Differences in amyloid-β clearance across mouse and human blood-brain barrier models: Kinetic analysis and mechanistic modeling

Abstract

Alzheimer’s disease (AD) has a characteristic hallmark of amyloid-β (Aβ) accumulation in the brain. This accumulation of Aβ has been related to its faulty cerebral clearance. Indeed, preclinical studies that used mice to investigate Aβ clearance showed that efflux across blood-brain barrier (BBB) and brain degradation mediate efficient Aβ clearance. However, the contribution of each process to Aβ clearance remains unclear. Moreover, it is still uncertain how species differences between mouse and human could affect Aβ clearance. Here, a modified form of the brain efflux index method was used to estimate the contribution of BBB and brain degradation to Aβ clearance from the brain of wild type mice. We estimated that 62% of intracerebrally injected ¹²⁵I-Aβ₄₀ is cleared across BBB while 38% is cleared by brain degradation. Furthermore, in vitro and in silico studies were performed to compare Aβ clearance between mouse and human BBB models. Kinetic studies for Aβ₄₀ disposition in bEnd3 and hCMEC/D3 cells, representative in vitro mouse and human BBB models, respectively, demonstrated 30-fold higher rate of ¹²⁵I-Aβ₄₀ uptake and 15-fold higher rate of degradation by bEnd3 compared to hCMEC/D3 cells. Expression studies showed both cells to express different levels of P-glycoprotein and RAGE, while LRP1 levels were comparable. Finally, we established a mechanistic model, which could successfully predict cellular levels of ¹²⁵I-Aβ₄₀ and the rate of each process. Established mechanistic model suggested significantly higher rates of Aβ uptake and degradation in bEnd3 cells as rationale for the observed differences in ¹²⁵I-Aβ₄₀ disposition between mouse and human BBB models. In conclusion, current study demonstrates the important role of BBB in the clearance of Aβ from the brain. Moreover, it provides insight into the differences between mouse and human BBB with regards to Aβ clearance and offer, for the first time, a mathematical model that describes Aβ clearance across BBB.

1. Introduction

Amyloid-β peptides (Aβ) are by-products of neuronal metabolism that have been linked to the pathogenesis of Alzheimer disease (AD) (Selkoe, 1993). Cerebral levels of these peptides are regulated by their production rate from proteolytic degradation of amyloid precursor protein (APP), influx from plasma that is mediated mainly by receptor for advanced glycation end product (RAGE) (Deane et al., 2003), and by their clearance from the brain (Sommer, 2002). In AD, the rate of cerebral accumulation of Aβ peptides, mainly Aβ₄₀ and Aβ₄₂, is accelerated resulting in toxic aggregates of different sizes ranging from soluble oligomers to insoluble plaques (Jan et al., 2010). In very rare cases of AD (familial AD), Aβ accumulation is related to its overproduction (Citron et al., 1992). However, mounting evidence suggests that Aβ accumulation in the brain of late-onset “sporadic” AD patients and in some cases of familial AD is related to its impaired clearance from brain (Deane and Zlokovic, 2007). Moreover, a previous study has shown that late-onset AD is associated with 30% decrease in the clearance of Aβ while the production rate did not differ between control and AD individuals (Mawuenyega et al., 2010). Clearance of Aβ from the brain takes place by three pathways, transport across the blood-brain barrier (BBB) (Deane et al., 2009), degradation in the brain tissue (Iwata et al., 2000), and bulk flow of cerebrospinal fluid (CSF) (Silverberg et al., 2003). It is estimated that the clearance rate of Aβ₄₀ across BBB is 6-fold higher than its clearance rate through bulk flow of CSF (Bell et al., 2007); however, the relative contribution of brain degradation was not determined.

Clearance of Aβ₄₀ across the BBB has been extensively studied over the past decade where many contributing transporters/receptors at the BBB have been identified (Deane et al., 2009). In addition, accelerated cerebral accumulation of Aβ₄₀ due to impaired clearance across the BBB has also been demonstrated to significantly affect its deposition and plaque formation in the brain of AD patients (Bell and Zlokovic, 2009). The main infrastructure of the BBB that regulates Aβ₄₀ clearance is the endothelial cells lining the brain capillaries. Endothelial cells are connected to each other by strong tight junctions and they are anchored to a continuous basement membrane that is supported by perivascular end-feet of the astrocytes forming a physical barrier for the movement of compounds (Ballabh et al., 2004). Given the important contribution of endothelial cells to the function of the BBB, transport of Aβ₄₀ across these cells is a crucial step in the clearance of Aβ₄₀. As a peptide, Aβ₄₀ has poor passive membrane permeability and it depends on transport system to pass across the endothelial cells of BBB (Banks et al., 2003). Aβ₄₀ is known to be a substrate for many receptors and transporters at the BBB such as low density lipoprotein receptor-related protein-1 (LRP1) and P-glycoprotein (P-gp) that contribute to its clearance from the brain to blood (Cirrito et al., 2005; Shibata et al., 2000), and RAGE that is involved in Aβ₄₀ influx from blood to brain (Deane et al., 2003).

Mouse is a widely used animal model to study cerebral clearance of Aβ₄₀. Available studies, independently (Barten et al., 2005; Cirrito et al., 2003; Mawuenyega et al., 2010), demonstrated a significant difference in Aβ₄₀ clearance rate between mice and humans. In cognitively normal individuals, about 7% of Aβ₄₀ is cleared from the brain per hour (corresponding half-life of ~10 h) and this percent decreased to 5.2% in patients with AD (corresponding half-life of ~13 h) (Mawuenyega et al., 2010). On the other hand, the half-life of human Aβ₄₀ in mouse brain ranges from 0.63 h in Tg2576 mice to 2 h in PDAPP mice (Barten et al., 2005; Cirrito et al., 2003). Thus, the rate of Aβ₄₀ clearance from mouse brain is at least 5-fold higher than human brain. However, differences in the rate of Aβ₄₀ clearance by degradation and/or transport across the BBB between mouse and human remain unknown.

Evaluation of kinetic parameters of Aβ₄₀ uptake, efflux and degradation by endothelial cells is essential to understand key steps involved in Aβ₄₀ clearance across the BBB. To date, available studies have modeled Aβ₄₀ clearance in vivo across the mouse BBB to calculate Aβ₄₀ kinetic parameters (Bell et al., 2007; Kandimalla et al., 2005; Shibata et al., 2000). However, the calculated parameters represented overall clearance across BBB and do not distinguish between clearance via transport or degradation. To estimate the kinetic parameters of the several processes involved in Aβ₄₀ clearance including uptake, efflux and degradation across the BBB, in vitro models of BBB are preferred. Moreover, as Aβ₄₀ disposition in brain endothelial cells involves several processes that simultaneously take place, it is difficult to use a conventional Michaelis-Menten approach to estimate Aβ₄₀ disposition kinetic parameters by these cells. Therefore, utilization of mechanistic models should provide a better approach to estimate and understand Aβ₄₀ kinetic disposition in endothelial cells (Poirier et al., 2008).

Accordingly and given the important role of the BBB in the clearance of Aβ₄₀, we hypothesized that BBB plays the major role in brain Aβ₄₀ clearance compared to Aβ₄₀ brain degradation, and that the higher clearance rate of Aβ₄₀ from mouse brain compared to human brain is due to differences in Aβ₄₀ clearance across the BBB. Three aims were set to examine this hypothesis, 1) investigate the in vivo contribution of the BBB to Aβ₄₀ clearance relative to its brain degradation, 2) compare mechanisms and rate of Aβ₄₀ clearance between mouse and human BBB in vitro models, and 3) establish a mechanistic kinetic model that describe differences in Aβ₄₀ disposition between mouse and human endothelial cells.

