High-Density Lipoprotein Mimetic Peptide 4F Reduces Toxic Amyloid-Beta Exposure to the Blood-Brain Barrier Endothelium in Alzheimer’s Disease Transgenic Mice

Abstract

Aβ accumulation in the blood-brain barrier (BBB) endothelium, which lines the cerebrovascular lumen, is a significant contributor to cerebrovascular dysfunction in Alzheimer’s disease (AD). Reduced high-density lipoprotein (HDL) levels are associated with increased AD risk, and the HDL mimetic peptide 4F has been developed as a promising therapeutic agent to improve cerebrovascular health in AD. In this study, we evaluated the impact of 4F on ¹²⁵I-Aβ₄₂ blood-to-brain distribution using dynamic SPECT/CT imaging in both wild-type and APP/PS1 transgenic mice. Graphical analysis of the imaging data demonstrated that 4F significantly reduced the blood-to-brain influx rate in wild-type mice and the distribution of ¹²⁵I-Aβ₄₂ in the BBB endothelium in APP/PS1 mice. To elucidate the molecular mechanisms underlying the effect of 4F, we evaluated its impact on the p38 pathway and its role in mediating Aβ₄₂ trafficking in human BBB endothelial cell monolayers. Treatment with 4F significantly decreased Aβ₄₂ induced p38 activation in BBB endothelial cells. Furthermore, inhibition of p38 kinase significantly reduced endothelial accumulation of fluorescence-labeled Aβ₄₂ and luminal-to-abluminal permeability across the cell monolayer. While our previous publication has hinted at the potential of 4F to reduce Aβ accumulation in the brain parenchyma, the current findings demonstrated the protective effect of 4F in reducing Aβ₄₂ accumulation in the BBB endothelium of AD transgenic mice. These findings revealed the impact of a clinically tested agent, the HDL mimetic peptide 4F, on Aβ exposure to the BBB endothelium and offer novel mechanistic insights into potential therapeutic strategies to treat cerebrovascular dysfunction in AD.

Graphical Abstract

INTRODUCTION

The accumulation of toxic amyloid-β (Aβ) peptides as senile plaques in the brain is one of the major pathological hallmarks of Alzheimer’s disease (AD). Despite substantial efforts to develop anti-Aβ therapies for AD treatment, the success rate of clinical trials remains disappointingly low.¹ Therefore, the focus has shifted toward AD pathologies beyond Aβ deposition, emphasizing the importance of addressing multiple pathological changes to effectively combat this heterogeneous disease. Cerebrovascular dysfunction is prevalent in 80% of AD patients without vascular dementia² and has been linked to cognitive decline in AD. Disruption of the blood-brain barrier (BBB), a monolayer of endothelial cells lining the cerebral microvasculature, is considered a significant contributor to cerebrovascular dysfunction in AD.

The BBB endothelium is a critical portal that regulates the equilibrium between blood and brain Aβ pools. Specifically, the blood-to-brain influx of Aβ is assumed to be mediated by the receptor for advanced glycation end products (RAGE).^3,4 In contrast, the brain-to-blood efflux is believed to be handled via the low-density lipoprotein receptor-related protein-1 (LRP1)^5,6 and P-glycoprotein (P-gp).⁷ Although these receptors and transporters involved in Aβ transport at the BBB endothelium have been well characterized, the signaling pathways governing the trafficking process remain understudied. Additionally, the perivascular drainage pathway plays a significant role in clearing Aβ from the brain.⁸ In the AD brain, there is an upregulation of RAGE⁹ but a downregulation of LRP1¹⁰ and P-gp.¹¹ This and impaired perivascular drainage may result in elevated Aβ exposure to the BBB endothelium, which results in cerebrovascular dysfunction^12–14 and augments AD progression. Hence, it is imperative to identify novel therapeutic agents that can effectively mitigate the consequences of Aβ exposure to the BBB endothelium and treat cerebrovascular dysfunction in AD.

Dyslipidemia is associated with an increased risk of developing AD. High-density lipoprotein (HDL) particles, also known as the “good cholesterol,” have demonstrated protective effects against AD.^15,16 In the periphery, apolipoprotein A-I (ApoA-I) is the principal protein constituent of HDL particles. Since ApoA-I is known to clear arterial plaques, it has been explored as a potential therapeutic agent to treat cerebral amyloid angiopathy (CAA) in AD. For instance, studies have shown that ApoA-I overexpression reduces CAA and preserves cognitive function in APP/PS1 mice.¹⁷ Conversely, ApoA-I deficiency was shown to increase CAA, induce amyloid-associated astrocyte reactivity, and exacerbate cognitive impairment.^18,19 It has also been reported that luminal ApoA-I promoted the abluminal efflux of Aβ across cerebrovascular endothelial cell monolayers.²⁰ Hence, ApoA-I appears to be effective in alleviating Aβ load in the cerebral vasculature and ameliorating cognitive deficits associated with AD. However, the translational potential of full-length ApoA-I is limited by its high manufacturing cost and poor BBB penetration. To overcome these challenges, small ApoA-I/HDL mimetic peptides have been developed as potential alternative therapeutics. One of the most promising HDL mimetics is the 18-amino-acid peptide known as 4F, which contains the amphipathic α-helix structure as an essential motif that recapitulates the biological activity of ApoA-I.²¹ Studies in AD transgenic mice have demonstrated that oral administration of 4F in combination with pravastatin significantly inhibits Aβ plaque formation, mitigates neuroinflammation, and improves cognitive function.²² Our recent publication intended to investigate mechanisms by which 4F reduces Aβ burden in the brain has demonstrated that 4F decreases blood-to-brain influx and increases brain-to-blood efflux of radiolabeled ¹²⁵I-Aβ₄₂ across the BBB endothelium in wild-type mice, and consequently reduces the overall parenchymal Aβ load.²³ However, the impact of 4F on Aβ accumulation within the BBB endothelium, which may engender BBB dysfunction, is yet to be elucidated. Moreover, the molecular mechanisms underlying the 4F effect on Aβ trafficking remain unknown.

