YY-1224, a terpene trilactone-strengthened Ginkgo biloba, attenuates neurodegenerative changes induced by β-amyloid (1-42) or double transgenic overexpression of APP and PS1 via inhibition of cyclooxygenase-2

The present study was performed in accordance with the Institute for Laboratory Research (ILAR) Guidelines for the Care and Use of Laboratory Animals. COX-2 (−/−)- and COX-2 (+/+)- mice were described previously [43, 44]. Breeding pairs of double transgenic mice expressing Swedish mutant amyloid precursor protein gene and mutant presenilin-1 (deletion of exon 9) gene [APPswe/PSEN1dE9 double Tg mice; B6C3-Tg (APPswe, PSEN1dE9) 85Dbo/J, The Jackson Laboratory, Bar Harbor, ME, USA] were bred and housed in an approved animal facility at Kangwon National University. Animals were maintained on a 12/12-h light/dark cycle and fed ad libitum and were adapted to these conditions for 2 weeks before the experiment.

Aβ (1-42) and Aβ (42-1) were purchased from American Peptide (Sunnyvale, CA, USA). Aβ (1-42) and Aβ (42-1) were dissolved in 0.1 M PBS at pH 7.4, and aliquots were stored at −20 °C. Aβ peptides in each aliquot were aggregated by incubation in sterile distilled water at 37 °C for 4 days. Six-month-old COX-2 (−/−)- and COX-2 (+/+)-mice were administered Aβ (1-42) and Aβ (42-1) (400 pmol, i.c.v. injection) according to the procedure established by Laursen and Belknap [45]. Each mouse was injected in the bregma using a 10-μl microsyringe (Hamilton, Reno, NV, USA) fitted with a 26-gauge needle inserted at a depth of 2.4 mm. The injection volume was 5 μl. The injection placement and needle track were visible and could be verified during dissection.

YY-1224 and a standard G. biloba extract (Gb) were obtained from the research center of Yuyu Pharma Inc. (Suwon, Republic of Korea). The content and representative HPLC chromatogram of each component in YY-1224 or Gb are shown in Table 1 and Additional file 1: Figure S1, respectively. YY-1224 (50 mg/kg) or Gb (50 mg/kg) was dissolved in 10% tween-80 and administered orally in a volume of 1 ml/kg. YY-1224 or Gb administration began 7 days before the Aβ i.c.v. injection, and the drug administration was continued once a day throughout the experimental period. The behavioral study commenced on day 3 after Aβ i.c.v. injection and was carried out sequentially. During the behavioral study, YY-1224 or Gb was administered 30 min after the behavioral test to avoid a direct effect on performance. The experimental design is shown in Fig. 1a and Additional file 1: Figure S2.

In addition, 6-month-old APPswe/PS1dE9 double Tg mice were treated with YY-1224 (50 mg/kg, p.o.) or Gb (50 mg/kg, p.o.) with or without the preferential COX-2 inhibitor, meloxicam (10 mg/kg, p.o.; Sigma-Aldrich, St. Louis, MO, USA), once a day for 3 months. Meloxicam was suspended in 0.5% sodium carboxymethyl cellulose (Na-CMC) immediately before use. The behavioral study was started when the mice were 9 months old, and additional treatment with YY-1224, Gb, or meloxicam was continued during the behavioral study. During the behavioral study, drugs were administered 30 min after the behavioral test to avoid a direct effect on performance. The experimental design is shown in Fig. 1b.

The Y-maze test was performed as described previously [46]. Briefly, the Y-shaped maze was constructed of black acrylic with three identical arms separated by 120°. Each arm was 40 cm long, 12 cm tall, and 10 cm wide. The mouse was placed at the end of one arm and allowed to move freely through the maze during an 8-min session. The percent alternation was calculated as the ratio of actual to possible alternations (defined as the total number of arm entries minus two) multiplied by 100.

