Blocking the Interaction between Apolipoprotein E and Aβ Reduces Intraneuronal Accumulation of Aβ and Inhibits Synaptic Degeneration

Materials and Reagents

All Aβ peptides were custom synthesized in the WM Keck Proteomic Facility of Yale University (New Haven, CT). Unlabeled peptides were provided by the laboratory of Dr. James I. Elliott (Yale University), and fluorescein-tagged Aβ1-40 was synthesized by the laboratory of Dr. Janet Crawford (Yale University). PCR primers were custom synthesized by Gene Link (Hawthorne, NY). Recombinant human APOE4 was bought from Leinco Technologies, Inc. (St. Louis, MO). Cell culture plasticware, including Transwell inserts with 0.4-μm pore size, were purchased from Corning Incorporated (Corning, NY). Cell culture media and human aggregated Aβ enzyme-linked immunosortbent assay (ELISA) kit were purchased from Invitrogen Life Technologies (Carlsbad, CA). Complete Protease Inhibitor Cocktail was obtained from Roche Applied Science (Indianapolis, IN). A kit for bicinchoninic acid assay and the SuperSignal chemiluminescent reagent for immunoblotting were obtained from Pierce Biotechnology Inc. (Rockford, IL). Nitrocellulose membrane and horseradish peroxidase-conjugated secondary antibodies (Abs) for immunoblot analysis were sheep anti-mouse and donkey anti-rabbit (GE Healthcare Bio-Sciences Corp., Piscataway, NJ). Immobilon polyvinylidene difluoride membrane and Microcon centrifugal filter devices were purchased from Millipore Corporation (Billerica, MA). Autoradiography films X-Omat Blue XB-1 were bought from Eastman Kodak Company (New Haven, CT). Mouse on mouse (MOM) peroxidase kit, MOM blocking reagent, streptavidin/biotin blocking kit, and biotinylated secondary Abs included goat anti-mouse IgG and goat anti-rabbit IgG used for immunocytochemistry and immunohistochemistry (Vector Laboratories, Ltd., Burlingame, CA). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO). Primary Abs were obtained from several different sources, and their listing, together with working dilutions, is provided in Table 1. HJ6.1 is a novel monoclonal antibody (mAb) against murine ApoE. It was generated with ApoE lipoprotein particles purified by affinity chromatography from media of primary astrocytic cultures established from C57BL/6 mice. Purified astrocyte-secreted ApoE lipoprotein particles were injected with complete Freund adjuvant to ApoE KO mice. For an initial screening of Abs, supernatant fluids from hybridoma cells (approximately 2000 wells) were added to 96-well plates coated with purified astrocyte-secreted ApoE lipoprotein particles. With the use of anti-mouse IgG-horseradish peroxidase as a detection antibody, we initially identified 36 clones and then subcloned them to find those that performed well in several biochemical and immunohistochemical assays.

Mice

All mouse care and experimental procedures were approved by Institutional Animal Care and Use Committees of the New York University School of Medicine and the Washington University School of Medicine. Cultures of primary hippocampal neurons and astrocytes were established from pups of C57BL/6 mice, whose breeding pairs were obtained from Charles River Laboratories (Wilmington, MA). Additional astrocytic cultures were established from ApoE KO mice described previously.¹⁷ To develop the APP_SW/PS1_dE9/ApoE KO line, ApoE KO mice were doubled crossed with APP_SW/PS1_dE9 mice (line 85, stock number 004462; The Jackson Laboratory, Bar Harbor, ME). In APP_SW/PS1_dE9 mice, the mouse amyloid precursor protein (APP) harboring the human Aβ sequence with the double Swedish familial AD mutation K594M/N595L and human presenilin 1 (PS1) with exon 9 deletion (dE9) are expressed under the control of the mouse prion protein promoter and transmitted as a single Mendelian locus. APP_SW/PS1_dE9/ApoE KO founders were identified by PCR on MyiQ2 real-time PCR system (Bio-Rad, Hercules, CA) by using pairs of primers for PS1 transgene (5′-TAAGTCAGTCAGCTTTTATACCCGGAAGGA-3′ and 5′-CAGGAGGATAGTCATGACAACAATGACACT-3′) and by showing lack of PCR product with APOE primers (5′-AGACGCGGGCACGGCTGT-3′ and 5′-CTCGCGGATGGCGCTGAG-3′). APP_SW/PS1_dE9/ApoE KO mice were then crossed back several times with ApoE-TR mice that carried one of the three human alleles (APOE2, APOE3, or APOE4) until the expression of APP_SW/PS1_dE9 transgene on the homozygote background of each of the three human ApoE isoforms was established. Each ApoE-TR mouse line contains mouse regulatory sequences and the noncoding murine exon 1 surrounding the inserted human exons 2′, 3′, and 4′.^18,19 Therefore, these mice express the human ApoE protein at physiological levels and retain the endogenous regulatory sequences required for modulating ApoE expression. APP_SW/PS1_dE9/ApoE-TR lines were maintained by breeding with ApoE isoform-matched ApoE-TR mice. Offspring were identified by PCR. The human ApoE isoform status was verified by restriction fragment length polymorphism analysis of the human APOE gene PCR product.²⁰ Animals were bred and housed in a germ-free facility with 12/12-hour light/dark cycle and ad libitum food and water access. Given documented sex differences in Aβ deposition in the APP_SW/PS1_dE9 strain,²¹ the load of Aβ plaques and intraneuronal Aβ accumulation were quantified only in female mice. All mice that were compared were on the same genetic background.

