Volume 118, Issue 3
ORIGINAL ARTICLE
Free Access

Amyloid‐β peptide(1‐40) elimination from cerebrospinal fluid involves low‐density lipoprotein receptor‐related protein 1 at the blood‐cerebrospinal fluid barrier

Masachika Fujiyoshi

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

Search for more papers by this author
Masanori Tachikawa

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

Department of Pharmaceutics, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan

Search for more papers by this author
Sumio Ohtsuki

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

SORST of the Japan Science and Technology Agency, Japan

Search for more papers by this author
Shingo Ito

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

SORST of the Japan Science and Technology Agency, Japan

Search for more papers by this author
Yasuo Uchida

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

Search for more papers by this author
Shin‐ichi Akanuma

Department of Pharmaceutics, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan

Search for more papers by this author
Junichi Kamiie

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

SORST of the Japan Science and Technology Agency, Japan

Search for more papers by this author
Tadafumi Hashimoto

SORST of the Japan Science and Technology Agency, Japan

Department of Neuropathology and Neuroscience, Graduate School of Pharmaceutical Sciences, University of Tokyo, Tokyo, Japan

Search for more papers by this author
Ken‐ichi Hosoya

Department of Pharmaceutics, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan

Search for more papers by this author
Takeshi Iwatsubo

SORST of the Japan Science and Technology Agency, Japan

Department of Neuropathology and Neuroscience, Graduate School of Pharmaceutical Sciences, University of Tokyo, Tokyo, Japan

Search for more papers by this author
Tetsuya Terasaki

Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, Sendai, Japan

SORST of the Japan Science and Technology Agency, Japan

Search for more papers by this author
First published: 17 May 2011
Citations: 28
Address correspondence and reprint requests to Tetsuya Terasaki, Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical Sciences, Tohoku University, 6‐3 Aoba, Aramaki, Aoba‐ku, Sendai 980‐8578, Japan.
E‐mail: [email protected]

Abstract

J. Neurochem. (2011) 118, 407–415.

Abstract

Amyloid‐β peptide (Aβ) concentration in CSF is potentially a diagnostic and therapeutic target for Alzheimer’s disease (AD). The purpose of this study was to clarify the elimination mechanism of human Aβ(1‐40) [hAβ (1‐40)] from CSF. After intracerebroventricular (ICV) administration, [125I]hAβ(1‐40) was eliminated from the rat CSF with a half‐life of 17.3 min. The elimination of [125I]hAβ(1‐40) was significantly inhibited by human receptor‐associated protein (RAP) and the elimination was attenuated in either anti‐low‐density lipoprotein receptor‐related protein (LRP)1 antibody‐treated or RAP‐deficient mice, suggesting that a member(s) of the low‐density lipoprotein receptor gene family is involved in the elimination of hAβ(1‐40) from CSF. The amounts of LRP1 and LRP2 proteins were determined by means of liquid chromatography‐tandem mass spectrometry, and the LRP1 content in rat choroid plexus was determined to be 3.7 fmol/μg protein, whereas the LRP2 content was below the detection limit (< 0.2 fmol/μg protein). Conditionally, immortalized rat choroid plexus epithelial cells exhibited predominant apical‐to‐basal and apical‐to‐cell transport of [125I]hAβ(1‐40). These results indicated that hAβ(1‐40) is actively eliminated from CSF and this process is at least partly mediated by LRP1 expressed at choroid plexus epithelial cells, which therefore play a role in determining CSF concentrations of hAβ(1‐40).

Abbreviations used

  • AD
  • Alzheimer’s disease
  • amyloid‐β peptide
  • BCSFB
  • blood‐cerebrospinal fluid barrier
  • CSF
  • cerebrospinal fluid
  • TR‐CSFB3
  • conditionally immortalized rat choroid plexus epithelial cell line
  • ECF
  • extracellular fluid
  • ISF
  • interstitial fluid
  • LC‐MS/MS
  • liquid chromatography‐tandem mass spectrometry
  • LRP
  • low‐density lipoprotein receptor‐related protein
  • PBS
  • phosphate‐buffered saline
  • PDAPP
  • human amyloid precursor protein
  • RAP
  • receptor‐associated protein
  • MRM
  • multiple reaction monitoring
  • An abnormally elevated level of Aβ in the brain is one of the prominent features of AD, and reduced clearance of cerebral Aβ has been proposed to cause accumulation of Aβ in the late‐onset AD brain (Hardy and Selkoe 2002). Clearance has been reported to be achieved by transfer across the blood–brain barrier (BBB) and by Aβ degradation (Zlokovic 2008).

    The brain interstitial fluid (ISF) is in direct with CSF and the soluble Aβ forms are present in normal human CSF (Seubert et al. 1992; Shoji et al. 1992). After infusion of [125I]Aβ and [14C]polyethelene glycol into rat lateral ventricle at equal levels of radioactivity, the normalized level of [125IAβ was significantly lower than that of [14C]polyethelene glycol in cisternal CSF (Ghersi‐Egea et al. 1996). It has also been reported that [125I]Aβ was bound to or taken up into isolated rat choroid plexus (Crossgrove et al. 2005). These results suggest the existence of a clearance pathway from CSF into choroid plexus epithelial cells, which form the blood‐CSF barrier (BCSFB). Total apical surface area of the choroid plexus was estimated to be half of that of the cerebral capillaries (Keep and Jones 1990a,b). Therefore, it is possible that efflux transport at the BCSFB, in addition to that at the BBB, plays an important role in cerebral Aβ clearance.

