TLR4 mutation reduces microglial activation, increases Aβ deposits and exacerbates cognitive deficits in a mouse model of Alzheimer's disease

Animals

Pathogen-free transgenic mice of an AD model, TgAPPswe/PS1dE9 mice [B6C3-Tg(APPswe,PSEN1dE9)85Dbo/J, strain name at Jackson] [15], and B6C3F1 mice were purchased from Jackson Laboratory (Bar Harbor, ME). The transgenic mice express chimeric mouse/human APP with the double mutations (K670N and M671L) and human PS1 with a deletion of exon 9 found in familial AD patients. The transgenic mice have been maintained by mating with B6C3F1 mice. C3H/HeJ mice are highly susceptible to Gram-negative infection and resistant to bacterial lipopolysaccharide (LPS) due to a destructive mutation of the TLR4 gene (TLR4^Lps-d). The TLR4 genotype was determined by polymerase chain reaction (PCR) followed by restriction enzyme digestion with Nla III as described previously [10]. In this study, four experimental groups at the ages of 5 and 9 months were used: 1) homozygous TLR4 mutant TgAPPswe/PS1dE9 transgenic mice (TLR4M Tg), 2) TLR4 wild-type transgenic mice (TLR4W Tg), 3) homozygous TLR4 mutant non-transgenic littermates (TLR4M non-Tg), and 4) TLR4 wild type non-transgenic mice (TLR4W non-Tg) (n = 8-11/group at each age). Half of the mice were deeply anesthetized and perfused transcardially with cold PBS followed by 4% paraformaldehyde and processed for histochemical and immunohistochemical analyses. Half of the mice were euthanized and their brains were processed for biochemical analyses. Nine month-old mice (n = 10 -11/group) were subjected to the Morris water maze test. A separate set of TLR4W (n = 6) and TLR4M (n = 5) Tg mice at 5 months of age were used for extraction of RNA. Another separate set of TLR4W (n = 7) and TLR4M (n = 4) Tg mice at 9 months of age were also used for biochemical analyses (protein and mRNA). All animal protocols used for this study were prospectively reviewed and approved by the Institutional Animal Care and Use Committee of the University of Illinois College of Medicine at Peoria.

Morris water maze behavioral test

Acquisition of spatial learning in the Morris water maze was assessed during 5 consecutive days. The Morris water maze consisted of a pool (diameter: 112 cm, wall height: 75 cm) filled with water (21°C) at a height of 31 cm. Powdered milk was evenly spread over the water surface in order to camouflage the escape platform (10 cm × 10 cm) made of white plastic and covered with a wire mesh grid to ensure a firm grip. The pool was contained in a room with visual cues such as light fixtures and a ladder. The mice were placed next to and facing the wall successively in north (N), east (E), south (S), and west (W) positions, with the escape platform hidden 1 cm below water level in the middle of the NW quadrant. An overhead video-camera and SmartTM videotracking software (San Diego, CA) were used to estimate path length and escape latencies in 4 trial sessions for 5 days with approximately 20 min intertrial intervals. Whenever the mice failed to reach the escape platform within 1 min, the mice were guided to the platform and remained on it for 5 seconds. The day after the acquisition phase, a probe trial was conducted by removing the platform and placing the mouse next to and facing the N side. The time spent in the previously correct quadrant was measured for a single 1 min trial. After the probe trial, the visible platform subtask was conducted, with the escape platform lifted 1 cm above water level and shifted to the SE quadrant. A 17 cm high pole was inserted on top of the escape platform as a viewing aid. With the exception that the subtest was conducted in a single day, the same procedure was adopted as with the acquisition phase.

