Endothelial delivery of simvastatin by LRP1-targeted nanoparticles ameliorates pathogenesis of alzheimer’s disease in a mouse model

Abstract

Blood-brain barrier (BBB) dysfunction is an early pathological hallmark of Alzheimer’s disease (AD), occurring prior to amyloid-β (Aβ) accumulation. A key factor contributing to BBB damage in AD is the loss of endothelial expression of low-density lipoprotein receptor-related protein 1 (LRP1). Endothelial LRP1 is crucial for maintaining BBB integrity and facilitating the transcytosis of Aβ across the BBB for peripheral clearance. However, LRP1 is also expressed in other neural cell types, such as neurons, where it paradoxically promotes Aβ generation and tau propagation. These dual roles of LRP1 for different cell types present a challenge for developing effective AD therapy targeting LRP1. Simvastatin (SIM), an HMG-CoA reductase inhibitor, has been shown to induce compensatory upregulation of LRP1 expression. To harness this potential, we designed SIM-loaded Angiopep-2-anchored nanoparticles (S@A-NPs) that can be effectively internalized by endothelial cells. Our findings demonstrate that intravenous (IV) injection with S@A-NPs upregulates endothelial LRP1 expression level, repairs BBB damage, attenuates Aβ accumulation, mitigates neurodegeneration, and ultimately preserves cognitive function in APP/PS1 mice. These results highlight the potential of endothelial delivery of SIM via nanoparticles to attenuate AD pathogenesis. Our study proposes a novel therapeutic strategy for AD by leveraging nanoparticle-mediated drug delivery.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13195-025-01840-5.

Introduction

Alzheimer’s disease (AD) is one of the most common neurodegenerative disorders, clinically characterized by progressive memory impairment and executive dysfunction. Its major pathological features include the accumulation of amyloid-β (Aβ) peptides, with particular pathological ‌contributions‌ from ‌neurotoxic Aβ, and‌ intracellular tau-containing neurofibrillary tangles [1, 2]. The accumulation of Aβ plays a central and early role in triggering pathological progression of AD, contributing to vascular dysfunction [3], neuroinflammation [4], impaired synaptic plasticity [5], and reduced cerebral blood flow (CBF) [3]. Therefore, reducing Aβ accumulation has become a primary therapeutic strategy for AD. The concentration of Aβ in the brain is tightly regulated by several processes, including its production through the sequential cleavage of β-amyloid precursor protein (APP) by β-secretase (BACE1) and γ-secretase [6, 7], its transport across the blood-brain barrier (BBB) [8−11] and its clearance via glial phagocytosis [12− 14].

The BBB is a critical structure for maintaining brain homeostasis, which acts as a defense structure that tightly regulates the exchange of molecules between the blood and brain parenchyma ( [15]. By preventing the entry of harmful toxins and pathogens, the BBB ensures a stable and protected environment for optimal neural function. Damage to the BBB has been identified as one of the earliest pathological events in AD, occurring even before Aβ accumulation ( [16–18]. Thus, BBB impairment has been recognized as an early biomarker of aging-related cognitive dysfunction and AD progression. Remarkably, BBB dysfunction alone is sufficient to trigger neurodegeneration and cognitive impairment through mechanisms independent of Aβ ( [19]. Furthermore, the cerebral accumulation of Aβ is determined by the balance between its production and clearance rates, with the Aβ clearance capacity playing a more critical role ( [20].‌ Among these, the transport of Aβ from the brain across the BBB to the bloodstream is recognized as the most efficient mechanism for its rapid clearance [21, 22]. About one-third of Aβ is transported out of the brain via the BBB and subsequently cleared by the peripheral organs, and enhancement of peripheral Aβ clearance decreases brain Aβ accumulation and prevents AD pathogenesis [23, 24].

BBB dysfunction is associated with not only the loss of its structural components, such as vascular endothelial cells, astrocytes, pericytes, and junctional complexes, including tight junctions and adherens junctions, but also the aberrant expression of BBB transporters and/or receptors [19, 25]. Low-density lipoprotein receptor-related protein 1 (LRP1), a transmembrane endocytic and cell signaling protein, is a receptor of 100 diverse ligands, including apolipoprotein E (APOE) [26]. APOE is one of risk factors for AD [27]. LRP1 expression in brain endothelial cells is essential for maintaining BBB integrity and function [28]. Mechanistically, the absence of LRP1 in endothelial cells results in the autonomous activation of the cyclophilin A-matrix metalloproteinase-9 (MMP-9) pathway in the endothelium, causing the loss of tight junctions and structural impairment of the BBB [29]. Endothelial LRP1 together with the receptor for advanced glycosylation end-products (RAGE), bidirectionally mediates the transport of Aβ between the blood and the brain: the influx of Aβ from the blood into the brain across the BBB relies on RAGE expression in the luminal surface of endothelial cells [9], while the efflux of Aβ from the brain depends on LRP1 expression on the abluminal side of the endothelial cells [8, 30, 31]. The Aβ efflux involves sequential steps: Following binding to abluminal LRP1, Aβ is internalized by endothelial cells through either clathrin-coated vesicle formation (PECAM-1 assisted) or phagosome generation via APOE-LRP1 complexes. This internalized Aβ then undergoes transcytosis via vesicular transport from the abluminal to the luminal membrane for direct release into the circulation. Alternatively, LRP1 facilitates Aβ transfer to luminal efflux transporters (P-gp/ABCB1 or ABCC1) for exocytosis-mediated clearance [32, 33]. Notably, RAGE mediates the transport of peripheral Aβ into the brain, thereby reducing its accumulation in the periphery, and regulates the influx of Aβ from the circulation into the brain compartment [9, 34, 35]. Thus, the BBB has a dual role in Aβ transport.

Earlier studies revealed that the levels of endothelial LRP1 decrease with aging [30] and are almost absent from blood vessels in the brains of AD patients [36, 37], and AD animal models [38]. The loss of endothelial LRP1 is a causative factor contributing to progressive BBB impairment [29] in both normal aging [28] and AD brains [8], which is followed by neurodegeneration and cognitive deficits. Thus, in ‌the human AD brain, endothelial cells exhibit downregulation of endothelial LRP1 and the upregulation of RAGE [37], which disrupts the Aβ influx-efflux equilibrium and exacerbates Aβ accumulation in the AD brain‌. It has been reported that deletion of LRP1 in the endothelial cells impairs the efflux of Aβ, resulting in the accumulation of Aβ in the brain and defective spatial memory in 5 × FAD mice, an AD mouse model [8]. These findings highlight a protective function of endothelial LRP1 against neurodegeneration and Aβ accumulation in the brain, suggesting that enhancing or restoring the expression of LRP1 in the endothelial cells is a promising strategy for mitigating AD-related cognitive decline.

However, it is worth noting that LRP1 is also expressed in neurons, astrocytes, and microglia in addition to vascular compartments‌. Astrocytic LRP1 mediates the uptake and degradation of Aβ [39]. Microglial LRP1 exhibits a conflicting role, either attenuating [40, 41] or exacerbating [42] neuroinflammation, depending on distinct context. Neuronal LRP1 plays a role in driving both the production [43–45] and the clearance [46] of Aβ, with its function in driving production outweighing its role in clearance [33, 4, 3]. Deficiency of endogenous LRP1 leads to attenuated overall Aβ load due to ‌reduced Aβ production and clearance [33]. LRP1-mediated endocytosis of neuronal Aβ₄₂ is associated with increased Aβ aggregation and neuronal toxicity [4, 7]. A recent study observed that neuronal LRP1, an important regulator of tau protein endocytosis, is required for tau propagation. Reducing neuronal LRP1 suppresses tau spread in the brain [4, 8]. Thus, due to the paradoxical functions of LRP1 in distinct cell types, precise therapeutic strategies targeting LRP1 in a cell type-dependent manner are needed‌. Restoring the endothelial LRP1 levels while avoiding ‌upregulation of neuronal LRP1‌ may represent ‌a potential therapeutic strategy‌ for AD.

Simvastatin (SIM) is a clinically commonly used lipid-lowering drug and an inhibitor of hydroxymethylglutaryl-coenzyme A (HMG-CoA) reductase [49]. SIM through two established molecular pathways to upregulates LRP1 expression. First, as an HMG-CoA reductase inhibitor, SIM suppresses the cholesterol biosynthesis pathway, thereby reducing levels of key isoprenoids, such as arnesyl pyrophosphate (FPP) and geranylgeranyl pyrophosphate (GGPP) [50–52]. This reduction inactivates Rho GTPases, which activate PPARγ, a transcription factor that directly promotes LRP1 gene expression [53, 54]. Second, SIM concurrently inhibits sterol regulatory element-binding protein 2 (SREBP-2) activity, a transcriptional repressor of LRP1 [55–57], effectively derepressing LRP1 synthesis.