2. Material and Methods

2.1. Animals

C57BL/6 male mice (6–7 weeks old; Harlan Laboratories, Houston, TX) were kept under standard environmental conditions (22°C, 35% relative humidity, 12 h dark/light cycle) with free access to tap water and standard rodent food. All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Louisiana at Monroe. All surgical procedures were consistent with the IACUC policies and procedures.

2.2. Brain efflux index (BEI) study

In vivo Aβ₄₀ clearance was investigated using the BEI method as described previously (Qosa et al., 2012). In brief, a stainless steel guide cannula was implanted stereotaxically into the right caudate nucleus of mice that had been anesthetized with intraperitoneal xylazine and ketamine (20 and 125 mg/kg, respectively) (Henry Schein Inc., NY). After 12 h recovery period, animals were re-anesthetized and tracer fluid (0.5 μl) containing ¹²⁵I-Aβ₄₀ (30 nM, PerkinElmer, MA) and ¹⁴C-inulin (0.02 μCi, American Radiolabeled Chemicals, MO) prepared in extracellular fluid buffer (ECF) was administered. Thirty minutes post ¹²⁵I-Aβ₄₀ injection (Cirrito et al., 2005; Shibata et al., 2000); brain tissues were rapidly collected for ¹²⁵I-Aβ₄₀ analysis. To characterize role of P-gp and LRP1, 0.5 μl of ECF containing valspodar (40 μM; XenoTech, KS), a well-established P-gp inhibitor, or RAP (1 μM; Oxford Biomedical Research, MI), an LRP1 inhibitor, were intracerebrally administered 5 min prior to ¹²⁵I-Aβ₄₀ injection.

2.3. Calculation of in vivo ¹²⁵I-Aβ₄₀ clearance

Calculations of ¹²⁵I-Aβ₄₀ clearance were performed as described previously (Qosa et al., 2012; Shibata et al., 2000). Using the trichloroacetic acid (TCA) precipitation assay intact (precipitate) and degraded (supernatant) ¹²⁵I-Aβ₄₀ were determined in brain tissue using a Wallac 1470 Wizard Gamma Counter (PerkinElmer Inc., Waltham, MA). ¹⁴C-inulin in precipitate and supernatant were also determined using a Wallac 1414 WinSpectral Counter (PerkinElmer Inc.). To calculate the total clearance of ¹²⁵I-Aβ₄₀ from the brain, a modified equation of the BEI method (Qosa et al., 2012) was used to calculate brain clearance index (BCI_Total(%)), i.e. clearance across BBB and brain degradation (equation 1). Clearance of ¹²⁵I-Aβ₄₀ across BBB (BCI_BBB(%)) and brain degradation (BCI_Degradation(%)) were defined by equations 2 and 3, respectively.

$${\mathit{BCI}}_{\mathit{Total}}(\%)=100-\left[\frac{\left(\frac{\text{Amount}\text{of}\text{intact}{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{in}\text{the}\text{brain}}{\text{Amount}\text{of}{}^{14}\mathrm{C}-\text{inulin}\text{in}\text{the}\text{brain}}\right)}{\left(\frac{\text{Amount}\text{of}\text{intact}{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{injected}\text{into}\text{the}\text{brain}}{\text{Amount}\text{of}{}^{14}\mathrm{C}-\text{inulin}\text{injected}}\right)}\times 100\right]$$ (Eq. 1)

$${\mathit{BCI}}_{\mathit{BBB}}(\%)=100-\left[\frac{\left(\frac{\text{Amount}\text{of}\text{total}{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{in}\text{the}\text{brain}(\text{intact}\text{and}\text{degraded})}{\text{Amount}\text{of}{}^{14}\mathrm{C}-\text{inulin}\text{in}\text{the}\text{brain}}\right)}{\left(\frac{\text{Amount}\text{of}\text{intact}{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{injected}\text{into}\text{the}\text{brain}}{\text{Amount}\text{of}{}^{14}\mathrm{C}-\text{inulin}\text{injected}}\right)}\times 100\right]$$ (Eq. 2)

$${\mathit{BCI}}_{\mathit{Degradation}}(\mathit{\%})={\mathit{BCI}}_{\mathit{Total}}(\mathit{\%})-{\mathit{BCI}}_{\mathit{BBB}}(\mathit{\%})$$ (Eq. 3)

2.4. Cell Culture

The mouse (bEnd3) and human (hCMEC/D3) brain endothelial cells were used as representative models for mouse and human BBB, respectively. bEnd3, passage 25–35, were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), penicillin G (100 units/ml) and streptomycin (100 μg/ml). hCMEC/D3, passage 25–35, were cultured in EBM-2 medium supplemented with 1 ng/ml human basic fibroblast growth factor (Sigma-Aldrich, MO), 10 mM HEPES, 1% chemically defined lipid concentrate (Gibco, NY), 5 μg/ml ascorbic acid, 1.4 μM hydrocortisone, 1% penicillin-streptomycin and 5% of heat-inactivated FBS “gold”. Cultures were maintained in a humidified atmosphere (5%CO₂/ 95% air) at 37°C.

2.5. Uptake and degradation of ¹²⁵I-Aβ₄₀ by bEnd3 and hCMEC/D3 cells

To study uptake and degradation of ¹²⁵I-Aβ₄₀, cells were seeded at a density of 5 × 10⁴ per well in 24-well plate, and maintained for 3 to 5 days. Uptake and degradation of ¹²⁵I-Aβ₄₀ were evaluated for 15 min and 12 h in bEnd3 and hCMEC/D3 cells, respectively. These time points were determined from ¹²⁵I-Aβ₄₀ depletion studies. In addition, initial studies to confirm the indivisible ¹²⁵I-label from Aβ₄₀, and the integrity of ¹²⁵I-Aβ₄₀ over different time points were performed by incubation of 0.1 nM of ¹²⁵I-Aβ₄₀ in media without cells, and by 16% bis-tris gel electrophoresis. The results demonstrated insignificant dissociation of ¹²⁵I from Aβ₄₀ during the investigated time points (Supplementary figure 1a), and the absence of dimers or higher oligomer formation at 12 h, the duration of the uptake study (Supplementary figure 1b). Uptake studies were initiated by addition of fresh warm medium containing 0.1 nM of ¹²⁵I-Aβ₄₀. For inhibition studies, RAP (1 μM), RAGE-IgG (5 μg/ml, Santa Cruz Biotechnology Inc., CA, USA) and valspodar (5 μM) were pre- and co-incubated with ¹²⁵I-Aβ₄₀ at 37°C. At the end of incubation period, media were collected and cells were washed twice with ice-cold PBS containing 0.2% BSA to minimize non-specific binding of ¹²⁵I-Aβ₄₀, and once with warm PBS. Cells were dissolved in 200 μl RIPA buffer containing complete mammalian protease inhibitors for 20 min on ice with shaking. Intact and degraded ¹²⁵I-Aβ₄₀ were measured in media and cell lysates as described below. Total amount of protein in each sample was determined using BCA protein assays. Net uptake of ¹²⁵I-Aβ₄₀ by the cells was normalized to the total amount of cellular protein.