    Cerebrospinal fluid formation and turnover rates have been reported to be reduced in patients with AD (Johanson et al. 2008). Cirrito have measured soluble Aβ in ISF by means of a microdialysis method (Cirrito et al. 2003), and found no correlation between soluble Aβ levels in ISF and those in CSF in 3‐month‐old human amyloid precursor protein (PDAPP) transgenic mice, whereas in 12‐ to 15‐month‐old PDAPP transgenic mice, a significant correlation was observed between ISF and CSF Aβ levels. This age‐difference could not be explained by elimination of Aβ by CSF bulk flow, but could be explained by age‐dependent attenuation of efflux transport of Aβ from CSF.

    It was reported that brain‐to‐blood efflux transport of hAβ(1‐40) across the BBB was mediated by LRP 1 (Shibata et al. 2000), and we demonstrated that LRP1 is the receptor responsible for the hepatic uptake of hAβ(1‐40) (Tamaki et al. 2006). As LRP1 has been reported to be expressed in the rat choroid plexus (Kounnas et al. 1994), LRP1 is likely to be involved in the clearance of hAβ(1‐40) in CSF. However, there is also a report that LRP2 is expressed at the BCSFB and is involved in the elimination of hAβ(1‐40) across cultured choroid plexus epithelial cells (Carro et al. 2005).

    Although Aβ concentration in the CSF is potentially a diagnostic and therapeutic target for AD, the molecular mechanism of Aβ elimination from the CSF has remained unclear. The purpose of this study was to clarify and characterize the efflux transport system of hAβ(1‐40) at the BCSFB by means of an in vivo ICV administration study and an in vitro study using a conditionally immortalized rat choroid plexus epithelial cell line (TR‐CSFB3).

    Materials and methods

    Animals

    Male Sprague‐Dawley rats and C57BL/6J mice were obtained from Charles River (Yokohama, Japan). RAP‐deficient mice (Stock #002987) were obtained from Jackson Laboratory (Bar Harbor, ME, USA). The investigation using animals described in this report conformed to the guidelines of the Animal Care Committee, Graduate School of Pharmaceutical Sciences, Tohoku University (Sendai, Japan).

    Reagents

    Monoiodinated and non‐oxidized [125I]hAβ(1‐40) labeled from hAβ(1‐40) (BioSource, Camarillo, CA, USA) by Perkin‐Elmer Life Sciences (Boston, MA, USA) (2200 Ci/mmol, molecular weight = 4455) was purchased from Perkin‐Elmer. [Carboxyl‐14C]Inulin ([14C]inulin, 1.92 mCi/g, molecular weight = 5000–5500) was purchased from ICN Biomedicals (Costa Mesa, CA, USA). Unlabeled Aβ(1‐40) and Aβ(1‐42) were purchased from Bachem AG (Bubendorf, Switzerland). RAP was purchased from Oxford Biomedical Research (Oxford, MI, USA). Human α2‐macroglobulin (α2M) was purchased from Sigma (St Louis, MO, USA), and activated with methylamine as described previously (Ashcom et al. 1990). Xylazine hydrochloride was purchased from Sigma. Ketamine hydrochloride was purchased from Sankyo (Tokyo, Japan). All other chemicals were commercial products of reagent grade.

    Intracerebroventricular administration

    The elimination of [125I]hAβ(1‐40) after ICV administration was studied using the methods described previously (Kitazawa et al. 2000). In brief, rats were anesthetized with an i.m. injection of xylazine (1.22 mg/kg) and ketamine (125 mg/kg), and the head was fixed with a stereotaxic apparatus (SR‐6; Narishige Co., Tokyo, Japan). A hole was drilled in the skull, 1.5 mm to the left and 0.5 mm posterior to the bregma, into which a needle was fixed as a cannula for injection. [125I]hAβ(1‐40) (0.05 μCi) or [14C]inulin (0.01 μCi) was dissolved in 10 μL of extracellular fluid (ECF) buffer (122 mM NaCl, 25 mM NaHCO3, 3 mM KCl, 1.4 mM CaCl2, 1.2 mM MgSO4, 0.4 mM K2HPO4, 10 mM D‐glucose, and 10 mM HEPES, pH 7.4) and administered to the left lateral ventricle. For inhibition studies, unlabeled inhibitor at the desired concentration was administered simultaneously. For treatment with antibody, either anti‐LRP1 antibody (N‐20; Santa Cruz Biotech Inc., Santa Cruz, CA, USA) or goat normal IgG (Chemicon International, Temecula, CA, USA) in ECF buffer at 60 μg/mL was infused into the left lateral ventricle for 10 min at 2.9 μL/min, and then [125I]hAβ(1‐40) (0.05 μCi) was administered with 60 μg/mL antibody (Bell et al. 2007). [125I]hAβ(1‐40) and each inhibitor or antibody were mixed and kept at 4°C for less than 2 h until injection. Just before the injection, the mixture was warmed to 37°C. At the designated time, CSF (50–100 μL) was withdrawn by cisternal puncture. Levels of 125I and 14C in the CSF and in the injection solutions were measured by using a γ‐counter (ARC300; Aloka, Tokyo, Japan) and a liquid scintillation counter (TRI‐CARB 2050CA; Packard, Meriden, CT, USA), respectively. When mice were used, a hole was drilled in the skull, 1 mm to the left and 0.5 mm posterior to the bregma and the volume of solution injected was reduced to 2 μL.

