OUP user menu

Amyloid Efflux Transporter Expression at the Blood-Brain Barrier Declines in Normal Aging

Gerald D. Silverberg MD, Arthur A. Messier PhD, Miles C. Miller ScB, Jason T. Machan PhD, Samir S. Majmudar BA, Edward G. Stopa MD, John E. Donahue MD, Conrad E. Johanson PhD
DOI: http://dx.doi.org/10.1097/NEN.0b013e3181f46e25 1034-1043 First published online: 1 October 2010

Abstract

Reduced clearance of amyloid β peptides (Aβ) across the blood-brain barrier contributes to amyloid accumulation in Alzheimer disease. Amyloid β efflux transport is via the endothelial low-density lipoprotein receptor-related protein 1 (LRP-1) and P-glycoprotein (P-gp), whereas Aβ influx transport is via the receptor for advanced glycation end products. Because age is the major risk factor for developing Alzheimer disease, we measured LRP-1 and P-gp expression and associated transporter expression with Aβ accumulation in aging rats. Quantitative LRP-1 and P-gp microvessel expression was measured by immunohistochemistry (IHC); LRP-1 and P-gp expression were assessed in microvessel isolates by Western blotting. There was an age-dependent loss of capillary LRP-1 across all ages (3-36 months) by IHC (linear trend p = 0.0004) and between 3 and 20 months by Western blotting (linear trend p < 0.0001). There was a late (30-36 months) P-gp expression loss by IHC (p < 0.05) and Western blotting (p = 0.0112). Loss ofLRP-1 correlated with Aβ42 accumulation (p = 0.0121) and verynearly with Aβ40 (p = 0.0599) across all ages. Expression of LRP-1correlated negatively with the expression of receptor for advanced glycation end products (p < 0.0004). These data indicate that alterations in LRP-1 and P-gp expression seem to contribute progressively to Aβ accumulation in aging.

Key Words
  • Aging
  • Alzheimer disease
  • Amyloid
  • Blood-brain barrier
  • LRP-1
  • P-gp
  • RAGE

Introduction

Amyloid β peptide (Aβ) concentrations in the brain interstitial fluid (ISF) are dependent primarily on the bidirectional transport of Aβ across the blood-brain barrier (BBB) in young adults (1). It has been postulated that impaired clearance of Aβ from the brain ISF may be an important factor in the accumulation of brain amyloid in both normal aging and in Alzheimer disease (AD) (2-6). Although there is enzymatic degradation of Aβ in situ, local degradation seems to account for a relatively small amount of Aβ clearance at physiologic concentration in aging and in AD (7-11). Therefore, Aβ clearance is largely a function of transport at blood, ISF, and cerebrospinal fluid (CSF) interfaces.

Transport of Aβ out of the brain occurs by 3 major pathways: (i) across the capillary endothelium of the BBB by receptor-mediated active transport, (ii) via solute diffusion and the bulk flow of CSF, and (iii) by active transport across the choroid plexus epithelium (1, 12-20). We have recently shown, by immunohistochemistry (IHC), ELISA, and Western blotting (WB), that 40- and 42-amino acid amyloid β peptides (Aβ40 and Aβ42) accumulate during normal aging in the Brown-Norway/Fischer (B-N/F) rat and that the Aβ accumulation demonstrated is associated with an increase in amyloid influx transporter expression at the BBB (21). We have also recently shown that impairing the CSF circulation by induced hydrocephalus leads to increased concentrations of Aβ and hyperphosphorylated Tau protein in the brains of elderly rats (22). Blood-brain barrier transport of Aβ out of the ISF across the BBB occurs via a basal endothelial receptor, the low-density lipoprotein receptor-related protein 1 (LRP-1) (17, 23), and via the apical (luminal) transporter P-glycoprotein (P-gp) (24, 25).

Low-density lipoprotein receptor-related protein 1, a member of the low-density lipoprotein receptor (LDLR) family, is expressed on neurons, endothelial cells, and activated glia in the brain (5, 26). Endothelial LRP-1 actively transports a number of molecules, including Aβ, from the ISF across the BBB. Transport of Aβ into the plasma occurs via P-gp, a phosphoglycolate phosphatase receptor that is a member of the multidrug resistance (MDR) gene family (24, 25). P-glycoprotein is only expressed on endothelial cells in the brain and forms a functional part of the BBB. Amyloid β influx transport, from the plasma into the brain ISF, is across the receptor for advanced glycation end products (RAGE) (18).