2.6. ¹²⁵I-Aβ₄₀ analysis

For all in vitro experiments, intact and degraded ¹²⁵I-Aβ₄₀ were measured in the medium and cells. Total ¹²⁵I-Aβ₄₀ was determined by counting sample radioactivity. Degraded ¹²⁵I-Aβ₄₀ was measured in the supernatant following precipitation with TCA (20%) in 1:1 ratio (v/v) (Qosa et al., 2012). The intact fraction was calculated by subtracting degraded ¹²⁵I-Aβ₄₀ from total ¹²⁵I-Aβ₄₀. ELISA kits (SensoLyte Anti-Human beta-Amyloid 1–40, Anaspec, CA) were used to calculate exact starting concentrations for all in vitro experiments, and to confirm ¹²⁵I-Aβ₄₀ degradation obtained from TCA assay. For ELISA analysis, 100 μl of each sample were used to measure intact ¹²⁵I-Aβ_40. All cpm values of ¹²⁵I-Aβ₄₀ were converted to fmole based on the specific activity of ¹²⁵I-Aβ₄₀, γ-counter efficiency and fmole versus cpm calibration curve obtained by ELISA assay (data not show).

2.7. Localization of Aβ₄₀-HiLyte Fluor into acidic organelles

For imaging studies, cells were plated on 8-well chambered glass slides (LAB-Tek, Nalge Nunc, NY) at 12,000 cells/cm², and allowed to grow overnight. bEnd3 and hCMEC/D cells were incubated with Aβ₄₀-HiLyte Fluor (1 nM, Alexa-Fluor, Anaspec, CA) for 15 min and 12 h, respectively. To track Aβ₄₀-HiLyte Fluor in acidic organelles, LysoTraker Red (Invitrogen, OR, USA), a weakly basic fluorophore that selectively accumulates in acidic organelles of the endosomal/ lysosomal pathway, was added at 75 nM to the incubation media 45 min prior to the end of Aβ₄₀-HiLyte Fluor incubation in hCMEC/D cells, and 30 min prior to the addition of Aβ₄₀-HiLyte Fluor and during incubation in bEnd3 cells. At the end of the incubation periods, cells were washed and fixed with 4% formaldehyde for 5 min on ice. Cells were mounted with DAPI. Images for Aβ₄₀-HiLyte Fluor and LysoTracker Red were captured using a Zeiss LSM 5 Pascal confocal microscope equipped with 488 nm Argon laser and 543 nm HeNe Laser using a 63X oil immersion objective lens with numerical aperture = 1.4 (Carl Zeiss MicroImaging LLC, Thornwood, NY).

2.8. ¹²⁵I-Aβ₄₀ transport across bEnd3 and hCMEC/D3 cells

For transport studies, transwell polyester membrane inserts, 6.5 mm diameter with 0.4 μm pores (Corning, NY), were coated with collagen (150 μg/ml). Cells were plated onto coated inserts at a seeding density of 50,000 cells/cm², medium was changed every other day. Trans-epithelial electrical resistance (TEER) was measured using an EVOM epithelial volt-ohmmeter with STX2 electrodes (World Precision Instruments, FL). According to the value of TEER (~35 Ω.cm²) both cell lines were used for ¹²⁵I-Aβ₄₀ transport experiments on day 6. Apical to basolateral (A→B) transport studies were initiated by addition of 0.1 nM ¹²⁵I-Aβ₄₀ and 0.05 mM ¹⁴C-inulin to the apical compartment, while basolateral to apical (B→A) transport studies were initiated by addition of 0.1 nM ¹²⁵I-Aβ₄₀ and 0.05 mM ¹⁴C-inulin to the basolateral compartment. At the end of incubation period (30 min for bEnd3 and 12 h for hCMEC/D3 cells), media in both compartments and cells were separately collected for ¹²⁵I-Aβ₄₀ analysis and inulin measurement. The clearance quotients of A→B (¹²⁵I-Aβ₄₀ CQ_A→B) and B→A (¹²⁵I-Aβ₄₀ CQ_B→A) transports were calculated using the following equations (Nazer et al., 2008):

$${}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}{\text{CQ}}_{\mathrm{A}\to \mathrm{B}}=\frac{\left(\frac{{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{in}\text{basolateral}\text{compartment}}{{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{total}}\right)}{\left(\frac{{}^{14}\mathrm{C}-\text{inulin}\text{in}\text{basolateral}\text{compartment}}{{}^{14}\mathrm{C}-\text{inulin}\text{total}}\right)}$$ (Eq. 4)

$${}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}{\text{CQ}}_{\mathrm{B}\to \mathrm{A}}=\frac{\left(\frac{{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{in}\text{apical}\text{compartment}}{{}^{125}\mathrm{I}-\mathrm{A}{\beta }_{40}\text{total}}\right)}{\left(\frac{{}^{14}\mathrm{C}-\text{inulin}\text{in}\text{apical}\text{compartment}}{{}^{14}\mathrm{C}-\text{inulin}\text{total}}\right)}$$ (Eq. 5)

where ¹²⁵I-Aβ₄₀ total is the total intact cpm in the apical and basolateral compartments, as well as cpm remaining in cells. ¹⁴C-inulin total is the total inulin dpm in the apical and basolateral compartments. To evaluate the effect of different inhibitors on ¹²⁵I-Aβ₄₀ transport and degradation across cell monolayer valspodar (5 μM), RAP (1 μM), and RAGE-IgG (5 μg/ml) were pre- and co- incubated with ¹²⁵I-Aβ₄₀.

2.9. Western blotting

For western blot analysis of P-gp, LRP1 and RAGE in bEnd3 and hCMEC/D3 cells, 25 μg of cellular protein was resolved on 7.5% SDS-polyacrylamide gels and transferred onto nitrocellulose membrane. Membranes were blocked with 2% BSA and incubated overnight with monoclonal antibodies for LRP1 (Calbiochem, NJ), P-gp (C-219, Covance Research Products, MA), RAGE (N-16) or β-actin (C-11) (Santa Cruz) at dilutions 1:200, 1:1500, 1:200 and 1:3000, respectively. For protein detection, the membranes were washed and incubated with HRP-labeled secondary IgG antibody for LRP1 and P-gp (anti-mouse) and RAGE and β-actin (anti-goat) (Santa Cruz) at 1:5000 dilutions. The bands were visualized using a SuperSignal West Femto detection kit (Thermo Scientific, IL). Quantitative analysis of the immunoreactive bands was performed using a GeneSnap luminescent image analyzer (Scientific Resources Southwest, TX) and band intensities were measured by densitometric analysis.