Handling of Synthetic Aβ Peptides

Aβ peptides were synthesized on solid-phase support and purified by reverse-phase high-pressure liquid chromatography, as previously described.^16,22 Full-length sequences of Aβ1-40, Aβ1-42, and Aβ1-40 amino-terminally tagged with 5,6-carboxyfluorescein (FITC-Aβ1-40) were synthesized with l-amino acids, whereas end-protected Aβ12-28P (NH₂-VHHQKLPFFAEDVGSNK-COOH) and its scrambled version Aβ12-28P_S (NH₂-QGKFSDHVNEPHFAVKL-COOH) were synthesized with d-amino acids as previously described.^16,22 The purity of each batch of peptides was verified by mass spectrometry. Aβ peptides were treated with 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), which renders peptides monomeric with minimal β-sheet content,^4,16 and lyophilized. Lyophilized aliquots of Aβ peptides were stored at −80°C. Immediately before commencing cell culture experiments, Aβ peptides were directly reconstituted in cell culture media to achieve the concentration used in particular experiments.

Aβ oligomers were prepared according to established protocols^4,23,24 by incubating 100 μmol/L Aβ1-42 in serum and phenol-free Dulbecco’s modified Eagle’s medium mixed 1:1 with Ham’s F12 medium, then centrifuged with Microcon YM-10 concentrators to remove unaggregated Aβ monomers.

Cell Culture Preparation

Cultures of primary hippocampal neurons and astrocytes were established from pups within 24 hours from birth (P0 to P1). Animals were sacrificed by decapitation, and their brains were immediately removed from skulls in aseptic conditions. To establish cultures of primary hippocampal neurons, the hippocampi were dissected from both hemispheres under AmScope stereoscopic microscope (AmScope, Chino, CA). Fragments of hippocampal tissue were placed in the ice-cold Hank's balanced salt solution free of Ca²⁺ and Mg²⁺ and gently cut further into fine pieces that were then treated with 0.05% trypsin in Hank's balanced salt solution for 10 minutes at 37°C to obtain cell suspension. After the addition of 0.1% type-I-S trypsin inhibitor and 0.05% DNase I, the cell suspension was centrifuged at 500 × g for 30 seconds. The resulting pellet was triturated, centrifuged again at 500 × g for 5 minutes, and resuspended in the minimal essential medium (MEM) supplemented with heat-inactivated 10% fetal bovine serum (FBS), 0.5 mmol/L glutamine, 50 μg/mL streptomycin, and 50 U/mL penicillin. Cell suspension was then filtered through a 70-μm mesh, and the neurons were manually counted with the use of a hemotocytometer (Hausser Scientific, Horsham, PA) and seeded on poly-l-lysine–coated 12-well plates. For biochemical analysis, 5 × 10⁵ neurons were seeded directly on the bottom of each well, whereas, for immunohistochemistry, removable poly-l-lysine–coated round coverslips were placed on the bottom of the wells, and the number of seeded neurons was reduced to 1 × 10⁵ per well. The neurons were allowed to adhere to the surface for 4 hours, and then the medium was replaced with the serum-free Neurobasal medium containing 2% B27 supplement, 0.5 mmol/L glutamine, 50 μg/mL streptomycin, and 50 U/mL penicillin. At 3 days in vitro (DIV), 5 μmol/L cytosine-β-D-arabinoside was added to inhibit division of nonneuronal cells. The neurons were fed by replacing one-third of the medium volume every 72 hours. At 7 DIV, one well from each preparation of primary hippocampal neurons was randomly selected to determine the purity of the culture and was immunostained against a neuronal-specific marker, the microtubule-associated protein 2 (MAP2), followed by nucleic acid stain DAPI (Figure 1C). The ratio of MAP2-positive to DAPI-positive cells was then calculated to assess culture purity. Those cultures that showed a purity of at least 90% were grown further until 14 DIV when neurons reach their mature form and express functional glutamatergic receptors.²⁵

Cultures of astrocytes were established from the brain cortex according to the protocol of Leroux et al.²⁶ After stripping the brain from the meninges and visible brain vessels, the cortical mantle was separated from the white matter and subcortical structures and gently cut into fine pieces in ice-cold Hank's balanced salt solution free of Ca²⁺ and Mg²⁺. Cell suspension was prepared by gentle trituration of cortical tissue and then centrifuged for 5 minutes at 500 × g. The pellet was resuspended in MEM supplemented with 20% FBS, 2 mmol/L glutamine, 50 μg/mL streptomycin, and 50 U/mL penicillin, and the cells were introduced to 25-cm² flasks. Two days after establishing the astrocytic culture, a single dose of 5 μmol/L cytosine-β-D-arabinoside was added. After 6 days of growth, the FBS content in MEM was decreased to 10%, and astrocytes were grown until confluence. To determine the purity of the astrocytic cultures, the cells were seeded on a coverslip, fixed with methanol, and immunostained against glial fibrillary acidic protein and counterstained with DAPI. Cultures containing 90% or more glial fibrillary acidic protein-positive astrocytes were used for co-culture experiments (Figure 1B). Production of ApoE by astrocytes was confirmed with Western immunoblot analysis with HJ6.1 mAb against murine ApoE in astrocytic lysates and media (Figure 1D).

To set up the neuronal-astrocytic co-culture system, astrocytes were harvested with the use of 0.25% trypsin and 0.03% EDTA, suspended in MEM with 10% FBS, and counted with the use of a hemocytometer. Astrocytes were then plated on the top side of the poly-l-lysine–coated polycarbonate membranes of the Transwell inserts²⁷ (2 × 10⁴ per insert) and cultured in MEM supplemented by 10% FBS, 2 mmol/L glutamine, 50 μg/mL streptomycin, and 50 U/mL penicillin for 48 hours followed by serum-free Neurobasal medium supplemented with 2% B27 and 0.5 mmol/L glutamine for 24 hours before combining the cultures. Transwell inserts with attached astrocytes were transferred into 12-well plates and positioned 2 mm above the neuronal monolayer growing on the bottom of the well (Figure 1A). Pores in the polycarbonate membranes of the Transwell insert allow for free passage of astrocyte-derived molecules, including lipidated ApoE complexes. The co-culture was maintained together in serum-free Neurobasal medium supplemented with 2% B27, 0.5 mmol/L glutamine, 50 μg/mL streptomycin, and 50 U/mL penicillin for 24 hours before commencing experiments. The astrocytes were always combined with neurons when neurons were 14 DIV, and the co-culture experiments were commenced when neurons were 15 DIV. The ApoE uptake by the primary hippocampal neurons was confirmed by Western immunoblot detection of ApoE in lysates from neurons co-cultured with astrocytes, whereas in the lysates of primary hippocampal neurons grown as a monoculture the ApoE signal was absent (Figure 1D).