    The kinetic parameters for [125I]hAβ(1‐40) and [14C]inulin were determined from eqn (1) (Ogawa et al. 1994) using the non‐linear least‐squares regression analysis program, MULTI (Yamaoka et al. 1981).

    image(1)
    where Ccsf(t), Vd and kel are the CSF concentration at time t, the volume of distribution and the elimination rate constant, respectively, of either [125I]hAβ(1‐40) or [14C]inulin. The apparent elimination clearance value was determined by multiplying kel by Vd (kel × Vd). As the value of [14C]inulin at time = 0 in Fig. 1 was 453% of dose/mL, the volume of CSF was determined to be 220 μL, so that the injected [125I]hAβ(1‐40) and inhibitor would have been diluted by 23‐fold.

    image

    Time‐courses of [125I]hAβ(1‐40) (• : 4.0 nM × 10 μL) and [14C]inulin ( : approximately 75 μM × 10 μL) remaining in rat CSF after intracerebroventricular administration. The solid line was obtained using non‐linear least‐squares regression analysis. Each point represents the mean ± SEM (n = 4–7).

    Cell culture

    A conditionally immortalized rat choroid plexus epithelial cell line was established from transgenic rat harboring the temperature‐sensitive simian virus 40 large T‐antigen in our laboratory (Kitazawa et al. 2001). TR‐CSFB3 cells were cultured in Dulbecco’s modified Eagle’s medium (Nissui Pharmaceutical Co., Tokyo, Japan) supplemented with 20 mM sodium bicarbonate, 4.5 g/L D‐glucose, 100 U/mL benzylpenicillin potassium, 100 μg/mL streptomycin sulfate, and 10% fetal bovine serum (FBS; Moregate, Bulimba, Australia). TR‐CSFB3 cells were maintained at 33°C, which is the permissive temperature at which the temperature‐sensitive simian virus 40 large T‐antigen is activated, in an atmosphere of 5% CO2 in air.

    Transcellular transport and uptake of [125I]hAβ(1‐40) in TR‐CSFB3 cells

    The transport studies were performed at 37°C, the physiological temperature. TR‐CSFB3 cells were seeded at a density of 3 × 104 cells/insert on polyester membrane transwell‐clear inserts (0.4 μm pore size, 12 mm diameter; Costar, Cambridge, MA, USA) inserted into 12‐well plates (Costar). The cells were washed three times with ECF buffer, and [125I]hAβ(1‐40) (0.75 μCi/mL) in ECF buffer was added to the apical (0.5 mL) or basal (1.5 mL) side; ECF buffer was added to the opposite side. At the designated time, aliquots of buffer were taken from the opposite side and the cells were washed three times with ice‐cold ECF buffer. 125I radioactivity was measured with a γ‐counter.

    The amount of [125I]hAβ(1‐40) transferred from the donor side to the receiver side was obtained as the clearance, by dividing the amount of [125I]hAβ(1‐40) in the receiver side by the [125I]hAβ(1‐40) concentration in the donor side at each time point (Ohtsuki et al. 2007). The total clearance was calculated by summing the clearance up to the specified time point. When total clearance was plotted versus time, the slope estimated by linear regression analysis represents the permeability × surface area product (PS). The PS value for the epithelial monolayer (PeS) was calculated from eqn (2).

    image(2)
    where PtotalS and PfilterS are the PS value in the presence and absence of the cells, respectively. The PeS value was divided by the surface area of the culture insert (1.1 cm2) to give the epithelial permeability coefficient (Pe).

    For measuring the amount of [125I]hAβ(1‐40) in TR‐CSFB3 cells, the cells were incubated with acetate‐barbital buffer (28 mM CH3COONa, 120 mM NaCl, and 20 mM barbital sodium, pH 3.0) for 20 min at 4°C. After the incubation, the cells were washed three times with ice‐cold acetate‐barbital buffer and lysed in 1% Triton X‐100 for 16 h at 20°C. 125I radioactivity was measured with a γ‐counter. The protein amount in the cell lysates was measured by the Bradford method using Bio‐Rad Protein Assay reagent (Bio‐Rad; Hercules, CA, USA). [125I]hAβ(1‐40) internalization was obtained as cell‐to‐medium ratio, determined from the amount of ligand accumulated in the cells divided by the medium concentration in the chamber in which [125I]hAβ(1‐40) was applied.