SIM has been shown to increase LRP1 expression in brain microvascular endothelial cells, thereby inhibiting the accumulation of neurotoxic Aβ caused by HIV-1 infection [58]. Oral gavage treatment with SIM elevates LRP1 expression in murine brain capillary cells isolated from 3 × Tg AD mice and facilitates the clearance of Aβ across the endothelial barrier [59]. Furthermore, oral delivery of SIM reverses cerebrovascular impairment and reduces CBF and the accumulation of soluble Aβ in AD mice. Utilizing SIM to enhance LRP1 expression represents a highly promising strategy for ameliorating AD pathology [60–62]. However, SIM administration fails to reduce Aβ plaque deposition and normalize cholinergic function and memory deficits at the advanced stage of pathology [63]. A recent study reported that subcutaneous (SC) administration of SIM impairs hippocampal synaptic plasticity, leading to cognitive decline in adult C57BL/6J mice [64]. These findings suggest that systemic administration of SIM has limited efficacy in AD therapy, even causing adverse effects, possibly due to the upregulation of LRP1 in multiple cell types, such as neurons and glia. Therefore, specific delivery of SIM to brain endothelial cells may be a promising therapeutic strategy for AD by selectively enhancing endothelial LRP1 expression level.

To achieve this purpose, we designed SIM-loaded Angiopep-2-anchored nanoparticles (S@A-NPs), which consist of a PLGA-PLL carrier sphere, a PEG linker material, an Angiopep-2 targeted ligand, and SIM. SIM was encapsulated in the PLGA-PLL carrier sphere, while Angiopep-2 was attached to the outer surface of the sphere via the PEG linker. Among them, the Angiopep-2 ligand binds LRP1 and facilitates LRP1-mediated endothelial internalization of S@A-NPs. Following intracellular release, SIM upregulates endothelial LRP1 expression, creating a positive feedback loop that enhances subsequent nanoparticle internalization (Fig. 1). Through this “self-promoting” manner, S@A-NPs display higher uptake efficiency by brain microvascular endothelial cells and a greater capability to penetrate the brain [65]. We analyzed the potential therapeutic function of S@A-NPs in AD using APP_swe/PS1_dE9 (APP/PS1) transgenic mice, which exhibit many key features of AD pathology, including BBB impairment [66], Aβ accumulation [67], neuroinflammation, synaptic loss, and cognitive deficits [68, 69]. These mice are widely used in AD preclinical studies. Upon intravenous (IV) injection, S@A-NPs were up taken by brain endothelial cells, where they upregulated LRP1 expression. Administration of S@A-NPs specifically upregulated endothelial LRP1 expression level, alleviated BBB damage, reduced Aβ burden, and ameliorated neuroinflammation, neurodegeneration, and cognitive deficits in APP/PS1 transgenic mice. Furthermore, we found that S@A-NPs suppressed the amyloidogenic cleavage of APP while enhancing the transport of Aβ across the BBB from the brain. These results indicate that S@A-NPs mitigate AD pathogenesis by increasing endothelial LRP1 levels. The present study highlights endothelial cell-specific SIM delivery as a novel approach for AD therapy.

Fig. 1.

Schematic description of LRP1-targeted S@A-NPs. S@A-NPs comprise a carrier sphere PLGA-PLL, a linker material PEG, a targeted ligand Angiopep-2, and SIM. SIM is encapsulated in a carrier sphere PLGA-PLL, and the outer side of the sphere is connected to Angiopep-2 by the linker material PEG. Among them, Angiopep-2 ligates LRP1 to drive cellular internalization and help efficient intracellular delivery of S@A-NPs into endothelial cells. Once released cellularly, SIM enhances LRP1 expression, which, in turn, accelerates the uptake of the nanoparticles by the endothelial cells. Through this “Self-promoting” manner, S@A-NPs display higher uptake efficacy by brain microvascular endothelial cells. The image was adapted as described [65]. The diagram created with BioRender.com and released under a Creative Commons Attribution-Non Commercial No Derivs 4.0 International license

Materials and methods

Materials

Poly (D, L-lactide-co-glycolide) (PLGA, Cat. 719900, acid-terminated, lactide: glycolide 50:50, MW 38,000–54,000) and poly (ε-carbobenzoxyl-L-lysine) (PLL, Cat. P4510, MW 500-4,000) were purchased from Sigma. α-Malemidyl-u-N-hydroxysuccinimidyl polyethyleneglycol (NHS-PEG-MAL, MW 5,000) and NHS-PEG-Cy3 (Cat. PG2-NSS3) were purchased from Nanocs. TCEP-HCl (Pierce™ 20490) was bought from Thermo Scientific. Polyvinyl alcohol (PVA-403) was obtained from Kuraray (Japan). Angiopep-2 with cysteine (Cys) on its C-terminal (TFFYGGSRGKRNNFKTEEYC) was synthesized by Nanjing Peptide Biotech Ltd (Nanjing, China). The hydrophobic C-terminus accommodates Cys’s hydrophobicity and preserves the hydrophilic N-terminal receptor-binding domain [70, 71]. This design enables stable thioether bond formation between the thiol group of the C-terminal Cys and the maleimide group in NHS-PEG-MAL, ensuring efficient nanoparticle conjugation without structural compromise. Thus, The C-terminal Cys can stabilize the structure of Angiopep-2 [70, 71] and assist in carrying nanoparticles [65, 72]. SIM was purchased from Innochem (Beijing, China).

Cell culture and animals

bEND.3 mouse cell line of BMECs was purchased from ATCC. Cells were grown in DMEM medium supplemented with 10% fetal bovine serum (FBS), 100 units/mL penicillin, and 100 µg/mL streptomycin in a 37 ℃ incubator containing 5% CO₂.

APP/PS1 double transgenic mice express mutant human amyloid precursor protein (APPswe) and presenilin 1 (PS1dE9) under the mouse prion protein promoter. The genetic background of APP/PS1 transgenic mice used for the present study is C57BL/6J. APP/PS1 transgenic mice (stock number 034829) and C57BL/6J mice (stock number 000664) were purchased from the Jackson laboratory and genotyped through tail clips and subsequent PCR analysis of genomic DNA. Female ICR mice of 20–25 g body weight were purchased from the Department of Experimental Animals, Soochow University. All experimental procedures were approved by the Ethics Committee of Soochow University and conformed to the Institutional Animal Care and Use Committee Guidelines of Soochow University and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals guidelines for the ethical treatment of animals. The approval numbers are 202302A0391 and SUDA20241121A07. Efforts were made to minimize the number of animals used. All mice were group-housed with 3–5 same-sex cage mates in standard mouse cages in a pathogen-free barrier facility under a 12-h light-dark cycle with lights on at 07:00 and a controlled temperature range of 22–25 ℃. Food and water were provided ad libitum.

Antibodies

Rabbit anti-LRP1 (ab92544, Abcam), Rat anti-CD31 (550274, BD), Mouse anti-NeuN (94403, CST), Rat anti-NeuN (ab279297, Abcam), Rabbit-anti-GFAP (3670T, CST), Goat-anti-IBA-1 (AB5076, Abcam), Goat anti-PDGFRβ (AF385, R&D systems), Mouse anti-Aβ-6E10 (803015, biolegend), Rabbit anti-synaptophysin (SYN) (ab32127, Abcam), Rabbit anti-ZO1 (ab276131, Abcam), Rabbit anti-Occludin (ab216327, Abcam), Rabbit anti-Claudin 5 (AF5216, AffinitY), Rabbit anti-RAGE (ab216329, Abcam), Mouse anti-APP (MAB348, Merke), Mouse anti-BACE1 (NBP2-37261, Novus), Mouse anti-GAPDH (FD0063-100, FD), β-actin(66009-1, Proteintech), Alexa Fluor 555 Donkey anti-Rabbit (A32794, Invitrogen), Alexa Fluor 488 Donkey anti Rabbit (A21206, Invitrogen), Alexa Fluor 555 Donkey anti Mouse (A32773, Invitrogen), Alexa Fluor 488 Donkey anti Mouse (A32766, Invitrogen), Alexa Fluor 488 Donkey anti Goat (A11055, Invitrogen), Alexa Fluor 488 Donkey anti-Rat (A21208, Invitrogen), and HRP AffiniPure Goat anti-Mouse IgG (H + L) (FD0142, FD), HRP AffiniPure Goat anti-Rabbit IgG (H + L) (FD0128, FD) were used.

Preparation of S@A-NPs

The starting material, PLGA-PLL, was synthesized via coupling using dicyclohexyl carbodiimide, following a previously reported method [73]. S@A-NPs were prepared using the emulsion solvent evaporation technique. In a typical synthesis, 100 mg PLGA-PLL dissolved in 2 mL ethyl ether containing 10 mg SIM was added dropwise into 4 mL 2.5% PVA under simultaneous vortex mixing and sonication to form an oil/water emulsion. This emulsion was then transferred into a beaker containing 0.3% PVA and stirred overnight. The resulting NPs suspension was first centrifuged at 1,000 rpm for 5 min to remove bigger particles and unencapsulated SIM aggregates. Subsequently, the supernatant was subjected to centrifugation at 30,000 rpm for 20 min to collect the NPs. The NP pellet was resuspended in PBS (pH 7.4) and reacted with 10 mg NHS-PEG-MAL or NHS-PEG-Cy3 for 1 h at room temperature to conjugate the PEG-MAL or PEG-Cy3 linker. The suspension was centrifuged to remove excess, unreacted PEG-MAL or PEG-Cy3, and the PEGylated NP pellet was collected. This pellet was resuspended in PBS (pH 7.4) and then reacted with Cys-modified Angiopep-2 in the presence of TCEP-HCl for 1 h at room temperature. The molar ratio of Angiopep-2 to PEG-MAL or PEG-Cy3 was maintained at 1:10 to achieve a 10% modification. Unreacted Angiopep-2 was removed by centrifugation. The final NPs were suspended in NS and lyophilized for storage and subsequent characterization. The surface density of PEG-Cy3 on the NPs was quantified by measuring the CY3⁺ fluorescence intensity. To analyze the distribution of CY3-conjugated S@A-NPs/A-NPs in the brain, five-month-old C57BL/6J mice were intravenously injected with CY3-conjugated A-NPs/S@A-NPs (100 mg NPs kg^− 1) twice at a frequency of once per 12 h. At 24 h after the final injection, mice were deeply anesthetized with 4% v/v isoflurane and transcardially perfused with 4% paraformaldehyde (PFA).