Immunohistochemistry, histochemistry and quantification of Aβ deposits and activated glial cells

Frozen serial sections (5 μm thick) were cut and subjected to immunohistochemistry using the avidin-biotin-peroxidase method (VECTASTAIN ABC Kit). Endogenous peroxidase was eliminated by treatment with 3% H₂O₂/10% methanol Tris-buffered saline (TBS) for 20 min at room temperature. After washing with water and 0.1 M TBS (pH 7.4), slides were blocked with 2% bovine serum albumin (BSA) and 2% goat serum in 0.1% triton-X-100 TBS (TBST) buffer for 60 min at room temperature to prevent non specific protein binding. The slides were then incubated with primary antibody 6E10 (1: 2000; Signet Laboratories, Dedham, MA) or CD11b (1:200; Serotec, MCA711, Raleigh, NC) in 2% BSA, 2% goat serum TBST overnight at 4°C. The sections were rinsed in 0.1 M TBST containing 0.1% BSA and incubated with biotinylated secondary antibody anti-mouse IgG (1: 400) for 6E10 or anti-rat IgG (1:200) for CD11b in 2% BSA, 1% goat serum TBST for 1 h at room temperature. Finally, the avidin biotin peroxidase method using 3,3'-diaminobenzidine as a substrate (Vector, Burlingame, CA) was performed according to manufacturer's protocol. For the negative control, slides were processed with isotype control antibodies (mouse IgG1 for 6E10 and rat IgG2b for CD11b) (BD Biosciences, San Jose, CA) to ensure specific staining by primary antibodies. Some sections were counterstained with hematoxylin. Brain sections were also stained with 1% thioflavin S followed by destaining in 70% ethanol for detection of Aβ fibrils.

Histomorphometry for quantification of amyloid deposition and microglial activation was performed using an Olympus BX61 automated microscope, Olympus Fluoview system and the Image Pro Plus v4 image analysis software (Media Cybernetics, Silver Spring, MD) capable of color segmentation and automation via programmable macros. Each brain section was entirely constructed from pictures that were taken using a 10X objective and 1X eyepiece lens in order to ensure no overlap. Six coronal brain sections from each mouse were analyzed, each separated by an approximately 250 μm interval, starting at 1.6 mm posterior to the bregma to caudal. Areas stained by 6E10, CD11b and thioflavin S were measured using Image Pro Plus v4 image analysis software and expressed as a percentage of total hippocampus or neocortex examined.

For double-label fluorescence immunohistochemistry, after quenching autofluorescence by 10 mg/ml sodium borohydride (NaBH4) and/or 0.05% Sudan Black B, brain sections were subjected to double-label fluorescence immunohistochemistry. For 6E10- and CD11b-double-label fluorescence, brain sections were incubated with these antibodies followed by incubation with Alexa Fluoro 594-conjugated chicken anti-mouse IgG antibody and Alexa Fluoro 488-conjugated goat-anti-rat IgG antibody, respectively. For TLR4-, CD11b, CD45-, and GFAP-double-label fluorescence, the sections were incubated overnight with rat anti-TLR4/MD2 antibody (eBioscience, San Diego, CA). After washing, the sections were incubated with chicken anti-rat IgG antibody conjugated with Alexa Fluoro 488 (Invitrogen, Carlsbad, CA) for 2 h. After washing, the same sections were similarly treated with rabbit anti- CD11b (Santa Cruz Biotechnology), anti-CD45 (Santa Cruz Biotechnology) or rabbit anti-GFAP (astrocytic marker: G-9269, Sigma) antibody followed by incubation with chicken anti-rabbit IgG Alexa Fluoro 594 conjugated secondary antibody (Invitrogen). Some sections were stained with CD11b antibody (MCA711) using the avidin-biotin-peroxidase method as described above and then subjected to thioflavin S staining. After washing, the sections were observed under Olympus IX71 automated fluorescence microscope. The pictures were taken through an Olympus DP70 digital camera system.

Quantification of buffer soluble brain Aβ by ELISA

Using the left cerebral hemispheres, the brain tissues were dounce-homogenized in carbonate buffer (100 mM Na₂CO₃, 50 mM NaCl, pH 11.5) containing protease inhibitors [10 μg/ml aprotinin and 1 mM 4-(2-aminoethyl) benzenesulphonyl fluoride hydrochloride (AEBSF)] and centrifuged at 16,000 g for 30 min at 4°C. Protein concentrations in the supernatants were determined by Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA), and levels of buffer-soluble Aβ were determined by Aβ42 and Aβ40 enzyme-linked immunosorbent assay (ELISA) kits (Invitrogen) according to the manufacturer's protocol. A duplicate sample from each mouse was used for quantification.

