To fully understand the role of BBB LPR1 in mediating Aβ transport, we followed its expression alongside other markers and Aβ in the APP/PS1 and wild-type animals. We conducted a multi-tiered comparative investigation employing enzyme-linked immunosorbent assay (ELISA) quantification and in situ imaging modalities. This integrated methodology facilitated comprehensive whole-brain profiling coupled with systematic evaluation of separated vascular-parenchymal (Fig. 2 and supplementary Fig. 1).
Fig. 2 Comparative analysis of protein expression profiles in AD and wild-type mice. Confocal microscopy of brain endothelial cells, pericytes, and Aβ in AD mice of different ages. LRP1 (gray), Aβ (red), and BBB endothelium (CD31, green) in 3-month-old (a) and 12-month-old (b) AD mice brain samples. 3D rendering (c) of the 12-month-old AD brain section highlighting the spatial relationship between the different markers. In later AD stages, increased Aβ accumulation near the BBB and reduced colocalization of CD31 with LRP1 are observed. CD31, CD146, and LRP1 expression in a 12-month-old AD brain sample (d). Scale bar = 10 μm. Age-dependent colocalization quantification between LRP1/Aβ and endothelial cells (CD31) as well as between LRP1/Aβ and pericytes (CD146) in AD mice using PCC (e). Studise was performed on images derived from three independent experiments, with 6–7 vessels analysis per trial, the data are presented as the means ± SEMs. Quantification of Aβ, LRP1, PACSIN2, and Rab5 levels as a function of age by ELISA of the brain vasculature and parenchyma (f) in AD and wild-type mice. Quantitative representation of Aβ-positive area fraction in brain coronal sections identified through immunohistochemistry (IHC) in both AD model and wild-type mice (g). Aβ concentrations via ELISA in both AD and wild-type mice in the whole brain (h). (For graphs f, g, h, n ≥ 3 per group, the data are presented as the means ± SEMs.) Full size image
We used confocal microscopy to assess the spatial localization of LRP1 and Aβ at the BBB endothelial cells (CD31) and pericytes (CD146) in 3- and 12-month-old AD brain samples. In Fig. 2a (3-month-old), Aβ (red) is highly colocalized with LRP1 (white) on the endothelial cells (green), suggesting the active involvement of LRP1 in Aβ transport and clearance at a younger age, with less Aβ accumulation around the vessels. Figure 2b shows 12-month-old brains, and the corresponding 3D image (Fig. 2c) indicates a noticeable increase in Aβ deposition on the basal side of the BBB vessels. The colocalization of LRP1 with Aβ appears to decrease, potentially indicating impaired LRP1-mediated clearance of Aβ as AD progresses. In the later stages of AD, increased Aβ accumulation and reduced association with LRP1 may affect BBB function and promote disease pathology. Interestingly, further imaging (Fig. 2d) suggested that LRP1 was predominantly deposited around the pericytes on the exterior side of the blood vessels. We conducted longitudinal analysis of representative brain sections spanning 3 to 12 months through immunolabeling with antibodies targeting LRP1, Aβ, pericyte marker (CD146), and endothelial cell marker (CD31). Colocalization analysis was performed, and Pearson correlation coefficients (PCC) were calculated to quantify the spatial relationships between Aβ/LRP1 and CD31, as well as between Aβ/LRP1 and CD146. In Fig. 2e, the results show a trend where the association between Aβ and endothelial cells weakens over time, while its correlation with pericytes appears to strengthen. Similarly, analyses with LRP1 revealed PCC over time, specifically between LRP1 and endothelial cells as well as between LRP1 and pericytes, suggesting potential associations.