2.10. Modeling of ¹²⁵I-Aβ₄₀ disposition in bEnd3 and hCMEC/D3 cells

Intracellular concentration of intact ¹²⁵I-Aβ₄₀ is controlled by three active processes; uptake, efflux and degradation. To establish a mechanistic model that describes these processes, cells were grown in 24-well plates and maintained as described above. Experiments were initiated by addition of fresh warm media containing ¹²⁵I-Aβ₄₀. bEnd3 cells were incubated at 37°C for 5, 15, 30, 60 and 120 min with different ¹²⁵I-Aβ₄₀ concentrations of 0.4, 0.8, 1.5, 3, 8 nM, while hCMEC/D3 cells were incubated at 37°C for 0.5, 1, 6, 12 and 24 h with ¹²⁵I-Aβ₄₀ concentrations of 0.8, 1.5, 3, 6, 12 nM. At the end of each incubation period, media were removed and cells were washed and lysed. Concentrations of intact and degraded ¹²⁵I-Aβ₄₀ were measured in both medium and cells. To calculate efflux parameters, bEnd3 and hCMEC/D3 cells were preloaded with different concentrations of ¹²⁵I-Aβ₄₀, as above, for 15 min and 12 h, respectively. At the end of the incubation period, culture media were removed, cells were washed and fresh warm media were added to the cells. Cells were incubated for 1, 3, and 5 min at 37°C. After that, media were collected and cells were washed and lysed. ¹²⁵I-Aβ₄₀ in media and cells were analyzed and efflux rates (fmole/min) of intact ¹²⁵I-Aβ₄₀ were plotted against intracelluar ¹²⁵I-Aβ₄₀ concentrations (nM). Efflux parameters were used as fixed values in building the model to simplify modeling procedure by decreasing the number of parameters to be estimated. For intracellular volume estimation (~ 0.18 μl), the protein content of endothelial cells was converted to volume using a previously published constant of 3 μl/mg protein (Brillault et al., 2008). At all examined concentrations, time versus intact ¹²⁵I-Aβ₄₀ concentration in intracellular compartment was modeled according to equation 6:

$$\frac{{\text{dC}}_{\text{cell}}}{\text{dt}}=\frac{{\mathrm{V}}_{max,\text{uptake}}\times {\mathrm{C}}_{\text{med}}}{{\mathrm{K}}_{\mathrm{m},\text{uptake}}+{\mathrm{C}}_{\text{med}}}-\frac{{\mathrm{V}}_{max,\text{efflux}}\times {\mathrm{C}}_{\text{cell}}}{{\mathrm{K}}_{max,\text{efflux}}+{\mathrm{C}}_{\text{cell}}}-\text{CL}$$ (Eq. 6)

and the time versus intact ¹²⁵I-Aβ₄₀ concentration in medium compartment was modeled according to equation 7:

$$\frac{{\text{dC}}_{\text{med}}}{\text{dt}}=-\frac{{\mathrm{V}}_{max,\text{uptake}}\times {\mathrm{C}}_{\text{med}}}{{\mathrm{K}}_{\mathrm{m},\text{uptake}}+{\mathrm{C}}_{\text{med}}}+\frac{{\mathrm{V}}_{max,\text{efflux}}\times {\mathrm{C}}_{\text{cell}}}{{\mathrm{K}}_{max,\text{efflux}}+{\mathrm{C}}_{\text{cell}}}$$ (Eq. 7)

whereas, change in the concentration of degraded ¹²⁵I-Aβ₄₀ in cells overtime was defined by equation 8:

$$\frac{{\text{dD}}_{\text{cell}}}{\text{dt}}=\text{CL}\times {\mathrm{C}}_{\text{cell}}$$ (Eq. 8)

In all equations, V_max was expressed in fmole/min, K_m in nM, and CL in nl/min. The mechanistic in vitro model was implemented in MatLab version 7.10 (MathWorks, MA, USA) to simultaneously estimate V_max, K_m and CL of ¹²⁵I-Aβ₄₀ uptake, while V_max and K_m of the efflux were fed to the model as fixed values that were estimated from the efflux studies. All data points were fitted in a single step using equation 6 with triplicates for each of the five time points for each concentration. SDs were calculated for all parameters obtained from individual experiments and coefficients of variation (CVs) for each parameter were subsequently calculated and expressed as a percentage to assess the quality of the parameter (Poirier et al., 2008).

2.11. Statistical analysis

Unless otherwise indicated, the data were expressed as mean±SEM. The experimental results were statistically analyzed for significant difference using two-tailed Student’s t-test for 2 two groups, and one-way analysis of variance (ANOVA) for more than two groups analysis. Values of P<0.05 were considered statistically significant.

3. Results

3.1. In vivo clearance of ¹²⁵I-Aβ₄₀ across BBB

The first objective of current study was to estimate the specific contribution of BBB to the clearance of ¹²⁵I-Aβ₄₀ from the brains of C57BL/6 mice. BCI method, a modified method for BEI, was used to calculate total clearance, i.e. efflux across BBB and clearance by degradation of ¹²⁵I-Aβ₄₀ in mouse brain. BCI_Total(%) demonstrated 58.7±3.9% of administered ¹²⁵I-Aβ₄₀ was cleared from mouse brain 30 min following ¹²⁵I-Aβ₄₀ microinjection (Figure 1A). BCI_BBB(%) showed that 35.6±3.3% of injected ¹²⁵I-Aβ₄₀, i.e. 62% of the BCI_Total(%), was cleared across the BBB (Figure 1B). On the other hand, 23.1±2.3% of injected ¹²⁵I-Aβ₄₀, i.e. 38% of BCI_Total(%), was cleared by brain degradation (Figure 1C). To elucidate the specific contribution of P-gp and LRP1 to BBB clearance and brain degradation of ¹²⁵I-Aβ₄₀, BCI_Total(%), BCI_BBB(%) and BCI_Degradation(%) of ¹²⁵I-Aβ₄₀ were determined in the presence of specific P-gp and LRP1 inhibitors. P-gp and LRP1 inhibition caused 25% (43.9±4.8%, P<0.05) and 49% (30.1±1.7, P<0.01) reduction, respectively, in ¹²⁵I-Aβ₄₀ BCI_Total(%) (Figure 1A). Moreover, our results showed that although P-gp and LRP1 inhibition significantly decreased ¹²⁵I-Aβ₄₀ BCI_BBB(%) to 18.3±1.5% (P<0.01) and 22.9±1.7 (P<0.05), respectively, only inhibition of LRP1 decreased ¹²⁵I-Aβ₄₀ BCI_Degradation(%) to 7.2±0.4% (P<0.001; Figures 1B and C). Collectively, these results confirm the significant contribution of BBB to cerebral clearance of ¹²⁵I-Aβ₄₀ compared to brain degradation, and signify the importance of P-gp and LRP1 at the BBB in this process.

Figure 1. In vivo clearance of ¹²⁵I-Aβ₄₀ from brain of C57BL/6 mice.

Total brain clearance (BCI_Total(%)) of ¹²⁵I-Aβ₄₀ in control (CNTRL), with P-gp inhibitor valspodar (VALSP) or LRP1 inhibitor RAP (A), BBB clearance (BCI_BBB(%)) of ¹²⁵I-Aβ₄₀ in control and inhibitor groups (B), clearance of ¹²⁵I-Aβ₄₀ by brain degradation (BCI_Degradation(%)) in control and inhibitor groups (C). Data represent mean±SEM for n=4; * P<0.05, ** P<0.01 and *** P<0.001. NS is not significant.

