Male ICR mice (Damool Science, Korea) weighing 25–30 g and Sprague-Dawley rats weighing 200–300 g, were used in all experiments. Animals were maintained in accordance with the National Institute of Toxicological Research, Korea Food and Drug Administration guidelines for the care and use of laboratory animals. Animals were housed in two cages (five per cage) and in a 22 ± 2°C and 45~65% relative humidity environment under a 12-hr light/12-hr dark cycle (8:00 a.m. ~8:00 p.m.). All animals had free access to food (Samyang Foods, Seoul, Korea) and water. The anti-inflammatory sulindac sulfide (3.75 or 7.5 mg/kg) was given orally for 3 weeks prior to the injection of LPS in in vivo study.

The mice were randomly divided within each cage and injected intraperitoneally with either 250 μg/kg of Lipopolysaccharide (LPS) or sterile saline (0.9% NaCl). For all experiments, LPS (Escherichia coli, serotype 055:B5, Sigma, St. Louis, MO, USA) was used to induce an inflammatory response and was injected once on day 1 of behavioral testing. All injections were administered 4 hrs prior to testing. This allows enough time for the development of neuro-inflammation expressing central IL-1β gene (most notably in circumventricular organs, meningeal tissue, and choroid plexus) at this dose and similar doses of intraperitoneal LPS 19.

After the behavioral tests, animals were perfused with PBS under inhaled diethyl ether anesthesia. Brains were immediately collected, stored at -20°C, and separated into cortical and hippocampal regions. The brain regions (hippocampus and cerebralcortex) were immediately stored at -80°C before an assay of secretase activities, Aβ_1–42 level as well as western blotting.

As described elsewhere 2122, 2-day-old rat pups were ice-anesthetized and decapitated. After the skin was opened and the skull was cut, the brain was released from the skull cavity. After washing with PBS, the cerebrum was separated from the cerebellum and brain stem, and the cerebral hemispheres were separated from each other by gently teasing along the midline fissure with the sharp edge of forceps. The meninges were gently peeled from the individual cortical lobes and the cortices were dissociated by mechanical digestion [using the cell strainer (BD Biosciences, Franklin Lakes, NJ, USA)] with Dulbecco's modified Eagle's medium (DMEM) containing F12 nutrient mixture (Invitrogen, Carlsbad, CA). The resulting cells were centrifuged (1,500 rpm, 5 mins), resuspended in serum-supplemented culture media, and plated into 100 mm dishes. Serum-supplemented culture media was composed of DMEM supplemented with F12, FBS (5%), NaHCO₃ (40 mM), penicillin (100 units/ml), and steptomycin (100 μg/ml). The cells were incubated in the culture medium in a humidified incubator at 37°C and 5% CO₂for 9 days. At confluence (9 days), the flask was subjected to shaking for 16–18 hrs at 37°C. The cultures were treated for 48 hrs with cytosine arabinoside and the medium was replaced with DMEM/F12HAM containing 10% FBS. The monolayer was treated with 1.25% trypsin-EDTA for a short duration after which the cells were dissociated and plated into uncoated glass coverslips. The astrocyte cultures formed a layer of process-bearing, GFAP-positive cells. The purity of astrocyte cultures was assessed by GFAP-immunostaining. Under these conditions, we can assume that over 95% of the cells were astrocytes. The cultured cells were treated with LPS or TNF-α or IFN-γ for 24 hrs, and cells were harvested for the assay of Aβ and western blotting.

The Sprague-Dawley pregnant rats were sacrificed by cervical dislocation and the embryos were removed on the 18^th day of gestation. The embryonic brain tissues were mechanically dissociated into individual cells in NEUROBASAL medium (Invitrogen, Carlsbad, CA, USA). The resulting cells were centrifuged (1,500 rpm, 5 min), resuspended in NEUROBASAL medium containing B-27 supplement (Invitrogen, Carlsbad, CA), L-glutamine (0.5 mM), penicillin (100 units/ml), steptomycin (100 μg/ml) and plated into 60 mm dishes. The culture media was changed every 2 days. Greater than 90% of the cells in these cultures were neurons as assessed by cell morphology and immunostaining with mouse monoclonal antibodies against neurofilaments (1: 5,000). 7 day cultured cells were treated with LPS or TNF-α or IFN-γgor 24 hrs, the cells were harvested for the assay of Aβ and western blotting.

