were incubated together with 10 μmol/L NS398, a COX-2 inhibitor. As prostaglandins are produced following the sequential oxidation of arachidonic acid by COX and terminal PGE synthase, the result suggested that activation of PGE synthase needs the upstream formation of prostaglandin H2 induced by COX.

Figure 3.

 Increased expression of microsomal PGE synthase (mPGES) and PGE2 release by bradykinin and lipopolysaccharide (LPS). (a) Microglial cells were double-stained with isolectin-B4 (FITC-labeled: green) and anti-mPGES antibody (Alexa Fluor 594-labeled: red). Expression of mPGES in microglia was increased by treatment with 100 ng/mL LPS and 300 nmol/L bradykinin (BK) for 42 h, but not in the presence of 10 μmol/L NS398. (b) The amount of PGE2 released into the culture medium was measured by EIA. The relative amounts of PGE2 released after treatment for 24 h with 100, 300 nmol/L BK, and 1 μg/mL LPS were compared with that in non-treated microglia (control). (c) PGE2 release was further increased after application of LPS and BK. LPS, together with BK induced more release of PGE2 than that induced by LPS alone. The B1 agonist (Des-Arg9-Bradykinin) mimicked the increasing effect of BK on LPS-induced PGE2 release. *p < 0.05, **p < 0.01 (significance from control, anova). #p < 0.05 (significance from 100 nmol/L B1 agonist together with LPS, anova).

The released PGE2 was measured using the supernatant of microglial cultures after application of BK (100 and 300 nmol/L) for 24 h. Application of 1 μg/mL LPS for the same period served as a reference. The relative amount of PGE2 compared with that of untreated microglia (control) increased to 147.7% ± 16.3% (n = 4) by 100 nmol/L BK, to 171.5% ± 28.3% (n = 4) by 300 nmol/L BK, and to 175.5% ± 1.2% by 1 μg/mL LPS (n = 3), respectively (Fig. 3b). The release of PGE2 was increased further by co-application of LPS and BK. When the microglial cells were treated with 100 ng/mL LPS for 24 h, the release of PGE2 increased from 47.3 ± 13.4 to 124 ± 22.5 pg/mL (n = 3). After co-treatment of microglia with LPS and BK (1 μmol/L), the release of PGE2 was further increased to 174 ± 25.3 pg/mL. Des-Arg9-Bradykinin, a B1 agonist, mimicked the dose-dependent effect of BK on LPS-induced PGE2 release; the amount of PGE2 release were 100.1 ± 29.1 pg/mL (n = 3), 174.0 ± 31.7 pg/mL (n = 3), and 186.0 ± 18.2 pg/mL (n = 3) with 100, 300, and 1000 nmol/L B1 agonist, respectively, together with LPS (Fig. 3c). These results indicate that BK and LPS stimulate PGE2 synthesis and that B1 receptors may play a more important role than B2.

Up-regulation of prostanoid EP2/EP4 receptors and bradykinin B1/B2 receptors in rat microglia

Microglia express the EP2/EP4 subtypes of prostanoid receptors (Caggiano and Kraig 1999; Patrizio et al. 2000). We examined the possibility that the expression of the receptors is affected by BK or LPS. Prostanoid receptor levels were determined by quantitative RT-PCR analysis in the microglial cells. EP1, EP2, and EP4 receptors could be identified, although at lower levels than those found in total brain extract. After treatment of the cells with 300 nmol/L BK or 100 ng/mL LPS for 24 h, the expression of EP2 receptor increased significantly; the relative expression level compared with that of rat brain increased from 4.1 ± 0.1 to 5.9 ± 0.1 and 41.3 ± 6.7 (n = 3) with BK and LPS, respectively. EP4 receptor expression increased after the treatment with LPS (from 0.16 ± 0.01 to 1.4 ± 0.2 (n = 3)), but not with BK (Fig. 4). EP1 receptor was not significantly increased by BK (from 0.28 ± 0.05 to 0.56 ± 0.05, n = 3) or by LPS (from 0.28 ± 0.05 to 0.31 ± 0.02, n = 3).

Figure 4.