3.2. In vitro uptake of ¹²⁵I-Aβ₄₀ by bEnd3 and hCMEC/D3 cells

Results from preliminary depletion studies showed that while 2 h incubation time was sufficient for complete depletion of ¹²⁵I-Aβ₄₀ from the medium of bEnd3 cells, more than 24 h incubation time was required with hCMEC/D3 cells (Figure 2). Consistent with the depletion study, bEnd3 cells showed very rapid uptake and degradation of ¹²⁵I-Aβ₄₀ compared to hCMEC/D3 cells. As shown in Figure 3A, bEnd3 cellular level of intact ¹²⁵I-Aβ₄₀ was 0.27±0.02 fmole/mg after 15 min incubation, while 12 h was needed for hCMEC/D3 to reach a similar cellular level of intact ¹²⁵I-Aβ₄₀ (0.26±0.01 fmole/mg). The degradation of ¹²⁵I-Aβ₄₀ by bEnd3 and hCMEC/D3, measured in the medium, reached 73.7±2.0 and 65.7±1.8% after 15 min and 12 h of ¹²⁵I-Aβ₄₀ addition, respectively (Figure 3B). Inhibition of LRP1 by RAP significantly decreased the cellular level of intact ¹²⁵I-Aβ₄₀ in bEnd3 and hCMEC/D3 cells by 67% and 42%, respectively (P<0.001, Figure 3A). The degradation of ¹²⁵I-Aβ₄₀ in bEnd3 and hCMEC/D3 was also decreased by 24% and 28%, respectively, by RAP (P<0.01, Figure 3B). On the other hand, inhibition of RAGE significantly decreased the cellular level of intact ¹²⁵I-Aβ₄₀ by 75% in both cells (P<0.001, Figure 3A), while ¹²⁵I-Aβ₄₀ degradation was not altered by RAGE inhibition in both cells (P>0.05, Figure 3B). Inhibition of P-gp by valspodar did not cause any significant change in ¹²⁵I-Aβ₄₀ accumulation in bEnd3 cells (P>0.05, Figure 3A), whereas this effect was obvious and significant in hCMEC/D3 where P-gp inhibition caused 15% increase in cellular levels of intact ¹²⁵I-Aβ₄₀ (P<0.05, Figure 3A). As expected, P-gp inhibition has no effect on ¹²⁵I-Aβ₄₀ degradation in both cells (P>0.05, Figure 3B). In order to explain the differences in ¹²⁵I-Aβ₄₀ disposition between the two cell lines, expression studies were performed by Western blotting. Both cell lines showed comparable levels of LRP1, significantly higher RAGE expression in bEnd3 compared to hCMEC/D3, and higher expression of P-gp in hCMEC/D3 compared to bEnd3 cells (Figure 4).

Figure 2. Depletion studies of ¹²⁵I-Aβ₄₀ in bEnd3 and hCMEC/D3 cells over 24 h.

Media containing 1.5 nM ¹²⁵I-Aβ₄₀ was added to cells in 24-well plate. Media were collected after 5, 15, 30 min and 1, 6, 12, 24 h to measure intact ¹²⁵I-Aβ₄₀ remaining in media by ELISA and radioactivity analyses. Figure inside inset demonstrates the linear range of depletion (0–0.5 h) observed with bEnd3.

Figure 3. In vitro uptake and degradation of ¹²⁵I-Aβ₄₀ by bEnd3 and hCMEC/D3 cells.

A) In vitro uptake of intact ¹²⁵I-Aβ₄₀ by bEnd3 and hCMEC/D3 cells. Cellular level of intact ¹²⁵I-Aβ₄₀ is expressed in fmole/mg protein following cells treatment with 0.1 nM ¹²⁵I-Aβ₄₀ without (control; CNTRL) and with RAGE (RAGE-IgG), or P-gp (valspodar; VALSP), or LRP1 (RAP) inhibitors for 15 min and 12 h in bEnd3 and hCMEC/D3, respectively. B) Percent of degraded ¹²⁵I-Aβ₄₀ in the media of bEnd3 and hCMEC/D3 cells following treatment with 0.1 nM of ¹²⁵I-Aβ_40, with or without inhibitor. Data represent mean±SEM from three independent experiments; *P<0.05, **P< 0.01 and *** P<0.001.

Figure 4. Expression of LRP1, RAGE and P-gp in bEnd3 and hCMEC/D3 cells.

A) Western blot analysis of LRP1, RAGE and P-gp protein expression in bEnd3 and hCMEC/D3 cells. B) Densitometry analyses showed similar expression level of LRP1 in two cells, higher RAGE expression in bEnd3 cells and higher P-gp expression in hCMEC/D3 cells. Data represent mean±SEM from three independent experiments; ** P<0.01, NS is not significant.

3.3. Accumulation of Aβ₄₀ in acidic organelles

While the localization of Aβ₄₀ in endosomal/lysosomal organells has been previously reported in bovine endothelial cells (Kandimalla et al., 2009), in this study, we have shown for the first time the endosomal/ lysosomal localization of Aβ₄₀ in human and mouse brain endothelial cells, a pathway that is associated with degradation preceded by peptide uptake. The cellular localization of Aβ₄₀ was studied by monitoring cellular accumulation of Aβ₄₀-HiLyte Fluor in the acidic organelles. Aβ₄₀-HiLyte Fluor showed a clear co-localization with LysoTracker Red in bEnd3 and hCMEC/D3 cells (Figure 5). Pearson’s correlation coefficients of Aβ₄₀-HiLyte Fluor co-localization with LysoTracker Red were about 0.64 in both cell lines, suggesting that uptaken Aβ₄₀ was directed to degradation via endosomal/lysosomal pathway.

Figure 5. Degradation of Aβ₄₀ within endosomal/lysosomal pathaway of bEnd3 and hCMEC/D3 cells.

Co-localization of intracellular Aβ₄₀-HiLyte Fluor and LysoTracker Red in endosomal/lysosomal pathway of brain endothelial cells (bEnd3 and hCMEC/D3) analyzed using confocal immunoFluorescence microscope. bEnd3 cells and hCMEC/D3 were incubated with 1 nM of Aβ₄₀-HiLyte Fluor for 15 min or 12 h; respectively. DAPI stains nuclei. Lysosomes were labeled with LysoTracker Red for 45 min before the end of study. The yellow color in the merged image represent co-localization of green (Aβ₄₀-HiLyte Fluor) and red (LysoTracker Red) fluorescence.

Additional experiments were performed to evaluate the contribution of specific ¹²⁵I-Aβ₄₀ degrading proteases to its total degradation relative to the lysosomal degradation. In these studies, the expression and activity of major Aβ₄₀ degrading enzymes including insulin-degrading enzyme (IDE) and neprilysin in bEnd3 and hCMEC/D3 cells were evaluated. Expression studies were achieved by Western blotting, and activity studies were examined using inhibition studies for each enzyme. The results demonstrated that IDE and neprilysin are expressed in both cell lines with significantly higher expression in bEnd3 (Supplementary figure 2a). In addition, while the inhibition of IDE by bacitracin (1 mg/mL) significantly decreased Aβ₄₀ degradation in both cells, its contribution to the total degradation was low (20% and 12% in bEnd3 and hCMEC/D3, respectively (Supplementary figure 2b). On the other hand, the inhibition of neprilysin by thiorphan (2 mM) did not show any significant effect on Aβ₄₀ degradation in both cells (Supplementary figure 2b). Neprilysin has low affinity for its physiological substrates (in the low mM range) (Shibata et al., 2000) and Aβ₄₀ concentrations used in our experiments were in the low nM range, thus it is more likely for Aβ₄₀ to bind its high affinity receptors that mediate its endocytosis and non-specific lysosomal degradation. Collectively, these findings demonstrate that both non-specific lysosomal degradation and specific protease dependent degradation are involved in Aβ₄₀ degradation by brain endothelial cells, however, with superior contribution of the non-specific lysosomal pathway.