    Quantification of LRP1 and LRP2 proteins by mass spectrometric analysis

    Quantification of LRP1 and LRP2 proteins was performed as described previously (Kamiie et al. 2008). A proteotypic peptide formed by trypsin digestion, specific for each receptor and giving an intense mass spectrometry signal, was selected according to the following criteria: peptide length between 6 and 16 amino acids; the peptide does not contain post‐translational modifications or SNPs; the peptide is not from a transmembrane domain; no continuous sequence of arginine or lysine in digestion sites; the peptide does not contain methionine or cysteine residue; and the peptide should preferably not contain histidine residue. The specific proteotypic peptide for each receptor protein is shown in Table 1, and stable‐isotope‐labeled peptides and unlabeled peptides were synthesized by Thermoelectron Corporation (Sedanstrabe, Germany). Under deep anesthesia induced with pentobarbital (100 mg/kg), adult rats were transcardially perfused with phosphate‐buffered saline to remove blood and the rat choroid plexus was isolated after this perfusion. Isolated rat choroid plexus and TR‐CSFB3 cells were suspended in 100 mM Tris‐HCl (pH 8.5), 7 M guanidium hydrochloride, 10 mM EDTA, and the proteins were S‐carbamoylmethylated as described (Mawuenyega et al. 2003). The alkylated proteins were precipitated with a mixture of methanol and chloroform. The precipitate was dissolved in 6 M urea, diluted with 100 mM Tris‐HCl (pH 8.0) and digested with TPCK‐treated trypsin (Promega, Madison, USA) at an enzyme/substrate ratio of 1 : 100 at 37°C for 16 h.

    Table 1. Proteotypic peptide sequences and selected ions for quantification of each peptide by means of LC‐MS/MS with multiple reaction monitoring (MRM)
    Target Peptide sequence Peptide mass MRM channel (m/z)
    Q1 Q3‐1 Q3‐2 Q3‐3
    Rat LRP1 GDYSVLVPGLR 1174.6 588.3 541.3 654.4 753.5
    GDYSVLVPGL*R 1181.6 591.8 548.3 664.1 760.5
    Rat LRP2 VTLITENLGHPR 1348.7 675.4 923.5 1036.6 1150.0
    VTLITENL*GHPR 1355.7 678.9 930.5 1043.6 1157.0
    • LRP, low‐density lipoprotein receptor‐related protein.
    • The conditions of MRM were optimized for high signal intensity following direct injection of peptide solution into the mass spectrometer through a turbo ion spray source. Theoretical m/z values of doubly charged ions of intact peptides were assumed as precursor ions. Three singly charged ions derived from the y series were obtained by MS/MS. Bold letters with asterisks indicate amino acid residues labeled with stable isotope (13C and 15N).

    The tryptic digests were acidified with formic acid for analysis with the HPLC system (Agilent 1100 system; Agilent, Santa Clara, USA), which was connected to an ESI‐triple quadrupole mass spectrometer (API5000; Applied Biosystems, Foster City, USA). HPLC was performed with C18 capillary columns (Agilent 300SB‐C18 0.5 mm ID × 150 mm, 5 μm particles). Linear gradients of 1–45% acetonitrile in 0.1% formic acid were applied to elute the peptides at a flow rate of 50 μL/min for 50 min. The mass spectrometer was set up for multiple reaction monitoring to detect peptides at ten dwell times per channel. Ion counts were determined by using the quantification procedures in Analyst software version 1.4.2 (Applied Biosystems). Although LRP2 was below the detection limit in the present study, LRP2 protein was able to be quantified in mouse kidney by this method (11.7 fmol/μg protein), where LRP1 was below the detection limit.

    Data analysis

    Unless otherwise indicated, all data represent the mean ± SEM values. An unpaired, two‐tailed Student’s t‐test was used to determine the significance of differences between mean values of two groups. The statistical significance of differences among means of more than two groups was determined by one‐way anova followed by the Bonferroni multiple comparisons test.

    Results

    Efflux of [125I]hAβ(1‐40) from rat CSF after intracerebroventricular administration

    Figure 1 shows the time‐courses of the percentage dose of [125I]hAβ(1‐40) and [14C]inulin, which is an impermeable marker reflecting CSF bulk flow, remaining in the rat CSF after ICV administration. The values of elimination rate constant (kel) of [125I]hAβ(1‐40) and [14C]inulin were 4.01 × 10−2 ±0.45 × 10−2 and 8.03 × 10−3 ± 3.16 × 10−3/min (mean ± SD), and the volumes of distribution (Vd) were 240 ± 13 and 221 ± 10 μL/rat (mean ± SD), respectively. The apparent elimination clearances of [125I]hAβ(1‐40) and [14C]inulin from the cerebral ventricle were 9.61 and 1.77 μL/min rat, respectively. These results indicate that [125I]hAβ(1‐40) was eliminated from CSF faster than would be expected for elimination via CSF bulk flow.

    Following co‐administration of unlabeled hAβ(1‐40) [10 μL × 20 μM hAβ(1‐40)] into rat ventricle, the CSF concentration of [125I]hAβ(1‐40) at 30 min was significantly increased to 195 (% dose)/mL compared with a tracer dose of [125I]hAβ(1‐40) [146 (% dose/mL)] (p < 0.01) at 30 min after the ICV administration (Table 2), indicating involvement of a saturable process in the elimination of [125I]hAβ(1‐40) from CSF. The CSF concentration of [125I]hAβ(1‐40) was also increased to 197 (% dose)/mL by co‐administration of unlabeled hAβ(1‐42) [10 μL × 20 μM hAβ(1‐42)] (Table 2).