Treatments of Neuronal Monocultures and Co-Culture Systems with Aβ Peptides

Monomerized Aβ1-40 and Aβ1-42 peptides were diluted with the Neurobasal medium to 10 μmol/L final concentration and added without or with 10 μmol/L Aβ12-28P or a control peptide Aβ12-28P_S to neuronal-astrocytic co-cultures or neuronal monocultures for 24 or 72 hours. In confocal microscopy experiments designed to directly visualize intraneuronal Aβ accumulation, Aβ1-40 was traced with 10% FITC-Aβ1-40. In the experiment designed to investigate the effect of ApoE on intraneuronal degradation of internalized Aβ, 10 μmol/L Aβ1-40 was added to the medium for 6 hours, and then the cells were further grown in the Aβ-free medium during the 48-hour washout period. At the conclusion of each experiment, co-cultures were disassembled by removing Transwell inserts with attached astrocytes, and the primary hippocampal neurons grown on the bottom of the well were subjected to lysis or fixation for biochemical or immunohistochemical analyses, respectively. Each treatment experiment in neuronal-astrocytic co-cultures or neuronal monocultures was repeated three to six times.

Western Blot and Dot Immunoblot Analysis

Primary hippocampal neurons were washed three times with PBS at 37°C and gently treated with 0.1% trypsin and 0.03% ethylenediaminetetraacetic acid in PBS to remove remnants of noninternalized Aβ attached to their external surfaces.²⁸ Neurons were then harvested with ice-cold lysis buffer that contained 50 mmol/L Tris-HCl, pH 7.4, 150 mmol/L NaCl, 1 mmol/L ethylenediaminetetraacetic acid, 1% Nonidet-P40, 0.1% SDS, 0.2% diethylamine, 1 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL Complete Protease Inhibitor Cocktail supplemented with leupeptin, antipain, and pepstatin (5 μg/mL each).²² The lysates were collected into 2-mL low-adhesion tubes, further homogenized with 10 strokes of a Teflon pestle, and centrifuged at 10,000 × g for 20 minutes at 4°C. The pellets containing cell debris were discarded, and the total protein concentration in the supernatant fluid was determined with the bicinchoninic acid assay. Samples of cell lysates that contained equal amounts of total protein were titrated with reducing Laemmli buffer to achieve equal volumes, boiled for 5 minutes, and then subjected to SDS-PAGE with the use of 15% gels. Under afore-described reducing conditions, most Aβ contained in the cell lysates appeared as monomers and dimers at the bottom of the membrane, which facilitated its densitometric quantification (Supplemental Figure S1, A and B). After Western blot transfer, nitrocellulose membranes were immunoblotted with 1:1 mixture of 6E10 and 4G8 mAbs²⁹ followed by horseradish peroxidase-conjugated sheep anti-mouse secondary Ab in 1:3000 dilution and the application of SuperSignal chemiluminescent reagent. To assure that Aβ12-28P, which has sequence overlap with the 4G8 epitope, does not alter 4G8 signal, a control experiment was conducted whereby synthetic Aβ1-40 or Aβ1-42 peptides were loaded on the gel alone or premixed with Aβ12-28P, and the membranes were immunoblotted with the 6E10/4G8 mixture, 6E10, and 4G8 alone, and mAbs HJ2 and HJ7.4, which are specific for C-termini of Aβ_X-40 and Aβ_X-42 peptides,³⁰ respectively (Supplemental Figure S2A). There was no evidence that Aβ12-28P changes detectability of Aβ1-40 or Aβ1-42 × 6E10 and 4G8, whereas Aβ12-28P alone was unrecognized by these mAbs. Furthermore, selected membranes that represented various treatment conditions were stripped and reprobed with Aβ C-terminal–specific mAbs that obtained a pattern of immunostaining comparable with that of the 6E10/4G8 mixture (Supplemental Figure S2B).

For detection of ApoE in lysates of primary hippocampal neurons and astrocytes and in the conditioned media, the samples were subjected to 10% SDS-PAGE under reducing conditions, and the membranes underwent immunoblot analysis with HJ6.1 mAb. For dot blot detection of Aβ oligomers, samples of neuronal lysates that contained equal amounts of total protein were titrated with PBS to achieve equal volumes of 20 μL and were blotted under vacuum onto polyvinylidene difluoride membranes. The membranes were immunoblotted with A11 oligomer-specific polyclonal Ab.³¹ Samples that contained preparations of HFIP monomerized Aβ1-42 and in vitro oligomerized Aβ1-42 were used as negative and positive controls, respectively. Lysates from 15-DIV primary hippocampal neurons cultured in the absence of Aβ peptides, astrocytes, and human recombinant ApoE4 were used as additional controls (Supplemental Figure S2C).

Densitometric Analysis of Immunoblots

Autoradiography films were converted into 8-bit grayscale digital files with the use of a Epson Perfection 4990 scanner (Epson America, Long Beach, CA) and Adobe Photoshop version CS4 (Adobe Systems, San Jose, CA) and saved in a TIFF format with a resolution of 600 dpi. Densitometric analysis of Aβ Western and dot blot signals was performed with NIH Image J software version 1.42 (Bethesda, MD) as previously described.^32,33

Quantification of Aβ Oligomers by ELISA

Samples of neuronal lysates that contained 15 μg of total protein were titrated with PBS to achieve equal volumes. Concentration of Aβ oligomers in the lysate samples was determined with Human Aggregated β-Amyloid ELISA Kit, following the manufacturer-provided manual. After background subtraction, absorbance values from serially diluted samples of aggregated Aβ provided by the manufacturer as a part of the ELISA kit were used to generate a standard curve in GraphPad Prism version 5.02 (GraphPad Software, Inc., San Diego, CA) with the use of a nonlinear curve-fitting algorithm. Concentration of Aβ oligomers in samples of neuronal lysates was determined by comparing their absorbance values against the standard curve and were given in nanograms per milligram of the total protein in the lysate.

