Aβ1-42 reduces P-glycoprotein in the blood-brain barrier through RAGE-NF-κB signaling

Abstract

The reduced clearance of amyloid-β peptide (Aβ) from the brain partly accounts for the neurotoxic accumulation of Aβ in Alzheimer's disease (AD). Recently, it has been suggested that P-glycoprotein (P-gp), which is an efflux transporter expressed on the luminal membrane of the brain capillary endothelium, is capable of transporting Aβ out of the brain. Although evidence has shown that restoring P-gp reduces brain Aβ in a mouse model of AD, the molecular mechanisms underlying the decrease in P-gp expression in AD is largely unknown. We found that Aβ1-42 reduced P-gp expression in the murine brain endothelial cell line bEnd.3, which was consistent with our in vivo data that P-gp expression was significantly reduced, especially near amyloid plaques in the brains of five familial AD mutations (5XFAD) mice that are used as an animal model for AD. A neutralizing antibody against the receptor for advanced glycation end products (RAGE) and an inhibitor of nuclear factor-kappa B (NF-κB) signaling prevented the decrease in Aβ1-42-induced P-gp expression, suggesting that Aβ reduced P-gp expression through NF-κB signaling by interacting with RAGE. In addition, we observed that the P-gp reduction by Aβ was rescued in bEnd.3 cells receiving inductive signals or factors from astrocytes making contacts with endothelial cells (ECs). These results support that alterations of astrocyte-EC contacts were closely associated with P-gp expression. This suggestion was further supported by the observation of a loss of astrocyte polarity in the brains of 5XFAD mice. Taken together, we found that P-gp downregulation by Aβ was mediated through RAGE-NF-κB signaling pathway in ECs and that the contact between astrocytes and ECs was an important factor in the regulation of P-gp expression.

Figure 1

P-glycoprotein (P-gp) expression was decreased in the brain capillaries of five familial Alzheimer's disease mutations (5XFAD) mice. (a and b) P-gp protein levels were measured by performing western blotting on total cerebral homogenates and cerebral capillaries that were isolated from the brains of wild-type and 5XFAD mice. β-Actin was used as a loading control. (c) Confocal microscopic analysis of P-gp and amyloid plaques in cortical sections (30 μm) from 6-month-old control and 5XFAD mice (n=3 for each). The amyloid plaques were labeled with the specific anti-amyloid-β peptide (Aβ) antibody (green; 4G8). P-gp (red) was clearly detected along the capillaries of the wild-type mice and far from the amyloid plaque deposits in 5XFAD mice (yellow arrow heads and the parts shown in the yellow dotted circles), but hardly detected near the amyloid plaque deposits in 5XFAD mice (the parts shown in the white dotted circles). Scale bars=10 and 30 μm. (d) Three-dimensional-structured illumination microscopy (3D-SIM) image of P-gp and amyloid plaques in the brains of wild-type and 5XFAD mice (n=3 for each). Amyloid plaques and P-gp were identified by the immunofluorescent detection of the 4G8 (green) and anti-P-gp antibody (red). Each panel shows different regions in the brains. The arrow heads indicate P-gp expression. Scale bar=2 μm

Figure 2

Aβ_1–42 decreased P-gp expression in bEnd.3 cells. (a) P-gp expression levels were measured by a western blot analysis in bEnd.3 cells that were incubated with various doses of Aβ_1–42 for 24 h. Tubulin was used as a loading control. (b) A histogram represents the densitometric analysis of blots, with the data expressed as mean±standard error of the mean (S.E.M.; n=4; **P<0.01 versus Aβ 0 μM). (c) Confocal microscopic analysis of the P-gp intensity in bEnd.3 cells that were treated with 5 μM of Aβ_1–42 for 24 h. The nuclei were labeled with 4′-6-diamidino-2phenylindole (DAPI; blue), and the P-gp immunoreactivity was labeled in red. Scale bar=20 μm. (d and e) Fluorescence intensity profile of each imaged cells that was determined by drawing lines parallel to the axis of the cell. The x-axis shows the length of the line, which was ∼30 μm in total, and the y-axis shows the intensity. The red line shows the intensity of P-gp, and the blue line shows that of DAPI