Amyloid β clearance seems to decline with age. In mice, net transport of radiolabeled Aβ40 across the BBB is reduced by approximately 50%, from 3 to 9 months (17). Whole-brain LRP-1 expression has been shown to decrease with age (27), and there is increasing evidence suggesting that both P-gp and LRP-1 deficits (as well as RAGE overexpression) may be important in the genesis and progression of AD (1, 5, 6, 21, 25, 28). Age is the single most important risk factor in developing nonfamilial AD (29-31). Age also renders the primate brain more vulnerable to Aβ toxicity (32). However, the reasons for these age-associated pathological events are not yet understood. We hypothesize that aging adversely affects the clearance of Aβ by altering the expression of Aβ transport proteins at the BBB and by diminishing CSF production and turnover (2, 4, 21, 22, 33, 34).

In our B-N/F rat model of aging, Aβ begins to accumulate early in life, but concentrations begin to increase markedly between 9 and 12 months. The expression of RAGE decreases between 3 and 9 months and only increases beyond 3-month levels after 12 months (21). Here, we report the decreased expression of endothelial LRP-1 over the whole adult age range and decreased P-gp expression late in the lifetime of the B-N/F rat based on IHC and WB analyses. In addition, we assessed the association of alterations in endothelial LRP-1, P-gp, and RAGE expression with Aβ accumulation over the course of aging in this model.

Materials and Methods

Animals and In Situ Perfusion

Male B-N/F rats (n = 120) aged 3 to 36 months were examined. B-N/F rats were chosen because they are long-lived and suffer less from age-dependent tumor formation than more inbred strains. We chose B-N/F rats, as opposed to larger animals that develop Aβ plaques, because we were primarily studying the aging changes in Aβ BBB transport proteins rather than AD and the life span and availability of the B-N/F rats made them an attractive model. We were able to obtain this rat strain at most ages up to 36 months from the National Institute of Aging colony. The rats were housed in the Central Animal Facility at Rhode Island Hospital and had food and water ad libitum. The Institutional Animal Care and Use Committee at Rhode Island Hospital approved all experiments.

Rats were killed with intraperitoneal pentobarbital (125 mg/kg) and perfused with phosphate-buffered saline (PBS) via left ventricular cannulation. For IHC, 50 rats were then perfused with 4% paraformaldehyde (PFA). For LRP-1 IHC, brains were fixed in a 4% PFA solution in 0.4 mol/L Sorensen's phosphate buffer (pH 7.2) (35) for 24 hours, rinsed in PBS, and then placed in 30% sucrose in 0.1 mol/L Sorensen's phosphate buffer (pH 7.4) until completely immersed. The brains were then cryoprotected. For P-gp IHC, the brains were immersed in 4% PFA and then processed and embedded in paraffin. The remaining 70 rats were used for microvessel isolations (MVIs).

Immunohistochemistry

For LRP-1 IHC, 15 specimens were snap-frozen in liquid N2, embedded in OCT compound (Sakura Finetek USA, Inc, Torrance, CA), and cryosectioned at a thickness of 25 μm. After a quick rinse in Tris-buffered saline with Tween-20 (TBST), sections were quenched in 10% H2O2 for 10 minutes, followed by a 24-hour blocking period with 5% normal goat serum (Vector Laboratories, Burlingame, CA) at 4°C. After overnight incubation at 4°C with a rabbit polyclonal LRP-1 antibody (catalog no. PAB-10774; Orbigen, San Diego, CA), diluted 1:75, the sections were washed with TBST. For the secondary antibody, a goat antirabbit IgG (Vector) diluted 1:500, was applied for 30 minutes. The ABC detection system (Vector) was used, and the tissue sections were stained using 3,3-diaminobenzidine as the chromogen. Sections were mounted, and slides were coverslipped and sealed using Cytoseal, a xylene-based mounting medium (Stevens Scientific, Riverdale, NJ). Primary antibody omission controls were run alongside the other samples to check for nonspecific binding along with positive control tissue (B-N/F rat lung and liver).