Here, we employed dynamic single photon emission computed tomography (SPECT) coupled with computed tomography (CT) imaging to assess the impact of 4F treatment on ¹²⁵I-Aβ₄₂ distribution kinetics in the BBB endothelium and brain parenchyma. The in vivo studies were conducted in wild-type and AD transgenic mice (APPswe/PSEN1dE9). Further, we investigated the cellular and molecular mechanisms underlying the impact of 4F on Aβ trafficking and accumulation kinetics in the BBB endothelium.

EXPERIMENTAL SECTION

Materials.

D-4F peptide (molecular weight: 2.3 kDa) was obtained from Bachem (Sunnyvale, CA) and 0.5 mg/mL stock solution was prepared in sterile PBS. Unlabeled and FITC labeled Aβ₄₂ peptides were custom synthesized by AAPPTec (Louisville, KY). ¹²⁵I-Aβ₄₂ was prepared using the chloramine-T reaction as described in our previous publication.²⁴ The unconjugated ¹²⁵I was removed by overnight dialysis in PBS. Trichloroacetic acid (TCA) precipitation was conducted to determine the purity of radiolabeled peptides.

ANIMALS AND TREATMENT

The APP/PS1 mice (B6C3-Tg (APPswe/PSEN1dE9) 85Dbo/J; stock number 004462) were purchased from Jackson Laboratory (Bar Harbor, ME). The animals were housed in the University of Minnesota animal care facility under standard conditions and provided ad libitum access to food and water. Littermates were used whenever possible to minimize the potential confounding effects of genetic backgrounds. In this study, both transgenic and nontransgenic wild-type mice aged around 12 months were randomly assigned into two groups: the treated group (D-4F, 2 mg/kg body weight) and the untreated group (sterilized PBS, the same volume as 4F solution). The study included both male and female mice, with a balanced gender distribution across all groups (n = 4 per group). The mice were treated with 4F or PBS via intraperitoneal injection for 7 days. All procedures were conducted in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals and the University of Minnesota Institutional Animal Care and Use Committee-approved protocols.

Dynamic SPECT/CT Imaging and Graphical Analysis.

On the last day of the 4F treatment, mice were intravenously injected with radiolabeled ¹²⁵I-Aβ₄₂, 1 h after receiving 4F. Immediately after the ¹²⁵I-Aβ₄₂ injection, animals were imaged for the following 40 min by dynamic single positron emission computed tomography (SPECT)-computed tomography (CT) imaging. The dynamic heart imaging data was used to deconvolve the plasma pharmacokinetics (PK) of ¹²⁵I-Aβ_42, as described by our previous publication.²⁵ Briefly, a three-compartment PK model was fitted to the heart radioactivity-time data, and biexponential equation parameters were predicted to describe the plasma PK profile. Dynamic brain imaging data and predicted plasma PK were leveraged for graphical analysis to determine the ¹²⁵I-Aβ₄₂ distribution kinetics in the brain. Two graphical analyses were conducted: the Logan plot and the Patlak plot.^26,27 In Logan plot analysis, only the initial 5 min of brain radioactivity data were used, and the tracers were assumed to distribute into the reversible compartment, which is the BBB endothelium in the current study. The Logan plot was constructed by plotting,

$$\frac{{\int }_0^t{\mathrm{Amount}}_{\mathrm{brain}}\left(\tau \right)\mathrm{d}\tau }{{\mathrm{Amount}}_{\mathrm{brain}}\left(t\right)}\mathrm{vs}\frac{{\int }_0^t{\mathrm{C}}_{\mathrm{p}}\left(\tau \right)\mathrm{d}\tau }{{\mathrm{Amount}}_{\mathrm{brain}}\left(t\right)}$$

where ${\mathrm{Amount}}_{\mathrm{brain}}\left(t\right)$ is brain radioactivity at time $t$; ${\int }_0^t{\mathrm{Amount}}_{\mathrm{brain}}\left(\tau \right)d\tau$ ${\mathrm{C}}_{\mathrm{p}}\left(t\right)$ is the area under the curve (AUC) of brain radioactivity between time 0 and $t$; ${\mathrm{C}}_{\mathrm{p}}\left(t\right)$ is the predicted plasma concentration of intact tracer at time ${\int }_0^{\mathrm{t}}{\mathrm{C}}_{\mathrm{p}}\left(t\right)d\tau$ is plasma AUC between time 0 and $t$. The slope of the Logan plot is referred to as the distributional volume in the BBB endothelium $\left({\mathit{V}}_{\mathrm{d}}\right)$.