 Quantitative RT-PCR of prostanoid receptors in rat microglia and brain. The expression level of each prostanoid receptor (EP1, EP2, EP3, and EP4) mRNA was normalized to the level of each receptor in rat brain. The relative values of each receptor in untreated microglia (control), or microglia treated for 24 h with bradykinin (BK: 300 nmol/L) or lipopolysaccharide (100 ng/mL) are shown. Difference from controls: *p < 0.05, **p < 0.01 (anova).

Our previous quantitative RT-PCR analysis showed that microglial B1 receptor mRNAs were up-regulated after the treatment with 100nmol/L BK for 24 h (Noda et al. 2003). In the present study, we also showed that expression of B1 and B2 receptor proteins was increased by BK treatment as revealed by immunolabeling with specific antibodies against B1 and B2 receptors. In untreated microglia, expression of both B1 and B2 receptors was low. However, after the treatment of the cells with 300 nmol/L BK for 24 h, strong immunolabeling of both B1 and B2 receptors was observed. LPS (100 ng/mL) had a similar effect on BK in increasing BK receptor expression (Fig. 5).

Figure 5.

 Immunofluorescence of B1 and B2 receptors in rat microglia. Microglial cells were double-stained with anti-B1 receptor (FITC labeled: green) and anti-B2 receptor antibody (Alexa Fluor 594-labeled: red) in untreated (control), lipopolysaccharide (100 ng/mL)- or BK (300 nmol/L)-treated microglia (24 h drug treatment).

Bradykinin inhibits LPS-induced IL-1β release but not IL-6 and IL-10 release

We also analyzed whether BK affected the release of other inflammatory cytokines, namely IL-1β, IL-6, and IL-10. BK application alone did not induce IL-1β release. In contrast, LPS-induced IL-1β release was significantly reduced in the presence of BK (Fig. 6a). BK (100 nmol/L) decreased the LPS-induced IL-1ß release by 48%. Higher concentrations of BK were not more effective, indicating that 100 nmol/L was already a saturating concentration for inhibiting IL-1ß release. On the contrary, 300 nmol/L or 1 μmol/L BK affected neither LPS-induced IL-6 (Fig. 6b) nor LPS-induced IL-10 release (Fig. 6c).

Figure 6.

 Bradykinin (BK) inhibited lipopolysaccharide (LPS)-induced IL-1β release, but not LPS-induced IL-6 or IL-10 release. (a) IL-1β release was measured after 6-h incubation with or without LPS and BK. (b) IL-6 release was measured after 6-h incubation with or without LPS and BK. (c) IL-10 release was measured after 18-h incubation with or without LPS and BK. The incubation time was determined by the time-dependent assay for each cytokine with or without BK and was set at the time with maximum release. *p < 0.05, **p < 0.01, ***p < 0.005 (difference from controls; Student’s t-test).

Bradykinin receptors in microglia were also functional in situ

To test for the presence of functional BK receptors in situ, we studied membrane current responses in microglial cells in acute slices from mouse brain. We focused on a subpopulation of microglial cells which were found on the surface of the slice in the mouse corpus callosum at postnatal days 6–9. The microglial cells do not represent the resting form, but are characterized by an amoeboid shape with a soma diameter in the range of 10–20 μm and migrate actively at this developmental stage (Brockhaus et al. 1993; Fig. 7a). Previously, we have reported that BK induced an outward current in cultured rat microglia presumably because of the activation of Ca2+-dependent K+ channels (Noda et al. 2003). We therefore tested for the presence of BK-induced membrane currents in microglia in brain slice. The mean value of the resting membrane potential was −39 ± 15 mV (lowest: −20 mV, highest: −72 mV). At a holding potential of −20 mV, application of 1 μmol/L BK induced an outward current with a mean amplitude of 14.8 ± 0.01 pA in 15 BK-responsive cells out of 52 patched cells (Fig. 7b). By clamping the membrane to a series of potentials ranging from −120 to 60 mV during the BK response (Figs 7b and c), we determined the current–voltage curve of the BK-induced current (Fig. 7d). The subtracted currents showed the reversal potential of −60 mV, suggesting the induction of a conductance that was substantially permeant to K+ (calculated EK and ECl were about −83 and 10 mV, respectively). The conductance increase (measured between 20 and 40 mV) was 0.32 ± 0.17 nS. The current–voltage relationship showed a slight outward rectification.