For P-gp IHC, paraffin-embedded tissue sections cut at 10 μm were obtained from 35 different specimens. Sections were incubated in an oven at 60°C for 1 hour and then deparaffinized and rehydrated. After a 20-minute pretreatment with hot (85°C) 10-mmol/L citrate buffer (pH 6.0), the sections were quenched for 10 minutes with 3% H2O2 and 50% methanol diluted in distilled water. Nonspecific binding sites were blocked by incubation with 5% normal horse serum (Vector) for 18 hours at 4°C. The sections were incubated overnight at 4°C with mouse monoclonal anti-P-gp (C219, catalog no. ab3364; Abcam, Cambridge, MA), diluted 1:100. The sections were then washed with TBST and subjected to a modified ABC technique using the Vectastain Elite ABC Mouse Peroxidase system (Vector) with 3,3-diaminobenzidine as the chromogen. The slides were then coverslipped and sealed. Primary antibody omission controls and positive controls were as described previously.

Image Analysis

Grayscale images were obtained using an Olympus BH2-RFCA microscope (Olympus America, Inc, Melville, NY) with a 40× objective. Images were acquired with a CoolSNAP cf camera (Roper Scientific, Tucson, AZ). For quantitative LRP-1 expression, 8 random fields were analyzed per specimen (n = 3 per age group, 24 fields per group); for P-gp, 5 or 6 per age group (40 or 48 fields per group) were analyzed to confirm the presence of LRP-1 and P-gp in relation to capillary endothelia. Additional rats and age points were added to the P-gp IHC analysis after the P-gp WB showed a trend toward a decrease at 6 months and a significant decrease between 30 and 36 months. The additional rats were to test whether these changes on WB were supported by IHC. Cross sections of cerebral microvessels 10 to 20 μm in diameter were selected out of each of the 8 random field images acquired per specimen for analyses. Image processing and analysis were performed using NIH Image shareware (National Institutes of Health, Springfield, VA), along with previously reported analytical methods (36, 37). For surface area calculations, all images were thresholded by the same observer at a single sitting to reduce errors associated with this mode of analysis. Statistics were performed using the average of the 8 collected surface areas of capillary endothelial cell immunoreaction product per animal.

Microvessel Isolation

For MVI, the meninges were removed and the brains were placed into ice-cold PBS. The cerebellum, most subcortical structures, and choroid plexus tissues were also removed, leaving the cortex and hippocampus. To obtain sufficient microvessels for analysis, 2 rat brains for a single “n” at each age point were used (i.e. an “n” of 5 represented 10 brains). Microvessels were isolated by homogenizing cortex and hippocampus (approximately 500 mg) in MVI buffer (21). The microvessels were then separated using a basic mechanical separation technique (38). The MVI pellets were treated with protein lysis buffer (Complete Protease Inhibitor; Roche Diagnostics, Mannheim, Germany) and frozen at −80°C in preparation for protein extraction.

Western Blots

The total protein content of each sample was determined using a BCA Protein Assay kit (Pierce, Rockford, IL), with absorbance read at 562 nm. Microvessel isolations were tested for contamination by other cell types by light microscopy after hematoxylin and eosin staining and by WB for neuron-specific enolase and the glial cell marker S100b. Although it is not possible to achieve 100% microvessel purity, we could achieve greater than 95% clean vessels. Only MVIs with less than 3% optical density (OD) readings of the neuronal and glial marker cortical homogenate OD values were used for the WB assessment of LRP-1 and P-gp expression (21).

