Abstract

Results

In young wild-type [nontransgenic (NTG)] or LDLR-TG [transgenic (TG)] mice (18), we used the BEI method (13–15, 20) to test the hypothesis that LDLR regulates steady-state Aβ levels by enhancing clearance from the brain. To compare the clearance kinetics from the brain over various time points (15–150 min), 12 nM [¹²⁵I]-radiolabeled, monomeric Aβ40 was injected simultaneously with [¹⁴C]inulin into brain ISF. Unlabeled and radiolabeled Aβ have been shown to exhibit nearly identical clearance kinetics (13). [¹⁴C]inulin serves as a reference marker of ISF bulk flow because it does not actively clear across the BBB. Total brain ISF Aβ clearance, corrected for degradation within the brain (SI Materials and Methods), was significantly faster from brains of TG mice compared with NTG mice (Fig. 1A; see Fig S1A for scatterplot of data). Analysis of major components of brain-to-blood efflux (BBB and ISF bulk flow) revealed that LDLR overexpression increased the BBB-mediated component of Aβ clearance compared with NTG mice, as indicated by the greater slope for TG vs. NTG mice from a plot of [¹²⁵I]Aβ40 remaining vs. time (Fig. 1B). Notably, the contribution of ISF bulk flow to total Aβ clearance was minimal (Fig. 1B), consistent with previous studies (14, 15, 20). Given the purported role of apoE in BBB integrity (21, 22), we monitored the elimination of [¹⁴C]inulin over the entire time course for both groups, which revealed that [¹⁴C]inulin was cleared in a slow, passive manner and to a similar extent in both groups (Fig. S1B), strongly suggesting an intact BBB in TG mice.

Open in a separate window

Fig. 1.

LDLR enhances clearance of radiolabeled Aβ from brain. (A) Percentage remaining for 12 nM [¹²⁵I]Aβ40 microinjected in ISF of caudate-putamen in NTG (■) and TG (○) mice killed at various time points. Percentage recovery was calculated from Eq. S1 (SI Materials and Methods). (B) Time-dependent clearance of [¹²⁵I]Aβ40 by passive ISF bulk flow (♦) and across the BBB after correction for degradation within brain by TCA precipitation method (NTG, ■; TG, ○) calculated from data in Fig. 1A and Eq. S4 (SI Materials and Methods). (C) Using fractional rate constants calculated in Table S1, relative contributions of degradation-corrected clearance of [¹²⁵I]Aβ40 by the BBB and ISF bulk flow, as well as retention within brain, were calculated for NTG (black bars) and TG (white bars) mice. Each component is indicated with a plus sign (+). Complete time course includes 32–41 mice (n = 4–6 mice per time point for each group). Values in A and C are represented as mean ± SEM. When two-way ANOVA was significant (with genotype and component as factors), differences among clearance components were assessed using Tukey’s post hoc test for multiple comparisons. ***P < 0.001, % BBB for NTG vs. TG. ^†††P < 0.001, % brain retention for NTG vs. TG. n.s., no significant difference between ISF bulk flow components between NTG and TG.