We collected brains from both AD and wild-type mice over a lifespan of 3–12 months and fractionated them into parenchyma and vasculature. We thus measured Aβ, LRP1, PACSIN2, and Rab5 levels via ELISA. The data shown in Fig. 2f reveal significantly more Aβ in the vasculature of wild-type mice with notable differences emerging at all lifespan stages compared with APP/PS1 mice, demonstrating the pathological hallmarks of AD. These differences correspond to increased Aβ levels in the parenchyma. The latter is the dominant index of temporal changes in Aβ from the macroscopic whole brain. Figure 2g (quantitative data of supplementary Fig. 1) and Fig. 2h display the aggregate Aβ levels in the brain, showing a marked Aβ increase in the AD models with age, a trend particularly pronounced between 6 and 12 months. The buildup of Aβ in brain, along with its restricted passage through blood vessels, corresponds to the downregulation of LRP1 and PACSIN2, alongside the upregulation of Rab5 as the animals aged, as measured by ELISA (Fig. 2f), and immune fluorescence (supplementary Fig. 2). Notably, this difference between AD and wild-type animals was especially significant during the 6- to 12-month period, particularly in the vascular system (Fig. 2f and supplementary Fig. 2). The interplay between LRP1, PACSIN2, and Rab5 at the BBB is crucial for understanding the mechanisms of aging and AD. Our previous studies40,42,50 in which the peptide angiopep-2 was used to target LRP1 revealed that the efficiency of crossing the BBB is greater for multivalent scaffolds. We demonstrated that LRP1 shuttles across the BBB through transcytosis31 via collective endocytosis and exocytosis regulated by the BAR domain protein PACSIN2 for mid-avidity cargo. PACSIN2 plays a pivotal role in facilitating transport via LRP1 for small Aβ structures (i.e., mid-avidity cargo) across the BBB.41 A large Aβ structure with greater affinity for LRP1 traffics toward Rab5-positive endosomes via the recruitment of PICALM and clathrin-mediated endocytosis.35 The loss of BBB integrity may trigger compensatory mechanisms, including the upregulation of Rab5, as the brain attempts to increase endosomal trafficking to manage increased cellular stress and the accumulation of neurotoxic substances, such as Aβ. However, Rab5 is significantly upregulated in vulnerable neuronal populations, particularly in individuals with AD.51,52 Combining the above ELISA and confocal evaluation results, the localization shift in LRP1 from the BBB vascular endothelium to pericytes with aging underscores a potentially pivotal role in the pathophysiology of AD. This progression suggests a decrease in LRP1-mediated Aβ clearance at the BBB endothelial level, with a concomitant increase associated with pericytes, which may impact AD progression. Most importantly, the timing of this alteration precedes or evolves alongside the early stage of cognitive decline, as measured in the APP/PS1 AD model we used.50
As we previously reported, both small Aβ and mid-avidity multivalent units trigger PACSIN2-mediated transcytosis. In both, this pathway is associated with the upregulation of the LRP1 receptor,31,32 as shown in Fig. 1b. We thus hypothesize that the use of multivalent LRP1-targeted nanoparticles may restore the ability of LRP1 to transport Aβ from the brain and potentially clear Aβ deposits in AD models. We prepared and characterized P[(OEG) 10 MA] 20 -PDPA 120 mixed with angiopep2-P[(OEG) 10 MA] 20 -PDPA 120 to make polymersomes bearing 40 ligands per particle. Hereinafter, these polymersomes are referred to as A 40 -POs (supplementary Fig. 3). The number of ligands optimized for transcytosis31 was adjusted via our phenotypic targeting theory calculations to account for the reduced LRP1 expression in AD mice.34
APP/PS1 transgenic AD mice were intravenously injected with 200 μL of A 40 -POs alongside four control treatments: a sham formulation (only PBS), angiopep-2 alone (A1), pristine P[(OEG) 10 MA] 20 -PDPA 120 polymersomes (A 0 -POs), and polymersomes with 200 angiopep-2 ligands (A 200 -POs). Two hours after administration, the animals were culled, and the Aβ levels in both the brain and blood plasma were measured via ELISA. The results plotted in Fig. 3a, b show a clear effect on only A 40 -POs treatment, with a reduction in brain Aβ of almost 50%, from 8603.6 to 4236.3 ng ml−1, and a mirrored increase in the blood plasma of 8 times from 85.3 to 673.5 ng ml−1 compared with that of the diseased animals treated with a sham formulation. If we assume that the brain volume of an APP/PS1 12-month-old mouse brain is 0.35–0.45 ml and has a total blood volume of 1.5–2.3 ml, the amount of Aβ removed from the brain corresponds almost entirely to the surplus measured in the plasma.
