Cyanobacterial Microcystis aeruginosa Lipopolysaccharide Elicits Release of Superoxide Anion, Thromboxane B 2 , Cytokines, Chemokines, and Matrix Metalloproteinase-9 by Rat Microglia

Microcystis aeruginosa ( M. aeruginosa ) is a cosmopolitan Gram-negative cyanobacterium that may contaminate freshwater by releasing toxins, such as lipopolysaccharide (LPS) during aquatic blooms, affecting environmental and human health. The putative toxic effects of cyanobacterial LPS on brain microglia, a glial cell type that constitutes the main leukocyte-dependent source of reactive oxygen species in the central nervous system, are presently unknown. We tested the hypothesis that in vitro concentration- and time-dependent exposure to M. aeruginosa LPS strain UTCC 299 would activate rat microglia and the concomitant generation of superoxide anion (O 2 2 ). After a 17-h exposure of microglia to M.aeruginosa and neurodegeneration.

An increasing body of literature suggests that massive growth of the Gram-negative cyanobacteria in water reservoirs has become an issue of concern for both animal and human health (Kuiper-Goodman et al., 1999;Sivonen and Jones, 1999) because these photosynthetic prokaryotic organisms may produce toxins such as cyclic hepatotoxic peptides (e.g., microcystins), neurotoxic alkaloids (e.g., anatoxin-a), and lipopolysaccharides (LPS), which may contaminate drinking water (Hitzfeld et al., 2000;Wiegand and Pflugmacher, 2005). In particular, if cyanobacterial LPS has not been effectively removed during water treatment (Rapala et al., 2002), humans may become exposed to LPS by several potential routes (Stewart et al., 2006): The oral route, via either drinking, recreational water, or the consumption of foods; through skin and respiratory exposure, by inhalation of aerosolized LPS during recreation or showering; or via the circulatory system, as in haemodialysis (Anderson et al., 2002). Since the first isolation of LPS from the cyanobacterium Anacystis nidulans (Weise et al., 1970), research on cyanobacterial LPS has established that while in some cases its structure appears to be similar to that of Gram-negative bacteria, there is also considerable variability in the chemistry of lipid A, the bioactive constituent of LPS, which is responsible for the biological responses (Bernardova et al., 2008;Sivonen and Jones, 1999;Snyder et al., 2009). In view of the role of LPS in Gram-negative septicaemia and endotoxin shock (Opal, 2010;Peleg and Hooper, 2010) as well as the high sensitivity of humans to LPS (Anderson et al., 2002), it is surprising that the toxicology of cyanobacterial LPS has received such limited attention (Stewart et al., 2006). Interestingly, cyanobacterial LPS may demonstrate differential toxicity: Thus while Synechococcus sp. LPS appears to be nontoxic to rodents (Schmidt et al., 1980), in contrast, Anabaena flos-aquae, A. cylindrica, Oscillatoria tenuis, O. brevisothe, and Microcystis aeruginosa LPS were reported to be lethal to mice (Keleti and Sykora, 1982;Raziuddin et al., 1983).
Systemic Gram-negative LPS may affect the blood-brain barrier (BBB) directly (Banks and Erickson, 2010) or enter the brain through regions with defective BBB function and then activate brain microglia, the macrophage of the brain immune system (Rock et al., 2004), which participate in neuroinflammation (Cunningham et al., 2005). Extensive research over the past 2 decades has shown that when microglia are activated by either in vivo or in vitro exposure to Gramnegative LPS (Rock et al., 2004;Ransohoff and Perry, 2009), inflammatory mediators may be released (for review, see Mayer, 1998) including reactive oxygen species, e.g., O 2 À (Colton and Gilbert, 1987) which may cause neuronal injury (Banati et al., 1993;Mayer et al., 1999) and progressive neurodegeneration (Perry et al., 2010;Qin et al., 2007). To our knowledge, no studies have been reported on the effects of cyanobacterial LPS on brain microglia generation of O 2 À . Although the chemistry of LPS isolated from several strains of the cosmopolitan water-bloom forming cyanobacterium M. aeruginosa has been investigated (Jurgens et al., 1989;Martin et al., 1989;Raziuddin et al., 1983), there has been limited research on the toxicology of M. aeruginosa LPS. Some studies have implicated the immune system in the pathobiology of M. aeruginosa LPS. Thus, crude M. aeruginosa extracts, perhaps containing LPS as well as the cyclic heptapeptide microcystins, were shown to induce secretion of the cytokines IL-1 and tumor necrosis factor-a (TNF-a) from macrophages in vitro and mice in vivo (Nakano et al., 1989(Nakano et al., , 1991 as well as affecting several parameters of the immune response in mice (Shen et al., 2003).