Based on the passive elimination kinetics of inulin from brain ISF and the total clearance of [¹²⁵I]Aβ40, we used our kinetic model (SI Materials and Methods) to calculate the relative contribution of ISF bulk flow and BBB transport to Aβ clearance in NTG and TG mice (Fig. 1C). A greater proportion of total Aβ clearance was attributed to BBB transport in TG mice compared with NTG mice (66.9% compared with 36.3%; Fig. 1C). Conversely, less Aβ was retained within brains of TG mice compared with NTG mice (27.8% compared with 58.6%, respectively). The proportion of Aβ clearance attributed to ISF bulk flow did not differ between NTG and TG mice (5.1% compared with 5.3%, respectively). Fractional rate constants (k, min⁻¹) were calculated (SI Materials and Methods) to determine the rates of Aβ clearance mediated by the BBB, ISF bulk flow, and brain retention (Table S1). We performed trichloroacetic acid (TCA) precipitation for brains of each group at early (30 min) and late (120 min) time points to compare cellular degradation within the remaining fraction of brain [¹²⁵I]Aβ. The proportion of TCA-precipitable (intact) Aβ did not differ significantly between NTG and TG mice, although a trend was noted toward greater degradation in brains of TG mice at both time points (Fig. S1C). Together, these results suggest LDLR enhances BBB-mediated Aβ clearance, whereas other modes of clearance do not appear to be significantly altered. Surprisingly, although LDLR overexpression clearly increased BBB-mediated clearance of [¹²⁵I]Aβ, expression of the HA-tagged LDLR transgene did not overlap with expression of BBB markers such as CD31 [platelet endothelial cell adhesion molecule (PECAM-1)] (Fig. S2A) (23), nor did it colocalize with aquaporin 4 (Aqp4) expression (Fig. S2B), which labels microvessel abluminal surfaces at the astrocyte-vessel interface (24). These results suggest the LDLR overexpression in neurons and astrocytes (18) likely increases BBB-mediated Aβ clearance through an indirect mechanism involving other apoE receptors at the BBB, such as LRP1—a receptor shown to directly mediate Aβ clearance across the BBB (14, 15, 26). To test this hypothesis, we delivered an anti-LRP1 antibody (N20) or vehicle into the brains of TG mice immediately before coinjection of 12 nM [¹²⁵I]Aβ40 and [¹⁴C]inulin. Blocking LRP1 resulted in a dramatic attenuation of the BBB component of Aβ clearance in TG mice compared with vehicle (Fig. S3). Though blocking LRP1 led to decreased BBB transport and thus greater brain retention, the proportion of Aβ cleared via ISF bulk flow was unaltered compared with vehicle. Together, these results suggest that the increased BBB component of Aβ clearance observed with LDLR overexpression is, in part, mediated by the action of LRP1 at the BBB.

The rapid degradation of Aβ in the periphery (t_1/2 = 2–3 min) precludes a direct and sensitive measurement of the rate of Aβ appearance from brain into blood (27, 28). To more directly assess the rate of brain-to-blood Aβ clearance, we first cross-bred LDLR-TG mice with the PDAPP (APP-V717F) mouse model of β-amyloidosis. PDAPP mice have been previously reported to produce human APP/Aβ solely within the CNS (29, 30), allowing the fate of Aβ to be followed from brain into blood. Hippocampus and cortex from 10-mo-old PDAPP/LDLR mice (PD-TG) exhibited 2.7-fold and 4.8-fold less Aβ burden, respectively, compared with PDAPP (PD-NTG) littermates (Fig. 2 A and B). Fibrillar amyloid burden in these regions was also significantly reduced as a result of LDLR overexpression (Fig. 2 C and D). Additionally, LDLR overexpression decreased apoE levels in hippocampal and cortical homogenates by 2.9-fold and 2.5-fold, respectively, compared with mice expressing normal levels of LDLR (Fig. 2E), further validating these mice as a useful model to understand the role of LDLR in brain-to-blood clearance of centrally derived human Aβ.

Open in a separate window

Fig. 2.

LDLR overexpression in PDAPP mice markedly decreases brain Aβ/amyloid deposition and apoE levels. (A) Representative coronal brain sections from 10-mo-old, sex-matched PDAPP^+/− mice expressing normal levels of LDLR (PD-NTG), and PDAPP^+/− mice overexpressing LDLR (PD-TG). Aβ immunostaining was performed using anti-Aβ antibody (biotinylated 3D6). (Scale bars, 300 μm.) (B) Quantification of the area of the hippocampus or cortex occupied by Aβ immunostaining (n = 9 mice per group). (C) Representative amyloid burden in coronal brain sections from 10-mo-old, sex-matched PD-NTG mice and PD-TG mice. Amyloid was visualized using the congophilic fluorescent dye, X-34. (Scale bars, 100 μm.) (D) Quantification of the area of hippocampus or cortex occupied by X-34 staining (n = 9–10 mice per group). In B and D, groups were compared using the Mann–Whitney U test. *P < 0.05, **P < 0.01, ***P < 0.001. (E) ApoE protein levels measured by sensitive sandwich ELISA in hippocampal and cortical homogenates from PD-NTG and PD-TG mice (at 3–4 mo of age to avoid confounding effects from amyloid plaque deposition; n = 9 mice per group). Differences between groups were assessed using two-tailed Student’s t test (with Welch's correction for E). ***P < 0.001. Values represent means ± SEM.