Quantification of cytokine and chemokine mRNA by real-time PCR

Isolation of CD11b⁺ splenocytes by flow cytometry

Spleens were individually isolated from 2-month old TLR4M and TLR4W Tg mice. Single cell suspension of splenocytes was prepared by homogenizing a spleen tissue in 10 ml of RPMI 1640 medium and forcing cells through a cell strainer with 70 μm pores. Splenocytes were centrifuged at 200 g for 5 min and suspended with 0.8 ml ACK lysing buffer (UAB Comprehensive Cancer Center) to lyse red blood cells. Cell suspension was centrifuged again at 300 g for 5 min and final cell pellets were suspended in 1 × PBS containing 1% BSA. Cells were adjusted to 1 × 10⁷ cells/ml and incubated with 1 μg/ml of PE rat anti-mouse CD11b (BD Pharmingen) at 4°C for 40 min in the dark. Then, cells were washed twice with 1 × PBS containing 1% BSA and centrifuged at 300 g for 5 min. The pellets were re-suspended in 1 × PBS containing 1% BSA at a concentration of 1 × 10⁷ cells/ml. The cells were sorted into CD11b⁺ and CD11b^-/low population by the FACSCalibur System (Becton-Dickinson Bioscience, Rockville, MD).

Treatment of CD11b⁺ monocytes with fibrillar Aβ

Synthetic Aβ42 was purchased from Anaspec (Anaspec Inc, San Jose, CA). Fibrillar Aβ was prepared as described previously [16]. The peptide was dissolved in 1 mM hexafluoroisopropanol (Sigma) and then dried under vacuum in a Speed Vac (Savant, Holbrook, NY). The residual peptide was re-suspended in dimethyl sulfoxide to a concentration of 5 mM. Fibrillar Aβ was made by adding 10 mM HCl to a concentration of 100 μM and incubated at 37°C for 24 h. CD11b⁺ splenocytes from TLRM and TLR4W Tg mice were plated at the density of 2.5 × 10⁵ cells/ml and incubated with 1 μM fibrillar Aβ for 4 h. Cells were harvested and RNA was extracted in Trizol reagent as described above. Complementary DNA (cDNA) was generated from 1 μg total RNA in a total volume of 20 μl using SuperScript^® III First-Strand Synthesis Kit according to the manufacturer's protocol. The experiment was performed in triplicate for each condition. mRNA levels of IL-1α, IL-1β, IL-6, CCL3, CCL4 were determined by real-time PCR as described above.

Statistical analysis

Data were expressed as mean ± standard error of the mean (SEM). Intergroup differences were assessed by a repeated measures analysis of variance (ANOVA) and two-tailed Student's t-test for normally distributed data. For the probe trial of the Morris water maze, the Mann-Whitney rank sum test was used for comparison. P ≤ 0.05 was considered statistically significant.

TLR4 mutation does not influence Aβ load in the brain of an AD mouse model at 5 months of age but diminishes microglial activation

TLR4W Tg mice start to develop Aβ deposits in the brain around the age of 5 months. To investigate if TLR4 signaling is involved in initiation of Aβ deposition and microglial activation at the early stage of AD, we determined Aβ load and microglial activation in the TLR4W (n = 5) and TLR4M (n = 7) Tg mice at 5 months. Diffuse and fibrillar Aβ deposits were detected by immunohistochemistry using 6E10 antibody that specifically reacts with human Aβ (Figure 1A and 1B) and were expressed by average percentages of areas showing Aβ immunoreactivity in the cerebral cortex (Figure 1E). No difference was found in Aβ load between TLR4W (0.243 ± 0.045%) and TLR4M (0.196 ± 0.030%) Tg mice (P > 0.05).

Activated microglia/myeloid cells in the cerebral cortex were immunostained for expression of CD11b (Mac-1) (Figure 1C and 1D) and the immunoreactive areas were quantified (Figure 1F). CD11b-immunoreactivity (0.237 ± 0.043%) in TLR4W Tg mice was greater than that (0.087 ± 0.011%) in TLR4M Tg mice (P = 0.001). The ratio of CD11b-immunoreactive area to 6E10-immunoreactive area in TLR4W Tg mice (0.921 ± 0.133) was greater than that in TLR4M Tg mice (0.459 ± 0.049, P = 0.006) (Figure 1G).