    Table 2. Effect of various compounds on elimination of [125I]hAβ(1‐40) from rat CSF
    Compound Concentration in injectate (μM) [125I]hAβ(1‐40) concentration remaining in CSF (% dose/mL)
    Control 146 ± 5
    hAβ(1‐40) 20 195 ± 5*
    hAβ(1‐42) 20 197 ± 5*
    RAP 5 203 ± 9*
    α2M* 0.4 198 ± 4*
    • *p <0.01, significantly different from the control.
    • hAβ, human amyloid‐β peptide; RAP, receptor‐associated protein; α2M*, methylamine‐activated α2‐macroglobulin.
    • Unlabeled hAβ(1‐40), hAβ(1‐42), RAP, and α2M* were co‐administered with [125I]hAβ(1‐40) (4.0 nM × 10 μL) into rat lateral ventricle.
    • Each value, determined 30 min after intracerebroventricular administration, represents the mean ± SEM (n = 4–7).

    Involvement of low‐density lipoprotein receptor‐related proteins (LRPs) in the elimination of [125I]hAβ(1‐40) from rat CSF

    To investigate the involvement of LRPs in [125I]hAβ(1‐40) elimination from the CSF, an inhibition study was conducted with LRP inhibitor and ligand. The CSF concentration of [125I]hAβ(1‐40) was significantly increased to 203 (% dose)/mL at 30 min after ICV administration of 5 μM RAP, which inhibits ligand binding to the LDL receptor gene family. Methylamine‐activated α2M (α2M*), which is an LRP1 ligand, significantly increased the CSF concentration of [125I]hAβ(1‐40) to 198 (% dose)/mL (Table 2).

    To investigate the involvement of LRP1 in [125I]hAβ(1‐40) elimination from CSF, either anti‐LRP1 antibody or control normal IgG was infused into the lateral ventricle, and [125I]hAβ(1‐40) was injected together with the antibody. The CSF concentration of [125I]hAβ(1‐40) at 30 min was increased 1.86‐fold by the treatment with anti‐LRP‐1 antibody compared with normal IgG treatment (Fig. 2a). Furthermore, we examined RAP‐deficient mice, which exhibit reduced expression of the LDL receptor gene family, predominantly LRP1, in the brain and liver (Willnow et al. 1995, 1996; Deane et al. 2004). The CSF concentration of [125I]hAβ(1‐40) at 30 min in RAP‐deficient mice was also increased to 1.67‐fold compared with that in control wild‐type mice, suggesting decrease in the elimination rate in RAP‐deficient mice (Fig. 2b).

    image

    Effect of anti‐low‐density lipoprotein receptor‐related protein (LRP) 1 antibody (a) and gene knockout of receptor‐associated protein (RAP) (b) on the elimination of [125I]hAβ(1‐40) from mouse CSF. (a) Either anti‐LRP1 antibody (N‐20) or goat normal IgG at 60 μg/mL was infused into the left lateral ventricle of rat brain for 10 min, and then [125I]hAβ(1‐40) was administered with 60 μg/mL antibody into the lateral ventricle. (b) [125I]hAβ(1‐40) was administered into the lateral ventricle of 11‐week‐old C57BL/6J mice (RAP+/+) and age‐matched RAP‐deficient mice (RAP−/−). Each value, determined 30 min after intracerebroventricular administration, represents the mean ± SEM (n = 3). As CSF (% dose/mL) is affected by the volume of the ventricular space and the mouse study therefore gives larger values than the rat study, the data are shown as % of control to allow comparison of (a) and (b). **p <0.01, significantly different from the control.

    Protein expression of LRP1 and LRP2 in rat choroid plexus and in TR‐CSFB3 cells

    The amounts of LRP1 and LRP2 were determined in rat choroid plexus and TR‐CSFB3 by means of multiple reactions monitoring with liquid chromatography‐tandem mass spectrometry (LC‐MS/MS), which was used to quantify a specific peptide for each target protein in the samples. The amounts of LRP1 protein in rat choroid plexus and in TR‐CSFB3 cells were determined to be 3.71 and 1.10 fmol/μg protein, respectively (Table 3). In contrast, LRP2 protein expression in rat choroid plexus and TR‐CSFB3 cells was below the detection limit (< 0.2 fmol/μg protein). Therefore, LRP1 is predominantly expressed in both rat choroid plexus and in TR‐CSFB3 cells.

    Table 3. Expression levels of low‐density lipoprotein receptor‐related protein (LRP) 1 and LRP2 protein in rat choroid plexus and conditionally immortalized rat choroid plexus epithelial cell line (TR‐CSFB3) cells
    Protein expression level (fmol/μg protein)
    Choroid plexus TR‐CSFB3 cells
    LRP1 3.71 ± 0.32 1.10 ± 0.23
    LRP2 < 0.20 < 0.20
    • Whole cell lysate of rat choroid plexus or TR‐CSFB3 cells (50 μg) was digested with trypsin. Trypsin‐digested peptides supplemented with 500 fmol of stable isotope‐labeled peptide mixture were injected into the LC‐MS/MS. Multiplexed‐MRM analysis was performed under optimized conditions. The quantitative value was calculated from the peak area ratio of analyte and stable isotope‐labeled peptide in each MRM channel. The amount of each protein was determined as the average of 12 quantitative values, from four different MRM channels of three samples with signal peaks over 5000 counts.
    • Each value represents the mean ± SEM (n = 12).