The novel object recognition test was performed as described previously [47, 48]. On the training trial, two different objects were fixed on the floor in a symmetric position from the center of the open field box (40 × 40 × 35 high cm), 20 cm from each other and 10 cm from the nearest wall, and each mouse was allowed to explore the objects for 10 min. Mice were subjected to the test trial 24 h after the training trial. One of the familiar objects used during training trial was replaced by a novel object, and each mouse was then allowed to explore the objects for 10 min. Time spent exploring each object was recorded and analyzed with a video tracking system (EthoVision, Noldus, The Netherlands). Novel object recognition was expressed as a “recognition index (%),” the percentage of time spent exploring the novel object over the total time spent exploring both objects.

Immunocytochemistry and staining quantification were performed as described previously [46, 49, 50]. Mice were perfused transcardially with 50 mL of ice-cold PBS (10 mL/10 g body weight) followed by 4% paraformaldehyde (20 mL/10 g body weight). Brains were removed and stored in 4% paraformaldehyde overnight. The brains were cut on a horizontal sliding microtome into 35-μm transverse free-floating sections. Every sixth section of the dorsal hippocampus (total of four sections) from each brain was collected for immunocytochemistry. Sections were blocked with PBS containing 0.3% hydrogen peroxide for 30 min and then incubated in PBS containing 0.4% Triton X-100 and 1% normal serum for 20 min. Antigen retrieval with formic acid (70%, 10 min) was performed to detect Aβ deposition in APPswe/PS1dE9 double Tg mice. After a 48-h incubation with primary antibody against PAFR (1:200; Cayman Chemical, Ann Arbor, MI, USA), PAF-AH (1:80; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), Aβ (1:50; Invitrogen, Thermo Fisher Scientific, Camarillo, CA, USA), or Iba-1 (1:500, Wako Pure Chemical Industries, Chuo-ku, Osaka, Japan), sections were incubated with the biotinylated secondary antibody (1:1000; Vector Laboratories, Burlingame, CA, USA) for 1 h. The sections were then immersed in a solution containing avidin–biotin peroxidase complex (Vector Laboratories) for 1 h, and 3,3′-diaminobenzidine was utilized as the chromogen. Digital images were acquired at ×4, ×10, or ×20 objective magnification using an Olympus microscope (BX51; Olympus) and a digital microscope camera (DP72; Olympus). ImageJ version 1.47 software (National Institutes of Health, Bethesda, MD, USA) was employed to quantify images. Briefly, images were converted to 8-bit grayscale images and were subjected to background subtraction to correct for uneven background. To measure PAFR- or PAF-AH-immunoreactivity, the pyramidal cell layer of CA1 and CA3 regions from each section were selected as the region of interest (ROI). Threshold values were set to select the immunoreactive area. The integrated density of each ROI was measured, and the value was expressed as density per micrometer square. For staining quantification of Aβ or Iba-1, the entire hippocampus and the cortical area containing primary somatosensory cortex were selected as the ROIs. Threshold values were set to identify immunoreactive areas, and particle analysis was employed to measure the area fraction of Aβ- or Iba-1-immunoreactivity (Additional file 1: Figure S13).

Tissues were lysed in buffer containing a 200-mM Tris–HCl (pH 6.8), 1% SDS, 5 mM EGTA, 5 mM EDTA, 10% glycerol, 1× phosphatase inhibitor cocktail I (Sigma-Aldrich), and 1× protease inhibitor cocktail (Sigma-Aldrich). The lysate was centrifuged at 12,000 × g for 30 min, and the supernatant fraction was used for Western blot analysis as described previously [48‐50]. Proteins (20 μg/lane) were separated by 8 or 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (PAGE) and transferred onto polyvinylidene fluoride (PVDF) membranes. Following transfer, the membranes were preincubated with 5% non-fat milk for 30 min and incubated overnight at 4 °C with primary antibody against PPAR-γ (1:200; Santa Cruz Biotechnology, Inc.) or β-actin (1:50000; Sigma-Aldrich). Membranes were then incubated with HRP-conjugated secondary anti-rabbit IgG (1:1000, GE Healthcare, Piscataway, NJ, USA) or anti-mouse IgG (1:1000, Sigma-Aldrich) for 2 h. Subsequent visualization was performed using an enhanced chemiluminescence system (ECL plus^®, GE Healthcare). Relative intensities of the bands were quantified with PhotoCapt MW (version 10.01 for Windows; Vilber Lourmat, Marne la Vallée, France), and then normalized to the intensity of β-actin.