Brain tissues and cells were homogenized with protein extraction solution (PRO-PREP™, Intron Biotechnology, Korea), and lysed by 60 min incubation on ice. The lysate was centrifuged at 15,000 rpm for 15 min. Equal amount of proteins (40 μg) were separated on a SDS/10% or 15%-polyacrylamide gel, and then transferred to a polyvinylidene difluoride (PVDF) membrane (GE Water & Process technologies, Trevose, PA, USA). Blots were blocked for 2 hrs at room temperature with 5% (w/v) non-fat dried milk in Tris-Buffered Saline Tween-20 [TBST: 10 mM Tris (pH 8.0) and 150 mM NaCl solution containing 0.05% tween-20]. After a short wash in TBST, the membrane was incubated at room temperature with specific antibodies. Rabbit polyclonal antibodies against iNOS and COX-2 (1: 1,000 dilution, Cayman Chemical, Ann Arbor, MI, USA), APP (1:500 dilution, ABR, Golden, CO, USA), BACE1 (1:500 dilution, Sigma, St. Louis, MO, USA), C99 (1:500 dilution, Sigma, St. Louis, MO, USA) and mouse monoclonal antibody against phospho-ERK (1:500 dilution, Santa Cruz Biothechnology Inc. Santa Cruz, CA, USA) were used in the study. The blot was then incubated with the corresponding conjugated anti-rabbit or mouse immunoglobulin G-horseradish peroxidase (1:2,000 dilutions, Santa Cruz Biotechnology Inc. Santa Cruz, CA, USA). Immunoreactive proteins were detected with the BM Chemiluminescence blotting substrate (Roche applied science, Mannheim, Germany).

Mice were euthanized with diethyl ether and perfused with 0.1 M PBS then with 4% paraformaldehyde. The brains were collected from mice following perfusion and immediately fixed in 4% paraformaldehyde for 24 hrs. The brains were transferred successively to 10%, 20% and 30% sucrose solutions. Subsequently, brains were frozen on a cold stage and sectioned in a cryostate (40 μm-thick). Sections were treated with endogenous peroxidase (3% H₂O₂ in PBS), and then with 0.01 M PBS blocking buffer containing 10% bovine serum albumin in PBS for 40 min. Then the sections were incubated with rabbit polyclonal antibody against Aβ_1–42 (1:2,000 dilution, Covance, Berkeley, CA, USA), and iNOS and COX-2 (1: 1,000 dilution, Cayman Chemical, Ann Arbor, MI, USA), overnight. After the incubation, sections were washed in PBS and incubated with the biotinylated secondary antibodies (ABC kit, Vector Laboaratories, Burlingame, CA) for 30 min. The sections were washed with PBS, incubated with the avidin-biotin complex (Vector Laboratories, Burlingame, CA) for 30 min, and visualized by chromogen DAB (Vector Laboratories, Burlingame, CA) reaction. The sections were dehydrated in ethanol, cleared in xylene, and mounted with permaunt (Fisher Scientific, Hampton, NH). For the detection of cellular location of Aβ_1–42, we did an immunofluorescence immunostaining. Sections were rinsed in 0.01 M PBS buffer. After washing in PBS, the sections were incubated for 1 hr at room temperature with 10% bovine serum albumin diluted in PBS. The sections were incubated overnight at 4°C with Rabbit Polyclonal Aβ_1–42 antibody (1:2000 dilution, Covance, Berkeley, CA, USA). After washing in PBS, the sections were washed and incubated with Alexa Fluro 568 conjugated Rabbit Polyclonal antibody (1:200 dilution, Molecular Probe, Carlsbad, CA, USA) for 2 hrs at room temperature. Next, the sections were incubated with DAPI for 15 min at 37°C. Finally, the sections were rinsed, mounted on slides, and coverslipped for fluorescence microscopy and photography using ApoTome microscope (Carl Zeiss, Inc., Thornwood, NY, USA). For detection of apoptotic cell death in tumor tissue, the paraffin embedded sections were then incubated in the mixture of labeling solution (450 μl) and enzyme solution (50 μl) for 1 hr at 37°C and washed 3 times in 0.1 M PBS for 5 min each according to manufacturer's instructions. Next, the sections were incubated with DAPI for 15 min at 37°C. Finally, the sections were rinsed, mounted on slides, and coverslipped for fluorescence microscopy (DAS microscope). Positive TUNEL stains were recorded by counting the number of positively stained DAPI in the definite area.