To directly compare the rate at which Aβ enters the blood from brain in PD-NTG and PD-TG mice, we developed a method to capture centrally derived, endogenously secreted Aβ over time in the periphery, thus protecting it from rapid degradation. Based on work characterizing the ability of anti-Aβ antibodies to rapidly bind Aβ in the periphery and prolong its half-life (30–32), we identified an anti-Aβ antibody specific for the central domain of Aβ (HJ5.1) that strongly bound Aβ40 and Aβ42 with thermodynamic dissociation constants (K_ds) of 20.8 ± 5.45 nM and 0.623 ± 0.230 nM, respectively (Fig. S4). Consistent with the long half-life of antibodies in the periphery (30), the plasma concentration of biotinylated HJ5.1 i.v. injected in PDAPP mice was stable over the entire serial retro-orbital plasma collection period and was in significant molar excess of circulating Aβ (Fig. 3A). Only a small fraction of injected antibody was found in hippocampal and cortical homogenates (2.65 × 10⁻³% to 1.75 × 10⁻²%; Table S2). To assess whether this small fraction alters brain Aβ metabolism, and to test whether HJ5.1 present in the periphery alters the brain-to-blood equilibrium of Aβ efflux over this acute time course (30), we sampled brain ISF during in vivo microdialysis in PD-NTG mice injected with HJ5.1. Levels of ISF Aβ over a 5-h period following i.v. antibody administration did not change compared with baseline ISF Aβ levels (Fig. 3 B and C). ISF Aβ metabolism was similarly unchanged following HJ5.1 injection in PD-TG mice (Fig. 3 D and E), arguing that administration of HJ5.1 would not confound brain-to-blood Aβ efflux by differentially altering Aβ metabolism in the brains of either group of mice. Notably, PDAPP mice overexpressing LDLR exhibited markedly lower steady-state ISF Aβ levels before HJ5.1 administration compared with PD-NTG mice (Fig. 3F), consistent with decreased Aβ accumulation observed in older PD-TG mice compared with PD-NTG mice. To compare the rate of brain-to-blood Aβ appearance in both groups of mice, we collected retro-orbital blood samples serially at various time points from mice of both groups following i.v. HJ5.1 administration. The concentration of CNS-derived human Aβ in plasma samples was determined using quantitative mass spectrometry. The kinetics of Aβ appearance were reliably linear for the duration of the time course in each mouse (Fig. 4A), reflecting the rapid capture of human Aβ entering the periphery from brain. The plasma appearance rate of human Aβ was significantly faster in PD-TG mice compared with PD-NTG mice (92 ± 4.8 pgmL⁻¹min⁻¹ vs. 69 ± 6.9 pgmL⁻¹min⁻¹; Fig. 4B). These results directly demonstrate in vivo that LDLR regulates the rate at which endogenously produced Aβ enters the blood from brain.

Open in a separate window

Fig. 3.

Intravenous HJ5.1 administration results in stable antibody steady-state levels in plasma without altering brain ISF Aβ metabolism. (A) Concentration of biotinylated HJ5.1 (HJ5.1B) in plasma collected by serial retro-orbital bleeds following intrajugular injection of HJ5.1B in PDAPP^+/− mice (n = 4; 3–4 mo old). (B) In vivo microdialysis was performed in PDAPP^+/− mice expressing normal levels of LDLR (PD-NTG) to monitor ISF Aβ_1−x levels during baseline sampling as well as the period following intrajugular administration of 250 μg HJ5.1 (n = 5; 3–4 mo old). (C) Mean effect of HJ5.1 treatment on ISF Aβ_1−x levels compared with mean baseline period preceding treatment. (D and E) Experiments in B and C were repeated in PDAPP^+/− mice overexpressing LDLR (PD-TG) (n = 5; 3–4 mo old). (F) ISF Aβ_1-x levels during the baseline period of microdialysis were compared between PD-NTG and PD-TG mice. Differences between groups were assessed by paired Student’s t test in C and E and Student’s t test in F. **P < 0.01. Values represent mean ± SEM.