The purpose of this investigation was to establish whether M. aeruginosa LPS might activate neonatal rat microglia and stimulate release of O 2 À , a reactive oxygen species that has been hypothesized to cause injury to the brain (Halliwell, 1992). Our preliminary studies (Aldulescu et al., 2009;Mayer et al., 2008;Patel et al., 2010) as well as this study provide experimental support for our working hypothesis, namely that M. aeruginosa LPS not only potentiates but also attenuates O 2 À generation by rat brain microglia in vitro in a concentration-dependent manner. Furthermore, the present findings show that inhibition of O 2 À generation appears to be concomitant with a progressive increase in the release of lactic dehydrogenase (LDH), as well as that of several proinflammatory mediators: thromboxane B 2 (TXB 2 ), matrix metalloproteinases (MMP), cytokines, and chemokines. Our findings are in accordance with our previous observation that Escherichia coli LPS potentiates as well as inhibits microglia O 2 À generation (Mayer et al., 1999), and further extend this phenomenon to another LPS, namely M. aeruginosa LPS, which although less potently than E. coli LPS, was observed for the first time to our knowledge to prime rat microglia O 2 À generation in a concentration-dependent manner in vitro.
LPS contamination. All glassware and metal spatulas were baked for 4 h at 180°C to inactivate LPS (Sharma, 1986). Sterile and LPS-free 225-cm 2 vented cell culture flasks were from BD Biosciences, Bedford, MA; 24-well flat-bottom culture clusters and disposable serological pipettes were from Costar Corporation, Cambridge, MA. Sterile and pyrogen-free Eppendorf Biopur pipette tips were from Brinkmann Instruments, Inc., Westbury, NY.
Isolation of rat neonatal microglia. Experiments were performed in adherence to National Institutes of Health guidelines on the use of experimental animals, with protocols approved by Midwestern University's Research and Animal Care Committee. Rat brain neonatal microglia were isolated and characterized as previously described (Mayer et al., 1999). Briefly, cerebral cortices of 1-to 2-day-old Sprague-Dawley rats (Charles Rivers, Hartford, CT) were surgically removed, placed in cold DMEM containing 120 U/ml P and 12 lg/ml S, the meninges removed, and brain tissue minced and dissociated with trypsin-EDTA at 35.9°C for 3-5 min. The mixed glial cell suspension was plated in 225-cm 2 vented cell culture flasks with DMEM medium supplemented with 10% FBS containing 120 U/ml P and 12 lg/ml S and grown in a humidified 5% CO 2 incubator at 35.9°C for 12-14 days. Upon confluence (day 14) and every week thereafter, microglia were detached using an orbital shaker (150 rpm, 0.5 h, 35.9°C, 5% CO 2 ), centrifuged (400 3 g, 25 min, 4°C), and cell number and viability assessed by trypan blue exclusion. Rat neonatal microglia yields averaged 1.1 3 10 6 microglia per tissue culture flask (225 cm 2 ) per week in our laboratory. Depending on the particular experimental protocol (see below), microglia averaging > than 95% viability were plated in 24-well cell culture clusters, with DMEM supplemented with 10% FBS containing 120 U/ml P and 12 lg/m S and placed in a humidified 5% CO 2 incubator at 35.9°C 24 h prior to the experiments.

Activation of microglia with LPS (experimental protocol).