Biological distribution analysis of SIM

SIM accumulation and concentration analyses in the brain were conducted as previously described [65]. The SIM loading efficiency in S@A-NPs was measured as 8.6% [65]. To examine the intracranial kinetics of SIM, female ICR mice were intravenously injected with free SIM (8.6 mg kg^− 1) or S@A-NPs (8.6 mg SIM kg^− 1) dissolved in normal saline (NS) once [65]. At 2, 12, and 24 h post-injection, mice were perfused with PBS (pH 7.4), and their brains and other organs were collected and weighed for SIM analysis using high-performance liquid chromatography (HPLC). To assess the accumulative effects of multiple dosing, ICR mice received IV injection of free SIM or S@A-NPs seven times either daily or every two days. At 24 h after the final injection, the mice were perfused with PBS (pH 7.4), and their brains were collected for SIM analysis using HPLC. HPLC analysis was performed using a C₁₈ column with a mobile phase of methanol and 0.02 mol L^− 1 monopotassium phosphate in water (85:15, v/v) at a flow rate of 1.0 mL min^− 1. The column temperature was maintained at 23 ℃, and detection was conducted at a wavelength of 238 nm. The injection volume for each sample was 20 µL.

Drug treatment

Male APP/PS1 heterozygous mice and age-matched wild-type (WT) mice were used in the present in vivo study. The mice were randomly divided into four groups: wild-type mice administered with NS (WT + NS), wild-type mice administered with S@A-NPs (WT + S@A-NPs), APP/PS1 mice administered with NS (TG + NS), and APP/PS1 mice administered with S@A-NPs (TG + S@A-NPs). At 5.5 months of age, APP/PS1 mice or WT mice were intravenously injected with S@A-NPs (7.85 mg SIM kg^− 1) through tail veils at a frequency of once per two days for 2–3 months. For BBB functional analysis and Aβ clearance assay, 6-month-old APP/PS1 or WT mice were intravenously injected with S@A-NPs (7.85 mg SIM kg^− 1) through tail veils at a frequency of once per two days for 1 month. The mice injected with NS were taken as control.

Immunofluorescence staining and image analysis

The mice were deeply anesthetized with 4% v/v isoflurane and transcardially perfused with PBS. Brain tissues were collected and post-fixed in 4% PFA for 4 h at 4 ℃, and then dehydrated through a gradient of 15% and 30% sucrose. The brain tissue was then embedded in OCT embedding medium and sectioned into 20 μm thick slices using a Microtome cryostat (CM1950, Leica). For immunofluorescence staining, brain slices were washed with PBS and PBST (PBS with 0.3% or 0.5% TritonX-100), and subsequently blocked with 10% fetal bovine serum (FBS) diluted with PBST for 1 h at room temperature. Brain slices were incubated with primary antibodies overnight at 4 ℃. After three washes with PBS, the slices were incubated with the corresponding secondary antibody for 1 h at room temperature. The slides were mounted with DAPI-containing mounting medium (0100 − 20, Southern Biotech), and images were captured under the Zeiss LSM 900 confocal laser scanning microscope with a 20 × objective (NA 0.8, WD 0.55 mm) provided a broader view, while a 40 × objective (NA 1.3, WD 0.21 mm) provided more detail. Laser excitation wavelengths include 405 nm (DAPI channel); 488 nm (FITC channel); 561 nm (TRITC/CY3 channel); 640 nm (Alexa Fluor 647 channel). Exposure time varied from 300 to 600 milliseconds. 5–10 random fields of view per brain region (hippocampus and cortex) were captured.

All images were analyzed using the ImageJ software (NIH, Bethesda, MA, USA). The percentage of CY3⁺marker⁺ cells was calculated by dividing the number of CY3⁺ marker⁺ cells by the total number of marker⁺ cells in the brain. The mean fluorescence intensity (MFI) of LRP1 and BACE1 in endothelial cells, neurons, or CY3⁺ cells was quantified by manually selecting the fluorescence signals. For the area of Aβ plaques quantification, the total area occupied by Aβ plaques in the respective regions of interest was measured, and the number of Aβ plaques in these regions was counted. For neuroinflammation analysis, the total area of GFAP and IBA-1 staining within the corresponding regions of interest was quantified. To quantify synaptic loss, the SYN⁺ MFI in the mossy fiber layer of the CA3 was measured. The areas of ZO1⁺, Claudin5⁺, and Occludin⁺ staining within brain endothelial profiles were determined using the ImageJ area measurement tool. The percentage of ZO1⁺, Claudin5⁺, and Occludin⁺ areas was calculated by dividing ZO1⁺, Claudin5⁺, and Occludin⁺ fluorescence signal areas by the total endothelial areas.

Western blotting

Protein lysates from the cortex and hippocampus were extracted in brain lysis buffer (10 mM Tris, pH 9.0, 150 mM NaCl, 0.5% TritonX-100, 1% sodium deoxycholate, 0.5% SDS, 2 mM EDTA) containing a protease and phosphatase inhibitor cocktail (B14002; B15001, Bimake). Cultured cells were lysed with RIPA lysis buffer (P0013D, Beyotime) containing a protease and phosphatase inhibitor cocktail for 30 min on ice. Samples were centrifuged at 14,000 × g for 15 min at 4 ℃, and supernatants were collected. Lysates were subjected to SDS-PAGE and transferred to a polyvinylidene fluoride membrane (IPVH00010, Merck Millipore). The membranes were blocked with 5% non-fat dry milk (BBI) in Tris-buffered saline (TBST, pH 7.5) (20 mM Tris, 150 mM sodium chloride, 0.1% Tween-20) at RT for 1 h, and then incubated with primary antibodies diluted in TBST overnight at 4 ℃. The membranes were washed with TBST and incubated with HRP-conjugated secondary antibodies (FD) at RT for 1 h. The protein signals were detected by ECL (Merck Millipore, WBKLS). Images were captured with an E-Gel Imager (E-Blot, China). The intensity of each protein signal on the membrane was measured as a grayscale value using ImageJ software (NIH, Bethesda, MA, USA), corrected for background intensity, and normalized to the internal control.

Effect of SIM on LRP1 expression in bEND.3 cells

To assess the in vitro expression of LRP1 in bEND.3 cells by western blot, 1.4 × 10⁵ bEND.3 cells were plated in 6-well plates for overnight. Subsequently, cells were treated with SIM at concentrations of 100 nM, 500 nM, and 1 µM for 36 h. In addition, cells were treated with either free SIM, A-NPs, or S@A-NPs at SIM concentration of 1 µM for 36 h. Cell lysates were mixed with the appropriate amount of SDS sample loading buffer (5×), boiled at 98 °C, centrifuged, and then separated by SDS-PAGE.

Enzyme-linked immunosorbent assay (ELISA)

To quantify Aβ, tissues from the cortex or hippocampus were homogenized in brain lysis buffer. Total protein levels were determined using a BCA assay kit (23227, ThermoFisher). Then, the soluble supernatant fractions were collected after centrifugation at 35,000 × g for 30 min at 4 ℃. The insoluble fraction in the precipitate was dissolved in 70% formic acid on ice for 2 h, and the supernatant was collected after centrifuged at 12,000 × g for 1 h at 4 ℃. The supernatant was neutralized with 1 M TP buffer (1 M Tris, 0.5 M Na₂HPO₄) to a neutral pH. Blood plasma was collected by centrifugation at 1,000 × g for 10 min at 4 ℃. The amounts of Aβ₄₂ and Aβ₄₀ were quantified using human Aβ₄₂ and Aβ₄₀ ELISA kits (KHB3441; KHB3482, Invitrogen) following the manufacturer’s instructions. The absorbance was read at 450 nm using a TECAN infinite M200 Pro multifunctional microplate reader.

Quantitative realtime PCR

Total RNA was extracted from brain tissue samples using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. Complementary DNA (cDNA) was synthesized from 1.0 µg total RNA using a reverse transcription kit (Vazyme Biotech Company Limited, China). SYBR qPCR Master Mix (Vazyme Biotech Company Limited, China) was used for quantitative PCR in a final volume of 10 µL on a 7500 Real-Time PCR System (ABI) under the following conditions: 95 ℃ for 30 s, followed by 40 cycles of amplification (95 ℃ for 10 s and 60 ℃ for 30 s). The average threshold cycle (Ct) of fluorescence units was applied to analyze the mRNA levels, which were normalized to GAPDH. qPCR primers are as follows: TNF-α (forward: 5’CTGAGGTCAATCTGCCCAAGTAC; reverse: 5’CTTCACAGAGCAATGACTCCAAAG), IL-β (forward: 5’CTAAAGTATGGGCTGGACTG; reverse: 5’AGCTTCAATGAAAGACCTCA), IL-6 (forward: 5’GAGGATACCACTCCCAACAGACC; reverse: 5’AAGTGCATCATCGTTGTTCATACA), and GAPDH (Forward: 5’GAAGGTCGGTGTGAACGGAT; reverse: 5’AATCTCCACTTTGCCACTGC).