Reverse transcription and real-time polymerase chain reaction (RT-rt-PCR) was performed as described previously [51]. Total RNA was isolated from the hippocampus using an RNeasy Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions. Reverse transcription reactions were carried out using the RNA to cDNA EcoDry Premix (Clontech, Palo Alto, CA, USA) with a 1-h incubation at 42 °C. For rt-PCR, 0.5 μL of complementary DNA (cDNA) was added to 50 μL of total PCR reaction mixture, containing 25 pmol of each primer and QuantiTect SYBR Green PCR Master Mix (Qiagen). rt-PCR was performed using a CFX96 Touch real-time PCR system (Bio-Rad Laboratories, Hercules, CA, USA). The reference gene (GAPDH) and target gene from each sample were run in parallel in the same plate with the same amount of cDNA. Primer sequences are listed in Table 2. After activation of HotStarTaq DNA polymerase at 95 °C for 15 min, PCR was performed as 40 cycles of denaturation at 94 °C for 1 min, annealing at 60 °C for 1 min, and extension at 72 °C for 1 min. The relative mRNA expression level was normalized to that of GAPDH (2^{[Ct(GAPDH)-Ct(each gene)]}).

The amount of lipid peroxidation in the hippocampus was determined by measuring the level of thiobarbituric acid-reactive substance in homogenates and is expressed in terms of malondialdehyde (MDA) content. The MDA level was measured using HPLC-UV/VIS detection system (model LC-20AT and SPD-20A, Shimadzu, Kyoto, Japan) as described by Richard et al. [52] with slight modifications [49, 53]. Briefly, each hippocampal tissue was homogenized in PBS, and 0.1 ml of this homogenate (or standard solutions prepared daily from 1,1,3,3-tetra-methoxypropane) and 0.75 ml of the working solution (thiobarbituric acid 0.37% and perchloric acid 6.4%, 2:1, v:v) were mixed and heated to 95 °C for 1 h. After cooling (10 min in ice water bath), the flocculent precipitate was removed by centrifugation at 3200 × g for 10 min. The supernatant was neutralized and filtered prior to injection. Isocratic separation was achieved on a 5-μm ODS column with mobile phase consisting of 50 mM PBS (pH 6.0): methanol (58:42, v/v) at a flow rate of 1.0 mL/min. The effluents were monitored at 532 nm.

The extent of protein oxidation in the hippocampus was assessed by measuring the content of protein carbonyl groups, which was determined spectrophotometrically with the 2,4-dinitrophenylhydrazine (DNPH)-labeling procedure, as described by Oliver et al. [54]. Results are expressed as nanomoles of DNPH incorporated/mg protein [49, 53] based on the extinction coefficient for aliphatic hydrazones at 21 mM⁻¹·cm⁻¹.

Synaptosomal fractions were obtained as described by Whittaker et al. [55], with minor modifications [49, 53]. The protein concentration of the synaptosomal preparation was measured using the BCA protein assay reagent (Pierce, Rockford, IL, USA) and was found to be approximately 5 mg of protein per milliliter. ROS formation was assessed by measuring the conversion from 2′,7′-dichlorofluorescin diacetate (DCFH-DA) to dichlorofluorescin (DCF) as described by Lebel and Bondy [56]. Briefly, hippocampal synaptosomes were incubated with 5 μM DCFH-DA (Molecular Probes, Eugene, OR, USA) for 15 min at 37 °C. Excess unbound probe was removed by centrifugation at 12,500 × g for 10 min. The fluorescence intensity due to ROS was measured at an excitation wavelength of 488 nm and an emission wavelength of 528 nm. DCF (Sigma-Aldrich) was used as a standard.