Figure 7.

 Bradykinin-induced outward currents in mice microglia in situ. (a) Phase contrast microscope image of microglia in a brain slice patched with a pipette (×600). (b) Currents recorded in response to 100 ms depolarizing (0, 20, 40, and 60 mV) and hyperpolarizing (−40, −60, −80, −100, and −120 mV) voltage commands from a holding potential of −20 mV (inset below). This clamp protocol was repetitively applied in 6-s intervals. Bradykinin (BK; 1 μmol/L) was applied as indicated by the horizontal bar. An outward current was induced at the holding potential and the currents in response to the voltage step increased in amplitude. (c) Currents induced by a series of voltage steps before application (solid circles) and during application (solid diamonds) indicated in (b) are shown on a faster time scale. (d) Current–voltage relationship for BK-induced currents. Current amplitudes measured before application of BK (solid circles) were subtracted from those during application of BK (solid diamonds) and plotted against membrane voltage (solid squares).

Protective effect of bradykinin on LPS-induced neuronal damage

To investigate whether BK has a protective effect on neurons by attenuating LPS-induced TNF-α release, we compared the effects of LPS alone and LPS together with BK in neuron–microglia co-cultures. Although cortical neurons were quite resistant to LPS compared with those in substantia nigra (Kim et al. 2000), the MAP2-positive cells decreased significantly after incubation of the microglia–cortical neuron co-culture in 10 μg/mL LPS for 48 h (Fig. 8a). However, pre-treatment with BK protected MAP2-positive cells. The number of the MAP2-positive cells decreased from 317 ± 10 to 157 ± 12 cells/mm2 by LPS, but showed a small decrease to 269 ± 9 and 296 ± 12 cells/mm2 after co-incubation with 100 and 300 nmol/L BK, respectively (n = 8 each) (Fig. 8b). These results suggest that BK protects neurons against the toxic effect of LPS application, namely decrease in TNF-α and IL-1β release. As neurons also possess B2 receptors, to exclude a direct effect of BK on the neurons these experiments were repeated using primary neurons cultured in the absence of microglia. In the CNS, microglia is the major LPS-responsive cell and specific activation of innate immunity by LPS is through a Toll-like receptor 4 (TLR4)-dependent pathway (Lehnardt et al. 2003), i.e., LPS-induced neurotoxicity is because of the activation of TLR4 in microglia and subsequent release of cytotoxic substances. In fact, we also observed the neuronal cell death induced by LPS in neuron–microglia co-culture but not without microglia (not shown). Therefore we simulated the neuronal death induced by LPS by adding high concentration of TNF-α, as LPS-induced cytokine release was much greater in TNF-α (from ∼9 pg/mL in control condition to 1000–3000 pg/mL) (Figs 2b and d) than that in IL-1β (from 40 pg/mL in control condition to 75 pg/mL) (Fig. 6). While TNF-α reduced the number of neurons, co-application of BK did not change the neuronal death rate (Fig. 8c). Treatment with TNF-α for 24 h decreased the number of neuronal cells from 272 ± 17 to 165 ± 14 cells/mm2. Addition of 100 or 300 nmol/L BK 30 min prior to TNF-α treatment did not significantly affect this response; the number of MAP2-positive cells was 186 ± 19 and 184 ± 14 cells/mm2, respectively (n = 4 each) (Fig. 8d). This result suggests that the neuroprotective effect of BK is not because of an effect on the neurons, but instead is mediated by the microglial cells.

Figure 8.

 Protective effect of bradykinin on lipopolysaccharide (LPS)-induced neuronal death mediated via microglia. (a) Neuron–microglia co-cultured cells were treated with 10 μg/mL LPS for 48 h. Bradykinin (BK: 100 or 300 nmol/L) were applied 30 min before LPS treatment. After 48 h, the cells were immunostained with MAP2, a neuron-specific antibody. (b) The numbers of MAP2-positive neurons in (a) were counted. (c, d) Without microglia, BK did not rescue the neuronal cell death induced by 1 ng/mL TNF-α. Numerical values in (b), (d) are expressed as number of MAP2-positive cells in 1 mm2. *p < 0.05, **p < 0.01, and ***p < 0.005 (difference from controls; anova). #p < 0.05 and ##p < 0.01 (difference from LPS; anova).


