Cerebrospinal Fluid Particles in Alzheimer Disease and Parkinson Disease.

J Neuropathol Exp Neurol Copyright Ó 2015 by the American Association of Neuropathologists, Inc.

Vol. 74, No. 7 July 2015 pp. 672Y687

ORIGINAL ARTICLE

Cerebrospinal Fluid Particles in Alzheimer Disease and Parkinson Disease Yue Yang, PhD, C. Dirk Keene, MD, PhD, Elaine R. Peskind, MD, Douglas R. Galasko, MD, Shu-Ching Hu, MD, PhD, Eiron Cudaback, PhD, Angela M. Wilson, MS, Ge Li, MD, PhD, Chang-En Yu, PhD, Kathleen S. Montine, PhD, Jing Zhang, MD, PhD, Geoffrey S. Baird, MD, PhD, Bradley T. Hyman, MD, PhD, and Thomas J. Montine, MD, PhD

Abstract

From the Department of Pathology, University of Washington, Seattle, WA (YY, CDK, EC, AMW, KSM, JZ, GSB, TJM); Department of Psychiatry & Behavioral Sciences, University of Washington, Seattle, WA; (ERP, GL); VA Northwest Network Mental Illness Research, Education, and Clinical Center, VA Puget Sound Health Care System, Seattle, WA (ERP); Department of Neurosciences, University of California San Diego, La Jolla, CA (DRG); Department of Neurology, University of Washington, Seattle, WA (S-CH); Department of Medicine, University of Washington, Seattle, WA (C-EY); Geriatric Research, Education, and Clinical Center, VA Puget Sound Health Care System, Seattle, WA (C-EY); Department of Laboratory Medicine, University of Washington, Seattle, WA (GSB); and Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA (BTH). Send correspondence and reprint requests to: Thomas J. Montine, MD, PhD, Department of Pathology, University of Washington, PO Box 357470, Seattle, WA 98195; E-mail: [email protected] This work was supported by P50 AG05136, P50 NS62684, P50 AG05131, the Department of Veterans Affairs, and the Nancy and Buster Alvord Endowment. Potential Conflicts of Interest. AMW, CDK, CEY, EC, GL, JG, JSB, KSM, SCH, TJM, and YY have nothing to report. DRG and ERP report grants from National Institutes of Health during the conduct of the study. BTH reports grants and personal fees from Biogen; grants from BMS, Ipierian, and Protena; and personal fees from Calico and Genentech, outside the submitted work. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the Journal’s Web site (www.jneuropath.com).

672

Key Words: Alzheimer disease, Biomarkers, Cerebrospinal fluid, Cognitive impairment, Neurodegenerative disease, Parkinson disease.

INTRODUCTION Human cerebrospinal fluid (CSF) lipoproteins are distinct from their plasma counterparts and harbor apoE isoforms and apoAI on largely distinct particle subclasses, as well as apoJ and other proteins, on individual particles estimated as ~0.02 Km in diameter (1Y5). APOE has not been associated consistently with CSF apoE concentration (6Y10) in contrast to plasma apoE (11). Moreover, there is not a reproducible association between CSF apoE levels and Alzheimer disease (AD) dementia or Parkinson disease (PD) (7, 10, 12Y16). AA peptides do bind apoE-containing CSF lipoproteins but not apoE itself (17, 18). Polymorphisms in the apoAI gene are associated with dementia (19, 20), and apoAI plasma or serum levels are inversely associated with AD dementia and PD (21Y27). ApoAI also is reported to bind AA in CSF and plasma. ApoAI interacts with the extracellular domain of amyloid precursor protein (APP), as well as with AA, where it suppresses AA aggregation and toxicity (28, 29). ApoJ, the product of the clusterin gene (CLU), confers risk for AD dementia (30, 31), modulates AA structure and toxicity in vitro (32Y34), and interacts with AA clearance in vivo (35). Cerebrospinal fluid apoJ may be a biomarker of predementia stages of AD (36). Cerebrospinal fluid clusterin concentration has not been associated reproducibly with PD (27, 37, 38). In addition to lipoprotein particles, virtually all cells elaborate extracellular vesicles (EVs), either constitutively or in response to a variety of signals and stressors (39Y41). These EVs are generally referred to as microvesicles, exosomes, ectosomes, or membrane particles and apoptotic vesicles. There is some ambiguity in the nomenclature of EVs because of incomplete knowledge of their origins and composition. Exosomes and microvesicles both present phosphatidylserine (PS) (42, 43), and overlapping characteristics such as size, morphology, density, J Neuropathol Exp Neurol Volume 74, Number 7, July 2015

Copyright © 2015 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.