Determination of 4-HNE was performed using slot blot analysis as described previously [49]. Following adsorption, the PVDF membranes were preincubated with 5% non-fat milk and incubated overnight at 4 °C with anti-4-HNE antibody (1:2000, Calbiochem, La Jolla, CA, USA). After incubation with the primary antibody, membranes were incubated with a HRP-conjugated secondary antibody. Subsequent visualization was performed using an enhanced chemiluminescence system (ECL plus®, GE Healthcare). The amount of oxidized proteins was measured using the Oxyblot kit [Chemicon (EMD Millipore), Temecula, MA, USA] according to the instructions provided by the manufacturer. Briefly, the protein carbonyl content was labeled with protein hydrazone derivatives using 2,4-dinitrophenylhydrazide (DNP). Each blot was preincubated with 5% non-fat milk and then incubated with the primary antibody (1:100) specific to the DNP moiety, followed by incubation with a HRP-conjugated secondary antibody. Subsequent visualization was performed using an enhanced chemiluminescence system (ECL plus®, GE Healthcare) [49].

The recognition test is based on the natural tendency of rodents to investigate a novel object instead of a familiar one. The choice to explore the novel object reflects the use of learning and (recognition) memory processes. There was no significant difference between Aβ-treated groups in exploratory preference for objects in the training trial. The exploratory preference for a novel object was significantly decreased [Fig. 2a, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01] in Aβ (1-42)-treated mice when compared with Aβ (42-1)-treated mice. In the COX-2 (+/+) mice, repeated treatment with Gb or YY-1224 significantly ameliorated the decrease in exploratory preference for a novel object that was induced by Aβ (1-42) [Fig. 2a, ANOVA and post hoc pairwise comparisons indicating the effect of Gb (P < 0.05) or YY-1224 (P < 0.01)]. Genetic depletion of COX-2 also significantly attenuated the Aβ (1-42)-induced decrease in the exploratory preference for a novel object (Fig. 2a, ANOVA and post hoc pairwise comparisons showing the effect of COX-2 gene knockout, P < 0.05), and YY-1224 or Gb did not provide additional cognitive enhancing effects in COX-2 (−/−) mice.

The Y-maze test is based on the natural navigation behaviors of rodents and is used to evaluate spatial working memory, which is known to be dependent on hippocampal function. There was no significant change in the performance of the Y maze task between Aβ (42-1)-treated groups. In the COX-2 (+/+) mice, alternation behavior was significantly decreased [Fig. 2b, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01] in Aβ (1-42)-treated mice compared to Aβ (42-1)-treated mice. ANOVA and post hoc pairwise comparisons revealed that the Aβ (1-42)-induced decrease in alternation behavior was significantly attenuated by YY-1224 (Fig. 2b, P < 0.01), Gb (Fig. 2b, P < 0.05), or genetic inhibition of COX-2 (Fig. 2b, P < 0.05). YY-1224-mediated attenuation appeared to be more pronounced than Gb in COX-2 (+/+) mice; however, YY-1224 and Gb did not affect the Y-maze performance in Aβ (1-42)-treated COX-2 (−/−) mice. The results from the Morris water maze are comparable to those from the novel object recognition test and the Y-maze test (Additional file 1: Figure S7).

To explore whether PAF modulatory actions were involved in the pharmacological effect of YY-1224 against Aβ (1-42)-induced memory impairments, the expressions of PAFR and PAF-AH were examined in the hippocampus of COX-2 (+/+)- and COX-2 (−/−)-mice using immunocytochemistry and RT-rt-PCR. ANOVA and post hoc pairwise comparisons indicated that Aβ (1-42) infusion significantly increased PAFR-immunoreactivity (Fig. 3a, P < 0.01 for CA1 or CA3) and PAFR mRNA levels (Fig. 3b, P < 0.01) in the hippocampi of COX-2 (+/+) mice. Aβ (1-42)-induced increases in PAFR-immunoreactivity and PAFR mRNA expression were significantly attenuated by Gb, YY-1224, or COX-2 gene knockout (Fig. 3a, b, ANOVA and post hoc pairwise comparisons showing the effect of Gb, YY-1224, or COX-2 gene knockout, P < 0.01). In addition, YY-1224 and Gb did not show additional effects compared to COX-2 gene depletion (Fig. 3). The results from RT-rt-PCR of PAFR mRNA are comparable to those from RT-PCR (Additional file 1: Figure S8a).