To determine the effect of M. aeruginosa LPS on rat neonatal microglia activation and inflammatory mediator release (O 2 À , eicosanoids, cytokines, chemokines, and MMP), 2 3 10 5 rat neonatal microglia were seeded in DMEM þ 10% FBS þ 120 U/ml P þ 12 lg/ml S into each well of nonpyrogenic polystyrene 24-well flat-bottom culture clusters (Costar, Corning Inc., Corning, NY) and stimulated with 0.1-100,000 ng/ml M. aeruginosa LPS for 17 h in a humidified 5% CO 2 incubator at 35.9°C. E. coli LPS (0.1-100 ng/ml) was used as control in all the experiments described herein (Mayer et al., 1999). After the 17-h incubation, conditioned media (1 ml) from each tissue culture well was aspirated and split into two aliquots. One aliquot (0.1 ml) was used to measure LDH levels, as a measure of cell viability (Morgenstern et al., 1966). The remaining aliquot (0.9 ml) was frozen (À84°C) until determination of eicosanoids, cytokines, 64 MAYER ET AL.
chemokines, and MMPs, as described below. Once the conditioned media had been removed, both M. aeruginosa and E. coli LPS-treated microglia cells were washed with warm (37°C) HBSS, and O 2 À was determined as described below.
Assay for superoxide anion (O 2 À ) generation. O 2 À generation was determined by the SOD-inhibitable reduction of FCC (Mayer et al., 1999). Briefly, PMA (1lM)-triggered O 2 À release from either E. coli or M. aeruginosa LPS-activated microglia was measured in the presence of FCC (50lM) and HBSS, with or without SOD (700 units), which inhibited > 95% of FCC reduction, during a 70-min incubation. All experimental treatments were run in duplicate and in a final volume of 1 ml. Changes in FCC absorbance were measured at 550 nm using a Beckman DU-800 spectrophotometer. Differences in the amount of reduced FCC in the presence and absence of SOD were used to determine microglia O 2 À generation by employing the molecular extinction coefficient of 21.0 3 10 3 M À1 cm À1 and expressed in nmol.
Assay for LDH. To assess cell viability following preincubation of microglia with either M. aeruginosa LPS or E. coli LPS as described in our experimental protocol, the conditioned media was harvested and LDH release was determined spectrophotometrically as previously described (Mayer et al., 1999;Morgenstern et al., 1966). Microglia LDH release was expressed as a percent of total LDH released into the conditioned media. Total LDH release resulted from 0.1% Triton X-100-lysed microglia (intracellular LDH) plus LDH present in the extracellular media. Because the FBS contained LDH (data not shown), unless LDH release from LPS-treated microglia was greater than 15% of that observed from Triton X-100 (0.1%)-treated microglia (total LDH), LPS treatment was considered to have no effect on microglia viability.
Assay for TXB 2 generation. Following incubation of microglia with either M. aeruginosa LPS or E. coli LPS for 17 h, TXB 2 generation in cell-free conditioned media was measured using TXB 2 immunoassays (Cayman Chemical, Ann Arbor, MI), as indicated by the manufacturer's protocol. Results were expressed as picogram per ml (pg/ml). The minimum detectable concentration was 7.8 pg/ml TXB 2 .
Gelatinase zymography for MMP-2 and MMP-9 analysis. Gelatincontaining zymograms are typically used to detect MMP-2 (68 kDa) and MMP-9 (92 kDa), and their identification is based on molecular weight. Following incubation of cultured rat neonatal microglia with either M. aeruginosa LPS or E. coli LPS, MMP-2 and -9 expression were determined in the cell-free conditioned media. As the rat neonatal microglia cultures were normalized for cell number, equal volumes of harvested media obtained from each condition were analyzed. Briefly, 90 lg of each protein sample were electrophoresed under nondenaturing conditions using a 10% polyacrylamide gel containing 0.1% gelatin. The gels were incubated twice for 30 min in 1X Novex Zymogram Renaturing Buffer (Invitrogen, Carlsbad, CA) and then incubated overnight in a 5% CO 2 incubator at 37°C in 1X Novex Zymogram Developing Buffer (Invitrogen). Gels were stained in 0.4% (wt/vol) Coomassie Brilliant Blue R-250 Solution (Bio-Rad, Hercules, CA), followed by destaining in 10% methanol, 10% acetic acid. MMP activity was visualized as clear bands against a blue background. Images of zymograms were obtained using a Kodak Gel Logic 1500 Imaging System and Molecular Imaging Software (Kodak, Rochester, NY). Semiquantitation of zymograms was performed using the UN-SCAN-IT gel automated digitizing system from Silk Scientific (Orem, UT). Microglia MMP release was normalized between experiments by dividing values (pixels) for treated samples by their respective controls.