Downloaded from http://jnen.oxfordjournals.org/ by guest on December 16, 2015

Human cerebrospinal fluid (CSF) contains diverse lipid particles, including lipoproteins that are distinct from their plasma counterparts and contain apolipoprotein (apo) E isoforms, apoJ, and apoAI, and extracellular vesicles, which can be detected by annexin V binding. The aim of this study was to develop a method to quantify CSF particles and evaluate their relationship to aging and neurodegenerative diseases. We used a flow cytometric assay to detect annexin V-, apoE-, apoAI-, apoJ-, and amyloid (A) A42-positive particles in CSF from 131 research volunteers who were neurologically normal or had mild cognitive impairment (MCI), Alzheimer disease (AD) dementia, or Parkinson disease. APOE D4/D4 participants had CSF apoE-positive particles that were more frequently larger but at an 88% lower level versus those in APOE D3/D3 or APOE D3/D4 patients; this finding was reproduced in conditioned medium from mouse primary glial cell cultures with targeted replacement of apoE. Cerebrospinal fluid apoE-positive and A-amyloid (AA42)-positive particle concentrations were persistently reduced one-third to one-half in middle and older age subjects; apoAI-

positive particle concentration progressively increased approximately 2-fold with age. Both apoAI-positive and annexin V-positive CSF particle levels were reduced one-third to one-half in CSF of MCI and/ or AD dementia patients versus age-matched controls. Our approach provides new methods to investigate CNS lipid biology in relation to neurodegeneration and perhaps develop new biomarkers for diagnosis or treatment monitoring.

J Neuropathol Exp Neurol Volume 74, Number 7, July 2015

changes in concentration and composition in aging and in relation to APOE genotype and neurodegeneration.

MATERIALS AND METHODS Participants and CSF Collection The Human Subjects Review Committee of the University of Washington and the Veterans Affairs Puget Sound Health Care System approved this study. Samples were from individuals enrolled in the University of Washington Alzheimer’s Disease Research Center or the Pacific Northwest Udall Center. All individuals provided informed consent and underwent extensive evaluation that consisted of medical history, family history, physical and neurologic examinations by clinicians who specialize in movement disorders or dementia, laboratory tests, and neuropsychological assessment; information was obtained from controls or from informants for patients (76, 77). Controls were compensated community volunteers in good health with no signs or symptoms suggesting cognitive decline or neurologic disease upon neurological and neuropsychological examination. Inclusion criteria were complete blood count, serum electrolytes, blood urea nitrogen, creatinine, glucose, vitamin B12, and thyroid stimulating hormone results within normal limits. Exclusion criteria for cases and controls included heavy cigarette smoking (910 packs/year) and alcohol use other than social. Any psychotherapeutic use was an exclusion criterion for controls. Any psychotherapeutic use other than for treatment of neurodegenerative disease was an exclusion criterion for cases. A total of 131 samples were selected randomly from subjects who met clinical diagnostic criteria (73Y75). This resulted in the following cohort: 59 healthy controls in young (G40 years old), middle-aged (40 to 65 years old), and older (965 years old) age ranges; 21 individuals with MCI (74); 27 patients with AD dementia (73); and 24 patients with PD (75). Information on the participants whose samples were used is presented in the Table. Cerebrospinal fluid was obtained by lumbar puncture and was collected between 8:00 and 11:00 am after a 12-hour fast, as described previously (78). Cerebrospinal fluid was separated into sequential 0.5-mL aliquots at the bedside, flash frozen on dry ice, and stored at j80-C before assay, according to National Institutes on Aging Best Practices Guidelines (http://www.nia.nih.gov/about/policies). Brain autopsy was not performed on any subject in this study; thus, neuropathological correlation with clinical diagnosis was not possible.

TABLE. Characteristics of 131 Participants Whose Samples Were Used Controls N Age (year, mean T SD) M:F (N) Education (mean years) APOE D3/D3 (N) APOE D3/D4 (N) APOE D4/D4 (N)

Young

Middle Aged

Older

MCI

AD Dementia

PD

15 28 T 6 4:11 17 10 4 1

21 55 T 7 11:10 16 11 10 V

23 73 T 5 11:12 15 9 10 4

21 75 T 10 14:7 16 10 10 1

27 69 T 10 16:11 16 10 10 7

28 64 T 8 12:12 16 14 10 V

AD, Alzheimer disease; M, male; F, female; MCI, mild cognitive impairment; PD, Parkinson disease.

Ó 2015 American Association of Neuropathologists, Inc.