In the Patlak plot analysis, 40 min of brain radioactivity data were used, and the tracers were assumed to eventually enter the irreversible compartment, which is the brain parenchyma in the current study. The Patlak plot was constructed by plotting,

$$\frac{{\mathrm{Amount}}_{\mathrm{brain}}\left(t\right)}{{\mathrm{C}}_{\mathrm{p}}\left(t\right)}\mathrm{vs}\frac{{\int }_0^t{\mathrm{C}}_{\mathrm{p}}\left(\tau \right)\mathrm{d}\tau }{{\mathrm{C}}_{\mathrm{p}}\left(t\right)}$$

The slope of the linear portion of the Patlak plot is referred to as the plasma-to-brain influx clearance, K₁.

Cell Culture.

The hCMEC/D3 cell line was a generous gift from Professor Pierre-Olivier Couraud, Institut Cochin, France. Cells were cultured as described in our previous publications.²⁸ Briefly, polarized human cerebral microvascular endothelial cell (hCMEC/D3) monolayers were grown on well-plates, on Transwell filters, on coverslip-bottomed dishes (MatTek Life Sciences, Ashland, MA) or chambered slides (Ibidi, Grafelfing, Germany) coated with collagen (Corning̈ Inc., Corning, NY). The Transwell inserts have a membrane with 24 mm diameter and 0.4 μM pore size. The nominal pore density is 4 × 10⁶ pores per cm². The cell seeding density was around 25,000 cells/cm² for well-plates as well as coverslip-bottomed dishes and 50,000 cells/cm² for Transwell inserts. The cells were cultured at 5% CO₂ and 37 °C for 5−7 days to reach confluency. The TEER values are routinely measured in our lab for the hCMEC/D3 monolayers, and the values were found to be around 80−120 Ω/cm². All the monolayers were moved to 1% serum medium the day before the experiment.

Western Blotting.

To investigate the impact of Aβ₄₂ on p38 phosphorylation, hCMEC/D3 cells were treated with 1 μM Aβ₄₂ in DMEM media for 15 min or 1 h. To evaluate the 4F effect, cells were pretreated with 10 μg/mL of 4F for 1 h, followed by Aβ₄₂ addition. After the treatment, cell lysates were prepared, and Western blotting analysis was conducted following the protocol described in our previous publications.²⁸ The following primary antibodies were used (1:1000 dilution): phospho-p38 MAPK (Thr180/Tyr182) (#9211), p38 MAPK (#8690) (Cell Signaling Technology, Danvers, MA).

FITC-Aβ₄₂ Exocytosis.

Polarized hCMEC/D3 monolayers were cultured on 24 mm Transwell inserts (Corning Inc., Corning, NY) and pretreated with 4F (10 μg/mL) for 1.5 h. Subsequently, both the luminal and abluminal chambers were spiked with FITC-Aβ₄₂ (F-Aβ₄₂, 1 μM) and incubated for 1 h at 37 °C. Following the incubation period, the media was aspirated, and the filter surface was washed twice with PBS to remove nonspecifically bound peptides. The cells were then supplied with phenored-free DMEM either without (control) or with 4F (treatment). Sample aliquots of 100 μL were collected from both the luminal and abluminal chambers at intervals of 0, 30, 45, 60, 75, and 90 min. Immediately after sampling, an equivalent volume of blank media (100 μL) was added to maintain the total volume of the culture media. The fluorescence of F-Aβ₄₂ was quantified using a microplate reader (Molecular Devices, San Jose, CA) with an excitation wavelength of 488 nm and an emission wavelength of 525 nm. The accumulative amount of F-Aβ₄₂ was plotted against time, and a simple linear regression was conducted to determine the exocytosis rate.

Flow Cytometry.

hCMEC/D3 monolayers grown in 6-well plates were treated with 1 μM p38 inhibitor SB203580 (Cayman Chemical, MI) for 30 min, followed by an additional 60 min incubation with 1 μM of F-Aβ₄₂. After the treatment, the cells were subjected to flow cytometry analysis as described in our previous publications.²⁸

Confocal Microscopy.

To study the impact of 4F on lysosomal accumulation of F-Aβ₄₂ in the BBB endothelial cells, hCMEC/D3 monolayers cultured on 35 mm coverslip bottomed dishes were coincubated with 10 μg/mL 4F and 1 μM F-Aβ₄₂ for 1.5 h, followed by an additional 30 min incubation with 75 nM LysoTracker Red (Thermo Fisher Scientific-Molecular Probes, Waltham, MA). After that, cells were washed two times with PBS, maintained under an atmosphere humidified with 5% CO₂ at 37 °C, and imaged live using a Nikon A1Rsi HD confocal microscope with SIM super-resolution (Nikon, Tokyo, Japan).

To evaluate the impact of p38 inhibition on intracellular accumulation of F-Aβ₄₂, hCMEC/D3 monolayers cultured on 8-well chambered slides were treated with 1 μM SB203580 (Cayman Chemical, MI) for 30 min, followed by an additional 60 min incubation with 1 μM F-Aβ₄₂. Cells were washed three times with ice-cold PBS and then fixed in 4% paraformaldehyde solution for 20 min at room temperature. After fixation, the cells were washed three times with PBS, and the nuclei were stained for 10 min using 300 nM DAPI (Thermo Fisher Scientific, Waltham, MA) solution. Cells were mounted using liquid mounting media (ibidi, Grafelfing, Germany) and then̈ imaged using a Nikon A1Rsi HD confocal microscope with SIM super resolution (Nikon, Tokyo, Japan).

Luminal-to-Abluminal Permeability across the BBB Cell Monolayers.