Statistical analysis of the data. Data was expressed as mean ± SEM from 2 to 4 independent experiments (n), each experiment with triplicate determinations. Data were analyzed with Prism software package version 5 from GraphPad, San Diego, CA. LPS-treated microglia were compared with the vehicle-treated microglia (control), shown as 0 in the corresponding figures. One way ANOVA followed by Dunnett's post hoc procedure was performed on all sets of data. Statistical significance between the effect of a single dose of E. coli and M. aeruginosa LPS on the release of each mediator studied (e.g., O 2 À ) was determined using 2-way ANOVA. Differences were considered statistically significant at p < 0.05 and reported in each figure legend.

Effect of M. aeruginosa LPS on Rat Neonatal Brain
Microglia O 2 À Generation Generation of reactive oxygen species by brain microglia have been implicated in oxidative stress reported in several chronic neurodegenerative diseases Colton and Wilcock, 2010;Mayer, 1998). We and others have reported that E. coli LPS-treated rat microglia release O 2 À in vitro (Colton and Gilbert, 1987;Mayer et al., 1999). As shown in Figure 1, PMA-stimulated O 2 À production was observed when rat microglia were pretreated with either M. aeruginosa or E. coli LPS for 17 h. A bell-shaped doseresponse curve was observed when rat neonatal microglia were treated with M. aeruginosa LPS for 17 h, with maximal O 2 À release observed at 1000 ng/ml M. aeruginosa LPS which thereafter progressively decreased. In contrast and as previously reported (Mayer et al., 1999), PMA-stimulated O 2 À generation in E. coli LPS-treated microglia cells was bell-shaped but shifted to the left with maximal O 2 À at 1 ng/ml LPS. Thus, M. aeruginosa LPS appeared to be a 1000-fold less potent than E. coli LPS in inducing O 2 À production from rat microglia in vitro.

Effect of M. aeruginosa LPS on Rat Neonatal Brain Microglia LDH Generation
In order to determine whether the progressive decrease in O 2 À release generation (Fig. 1) resulted from concentration-dependent toxicity of M. aeruginosa and E. coli LPS to microglia during the 17-h incubation, LDH release was determined in the tissue culture supernates (Mayer et al., 1999). LDH has frequently been used as a marker for cellular toxicity (Mayer and Spitzer, 1994;Morgenstern et al., 1966). As shown in Figure 2, there was a concentration-dependent increase in LDH release in vitro that closely paralleled the progressive decrease in O 2 À generation observed at higher M. aeruginosa and E. coli LPS concentrations. Thus, in M. aeruginosa LPS-stimulated cells, a maximum 55 ± 19.3% of control LDH release was observed at 100,000 ng/ml. In contrast, in E. coli LPS-stimulated microglia, a dose-dependent LDH increase was observed at greater than 1 ng/ml LPS, reaching 67.7 ± 8.2% of control at 100 ng/ml LPS.

Effect of M. aeruginosa LPS on Rat Neonatal Brain
Microglia TXB 2 Generation Eicosanoids released by activated microglia have been proposed to play a proinflammatory role in the pathology of neurological and neurodegenerative diseases (Choi et al., 2009). We and others have shown that E. coli LPS-treated rat microglia release TXB 2 in vitro (Mayer et al., 1999;Minghetti and Levi, 1995). As shown in Figure 3, unstimulated microglia released low levels of TXB 2 . In M. aeruginosa LPS-treated microglia cells, TXB 2 generation yielded a sigmoid curve and became statistically significant at 1000 ng/ml, when there was maximal O 2 À generation ( Fig. 1) but low LDH release (Fig. 2). Confirming our previous study (Mayer et al., 1999), in E. coli LPS-stimulated cells, a concentration-dependent TXB 2 release was observed at 1 ng/ml. Thereafter, TXB 2 release progressively increased, becoming maximal at 100 ng/ml, when O 2 À generation was attenuated (Fig. 1), and LDH release was maximal (Fig.  2). Thus, M. aeruginosa LPS appeared to be less potent than E. coli LPS in inducing a concentration-dependent TXB 2 release from rat microglia in vitro as well as less efficacious than E. coli LPS.