3.4. ¹²⁵I-Aβ₄₀ transport across endothelial cells

Results of transport studies showed A→B transport of ¹²⁵I-Aβ₄₀ was about 2-fold higher across bEnd3 than hCMEC/D3 cells. CQ_A→B of intact ¹²⁵I-Aβ₄₀ was 0.62±0.06 and 0.32±0.02 across bEnd3 and hCMEC/D3 cells, respectively (P<0.01, Figure 6A). RAGE inhibition decreased A→B transport of ¹²⁵I-Aβ₄₀ by 50% in bEnd3 and 33% in hCMEC/D3 cells (P<0.01, Figure 6A). Consistent with the results of the uptake study, inhibition of P-gp increased transport of ¹²⁵I-Aβ₄₀ from A→B by 22% in hCMEC/D3 while no alteration in ¹²⁵I-Aβ₄₀ transport in bEnd3 was observed (Figure 6A). On the other hand, ¹²⁵I-Aβ₄₀ transport from B→A was more than 2-fold higher in hCMEC/D3 compared to bEnd3 (P<0.01, Figure 6B). This observation can be explained by the fact that at 30 min the concentration of intact ¹²⁵I-Aβ₄₀ available in bEnd3 for transport is lower than that at 12 h in hCMEC/D3, which is consistent with the rapid degradation of ¹²⁵I-Aβ₄₀ seen in the depletion studies (Figure 2). In agreement with the role of LRP1, RAP decreased intact ¹²⁵I-Aβ₄₀ transport from B→A direction by 42% in both cells (P<0.01, Figure 6B). In addition, consistent with rapid depletion in bEnd3, degradation of ¹²⁵I-Aβ₄₀ was significantly higher in bEnd3 than hCMEC/D3 in both directions (P<0.01, Figures 6C and D). Inhibition of RAGE, P-gp, or LRP1 did not affect ¹²⁵I-Aβ₄₀ degradation in both cell lines and in either direction (Figures 6C and D).

Figure 6. Transport of Aβ₄₀ across bEnd3 and hCMEC/D3 cell monolayers.

Apical to basolateral (A→B) and basolateral to apical (B→A) permeability and degradation of 0.1 nM ¹²⁵I-Aβ₄₀ across bEnd3 and hCMEC/D3 cells determined at 30 min and 12 h, respectively. Clearance quotient CQ_A→B and degradation of ¹²⁵I-Aβ₄₀ across bEnd3 and hCMEC/D3 cells in the absence (CNTRL) or presence of RAGE (RAGE-IgG) or P-gp (VALSP) inhibitors (A, C). CQ_B→A and degradation of ¹²⁵I-Aβ₄₀ across bEnd3 and hCMEC/D3 in the absence (CNTRL) or presence of LRP1 inhibitor (RAP) (B, D). Data represent mean±SEM from three independent experiments, * P<0.05, ** P<0.01 and *** P<0.001.

3.5. Mechanistic model for Aβ₄₀ cellular disposition

Initial uptake studies of ¹²⁵I-Aβ₄₀ (0.1 nM) over 2 and 24 h incubation in bEnd3 and hCMEC/D3, respectively, showed a saturable uptake process of ¹²⁵I-Aβ₄₀ by both cells, with significantly higher rate of uptake in bEnd3 cells (Supplementary figure 3). On the other hand, efflux studies demonstrated a saturable efflux process in both cells with similar rates (Supplementary figure 4). Degradation studies, however, demonstrated unsaturable intracellular degradation of ¹²⁵I-Aβ₄₀ in both cells as the percentage of degraded ¹²⁵I-Aβ₄₀ were constant (73.7±2.0 and 65.7±1.8% in bEnd3 and hCMEC/D3, respectively) as a function of concentration. High intracellular concentration of Aβ₄₀ and presence of multiple processes that work simultaneously necessitates the use of a mechanistic model approach to estimate kinetic parameters for each of these integrated processes. Kinetic parameter determination using a mechanistic model approach depends on measurement of ¹²⁵I-Aβ₄₀ uptake and degradation at multiple time points and for different ¹²⁵I-Aβ₄₀ concentrations at each time point. Based on our results of ¹²⁵I-Aβ₄₀ uptake and degradation in bEnd3 and hCMEC/D3 cells at different time points and ¹²⁵I-Aβ₄₀ concentrations, we established a mechanistic two-compartment model that describes uptake, efflux and degradation of ¹²⁵I-Aβ₄₀ by endothelial cells. The model structure consisted of saturable uptake and efflux processes in addition to linear intracellular degradation process (Figure 7A). Appropriate equations were fit to the mean ¹²⁵I-Aβ₄₀ accumulation in endothelial cells versus time data plotted in Figure 7 (B-E). Based on visual inspection of the residuals, coefficients of variation (CV%) of the recovered parameters and Akaike’s information criterion values, the model structure presented in Figure 7A provided the best description of the entire data set. This model scheme incorporated saturable uptake and efflux steps (K_m,uptake, V_m,uptake and K_m,efflux, V_m,efflux) and a linear intracellular degradation (CL). Estimates of kinetic parameters obtained from the model are shown in Table 1. The model recovered good estimates of all parameters. Consistent with the observed data, established model indicated a very rapid ¹²⁵I-Aβ₄₀ uptake in bEnd3 when compared to hCMEC/D3 cells with intrinsic uptake (V_max/K_m) from media of 35-fold higher in bEnd3. Moreover, this model showed 15-fold higher rate of intracellular degradation (CL) by bEnd3 cells. Estimated K_m values indicated that ¹²⁵I-Aβ₄₀ has higher affinity for uptake receptors in bEnd3 than in hCMEC/D3 while it has similar affinity for efflux transporters/receptors (Table 1). Similarly, uptake was more efficient in bEnd3 than hCMEC/D3 evident by the higher V_max for ¹²⁵I-Aβ₄₀ uptake in bEnd3 cells while the rate of efflux were comparable between the two cells.

Figure 7. Mechanistic modeling of Aβ₄₀ disposition by bEnd3 and hCMEC/D3 cells.

Schematic description of proposed mechanistic model that describes ¹²⁵I-Aβ₄₀ uptake, efflux and degradation by endothelial cells (A). This mechanistic model contains two-compartments corresponding to the incubation media (C_med, concentration of intact ¹²⁵I-Aβ₄₀ in media in nM), and intracellular space of the cells (C_cell, cellular concentration of intact ¹²⁵I-Aβ₄₀ in nM; D_cell, cellular concentration of degraded ¹²⁵I-Aβ₄₀ in nM). This model provides the optimal model structure for ¹²⁵I-Aβ₄₀ disposition in endothelial cells. Kinetic profiles of cellular levels of intact (B) and degraded (D) ¹²⁵I-Aβ₄₀ in bEnd3 cells obtained when measured in plated cells at 0.4 nM (●), 0.8 nM (▲), 1.5 nM (▼), 3 nM (◆) and 6 nM (■) over 2 h incubation period; and cellular levels of intact (C) and degraded (E) ¹²⁵I-Aβ₄₀ in hCMEC/D3 cells when measured in plated cells at 0.8 nM (●), 1.5 nM (▲), 3 nM (▼), 6 nM (◆) and 12 nM (■) over 24 h incubation period. Lines represent the predicted cellular level obtained from the mechanistic two-compartment model. Observed data points are mean of triplicate measurements.

Table 1.

Kinetic apparent parameters of ¹²⁵I-Aβ₄₀ disposition by hCMEC/D3 and bEnd3 cells estimated from mechanistic model.