Polarized hCMEC/D3 cell monolayer grown on 24 mm Transwell inserts was pretreated with 1 μM SB203580 for 30 min, after which 1 μM F-Aβ₄₂ was added to the luminal side. The receiver medium was sampled from the abluminal side at intervals of 10, 20, 30, 45, 60, and 90 min. Immediately after sampling, an equivalent volume of blank media (100 μL) was added to maintain the total volume of the culture media. The fluorescence of F-Aβ₄₂ was quantified using a microplate reader (Molecular Devices, San Jose, CA) with an excitation wavelength of 488 nm and an emission wavelength of 525 nm. The flux of F-Aβ₄₂ across the BBB endothelium was calculated as the slope of the cumulative amount of F-Aβ₄₂ reaching the abluminal side per cm² versus time. The apparent permeability coefficients were then calculated by normalizing flux with the initial concentration of F-Aβ₄₂ added to the luminal compartment.

Statistics.

Unless otherwise specified in the figure legends, all the data in the table are expressed as mean ± standard deviation. Error bars in all the plots represent the standard deviations. The statistical significance of differences between two groups was determined using Student’s t test. For comparisons among more than two groups, one-way ANOVA followed by Bonferroni’s multiple comparison test was employed. All statistical analyses were conducted using Prism GraphPad software version 9 (La Jolla, CA).

RESULTS

Impact of 4F on Plasma PK of ¹²⁵I-Aβ₄₂ in Wild-Type and APP/PS1 Mice.

The plasma PK of ¹²⁵I-Aβ₄₂ in both wild-type (WT) and APPswe/PSEN1dE9 (AD) mice with or without 4F treatment were deconvolved from dynamic heart imaging data obtained by SPECT/CT imaging. This analysis employed a model-based deconvolution method detailed in our previous publication.²⁵ The model adequately described the heart radioactivity-time data across all four animal groups (Figure 1A,D). Interestingly, WT mice treated with 4F exhibited higher heart radioactivity of ¹²⁵I-Aβ₄₂ compared to their untreated counterparts. Conversely, in AD mice, 4F treatment reduced heart radioactivity of ¹²⁵I-Aβ₄₂, which declined rapidly with time. This shift was consistently reflected in the deconvolved plasma PK, where 4F elevated plasma concentrations of ¹²⁵I-Aβ₄₂ in WT mice but reduced them in AD mice. Finally, the calculated plasma PK parameters revealed that 4F treatment led to a substantial increase in systemic clearance of ¹²⁵I-Aβ₄₂ in AD mice (1.6 mL/min in PBS-treated versus 4.3 mL/min in 4F-treated mice). However, the clearance was decreased from 16.1 to 9.5 mL/min by 4F treatment in WT mice (Figure 1C,F).

Figure 1.

Impact of 4F on plasma pharmacokinetics (PK) of ¹²⁵I-Aβ42 in wild-type (WT) and APP/PS1 (AD) mice. Mice were administered daily with either PBS or 2 mg/kg of D-4F via intraperitoneal injection for 7 days. On the last day, ¹²⁵I-Aβ₄₂ was injected via the femoral vein an hour after the D-4F administration. Dynamic SPECT/CT imaging of the mice was performed immediately following ¹²⁵I-Aβ₄₂ administration. Plasma PK was deconvolved from dynamic heart imaging data using a previously established modeling approach. (A) Model fitting of dynamic heart imaging data in WT animals treated with or without 4F. (B) Deconvolved plasma PK profiles in WT animals treated with or without 4F. (C) Predicted PK parameters in WT animals treated with PBS or 4F, where V_c is the volume of distribution in the central compartment, CL is systemic clearance, T_1/2,α and T_1/2,β are the half-lives of the distributional and terminal phases, respectively. (D) Model fitting of dynamic heart imaging data in AD animals treated with or without 4F. (E) Deconvolved plasma PK profiles in AD animals treated with or without 4F. (F) Predicted PK parameters in AD animals treated with PBS or 4F.

Impact of 4F on ¹²⁵I-Aβ₄₂ Distribution from Plasma to Brain in WT and AD Mice.

To assess the effect of 4F on the plasma-to-brain distribution of ¹²⁵I-Aβ₄₂, dynamic brain SPECT/CT imaging and the deconvolved plasma PK data were used to conduct graphical analyses. The first analysis utilized the Logan plot, and it informed the distribution kinetics of ¹²⁵I-Aβ₄₂ into the reversible compartment, which was assumed to be the BBB endothelium (Figure 2A).

Figure 2.

Impact of 4F on ¹²⁵I-Aβ₄₂ distribution in the BBB endothelium determined by Logan-plot analysis. (A) Demonstration of the PK model underlying Logan plot analysis. The three-compartment model system describes the distribution of intravenously injected ¹²⁵I-Aβ₄₂ between the plasma, peripheral tissue, and BBB compartments. The distribution between plasma and the BBB compartment is assumed to be reversible. The model assumptions hold for the initial 5 min after ¹²⁵I-Aβ₄₂ injection. CL, systemic clearance. CL_d, distributional clearance between the plasma and peripheral tissue compartment. K₁, influx clearance representing ¹²⁵I-Aβ₄₂ entry into the BBB endothelium. k₂, first-order rate constant describing the efflux of ¹²⁵I-Aβ₄₂ from the BBB endothelium. (B, C) Representative Logan plots were generated using brain dynamic SPECT/CT imaging data obtained from (B) WT and (C) AD mice with or without 4F treatment. (D) K₁/k₂ values (representing the ¹²⁵I-Aβ₄₂ distributional volume (V_d) in the BBB endothelium) of each animal group (n = 4). The values were determined from the slopes of the Logan plots. **p < 0.01, unpaired two-tailed t test.