In contrast, total PAF-AH I-immunoreactivity was significantly decreased by Aβ (1-42) in the hippocampi of COX-2 (+/+) mice [Fig. 4, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01]. To examine which subunit of PAF-AH is transcriptionally regulated by Aβ (1-42), we examined changes in the mRNA expression level of PAF-AH I subunits using RT-rt-PCR. Aβ (1-42) infusion significantly decreased the mRNA expression of the α2 subunit, but not the α1 or LIS1 subunit, in the hippocampus of COX-2 (+/+) mice [Fig. 5b, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01]. In addition, the PAF-AH II mRNA level was significantly increased after treatment with Aβ (1-42) [Fig. 5d, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01]. These changes in PAF-AH I-immunoreactivity, PAF-AH I α2 mRNA expression, and PAF-AH II mRNA expression were significantly reversed by Gb, YY-1224, and COX-2 gene depletion (Fig. 5b, d, ANOVA and post hoc pairwise comparisons showing the effect of Gb, YY-1224, or COX-2 gene knockout, P < 0.01). YY-1224 appeared to be more effective in enhancing PAF-AH 1α2 mRNA expression than Gb in COX-2 (+/+) mice (Fig. 5b, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05). YY-1224 and Gb did not show additional effects compared to COX-2 gene depletion in the presence of Aβ (1-42) (Figs. 4 and 5). The results from RT-rt-PCR of the mRNA expression of PAF-AH subtypes are comparable to those from RT-PCR (Additional file 1: Figure S8b-d).

To evaluate whether COX-2 inhibitory action is involved in the pharmacological effect of YY-1224 against Aβ (1-42)-induced oxidative burdens, the level of oxidative stress markers, including lipid peroxidation, protein expression, and ROS formation was measured in the hippocampus of COX-2 (+/+)- and COX-2 (−/−)-mice. Treatment with Aβ (1-42) resulted in an increase in lipid peroxidation, as assessed by MDA and 4-HNE level, protein oxidation, as shown by protein carbonyl level and expression, and synaptosomal ROS formation in COX-2 (+/+) mice [Fig. 6, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01]. ANOVA and post hoc pairwise comparisons indicated that YY-1224 and Gb treatment significantly attenuated Aβ (1-42)-induced lipid peroxidation (Fig. 6a, YY-1224 and Gb, P < 0.01 for MDA and 4-HNE), protein oxidation (Fig. 6b, YY-1224, P < 0.01 or Gb P < 0.05 for protein carbonyl level; YY-1224 and Gb, P < 0.01 for protein carbonyl expression), and synaptosomal ROS formation (Fig. 6c, YY-1224, P < 0.01 or Gb P < 0.05) in COX-2 (+/+) mice. COX-2 gene depletion also significantly attenuated oxidative stress induced by Aβ (1-42) (Fig. 6, ANOVA and post hoc pairwise comparisons showing the effect of COX-2 gene knockout, P < 0.01 for all parameters). YY-1224 attenuated Aβ (1-42)-induced protein oxidation and ROS formation in COX-2 (+/+) mice more effectively than Gb (Fig. 6b, c, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05), but neither showed additional effects over COX-2 gene depletion (Fig. 6).