The total activities of α-, β- and γ-secretase present in cortical and hippocampal regions were determined using a commercially available α-secretase activity kit (R&D systems, Wiesbaden, Germany), β-secretase fluorescence resonance energy transfer (BACE 1 FRET) assay kit (PANVERA, Madison, USA) and γ-secretase activity kit, (R&D systems, Wiesbaden, Germany) according to the manufacturer's instructions, respectively. Each tissue was homogenized in cold 1 × cell extraction buffer (a component of the kit) to a final protein concentration of 1 mg/ml.

To determine α (or γ)-secretase activity, 50 μl of lysate was mixed with 50 μl of reaction buffer. The mixture was incubated for 1 hr in the dark at 37°C after 5 μl of substrate was added. Substrate conjugated to the reporter molecules EDANS and DABCYL was cleaved by α (or γ)-secretase and released a fluorescent signal. This fluorescence was measured using a Fluostar galaxy fluorometer (excitation at 355 nm and emission at 510 nm) equipped with Felix software (BMG Labtechnologies, Offenburg, Germany). The level of α (or γ)-secretase enzymatic activity was proportional to fluorescence with the intensity of fluorescene which was expressed as fluorescence units.

To determine β-secretase, 10 μl of lysate was mixed with 10 μl of BACE1 substrate (Rh-EVNLDAEFK-Quencher). The reaction mixture was then incubated for 1 hr at room temperature in a black 96-microwell plate. The reaction was stopped by adding 10 μl of BACE1 stop buffer (2.5 M sodium acetate). Fluorescence was determined using a Fluostar galaxy fluorometer (excitation at 545 nm and emission at 590 nm) equipped with Felix software (BMG Labtechnologies, Offenburg, Germany). Enzyme activity was linearly related to fluorescence increases, and the activity was expressed as fluorescence units. All controls, blanks and samples were run in triplicate.

Lysates of brain tissue prepared as described in the Western blotting section were obtained through protein extraction buffer containing protease inhibitor. Media from neuronal cell culture was collected, then briefly spun to remove cell debris and mixed with 4-(2-aminoethyl)-benzene sulfonyl fluoride serine protease inhibitor. Aβ_1–42 and Aβ_1–40 levels were determined using specific ELISAs (IBL, Immuno-Biological Co., Ltd., Japan). In short, 100 μl of sample was added into the precoated plate and was incubated overnight at 4°C. After washing each well of the precoated plate with washing buffer, 100 μl of labeled antibody solution was added and the mixture was incubated for 1 hr at 4°C in the dark. After washing, chromogen was added and the mixture was incubated for 30 mins at room temperature in the dark. After the addition of stop solution, the resulting color was assayed at 450 nm using a microplate absorbance reader (Sunrise™, TECAN, Switzerland).

The experimental results were expressed as mean ± S.E. A one-way analysis of variance (ANOVA) was used for multiple comparisons followed by Dunnett. Differences with P < 0.05 were considered statistically significant.

In the passive avoidance test, at the learning trial (day 0), mice of all groups entered the dark compartment, and there were no significant differences among the animals. However, in the testing trial (day 1), the mice which received a single intraperitoneal injection of LPS (250 μg/kg) showed a significantly reduced step-through latency compared to those injected with vehicle (Fig. 1A). In the water maze test, the mice exhibited progressively decreased escape latency by the training (3 days after training; 2 times/day, total of 6 times training), and the escape latency at the end of training (7^th escape latency) to the platform was about 403 ± 44 cm and 19 ± 1 s (data not shown). LPS was then administered into the mice. Similar to the result in the step through test, LPS-treated mice showed longer escape latency (the time required to find the platform) and escape distance (the distance swam to find the platform) than the control group (Fig. 1B). This tendency continued throughout the 3-day trial period although the difference between the two groups was getting smaller. The LPS-treated group traveled a further distance to reach the platform than the control group did (Fig. 1C). It is considered that these differences between the two groups reflected the difference in memory function since there was not much difference in swimming speed (Fig. 1D).