Notably, in AD mice treated with 4F, the distributional volume (V_d), determined from the slope of the Logan plot, exhibited a significant decrease compared to PBS-treated controls (Figure 2C). Specifically, the V_d value was decreased from 81.9 ± 15.5 to 46.8 ± 6.37 mL upon 4F treatment. However, there was no significant impact of 4F on ¹²⁵I-Aβ₄₂ distribution kinetics in WT mice (Figure 2D). The second analysis involved the Patlak plot, and it determined ¹²⁵I-Aβ₄₂ influx from plasma to the irreversible compartment, which was assumed to be the brain parenchyma (Figure 3A). Interestingly, in 4F-treated WT mice, the influx clearance (K₁) was significantly lower (0.70 ± 0.10 × 10⁻³ mL/min) compared to PBS-treated controls (1.27 ± 0.08 × 10⁻³ mL/min). Conversely, in AD mice, 4F exhibited an opposite effect, significantly increasing the K₁ value from 0.70 ± 0.05 × 10⁻³ to 1.94 ± 0.22 × 10⁻³ mL/min (Figure 3B–D).

Figure 3.

Impact of 4F on ¹²⁵I-Aβ₄₂ influx from plasma to brain parenchyma determined by Patlak-plot analysis. (A) Demonstration of the PK model underlying Patlak-plot analysis. The three-compartment model system describes the distribution of intravenously injected ¹²⁵I-Aβ₄₂ between the plasma, peripheral tissue, and brain parenchymal compartments. The distribution between plasma and the brain parenchymal compartment is assumed to be irreversible. CL, systemic clearance. CL_d, distributional clearance between the plasma and peripheral tissue compartment. K₁, influx clearance representing ¹²⁵I-Aβ₄₂ entry into the brain parenchyma. The model assumptions hold for a later time (>5 min) after ¹²⁵I-Aβ₄₂ injection. (B, C) Representative Patlak plots were generated using brain dynamic SPECT/CT imaging data obtained from (B) WT and (C) AD mice with or without 4F treatment. (D) K₁ values (representing the ¹²⁵I-Aβ₄₂ influx clearance into the brain parenchyma) of each animal group (n = 4). The values were determined from the slopes of the Patlak plots. ***p < 0.001, ****p < 0.0001, unpaired two-tailed t test.

Impact of 4F on FITC-Aβ₄₂ Lysosomal Accumulation and Exocytosis in BBB Endothelial Cells.

The in vivo study suggested that 4F treatment decreased ¹²⁵I-Aβ₄₂ accumulation in the BBB while increasing the uptake by brain parenchyma in AD mice. This led us to hypothesize that 4F might achieve this effect by modulating Aβ₄₂ trafficking in BBB endothelial cells. To test this hypothesis, we first examined the impact of 4F on F-Aβ₄₂ accumulation in the lysosomes of hCMEC/D3 endothelial cells. As shown in Figure 4A, after 2 h of incubation, F-Aβ₄₂ (green) was observed to colocalize with LysoTracker dye (red), which is predominately a lysosomal marker. However, when cells were coincubated with 4F, the colocalization was substantially reduced. Subsequently, we evaluated the impact of 4F on F-Aβ₄₂ exocytosis at the BBB endothelial cell monolayer. Our findings revealed that pretreatment with 4F significantly increased F-Aβ₄₂ exocytosis to the abluminal side (Figure 4I), whereas no significant difference was observed on the luminal side.

Figure 4.

Impact of 4F on FITC-Aβ₄₂ lysosomal accumulation and exocytosis in BBB endothelial cells. (A−F) Representative images showing colocalization of FITC-Aβ₄₂ (green) in hCMEC/D3 cells labeled with LysoTracker (red) under (A−C) untreated and (D−F) 4F treated conditions. (G) Experiment design of FITC-Aβ₄₂ exocytosis study. (H, I) Exocytosis rate of FITC-Aβ₄₂ to the (H) luminal and (I) abluminal side of the hCMEC/D3 cell monolayer with or without 4F pretreatment. The data was represented as fold change compared to untreated cells. **p < 0.01, student’s t test.

4F Reduced Aβ₄₂-Induced p38 Activation in BBB Endothelial Cells.

To identify molecular mechanisms underlying the 4F impact on ¹²⁵I-Aβ₄₂ brain distribution, we focused on the p38 MAPK kinase because the RPPA proteomics assay suggested that Aβ exposure elevated p38 phosphorylation in hCMEC/D3 endothelial cells (Figure 5A). This result was further confirmed by Western blot studies, which showed that Aβ significantly increased phosphorylated p38 levels following 15 min exposure, and the effect persisted 1 h after treatment (Figure 5B,C). To assess whether 4F influenced p38 kinase activation, cells were pretreated with 4F for 1 h, followed by Aβ₄₂ stimulation. Even though 4F alone did not affect p38 activation, it significantly attenuated p38 phosphorylation in the presence of Aβ₄₂ (Figure 5D,E). These results indicated that 4F could mitigate p38 activation induced by Aβ₄₂ in BBB endothelial cells.

Figure 5.