    Transcellular transport and intracellular accumulation of [125I]hAβ(1‐40) in TR‐CSFB3 cells

    We next examined transepithelial transport of [125I]hAβ(1‐40) from the apical side to the basal side and from the basal side to the apical side, which would correspond to the CSF‐to‐blood and blood‐to‐CSF directions, respectively, using TR‐CSFB3 cells. TR‐CSFB3 cells exhibit apical localization of Na+/K+ ATPase and apical‐to‐cell uptake of L‐proline predominated over basal‐to‐cell uptake, as is the case in vivo (Kitazawa et al. 2001). The apparent permeability coefficients of [125I]hAβ(1‐40) in the presence of the cells, PtotalS, and that in the absence of the cells, PfilterS, were determined from the slopes in Fig. 3a. Using these values of PtotalS, PfilterS, and eqn (2), PeS was determined. The epithelial permeability coefficient (Pe) of [125I]hAβ(1‐40) in TR‐CSFB3 cells, which was calculated by dividing PeS by the surface area of the culture insert, was 6.24 μL/min/cm2 in the apical‐to‐basal direction and 1.49 μL/min/cm2 in the basal‐to‐apical direction, suggesting that CSF‐to‐blood transport predominates over blood‐to‐CSF transport.

    image

    Time‐courses of [125I]hAβ(1‐40) transport (a) and uptake (b) in conditionally immortalized rat choroid plexus epithelial cell line (TR‐CSFB3) cells cultured on Transwells. (a) [125I]hAβ(1‐40) was added to the apical side, squares; to the basal side, circles. Clearance with TR‐CSFB3 cells, closed symbols; clearance without TR‐CSFB3 cells, open symbols. (b) [125I]hAβ(1‐40) uptake following application to the apical side, squares; to the basal side, circles. Each point represents the mean ± SEM (n = 3). Where error bars are not shown, they are smaller than the symbol.

    As shown in Fig. 3b, [125I]hAβ(1‐40) accumulation in TR‐CSFB3 cells was concentrative and time‐dependent in both directions. The cellular accumulation of [125I]hAβ(1‐40), determined at 10 min, was 8.57‐fold greater in the apical‐to‐cell direction (67.9 ± 4.0 μL/mg protein) than in the basal‐to‐cell direction (7.92 ± 0.45 μL/mg protein).

    Discussion

    The present study demonstrates that hAβ(1‐40) was eliminated from CSF after ICV administration (Fig. 1). The elimination rate constant of hAβ(1‐40) was greater than that of inulin, which reflects CSF bulk flow (Fig. 1), and the elimination was inhibited by excess unlabeled hAβ(1‐40) (Table 2). These results indicate that a carrier‐mediated process is involved in the elimination of hAβ(1‐40) from the CSF. Other possible mechanisms of elimination include diffusion back into the brain parenchyma and/or binding and transport to cells in the ventricular space, such as choroid plexus epithelial cells and ependymal cells. A 125I autoradiogram of brain slices 30 min after administration of [125I]hAβ(1‐40) into the lateral ventricle showed an intense signal in choroid plexus, but the signal decreased sharply from the ependymal border to the adjacent brain (Ghersi‐Egea et al. 1996). This suggests that the process of diffusion into the brain parenchyma would have had a limited impact on elimination within the period (30 min) of our experiments.

    As shown in Table 2, LRP competitive ligands (RAP and α2M*) inhibited the elimination of hAβ(1‐40) from CSF, suggesting the involvement of LRP in the elimination process. The co‐administered LRP ligands would presumably also be eliminated from the CSF, resulting in attenuation of the inhibitory effect. However, the Kd values of RAP and α2M* toward LRP1 were reported to be 1–10 nM (Wu and Pizzo 1996; Kanekiyo and Bu 2009), and the concentrations of RAP and α2M* in the CSF was estimated to be 230 nM and 18 nM, respectively, on the basis of 23‐fold dilution of the injected concentrations, as described in Materials and methods. These concentrations of ligands in CSF would be expected to show an inhibitory effect on LRP1 even if 70% of each ligand was eliminated from the CSF, as was the case for [125I]hAβ(1‐40).

    It is necessary to rule out the possibility that the inhibitory effect resulted from binding of hAβ(1‐40) to the co‐administered LRP ligand, as RAP or α2M* was reported to show complex formation with hAβ(1‐40) (Narita et al. 1997; Kanekiyo and Bu 2009; Kerr et al. 2010). However, only a small part (approximately 15%, estimated from peak height) of hAβ(1‐40) was reported to be bound to RAP after incubation with [125I]hAβ(1‐40) (500 nM) and RAP (2.5 μM) for 2 h at 37°C (Kanekiyo and Bu 2009). As for the complex formation of [125I]hAβ(1‐40) and α2M*, we could not detect any complex formation by gel filtration chromatography after incubation of [125I]hAβ(1‐40) (25 nM) and α2M* (2.5 nM) at 4°C for 3 h (data not shown). In the present study, the mixtures of [125I]hAβ(1‐40) and each LRP ligand were kept at 4°C for less than 2 h before injection. However, the possibility that LRP ligand bound [125I]hAβ(1‐40) after the injection cannot be excluded, and it is also possible that other carrier proteins in the CSF influence the interaction of LRP ligand with [125I]hAβ(1‐40).