Preparation of Aβ oligomers

Aβ oligomers were prepared according to a previously established method [74]. Human Aβ₄₂ (Shanghai RovoBiotech, Cat. No. 22JW06579) was dissolved in hexafluoro isopropanol (HFIP) (H811026, Macklin) to generate monomeric Aβ. The HFIP-treated monomer solution was lyophilized and reconstituted in DMSO in a stock concentration of 5 mM. Aβ oligomerization was induced by diluting the monomer stock to 200 µM in ice-cold PBS, followed by overnight incubation at 4 ℃. The solution was centrifuged at 14,000 × g for 10 min at 4 ℃, and the supernatant containing oligomeric Aβ₄₂ was collected. Oligomer quality was confirmed by 13% Tris-SDS PAGE and western blotting with an anti-Aβ antibody. For fluorescent labeling, 100 µL of oligomeric Aβ (200 µM) was incubated with 4 µL Super Fluor 555 NHS ester (Bioesn) in PBS containing 0.1 M sodium bicarbonate buffer (pH 8.3) for 30 min at room temperature with gentle mixing. Unconjugated dye was removed using an Amicon Ultra 3k ultrafiltration cube (UFC5003, Millipore).

Aβ clearance assay

Mice were deeply anesthetized using 2% v/v isoflurane and mounted on a stereotactic frame (71000-M, RWD). Holes were drilled at the injection site after the skull was exposed. Super Fluor 555 NHS ester-labelled Aβ oligomers were loaded into a 10 µL syringe (Gaoge, China), and 1 µL was injected with a rate of 200 nL min^− 1 into the targeted regions (dentate gyrus (DG): Anteroposterior (AP) = − 2.0 mm, mediolateral (ML) = -1.3 mm, dorsoventral (DV) = -2.2 mm) [75]. The needle was left in place for an additional 5 min before being slowly withdrawn. To minimize potential brain tissue damage, a unilateral injection (left hemisphere) was performed in this study rather than a bilateral injection. After 48 h, mice were deeply anesthetized using 4% v/v isoflurane and transcardially perfused with 4% PFA. The entire injection site was sectioned and imaged as described above, and consecutive sections spaced 20 μm apart. The area occupied by fluorescence signals in all brain slices was manually delineated and measured using Image J software. The sum of the fluorescent signal area across all brain sections represents the area of Aβ residue that could not be cleared.

BBB leakiness assay

Mice were injected with 0.1 mL of 10-kDa and 40-kDa FITC-conjugated dextran (40 mg mL^− 1, Sigma) via tail vein. After 15 min, the mice were deeply anesthetized using 4% v/v isoflurane and transcardially perfused with 20 mL of ice-cold PBS. The brains were collected and frozen in liquid nitrogen for at least 5 min, and subsequently stored at -80 ℃ refrigerator until sectioning. FITC-dextran extravasation in brain cryosections (10 μm) was imaged using Olympus VS 200. The fluorescence area of FITC-dextran in the cortex and hippocampus coronal sections was measured using Image J software. The extent of FITC-dextran extravasation, indicating BBB disruption, was calculated based on the fluorescence area. More than three images were acquired within the region of interest (ROI) for each mouse tissue section.

Behavioral tests

The experimenter was blinded to the groups during testing. All behavioral tests were performed in a dimly lit room without noise interference. Tests were recorded using Any-Maze 7.2 (Stoelting, CO, USA). The experimental apparatus was cleaned with 30% ethanol between trials to remove odors of the tested subjects.

Open Field Test: Experiments were conducted in a 40 × 40 × 40 cm cubic chamber. Each mouse was gently placed at the center of the arena, and spontaneous locomotor parameters, including total distance traversed and average speed, were recorded over a 10-minute period.

New object recognition (NOR): Mice were placed in a 40 × 40 × 40 cm box equipped with a camera on top for habituation trial over three consecutive days. Following habituation, two identical objects, A₁ and A_2, were placed diagonally in the box, 8 cm away from both walls. Mice were allowed to explore for 10 min, and the time spent sniffing or chewing the object was recorded as exploratory activity. After 90 min, one object (A₁) was replaced with a novel object (B). Mice were then allowed to explore for another 10 min. The object preference index was defined as the ratio of time spent on object A₁ to the total time spent on objects A₁ and A₂ in the first round of testing. The recognition index was defined as the ratio of time spent on object B to the total time spent on both object B and A₂ in the second round of testing. Object Location Test (OLT) was performed in the same box used for the NOR test. Mice were allowed to freely explore the apparatus, which contained intramural cues on the walls for 5 min to enable habituation. After a 15–30 min interval, the mice were returned to the box and exposed to two identical objects, A₁ and A_2, placed equidistant from the side walls. Mice were allowed to explore for 10 min freely. After 24 h, object A₂ was removed to a different location within the box, and the mice were allowed to freely explore for 5 min. The object preference index and the recognition index were defined as the ratio of time spent on object A₂ to the total time spent on object A₁ and A₂ during two respective rounds of testing. Mice with normal memory typically prefer the new object or the object in the novel location due to their natural curiosity, resulting in a higher recognition index. In contrast, mice with cognitive deficits tend to spend similar duration of time exploring both the novel and old locations, resulting in a reduced recognition index. Thus, the object preference index value approaching 50% indicates the absence of inherent preference for either object in mice, while the recognition index ≥ 60% demonstrates intact healthy recognition memory.

Statistical analysis

All data analyses were performed using GraphPad Prism 8.0 to produce graph values. The graph values are presented as the mean ± standard error of the mean (SEM). All statistical analysis was performed using SPSS 26.0. The number of mice included in each experiment was based on standards established in the literature rather than being predetermined by statistical methods. The experimental design for each assay is described in the subsection below. Tests for normality and equal variances were used to determine the appropriate statistical test to apply. Data were analyzed using Student’s t-test (a comparison of the difference between the two groups) and one-way ANOVA (a comparison of the differences between multiple groups) followed by LSD or Dunnett T3 post hoc tests. The P-value indicates significance, and p < 0.05 was considered for the significance level for all analyses; * p < 0.05; ** p < 0.01, *** p < 0.001.

Results

Intravenously delivered S@A-NPs penetrate the brain and distribute in the endothelial cells

LRP1 is expressed on the luminal and abluminal membrane of brain microvascular endothelial cells [8, 30, 31]. The interaction between LRP1 and Angiopep-2-functionalized nanocarriers promotes endothelial cell uptake of drug cargo, facilitating its subsequent internalization for brain delivery [65, 76, 77]. S@A-NPs exhibited a spherical morphology with diameters ranging from 53.0 ~ 100.2 nm under transmission electron microscopy (TEM) observations [65]. Dynamic light scattering (DLS) measurements showed that the freshly prepared S@A-NPs had an average hydrodynamic diameter of 104.0 nm and a ζ-potential of -3.6 mV [65]. In vitro release kinetics assessed via HPLC revealed that S@A-NPs released 66.24% SIM within 24 h, achieving near-complete release (~ 100%) over 5 days in a pH-independent manner [65]. In this study, we further confirmed the brain delivery efficacy of S@A-NPs by comparing the biodistribution between free SIM and S@A-NPs in the brains of adult mice. Mice treated with S@A-NPs via a single IV injection exhibited a higher brain SIM concentration than those receiving free SIM. The drug concentration in the brain reached the maximum at 2 h after IV injection and gradually decreased over time (Fig. 2a-b, Supplementary Fig. 1a), suggesting a typical clearance dynamic of S@A-NPs. We next examined the accumulation of SIM in the brains of the mice that received seven consecutive IV injections of S@A-NPs. The results showed a significant intracranial increase in SIM accumulation in S@A-NPs-treated mice compared to those injected with free SIM at all investigated post-injection time points. Notably, SIM concentrations in the brains of S@A-NP-administered mice were higher on day 7 post-injection than on day 1 post-injection, indicating a gradual accumulation of SIM in the brain following the 7-day consecutive injection regimen (Fig. 2c-d, Supplementary Fig. 1b). This intracranial increase in SIM accumulation can be attributed to the autocatalytic effect of LRP1 up-regulation mediated by S@A-NPs [65]. Importantly, compared with free SIM, the S@A-NP delivery system did not alter peripheral distribution pattern of SIM with the liver, spleen, kidney, and lung as the major distribution tissues (Supplementary Fig. 1c-f). Thus, continuous injections of S@A-NPs at intervals of less than or equal to 2 days could maintain the steady intracranial SIM concentration, supporting its potential use for long-term AD therapy.

Fig. 2.