4F reduced Aβ₄₂ stimulated p38 MAPK activation. (A) RPPA assay results showing p38 and phosphorylated p38 levels upon Aβ₄₀ and Aβ₄₂ treatment. False discovery rate (FDR) was used for statistical testing. LogFC: Logarithmic fold change compared to untreated cells. (B) Representative immunoblots showing phosphorylated p38 levels in hCMEC/D3 endothelial cell monolayers treated with Aβ₄₂ (1 μM) for 15 min or 1 h. (C) Quantification of immunoblots as described in (B). **p < 0.01, ****p < 0.001, oneway ANOVA with Bonferroni post-tests. (D) Representative immunoblots depicting phosphorylated p38 levels in hCMEC/D3 endothelial cells treated with Aβ₄₂ and 4F. Cells were pretreated with 10 μg/mL 4F for 1 h, followed by incubation with Aβ₄₂ (1 μM) for an additional 15 min. (E) Quantification of immunoblots as described in (D). *p < 0.05, one-way ANOVA with Bonferroni post-tests.

p38 Inhibition Reduced Aβ₄₂ Uptake and Luminal-to-Abluminal Permeability across the BBB Monolayer.

Based on the inhibitory effect of 4F on Aβ₄₂-induced p38 activation, we hypothesized that p38 activation mediates Aβ₄₂ distribution into the brain and that 4F reduces Aβ₄₂ accumulation via p38 inhibition. To test this hypothesis, we first evaluated the impact of p38 inhibition on intracellular uptake of F-Aβ₄₂ in BBB endothelial cells. The flow cytometry analysis demonstrated that p38 inhibitor SB203580 significantly reduced the F-Aβ₄₂ uptake by the hCMEC/D3 cells (Figure 6A,B). The results were further corroborated by confocal microscopy, which showed a substantial reduction in fluorescent intensity of F-Aβ₄₂ in the cytoplasm of endothelial cells after SB203580 treatment (Figure 6C). Finally, we evaluated the permeability of F-Aβ₄₂ across the BBB monolayer with and without p38 inhibition. The permeability assay results demonstrated that SB203580 significantly decreased the luminal-to-abluminal permeability of F-Aβ₄₂ across the BBB endothelium (Figure 6D,E).

Figure 6.

p38 inhibitor, SB203580, reduced F-Aβ₄₂ cellular uptake and luminal-to-abluminal (L-A) permeability across the hCMEC/D3 monolayer. In the cellular uptake study, hCMEC/D3 monolayers were treated with 1 μM SB203580 for 30 min, followed by another 60 min with 1 μM of F-Aβ₄₂. The intracellular uptake was assessed by flow cytometry or confocal microscopy. (A) Representative histograms of hCMEC/D3 cells incubated with F-Aβ₄₂ in the control and SB203580 treated group. (B) Median fluorescence intensity (MFI) of the cell population in each group. **p < 0.01, unpaired two-tailed t test. (C) Confocal microscopic images depicting F-Aβ₄₂ (green) internalization in hCMEC/D3 cells treated with or without SB203580. Cell nuclei were stained with DAPI (blue). (D) Cumulative F-Aβ₄₂ amount was plotted against time, and a simple linear model was fitted to the linear portion of the curve. In the permeability study, polarized hCMEC/D3 monolayers cultured on Transwell filters were pretreated with 1 μM SB203580 for 30 min, followed by the addition of 1 μM F-Aβ₄₂ on the luminal side. The receiver medium was periodically sampled from the abluminal side, and F-Aβ₄₂ concentrations were determined. Data represents means ± SD (n = 4). (E) Bar charts showing permeability values of F-Aβ₄₂ across control and treated monolayers. Permeability was calculated as the flux [linear regression slope normalized to the surface area of the insets (0.33 cm²)] normalized by the initial luminal concentration. ****p < 0.0001, unpaired two-tailed t test.

DISCUSSION

Cerebrovascular dysfunction has emerged as a significant contributor to AD pathogenesis.^29,30 Anomalous Aβ exposure to the BBB endothelium is associated with various BBB dysfunctions such as increased permeability,¹² disrupted insulin signaling,¹³ and inflammation,¹⁴ ultimately culminating in cerebrovascular pathology observed in AD brains. Therefore, utilizing pharmacological agents to alleviate pathological changes triggered by Aβ accumulation in the BBB presents a promising therapeutic strategy to treat cerebrovascular dysfunction in AD.

Reduced ApoA-I plasma levels have been reported in AD patients³¹ and found to be correlated with disease severity.³² Given the protective role of ApoA-I in the peripheral vasculature, elevating ApoA-I levels might offer therapeutic benefits in addressing cerebrovascular dysfunction in AD. By conducting genetic manipulation studies in AD transgenic mice, previous research has demonstrated a negative correlation between the ApoA-I level and the severity of CAA.^17,18 The clinical translational potential of ApoA-I-based therapies is further enhanced by the development of mimetic peptides, which have lower manufacturing costs and greater brain penetration.³³ Our previous research has indicated that ApoA-I mimetic peptide 4F could decrease blood-to-brain influx but increase brain-to-blood efflux of Aβ₄₂ across the BBB, thus potentially reducing the overall accumulation of Aβ₄₂ in brain parenchyma.²³ However, the influence of 4F on Aβ accumulation within the BBB endothelium, which engenders cerebrovascular dysfunction, remains unknown. In this study, we demonstrated that 4F preserved the protective effect of ApoA-I against CAA by reducing Aβ exposure to the BBB endothelium. Further, we investigated the underlying cellular and molecular mechanisms and identified that 4F modulated Aβ trafficking by inhibiting the p38 MAPK pathway in BBB endothelial cells.