Effect of M. aeruginosa LPS on Rat Neonatal Brain
Microglia MMP-2 and MMP-9 Generation MMPs released by activated microglia have been proposed to play a proinflammatory role in the pathology of sepsis and neuroinflammation (Candelario-Jalil et al., 2009;Vanlaere and Libert, 2009). We and others have reported MMP-2 and MMP-9 release from E. coli LPS-stimulated rat microglia in vitro (Gottschall et al., 1995;Mayer et al., 1999). As shown in Figure 4, in M. aeruginosa LPS-stimulated microglia, MMP-9 but not MMP-2 levels were statistically significant at LPS concentrations equal or greater than 1000 ng/ml. In contrast, when microglia were stimulated with E. coli LPS for 17 h, a concentration-dependent increase of MMP-9 but not of MMP-2 was observed at LPS concentrations equal or greater than 1 ng/ml. Thus, similar to O 2 À (Fig. 1) and TXB 2 (Fig. 3), M. aeruginosa LPS appeared to be a 1000-fold less potent than E. coli LPS in inducing concentration-dependent release of MMP-9 from rat microglia in vitro. M. aeruginosa LPS appeared to be more efficacious than E. coli LPS because maximal release MMP-9 was 1.23-fold higher.

Effect of M. aeruginosa LPS on Rat Neonatal Brain
Microglia TNF-a, IL-1a, and IL-6 Generation Presence of the cytokine TNF-a has been considered a hallmark of neuroinflammation as well as numerous neurodegenerative conditions (McCoy and Tansey, 2008). Release of TNF-a triggered by E. coli LPS in vitro has been reported in mouse (Esen and Kielian, 2007;Hausler et al., 2002;Hayashi et al., 1995), human (Lee et al., 1993), and rat microglia (Horvath et al., 2008;Mayer et al., 1999). In the present study, unstimulated rat microglia released low levels of TNF-a. As shown in Figure 5 (panel A), in M. aeruginosa LPS-stimulated rat microglia cells, TNF-a levels increased at greater than 100 ng/ml and became statistically significant at 100,000 ng/ml. In contrast, a concentration-dependent TNF-a release in E. coli LPS-stimulated microglia was observed at greater than 0.1 ng/ ml LPS which peaked at 10 ng/ml LPS thus confirming our previous observations (Mayer et al., 1999). Thus, similar to O 2 À (Fig. 1) and TXB 2 (Fig. 3), M. aeruginosa LPS appeared to be a 1000-fold less potent than E. coli LPS in inducing TNF-a production from rat microglia in vitro. Notably, M. aeruginosa LPS was clearly more efficacious than E. coli LPS because maximal TNF-a release was 375% higher.
The cytokine IL-1a appears to be a pivotal mediator in neuroimmune responses and chronic neurodegenerative disorders (Allan et al., 2005;Schultzberg et al., 2007;Simi et al., 2007). Release of IL-1a has been reported from rat microglia stimulated with E. coli LPS in vitro (Giulian et al., 1986). In FIG. 4. The effect of Escherichia coli and Microcystis aeruginosa LPS on rat neonatal microglia MMP-2 and -9 release. Neonatal rat microglia (2 3 10 5 cells/well) were treated with E. coli LPS (0.1-100 ng/ml) or M. aeruginosa LPS (0.1-10 5 ng/ml) for 17-h in vitro. SDS-PAGE zymography (panel A) and bar graphs depicting the quantitated results for E. coli LPS (panel B) and for M. aeruginosa LPS (panel C). MMP-2 and MMP-9 were determined as described in Materials and Methods. Data expressed as mean ± SEM of normalized MMP release from three independent experiments (n). *p < 0.05, **p < 0.01 LPS versus untreated control (0).