Open in a separate window

Fig. 4.

Antibody-assisted plasma accumulation of brain Aβ reveals faster brain-to-blood appearance rate in PDAPP mice overexpressing LDLR. (A) Representative plasma accumulation experiment illustrating kinetics of brain-derived Aβ appearance in plasma collected by serial retro-orbital bleeds following HJ5.1 treatment. Appearance rates were calculated from the slopes of individual linear regressions, e.g., for A, 82.1 pgmL⁻¹min⁻¹. (B) Mean rate of Aβ appearance in PDAPP^+/− mice expressing normal levels of LDLR (PD-NTG) or PDAPP^+/− mice overexpressing LDLR (PD-TG) (n = 6–7 per group; 3.5–4.5 mo old). Difference between groups was analyzed using two-tailed Student’s t test. *P < 0.05. Values in B represent mean ± SEM.

Discussion

The accumulation of Aβ into high-order species and amyloid plaques throughout life is hypothesized to be a critical initiating event in AD pathogenesis (2, 3, 33). Recent data have emerged suggesting that Aβ accumulates in the vast majority of AD cases as a result of impaired Aβ clearance and not increased synthesis (4). We recently provided in vivo evidence that human apoE isoforms differentially regulate soluble Aβ clearance from brain ISF (11, 15), with the slowest Aβ clearance observed in mice expressing APOE ε4 (11), the strongest identified genetic risk factor for AD (5). Based on previous evidence that receptors for apoE modulate Aβ metabolism (34), we sought to elaborate the previously unappreciated role of LDLR in Aβ metabolism. Although LDLR is well-studied for its role in mediating removal of cholesterol and cholesterol esters in the periphery (35), little is known about its function in the CNS. Recent work has identified that LDLR is a major apoE receptor in the CNS (16) that profoundly affects the accumulation of Aβ (17–19). In the current study, we found that LDLR regulates clearance of exogenously administered Aβ across the BBB but does not significantly alter clearance by ISF bulk flow. We then created mice that overexpress LDLR in the setting of CNS expression of human Aβ using the PDAPP mouse model of β-amyloidosis. We found that LDLR overexpression in young PDAPP mice markedly decreases apoE levels and decreases Aβ deposition in aged PDAPP mice. We next developed a method to stabilize human Aβ entering the peripheral circulation from brain using a high-affinity anti-Aβ antibody. Using this method, we found that LDLR overexpression significantly increases the appearance rate of endogenously produced human Aβ from brain to blood. Together, our results suggest a mechanism whereby LDLR regulates brain Aβ accumulation via BBB-mediated elimination of brain Aβ.

Previous work has identified that several members of the LDLR family of receptors, including LRP1, LRP1B, SorLA, and apoER2, regulate the trafficking and processing of the amyloid precursor protein (APP) (34, 36–39). For example, LRP1 has been shown to interact with APP, regulating its internalization, trafficking, and its subsequent processing to Aβ (36, 40–42). We did not observe any changes in APP expression or processing in brains of mice overexpressing LDLR (18); our work strongly suggests that LDLR influences Aβ metabolism by affecting its clearance from the brain into blood, a mechanism previously suggested only for LRP1 and VLDLR (13–15, 26). ApoE has been shown to impede the clearance of Aβ from brain ISF (13, 15, 43); therefore, it is likely that the reduction of apoE levels with LDLR overexpression facilitates greater ISF Aβ clearance across the BBB. Given that LDLR overexpression was limited to neurons and astrocytes in our model (18), with no transgene expression in cells constituting the BBB (Fig. S2), we speculate that LDLR-mediated removal of extracellular apoE results in increased Aβ clearance across the BBB via LRP1, which has been implicated in mediating direct BBB-mediated Aβ clearance (14). Indeed, we provide evidence that LRP1 may be, in part, responsible for mediating Aβ clearance across the BBB in the context of LDLR overexpression by using an anti-LRP1 antibody approach coupled with the BEI method (Fig. S3). We speculate that the reduction in apoE concentration as a result of LDLR overexpression allows free Aβ in the ISF to clear more rapidly across the BBB via LRP1 and other receptors, given that apoE has been suggested to impair Aβ clearance (15, 43). Further studies are needed to elaborate the putative cross-talk between LDLR and other apoE receptors that governs ISF Aβ elimination from the brain. Moreover, though our data revealed only subtle trends toward greater Aβ degradation as a result of LDLR overexpression, we cannot rule out a role for LDLR in mediating Aβ degradation within particular cell types (44), the magnitude of which may have been too subtle to detect in whole-brain homogenates. Conditional deletion strategies targeting LDLR expression within particular CNS cell types will be useful to address these possibilities. Though the current study focused on murine apoE, our results demonstrate a role for LDLR in BBB-mediated Aβ clearance, warranting further investigation into the contribution of this clearance pathway to apoE isoform-dependent Aβ clearance. This regulation may be especially relevant given that the affinity of apoE for LDLR is related to apoE isoform (45, 46). Given that haploinsufficiency of either apoE3 or apoE4 leads to reduced amyloid burden (25, 47), we would expect that reducing apoE levels by LDLR overexpression may result in enhanced Aβ clearance across the BBB in the context of either apoE3 or apoE4, and that increasing apoE expression would impair clearance. Future experiments to assess this possibility will be useful as apoE-reducing strategies are considered for AD treatment and prevention.