Fig. 3 A 40 -POs treatment reduces cerebral Aβ burden in APP/PS1 mice. ELISA measurement of whole-brain (a) and plasma (b) Aβ levels at 2 h post injection, comparing wild-type (WT), sham, angiopep-2 alone (A 1 ), pristine P[(OEG) 10 MA] 20 -PDPA 120 polymersomes (A 0 -POs), and angiopep-2-functionalized polymersomes with ligand densities of 40 (A 40 -POs) or 200 (A 200 -POs) per vesicle. PET‒CT visualization (c) in wild-type (WT) and APP/PS1 (pre- and 12 h post A 40 -POs administration) mice with [18 F] AV-45 (2.8--3.2 MBq) tracer. Quantified SUV (d) reductions confirm significant Aβ clearance. 3D rendering imaging (e) of the brain after tissue clearing shows reduced Aβ signals in 12 h post A 40 -POs injected mice. Brain parcellation into 14 regions was performed according to the Allen Brain Atlas, revealing Aβ distribution across brain regions (f). A 40 -POs treatment induced 41% Aβ volume reduction in total after 12 h of treatment (g) (data expressed as mean ± SEM). Heat map representation of Aβ fluorescence intensity across a single coronal brain layer (h). Tissue-clearing and PET-CT imaging were conducted three repeats per group (n = 3). Statistical significance for graphs a, b, d was determined via one-way analysis of variance (ANOVA), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Full size image
A parallel IHC analysis confirmed that the Aβ area fraction also decreased (supplementary Fig. 4). Furthermore, we employed positron emission tomography-computed tomography (PET-CT) to assess the clearance of Aβ in the brains of live animals. The animals were injected with [18 F]-4-(2-(6-(2-(2-(2-18F-fluoroethoxy)ethoxy)ethoxy)pyridin-3-yl)vinyl)-N-methyl benzamine ([18 F] AV-45), an established Aβ marker.53 PET‒CT revealed that the brain of 12-month-old APP/PS1 mice exhibited intense Aβ signal. In contrast, this signal sharply decreased after treatment with A 40 -POs (Fig. 3c). After 12 h of administration of A 40 -POs, the reduction in [18 F] AV-45 standardized uptake value associated with Aβ was 46.25% (Fig. 3d). Confocal images revealed that Aβ deposition around the BBB disappeared, and a large amount of Aβ signal in the vascular lumen (supplementary Fig. 5). We performed tissue clearing on the brains of 12-month-old APP/PS1 mice treated with sham formulation or A 40 -POs. The Aβ (red) and blood vessels (green) of these brains were labeled (Fig. 3e and supplementary Fig. 6). The brains of the mice treated with A 40 -POs presented fewer Aβ signals than did the Sham APP/PS1 brains. The 3D brain images were embedded into the Allen Brain Atlas-based parcellation model integrated with Amira software, with each brain parcellated into 14 distinct regions (Fig. 3f). The Aβ volume in 14 brain regions of the mouse brain was measured separately (Fig. 3g). There was a 41% Aβ volume reduction in the brains of A 40 -POs-treated mice. Finally, the coronal Aβ distribution is shown as a heatmap in Fig. 3h.
These findings motivated us to study the BBB vascular phenotype after A 40 -PO treatment. We first observed increasing of the colocalization of LRP1 with CD31 in the treated brain, as shown in Fig. 4a and supplementary Fig. 7. The overlap of LRP1 and BBB endothelial cells (CD31) returned to the wild-type state. Quantitative analysis of Aβ distribution revealed a significant increase in the brain vasculature after treatment (Fig. 4b), contrasting with a progressive reduction in parenchymal Aβ deposition (Fig. 4c). ELISA tests were subsequently performed to detect proteins in both the vasculature and parenchyma. As discussed previously, the analysis focused on the concentrations of various proteins, including LRP1, PACSIN2, and Rab5. The nanomedicine cleared Aβ and caused a rapid change in the BBB phenotype by upregulating PACSIN2 and downregulating Rab5 (Fig. 4d). This finding is consistent with our fluorescent imaging data, which show that PACSIN2 relocates to blood vessels (supplementary Fig. 7). The morphology of LRP1, as observed under a stimulated emission depletion (STED) microscope, revealed a clustered distribution in the vessel wall, suggesting robust ongoing transcytosis (Fig. 4e and supplementary Fig. 5).
Fig. 4 A 40 -POs treatment restored the BBB phenotype. PCC for the colocalization of LRP1 and endothelial cells (CD31) (a). Analyzis was performed on images derived from three independent experiments, with 6–7 vessels studied per trial (statistical analysis performed via unpaired t-tests, **** p < 0.0001). Aβ content in cerebrovasculature (b) and parenchyma (c) quantified by ELISA. ELISA measurements of vascular and parenchymal LRP1, PACSIN2, and Rab5 levels in wild-type, Sham APP/PS1, and APP/PS1 mice after A 40 -PO treatment (d). STED microscopy imaging of LRP1 (white), Aβ (red), and vessel wall (green). After treatment, Aβ deposits around the BBB are cleared, and notable Aβ signals are present within the vascular lumen (e). For b, c, and d, statistical significance was determined via one-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n ≥ 3 Full size image
Finally, we investigated the effects of A 40 -POs administration on animal cognition via Morris water maze. As indicated in Fig. 5a the stage I, with the number of experimental days increased, the time they took to find the platform gradually decreased, suggesting that animals made progress in learning and remembering the platform’s location.