Immunocytochemistry and Quantification of Synaptic Protein Expression

Coverslips with attached primary hippocampal neurons were immersed three times in PBS at 37°C, then in 80% ice-cold methanol for 10 minutes to fix the neurons, and washed again three times in PBS. Nonspecific bindings of the primary Abs and streptavidin were blocked with MOM blocking mixture for 1 hour, followed by the streptavidin/biotin blocking kit for 30 minutes. Intraneuronal organelles were identified with a panel of Abs against organelle-specific markers: MAP2 for the cytoskeleton, early endosomal antigen 1 for early endosomes, Rab7 for late endosomes, Cathepsin D for late endosomes and lysosomes, and heat shock protein 60 for mitochondria (Table 1). Synaptic proteins specific for excitatory glutamatergic synapses of primary hippocampal neurons, including the NR1 subunit of the N-methyl-d-aspartic acid receptor (NMDAR), postsynaptic density protein 95 (PSD-95), which is functionally and structurally associated with NMDARs, and a major synaptic vesicle protein synaptophysin were immunodetected with the use of specific mAbs listed in Table 1. We used 1:1000 biotinylated secondary Abs followed by 1:500 cyanine 3-conjugated streptavidin to detect binding of primary Abs to their antigens. Neuronal nuclei were counterstained with DAPI. Immunostained neurons were analyzed with an 80i Nikon fluorescent microscope (Nikon Corp., Tokyo, Japan). Negative controls for immunocytochemistry included primary hippocampal neurons immunostained with anti-rabbit and anti-mouse biotinylated secondary antibodies with omission of the primary antibodies (Supplemental Figure S3A) and astrocytic cultures immunostained with mAbs for NR1, PSD-95, and synaptophysin (Supplemental Figure S3B).

To determine the expression of the synaptic proteins, at least 20 neurons from each treatment group per experiment from three independent experiments were photographed with the use of ×100 immersion oil objective. The images were captured with a high-sensitivity, cooled, monochrome DS-Qi1Mc camera and NIS Elements Imaging Software version 3.01 (Nikon Corp.) and saved in a TIFF format with a resolution of 600 dpi. With the use of NIH ImageJ software version 1.42, five rectangular test areas that measured 20 × 4 μmol/L each were randomly superimposed along the primary and secondary dendrites of each examined neuron, with the long axis of the test area oriented parallel to the long axis of the dendrites. Synaptic densities within the test area were automatically thresholded and filtered according to the preset algorithm to discriminate nonspecific staining. Data were expressed as the percentage values, using as 100% the average from the total area of synaptic protein puncta per test field in DIV-matched control primary hippocampal neurons.

Subcellular localization of internalized Aβ was examined by two- or three-channel confocal microscopy. Serial Z stacks of 0.5 μm thick tomograms were collected simultaneously in fluorescein and rhodamine channels with the use of a Bio-Rad Radiance 2000 confocal system mounted on the Olympus BX50WI fluorescence microscope (Olympus America Inc., Center Valley, PA). Zeiss LSM 510 confocal microscope (Carl Zeiss Microscopy GmbH, Jena, Germany) was used to collect Z stacks of 0.2 μm thick tomograms in fluorescein, rhodamine, and Texas Red channels. Captured stacks of images were further analyzed with Zeiss LSM Image Browser software version 4.2.0.121 or NIH ImageJ software version 1.42.

Histology, Immunohistochemistry, and Quantitative Analysis of Aβ Plaque Load and Intraneuronal Aβ Accumulation in APP_SW/PS1_dE9/ApoE KO and APP_SW/PS1_dE9/ApoE-TR Tg Mice

Eleven-month-old female Tg mice were sacrificed with an intraperitoneal injection of 150 mg/kg sodium pentobarbital and transcardially perfused with heparinized PBS. Brains were fixed by immersion in 4% paraformaldehyde in 0.1 phosphate buffer, pH 7.4, for 48 hours. Brains were dehydrated in a mixture of 20% glycerol and 20% dimethyl sulfoxide in PBS for 72 hours and cut into 40-μm–thick coronal sections on a freezing microtome. Serial sections were collected in 10 separate series, each containing every tenth section cut along the rostrocaudal brain axis. Randomly selected series were stained with thioflavine-S,²² and X-34 dye²¹ for fibrillar Aβ deposits and immunostained with HJ3.4 monoclonal Ab,²¹ followed by MOM peroxidase kit to visualize the full spectrum of extracellular Aβ deposits containing its fibrillar and nonfibrillar forms. Negative controls for HJ3.4 immunohistochemistry included sections from C57BL/6 mice and sections from APP_SW/PS1_dE9/ApoE4 mice immunostained with omission of HJ3.4 (Supplemental Figure S3C). The Aβ plaque load in the neocortex (percentage of cross-sectional area covered by Aβ) was quantified on X-34-stained sections in a blinded manner according to unbiased stereologic principles (Cavalieri-point counting method) as previously published.²¹