MICROGLIA ACTIVATION BY MICROCYSTIS AERUGINOSA LPS our study, unstimulated microglia released low levels of IL-1a. As shown in Figure 5 (panel B), in M. aeruginosa LPSstimulated microglia, IL-1a levels increased after 100 ng/ml and became statistically significant at 100,000 ng/ml. In contrast, in E. coli LPS-stimulated microglia, IL-1a levels rose at 0.1 ng/ml and peaked at 10 ng/ml. Thus, similar to O 2 À (Fig. 1), TXB 2 (Fig. 3) and TNF-a (Fig. 5) (panel A) M. aeruginosa LPS appeared to be a 1000-fold less potent than E. coli LPS in inducing IL-1a generation from rat microglia in vitro. Interestingly, as observed for TNF-a (Fig. 5) (panel A) generation, M. aeruginosa LPS appeared more efficacious than E. coli LPS because maximal release of IL-1a was 165% higher.
IL-6 is a multifunctional cytokine involved in a variety of inflammatory conditions, including neuroinflammation (Nishimoto, 2010) and chronic neurodegenerative disorders such as Alzheimer's disease and multiple sclerosis (Harris and Sadiq, 2009;Maccioni et al., 2009). It has previously been shown that human (Lee et al., 1993), murine (Hausler et al., 2002), and rat (Gottschall et al., 1995;Horvath et al., 2008) microglia release the cytokine IL-6 when stimulated with E. coli LPS in vitro. In our studies, unstimulated microglia released low levels of IL-6. As shown in Figure 5 (panel C), in M. aeruginosa LPSstimulated microglia, IL-6 levels increased after 100 ng/ml and became statistically significant at 100,000 ng/ml. In contrast, IL-6 levels in E. coli LPS-stimulated microglia rose after 0.1 ng/ml LPS and peaked at 10 ng/ml LPS. Thus, similar to O 2 À (Fig. 1), TXB 2 (Fig. 3), TNF-a (Fig. 5) (panel A), and IL-1a (Fig. 5) (panel B) M. aeruginosa LPS appeared to be a 1000-fold less potent than E. coli LPS in inducing concentration-dependent release of IL-6 from rat microglia in vitro, but as observed with TNF-a (Fig. 5) (panel A) and IL-1a (Fig. 5) (panel B) release, M. aeruginosa LPS was more efficacious than E. coli LPS because maximal release of IL-6 was 233% higher.
MCP-1/CCL2 is an inflammatory chemokine involved in neuroinflammation (Ubogu et al., 2006) and multiple sclerosis (Szczucinski and Losy, 2007). MCP-1/CCL2 has been shown to be generated by mouse (Hausler et al., 2002;Hayashi et al., 1995), rat (Horvath et al., 2008;Sun et al., 1997), and human microglia in vitro (Peterson et al., 1997). As shown in Figure 6 (panel C), M. aeruginosa LPS-induced MCP-1/CCL2 release yielded a flat sigmoid curve with maximal nonstatistically significant release at 100,000 ng/ml. In contrast, E. coli LPSinduced statistically significant release of MCP-1/CCL2 yielding a bell-shaped curve which peaked at 1 ng/ml. Thus, similar to the other cytokines and chemokines investigated, M. aeruginosa LPS appeared to be less potent than E. coli LPS in inducing concentration-dependent release of MCP-1/CCL2. However, in contrast to the other inflammatory mediators investigated, release of MCP-1/CCL2 was less efficacious at the highest M. aeruginosa LPS concentration used.

Effect of M. aeruginosa LPS on Rat Neonatal Brain
Microglia TGF-b1 and TGF-b2 Generation In order to determine whether M. aeruginosa LPS-treated rat neonatal brain microglia affected release of anti-inflammatory cytokines, we investigated the presence of TGF-b1 and TGF-b2, which have been studied for their neuroprotective effects (Qian et al., 2008). TGF-b1 and TGF-b2 are constitutively expressed in rat microglia (Polazzi et al., 2009) in vitro, as well as in E. coli LPS-activated human (Walker et al., 1995) and murine microglia (Chao et al., 1995). As shown in Table 1, unstimulated rat microglia released TGF-b1 constitutively but neither M. aeruginosa LPS nor E. coli LPS enhanced TGF-b1 or TGF-b 2 release during the 17-h in vitro incubation. In fact, a small though nonstatistically significant decrease of constitutive TGF-b1 was observed in M. aeruginosa LPS-treated microglia.