A single intraperitoneal injection of LPS increased the Aβ_1–42 level in the cortex and hippocampus (Fig. 2A). In contrast, a single intraperitoneal injection of LPS decreased the Aβ_1–40 level in the cortex and hippocampus (Fig. 2A). To assess the pattern of Aβ_1–42 deposition, we analyzed the Aβ_1–42 immunoreactivity in the cortex and hippocampus following daily LPS injections for 3–7 days. An increase of Aβ_1–42 immunoreactivity was observed in the LPS injected group compared to that of the control group (Fig. 2B). Aβ_1–42 immunoreactivity progressively increased with the duration of LPS adminstration and was much more intense in the hippocampus compared to the control.

Following a single injection of LPS, the activities of β and γ-secretase in the cortex and hippocampus increased (Fig. 3B and 3C), whereas, the activity of α-secretase decreased in mice brains (Fig. 3A). Moreover, LPS treatment increased expression of APP, BACE and C99 accompanied with the increase of inflammatory proteins iNOS and COX-2 expression (Fig. 3D).

It is known that microglia and astrocytes are major sources of neuro-inflammation. Moreover, recent data showed that neuronal cells also have cytokine receptors such as LPS receptor (toll like receptor) as well as TNF receptor 2324. Neurons may be directly involved in neuro-inflammation or indirectly via the interaction with microglia and astrocytes. In order to analyze the effect of LPS induced inflammation on amyloidogenesis in vitro, cultured astrocytes from rat pups and neuronal cells from rat embryos were used. Astrocytes lend both mechanical and metabolic support for neurons, regulating the environment in which they function. Interferron-gamma (IFN-γ) and tumor necrosis factor-alpha (TNF-α) as well as LPS were treated to induce an inflammatory reaction. Similiar to the in vivo results, inflammatory stimuli concomitantly increased expression of amyloidogenic proteins (such as APP, BACE and C99) accompanied with the increase of expression of inflammatory proteins (such as COX-2 and iNOS) in both astrocytes (Fig. 4A) and neuronal cells (Fig. 4B). These results further indicate amyloidogenic pathway could be promoted by neuro-inflammatory stimulation in in vitro and in vivo.

The effect of sulindac sulfide, a COX-1, 2 non-selective drugs, in vivo and in vitro system was assessed. Sulindac sulfide has been known to decrease the Aβ secretion in N2a neuroblastoma cells stably transfected with human APP695 bearing the Swedish mutation 25. As shown in Fig. 4C, the cells expressed low levels of APP, β-site APP cleavage enzyme (BACE) and C99 protein in an unstimulated condition, whereas, expression of BACE, APP and C99 proteins increased in response to LPS (1 μg/ml) after 24 hrs. Treatment with sulindac sulfide (12.5, 25, 50 μM) caused concentration-dependent decreases in LPS-induced BACE, and C99 expression in neuronal cells, but did not change the expression of APP. In addition, sulindac sulfide decreased LPS-induced Aβ_1–42 secretion into culture media (Fig. 4D). Furthermore, oral pretreatment with sulindac sulfide (3.75 and 7.5 mg/kg) for 3 weeks suppressed memory impairment caused by LPS, and reduced increased Aβ_1–42 levels (Fig. 5D) in concentration-dependent manners. This was evaluated with the passive avoidance test (Fig. 5A) and the water maze test (Fig. 5B and 5C). It is considered that sulindac sulfide may have an endogenous Aβ-lowering effect in neuronal cells, and suggests that inflammatory reaction could influence the amyloidogenesis, and thus could improve memory function.

To verify the relationship between LPS-induced accumulation of Aβ and neuronal cell death, we investigated the induction of cell death by LPS in vivo. Substantial increase of apoptotic cells was found in the hippocampus of LPS treated mice. A significant increase in the percentage of the number of apoptotic cells was detected in the LPS treated animals (36.2 ± 3.6%) verses the control (2.1 ± 0.8%). The percentage of the number of apoptotic cells in the brains of LPS treated animals was significantly reduced by the sulindac sulfide pretreatment. The values were 11.4 ± 2.8% (3.75 mg/kg), and 6.1 ± 1.8% (7.5 mg/kg), respectively (Fig. 6A).

The activation of astrocytes was analyzed by their immunoreactivity for GFAP which was more intensive in LPS treated mice brains than in controls. It was also reduced by sulindac sulfide pre-treatment (Fig. 6B).