For the analysis and quantification of intraneuronal Aβ accumulation, brains were embedded in paraffin and cut into 10-μm–thick coronal sections, which were mounted on glass slides. After deparaffinization, sections were washed three times in PBS, and nonspecific bindings of the primary Abs and streptavidin were blocked with MOM blocking mixture for 1 hour, followed by the avidin/biotin blocking kit for 30 minutes. 6E10/4G8 mAb mixture was then added for 2 hours followed by 1:1000 biotinylated goat anti-mouse IgG secondary Ab for 1 hour and 1:500 FITC-conjugated streptavidin for 30 minutes. MOM blocking mixture was applied again for 1 hour, followed by anti-MAP2 Ab and cyanine 3-conjugated sheep anti-mouse secondary Ab. Vigorous washing was performed between each step of double immunofluorescence staining. The sections were analyzed under Bio-Rad Radiance 2000 confocal system mounted on the Olympus BX50WI fluorescence microscope. Ten stacks of confocal images spanning the entire thickness of the pyramidal layer of the CA1 and CA2/3 cornu ammonis sectors per section were taken from five to seven serial sections per brain. Neurons positive for intraneuronal Aβ_1-X were identified only as those which were displaying bright punctuate immunostaining pattern inside the perikaryon, but not those showing weak, diffuse membrane staining. The number of neurons positive for intraneuronal Aβ_1-X against the total number of neurons visualized by anti-MAP2 neuronal cytoskeletal staining was quantified and averaged per hippocampal sector in a given mouse brain. Additional hippocampal sections were immunostained with antibodies against C-terminus of Aβ_X-42, and amino and carboxy termini of APP, followed by DAPI counterstaining to control for the specificity of 6E10/4G8 immunostaining and the effect of ApoE KO on APP expression. Negative control for immunohistochemistry included hippocampal sections from C57BL/6 mice immunostained with anti-Aβ and anti-APP Abs (Supplemental Figure S3D).

Statistical Analysis

The statistical analysis was performed with GraphPad Prism version 5.02 (GraphPad Software, Inc.). Differences in the levels of the intraneuronal Aβ and expression of synaptic proteins were analyzed using repeated measures one-way analysis of variance followed by the Tukey’s HSD post hoc test. Two-way analysis of variance followed by Bonferroni’s post hoc test was used to analyze differences in the number of hippocampal neurons with intraneuronal Aβ deposits between CA1 and CA2/3 hippocampal sectors and between APP_SW/PS_dE9/ApoE KO and various APP_SW/PS_dE9/ApoE-TR lines. The U-test was used to analyze the therapeutic effect of Aβ12-28P in co-culture system on the intraneuronal Aβ oligomer level and differences in the load fibrillar Aβ deposits between various APP_SWPS1_dE9/ApoE-TR lines.

Primary Hippocampal Neurons Co-Cultured with Astrocytes Show Increased Intraneuronal Accumulation of Aβ Peptides, Which Can Be Reduced with the ApoE/Aβ Binding Antagonist Aβ12-28P

Uptake of Aβ1-40 and Aβ1-42 from the conditioned media and their intraneuronal accumulation was compared between the primary hippocampal neurons cultured alone and the primary hippocampal neurons co-cultured with astrocytes combined in a noncontact neuronal-astrocytic co-culture system (Figure 1). In this in vitro experimental system, synthetic Aβ1-40 or Aβ1-42 peptides monomerized by HFIP treatment were added to the culture media once at the commencement of the experiment in the final concentration of 10 μmol/L. Both the primary hippocampal neurons and astrocytes (unless specifically stated otherwise) were derived from C57BL/6 wild-type mice. We also confirmed that neurons co-cultured with astrocytes internalize ApoE secreted by astrocytes to the conditioned media. This was done by Western immunoblot detection of ApoE in lysates of neurons co-cultured with astrocytes, whereas in the lysates of neurons cultured alone the ApoE signal was absent (Figure 1D).

Shown in Figure 2, A and B, are comparisons of intraneuronal levels of Aβ1-40 and Aβ1-42 between primary hippocampal neurons cultured alone or co-cultured with astrocytes for 24 hours in the presence of Aβ1-40 and Aβ1-42 peptides, respectively. Although the primary hippocampal neurons cultured alone showed internalization of Aβ1-40 and Aβ1-42 peptides from the media, the intraneuronal accumulation of both Aβ peptides was significantly increased in neurons, co-cultured with astrocytes. The intraneuronal level of Aβ1-40 in neurons from neuronal-astrocytic co-cultures was at 150.8% ± 3.6% (means ± SEM) of the level of neurons cultured alone (P < 0.001) (Figure 2A and Supplemental Figure S1A), whereas the level of Aβ1-42 was at 227.1% ± 26.4% of the level of neurons grown alone (P < 0.01) (Figure 2B and Supplemental Figures S1B and S2B). When Aβ12-28P, the antagonist of the ApoE/Aβ interaction, was added to the culture media together with Aβ1-40 and Aβ1-42 at the commencement of the experiment (10 μmol/L final concentration), intraneuronal levels of Aβ1-40 and Aβ1-42 peptides were reduced to 46.6% ± 4.8% (P < 0.001) and 82.8% ± 9.0%, respectively (P < 0.01), of those in neurons cultured alone. Aβ12-28P_S, a control peptide that represents a scrambled sequence of Aβ12-28P, had no significant effect on intraneuronal levels of Aβ1-40 and Aβ1-42 peptides in primary hippocampal neurons co-cultured with astrocytes. Moreover, Aβ12-28P had no effect on the intraneuronal level of Aβ peptides, when it was added to primary hippocampal neurons cultured alone in the presence of Aβ1-40 or Aβ1-42 peptides. No significant differences were observed in the ApoE level in the conditioned media among various experimental conditions, which could account for the differences in the intraneuronal levels of Aβ peptides (Figure 2C).