Figure 3

The nuclear factor-kappa B (NF-κB) pathway mediated the Aβ_1–42-induced decrease in P-gp expression. (a) Western blot analysis of the P-gp contents in bEnd.3 cells that were incubated with 5 μM of Aβ_1–42 for 24 h with or without 5 μM of BAY 11-7082. (b) The histogram represents the densitometric analysis of the blots, with the data expressed as mean±S.E.M. (n=4; **P<0.01 versus Aβ 5 μM). (c) Confocal microscopic analysis of the P-gp intensity in the bEnd.3 cells that were treated with 5 μM of Aβ_1–42 for 24 h with or without 5 μM of BAY 11-7082. The nuclei were labeled with DAPI (blue), and P-gp was immunofluorescently labeled in red. Scale bar=20 μm. (d) The fluorescence intensity profile of each imaged cell that was determined by drawing lines parallel to the axis of the cell. The x-axis shows the length of the line, which was ∼30 μm in total, and the y-axis shows the intensity. The red line shows the intensity of P-gp, and the blue line shows that of DAPI. (e) The average peak fluorescence intensity of P-gp is presented. The peak fluorescence intensity values were taken from 10 cells per group from three independent preparations (***P<0.001 versus Aβ 5 μM). (f) The p-IκBα protein levels were measured by a western blot analysis in bEnd.3 cells that were incubated with 5 μM of Aβ_1–42 for 15 min with or without 5 μM BAY 11-7082, alongside total IκBα. (g) The histogram shows a densitometric analysis of the blots, with the data expressed as mean±S.E.M. (n=4; ***P<0.001 versus Aβ 5 μM). (h) NF-κB response element luciferase gene reporter assay done in bEnd.3 cells that were treated with Aβ_1–42 for 30 min in the presence or absence of BAY 11-7082 (n=3; **P<0.01, ***P<0.001 versus Aβ 5 μM)

Figure 4

The receptor for advanced glycation end products (RAGE) mediated the Aβ_1–42-induced decrease in P-gp expression. (a) Confocal microscopic analysis of the P-gp intensity in bEnd.3 cells that were treated with 5 μM of Aβ_1–42 for 24 h with or without 15 μg/ml of anti-RAGE antibody. The nuclei were labeled with DAPI (blue), and the P-gp immunoreactivity was labeled in red. Scale bar=20 μm. (b) The fluorescence intensity profile of each imaged cell was determined by drawing lines that were parallel to the axis of the cell. The x-axis shows the length of the line, which was ∼30 μm in total, and the y-axis shows the intensity. The red line shows the intensity of P-gp, and the blue line shows that of DAPI. (c) The average peak fluorescence intensity of P-gp is presented. The peak fluorescence intensity values were taken from 10 cells per group, and from three independent preparations (***P<0.001 versus Aβ 5 μM). (d) P-gp expression levels were measured by western blots in bEnd.3 cells that were incubated with 5 μM of Aβ_1–42 for 24 h after transfection with mock or full-length human RAGE. (e) The histogram represents the densitometric analysis of the blots, with the data expressed as mean±S.E.M. (n=3; **P<0.01, ***P<0.001 versus mock-ctrl)

Figure 5

Aβ_1–42 induced a decrease in P-gp expression only when the astrocyte–endothelial cell (EC) contact was not made. (a) Schematic diagrams illustrating the conditioned medium (CM) transfer from the control, and the contact and non-contact bEnd.3-astrocyte co-culture systems. (b) Western blot analysis of the P-gp contents in bEnd.3 cells that were incubated with 5 μM of Aβ_1–42 for 24 h in the presence of the described CM. (c) Histogram showing densitometric analysis of the blots, with the data expressed as mean±S.E.M. (n=4; **P<0.01 versus Aβ 5 μM)

Figure 6

Astrocyte polarization was disrupted in the brains of 5XFAD mice. (a) Confocal microscopic analysis of aquaporin 4 (Aqp4) and blood vessels in the cortical sections (30 μm) from 6-month-old control and 5XFAD mice (n=3 for each). Aqp4 (red) was localized to the blood vessels (green) in controls, but was diffused (white arrow heads) and distributed to bridging microvessels in 5XFAD mice. Scale bar=50 μm. (b) Compared to wild-type mice, Aqp4-positive cells were significantly decreased in the brains of 5XFAD mice. (c) Confocal microscopic analysis of laminin α2 (Lama2) and the blood vessels in the cortical sections of 6-month-old control and 5XFAD mice (n=3 for each). Lama2 (red) was localized to the blood vessels (green) in controls, but it was reduced in intensity and in the extent of the covered area in 5XFAD mice. Scale bar=50 μm. (d) Compared to wild-type mice, Lama2-positive cells were significantly decreased in the brains of 5XFAD mice. Both P-values (*P<0.05 and ***P<0.001) means the 5XFAD mice versus the wild-type mice)