The present study demonstrates that BK may have a neuroprotective role in the CNS, mediated by glial cells. During inflammatory diseases, kinins (including BK) might well be produced in the brain, or access the brain from the periphery, and thereby could exert inflammatory effects on neurons analogous to its effects in the periphery. However, when we focus on the effects of BK on glial cells – mainly microglia in the present study – then it appears that, on the contrary, BK has an anti-inflammatory role.

Anti-inflammatory effects of bradykinin mediated by microglia

Unexpectedly, BK inhibited LPS-induced TNF-α release from microglia (Fig. 1), even although it did not affect resting TNF-α release. It also inhibited LPS-induced release of IL-1β (Fig. 6a). As the release of these cytokines is intimately associated with elements of the acute phase immune responses, including fever, and are responsible for LPS-induced fever (Saigusa 1990; Luheshi et al. 1997) and with some forms of neurodegeneration (Boka et al. 1994; Mogi et al. 1994; Mehlhorn et al. 2000; Nagatsu et al. 2000), this implies that the effect of BK on microglial cells might promote anti-inflammatory and neuroprotective effects. Indeed, we show (in Fig. 8) that BK does exert a neuroprotective action, and that this depends on the presence of microglial cells. These inhibitory effects seems contrary to the view that BK is a well-known pro-inflammatory mediator, although a protective role of BK has previously been reported in cardiac and renal system (Pinto et al. 2000; Bascands et al. 2003), and in the retina (Yasuyoshi et al. 2000). Therefore, it is likely that BK may work as an anti-inflammatory mediator in the CNS by modulating the brain’s immune system.

Mechanism of the neuroprotective effects of BK

Experiments on other cells suggest that a likely mechanism for the inhibition of LPS-induced secretion by BK might be through the production of PGE2 and subsequent increase in intracellular cAMP (Pyne et al. 1997; Caggiano and Kriag 1999; Webb et al. 2003). Our results are consistent with this mechanism. Thus, BK inhibition of LPS-induced TNF-α secretion was replicated by cAMP or an EP2 agonist, and was antagonized by a COX-2 inhibitor or a prostanoid receptor antagonist (Fig. 2). It has previously been reported that microglia in vitro constitutively express COX-1 and LPS induces COX-2 expression (Hoozemans et al. 2002). As the effect of the COX-2 inhibitor (NS398) was almost equal to that of COX-1/COX2 inhibitor indomethacin, it seems likely that COX-2 plays the major role in the effect of BK.

Quantitative RT-PCR analysis showed that Gs-coupled EP2 receptor was the main prostanoid receptor for PGE2 in microglia (Fig. 4). The EP2 receptor has been reported to exert a cAMP-dependent neuroprotective role in cerebral ischemia (McCullough et al. 2004). BK receptors link to Gq/11 and Gi/o, and do not activate adenylyl cyclase (Pal-Ghosh et al. 2003) although Gs-mediated signaling was also reported in (Fang et al. 2005). Instead, in microglia, the BK receptor appears to modulate the production of PGE2/EP2 to increase intracellular cAMP indirectly, rather than coupling to adenylyl cyclase directly. This scenario is supported by several previous observations. Thus, a dose-dependent production of cAMP by PGE2 and EP2/EP4-specific agonists has previously been observed in cultured rat microglia and trabecular meshwork cells (Caggiano and Kriag 1999; Webb et al. 2003), and incubation of the cells with PGE2 in combination with BK resulted in a three- to fivefold enhancement of PGE2-stimulated cAMP production (Webb et al. 2003). Further, elevation of intracellular cAMP has been reported to reduce both LPS-induced TNF-α and IL-1β production in rat cultured microglia (Caggiano and Kraig 1999) and to prevent LPS-induced neuronal death (Kim et al. 2002). In our experiments, we confirmed that dbcAMP inhibited LPS-induced TFN-α release and that the membrane-permeable cAMP antagonist, RpcAMP, prevented the inhibitory effect of BK on this response (Fig. 2). The source of PGE2 would be also astrocytes in vivo, because the attenuation of LPS-induce TFN-α release by BK was much greater in mixed glial culture (not shown).