    Therefore, to evaluate the direct interaction of [125I]hAβ(1‐40) with LRP1, we further examined the attenuation of [125I]hAβ(1‐40) elimination from CSF by anti‐LRP1 antibody treatment as well as in RAP‐deficient mice (Fig. 2). Anti‐LRP1 antibody (N20) was reported to inhibit hAβ(1‐40) elimination from the brain through the blood‐brain barrier on co‐administration following pre‐infusion (Bell et al. 2007). As the control normal IgG would be expected to interact with [125I]hAβ(1‐40) in the CSF in a similar manner to anti‐LRP1 antibody, the decrease in the elimination rate of [125I]hAβ(1‐40) caused by anti‐LRP1 antibody treatment, which resulted in an increase of the CSF concentration as shown in Fig. 2a, was suggested to be mainly because of the interaction of the antibody with LRP‐1. RAP gene deficiency results in decreased expression of members of the LDL receptor gene family, including decreased LRP1 expression in the brain and liver (Willnow et al. 1995). If [125I]hAβ(1‐40) interacts with endogenous RAP in the CSF of wild‐type mice, its elimination rate in wild‐type would be expected to be slower than or the same as that in RAP‐deficient mice. Therefore, the decrease in the elimination rate of hAβ(1‐40) in RAP‐deficient mice shown in Fig. 2b can be explained by decreased expression of LRPs in RAP‐deficiency, but not by the interaction of hAβ(1‐40) with endogenous RAP in wild‐type. Overall, these results suggest that LRP1 is involved in the elimination of hAβ(1‐40) from the CSF, at least in part.

    The contribution of each process to the elimination of hAβ(1‐40) from the CSF could be estimated based on the kel in eqn (1) given in Materials and methods. The elimination process by CSF flow was 20.0% of the total elimination (0.803 × 10−2/min for kel,flow and 4.01 × 10−2/min for kel,total shown in Fig. 1). The kel of non‐CSF flow (kel,non‐flow) was 3.21 × 10−2/min, obtained by subtracting kel,flow from kel,total. To estimate the contribution of the RAP‐sensitive process, the apparent kel of hAβ(1‐40) co‐administered with RAP (kel,total/RAP) was taken to be 2.41 × 10−2/min on the assumption that the CSF concentration of hAβ(1‐40) at time = 0 was not affected by RAP (Fig. 1). The kel,total/RAP reflects the process involving CSF flow and the RAP‐insensitive process. The kel of the RAP‐insensitive process in non‐CSF flow processes (kel,non‐flow/RAP) was calculated to be 1.61 × 10−2/min by subtracting kel,flow from kel,total/RAP. As a result, the RAP‐insensitive and RAP‐sensitive processes are estimated to contribute almost equally to the non‐CSF flow elimination process [1.61 × 10−2/min and 1.60 × 10−2/min (kel,non‐flow−kel,non‐flow/RAP), respectively]. This implies that LRP plays a substantial role in the elimination process of hAβ(1‐40) from the CSF. This also suggests the existence of a RAP‐insensitive process, which is likely to involve a low‐affinity transport system mediated by other molecule(s) and/or binding in the ventricular space. The extent of binding in the ventricular space will need to be assessed in further studies, e.g., by measuring the remaining radioactivity in the brain after the perfusion.

    Previously, we determined the elimination clearance of hAβ(1‐40) across rat BBB to be 11.0 μL/(min·g brain) (Ito et al. 2006). On the assumption that the weight of rat brain is 1.6 g per rat, the present findings indicate that the elimination clearance of hAβ(1‐40) from the CSF is 6.01 μL/(min·g brain) including the elimination via CSF bulk flow [1.11 μL/(min·g brain)]. The elimination rate of hAβ(1‐40) was calculated by multiplying the elimination clearance by the reported hAβ(1‐40) concentrations in 3‐month‐old PDAPP transgenic mice: 0.224 ng/mL in ISF and 5.06 ng/mL in CSF (Cirrito et al. 2003). The elimination rate of hAβ(1‐40) across the BBB and that from CSF were estimated to be 2.46 and 30.4 pg/(min·g brain), respectively. Although the clearance rate is affected by the hAβ concentrations in ISF and CSF, these results suggest that the elimination process from the CSF is a significant pathway for hAβ(1‐40) clearance from the brain.

    One of the possible elimination routes from the CSF is uptake by choroid plexus epithelial cells. As shown in Fig. 3, the hAβ(1‐40) epithelial permeability coefficient in TR‐CSFB3 cells was significantly greater in the apical‐to‐basal direction than in the basal‐to‐apical direction. In addition, hAβ(1‐40) internalization in TR‐CSFB3 cells was significantly higher in the apical‐to‐cell direction than in the basal‐to‐cell direction (Fig. 3). These in vitro results suggest that hAβ(1‐40) in the CSF was eliminated not only by binding to ependymal cells and choroid plexus epithelial cells but also by uptake into and transport across the choroid plexus epithelial cells. Furthermore, the CSF‐to‐blood transport of hAβ(1‐40) is likely to be predominant over the blood‐to‐CSF transport.

    Low‐density lipoprotein receptor‐related protein 1 and 2, which belong to the LDL gene family, have been reported to be involved in transport of hAβ(1‐40) at the BBB (Zlokovic et al. 1996; Hammad et al. 1997; Shibata et al. 2000). hAβ(1‐40) elimination from rat CSF was inhibited by α2M* (Table 2), which binds to LRP1 at the same binding site as Aβ(1‐40) (Deane et al. 2004). LRP1 protein expression was also detected in isolated rat choroid plexus and in TR‐CSFB3 cells by the LC‐MS/MS method (Table 3), whereas LRP2 expression was below the detection limit, suggesting that LRP1 is involved in the elimination process of hAβ(1‐40) from CSF through uptake by choroid plexus epithelial cells.