To confirm that the increased intraneuronal accumulation of Aβ peptides in neurons co-cultured with astrocytes was directly related to astrocyte-secreted ApoE, we conducted a control experiment in which neuronal-astrocytic co-cultures were set up from primary hippocampal neurons derived from C57BL/6 wild-type mice and astrocytes derived from ApoE KO mice, which were on the C57BL/6 background. Concentrations and length of exposure to Aβ1-42 and Aβ12-28P were exactly matching the conditions in experiments with neuronal-astrocytic co-cultures that used wild-type astrocytes. No statistically significant differences in the intraneuronal Aβ1-42 level were appreciated between primary hippocampal neurons cultured alone and those co-cultured with ApoE KO astrocytes. Aβ1-42 level in neurons co-cultured with ApoE KO astrocytes was at 87.8% ± 6.8% of the Aβ1-42 level of neurons cultured alone, whereas, in neurons co-cultured with astrocytes and additionally treated with Aβ12-28P, the intraneuronal Aβ1-42 level was at 93.0% ± 10.9% of the Aβ1-42 level of neurons cultured alone (differences statistically not significant) (Figure 2D).

Furthermore, we investigated whether co-culturing of neurons with astrocytes affected intraneuronal level of Aβ oligomeric assemblies as well as the rate of intraneuronal degradation of internalized Aβ. Two independent methods were used to characterize intraneuronal content of Aβ oligomers, namely a dot immunoblot analysis that used A11 oligomer-specific Ab and ELISA specific for human-aggregated Aβ. We saw 11.8- and 1.7-fold increase (P < 0.001) in the level of intraneuronal Aβ1-40 and Aβ1-42 oligomers in primary hippocampal neurons co-cultured with astrocytes compared with those cultured alone, respectively (Figure 3A). Note that the primary hippocampal neurons cultured alone showed an approximately sevenfold greater level of intraneuronal Aβ1-42 oligomers than Aβ1-40 oligomers (P < 0.001), whereas no statistically significant difference between Aβ1-42 and Aβ1-40 oligomer levels was noted in neurons co-cultured with astrocytes. This observation indicated that ApoE more effectively promotes oligomerization of Aβ1-40, which has weaker self-aggregation properties than Aβ1-42, which is inherently more prone to form oligomers. The treatment of primary hippocampal neurons co-cultured with astrocytes with Aβ12-28P, which was added to the culture media together with Aβ1-40 and Aβ1-42 at the commencement of the experiment, produced a significant reduction in the intraneuronal Aβ oligomer level. The concentration of intraneuronal Aβ1-40 oligomers in neurons from untreated co-cultures was 57.5 ± 8.3 ng/mg, whereas under Aβ12-28P treatment it was reduced to 30.0 ± 6.3 ng/mg (P < 0.05) as determined by ELISA specific for aggregated Aβ. The concentration of Aβ1-42 oligomers was 57.5 ± 5.9 ng/mg, whereas under Aβ12-28P treatment it was reduced to 40.0 ± 6.3 ng/mg (P < 0.05) (Figure 3B).

The effect of ApoE on the intraneuronal degradation of internalized Aβ was investigated by comparing how effectively primary hippocampal neurons cultured alone and those co-cultured with astrocytes degraded Aβ1-40 internalized during time-limited exposure (Figure 4, A and B). Neuronal monocultures and neuronal-astrocytic co-cultures were exposed to monomerized, synthetic Aβ1-40 peptide for 6 hours (10 μmol/L final concentration in the conditioned media at the commencement of the experiment), then the conditioned media were replaced with Aβ1-40–free media, and neurons were grown further in monoculture or in co-culture with astrocytes (washout period). Primary hippocampal neuron culture alone degraded 50% of the internalized Aβ1-40 within 3.2 hours, and no Aβ1-40 was detectable in their lysates at 24 hours after the commencement of the washout period. In contrast, the primary hippocampal neurons co-cultured with astrocytes still showed the presence of intraneuronal Aβ1-40 at 24 and 48 hours after the commencement of the washout period, which was 43.7% ± 4.9% and 32.9% ± 9.2%, respectively, of the Aβ1-40 signal at t = 0 hours.

We also analyzed the levels of Aβ1-40 and Aβ1-42 peptides in the conditioned media of the primary hippocampal neurons cultured alone and in the conditioned media of neuronal-astrocytic co-cultures (Supplemental Figure S4, A and B). We noted that at the conclusion of the experiment, a significant reduction was seen in the levels of both Aβ1-40 and Aβ1-42 peptides in the conditioned media from neuronal-astrocytic co-cultures compared with the levels of Aβ peptides in the conditioned media from neuronal monocultures. This effect was likely caused by the active clearance of Aβ peptides from the media by astrocytes. To confirm that astrocytes can robustly clear Aβ peptides from the conditioned media, we conducted a complementary experiment in which monomerized, synthetic Aβ1-40 or Aβ1-42 peptides were added to the media of astrocytes cultured alone (10 μmol/L final concentration at t = 0 hours), and then changes in the level of Aβ peptides in the media were analyzed over time. At t = 24 hours levels of Aβ1-40 and Aβ1-42 were reduced by 30.8% ± 6.3% and 24.4% ± 4.9%, respectively, compared with the levels of Aβ peptides at t = 0 hours, whereas at t = 96 hours levels of Aβ1-40 and Aβ1-42 were reduced by 68.2% ± 2.7% and 49.4 ± 0.9%, respectively (two-way analysis of variance P < 0.0001 for the effect of Aβ1-40 versus Aβ1-42) (Supplemental Figure S4C).

Furthermore, we analyzed the rate of intracellular degradation of internalized Aβ by astrocytes akin to the experiment in which we compared the rate of intraneuronal degradation of Aβ between primary hippocampal neurons cultured alone and those co-cultured with astrocytes. After 6 hours of exposure to 10 μmol/L Aβ1-40, the Aβ1-40 signal was seen only in the astrocytic lysate at the conclusion of the Aβ exposure, whereas no Aβ1-40 signal was detectable within t = 2 hours after commencement of the washout period (Supplemental Figure S4D). These experiments confirmed robust Aβ clearance and intracellular degradation capacity of astrocytes.