Although an established method to examine different components of Aβ elimination, the BEI method uses a kinetic model that cannot perfectly describe the complex physiology of the BBB and other routes of efflux. Moreover, the rapid and widespread distribution of Aβ species throughout various peripheral spaces and compartments makes following the fate of radiolabeled Aβ into the blood difficult. Acknowledging these limitations, we sought to verify our observations with an independent method to assess brain-to-blood Aβ elimination. The rapid degradation of Aβ once it enters the blood from brain precludes direct and reliable measurement of its influx rate by conventional means (27, 28). Thus, we reasoned that an anti-Aβ antibody would bind CNS-derived Aβ within the blood, preventing its degradation to allow direct measurement of its appearance rate. We previously hypothesized that anti-Aβ antibody treatment in the periphery leads to a rapid rise in plasma Aβ, in part by altering the efflux of Aβ from the brain (30, 32). In the current study, our microdialysis results suggest that peripheral administration of the HJ5.1 (anti-Aβ_13–28) antibody does not alter the metabolism of Aβ within the brain in the acute phase (5 h) during which we analyzed Aβ influx into the circulation. Though it is possible that over longer periods of time (days to weeks), certain anti-Aβ antibodies may alter Aβ metabolism in the brain, this possibility was not assessed in these experiments. A recent study suggested that the anti-Aβ antibody, m266, alters Aβ metabolism in the CNS by entering the brain and sequestering soluble Aβ (49). The small fraction of antibody that entered the brain in our study did not alter Aβ levels in the ISF over a very short timeframe, perhaps a reflection of its lower affinity for Aβ compared with the m266 antibody. Alternatively, the microdialysis technique may be insensitive to identifying the initial altered equilibrium changes in ISF Aβ. Importantly, our plasma accumulation results were consistent with results obtained using the BEI method (Fig. 1), further validating the BEI method as a useful technique to assess the contribution of different clearance components in overall Aβ clearance from the brain. Provided a suitable antibody is available that does not significantly alter brain Aβ metabolism, the plasma accumulation technique we report herein may be useful to screen drugs targeting Aβ elimination from brain to blood, while also serving as a useful tool for probing the biology of brain apoE receptors and their role in Aβ metabolism.

Our findings that LDLR regulates BBB-mediated Aβ clearance provide rationale for targeting apoE receptors in the brain, and specifically in brain endothelial cells, as an additional means to reducing Aβ accumulation. Recent studies have suggested that targeting apoE-mediated Aβ clearance may be an efficacious therapeutic strategy for reducing Aβ accumulation (50, 51). Given that LDLR has very few identified ligands compared with other apoE receptors (34), strategies aimed at modulating LDLR expression will likely be relatively specific to Aβ/apoE metabolism, presenting innovative avenues for AD prevention and treatment.

Materials and Methods

Supplementary Material

Acknowledgments

Footnotes

References