    Low‐density lipoprotein receptor‐related protein 2 was detected in rat choroid plexus by means of Western blotting (Carro et al. 2005), but an earlier immunohistochemical study had failed to find LRP2 expression in rat choroid plexus (Zheng et al. 1994). There has been no quantitative analysis to compare the expression levels of LRP1 and LRP2 in choroid plexus. In the present study, absolute protein expression levels of LRP1 and LRP2 were selectively quantified by LC‐MS/MS. As shown in Table 3, a peptide specific for LRP1 was detected from choroid plexus and TR‐CSFB3 cells, whereas a peptide specific for LRP2 was not. This result demonstrates that the protein amount of LRP1 in the rat choroid plexus is at least 19‐fold greater than that of LRP2, suggesting that LRP1 is predominantly expressed in choroid plexus.

    The involvement of LRP1 in hAβ(1‐40) elimination from the CSF is likely to explain the contradictory nature of previous reports on the role of LRP1 in hAβ(1‐40) elimination at the BBB. Shibata et al. (2000) reported that LRP1 plays a major role in hAβ(1‐40) elimination from brain ISF at the BBB because the hAβ(1‐40) elimination from the brain was inhibited by RAP to the extent of 90%, and clearance was decreased by about 75% in RAP knockout mice. In contrast, we found that RAP inhibited hAβ(1‐40) elimination from rat brain by only 20% (Ito et al. 2006). Our present results indicate that intracerebrally microinjected Aβ was eliminated from the brain to the circulating blood via efflux transport across the BBB, CSF bulk flow, and efflux transport across the BCSFB. In our previous study, [125I]hAβ(1‐40) was injected into Par2 in the cerebral cortex, and it was confirmed that no significant radioactivity was detected in the contralateral, cerebrum, cerebellum or CSF compartment 60 min after microinjection. In the study by Shibata et al. (2000), hAβ and inulin were detected in CSF after intracerebral microinjection into the caudate nucleus. Although CSF bulk flow was compensated based on inulin clearance, they did not take into account LRP1‐mediated elimination from the CSF. It was reported that less than 3% of [14C]inulin injected in the Par2 region was detected in CSF at 20 min after injection, whereas as much as 12–46% was detected in CSF after injection into other regions, such as field CA2 of Ammon's Horn, hippocampal fissure and frontal cortex area 1 (Kakee et al. 1996). These results raise the possibility that hAβ(1‐40) injected into the caudate nucleus was eliminated not only across the BBB and by CSF bulk flow but also by LRP1‐mediated elimination across the BCSFB. Therefore, the elimination clearance of hAβ(1‐40) injected into the caudate nucleus is likely to include a contribution because of the function of LRP1 at the BCSFB.

    Cerebrospinal fluid is in direct communication with the brain ISF, so CSF concentrations of hAβ(1‐40) and hAβ(1‐42) are considered to give information about the development of AD (Lewczuk et al. 2004). Though many lines of evidence demonstrate that CSF hAβ(1‐42) level is reduced in AD, reduced hAβ(1‐42) level is also a feature of other dementias, with and without hAβ accumulation (Otto et al. 2000). hAβ(1‐40) elimination from rat CSF was inhibited by hAβ(1‐42) (Table 2), suggesting that hAβ(1‐40) and hAβ(1‐42) concentrations in CSF would both be affected by changes in elimination activity from CSF. Actually, Aβ(1‐40) and Aβ(1‐42) concentrations in the CSF were decreased in LRP1‐overexpressing mice without Aβ aggregation in brain parenchyma (Zerbinatti et al. 2006). Therefore, LRP1 activity in the choroid plexus should be taken into account in assessing the utility of CSF concentrations of hAβ(1‐40) and hAβ(1‐42) as AD diagnostic markers. As the biophysical properties of hAβ(1‐42) are different from those of hAβ(1‐40), further study is needed to examine the elimination mechanism of hAβ(1‐42) from CSF and to understand its contribution to the hAβ(1‐42) level in CSF.

    In conclusion, our results indicate that hAβ(1‐40) is at least partly eliminated from the CSF via a carrier‐mediated process, in which LRP is involved. Furthermore, LRP‐1 was suggested to play a role in the elimination via uptake by choroid plexus epithelial cells. The LRP‐mediated elimination process from CSF is likely to account for a substantial part of the overall hAβ(1‐40) clearance from the brain, and therefore is expected to influence the concentration of hAβ(1‐40) in CSF and possibly brain ISF. Our results also indicate that TR‐CSFB3 cells represent a good in vitro model to analyze hAβ(1‐40) elimination at the BCSFB.

    Acknowledgements

    This study was supported in part by a Grant‐in‐Aid for Scientific Research on Priority Areas 17081002 from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and a 21st century Center of Excellence (COE) Program grant from the Japan Society for the Promotion of Sciences. This study was also supported in part by the Industrial Technology Research Grant Program from New Energy and the Industrial Technology Development Organization (NEDO) of Japan.

        The full text of this article hosted at iucr.org is unavailable due to technical difficulties.