Approximately 3 to 5 mg of microvessels or rat liver and cortex (as positive controls) was homogenized in protein lysis buffer, and 1 mL of each homogenate was then centrifuged at 10,000 × g for 15 minutes. Protein concentrations were determined using bicinchoninic protein assay reagent (Pierce); 50 μg of each sample was added to NUPAGE sample buffer (Invitrogen, Grand Island, NY). Samples were heated to 95°C for 2 minutes and spun briefly at 1,000 × g. Samples were run on 4% to 12% NUPAGE Novex Bis-Tris gels in running buffer (Invitrogen) at 200 V for 35 minutes, and protein was then transferred wet in NUPAGE transfer buffer (Invitrogen). Membranes were stained with Ponceau S, and then blocking serum (5% nonfat milk; Sigma, St Louis, MO) was applied. The following primary antibodies and dilutions were used: LRP-1, 1:1000 (mouse monoclonal, no. 3501; American Diagnostica, Stamford, CT); P-gp, 1:1000 (mouse monoclonal, no. ab3364; Abcam); S-100b, 1:10,000 (mouse monoclonal, no. ab4066; Abcam); β-actin, 1:10,000 (mouse monoclonal, no. A 5441; Sigma); and neuron-specific enolase, 1:10,000 (rabbit polyclonal, no. ab16873; Abcam). Horseradish peroxidase secondary antibody was used at a concentration of 1:50,000 (antimouse, no. PI2000; Vector). Western blotting detection reagent (Supersignal West Pico Enhanced ECL, no. 1856135-6; Pierce) was added, and blots were developed. Rat cortex or liver controls established the antibody specificities. Negative controls ensured that no protein species could be detected on immunoblots when proteins were incubated with preimmune serum or without the primary antibody. The estimate of OD of the immunoblots is done by a calibration method described in the ImageJ software (NIH shareware). β-Actin along with Ponceau S staining was also used as a control to ensure equal loading of proteins on the gels. All WBs were first standardized to β-actin and Ponceau S before entering the aging analysis. Measurements of all age point OD values were compared with 3-month-old animals as the normalizing control.

Statistical Methods

Mean estimates of each age group with 95% confidence intervals (95% CI) are presented. Differences between age groups in LRP-1, P-gp, RAGE, Aβ40, and Aβ42 were modeled using mixed linear models. Data from Aβ40 and Aβ42 were from a previous publication (21) but were reanalyzed to obtain means for use in later correlations because they were previously modeled as a continuous function. These were modeled after applying a square root transformation, with means back-translated for presentation. Models permitting heterogeneous variances were compared with those assuming homogeneity of variance; the one with the lowest Bayesian Information Criterion was selected. Follow-up comparisons of significant main effects were adjusted using the Holm test. This included the test for linear trends (where present) with group differences; linear trends were not adjusted when group comparisons were not carried out. Adjusted p values are indicated as “adj. p."

For determining relationships between measures of same samples, measurements of more than 1 variable per rat or rat pair were permitted in some instances. In these instances, the relationship between these variables across animals of varying ages was assessed using the Pearson correlation coefficient. For different samples, the relationship between variables drawn from different rats or rat pairs could not be directly assessed because observations were not nested within a rat or rat pair; therefore, this effect was estimated using the Pearson correlation coefficient for points defined by group means of comparable age groups. For example, the group mean for ELISA Aβ42 in rats killed at 3 months was paired with the group mean for WB OD values for LRP-1 of comparably aged rats to form a single observation, with 6 other observations formed using the group means of the other comparable ages at death. A single Pearson correlation coefficient was then used to estimate the strength of the relationship between Aβ40 and LRP-1 across these 6 observed pairs of means, the assumption being that the relationship being observed among group means was driven by a similar relationship occurring within individuals. These correlations have extremely low statistical power because of the sample sizes determined by the number of age groups.

Results

LRP-1 Expression by IHC

Low-density lipoprotein receptor-related protein 1 immunoreactivity (IR) was observed in the rat brains across all age groups tested. The immunostained cell types included microvessel endothelial cells, neurons, and glia, with differing degrees of staining noted as a function of age. There was a progressive decrease in microvessel LRP-1-IR with advancing age, whereas both neuronal and glial LRP-1-IR had a tendency to increase as the animals aged. The most dramatic difference is evident when juxtaposing representative tissue sections from 3- and 34-month-old rats. In 3-month-old rats, on average, there was strong LRP-1-IR in the microvasculature (Fig. 1A), whereas there was notably lower IR in the microvasculature of 34-month-old rats (Fig. 1B).

FIGURE 1.

Expression of capillary low-density lipoprotein receptor-related protein 1 (LRP-1) from 3 to 34 months. (A) At 3 months, there is a robust expression of LRP-1 on the basal surface of endothelial cells (arrow) and evidence of neuronal staining (asterisks). (B) At 34 months, LRP-1 expression on the endothelium is significantly reduced (arrow), whereas neuronal expression (asterisks) seems to be slightly increased. (C) Mean normalized LRP-1 microvessel surface area with 95% confidence interval plotted as a function of age at death with a linear trend line. There is a significant decreasing linear trend of −0.35 μm2/mo (95% confidence interval, −0.48 to −0.23 μm2/mo; p = 0.0004).