Distribution of S@A-NPs in the brains of mice. (a) Schematic description of the timeline of experimental procedures. Adult mice were intravenously injected with free SIM or S@A-NPs for one time. (b) The brain uptake kinetics of SIM (concentration in µg g^− 1 brain tissue) were measured at various time points following a single injection. (c) Schematic description of the timeline of experimental procedures. Adult mice were intravenously injected with free SIM or S@A-NPs for seven consecutive times at a frequency once per one or two days. (d) The brain uptake kinetics of SIM (concentration in µg g^− 1 brain tissue) were measured at various time points following multiple injection. (e) Schematic description of the timeline of experimental procedures. Adult mice were intravenously injected with CY3-conjugated S@A-NPs for two consecutive times at a frequency once per 12 h. (f) Distribution of CY3⁺ signals in the brain of CY3-conjugated S@A-NPs-administered mice. (g) The coronal sections of hippocampus of CY3-conjugated S@A-NPs-administered mice were immunostained for CD31, NeuN, GFAP or IBA-1 and DAPI. (h) The coronal brain sections were immunostained for CD31, PDGFRβ, and DAPI in the hippocampus. Smooth endothelial cells and pericytes with nuclei were indicated by arrowheads and arrows, respectively. (i) Percentage of CY3⁺ marker⁺ cells among total marker⁺ cells in the brains. Data are presented as mean ± SEM. n = 5 mice (b, d), n = 7–13 slices from 3 mice/group (i). *p < 0.05. **p < 0.01. Student t-test. Scale bars: 1 mm (f), 20 μm (g), 10 μm (h)

We further analyzed the distribution of S@A-NPs in the brain parenchyma of the adult mice following IV administration of CY3-conjugated S@A-NPs (Fig. 2e-f). Immunofluorescence staining for markers of distinct cell types was performed 24 h post-injection. The results showed that the CY3⁺ cells were predominantly endothelial cells (CD31⁺, 97.98%) (Fig. 2g, i). Only a small proportion of neurons (NeuN⁺, 2.02%) displayed CY3⁺ signals, which localized primarily in the perirhinal cortex (Supplementary Fig. 2a). The density of CY3⁺ nanoparticles in perirhinal cortical neurons was significantly lower than that in endothelial cells, indicating that while some perirhinal cortical neurons internalized the nanoparticles, the quantity of nanoparticles they absorbed was minimal (Supplementary Fig. 2a-b). Few CY3⁺ neurons were observed in other brain regions. Furthermore, we did not detect CY3⁺ signals in astrocytes (GFAP⁺), microglia (IBA-1⁺), or pericytes (PDGFRβ⁺) (Fig. 2g-i). These results demonstrate that IV administered S@A-NPs were successfully internalized by endothelial cells.

Administration of S@A-NPs enhances endothelial LRP1 levels in the brain

The critical role of brain endothelial LRP1 in AD pathogenesis makes it an attractive therapeutic target for AD therapy. We previously reported that S@A-NPs could upregulate the expression of LRP1 in brain microvascular endothelial cells, thereby enhancing BBB targeting and facilitating the internalization of LRP1-targeted nanoparticles through a “self-promoting” mechanism in mouse model with brain metastases [65]. To confirm this, we treated bEND.3 endothelial cells with free SIM and observed dose-dependent LRP1 upregulation. At the SIM concentrations of 1 µM (0.42 µg/mL), LRP1 expression significantly increased compared to untreated controls in bEND.3 cells (Supplementary Fig. 3a). The expression of LRP1 in bEND.3 cells treated with 1 µM S@A-NPs was comparable to that in cells treated with free SIM (Supplementary Fig. 3b). Notably, the blank A-NPs showed no effect on endothelial LRP1 expression level in vitro or in vivo (Supplementary Fig. 3b-d) [65], further approving a capability of S@A-NPs via releasing SIM in upregulating LRP1 expression as described previously [65, 78].

Consistent with the previous finding that S@A-NPs upregulate LRP1 expression by releasing SIM [65, 78], we further confirmed LRP1 expression in the brains of APP/PS1 mice treated with S@A-NPs. While no preclinical model fully recapitulates the complexity of human AD pathogenesis, the APP/PS1 transgenic mouse remains a scientifically validated and widely adopted system for studying early AD mechanisms. These mice exhibit well-characterized temporal pathology: amyloid plaques emerge in the cortex at about 4 months of age and ‌in‌ the hippocampus at about 6 months of age, increasing in density with aging‌ [29, 79];‌ ‌inflammatory responses increase‌ at 3 months ‌of age‌, ‌and significant memory deficits appear‌ at 5 months ‌of age [80]‌. Crucially, BBB dysfunction, a key focus of this study, manifests at about 4 months of age [29], preceding significant Aβ deposition. This model provides a rigorous platform for investigating the BBB-Aβ feedback loop central to AD pathological progression: BBB impairment accelerates Aβ accumulation, which further damages cerebrovascular integrity. We thus examined whether S@A-NPs could specifically target this early pathological cascade. Mice were intravenously injected with S@A-NPs for three months, starting at 5.5 months of age (Fig. 3a). This treatment regimen targets the critical period of exacerbating pathology to evaluate the efficacy of earlier and sustained intervention. Western blot analysis showed elevated LRP1 levels in the cortex and hippocampus of S@A-NPs-treated APP/PS1 transgenic mice compared to saline-treated controls (Fig. 3b-c). In contrast, there was no change in the expression of RAGE (Supplementary Fig. 4), another key transporter mediating Aβ influx across the BBB. As described previously, endothelial LRP1 expression is significantly reduced in the brains of AD animal models [38]. Paradoxically, upregulation of LRP1 in neurons facilitates the propagation of pathological tau [48]. Immunofluorescence analysis showed a loss of endothelial LRP1 and an upregulated neuronal LRP1 in the hippocampi of APP/PS1 mice compared to WT mice, which was attenuated by S@A-NPs treatment (Fig. 3d-f). Given IV delivered S@A-NPs are mainly distributed in endothelial cells rather than in hippocampal neurons, the observed reduction in neuronal LRP1 may result from crosstalk between endothelial cells and neurons [81 − 83]. Thus, these results indicate that IV delivered S@A-NPs distribute in brain endothelial cells and effectively enhance endothelial LRP1 expression level, thereby highlighting their potential therapeutic impact in AD.

Fig. 3.

S@A-NPs treatment enhances endothelial LRP1 expression level. (a) Schematic description of experimental timeline. APP/PS1 mice were intravenously injected with S@A-NPs at 5.5 months of age and subjected to behavioral tests and analysis of AD-related pathology at 7.5 and 8.5 months of age, respectively. (b, c) WB analysis of LRP1 levels in the cortex and hippocampus of APP/PS1 mice treated with either normal saline (Ctrl) or S@A-NPs. GAPDH was detected as a loading control. Relative levels of LRP1 were quantified. (d) The coronal hippocampal sections were immunostained for LRP1, DAPI and CD31or NeuN. (e-f) The LRP1⁺ MFI of in hippocampal endothelial cells (e) or neurons (f). Data are presented as mean ± SEM. n = 6–7 mice/group (c), n = 3 mice/group (e, f), * p < 0.05; ** p < 0.01; *** p < 0.001. One-way ANOVA (c). Student t-test (e, f). Scale bars: 20 μm (d)

Administration of S@A-NPs repairs the damaged BBB in APP/PS1 transgenic mice

Impaired BBB function is an earlier pathological event in AD, occurring even before the accumulation of Aβ. One mechanism underlying BBB dysfunction in AD is the loss of endothelial LRP1, which leads to a substantial reduction in tight junction proteins, including zonula occludens 1 (ZO-1), claudin-5, and occludin. Notably, previous studies have shown that endothelial delivery of LRP1 via AAVs can restore the tight junction proteins and improve the BBB integrity in mice with endothelial LRP1 deficiency [28, 29]. Given the potent capability of S@A-NPs to upregulate endothelial LRP1, we investigated whether administration of S@A-NPs could benefit BBB integrity in APP/PS1 mice. Consistent with the loss of endothelial LRP1 in AD brains, APP/PS1 mice displayed a significant reduction in the tight junction protein levels (ZO-1, claudin-5, and occludin) compared to WT mice. However, these reductions were partially rescued by S@A-NPs (Fig. 4a-f). To further evaluate BBB integrity, we injected 10 kDa and 40 kDa FITC-conjugated dextran tracers into the tail veins of APP/PS1 mice and quantified their densities in the brain parenchyma. The results demonstrated impaired BBB integrity in APP/PS1 mice, as evidenced by increased tracer density in the brain parenchyma compared to WT mice. Importantly, administration of S@A-NPs partially rescued the BBB dysfunction (Fig. 4g-i). These results suggest that specifically increasing endothelial LPR1 expression level alleviates BBB impairment in APP/PS1 mice by preventing the loss of the tight junction proteins, thereby enhancing BBB integrity.

Fig. 4.