To confirm this reduction in CD11b-positive microglia in TLR4M Tg mice, we carried out double immunofluorescence staining of brain sections from Tg mice using anti-CD11b antibody (green) and anti-Aβ antibody (red) (Figure 1H and 1I). Most Aβ deposits in TLR4W Tg mice were closely associated with microglia showing high expression of CD11b and some degree of overlap was found in double immunofluorescence images (Figure 1H), suggesting uptake of Aβ by activated microglia. On the contrary, expression of CD11b in microglia closely associated with Aβ deposits was limited in TLR4 M Tg mice (Figure 1I).

To investigate if the TLR4 mutation alters production of Aβ in the brain, we determined levels of buffer-soluble Aβ40 and Aβ42 by ELISA in early stages of amyloidogenesis. There was no difference between TLR4W (n = 4) and TLR4M (n = 6) Tg mice at 5 months of age in the cerebral buffer-soluble Aβ40 (24.1 ± 3.4 and 27.3 ± 8.6 pg/mg protein, respectively) and Aβ42 (19.1 ± 4.0 and 17.6 ± 7.2 pg/mg protein, respectively) content.

TLR4 mutation increases Aβ deposits as well as soluble Aβ42 in the brain of an AD mouse model at 9 months of age

We previously reported that cerebral Aβ load increased in TLR4M Tg mice as compared to TLR4W Tg mice at 14-16 months of age [10] but we did not find such an increase at 5 months of age. Therefore, we further examined Aβ load in these AD mouse models at 9 months of age. Diffuse and fibrillar Aβ deposits were detected by immunohistochemistry using 6E10 antibody (Figure 2A and 2B) and Aβ loads were expressed by average percentage of areas showing Aβ immunoreactivity in the hippocampus and neocortex (Figure 2G and 2H). The Aβ loads in the neocortex (2.20 ± 0.15%) and hippocampus (1.42 ± 0.27%) increased in TLR4M Tg mice as compared to TLR4W Tg mice (1.40 ± 0.15%, P < 0.005 for the neocortex and 1.08 ± 0.06%, P < 0.01 for the hippocampus, n = 6 for each group). Fibrillar Aβ deposits were visualized by thioflavin S fluorescence (Figure 2C and 2D). The Aβ load in TLR4M Tg mice (0.728 ± 0.064% for the neocortex and 0.601 ± 0.080% for the hippocampus) was greater than that in TLR4W Tg mice (0.487 ± 0.056%, P = 0.021 for the neocortex and 0.368 ± 0.033%; P = 0.026 for the hippocampus, n = 6 for each group) (Figure 2I and 2J).

Activated microglia/myeloid cells in the neocortex and hippocampus were immunostained for expression of CD11b (Figure 2E and 2F) and immunoreactive areas were quantified (Figure 2K and 2L). There was no difference in expression levels of CD11b between TLR4W and TLR4M Tg mice. No difference was found between the two groups in the ratios of CD11b-immunoreactive area to 6E10-immunoreactive area (Figure 2M and 2N).

Levels of buffer-soluble Aβ in the cerebrum of TLR4M Tg mice increased at 14-16 months of age as compared to TLR4W Tg mice [10] but not at 5 months of age. Therefore, we further determined levels of buffer-soluble Aβ in the two Tg mouse groups at 9 months by Aβ40- and Aβ42-specific sandwich ELISA. The cerebral buffer-soluble Aβ42 content in TLR4M Tg mice (597.8 ± 21.3 pg/mg protein, n = 4) was significantly higher than that in TLR4W Tg mice (278.3 ± 79.3 pg/mg protein, n = 4, P < 0.01) while there was no difference between TLR4M and TLR4W Tg mice in the cerebral buffer-soluble Aβ40 content (183.2 ± 24.5 pg/mg protein and 218.6 ± 32.0 pg/mg protein, respectively, P > 0.05) (Figure 3).