Quantitative IHC measurements for microvessel LRP-1 expression were expressed as normalized surface area in squared micrometers. Low-density lipoprotein receptor-related protein 1-IR showed a progressive decrease of approximately −0.35 μm2/mo (95% CI, −0.48 to −0.23 μm2/mo; linear trend p = 0.0004; Fig. 1C). The normalized surface area staining at 3 months was 23.68 μm2 (95% CI, 20.30-27.06 μm2). At 12 months, the measurement was reduced to 20.64 μm2 (95% CI, 18.95-22.34 μm2), and at 23 months, it was 15.63 μm2 (95% CI, 14.08-17.18 μm2). At 30 months, it was 16.10 μm2 (95% CI, 13.98-18.23 μm2), and at 34 months, it was 11.70 μm2 (95% CI, 8.03-15.38 μm2).

Neuronal LRP-1 immunostaining in a 3-month-old rat is shown in Figure 2A. There was much stronger neuronal LRP-1-IR at 34 months (Fig. 2B). Glial IR was particularly evident in the molecular layer of the dentate gyrus of hippocampus (Fig. 2C). There was a mild increase in glial LRP-1-IR in the 34-month-old versus the 3-month-old rats (Fig. 2D).

FIGURE 2.

Neuronal and glial expression of low-density lipoprotein receptor-related protein 1 (LRP-1). (A) A representative photomicrograph of cortical neuronal LRP-1 immunoreactivity (IR) at 3 months. Low-density lipoprotein receptor-related protein 1-IR of neurons is present but is not intense. (B) At 34 months, cortical neuron LRP-1-IR is increased. (C) There is glial LRP-1-IR in the molecular layer of the dentate gyrus (arrow). (D) Glial LRP-1-IR in the molecular layer of the dentate gyrus is mildly increased at 34months (arrows).

P-gp Expression by IHC

P-glycoprotein-IR was confined to the brain capillary endothelium in all age groups with no significant differences from 3 to 30 months (Fig. 3A); no nonendothelial cells appeared to be stained in any of the age groups. There was, however, a significant decrease in microvascular IR by 36 months (Fig. 3B). The uniform expression of P-gp-IR seen across the luminal surface of transverse sections through microvessels was less in the 36-month-old animals. In many cases, a single luminal side of a transverse section stains, whereas the luminal region directly across from it was devoid of P-gp-IR at 36 months.

FIGURE 3.

Expression of endothelial P-glycoprotein (P-gp) from 3 to 36 months. (A) At 3 months, there is abundant capillary P-gp staining (arrows). (B) At 36 months, there is decreased endothelial P-gp staining. (C) Mean normalized P-gp microvessel surface area staining with 95% confidence interval plotted with means joined as a function of age at death; 36-month-old rats were significantly lower than all other ages (adjusted p < 0.05 for all comparisons). No other age groups differed significantly (adjusted p ≥ 0.05).

Quantitative IHC measurements for microvessel P-gp-IR over time were very different from the LRP-1-IR pattern. There were no significant differences among ages 3 to 30 months (adj. p ≥ 0.05), but 36-month-old rats had markedly lower IR than all other age groups (adj. p < 0.05; Fig. 3C).

WB for LRP-1 and P-gp Receptor Expression

A representative WB for P-gp, LRP-1, and β-actin is shown in Figure 4A. The expression of LRP-1 in MVI measured by WB OD values was consistent with microvessel LRP-1-IR assessed by IHC, that is, both showed a continued decrease in receptor expression with advancing age. Relative LRP-1 expression by WB was reported as blot ODs (Fig. 4B). There was a significant linear decrease in LRP-1 of approximately -0.04 OD units/mo (95% CI, -0.05 to -0.03 OD units/mo) until 20 months (adj. p < 0.0001), with no significant differences between 20, 30, and 36 months (adj. p ≥ 0.05). The OD values were as follows: 0.87 (95% CI, 0.77-0.96) at 3 months, 0.77 (95% CI, 0.50-1.05) at 6 months, 0.79 (95% CI, 0.61-0.97) at 9 months, 0.59 (95% CI, 0.30-0.87) at 12 months, 0.19 (95% CI, 0.09-0.29) at 20 months, 0.16 (95% CI, 0.08-0.24) at 30 months, and 0.14 (95% CI, 0.05-0.24) at 36 months.