DISCUSSION
The increased generation of reactive oxygen species such as O 2 À in the central nervous system (Halliwell, 2001) has been hypothesized to be involved in neuronal injury and neurodegeneration in several neuropathologies (Cunningham et al., 2005). Thus, microglia activation (Rock et al., 2004) and concomitant O 2 À generation as part of the mechanism of neuroinflammation have been extensively investigated over the past 25 years (Ransohoff and Perry, 2009). One activator of microglia O 2 À generation that has received considerable attention is LPS (Holst et al., 1996) because it may activate microglia via the lipid A portion of the macromolecule leading to the in vivo and in vitro generation of inflammatory mediators such as MMPs, arachidonic acid metabolites, and cytokines (Mayer, 1998).
The first goal of his study was to determine whether M. aeruginosa LPS might induce O 2 À release from brain microglia in vitro. Our current data supports the following conclusions: First, confirming our previous observations (Mayer et al., 1999) after microglia were incubated with E. coli LPS for 17 h, PMA-stimulated O 2 À generation yielded a bell-shaped dose-response curve, with potentiation of O 2 À release at 0.1 to 1 ng/ml LPS, followed by a progressive inhibition of O 2 À generation at 10 and 100 ng/ml. The progressive decrease in O 2 À generation was concomitant to an increase in LDH release, revealing that increased concentrations of E. coli LPS caused self-injury to microglia cells. Second and for the first time to our knowledge, cyanobacterium M. aeruginosa LPS-treated microglia O 2 À generation yielded a bell-shaped curve: thus, following an initial potentiation of O 2 À release at M. aeruginosa LPS 10 to 1000 ng/ml, a progressive attenuation of O 2 À generation was observed at M. aeruginosa LPS 10,000 and 100,000 ng/ml. Also similar to E. coli LPS, after a 17-h incubation with M. aeruginosa LPS toxicity to microglia was demonstrated by increased LDH in the tissue culture media. Third, M. aeruginosa LPS was less potent than E. coli LPS in activating rat microglia O 2 À generation after 17 h of in vitro stimulation. Additional studies will be required to further characterize the potentiation and attenuation of O 2 À generation by M. aeruginosa LPS to determine whether the kinetics are similar to E. coli LPS, where potentiation and inhibition of rat microglia O 2 À generation required 18-24 h (Mayer et al., 1999). a Neonatal rat microglia (2 3 10 5 cells/well) were treated with E. coli LPS (0.1-100 ng/ml) or M. aeruginosa LPS (0.1-10 5 ng/ml) for 17-h in vitro. TGFb 1 and TGF-b 2 were determined as described in Materials and Methods. Data expressed as pg/ml is the mean ± SEM from 1 to 3 independent experiments (n), each with triplicate determinations.