S@A-NPs treatment ameliorates impairment in BBB of APP/PS1 transgenic mice. (a-f) Coronal hippocampal sections were immunostained for CD31 and ZO1 (a), Claudin 5 (b) or Occludin (c), and DAPI. Percentages of ZO1⁺ (d), Claudin 5⁺ (e), and Occludin⁺ (f) area occupied in the total CD31⁺ area. (g) Representative images in the cortex and hippocampus coronal sections assessed BBB leakage by 10 KDa and 40 KDa FITC-dextran. (h, i) Percentage of 10 KDa (h) or 40 KDa (i) FITC-dextran⁺ areas in the cortex and hippocampus coronal sections. Data are presented as mean ± SEM. n = 13–15 slices from 3 mice/group (d, e); n = 14–17 slices from 3 mice/group (f); n = 3 mice/group (h, i); * p < 0.05; ** p < 0.01; *** p < 0.001. One-way ANOVA (d-f, h, i). Scale bars: 20 μm (a, b, c); 500 and 100 μm in the images with lower and higher magnification, respectively (g)

Administration of S@A-NPs ameliorates Aβ accumulation in APP/PS1 transgenic mice

In addition to preserving BBB integrity, endothelial LRP1 plays a critical role in regulating Aβ clearance and accumulation [8]. To evaluate whether administration of S@A-NPs could alleviate AD pathology, we analyzed Aβ accumulation in APP/PS1 mice. These mice begin to exhibit Aβ plaques as early as 3 months of age, followed by neurodegeneration and cognitive deficits by 7 months of age [67]. Starting at 5.5 months of age, when Aβ accumulation, neuroinflammation, and neurodegeneration are already present, APP/PS1 mice were intravenously injected with S@A-NPs for 3 months at a frequency of once every two weeks (Fig. 3a). Immunofluorescence staining using the 6E10 antibody revealed that the density of Aβ plaques in the hippocampus and cortex, measured by both the number and covered area of Aβ plaques, was significantly reduced in S@A-NPs-treated mice compared to saline-treated mice (Fig. 5a-c). ELISA analysis showed that the levels of soluble Aβ₄₂ and Aβ₄₀ in the cortex of APP/PS mice were decreased following administration of S@A-NPs (Fig. 5d, e). In the hippocampus, S@A-NPs treatment reduced soluble Aβ₄₀ levels but had no significant effect on soluble Aβ₄₂ levels (Fig. 5f, g). This discrepancy could be attributed to the differential distribution of S@A-NPs across different brain regions. In contrast to the reduction of soluble Aβ, the insoluble Aβ₄₂ and Aβ₄₀ levels in the cortex and hippocampus were not decreased by S@A-NPs treatment (Fig. 5h-k). Interestingly, administration of S@A-NPs even increased levels of insoluble Aβ40 in the hippocampus (Fig. 5k). LRP1 exhibits a significantly higher affinity for Aβ40 compared to Aβ42, with reported transport efficiencies exceeding twofold [84, 85]. The increased insoluble Aβ40 in the hippocampus may represent a complementary response to the therapeutic enhancement of soluble Aβ40 clearance. One plausible explanation is that accelerated efflux of soluble Aβ40 could transiently alter the equilibrium between soluble and insoluble Aβ pools or influence aggregation kinetics, potentially leading to a measurable shift toward the insoluble fraction in this specific region. However, the precise mechanism underlying this specific increase in hippocampal insoluble Aβ40 requires further investigation. These findings indicate that specifically increasing endothelial LPR1 expression level effectively attenuates Aβ accumulation, with a more pronounced impact on reducing soluble Aβ levels than insoluble Aβ. This highlights the potential of S@A-NPs as a therapeutic strategy for targeting the soluble forms of Aβ, which are considered to play a more direct role in AD pathogenesis.

Fig. 5.

S@A-NPs treatment attenuates Aβ accumulation in the brains of APP/PS1 transgenic mice. (a-c) Coronal brain sections were immunostained for Aβ and DAPI. Representative images in the cortex and hippocampus were shown. The number and covered area of Aβ plaques in the cortex and hippocampus coronal sections (b, c). (d-k) ELISA analysis of the soluble (d-g) and insoluble (h-k) Aβ₄₂ and Aβ₄₀ levels in the cortex and hippocampus. Data are presented as mean ± SEM. n = 17–19 slices from 3 mice/group for Aβ plaques analysis; n = 3–7 mice/group for ELISA analysis. * p < 0.05; ** p < 0.01; *** p < 0.001. Student t-test (b, c). One-way ANOVA (d-k). Scale bars: 500 μm and 50 μm, respectively, in the images with lower and higher magnification

Administration of S@A-NPs enhances Aβ clearance while suppressing Aβ production

Aβ accumulation in AD is driven by both increased production and impaired clearance of Aβ. Given the critical role of endothelial LRP1 in facilitating Aβ transcytosis out of the brain [8], we investigated whether administration of S@A-NPs could modulate this process. To assess Aβ clearance, we prepared Aβ oligomers and injected SF555-conjugated Aβ oligomers into the hippocampal DG of 6-month-old APP/PS1 mice (Supplementary Fig. 5). The mice were IV administered with either S@A-NPs or NS for one month, and the levels of SF555-conjugated Aβ in the brains were analyzed 48 h post-injection (Fig. 6a). Results showed that administration of S@A-NPs significantly reduced the levels of SF555-conjugated Aβ remnants in the brains of APP/PS1 mice (Fig. 6b, c). Correspondingly, plasma levels of Aβ₄₂ and Aβ₄₀ were increased in S@A-NPs-treated mice, indicating an enhanced Aβ clearance via transcytosis into the bloodstream (Fig. 6d, e). These results indicate that specifically increasing endothelial LPR1 expression level promotes Aβ clearance by accelerating its transcytosis across the BBB into the systemic circulation.

Fig. 6.

S@A-NPs treatment suppresses amyloidogenic cleavage of APP, while enhancing transcytosis of A β out of the brains of APP/PS1 transgenic mice. (a) Schematic description of experimental timeline. APP/PS1 mice were intravenously injected with S@A-NPs at 6 months of age and subjected to analysis of Aβ clearance at 7 months of age. SF555-conjugated Aβ oligomers were stereotactically injected into the hippocampal DG. (b, c) SF555-conjugated Aβ remnants in the brains were shown 48 h after injection. Area of SF555-conjugated Aβ oligomers occupied in the hippocampal DG (c). (d, e) ELISA analysis of Aβ₄₂ (d) and Aβ₄₀ (e) levels in the plasma. (f) WB analysis of levels of full-length APP, BACE1 and α/β-CTF in the cortex and hippocampus. GAPDH was detected as a loading control. Relative levels of full-length APP (f, g), BACE1 (f, h), β-CTF (f, i), and α-CTF (f, j). Data are presented as mean ± SEM. n = 3 mice/group (c); n = 7–9 mice/group (d); n = 6–11 mice/group (e); n = 3–8 mice/group (g-j). * p < 0.05; ** p < 0.01; *** p < 0.001. One-way ANOVA (c-e, g-j). Scale bars: 100 μm (b)

Aβ is generated by the sequential cleavage of APP by β-secretase (β-APP-cleaving enzyme-1, BACE1) and γ-secretase, a process referred to as the amyloidogenic pathway. In contrast, the cleavage of APP by α-secretase and γ-secretase, known as the non-amyloidogenic pathway, prevents Aβ generation since the cleavage site of APP by α-secretase is within the Aβ region [86]. To examine whether S@A-NPs treatment affects these two pathways, we analyzed the levels of key proteins in these pathways by western blotting. Compared to control APP/PS1 mice, S@A-NPs-administered APP/PS1 exhibited decreased levels of APP and β-CTFs (APP fragments generated by BACE1 cleavage), in both the cortex and the hippocampus, while reduced BACE1 levels in the cortex but not in the hippocampus (Fig. 6f-i). The decreased levels of α-CTF were observed in the cortex, but not in the hippocampus, likely due to the reduction of full-length APP (Fig. 6f, j). In contrast, S@A-NPs treatment failed to alter the levels of the above proteins, except for BACE1, in the brains of WT mice (Fig. 6f-j). BACE1 is significantly upregulated in brain endothelial cells and neurons of AD mouse models, leading to excessive production of neurotoxic Aβ peptides [87, 88]. Consistently, we observed APP/PS1 mice exhibit increased BACE1 expression in both brain endothelial cells and neurons, compared to WT mice. S@A-NPs treatment decreases BACE1 levels in both endothelial cells and neurons within the cortex of APP/PS1 mice (Supplementary Fig. 6), contrasting with the primarily endothelial distribution of the nanocarriers themselves (Fig. 2g-i). The downregulation of endothelial BACE1 is likely attributable to accelerated BACE1 turnover mediated by selective S@A-NP-induced upregulation of endothelial LRP1, which is known to inhibit BACE1-mediated APP cleavage and promote BACE1 degradation [89, 90]. While neuronal LRP1 expression is decreased (Fig. 3d, f), neuronal BACE1 levels were nonetheless reduced by S@A-NPs treatment (Supplementary Fig. 6). We hypothesize that this neuronal BACE1 downregulation may result from the attenuation of cellular stresses, such as Aβ accumulation and neurodegeneration, induced by the treatment, as BACE1 expression is responsive to various stressors, including Aβ accumulation [67, 91], hypoxia, and neuronal injury [92]. These findings suggest that the effects of S@A-NPs on APP processing are context-dependent, occurring specifically under pathological conditions.

Administration of S@A-NPs attenuates neuroinflammation and neurodegeneration in APP/PS1 transgenic mice

Excessive neuroinflammation is an essential contributor to the pathological mechanisms of Alzheimer’s disease. Astrocytes and microglia, the primary immune cells in the central nervous system, become activated in response to Aβ plaques. Upon activation, these cells proliferate and accumulate around Aβ plaques, further contributing to neuroinflammation [93, 94]. To assess whether S@A-NPs treatment alleviates neuroinflammation in APP/PS1 mice, we examined astrocyte and microglia activation in the cortex and hippocampus. The results showed that the density of astrocytes and microglia, as reflected by the area covered by GFAP⁺ (a marker for astrocytes, Fig. 7a, c) and IBA-1⁺ (a marker for microglia, Fig. 7b, d) cells, decreased in both the cortex and hippocampus of S@A-NPs-treated APP/PS1 transgenic mice, compared to control APP/PS1 transgenic mice. Activated microglia and reactive astrocytes play a crucial role in neuroinflammation by secreting inflammatory factors [95]. The transcript levels of TNF-α, IL-1β, IL-6, and iNOS in the brains of S@A-NPs-treated APP/PS1 transgenic mice also decreased significantly, compared to control APP/PS1 transgenic mice (Fig. 7e). These results indicate that S@A-NP-mediated restoration of BBB integrity via endothelial LRP1 upregulation suppresses neuroinflammation by reducing activation of astrocytes and microglia and secretion of inflammatory factors.