Additionally, we observed that both B1 and B2 receptors are up-regulated by LPS and BK as determined immunocytochemically (Fig. 5). These effects are probably mediated by nuclear factor-κB (NFκB) and PGE2, respectively. Thus, in other cells, up-regulation of B1 receptors by LPS and TNF-α mediated by the activation nuclear factor-κB (NFκB) has previously been reported (Sardi et al. 1999; Passos et al. 2004), while up-regulation of B2 receptors was induced by BK-induced PGE2 production (Castano et al. 1998).

Taken together, a signaling schema explaining how BK might attenuate the LPS-induced TNF-α release is shown in Fig. 9. We propose that the principal effect of BK is to enhance both prostaglandin synthesis and prostanoid receptor expression, thereby enhancing microglial cAMP production, which in turn inhibits LPS-induced TGF-α release; and that this may be amplified by upregulation of BK receptors. We surmise that the same mechanism might also apply to the inhibition of IL-1β release, although we have not investigated this in detail in the present experiments. This provides an amplified negative feed-back mechanism on TNF-α and IL-1β production in microglia, thereby accounting for the neuroprotective action of BK shown in Fig. 8.

Figure 9.

 A proposed schema for the inhibitory effects of bradykinin on lipopolysaccharide (LPS)-induced TNF-α and IL-1β. Bradykinin (BK) activates B1 receptors (although B2 receptors were not excluded yet), activates Gq/11, mobilizes intracellular Ca2+ (inducing a Ca2+-dependent K+ current IK(Ca)), and releases PGE2. This induces a negative-feedback inhibition of LPS-induced TNF-α and IL-1β release, mediated by cyclic AMP, and a positive-feedback increase in B2 receptor expression (Castano et al. 1998). LPS-induced TNF-α release also up-regulates EP2/EP4 receptors and BK receptors (B1 + B2 receptors), probably through the activation of NF-κB. LPS itself may exert a negative-feedback inhibition of its own TNF-α release via PGE2 and prostanoid receptors and cAMP. However, it was not proved because the amount of TNF-α release in the presence of cAMP antagonist was not greater than that induced by LPS alone.

It seems likely that the principal BK receptor involved in this anti-inflammatory effect is the B1 receptor, because: (i) B1 receptors are rarely expressed in non-traumatized tissues but highly expressed in activated immune cells, (ii) inhibition of LPS-induced TNF-α release was also induced by B1 agonist (Fig. 1a) and was substantially cancelled by B1 antagonist (not shown), (iii) production of PGE2 was largely, not completely, dependent on B1 receptor activation (Fig. 3c), and (iv) BK-induced microglial migration (via activation of Ca2+-dependent K+ channels) was also because of the activation of B1 receptors but not B2 receptors (Ifuku et al. 2005). Thus, even although B1 receptors might be responsible for nerve injury-induced neuropathic pain in the periphery (Ferreira et al. 2005), in the CNS, microglial B1 receptors confer the opposite (anti-inflammatory and neuroprotective) action.

In conclusion, we suggest that – in addition to its pro-inflammatory action – BK can also work as an anti-inflammatory or neuroprotective mediator in the brain, through its effect on glial cells. The key signaling pathway appears to be the production of prostaglandine E2 and the consequential increase in intracellular cAMP in the microglia. As BK receptors are expressed in almost all kinds of cells, it can evoke complex cascades in different cells within the central nervous system. However, if we can clarify the inflammatory and anti-inflammatory cascades of BK, it might be possible to manipulate inflammation and inflammatory diseases and could thereby assist in identifying new therapeutic targets. Our data provide essential information for understanding the anti-inflammatory action of kinins within the CNS, which could contribute to the treatments against infectious diseases, inflammation, trauma, or neurodegeneration accompanying inflammation.


We thank Prof. D. A. Brown (University College London, UK), Prof. H. Higashida (Kanazawa University, Japan), and Prof. H. Nakanishi (Kyushu University, Japan) for valuable suggestions. This work was supported by Grants-in Aid for Scientific Research of Japan Society for Promotion of Science, Research Grant in Priority Area Research of the Ministry of Education, Culture, Sports, Science and Technology, Japan, Kyushu University Foundation, Grants-in-Aid for Scientific Research of the Ministry of Health, Labour and Welfare, Japan and the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO).