Selective antihypertensive dihydropyridines lower Aβ accumulation by targeting both the production and the clearance of Aβ across the blood-brain barrier

Abstract

Several large population-based or clinical trial studies have suggested that certain dihydropyridine (DHP) L-type calcium channel blockers (CCBs) used for the treatment of hypertension may confer protection against the development of Alzheimer disease (AD). However, other studies with drugs of the same class have shown no beneficial clinical effects. To determine whether certain DHPs are able to impact underlying disease processes in AD (specifically the accumulation of the Alzheimer Aβ peptide), we investigated the effect of several antihypertensive DHPs and non-DHP CCBs on Aβ production. Among the antihypertensive DHPs tested, a few, including nilvadipine, nitrendipine and amlodipine inhibited Aβ production in vitro, whereas others had no effect or raised Aβ levels. In vivo, nilvadipine and nitrendipine acutely reduced brain Aβ levels in a transgenic mouse model of AD (Tg PS1/APPsw) and improved Aβ clearance across the blood-brain barrier (BBB), whereas amlodipine and nifedipine were ineffective showing that the Aβ-lowering activity of the DHPs is independent of their antihypertensive activity. Chronic oral treatment with nilvadipine decreased Aβ burden in the brains of Tg APPsw (Tg2576) and Tg PS1/APPsw mice, and also improved learning abilities and spatial memory. Our data suggest that the clinical benefit conferred by certain antihypertensive DHPs against AD is unrelated to their antihypertensive activity, but rely on their ability to lower brain Aβ accumulation by affecting both Aβ production and Aβ clearance across the BBB.

Figure 1

Dose-dependent effect of antihypertensive calcium channel blockers on Aβ production by 7W CHO cells overexpressing APP. (A) Effect of a 24 h treatment with different calcium channel blockers on Aβ_1–40 production. A β_1–40 values were expressed as a percentage of the Aβ_1–40 values obtained in the vehicle treatment conditions ± S.E.M. ANOVA reveals statistically significant main effect of amlodipine (P < 0.001), nilvadipine (P < 0.001), nitrendipine (P < 0.05) and nimodipine (P < 0.002), but no statistically significant main effect of felodipine (P = 0.1), isradipine (P = 0.055), nifedipine (P = 0.245), diltiazem (P = 0.09) or verapamil (P = 0.173) on Aβ_1–40 production. Post hoc analyses show significant differences between Aβ_1–40 values in untreated cells and cells treated with amlodipine at 1 (P < 0.02), 2.5 (P < 0.001), 5 (P < 0.001) and 10 μmol/L (P < 0.001), with nilvadipine at 5 and 10 μmol/L (P < 0.001), with nitrendipine at 5 and 10 μmol/L (P < 0.001) and with nimodipine at 5 and 10 μmol/L (P < 0.001). (B) Effect of a 24 h treatment with different calcium channel blockers on Aβ_1–42 production. Aβ_1–42 values were expressed as a percentage of the Aβ_1–42 values obtained in the vehicle treatment conditions ± S.E.M. ANOVA reveals statistically significant main effect of amlodipine (P < 0.001), nilvadipine (P < 0.001), isradipine (P < 0.001) and nimodipine (P < 0.001), but no statistically significant main effect of felodipine (P = 0.577), nitrendipine (P = 0.293), nifedipine (P = 0.663), verapamil (P = 0.968) and diltiazem (P = 0.308) on Aβ_1–42 production. Post hoc analyses show significant differences between Aβ_1–42 values in untreated cells and cells treated with amlodipine at 1, 2.5, 5 and 10 μmol/L (P < 0.005), with nilvadipine at 2.5, 5 and 10 μmol/L (P < 0.01), with nimodipine at 2.5, 5 and 10 μmol/L (P < 0.002) and with isradipine at 2.5, 5 and 10 μmol/L (P < 0.001). (Average Aβ_1–40 values observed in the control conditions were 2,092.3 pg/mL whereas Aβ_1–42 values were 189.5 pg/mL).

Figure 2

Effect of nilvadipine and amlodipine on APP processing. Western-blot depicting the effect of a 24-h treatment with 5, 10 and 20 μmol/L of amlodipine and nilvadipine on the secretion of APPsα an APPsβ in the culture medium of 7W CHO cells overexpressing APP, showing an inhibition of extracellular sAPPβ production by nilvadipine and amlodipine. In cell extracts, no alteration of full-length APP (FL-APP) level was observed with nilvadipine and amlodipine treatments, whereas an accumulation of APP-CTFα was observed with amlodipine. Actin was used as a loading control.

Figure 3

Effect of antihypertensive calcium channel blockers on human BACE-1 activity using cell free assays. (A) BACE-1 mediated proteolytic process of a FRET peptide substrate containing the Swedish peptide sequence (EVNLDAEFK) as a function of time. The inhibitor BACE-IV was used as a reference inhibitor for BACE-1. (B) Quantification of BACE-1 activity using the chemoluminescent HitHunter assay. Data show that calcium channel blockers do not inhibit BACE-1 activity directly and, in particular nilvadipine and amlodipine (no statistically significant main effect of nilvadipine and amlodipine doses was observed by ANOVA [P > 0.05]) do not have any impact on BACE-1 activity for the dose range tested.

Figure 4

Effect of antihypertensive DHPs on Aβ production in neuroblastoma SHSY cells overexpressing the C99 C-terminal fragment of APP. C99 overexpressing SHSY cells were treated for 24 h with different DHPs. Results were expressed as a percentage of the Aβ_1–40 values observed in the vehicle treatment conditions ± S.E.M. ANOVA reveals no statistically significant main effect for the different doses of the DHPs tested on Aβ_1–40 production (P > 0.05) showing that antihypertensive DHPs do not inhibit the production of Aβ_1–40 by C99-overexpressing cells. (Average Aβ_1–40 values observed in the control conditions were 770.1 pg/mL.)

Figure 5

(A) Effect of an acute treatment with nilvadipine, amlodipine, nifedipine, nitrendipine and DAPT on brain soluble Aβ levels in Tg PS1/APPsw mice. DAPT (10 mg/kg of body weight), nilvadipine (2 mg/kg), amlodipine (2 mg/kg) and nifedipine (2 mg/kg) were dissolved in 50% DMSO/PBS and administered to Tg PS1/APPsw mice (6-month-old) intraperitoneally for 4 d. Vehicle treated Tg PS1/APPsw mice received an i.p. injection of the vehicle for 4 d. Within 1 h after the last injection, brains and plasma samples were collected. Brain soluble Aβ levels and plasma Aβ levels were quantified by ELISAs. ANOVA reveals statistically significant main effect of the acute treatment with nilvadipine (P < 0.05), nitrendipine (P < 0.01) and DAPT (P < 0.004) on brain Aβ level but no main effect for amlodipine (P = 0.184) and nifedipine (P = 0.36) treatments. Post hoc analysis show statistically significant differences in brain Aβ_1–40 levels between control and nilvadipine treated mice (P < 0.05) as well as significant differences between brain Aβ_1–42 and Aβ_1–40 levels between vehicle and nitrendipine-treated mice (P < 0.01) and between vehicle- and DAPT-treated mice (P < 0.004). (Average brain soluble Aβ_1–40 and Aβ_1–42 values observed in vehicle-treated mice were respectively 10249.1 and 8770.8 pg/mg of protein). (B) Effect of an acute treatment with nilvadipine, amlodipine, nifedipine, nitrendipine and DAPT on plasma Aβ levels in Tg PS1/APPsw mice. An elevation of plasma Aβ_1–40 and Aβ_1–42 was observed following nilvadipine and nitrendipine treatments, whereas a decreased plasma A β_1–42 level was observed following DAPT treatment. ANOVA reveals a statistically significant main effect of the acute treatment with nilvadipine (P < 0.05) and nitrendipine (P < 0.01) on plasma Aβ_1–40 level and significant main effect of nilvadipine (P < 0.05), nitrendipine (P < 0.01) and DAPT (P < 0.04) on plasma Aβ_1–42, but no main effect for amlodipine and nifedipine treatments (P > 0.05). Post hoc comparisons shows significant differences in plasma Aβ_1–40 values between vehicle, nilvadipine and nitrendipine treated mice (P < 0.05) and significant differences in plasma Aβ_1–42 values between vehicle, nilvadipine, nitrendipine and DAPT treatments (P < 0.05). (Average plasma Aβ_1–40 and A β_1–42 values observed in vehicle treated mice were respectively 1384.9 and 293.9 pg/mL.)

Figure 6

Effects of amlodipine, nitrendipine and nilvadipine on Aβ transcytosis across an in vitro model of the BBB. Fluorescein labeled human Aβ_1–42 was added to the basolateral compartment (“brain” side) whereas different doses of amlodipine, nitrendipine and nilvadipine were added to the apical side (“blood” side) of the in vitro BBB model. The amount of fluorescein-Aβ_1–42 was quantified in the apical side over a period of 90 min to calculate the apparent permeability of Aβ_1–42 for the different treatments conditions. ANOVA shows a statistically significant effect of nilvadipine (P < 0.05), of nitrendipine (P < 0.001) but not of amlodipine (P = 0.910) on Aβ transcytosis in vitro across the BBB layer of human brain microvascular endothelial cells. Post hoc analysis reveals a significant effect of nilvadipine at 10 μmol/L (P < 0.05) and a significant effect of nitrendipine at 5 μmol/L (P < 0.01) and 10 μmol/L (P < 0.001), showing that nilvadipine and nitrendipine stimulate the transport of Aβ from the brain to the periphery in an in vitro model of the BBB, whereas amlodipine is inefficient.

Figure 7

(A) Effect of amlodipine, nitrendipine and nilvadipine on Aβ clearance across the BBB in vivo. Following an i.p. injection with the vehicle alone, amlodipine, nitrendipine and nilvadipine (2 mg/kg), wild-type B6/SJL F1 mice were injected stereotaxically with human Aβ_1–42. Plasma human Aβ_1–42 levels were quantified by ELISA following the intracranial injection of Aβ to reveal the amount of human Aβ1–42 clearance across the BBB. ANOVA shows a significant main effect of nilvadipine (P < 0.001), of nitrendipine (P < 0.001), but not of amlodipine (P = 0.969) on plasma Aβ_1–42 level. Post hoc comparisons reveal statistically significant differences between plasma Aβ_1–42 for the vehicle and the nilvadipine treatment groups (P < 0.004), as well as significant differences between the vehicle and nitrendipine treatment groups (P < 0.001), showing that nitrendipine and nilvadipine stimulate the clearance of Aβ across the BBB whereas amlodipine is inefficient. (B) Effect of the nilvadipine treatment and of the intracranial Aβ injection on BBB leakiness. Mice were subjected either to cold injury of the cortex to induce a breakdown of the BBB (positive control), were untreated (control mice), were injected intraperitoneally with 100 μL of the vehicle (50% DMSO in PBS) (IP vehicle) or with 2 mg/kg of nilvadipine (IP nilvadipine) or injected intracranially with 3 μL of the vehicle (IC vehicle) or with 3 nmol of Aβ_1–42 (IC Aβ_1–42). Potential BBB leakage was evaluated by injecting the Evans blue dye intraperitoneally and measuring its extravasation in the brain 1 h later. Representative pictures of the brains for the different treatments are shown. Leakage of the Evans blue dye at the sites of the cold injury is evident, the site of the needle puncture created by the intracranial injection is indicated by an arrow. The histogram represents the amount of Evans blue dye extravasated in the brain of the animals for the different treatment groups. ANOVA reveals a significant main effect of the treatments on brain levels of Evans blue (P < 0.001). Post hoc analyses show a statistically significant elevation of Evans blue in the brains of the mice subjected to the cold injury (P < 0.005) compared with all the other groups but no statistically significant difference between untreated mice and IC vehicle injected animals (P = 0.999), IC Aβ_1–42 injected mice (P = 0.999), IP vehicle injected mice (P = 0.982) and nilvadipine treated mice (P = 0.999) showing that none of these treatments increased the leakiness of the BBB to Evans blue.

Figure 8

Effect of a chronic oral treatment with nilvadipine on Aβ burden in Tg APPsw and Tg PS1/APPsw mice. Three-month-old Tg APPsw were fed for 17 months with a powder diet containing 0.03% (weight/weight) of nilvadipine formulation of oral dosage (n = 10) or a placebo (n = 12), whereas 10-month-old Tg PS1/APPsw were fed for 10 months with nilvadipine (n = 8) or a placebo (n = 7). (A) Representative pictures (20 × objective) showing the amount of Aβ deposits in the hippocampus of Tg APPsw and Tg PS1/APPsw following treatment with a placebo or nilvadipine. Histograms representing the amount of Aβ burden in the cortex and hippocampus of Tg APPsw mice (B) and Tg PS1/APPsw mice (C) treated with nilvadipine and the placebo. ANOVA reveals a statistically significant main effect of the nilvadipine treatment in Tg APPsw (P < 0.003) and in Tg PS1/APPsw mice (P < 0.001). Post hoc analyses reveals a statistically significant difference in Aβ burden for the cortex and hippocampus of placebo and nivadipine treated Tg APPsw (P < 0.003) and Tg PS1/APPsw mice (P < 0.001) showing that nilvadipine is reducing Aβ accumulation in the brains of Tg APPsw and Tg PS1/APPsw mice.

Figure 9

Effect of nilvadipine on the exploratory activity of Tg PS1 and Tg PS1/APPsw mice in the open field apparatus. (The distance traveled in a 1 meter open field over 30 min is represented). Tg PS1/APPsw mice show an hyperactivity behavior in the open field compared with Tg PS1 littermate controls. The nilvadipine treatment appear to reduce the hyperactivity of Tg PS1/APPsw mice. Repeated ANOVA measures followed by post hoc analysis show statistically significant differences for the total distance traveled in the open field arena between Tg PS1/APPsw treated with nilvadipine and Tg PS1/APPsw treated with a placebo (P < 0.001), between Tg PS1/APPsw treated with nilvadipine and Tg PS1 treated with nilvadipine (P < 0.001), between Tg PS1/APPsw treated with nilvadipine and Tg PS1 treated with nilvadipine (P < 0.001) and between Tg PS1/APPsw treated with a placebo and Tg PS1 treated with a placebo (P < 0.001).

Figure 10

Effect of nilvadipine on cognitive behavior of Tg PS1 and Tg PS1/APPsw mice in the Morris water maze. (A) Learning curve for Morris water maze acquisition trials across a period of 9 d. (B) Learning curve for Morris water maze during the reversal period (hidden platform moved to a new location). (C) Effect of nilvadipine on memory retention in the Morris water maze (probe trials). Placebo-treated Tg PS1/APPsw mice have the worst learning performance in the Morris water maze compared with placebo-treated Tg PS1 litter-mate controls (P < 0.001) during the acquisition trials and the reversal. Nilvadipine treated Tg PS1/APPsw mice locate the hidden platform more efficiently than placebo-treated Tg PS1/APPsw mice (P < 0.04) during the acquisition trials and the reversal. Pooled probe trials for the Morris water maze show a significant increase in retention for Tg PS1 control littermates compared with Tg PS1/APPsw treated with the placebo (P < 0.01) as well as an improved memory retention for nilvadipine-treated Tg PS1/APPsw mice compared with placebo treated Tg PS1/APPsw (P < 0.02).