To understand whether COX-2 inhibitory action was involved in the pharmacological effect of YY-1224 against Aβ (1-42)-induced pro-inflammatory changes, the mRNA levels of pro-inflammatory markers were examined in the hippocampus of COX-2 (+/+)- and COX-2 (−/−)-mice using RT-rt-PCR. As shown in Fig. 7, the mRNA expression of TNF-α, IL-1β, IL-6, and iNOS, but not IFN-γ, was significantly increased by Aβ (1-42) in COX-2 (+/+) mice [Fig. 7a–c, e, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01]. YY-1224, Gb, and COX-2 gene depletion significantly attenuated these increases (Fig. 7a–c, e, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of YY-1224, Gb, and COX-2 gene knockout, P < 0.01). ANOVA and post hoc pairwise comparisons showed that YY-1224 had a more pronounced effect than Gb on the increases in IL-1β, IL-6, and iNOS induced by Aβ (1-42) in COX-2 (+/+) mice (Fig. 7b, c and e, P < 0.05 for IL-1β and iNOS; P < 0.01 for IL-6). YY-1224 and Gb did not show additional effects compared to COX-2 gene depletion in the presence of Aβ (1-42) (Fig. 7). The results from RT-rt-PCR of the mRNA expression of pro-inflammatory factors were comparable to those from RT-PCR (Additional file 1: Figure S9).

As mentioned above, it has been suggested that PPARγ is related to inflammatory and neurodegenerative processes [22]. To explore whether PPARs modulatory actions are involved in the anti-inflammatory effect of YY-1224, the mRNA or protein expression of PPARα or PPARγ was examined after Aβ (1-42) infusion in the hippocampus of COX-2 (+/+)- and COX-2 (−/−)-mice. Aβ (1-42) infusion significantly decreased the mRNA and protein expression of PPARγ in COX-2 (+/+) mice [Fig. 8b, c, ANOVA and post hoc pairwise comparisons showing the effect of Aβ (1-42), P < 0.01]. Aβ (1-42)-induced decreases in PPARγ mRNA and protein expression were significantly attenuated by YY-1224, Gb, or COX-2 gene knockout (Fig. 8b, c, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of YY-1224, Gb, and COX-2 gene knockout, P < 0.01). YY-1224 mediated significant attenuation than Gb in COX-2 (+/+) mice (Fig. 8b, c, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05). YY-1224 and Gb did not show additional effects compared to COX-2 gene depletion (Fig. 8). The results from RT-rt-PCR of the PPARα or PPARγ mRNA expression were comparable to those from RT-PCR (Additional file 1: Figure S10).

To understand the pharmacological effects of YY-1224, we examined the effect of Gb or YY-1224 on memory dysfunction and Aβ plaque deposition in APP/PS1 Tg mice. ANOVA and post hoc pairwise comparisons showed that treatment with YY-1224 or Gb significantly increased the alternation ratio in the Y-maze test (Fig. 9a, YY-1224 or Gb, P < 0.01 or P < 0.05) and the recognition index in the novel object recognition test (Fig. 9b, YY-1224 or Gb, P < 0.01 or P < 0.05). A preferential COX-2 inhibitor, meloxicam, also significantly enhanced cognitive function in the Y-maze (Fig. 9a , P < 0.01) and novel object recognition tests (Fig. 9b, P < 0.01). Consistent with results obtained from Aβ-treated mice, YY-1224 was more effective at enhancing novel object recognition in APP/PS1 Tg mice (Fig. 9b, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05), and meloxicam treatment did not affect the performance of YY-1224- or Gb-treated APP/PS1 Tg mice (Fig. 9a, b).

YY-1224, Gb, or meloxicam treatment significantly attenuated Aβ plaque deposition in the hippocampus and cortex of APP/PS1 Tg mice (Fig. 9c, d, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of YY-1224, Gb, and meloxicam, P < 0.01 for hippocampus and cortex). Attenuation mediated by YY-1224 was more pronounced than attenuation mediated by Gb (Fig. 9c, d, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05 for hippocampus and cortex). Meloxicam did not affect responses to pharmacological activity in YY-1224- or Gb-treated APP/PS1 Tg mice (Fig. 9c, d).