Amyloid plaques are associated with activated microglia expressing TLR4

Aβ deposits have been shown to be closely associated with activated microglia and reactive astrocytes. To assess whether Aβ deposition in the brain can recruit microglia and astrocytes expressing TLR4, we performed a colocalization analysis using anti-CD11b, anti-CD45 (a transmembrane protein tyrosine phosphatase specific for migratory leukocytes including activated microglia), anti-GFAP (an astrocyte-specific intermediate filament protein) and anti-TLR4 antibodies. Brain sections from 9-month-old TLR4W Tg mice were double-stained with anti-TLR4 antibody and 1of 3 antibodis (anti-CD11b, anti-CD45, or anti-GFAP antibody). One brain section was statined with both anti-CD11b antibody and thioflavin S. Almost all fibrillar Aβ deposits stained with thioflavin S were closely accompanied by CD11b-immunoreactive microglia (Figure 4A, B and 4C). Immunofluorescence staining of the brain sections revealed that CD11b- and CD45-positive microglia co-expressed TLR4 (Figure 4D, E and 4F for CD11b, and G, H and I for CD45) whereas limited expression of TLR4 was found in GFAP-positive astrocytes (Figure 4J, K and 4L).

TLR4 mutation makes an AD mouse model vulnerable to cognitive deficits in the Morris water maze

We previously reported that the APPswe/PS1dE9 transgenic mice had spatial learning and memory deficits by the Morris water maze at 12 months of age but not at 7 months [17, 18]. Because soluble Aβ42 is thought to be neurotoxic and levels of soluble Aβ42 in the brains of 9-month-old TLR4M Tg mice increased compared to those in TLR4W Tg mice, we evaluated the effects of the TLR4 mutation on spatial learning and memory by the Morris water maze in the AD mouse model at 9 months of age. In the acquisition phase, three-way ANOVA with transgene and TLR4 as main factors and days as the repeated measure revealed significant TLR4 (F _{(1, 36)} = 5.46, P < 0.05) and day (F _{(1, 144)} = 19.72, P < 0.001) effects for path lengths as well as TLR4 (F _{(1, 36)} = 5.50, P < 0.05) and day (F _{(1, 144)} = 32.84, P < 0.001) effects for escape latencies. Paired comparisons by one-way ANOVA revealed higher latencies on days 2 (F _{(1, 18)} = 8.33, P < 0.01) and 3 (F _{(1, 18)} = 4.85, P < 0.05) and higher path lengths on days 1 (F _{(1, 18)} = 4.58, P < 0.05) and 3 (F _{(1, 18)} = 4.58, P < 0.05) by TLR4M Tg mice (n = 10) than TLR4W Tg mice (n = 10), indicating that mutated TLR4 led to poorer scores than wild-type TLR4 in APP/PS1 transgenic mice. Moreover, TLR4M Tg mice had higher escape latencies than TLR4M non-Tg mice (n = 10) on days 1 (F _{(1, 18)} = 5.97, P < 0.05) and 3 (F _{(1, 18)} = 7.85, P < 0.02) (Figure 5A and 5B). No other group differences were found in the acquisition phase or in either the probe or visible platform subtasks. These data indicate that mutated TLR4 impaired acquisition in the early not the late stage of training and spared long-term memory, visual acuity, and swimming abilities.

Aβ fibrils upregulate expression of certain cytokines and chemokines via TLR4 signaling in the hippocampus and splenic monocytes

Induction of TLR signaling by its ligands in macrophages/microglia culminates in activation of transcription factors that upregulate expression of certain cytokines and chemokines, which are required for protective immune responses to pathogens and efficient clearance of damaged tissues. Fibrillar Aβ can activate microglia through interaction with cell surface receptor complexes whose essential components include TLR4 [7–9]. Therefore, we determined levels of cytokines and chemokines in 5- and 9-month-old TLR4M (n = 5 and 4, respectively) and TLR4W (n = 6 and 7, respectively) Tg mice, which are possibly modulated by TLR signaling in the AD pathogenesis. Cycle threshold (Ct) values of mRNAs for the cytokines and chemokines are provided in Table 2. At 5 months of age, mRNA levels of IL-6 in TLR4M Tg mice decreased compared with TLR4W Tg mice in the hippocampus (2-fold, P = 0.01) but the difference in the neocortex was not significant (P = 0.18) (Figure 6). There were no differences in mRNA levels of IL-1α, IL-1β, TGF-β, CCL2, CCL3, CCL4, and CCL6 between the two groups at 5 months of age. At 9 months, mRNA levels of IL-1β (2.5-fold, P < 0.0005), CCL3 (3.5-fold, P < 0.05), and CCL4 (3-fold, P < 0.05) decreased in TLR4M Tg mice as compared to those in TLR4W Tg mice in the hippocampus (Figure 6). In the neocortex, however, such a decrease in 9-month-old TLR4M Tg mice was found only in the CCL3 mRNA levels (5-fold, P = 0.01). Cerebral mRNA levels of IL-4, TNF-α, and IFN-γ were too low to satisfactorily make comparisons under the current experimental conditions.