Fig. 5 Behavior tests demonstrated that A 40 -PO treatment improved the performance of APP/PS1 mice. Morris water maze test in which mice were injected with saline (sham APP/PS1 group and WT group, 200 μL) or A 40 -POs (APP/PS1 POs group, 10 g/L 200 μL) once daily for the 365th–367th morning of their lifespan. Recovery was executed for 1 week under the original rearing conditions. The place navigation test (Stage I) was performed on days 375th–378th, and the results revealed a gradual decrease in passing length and escape latency for finding the platform in all groups, with the APP/PS1 POs group matching the WT group and significantly outperforming the Sham APP/PS1 group. During the spatial probe test (Stage II), both the APP/PS1 POs and WT groups demonstrated more passing times and a greater percentage of time spent at the platform’s original location. In the reverse-place navigation (Stage III) trial from days 380th–383rd, with the platform moving to the opposite side, the APP/PS1 POs and WT groups initially took longer, indicating stronger spatial memory from Stage I and Stage II. Their reverse passing length and escape latency decreased rapidly over time and were significantly lower than those of the Sham APP/PS1 mice at the last day of this stage. On day 384th, in the reverse spatial probe (stage IV) test without the platform, A 40 -POs treated mice still outperformed the Sham APP/PS1 group. After 180 days, the mice were reperformed for place navigation and spatial probes (Stages V and VI). The conditions were consistent with those of stage I and stage II. Six months after the injection of A 40 -POs, the mice could still find the escape platform within a shorter time in the navigation experiment. They stay longer and traverse more times at the correct location in the spatial probe test. The performance of the mice injected with A 40 -POs (6 months p.i.) was close to that of the wild-type mice with same age, and better than that of the Sham APP/PS1 group (a). Place navigation trials (Stages I, III and V) were analyzed via two-way ANOVA, whereas spatial probe trials (Stages II, IV, and VI) were compared via one-way ANOVA. Significance levels are denoted as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, with n ≥ 11 (for all 12-month-old mice) and n ≥ 6 (for all 18-month-old mice). Nest-construction images (b), nest scores (c) and sucrose preferences (d) were recorded for the APP/PS1 Sham, APP/PS1 POs and wild-type groups at two time points post-injection. (groups were compared via one-way ANOVA. Significance levels are denoted as *p < 0.05, **p < 0.01, ***p < 0.001 with n = 5, the data are presented as the means ± SEMs.) Full size image
The APP/PS1 POs (A 40 -POs-treated) group exhibited a significantly shorter escape path length than the sham-operated APP/PS1 group after training (Fig. 5a, Stage I), suggesting improved spatial navigation strategies. Their search efficiency was similar to that of the wild-type mice. The escape latency (time taken to reach the escape platform) of the APP/PS1 POs group was also shorter than that of the sham group. A shorter escape latency indicates better spatial learning and memory ability. Although the relationship between swimming speed and spatial learning and memory abilities is weak, analyzing swimming speed can rule out the impact of an animal’s motor ability or fear during experiments. For all the parts, there was no significant difference in the swimming speeds of the three groups of mice.
When the platform was removed in Stage II, the APP/PS1 POs group crossed the platform more times and spent a significantly greater percentage of time at the platform’s original location than the Sham group did, reflecting stronger memory of the platform’s location. In Stage III, the escape platform was placed on the opposite side of its original location. Initially, longer search times reflect the long-term memory of the original platform’s location. Nevertheless, as the number of training sessions increased, the group with stronger learning abilities would present a greater reduction in path length and escape latency. In the last two days of this stage, animals treated with A 40 -POs and the wild-type group presented shorter escape paths and escape latencies. When the escape platform was removed (stage IV), the APP/PS1 POs group stayed longer than the sham group at the escape location. This location was crossed more often, reflecting the stronger memory abilities. Six months after the mice were treated with A 40 -POs, we performed this water maze experiment to evaluate the persistence of cognitive improvement in the treated mice. Place navigation and spatial probe tests (Stages V and VI) were performed on the mice adhering to the same methods as those implemented in Stages I and II. Between-group comparisons revealed that the cognitive enhancement provided by A 40 -POs treatment persisted in APP/PS1 mice, and A 40 -POs-treated mice demonstrated a level of cognitive similar to that of wild-type mice, which was significantly greater than that of sham APP/PS1 mice.
Enhancing quality of life is a crucial objective in AD treatment and improvement. To assess the life quality of the mice, we conducted nest construction (Fig. 5b, c) and sucrose preference (Fig. 5d) experiments following the Stages IV and VI of the Morris water maze test. Nest-construction behavior is commonly used to evaluate daily activities, fine motor skills, cognition, and emotional state in mice with cognitive impairments. For mice, a high-quality nest provides thermoregulation and predator avoidance, serving as a security indicator that correlates with executive function performance. The treated group exhibited a significantly higher nest-construction score compared to the Sham group. The sucrose preference experiment was conducted to assess the hedonic response of the animals to sweetness by administering a low-concentration sucrose solution. The group treated with A 40 -POs exhibited significantly higher preference scores compared to the sham APP/PS1 group (Fig. 5d).