Fig. 7.

S@A-NPs treatment attenuates neuroinflammation and neurodegeneration in the brains of APP/PS1 transgenic mice. The coronal sections of the cortex and hippocampus were immunostained for GFAP (a), IBA-1(b), synaptophysin (f, SYN), and DAPI. The percentage of the covered area of astrocytes (c) or microglia (d) occupied in the total area in the cortex and hippocampus coronal sections was quantified. (e) The relatively mRNA levels of TNF-α, IL-1β, IL-6 and iNOS in the brains. (f) The hippocampal CA3 of mice were immunostained for SYN. Representative images. (g) The SYN⁺ MFI in the hippocampal CA3 were quantified. >Data are presented as mean ± SEM. n = 18–19 slices from 3 mice/group (c, d); n = 3–4 mice/group (e). n = 14–17 slices from 3 mice/group (g). *p < 0.05; **p < 0.01; *** p < 0.001. Student t-test (c-e). One-way ANOVA (g). Scale bars: 200 and 50 μm in the images with lower and higher magnification, respectively (a, b); 50 μm (f)

Synaptic loss is closely related to the cognitive dysfunction of AD [96]. Synaptophysin, an integral membrane glycoprotein in presynaptic vesicles, serves as a reliable marker for quantifying synapse numbers. To assess the effect of S@A-NPs on synaptic preservation, we examined synaptic density by measuring the synaptophysin signals⁺ in the hippocampal CA3 region, an area enriched in synapses. The results showed that S@A-NPs-administered APP/PS1 transgenic mice exhibited increased synaptic density compared to control APP/PS1 mice but decreased synaptic density compared to WT mice (Fig. 7f-g). These results indicate that specifically increasing endothelial LPR1 expression level partially rescues synaptic loss in APP/PS1 mice, providing a potential mechanism for improving cognitive function in AD models.

Administration of S@A-NPs rescues cognitive deficits of APP/PS1 transgenic mice

In summary, administration of S@A-NPs enhances BBB integrity by specifically upregulating the expression of endothelial LRP1 and alleviates Aβ accumulation, neurodegeneration, and neuroinflammation in APP/PS1 transgenic mice. We next examined whether administration of S@A-NPs could rescue the cognitive deficits in these mice. As described previously, APP/PS1 transgenic mice were intravenously injected with S@A-NPs at 5.5 months of age and subjected to behavioral tests two months after injection. The open field test showed that the S@A-NPs treatment did not affect the spontaneous locomotor function of the mice (Supplementary Fig. 7).

The NOR test was used to assess declarative memory [97]. The tested mice were initially habituated to two objects, with one replaced by a novel object 90 min later. However, cognitively impaired mice fail to remember the old object and show a reduced recognition index. APP/PS1 mice exhibited a reduced recognition index compared to WT mice, and this reduction was rescued by S@A-NPs treatment (Fig. 8a, c). Furthermore, the preference index, which reflects the mice’s preference for the original objects, remained identical among groups of mice (Fig. 8b), confirming the enhanced recognition index in S@A-NPs-treated APP/PS1 mice was due to the improved cognition, rather than a bias towards the objects themselves.

Fig. 8.

S@A-NPs rescues cognitive deficits of APP/PS1 transgenic mice. (a-c) Novel object recognition test. Schematic description of test design (a). preference index (b), and recognition index (c). (d-f) Object location recognition test. Schematic description of test design (d). preference index (e), and recognition index (f). Data are presented as mean ± SEM. n = 9–14 mice/group. * p < 0.05; ** p < 0.01. One-way ANOVA (b, c, e, f)

The mice were subsequently subjected to the OLT test to assess spatial memory [98]. In the OLT test, the mice were first habituated to two identical objects placed in different locations during the training phase. 24 h later, one object was removed to a novel location (Fig. 8e). Compared to WT mice, APP/PS1 mice showed a reduced recognition index in the OLT, and this impairment was rescued by S@A-NPs treatment (Fig. 8f). Interestingly, S@A-NPs-treated WT mice exhibited similar recognition index levels in both the NOR and OLT tests (Fig. 8c, f), suggesting that these nanoparticles do not affect the cognition of normal mice. Thus, these results indicate that specifically increasing endothelial LPR1 expression level rescues the cognitive deficits of APP/PS1 mice.

Discussion

In this study, we demonstrate that S@A-NPs, LRP1-targeted nanoparticles, are internalized by endothelial cells after IV delivery, where they increase endothelial LRP1 expression through local SIM release. We further show that endothelial delivery of S@A-NPs improves BBB integrity, suppresses amyloidogenic cleavage of APP, enhances transcytosis of Aβ out of the brain, reduces neuroinflammation and neurodegeneration, and ultimately rescues cognitive deficits in an AD mouse model (Fig. 9). Therefore, this study proposes the endothelial delivery of SIM as a novel potential therapeutic approach for AD. S@A-NPs rely on the binding of Angiopep-2 to endothelial LRP1 to facilitate their endocytosis by brain endothelial cells. Once internalized, S@A-NPs release SIM locally, which further upregulates endothelial the LRP1 expression. This, in turn, enhances the internalization of additional S@A-NPs, creating a self-amplifying feedback loop that results in the accumulation of SIM within endothelial cells [65]. We propose that this “self-promoting” feedback mechanism, formed by endothelial LRP1 and S@A-NPs, is key to the selective distribution of S@A-NPs to brain endothelial cells and the upregulation of endothelial LRP1. In this context, the loss of endothelial LRP1 is an earlier biomarker and causative factor for aging [30] and AD [36–38] pathogenesis, as it is essential for BBB integrity [28] and Aβ efflux [8] from the brain. Consistent with this, our study demonstrates that S@A-NPs repair the damaged BBB and enhance Aβ transcytosis for peripheral clearance.

Fig. 9.

Administration of S@A-NPs attenuates AD pathogenesis by specifically upregulating endothelial LPR1 expression level. The loss of endothelial LRP1 impairs BBB and transcytosis of Aβ, resulting in Aβ accumulation, neuroinflammation, neurodegeneration, and cognitive deficits. Administration of S@A-NPs penetrates the brain across BBB, predominately distributed in the endothelial cells. S@A-NPs are internalized by endothelial cells and retained within the endothelial cells. S@A-NPs release SIM locally in the endothelial cells, upregulating the endothelial LRP1 levels. The latter repairs the damaged BBB, promoting the efflux of Aβ. In addition, administration of S@A-NPs decreases neuronal LRP1, possibly involving interaction between endothelial cells and neurons, suppressing Aβ production. Through these mechanisms, S@A-NPs attenuate Aβ accumulation, neuroinflammation, and neurodegeneration, eventually protecting the cognitive function of AD. The diagram created with BioRender.com and released under a Creative Commons Attribution-Non-Commercial NoDerivs 4.0 International license

Surprisingly, while endothelial LRP1 level was upregulated, neuronal LRP1 level was decreased following administration of S@A-NPs. This suggests that after uptake by endothelial cells, SIM is locally released and not further taken up by other cell types. Given that S@A-NPs primarily target brain endothelial cells, the reduction of neuronal LRP1 may result from endothelial cells-neuronal crosstalk. Endothelial cells secrete factors involved in neuronal development [99, 100], myelination [101–103], and inflammation [104, 105], potentially explaining this phenomenon. However, further investigation is required to elucidate how endothelial-targeted S@A-NPs reduce neuronal LRP1 expression. It is noteworthy that, in contrast to endothelial LRP1, which is crucial for BBB integrity and Aβ transcytosis, neuronal LRP1 promotes Aβ production [43–45] and tau propagation [48]. In line with this, S@A-NPs treatment reduced amyloidogenic cleavage of APP. As amyloidogenic cleavage of APP predominantly occurs in neurons, where BACE1 is mainly expressed [88], diminished neuronal LRP1 may contribute to S@A-NPs-mediated suppression of Aβ production. Additionally, decreased brain APP levels following S@A-NPs treatment may further reduce Aβ production. APP, a marker of axonal injury, is upregulated in damaged neurons [106]. Thus, its downregulation may reflect attenuated neurodegeneration after S@A-NPs intervention. BACE1 is also expressed in endothelial cells, where it is predominantly localized abluminally [87]. This localization enables cleavage of APP within endothelial cells, producing Aβ that directly contributes to cerebral amyloid deposition [87, 107]. Furthermore, endothelial BACE1 upregulation impairs BBB integrity independently of amyloidogenesis: it cleaves occludin, a tight junction protein, and reduces expression of ZO-1, JAM-A, and claudin-1 via cleavage-independent mechanisms [108]. We observed that S@A-NP treatment decreased endothelial BACE1 levels, potentially due to LRP1 upregulation promoting BACE1 degradation [89, 90]. Stressors like Aβ accumulation [67, 91], hypoxia, and neuronal injury [92] also upregulate BACE1. Thus, S@A-NPs-mediated alleviation of Aβ-driven stress may contribute to endothelial BACE1 downregulation as well. Intriguingly, S@A-NPs reduced BACE1 levels in brain endothelial cells of APP/PS1 mice while increasing them in WT mice. We propose several non-exclusive mechanisms for this differential effect: (1) LRP1-mediated biodistribution differences: S@A-NPs rely on Angiopep-2 ligand binding to LRP1 for endothelial uptake. Differential endothelial LRP1 expression between AD transgenic mice and WT mice may drive distinct cellular responses to S@A-NPs; (2) Inherent context-dependent effects of simvastatin: Free SIM itself exhibits context-dependent effects. For example, it modulates cholesterol in dysregulated systems but exerts minimal change in normal mice [109, 110]. S@A-NPs may thus act differentially based on baseline pathophysiology; (3) Stress-response modulation: In APP/PS1 mice, S@A-NPs-enhanced Aβ efflux alleviates Aβ-driven stress, thereby reducing BACE1 levels. This mechanism is likely inactive in Aβ pathology-free WT mice. Thus, the increased BACE expression in S@A-NPs-treated WT mice likely reflects the interplay of these mechanisms, particularly the absence of Aβ load attenuation and differential nanoparticle trafficking.