FIGURE 4.

Low-density lipoprotein receptor-related protein 1 (LRP-1) and P-glycoprotein (P-gp) expression measured by immunoblotting of microvessel proteins. Western blot (WB) OD values are plotted against age. (A) Representative WB for LRP-1, P-gp, and β-actin. (B) Mean LRP-1 OD with 95% confidence interval (CI) plotted as a function of age at death. There is a significant decreasing linear trend of −0.04 OD values/mo (95% CI, −0.05 to −0.03 OD values/mo) between 3 and 20 months (adjusted p < 0.0001) and there were no significant differences between 20, 30 and 36 months (adjusted p ≥ 0.05). (C) Mean P-gp OD values with 95% CI plotted with means joined as a function of age at death; 36-month-old rats had significantly lower values versus30-month-old rats (adjusted p = 0.0112), but no other age groups differed (adjusted p ≥ 0.05).

Western blot of MVI proteins for P-gp expression showed a different expression profile from that of LRP-1 (Fig. 4C) but correlated with P-gp IHC. The 36-month-old rats were significantly lower than 30-month-old rats (adj. p = 0.0112), but no other age groups differed (adj. p ≥ 0.05). The mean P-gp OD values were as follows: 0.74 (95% CI, 0.53-0.95) at 3 months, 0.49 (95% CI, 0.28-0.70) at 6 months, 0.81 (95% CI, 0.60-1.02) at 9 months, 0.72 (95% CI, 0.51-0.03) at 12 months, 0.62 (95% CI, 0.41-0.83) at 20 months, 0.92 (95% CI, 0.72-1.13) at 30 months, and 0.36 (95% CI, 0.16-0.57) at 36 months.

Mean Aβ40 and Aβ42 concentrations (from reference [21]) versus WB mean expression of LRP-1 and P-gp over time are plotted in Figure 5. Parallel to the decrease in efflux transporter expression with age, there was a concomitant increase in brain Aβ concentration.

FIGURE 5.

Mean Western blot OD values for low-density lipoprotein receptor-related protein-1 (LRP-1; dashed line, diamonds, left ordinate), P-glycoprotein (P-gp; solid black line, squares, left ordinate), ELISA Aβ40 (solid gray line, squares, right ordinate), and Aβ42 (solid black line, circles, right ordinate) as a function of age. Results for LRP-1 and P-gp are from aset of rats distinct from those used for Aβ40 and Aβ42. As LRP-1 expression decreased from 3 months onward, Aβ40 and Aβ42 increased. P-glycoprotein expression did not change significantly until after 30 months.

Correlations Among LRP-1, P-gp, RAGE, and Aβ

Sufficient tissue was available to take WB measures of LRP-1, P-gp, and RAGE (RAGE data from [21]) within given rat pairs, but there were insufficient tissues in these same rats to take additional measures (e.g. IHC for LRP-1 and P-gp and ELISA for Aβ40 and Aβ42). Therefore, the correlations were tested based on pairing means from comparably aged groups from 2 different sets of rats. Although these correlations have extremely low statistical power, they show both significant correlations and trends. There was a negative correlation between age group means for Aβ42 and LRP-1 WB OD values, r = -0.91, p = 0.0121 (Fig. 6A). There was no clear correlation between LRP-1 and Aβ40, although it approached significance, r = -0.80, p = 0.0599 (Fig. 6B). There were no significant correlations on WB between P-gp and LRP-1 over the lifetime of the B-N/F rat.

FIGURE 6.

Plots of mean Western blot (WB) OD values for low-density lipoprotein receptor-related protein-1 (LRP-1) as a function of Aβ40 and Aβ42 ELISA measurements and as a function of receptor for advanced glycation end products (RAGE) OD values. (A) MeanLRP-1 plotted as a function of mean Aβ42 from comparably aged groups of rats. There was a significant negative correlation between age group means (r = −0.91, p = 0.0121). (B) Mean LRP-1 plotted as a function of mean Aβ40 from comparably aged groups of rats. The negative correlation between age group means approached significance (r = −0.80, p = 0.0599). (C) Within animal WB, RAGE plotted as a function of LRP-1 for all ages (gray open circles) and from 9 to 36 months of age (MOA; black closed circles). Expressions of RAGE and LRP-1 are negatively correlated (gray open circles and dashed line, r = −0.57, p = 0.0009), with a higher correlation within animals at ages across which RAGE increases (solid black circles and line, r = −0.7, p = 0.0004).