In the in vivo study, we employed dynamic SPECT/CT imaging techniques to capture brain radioactivity of ¹²⁵I-Aβ₄₂ at 1 min intervals over a 40 min period. This is a significant improvement over our previous study, in which brain radioactivity was only obtained at the last time point.²³ Continuous “sampling” enabled a more comprehensive and accurate characterization of ¹²⁵I-Aβ₄₂ distribution kinetics. Moreover, in the current study, we also used AD transgenic mice (APP/PS1) along with WT mice to evaluate 4F efficacy under Aβ pathology. We applied three methods to analyze the SPECT/CT imaging data obtained from WT and APP/PS1 mice treated with or without 4F. First, a model-based approach was used to deconvolve plasma PK of ¹²⁵I-Aβ₄₂ from dynamic heart imaging data.²⁵ Second, the Logan plot was constructed using the deconvolved plasma PK and brain radioactivity data to quantify ¹²⁵I-Aβ₄₂ distribution kinetics in the BBB endothelium. Finally, the Patlak plot was applied to evaluate the irreversible uptake kinetics of ¹²⁵I-Aβ₄₂ into the brain parenchyma. The results of plasma PK deconvolution indicated that 4F decreased systemic clearance of ¹²⁵I-Aβ₄₂ in WT mice. This is consistent with our previous study, which was conducted using a conventional time-series blood sampling method.²³ However, in APP/PS1 mice, 4F was shown to increase clearance and thus reduce the systemic exposure of ¹²⁵I-Aβ₄₂ (Figure 1).

This is of significance because increasing Aβ clearance in the periphery has been actively pursued as a promising strategy to reduce brain Aβ accumulation in AD brain. For instance, a previous study demonstrated that peritoneal dialysis significantly reduced Aβ levels in both plasma and brain in APP/PS1 transgenic mice. This intervention alleviated various AD pathologies, including tau hyperphosphorylation, neuroinflammation, synaptic dysfunction, and behavioral deficits.³⁴ However, the underlying molecular mechanisms by which 4F increased systemic clearance of ¹²⁵I-Aβ₄₂ in AD mice remain unknown.

The Aβ elimination from the peripheral circulation is primarily mediated by hepatic clearance, which is presumably mediated by LRP1 expressed on the hepatocytes.³⁵ While there is currently no evidence indicating a direct interaction between 4F and LRP1 to our knowledge, a recent study has identified LRP1 as a novel ligand for ApoA-IV in adipose tissue.³⁶ Given the structural similarity between ApoA-IV and 4F,³⁷ it is reasonable to speculate that 4F might interact with LRP1 in the liver, potentially influencing the peripheral clearance of Aβ₄₂. In addition, 4F may exert its effect on Aβ clearance through the HDL receptor SR-BI (scavenger receptor class B type I) in the liver. This possibility is supported by previous findings showing that reduction of SR-BI aggravates brain Aβ accumulation and CAA³⁸ and that 4F treatment increases SRBI expression in the liver.³⁹ Specific roles of hepatic receptors, including LRP1 and SR-BI, and their interactions with 4F in promoting Aβ clearance warrant further investigation.

Next, we conducted a graphical analysis of the dynamic brain imaging data, including the Logan plot and Patlak plot, to assess the blood-to-brain distribution kinetics of ¹²⁵I-Aβ₄₂. The main difference between these two methods is that the Logan plot informs about the distributional volume of ¹²⁵I-Aβ₄₂ in the reversible compartment. In contrast, the Patlak plot quantifies the irreversible uptake rate of ¹²⁵I-Aβ₄₂.^26,27 In the current study, reversible and irreversible compartments were assumed to be the BBB endothelium and brain parenchyma, respectively. The rationale behind this assumption is that during the initial time intervals following intravenous administration, the ¹²⁵I-Aβ₄₂ is expected to be reversibly associated with BBB endothelium via receptor binding and unbinding. However, as time progresses and ¹²⁵I-Aβ₄₂ enters the brain parenchyma, its uptake becomes irreversible. This is because the ¹²⁵I-Aβ₄₂ will undergo internalization by neuro-glia cells and bind to existing plaques, particularly in the brains of AD mice.

The Logan plot was constructed using brain radioactivity data from the first 5 min, after which linearity was lost, indicating a deviation from the reversible assumption. The Logan plot results demonstrated that 4F treatment significantly reduced ¹²⁵I-Aβ₄₂ distribution to the BBB endothelium in APP/PS1 mice but had no impact in WT mice (Figure 2). These results suggested that 4F maintained the beneficial effect of ApoA-I in mitigating Aβ exposure to the cerebral vasculature, as shown in a previous study.¹⁷ Considering its higher brain penetration and lower manufacturing costs, 4F emerges as a more promising drug candidate to treat cerebrovascular pathology in AD compared to ApoA-I. Interestingly, in Patlak plot analysis, 4F decreased the irreversible uptake rate of ¹²⁵I-Aβ₄₂ into parenchyma in WT mice but increased that in APP/PS1 mice (Figure 3). The results obtained in WT mice were consistent with those presented in our prior publication, in which 4F treatment decreased the permeability surface product (PS) value of ¹²⁵I-Aβ₄₂.²³ The observed discrepancy between WT and APP/PS1 mice could be due to the saturation of the BBB influx transport system by excessive endogenous Aβ₄₂ peptides in AD mice, leading to a reduced influx clearance compared to the WT mice, as indicated in Figure 3D. Upon 4F treatment, the Aβ₄₂ level in the systemic circulation of APP/PS1 mice was significantly decreased, thus relieving the influx receptors to increase Aβ₄₂ influx into the brain. The findings in APP/PS1 mice, where 4F decreased ¹²⁵I-Aβ₄₂ distribution in the BBB endothelium but increased its blood-to-brain influx trafficking, also imply that 4F enhances the trafficking function of the BBB endothelium, which is critical in maintaining dynamic equilibrium between blood and brain Aβ pools.⁴⁰