Confocal Microscopy Analysis of Aβ1-40 Accumulation and Distribution in the Primary Hippocampal Neurons Cultured Alone and Those Co-Cultured with Astrocytes without and with Aβ12-28P Treatment

Along with biochemical analysis of the intracellular Aβ content, we have used confocal microscopy to directly image differences in the intracellular accumulation and distribution of Aβ between primary hippocampal neurons cultured alone and those co-cultured with astrocytes. In the confocal microscopy experiments, Aβ1-40 added to the culture medium was traced with 10% FITC-Aβ1-40, which allowed us to directly visualized Aβ accumulation and distribution. The Aβ1-40/FITC-Aβ1-40 mixture was added to the conditioned media once at the commencement of the experiment at the final concentration of 10 μmol/L, and the neurons were cultured in its presence for 72 hours. In primary hippocampal neurons cultured alone, most of the Aβ1-40 was found outside neurons visible in the form of fine aggregates that covered the external surface of their dendrites (Figure 5A). A small amount of Aβ1-40 was also found inside neuronal perikarya (Figure 5, B and C). In contrast, in neurons co-cultured with astrocytes, a significantly smaller number of Aβ1-40 aggregates were seen on the external surface of dendrites (Figure 5D). Instead, confocal microscopy tomograms taken at the level of perikarya showed large Aβ1-40 aggregates present inside the neurons. They were clearly visualized inside the perikarya and in proximal segments of the primary dendrites (Figure 5, E and F). Intraneuronal Aβ aggregates stained negatively with thioflavin-S, which indicated that they contain no fibrillar Aβ (not shown). In the treatment experiment, 10 μmol/L Aβ12-28P was added together with Aβ1-40/FITC-Aβ1-40 to the neuronal-astrocytic co-cultures at the commencement of the experiment. Under Aβ12-28P treatment, the amount of intraneuronal Aβ1-40 aggregates was greatly reduced (Figure 5, G–I).

Furthermore, we combined confocal microscopy with immunocytochemistry against organelle-specific markers to analyze the subcellular distribution of internalized Aβ1-40 in primary hippocampal neurons when they were cultured alone or co-cultured with astrocytes without and with Aβ12-28P treatment. As in the previous experiments, neuronal monocultures and neuronal-astrocytic co-cultures were maintained in the presence of Aβ1-40 traced with 10% FITC-Aβ1-40 (10 μmol/L final Aβ1-40/FITC-Aβ1-40 concentration) for 72 hours. At the conclusion of the experiment, they were fixed and immunostained against early endosome antigen 1 early endosomal marker, Rab7 late endosomal marker, Cathepsin D late endosomal and lysosomal marker, and heat shock protein 60 as a mitochondrial marker. Primary hippocampal neurons cultured alone showed a modest amount of intraneuronal Aβ1-40, which colocalized weakly with all four organelle markers (Figure 6, A and B, and Supplemental Figure S5, A and B). In contrast neurons co-cultured with astrocytes showed a substantial increase in the intraneuronal content of Aβ1-40 compared with neurons cultured alone. In neurons from the neuronal-astrocytic co-cultures, the Aβ1-40 content was particularly increased in late endosomes and in lysosomes (Figure 6C and Supplemental Figure S5C). To less extent, Aβ1-40 colocalized with markers for early endosomes and mitochondria. Furthermore, some Aβ1-40 was seen outside the nuclear envelope and did not show colocalization with any of the aforementioned organelle markers, which suggested its possible cytoplasmic localization. As in previous experiments, Aβ12-28P treatment of primary hippocampal neurons co-cultured with astrocytes resulted in a marked reduction in the intraneuronal Aβ accumulation in all investigated organelles and cytoplasm (Figure 6D and Supplemental Figure S5D).

Primary Hippocampal Neurons Co-Cultured with Astrocytes Show Enhanced Loss of Synaptic Proteins During Exposure to Aβ1-40, Which Can Be Inhibited by Aβ12-28P Treatment

To investigate whether increased intraneuronal accumulation and oligomerization of Aβ by primary hippocampal neurons co-cultured with astrocytes is associated with enhanced synaptic degeneration, we determined surface expression of the following synaptic proteins: NR1 subunit of the NMDAR, PSD-95, and synaptophysin. Quantification of synaptic protein expression was performed in 18-DIV primary hippocampal neurons cultured alone or co-cultured with astrocytes after 72 hours of exposure to Aβ1-40 traced with 10% FITC-Aβ1-40 (10μmol/L final concentration), without and with Aβ12-28P cotreatment. In primary hippocampal neurons cultured alone, exposure to Aβ1-40 resulted in significant down-regulation of the NR1 subunit of NMDAR, PSD-95, and synaptophysin expression by 46.3%, 51.9%, 54.4% of the control neuron expression, respectively (P < 0.001) (Figure 7, A, B, and F). In primary hippocampal neurons cultured alone, loss of synaptic proteins expression was mainly associated with the presence of Aβ1-40 aggregates that covered the external surfaces of the dendrites (Figure 7B), whereas the amount of intraneuronally accumulated Aβ was modest.

Co-culturing of primary hippocampal neurons with astrocytes in the absence of Aβ1-40 was associated with marked up-regulation of PSD-95 and synaptophysin expression compared with primary hippocampal neurons cultured alone by 21.7% and 12.2%, respectively (P < 0.001) (Figure 7, C and F), whereas no statistically significant difference in the expression of NMDAR NR1 subunit was observed (Figure 7, C and F). When primary hippocampal neurons grown in co-cultures with astrocytes were exposed to Aβ1-40, the reduction in the synaptic protein expression was markedly greater than that observed in primary hippocampal neurons cultured alone. Expression of NR1, PSD-95, and synaptophysin was reduced by 76.8%, 80.4%, and 75.3%, respectively, compared with neurons grown in co-cultures with astrocytes in the absence of Aβ1-40 (P < 0.001) (Figure 7, D and F). This profound loss of synaptic protein expression was associated with massive intraneuronal Aβ1-40 accumulation (Figure 7D). The loss of synaptic protein markers was not associated with degeneration of the cytoskeleton as shown with the use of α-tubulin, as a control cytoskeletal protein. The Western immunoblot signals for α-tubulin in lysates of primary hippocampal neurons treated with Aβ1-40 in the absence or presence of astrocytes were comparable with those of control neurons (Supplemental Figure S6). Aβ12-28P treatment of neurons co-cultured with astrocytes inhibited the loss of synaptic proteins. Compared with the primary hippocampal neurons co-cultured with astrocytes in the absence of Aβ1-40, expression of the NR1 subunit of NMDAR, PSD-95, and synaptophysin was reduced only by 29%, 26.1%, and 36.5%, respectively (P < 0.001) (Figure 7, E and F). The therapeutic effect of Aβ12-28P on synaptic protein expression was associated with substantial reduction in the intraneuronal Aβ1-40 accumulation (Figure 7E).