Because YY-1224 exerted its pharmacological effect through the enhancement of PAF-AH I and PPARγ expression in Aβ (1-42)-treated mice, we examined the effect of YY-1224 on the mRNA and protein expression of PAF-AH I and PPARγ in the hippocampus of APP/PS1 Tg mice. In line with the results from Aβ-treated mice, ANOVA and post hoc pairwise comparisons indicated that PAF-AH I-immunoreactivity was significantly increased by YY-1224 (Fig. 10a, b, P < 0.05 for CA1 and CA3), Gb (Fig. 10a, P < 0.05 for CA1), and pharmacological inhibition (i.e., meloxicam) of COX-2 (Fig. 10a, b, P < 0.05 for CA1 and CA3) in APP/PS1 Tg mice. Since the α2 subunit of PAF-AH I was positively regulated by YY-1224 in Aβ-treated mice, we examined the mRNA expression of this subunit in APP/PS1 Tg mice. ANOVA and post hoc pairwise comparisons revealed that the mRNA expression of the PAF-AH I α2 subunit was significantly enhanced by YY-1224 (Fig. 10c, P < 0.01), Gb (Fig. 10c, P < 0.01), and COX-2 inhibition (i.e., meloxicam; Fig. 10c, P < 0.01). YY-1224 enhanced PAF-AH I-immunoreactivity and PAF-AH I α2 mRNA expression more than Gb (Fig. 10a–c, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05). The results from RT-rt-PCR of PAF-AH I α2 subunit mRNA expression were comparable to those from RT-PCR (Additional file 1: Figure S11a).

As shown in Fig. 11, ANOVA and post hoc pairwise comparisons showed that the mRNA and protein expressions of PPARγ were significantly increased by treatment with YY-1224 (Fig. 11b, c, P < 0.01), Gb (Fig. 11b, c, P < 0.01 for mRNA, P < 0.05 for protein), or meloxicam (Fig. 11b, c, P < 0.01) in APP/PS1 Tg mice, and these results were consistent with the results from Aβ-treated mice. The pharmacological action of YY-1224 was more effective than that of Gb in PPARγ protein expression (Fig. 11c, ANOVA and post hoc pairwise comparisons showing the difference between Gb and YY-1224, P < 0.05). In addition, meloxicam did not affect the expression of PAF-AH or PPARγ in YY-1224- or Gb-treated APP/PS1 Tg mice (Figs. 10 and 11). The results from RT-rt-PCR of PPARα or PPARγ mRNA expression were comparable to those from RT-PCR (Additional file 1: Figure S11b and c).

To elucidate the role of microglia in the pharmacological effects of YY-1224, we examined the effect of Gb and YY-1224 on activated patterns of microglia (Fig. 12). Treatment with Gb and YY-1224 significantly attenuated Iba-1-labeled microglia in the hippocampus and cerebral cortex of 0.5% Na-CMC-treated APP/PS1 Tg mice (Fig. 12a, b, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of YY-1224, Gb, and meloxicam, P < 0.01 for hippocampus and cortex). Meloxicam did not alter significantly the Iba-1 immunoreactivity mediated by YY-1224 or Gb in APP/PS1 Tg mice (Fig. 12).

Different roles of microglia in tissue repair or damage may be due to distinct microglial subsets, i.e., “classically activated” pro-inflammatory (M1) or “alternatively activated” anti-inflammatory (M2) cells. Treatment with Gb or YY-1224 significantly decreased mRNA levels of M1 markers (CD16, CD32, and CD86) in the hippocampus of 0.5% Na-CMC-treated APP/PS1 Tg mice (Fig. 13a–c, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of YY-1224 and Gb, P < 0.01). In contrast, Gb and YY-1224 significantly enhanced the mRNA levels of YM1 and CD206 of the M2 phenotype in the hippocampus of 0.5% Na-CMC-treated APP/PS1 Tg mice (Fig. 13d, e, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of YY-1224 and Gb, P < 0.01). In meloxicam-treated groups, mRNA levels of CD16, CD32, and CD86 were significantly decreased in the hippocampus, while YM1 and CD206 mRNA levels were significantly increased (Fig. 13, ANOVA and post hoc pairwise comparisons showing the inhibitory effect of meloxicam, P < 0.01). In addition, meloxicam did not significantly alter the mRNA expression of CD16, CD32, CD86, YM1, and CD206 mediated by YY-1224 or Gb in APP/PS1 Tg mice (Fig. 13). The results from RT-rt-PCR of the mRNA expression of microglial phenotypic markers were comparable to those from RT-PCR (Additional file 1: Figure S12).