There was a negative correlation between LRP-1 expression and RAGE expression, r = -0.57, p = 0.0009. This correlation increased when considering only animals at ages spanning 9 to 36 months, when both Aβ and RAGE expression seemed to be increasing, r = -0.71, p = 0.0004 (Fig. 6C). There was also a significant negative relationship between levels of LRP-1 and RAGE on quantitative IHC as well, r = -0.74, p = 0.0016.

Discussion

An increasing amyloid burden, particularly the accumulation of toxic soluble Aβ oligomeric and intermediate forms, seems to be an important component of AD pathogenesis (39-43). Pathological studies have, however, also shown that amyloid plaques and neurofibrillary tangles occur in the brains of aging subjects without dementia (44, 45). There seems to be a continuum in Aβ and Tau protein accumulation from normal aging to AD, although the molecular basis for the transition from normal aging to AD is not understood. Abnormalities in amyloid precursor protein (APP) and Aβ cellular processing likely contribute to the increased concentration of extracellular Aβ in aging and AD brains (41). Mutations in APP have also been shown to increase brain Aβ concentrations (42), and apolipoprotein E isoforms play a role in Aβ transport and accumulation (19, 43).

Our model of aging demonstrates an increase in brain Aβ40 and Aβ42 over the lifetime of the B-N/F rat, the most rapid rise being between 9 and 12 months (21). The early rise in Aβ concentration in our model also seems to correlate with an early decline in endothelial LRP-1 expression, albeit with an age-related increase in neuronal and glial LRP-1 expression. After 9 months, the rapid increase in both Aβ peptides seems to be related to a further decrease in endothelial LRP-1 expression, although RAGE expression also begins to rise at this time (21). Very late in the life of the B-N/F rat (between 30 and 36 months), the decline in P-gp expression also seems to contribute to Aβ accumulation. These changing efflux and influx expression profiles likely diminish amyloid clearance from the CNS. They might even reverse amyloid efflux, that is, the net transport of Aβ then being from the vascular compartment back into the brain. The statistical correlations between endothelial LRP-1 expression decline, endothelial RAGE expression increase, and the increasing accumulation of Aβ suggest a relationship among these measurements. That there is a causal relationship can only be inferred based on the known biologic activities of these Aβ transporters.

P-glycoprotein expression did not correlate with amyloid accumulation or with the alterations in LRP-1 or RAGE expression over the lifetime of the B-N/F rat. Other than the late decrease in expression, P-gp seemed to rise (although not significantly) up to 30 months. Others have also reported no significant difference in P-gp expression with aging in Fischer-344 rats aged 3, 13 to 14, and 25 to 26 months (46). We also found no significant changes before the age of 30 months and only saw decreased expression between 30 and 36 months. Inasmuch as a similar age-related P-gp decrease in association with local amyloid deposition is also seen in human aging (28), P-gp loss may be part of the amyloid efflux failure that facilitates the pathological conversion from normal aging to AD.

We found significant inverse correlations not only between LRP-1 expression and increased Aβ42 concentration but also between decreasing LRP-1 expression and the later increase in RAGE expression. The reciprocal changes in these 2 Aβ transporters, as well as their temporal sequence, suggest a common or related signaling pathway or ligand-receptor interaction. For example, the early increase in Aβ (or perhaps another ligand transported by LRP-1) might upregulate subsequent RAGE expression. Indeed, ligand binding to RAGE increases its expression (47).

The LDLR gene family is highly conserved across species (48). It acts both as a multifunctional scavenger and as a signaling receptor. Low-density lipoprotein receptor-related protein 1 binds a wide variety of ligands, including apoE, α2-macroglobulin, APP, and Aβ (1, 48-51). Because LRP-1 binds and internalizes a number of structurally diverse ligands, another role for it may be to remodel the cell membrane proteome to meet the demands of a changing microenvironment (52).