Knockout of the Apoe Gene in APP_SW/PS1_dE9AD Tg Mice Results in the Absence of Extracellular Fibrillar Aβ Deposits, Whereas TR with ApoE4 Isoform Doubles the Load of Fibrillar Aβ Deposits Compared with TR with ApoE2 and ApoE3 Isoforms

To investigate the role of ApoE in the in vivo intraneuronal Aβ accumulation, we have generated four novel AD Tg mice lines, including APP_SW/PS1_dE9/ApoE2, APP_SW/PS1_dE9/ApoE3, APP_SW/PS1_dE9/ApoE4, and APP_SW/PS1_dE9/ApoE KO. These Tg mice were derived from ApoE-TR mice, where human ApoE isoforms are expressed at the physiological levels of murine ApoE expression.^18,19 For the first time, we have quantified deposition of extracellular Aβ in these previously uncharacterized AD Tg mice lines, calculating the average load of fibrillar Aβ deposits in the neocortex in 11-month-old female mice with the use of an unbiased stereologic principle. The mean fibrillar Aβ load in APP_SW/PS1_dE9/ApoE2 mice and APP_SWPS1_dE9/ApoE3 was 0.86% ± 0.19% and 1.05% ± 0.18%, respectively (Figure 8, A and B) (difference not statistically significant). In APP_SW/PS1_dE9/ApoE4, the fibrillar Aβ load was 1.92% ± 0.34%, almost twice as high as in mice expressing E2 and E3 isoforms (P < 0.05). APP_SWPS1_dE9/ApoE KO mice had little true amyloid deposits shown by amyloid-binding dyes X-34 or thioflavin-S. Scant deposits were too rare and too small to be reliably thresholded and quantified (Figure 8A). A fair number of nonfibrillar Aβ deposits could be shown in APP_SWPS1_dE9/ApoE KO mice with the use of anti-Aβ immunohistochemistry (Figure 8C). In the various lines of APP_SWPS1_dE9/ApoE-TR mice differences in the load of Aβ deposits shown by immunohistochemistry corresponded to that noted by amyloid-binding dyes.

Knockout of the Apoe Gene in APP_SW/PS1_dE9AD Tg Mice Markedly Reduces Intraneuronal Aβ Presence Compared with TR with Human ApoE Isoforms

The effect of ApoE KO and expression of various human ApoE isoforms on intraneuronal Aβ presence was investigated by confocal microscopy in the pyramidal neurons in the CA1 and CA2/CA3 sectors of the cornu ammonis with the use of double immunostaining against Aβ_1-X and MAP2 for neuronal identification. Aβ_1-X was immunostained with 6E10/4G8 mAbs that gave a bright punctate immunostaining pattern inside perikarya (Figure 9A). In APP_SW/PS1_dE9/ApoE KO mice few neurons showed intraneuronal Aβ_1-X presence. They constituted 1.2% ± 0.3% and 4.3% ± 0.6%, respectively, of the total population of pyramidal neurons in the CA1 and CA2/3 sectors (P < 0.001 against all APP_SW/PS1_dE9/ApoE-TR lines) (Figure 9B). In APP_SW/PS1_dE9/ApoE-TR mice, the percentage of CA1 neurons with intraneuronal Aβ_1-X presence ranged from 39.7% ± 2.3% in APP_SW/PS1_dE9/ApoE2 mice to 45.4% ± 1.6% in the APP_SW/PS1_dE9/ApoE4 line. In the CA2/3 sector, the percentage of Aβ_1-X-positive neurons ranged from 55.9% ± 2.0% in the APP_SW/PS1_dE9/ApoE3 line to 63.1% ± 2.3% in the APP_SW/PS1_dE9/ApoE4 line. The fraction of pyramidal neurons that showed intraneuronal Aβ_1-X presence was significantly higher in the CA2/3 sector than in the CA1 sector for all ApoE isoforms (P < 0.001) but not in APP_SW/PS1_dE9/ApoE KO mice. Statistically significant differences between particular ApoE isoforms were seen only in the CA2/3 sector between APP_SWPS1_dE9/ApoE2 and APP_SWPS1_dE9/ApoE4 mice (P < 0.01). To show specific immunodetection of intraneuronal Aβ presence with a mixture of 6E10 and 4G8 mAbs, with no conflicting signals from APP, additional sections were immunostained with mAb HJ7.4 against the carboxy terminus of Aβ_X-42 and Abs against carboxy and amino termini of APP (Figure 9C). Aβ_X-42 carboxy-terminal immunostaining produced a fine, granular pattern that was associated with a large fraction of pyramidal neurons in APP_SW/PS1_dE9/ApoE-TR mice but with few neurons in APP_SW/PS1_dE9/ApoE KO mice, akin to that observed with 6E10/4G8 mAb staining. Both anti-APP Abs produced robust staining of virtually all CA1 and CA2/3 pyramidal neurons in APP_SW/PS1_dE9/ApoE KO mice, which was comparable with that seen in APP_SW/PS1_dE9/ApoE-TR mice.