Activation of astrocytes and microglia is a double-edged sword in AD pathogenesis. Their activation facilitates Aβ clearance at the earlier stage while accelerating neurodegeneration by inducing neuroinflammation in later stages. Astrocytes and microglia are recruited to amyloid plaques, where they proliferate and become activated [95, 111]. Thus, the reduced density of astrocytes and microglia observed in S@A-NPs-treated APP/PS1 mice may be attributed to decreased Aβ accumulation. However, the crosstalk between endothelial cells and glial cells may also contribute to the suppression of neuroinflammation resulting from the endothelial delivery of S@A-NPs [112–114]. Further investigation is needed to understand the underlying mechanisms of the therapeutic effects of S@A-NPs.

Administration of S@A-NPs reduced the accumulation of soluble, rather than insoluble, Aβ aggregates in the brain. This aligns with the observation that the APP/PS1 mice injected with S@A-NPs at 5.5 months of age had already accumulated Aβ plaques [67]. The aggregation of Aβ impairs its clearance by endothelial cells. Earlier intervention, such as at 3 months of age in APP/PS1 mice when Aβ begins to accumulate, might be more effective in reducing insoluble Aβ aggregates. Similarly, the effects of S@A-NPs administration may be partially, but not entirely, due to the impaired BBB integrity and the neurodegeneration in APP/PS1 mice. Despite these observations, administration of S@A-NPs ultimately rescued cognitive deficits in APP/PS1 mice. This suggests that the partial recovery of BBB integrity and mitigation of neurodegeneration are sufficient to enhance functional cognition.

While SIM is clinically recognized for its lipid-lowering effects, it prevents cholesterol elevation in hypercholesterolemic models but induces no significant change in total serum cholesterol in normal mice [109, 110]. Crucially, in the APP/PS1 mice used in this study, serum cholesterol levels remain comparable to those of wild-type littermates until advanced ages (e.g., 12 months) [115, 116], indicating minimal baseline dysregulation of serum cholesterol at our experimental timepoints. Consistent with this profile, our prior work demonstrated that sustained IV administration of S@A-NPs, despite delivering therapeutic SIM doses, did not alter systemic cholesterol levels in healthy mice [65]. This can be attributed to the nanoparticles’ sustained-release profile and targeted biodistribution, which minimizes peripheral SIM exposure while concentrating therapeutic effects at the cerebrovascular endothelium. Importantly, this absence of systemic metabolic disruption confirms the safety profile of our approach and demonstrates that the observed endothelial LRP1 upregulation and BBB repair occur independently of whole-body cholesterol modulation. In this context, we demonstrated that long-term S@A-NPs treatment did not affect spontaneous locomotor activity, suggesting that such treatment does not cause myotoxicity, a common side effect associated with free SIM administration [117]. Our prior study demonstrated that an identical S@A-NPs treatment improves survival in metastasis-bearing mice without observable toxicity, achieving efficacy comparable to doxorubicin chemotherapy [65]. These data suggest the biosafety of S@A-NPs. However, further detailed histopathological evaluation of the biosafety of S@A-NPs is still required for IND-enabling studies.

The loss of LRP1 is a well-established driver of BBB dysfunction in aging and AD [8, 28–30, 36–38], positioning its restoration as a strategic therapeutic intervention. SIM, however, exhibited limited efficacy in AD therapy and caused adverse effects during long-term therapeutic use for AD [118, 119]. These effects may be linked to its role in upregulating neuronal LRP1 expression. Here, we demonstrate that S@A-NPs selectively localize to brain endothelial cells and actively repair BBB impairment in an AD model via targeted upregulation of endothelial LRP1. In this context, the present study represents a strategic and technical advance: First, unlike viral vector approaches, e.g., AAV-mediated LRP1 delivery [120], which face translational safety barriers, S@A-NPs leverage clinically validated components: biodegradable PLGA polymer and SIM, both with decades-long safety records. Second, this study focuses on BBB breakdown, an early pathological event that accelerates Aβ deposition, neuroinflammation, and cognitive decline, thereby offering a potential opportunity for earlier intervention in disease modification. Third, and most significantly, while our previous study demonstrated the utility of S@A-NPs in improving drug distribution and survival outcomes in metastatic models, it did not explore the intrinsic biological activity of the S@A-NPs (devoid of drug cargo) on BBB integrity, a gap this study decisively addresses [65]. This novel study found that the S@A-NPs themselves function therapeutically without requiring further loading with any drug, thereby attenuating AD pathogenesis.

LRP1 is also expressed by hepatocytes. Thus, it is possible that S@A-NPs were uptaken by hepatocytes as well. However, due to the different cellular responses between endothelial cells and hepatocytes, further investigation is needed to determine whether S@A-NPs also enhance hepatocytic LRP1 expression. In this context, we emphasize the cell-type-specific functions of LRP1: brain endothelial cells employ LRP1 for Aβ transcytosis out of the CNS, while hepatocytes utilize it for systemic Aβ clearance from circulation [121]. Thus, while S@A-NPs may interact with hepatocytes, their therapeutic effect on BBB repair stems principally from endothelial-specific LRP1 upregulation rather than hepatic modulation. Nevertheless, this study requires multiple IV injections of S@A-NPs, which may limit its clinical appeal. As a proof-of-concept study, our primary aim was to establish the therapeutic potential and mechanism of S@A-NPs. Future work will focus on optimizing nanoparticle design or exploring alternative delivery strategies (e.g., implantable devices) to reduce dosing frequency and enhance translational potential.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AD

Alzheimer’s disease

APP/PS1

APPswe/PS1DE9

WT

Wild-type

Aβ

Amyloid β

BBB

Blood brain barrier

SIM

Simvastatin

NS

Normal saline

NPs

Nanoparticles

HMG-CoA

Hydroxy methylglutaryl-coenzyme A

LRP1

Low-density lipoprotein receptor related protein 1

RAGE

Receptor for advanced glycation end-products

ZO1

Zona occludens 1

CBF

cerebral blood flow

BACE1

β-site APP cleaving enzyme 1

CTF

C-terminal fragment.

Cy3

Cyanine 3 phosphoramidite

PLGA

Polylactic-co-glycolic acid

PLL

Poly-L-lysine

PEG

Polyethylene glycol

Author contributions

Conceptualization, Q-H.M.; methodology, T-T.P. and Y-Y.S.; validation, Y-F.S., M.Z. and W-L.J.; formal analysis, T-T.P. and Y-F.S.; investigation, T-T.P., Y-Y.S., Y-F.S., M.Z., H-Y.C., W-L.J. and N-U.K.; resources, Q-H.M. and L.H.; data curation, T-T.P. and Y-Y.S.; writing-original draft preparation, Q-H.M.; T-T.P.; writing-review and editing, Y-Y.S., J.L., L.H., and Q-H.M.; supervision, Q-H.M.; funding acquisition, Q-H.M., Y-Y.S. and L.H.; All authors have read and agreed to the published version of the manuscript.

Funding

The authors are grateful to the following funds for support: the STI2030-Major Projects (2021ZD0204001), the National Natural Science Foundation of China (92049120, 81870897), Sino German cooperation and exchange project (M-0679), the Project of MOE Key Laboratory of Geriatric Diseases and Immunology (A0103), the Guangdong Key Project in the Development of New Tools for the Diagnosis and Treatment of Autism (2018B030335001), Research Project of Neurological Diseases in the Second Affiliated Hospital of Suzhou University, Research Center (ND2024A01), the Basic Frontier Innovation Cross-Scientific Research Project of Suzhou Medical College of Soochow University (YXY2304058), Suzhou International Joint Laboratory for Diagnosis and Treatment of Brain Diseases, A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions, National Natural Science Foundation of China (32171381), and Natural Science Foundation of Jiangsu Province (BK20240149).

Data availability

The data that supports the conclusions of this study are available in the paper and/or the Supplementary material.

Declarations

Ethics approval and consent to participate

All procedures were approved by the Institutional Animal Care and Use Committee of Soochow University and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals guidelines for the ethical treatment of animals. Efforts were made to minimize the number of animals used.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.