		IL-1α	IL-1ß	IL-6	TGFß	CCL2 (MCP-1)	CCL3 (MIP-1α)	CCL4 (MIP-1ß)	CCL6 (C10)	GAPDH
9M Neocortex	TLR4M	25.7 ± 0.35	28.8 ± 0.49	29.0 ± 0.44	22.3 ± 0.39	27.4 ± 0.74	28.1 ± 0.42	28.6 ± 0.37	23.5 ± 0.50	13.2 ± 0.22
	TLR4W	24.5 ± 0.38	27.5 ± 0.30	28.6 ± 0.15	21.7 ± 0.30	27.2 ± 0.33	24.7 ± 0.36	26.2 ± 0.29	22.4 ± 0.23	13.4 ± 0.30
Hippocampus	TLR4M	25.5 ± 0.49	29.8 ± 0.57	29.2 ± 0.41	21.4 ± 0.20	28.03 ± 0.25	27.6 ± 0.34	29.2 ± 0.40	23.1 ± 0.22	12.8 ± 0.33
	TLR4W	24.1 ± 0.29	27.4 ± 0.09	28.2 ± 0.13	20.6 ± 0.07	27.7 ± 0.29	24.7 ± 0.37	26.6 ± 0.30	20.7 ± 0.24	12.6 ± 0.35
5M Neocortex	TLR4M	29.2 ± 0.50	29.5 ± 0.54	29.1 ± 0.45	23.2 ± 0.41	31.6 ± 0.93	27.4 ± 0.51	28.9 ± 0.42	24.24 ± 0.23	15.5 ± 0.45
	TLR4W	27.5 ± 0.20	28.9 ± 0.24	27.1 ± 0.28	21.9 ± 0.10	29.5 ± 0.83	26.8 ± 0.13	27.7 ± 0.44	22.6 ± 0.29	14.6 ± 0.43
Hippocampus	TLR4M	26.2 ± 0.34	29.2 ± 0.33	28.2 ± 0.06	22.0 ± 0.40	28.3 ± 0.49	26.9 ± 0.62	29.0 ± 0.48	23.8 ± 0.34	14.1 ± 0.22
	TLR4W	24.9 ± 0.24	28.4 ± 0.24	27.2 ± 0.15	21.3 ± 0.22	26.9 ± 0.32	25.7 ± 0.54	28.5 ± 0.40	23.2 ± 0.69	14.0 ± 0.06

Microglia are cells of myeloid origin and considered to be macrophages/monocytes in the central nervous system. Therefore, we have isolated CD11b⁺ splenocytes (monocytes) from 2-month-old TLR4W and TLR4 M Tg mice by fluorescence-activated cell sorting (FACS). After treating CD11b⁺ cells with fibrillar Aβ, expression levels of cytokines and chemokines were determined by real-time PCR as described above. Ct values of mRNA for cytokines and chemokines are shown in Table 3. mRNA levels of IL-1α, IL-1β, IL-6, CCL3 and CCL4 in fibrillar Aβ-treated TLR4M CD11b⁺ cells were significantly lower than those in fibrillar Aβ-treated TLR4W CD11b⁺ cells (P < 0.005 for all measures) (Figure 7).

	IL-1α	IL-6	CCL4	CCL3	IL-1β	GAPDH
TLR4M	27.7 ± 0.10	31.2 ± 0.40	29.3 ± 0.76	28.1 ± 0.13	24.6 ± 0.06	20.2 ± 0.09
TLR4W	25.9 ± 0.08	26.4 ± 0.46	26.2 ± 0.15	23.2 ± 0.06	21.6 ± 0.25	19.1 ± 0.06