These shifts in ¹²⁵I-Aβ₄₂ distribution engendered by 4F treatment may eventually enhance Aβ₄₂ clearance because brain parenchyma contains more abundant enzymes and cells to clear Aβ effectively. For instance, insulin-degrading enzymes and neprilysin are the main enzymes present in brain parenchyma that degrade Aβ peptides.⁴¹ Additionally, microglia are immune cells that clear Aβ peptides through phagocytosis and enhance their subsequent lysosomal degradation. We hypothesized that 4F reduces endothelial accumulation of Aβ₄₂ by facilitating its trafficking across the BBB into the brain parenchyma to facilitate their efficient clearance in the brain parenchyma. This hypothesis was supported by the confocal micrographs that demonstrated lower lysosomal accumulation of F-Aβ₄₂ in BBB endothelial cells treated with 4F compared to untreated control cells (Figure 4A–F). The in vivo findings were further corroborated by an in vitro exocytosis study where 4F was shown to increase the exocytosis rate of F-Aβ₄₂ to the abluminal side of the polarized BBB endothelial monolayer (Figure 4I). Importantly, more efficient trafficking at the BBB does not necessarily cause increased Aβ load in the brain parenchyma. Instead, previous studies have shown that oral administration of 4F peptide significantly reduced the Aβ load in APP/PS1 transgenic mice,²² supporting its therapeutic potential. However, the underlying mechanisms by which 4F reduced Aβ in the brain parenchyma in APP/PS1 mice remains unclear. The impact of 4F on other pathways that drive Aβ clearance from the brain parenchyma, including its effect on Aβ degrading enzymes such as insulin-degrading enzymes and neprilysin, needs further investigations. Taken together, these results indicated that 4F reduced Aβ exposure to the BBB endothelium by modulating its itinerary within the endothelial cells. Therefore, 4F appears to be a promising therapeutic molecule to treat cerebrovascular dysfunction in AD, including conditions like CAA.

Further, we investigated the underlying molecular mechanisms by which 4F reduced Aβ₄₂ accumulation in the BBB endothelium. We first examined the impact of Aβ₄₂ exposure on proteomic changes in the BBB endothelial cells using RPPA. The RPPA results suggested that p38 MAPK kinase was significantly activated upon Aβ₄₂ exposure. These RPPA results were further verified by Western blot studies, which demonstrated that Aβ42 increased p38 phosphorylation (Figure 5A–C). In addition to Aβ exposure, various pathological conditions have been shown to mediate BBB dysfunction via the p38 MAPK pathway. For instance, a previous study has demonstrated that Aβ₄₀ stimulated inflammatory responses in the BBB endothelium, and the inflammation was attenuated by inhibiting p38 MAPK.⁴² Moreover, p38 activation was involved in the transcellular transport of HIV-1 across the BBB⁴³ and is associated with BBB disruption in focal cerebral ischemia.⁴⁴

The current study has shown that 4F treatment significantly inhibits Aβ₄₂ induced p38 phosphorylation (Figure 5D,E). Previous studies have demonstrated that ApoA-I inhibits tolllike receptor 4 (TLR4) mediated p38 activation and downstream inflammation in adipocytes.⁴⁵ Further, the ApoA-I binding protein was found to prevent p38 activation in macrophages.⁴⁶ These findings suggest that 4F maintains a physiological impact similar to ApoA-I in suppressing p38 activation. This prompted us to hypothesize that 4F diminishes Aβ₄₂ exposure within the BBB endothelium via the p38 pathway. The results of this study supported the hypothesis that the p38 inhibitor significantly reduced F-Aβ₄₂ cellular uptake and luminal-to-abluminal permeability at the BBB endothelium (Figure 6). Takuma et al. previously demonstrated that activated p38 MAPK promoted the neuronal uptake of Aβ.⁴⁷ The current study suggested that inhibiting p38 could inhibit Aβ₄₂ uptake by endothelial cells and, subsequently, its transcytosis across the BBB into the brain parenchyma.

In conclusion, the current study has demonstrated that 4F reduces the accumulation of Aβ42, which engenders BBB dysfunction, in the BBB endothelium of APP/PS1 mice. In vitro studies conducted in BBB cell culture models have indicated that 4F accomplishes this by inhibiting p38 MAPK-mediated endocytosis of Aβ42 and enhancing Aβ42 exocytosis out of the BBB endothelium. These findings provide mechanistic insights into the impact of 4F treatment on cerebrovascular function as it relates to AD pathogenesis.

ABBREVIATIONS

Aβ

amyloid beta

AD

Alzheimer’s disease

ApoA-I

Apolipoprotein A-I

BBB

blood-brain barrier

CAA

cerebral amyloid angiopathy

hCMEC/D3

human cerebral microvascular endothelial cells

HDL

high-density lipoprotein

LRP1

low-density lipoprotein receptor-related protein-1

Pgp

P-glycoprotein

RAGE

advanced glycation end products

RPPA

reverse phase protein array

SPECT/CT

single positron emission computed tomography/computed tomography