Parameter	hCMEC/D3		bEnd3
	Estimate	CV%	Estimate	CV%
Km, uptake (nM)	49.1	0.1	19.8	14.3
Vmax, uptake (fmole/min)	4.0	0.01	56.4	3.9
Intrinsic uptake (nl/min)	81		2848
Km, efflux (nM)	7.5	17.4	7.6	22.7
Vmax, efflux (fmole/min)	0.12	12.2	0.14	17.1
Intrinsic efflux (nl/min)	16		18
CL (nl/min)	17	4.5	260	5

4. Discussion

Efficient brain clearance of Aβ₄₀ is important to prevent its brain accumulation (Mawuenyega et al., 2010). Therefore, quantitative assessment of the contribution of different clearance pathways is important to elucidate key factors that could significantly enhance brain Aβ₄₀ accumulation. Although Aβ₄₀ is less pathogenic than Aβ₄₂ due to its low propensity to aggregate (Jan et al., 2010; Sommer, 2002), Aβ₄₀ is found at several fold higher than Aβ₄₂ in the brain, and its accumulation disrupts the integrity of BBB (Gregory and Halliday, 2005). Therefore, it is important to investigate clearance of Aβ₄₀ across the BBB. Previous studies reported significant contribution of the BBB to Aβ₄₀ clearance with less contribution of CSF bulk flow; however, the role of brain degradation was not elucidated (Bell et al., 2007). In the current study, we estimated contribution of BBB and brain degradation to the total brain clearance of Aβ₄₀ in mouse brain using a modified form of the BEI method. The BEI method depends on measuring intact ¹²⁵I-Aβ₄₀ remained in the brain after specific time of intracerebrally injected ¹²⁵I-Aβ₄₀, which does not only consider clearance across the BBB but also brain degradation of ¹²⁵I-Aβ₄₀. The modified method of BEI allowed quantification of BBB relative contribution (62%) and brain degradation (38%) to the total clearance of brain Aβ₄₀, excluding clearance via CSF flow which was previously estimated to contribute by 10–15% of overall brain Aβ₄₀ clearance (Shibata et al., 2000). The relative contribution of these Aβ₄₀ clearance pathways from the brain is summarized in Fig. 8. Consistent with its function as Aβ₄₀ efflux transporter (Cirrito et al., 2005), inhibition of P-gp decreased the clearance of Aβ₄₀ across the BBB without altering its brain degradation, while LRP1 inhibition by RAP decreased both, clearance across the BBB and brain degradation of Aβ₄₀. These results are in agreement with brain distribution and function of LRP1. LRP1 is expressed in brain microvessels where it mediates intact Aβ₄₀ clearance via receptor-mediated transcytosis (Castellano et al., 2012; Shibata et al., 2000), and in neurons and brain vascular smooth muscle cells where it mediates endocytosis and degradation of Aβ₄₀ (Fuentealba et al., 2010; Kanekiyo et al., 2012). Low-density lipoprotein receptor (LDLR) family members have also been reported to mediate Aβ₄₀ degradation in particular cell types within the brain (Basak et al., 2012), and similar to LRP1, these LDLR members are subjected to RAP inhibition (Bu, 1998). Thus, reduction in brain Aβ₄₀ degradation observed following RAP intracerebral administration could be related to LRP1 inhibition and possibly other LDLRs expressed in different brain cellular components. However, at the BBB, it is possible that RAP primarily inhibited LRP1transport of Aβ₄₀ as proposed by a previous study where the inhibition of LRP1 at the mouse brain capillaries by RAP or a specific LRP1 antibody showed comparable reduction in Aβ₄₀ uptake and clearance suggesting RAP specific inhibition of LRP1 (Deane et al., 2004).

Figure 8. Schematic representation of Aβ₄₀ disposition in the brain.

Different cellular compartments in the brain contribute to Aβ₄₀ clearance. Endothelial cells express LRP1 which mediates Aβ₄₀ transport across the BBB from the brain to the blood by transcytosis (1 and 2), at the luminal side P-gp effluxes Aβ₄₀ to the blood (3), and RAGE mediates Aβ₄₀ uptake toward the brain (4). In addition, endothelial cells express unknown components that mediate uptake and degradation of Aβ₄₀ (5). LRP1 expressed at the cell membranes of neurons and astrocytes (6 and 7, respectively) mediates uptake and degradation of Aβ_40. In addition to cellular components, interstitial enzymes, IDE and NEP participate in Aβ₄₀ disposition by extracellular degradation (8). About 10–15% of Aβ₄₀ is cleared from interstitial fluid by CSF bulk flow (9).

Transgenic AD mice models are widely used in AD research. Available studies showed that transgenic mice clear human Aβ₄₀ from the brain at higher rate compared to human brain (Cirrito et al., 2003; Mawuenyega et al., 2010). Given the differences in clearance rate of Aβ₄₀ between mouse and human and the important contribution of BBB to Aβ₄₀ clearance, a set of experiments were designed to compare Aβ₄₀ disposition in mouse and human endothelial cells as BBB in vitro models.

Brain endothelial cells, astrocytes, pericytes, and neurons form the neurovascular unit that contributes to Aβ levels regulation in brain ISF (Bell and Zlokovic, 2009). While the astrocytes, pericytes and basement membrane play important role in regulating the BBB function, only the capillary endothelium forms the physical barrier separating brain from blood. To cross the BBB toward the blood, Aβ utilizes specific transporters/receptors to facilitate its transport across the endothelial physical barrier. Thus, in the current study we focused our investigation on the specific contribution of endothelial cells to the clearance of Aβ₄₀ mediated by different transport proteins from brain to blood. Our preliminary studies demonstrated a rapid depletion of Aβ₄₀ from bEnd3 cells to the media compared to hCMEC/D3. To explain this difference in Aβ₄₀ disposition, we characterized the activity of known Aβ₄₀ receptors and transporters in both cells using uptake and transport studies. For degradation studies, we used the percent of degraded Aβ₄₀ in the media as a measure for cellular degradation due to the rapid release of degraded Aβ₄₀ fragments into the media (Yamada et al., 2008). Cells grown on wells surface do not form polarized monolayers, thus cellular level of intact Aβ₄₀ is controlled by uptake by LRP1 and RAGE, efflux by P-gp, lysosomal degradation, and to a lesser extent by Aβ degrading proteases. On the other hand, cells plated on a transwell filters are polarized and form tight junctions mimicking the BBB where RAGE and P-gp are expressed on the apical side while LRP1 is expressed on the basolateral side. In transport studies, the activity of RAGE and P-gp was investigated in A→B direction, and activity of LRP1 was investigated in B→A direction. Our findings demonstrated that in the uptake studies, LRP1 in bEnd3 and hCMEC/D3 mediates uptake of intact Aβ₄₀ and partial Aβ₄₀ degradation as confirmed by RAP inhibition. On the other hand, LRP1 only mediates transport of intact Aβ₄₀ from B→A in the transport studies, but not degradation. This result is consistent with previous in vivo studies demonstrated LRP1 function in Aβ₄₀ transcytosis across the BBB (Castellano et al., 2012; Shibata et al., 2000), and is further supported by in vitro transport studies conducted using primary brain endothelial cells plated on filters (Pflanzner et al., 2011). In addition, this result confirms our in vivo data where LRP1 mediates transport of intact Aβ₄₀ across BBB. The partial reduction in Aβ₄₀ degradation by RAP observed in the uptake studies is possibly related to inhibition of another receptor that play role in Aβ₄₀ degradation, most likely expressed at the apical side in the transport model, and on the cell surface in the uptake model (Figs, 3 and 6C&D). Another possible explanation is based on previously reported studies, which utilized LRP1 minireceptors transfected into epithelial cells such as MDCK II and CHO cells. These studies showed LRP1 to mediate Aβ₄₀ degradation mainly via endocytosis (Cam et al., 2005; Nazer et al., 2008). Conflicting findings regarding LRP1 function suggest LRP1 mediates both Aβ₄₀ endocytosis and transcytosis probably depending on cell type and polarization state. This behavior of LRP1 is consistent with other members of the LDL receptors family shown previously to mediate endocytosis and transcytosis (Doherty and McMahon, 2009; Tuma and Hubbard, 2003).