A second goal of this investigation was to assess whether M. aeruginosa LPS-treated microglia might release proinflammatory and antiinflammatory mediators concomitantly with O 2 À generation. The following results were observed: First, confirming our previous observations (Mayer et al., 1999), after microglia were incubated with E. coli LPS for 17 h, a statistically significant release of TXB 2 , MMP-9, and TNF-a was observed concomitant with maximal O 2 À potentiation by 1 ng/ml E. coli LPS. Moreover, we observed several additional proinflammatory cytokines and chemokines in E. coli LPS-treated microglia tissue culture supernates, in the following rank order: MIP-2/CXCL2 > MIP-1a/CCL3 > IL-6 > IL-1a > MCP-1/CCL2. Second and for the first time to our knowledge, M. aeruginosa LPS-treated microglia were shown to release TXB 2 , MMP-9, as well as well as several cytokines and chemokines concomitant with maximal O 2 À potentiation and in quantities which were ''enhanced'' when compared with E. coli LPS-treated microglia in the following rank order: MIP-2/CXCL2 > MIP-1a/CCL3 > TNF-a > IL-6 > IL-1a > MCP-1/CCL2. Third and similar to O 2 À generation, M. aeruginosa LPS was less potent than E. coli LPS in stimulating release of microglia inflammatory mediators after a 17-h in vitro stimulation. However, M. aeruginosa LPS was more efficacious than E. coli LPS because most of the inflammatory mediators investigated were generated in larger quantities in vitro. Fourth, there appeared to be a correlation between both the increased release of LDH and that of TXB 2 , MMP-9, as well as the cytokines and chemokines we measured in the culture media, and the progressive concentrationdependent attenuation of PMA-elicited microglia O 2 À generation in M. aeruginosa LPS-treated cells. Whether one or several of these inflammatory mediators might be responsible for the mechanism leading to the increased cytotoxicity to neonatal microglia cells in vitro is presently unknown. The current study suggests, but does not conclusively prove, that these mediators, and perhaps others, may in some manner contribute to the mechanism of M. aeruginosa LPS-induced cytotoxicity to the microglia in vitro.
It is important to reflect on potential new lines of inquiry that emerge from the observed effects of M. aeruginosa LPS on rat neonatal microglia in vitro. First, the present study was completed with M. aeruginosa LPS strain UTCC 299, one of several M. aeruginosa strains currently available (Rapala et al., 2002). Thus, it would be important to determine whether LPS isolated from other M. aeruginosa strains shown to be present in drinking water sources will also be bioactive in the rat microglia model in vitro. Second, our study was limited to the effect of M. aeruginosa LPS on ''neonatal'' brain microglia. Thus, it would be important to determine whether ''adult'' rat microglia, which are known to release higher levels of PGE 2 than neonatal microglia (Slepko et al., 1997) may perhaps differ in their capacity to generate O 2 À and other inflammatory mediators. Third, because human, mice, and hamster microglia produce significantly different amounts of O 2 À in response to the same activating agents, differences which have been hypothesized to be important when modeling human disease processes from rodent studies (Colton et al., 1996), it should be important to determine whether the observed biphasic effects of M. aeruginosa LPS on rat neonatal microglia O 2 À generation in vitro will also occur with human microglia. Interestingly, we have observed that E. coli LPS will significantly prime human microglia for O 2 À release in a concentrationdependent manner (Mayer et al., 2004). Fourth, in vivo studies will be required to determine whether M. aeruginosa LPS may be pathogenic to the brain immune system because M. aeruginosa extracts have been shown to affect the immune response in mice (Shen et al., 2003) and caused IL-1 and TNF-a release from macrophages in vitro and mice in vivo (Nakano et al., 1989(Nakano et al., , 1991. Fifth, determining whether LPS isolated from other cyanobacteria such as A. flos-aquae, A. cylindrica, Oscillatoria tenui, and O. brevisothe, which have been shown to cause mortality in mice (Keleti and Sykora, 1982;Raziuddin et al., 1983), are bioactive in microglia in vitro becomes a relevant question meriting further investigation.
In our present investigation, inflammatory products elicited by ''classical'' activation of rat brain microglia have been observed; however, based on the ''complexity of microglia activation states'' and the limited nature of the present study (17-h incubation with M. aeruginosa LPS in vitro), it remains unknown whether additional microglia proinflammatory mediators as well as anti-inflammatory cytokines (e.g., IL-4, IL-13, and IL-10) typical of ''inflammo-resolution states'' may play an important role in our in vitro model (Colton and Wilcock, 2010). We suggest that additional mechanistic investigation of the effect of M. aeruginosa LPS on microglia at both the functional and molecular level, both in vitro and in vivo, will help define the ''continuum'' of activation states of microglia upon M. aeruginosa LPS treatment and the factors involved (receptors, enzymes, proteins, etc.) (Colton 2009). Furthermore, we are hopeful that accumulated data on microglia activation states after M. aeruginosa LPS exposure will help identify candidate therapeutic targets and novel treatment strategies to protect and treat humans after environmental exposure to cyanobacterial LPS (Anderson et al., 2002;Stewart et al., 2006Stewart et al., , 2009.