There are 2 possible LRP-1-mediated Aβ clearance mechanisms: LRP-1-mediated Aβ uptake into cells followed by local degradation, or Aβ transport out of the brain across the BBB into the peripheral circulation where Aβ is degraded in the liver. Although the involvement of LDLR family members in AD is not yet fully understood, it is clear that they can directly affect APP production, Aβ degradation and Aβ transport out of the brain (53). It is not yet clear why LRP-1 expression decreases with aging, however. Expression of LRP-1 may be reduced by sublethal chemical stress (54), and diabetic mice show a decrease in BBB LRP-1 expression that correlates with decreased efflux of radiolabeled Aβ (55). Moreover, activation of protein kinase C-α phosphorylates and later downregulates LRP expression in glioblastoma cells (56), whereas glycation of low-density lipoproteins in vascular smooth muscle cells upregulates LRP-1 expression (57). Expression of LRP-1 is also regulated by the low-density lipoprotein receptor-related protein-associated protein 1, insulin 1 and by APP (58-60). It is also of some interest that increased APP and increased gamma cleavage of APP downregulate LRP-1 (60).

Which receptor ligands and signaling pathways are responsible for LRP-1 downregulation with aging are not known. The increase in RAGE expression in aging has been linked to activation of nuclear factor κB and to demethylation of the gene promoter in aged human brains (61, 62). Epigenetic alterations, such as DNA methylation and histone modification in genes associated with AD pathology, have been proposed as mechanisms related to both aging and AD (63). Whether activation of protein kinase C-α and/or increased methylation of the LRP-1 promoter occurs with advancing age is not yet known.

P-glycoprotein is a functional component of the BBB and limits CNS penetration by various molecules, including chemotherapeutic agents, small peptides, antibiotics, human immunodeficiency virus protease inhibitors, and antidepressant drugs (64, 65). In mice, P-gp is encoded by both mdr1a and mdr1b, which have 90% sequence homology to each other and 80% to human MDR1. Sphingosine kinase 1 has been implicated in the regulation of P-gp transport activity in brain endothelial cells (66). In vitro biochemical data, using an inside-out membrane vesicle preparation, showed Aβ to be a substrate for P-gp transport (24). There is also a region-specific level of Aβ deposition in AD brains inverse to the level of endothelial P-gp expression on IHC (28). Amyloid β removal from the brains of P-gp-null mice was reduced, although not eliminated, suggesting that Aβ transport was at least partially mediated by endothelial P-gp. Also, immediate inhibition of P-gp activity in wild-type mice increased Aβ levels in brain ISF (25). Regulation of P-gp expression in both rat and human brain capillary endothelial cells seems to be by Wnt-β catenin signaling (67). Others have shown involvement by nuclear factor κB in regulating P-gp expression (68, 69). The causes of reduced P-gp expression late in the life of the B-N/F rat are not yet known.

There is a parallel between altered brain endothelial LRP-1 and RAGE expression in the aging rat and findings in AD brains (5, 6). In Braak and Braak stage V and VI human autopsy material, we demonstrated a loss of capillary LRP-1 expression, a concomitant increase in neuronal LRP-1 expression, and an overexpression of capillary RAGE (5). We also demonstrated a parallel between RAGE expression and amyloid burden in AD brains (6). The present findings of an early and persistent decrease in capillary LRP-1 expression, a late decrease in P-gp expression, and an intermediate increase in RAGE expression suggest that age-related changes in the cerebral vasculature (70) can lead to an inability of the CNS to clear Aβ in normal aging and in AD. The present results also support the concept that normal aging and AD are on a continuum of metabolite clearance failure. It will be important to analyze the cellular signaling mechanisms behind these deleterious alterations in efflux and influx transport proteins, as this may shed some light on the poorly understood and very complex relationships between aging and AD. Such studies may also uncover possible therapeutic targets.

Acknowledgments

The authors thank Tracey Brooks and Tom Davis (Department of Medical Pharmacology, University of Arizona College of Medicine, Tucson, AZ) for sharing the composition of the MVI buffer, and Stephanie Slone and Elizabeth Kenney (Department of Clinical Neuroscience, Aldrich Labs, Providence, RI) for the microvessel isolations.

Footnotes

  • Funding for this study was provided by National Institutes of Health grant 1RO1 AG027910-01A1. Capital equipment purchases were provided by the Saunders Family Fund and the Rae and Jerry Richter AD Research Fund at the Neurosurgical Foundation, Rhode Island Hospital, and Warren Alpert Medical School, Brown University.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
View Abstract