RAGE is a BBB receptor that is expressed mainly at the apical side of brain endothelial cells and mediates influx of Aβ₄₀ from blood to brain (Deane et al., 2003). In non-polarized and polarized bEnd3 and hCMEC/D3 cells, RAGE has significant contribution to the uptake and A→B transport of intact Aβ₄₀ but not degradation. The results of current study support the finding of previous studies reported role for RAGE in the transcytosis of Aβ₄₀ from blood to brain (Candela et al., 2010; Deane et al., 2003).

P-gp, an efflux transporter, has been shown to contribute to Aβ₄₀ clearance across BBB (Cirrito et al., 2005; Qosa et al., 2012; Tai et al., 2009). hCMEC/D3 cells demonstrated higher P-gp expression and activity compared to bEnd3. While the effect of P-gp inhibition in bEnd3 cells was not obvious, this does not necessarily exclude the function of P-gp at the mouse BBB in Aβ₄₀ clearance. This finding could be explained rather by the rapid uptake and degradation of Aβ₄₀ by bEnd3, which hinders the role of P-gp efflux. Findings from our current and previous in vivo and in vitro P-gp inhibition studies supported P-gp function at the mouse BBB in the efflux of intact Aβ₄₀ (Abuznait et al., 2013; Qosa et al., 2012), which entered the cells by transcytosis via LRP1, and possibly other receptors, or Aβ₄₀ molecules leaked from cell organelles such as endosomes (Hansson Petersen et al., 2008). Though results from previous studies observed conflicting data about P-gp contribution to Aβ₄₀ clearance across BBB (Cirrito et al., 2005; Ito et al., 2006; Nazer et al., 2008; Tai et al., 2009), several imaging studies utilizing positron emission tomography technology demonstrated an association between P-gp function and Aβ plaque formation in AD brains (van Assema et al., 2012). Collectively, results of the current study and findings from previous mouse and human studies strongly suggest the contribution of P-gp to the clearance of Aβ₄₀ across mouse and human BBB.

Aβ₄₀ transport proteins including LRP1, RAGE and P-gp, in addition to degradation, have comparable functions in the mouse and human endothelial cells; however, differences in expression and/or activity between the two species may explain differences in Aβ₄₀ cellular disposition, thus, clearance rate. To identify the main process(s) that contributes to such differences, a mechanistic two-compartment model was established. Parameters estimated by this model describe the activity of an overall process, i.e. overall uptake (include receptor mediated endocytosis and transcytosis), overall efflux (P-gp, other efflux proteins and exocytosis), and degradation, thus these parameters were termed as apparent parameters. Nevertheless, degradation of Aβ₄₀ described by this model represents mainly lysosomal degradation and does not differentiate between different Aβ₄₀ metabolites. Consistent with observed data, the mechanistic model suggests higher Aβ₄₀ affinity to surface uptake receptors, and higher receptor activity in bEnd3 compared to hCMEC/D3 cells. As demonstrated in the model scheme (Figure 7A), removal of Aβ₄₀ from cellular compartment takes place by degradation and efflux. Apparent parameters obtained by mechanistic modeling indicated the rate of Aβ₄₀ removal by efflux process was similar in the two cells while degradation rate is higher in bEnd3 compared to hCMEC/D3 cells (Table 1). The established model also showed saturation of the uptake process in both cells was at low nanomolar concentrations of Aβ₄₀ (K_m values of 19.8 and 49.1 nM in bEnd3 and hCMEC/D3 cells, respectively). Although these concentrations are higher than those found in brains of normal individuals and in young transgenic mice (0.65 nM and 0.4 nM, respectively) (Brody et al., 2008; Kawarabayashi et al., 2001), they are much lower than Aβ₄₀ concentrations in the ISF of AD patients and aged transgenic mouse brains (2000 nM and 2500 nM, respectively) (Gregory and Halliday, 2005; Kawarabayashi et al., 2001). This suggests that in normal brains with low Aβ₄₀ concentration, clearance of Aβ₄₀ across BBB is efficient, while it is significantly diminished in AD brains due to saturable clearance across the BBB.

Collectively, parameters estimated from the mechanistic model and results obtained from uptake and transport studies permit us to conclude that; i) the uptake process of Aβ₄₀ determines subsequent steps including degradation and efflux processes, ii) brain endothelial cells mediate Aβ₄₀ degradation, however degradation rate is dependent on uptake rate which is higher in bEnd3, iii) both cells mediate efficient efflux of intact Aβ₄₀ supporting the role of efflux transporters as well as receptor mediated transcytosis in the clearance of Aβ₄₀ across endothelial cells.

Both, the in vitro and mechanistic models provided novel tools to study Aβ transport and clearance across BBB that should facilitate our understanding of physiological and pathological factors affecting Aβ clearance, however, it is worth noting that there are some limitations with these models including; i) bEnd3 and hCMEC/D3 are immortalized cell lines, thus whether current findings can be translated to in vivo differences or not remain to be evaluated; ii) the transport kinetics of free Aβ₄₀ was evaluated here, however, in physiological and pathological conditions, Aβ₄₀ is also found in complex with chaperone proteins that influence its transport kinetics; and iii) our studies focused on Aβ₄₀ disposition and not Aβ₄₂. Compared to Aβ₄₀, Aβ₄₂ clearance across the BBB is slower (Ito et al., 2006), and prone to rapid aggregation. However, as both peptides use the same clearance pathways, the findings of this study with Aβ₄₀ are expected to be applicable for Aβ₄₂.

5. Conclusions

In conclusion, our results confirmed the significant contribution of BBB to the total clearance of Aβ₄₀ from brain compared to parenchymal degradation, and demonstrated that the higher clearance rate of Aβ₄₀ across the mouse BBB compared to human BBB could be explained by the rapid uptake process precedes its efflux and degradation. Furthermore, we successfully established a mechanistic model that could predict Aβ₄₀ disposition in brain endothelial cells.

Highlights.

• Brain clearance index, a modification of brain efflux index method, was established.

• Compared to brain degradation, clearance across BBB is major clearance pathway for Aβ.

• Aβ₄₀ clearance rate by mouse brain endothelial cells is 15 x faster than human endothelial cells.

• The rapid clearance rate of Aβ₄₀ across mouse BBB is controlled by uptake and degradation.

• Aβ₄₀ disposition in brain endothelial cells was best described by 2-compartment model.

Abbreviations

AD

Alzheimer’s disease

Aβ

amyloid-β

APP

amyloid-β precursor protein

BBB

blood-brain barrier

BCI

brain clearance index

BEI

brain efflux index

CSF

cerebrospinal fluid

IDE

insulin degrading enzyme

LRP1

low density lipoprotein receptor-related protein-1

P-gp

P-glycoprotein

RAGE

advanced glycation end product

TCA

trichloroacetic acid