673


and protein composition make it difficult to discriminate between exosomes and microvesicles (44). Therefore, in this study, we conservatively use the term extracellular vesicles rather than microvesicles or exosomes. Several subclasses of EV have overlapping size ranges from approximately 40 nm up to 5 Km with particles larger than 1 Km deriving mostly from apoptotic bodies (45). Most subclasses of EVs express surface PS, and annexin V, a protein with strong affinity for PS, is often used as a probe for PS exposure (46). Extracellular vesicles have been identified in ovine CSF (47), and a variety of cell culture experiments have explored neurotrophic and neuroinflammatory effects of neuronal and glial EVs (48Y51). In the context of neurodegenerative diseases, cell culture experiments have shown that EVs contain proteolytic fragments of APP, including AA peptide, tau, and >-synuclein (52Y56). Extracellular vesicles enriched with proteolytic APP fragments have been isolated from the brains of transgenic mice overexpressing mutant human APP and from the brains of patients who died with AD dementia (57). Extracellular vesicle proteins were enriched in amyloid plaques from the brain sections of 3 AD patients (52), and tau protein was identified in EVs isolated by ultracentrifugation from ventricular CSF (55). One group reported EVs in CSF obtained by lumbar puncture from a patient with ALS (51). Given these intriguing results, we hypothesized that EVs in CSF might be useful as biomarkers for AD or PD. Several studies have characterized the molecular composition of CSF particles using a variety of techniques, but each is limited. Some techniques used to enrich particles may modify protein-lipid and protein-protein interactions; many studies did not distinguish particle-free from particle-bound proteins, and most investigations required large CSF volumes, which is impractical for application as a biomarker in the clinical setting (4, 5, 15Y17, 58Y63). Recently, flow cytometry has evolved to assay particles much smaller than a cell (64, 65); indeed, flow cytometry has been used extensively to study synaptosomes prepared from AD and control brains (66Y70). Engineering advances have produced flow cytometers that detect polystyrene microspheres as small as 0.2 Km (71) and accurately separate and count EVs (72). In this study, we developed a novel CSF flow cytometry method and assayed 0.5 mL aliquots from 131 individuals who had undergone clinical and neuropsychological evaluations and were without neurologic disease or met diagnostic criteria for mild cognitive impairment (MCI), AD dementia, or PD (73Y75). We hypothesized that lipid particles and EVs would show

CSF Particles in AD and Parkinson Disease

Yang et al

Antibodies and Fluorescent Reagents

CSF Particle Staining for Flow Cytometric Assay Zenon IgG labeling kit (Invitrogen) was used to generate fluorophore conjugated anti-AA42, tau, apoE, and apoJ antibodies before use. Specifically, mouse anti-apoE monoclonal antibody and mouse anti-apoJ monoclonal antibody were fluorescently labeled with the Zenon Alexa Fluor 488 mouse IgG1 labeling kit, respectively. Rabbit anti-AA42 monoclonal antibody was fluorescently labeled with the Zenon PE rabbit lgG labeling kit according to the manufacturer’s protocol. Antibodies were used in the following formats, and final concentrations were determined by titration experiments: rabbit-anti-AA42-PE (IgG, 10 Kg/mL), mouse-anti-tau-Alexa Fluor 488 (IgG1, 10 Kg/mL), mouse anti-apoE-Alexa Fluor 488 (IgG1, 10 Kg/mL), mouseanti-apoJ-Alexa Fluor 488 (IgG1, 10 Kg/mL), mouse-antiapoAI-FITC (IgG1, 10 Kg/mL), and annexin V-FITC (10 ng/mL). To avoid cross-reactivity, primary antibodies used for double immunostaining were raised in different species, with the secondary antibodies recognizing one of the species exclusively. All samples were colabeled with anti-AA42 antibody and an apolipoprotein antibody to detect vesicle membrane (surface) protein or annexin V to detect PS; thus, quadruplicates were obtained for AA42 while each apolipoprotein and annexin V binding was determined once for each CSF sample. Never thawed 500 KL CSF aliquots were maintained at j80-C. Annexin V binding to phosphatidylserine was assayed by incubation with annexin V-FITC in the presence of 1 mM Ca2+ for 10 minutes on ice after incubation with anti-AA42-PE for 30 minutes in the dark. We used 2 negative controls for antibody assays: isotype controls of corresponding species of immunoglobulin were used at the same final concentration as the specific antibodies, and the same volume of phosphate-buffered saline (PBS) was labeled by specific antibodies. Buffer without Ca2+ was used for dilution instead of binding buffer as negative control for annexin V binding. Two sets of experiments used

674

small subsets of CSF for which we had sufficient residual volume. One set of 5 CSF samples was used to determine colabeling of particles by apoE and apoJ. The second set was used to estimate the amount of particle associated tau protein. As recommended by others to reduce background, debris, and precipitates in particle analysis by flow cytometry (64), we used 0.1-Km pore size filters for all solutions (Millipore, Billerica, MA).

CSF Particle Analysis Using Flow Cytometry Samples were analyzed using an Apogee A50 flow cytometer (Apogee Flow Systems). The Apogee A50 is significantly more sensitive and accurate in detecting EVs compared with other flow cytometric systems (71). Flow cytometer sheath solutions were 0.1 Km filtered before use. Size gates were defined based on forward angle light scattering from silica microbeads. The particle gate was determined using ‘‘Calibration BeadMix’’ (Apogee Flow Systems), which is a mix of 6 silica beads with diameters of 0.18, 0.24, 0.30, 0.59, 0.88, and 1.30 Km and 2 fluorescent polystyrene microbeads with diameters of 0.11 and 0.5 Km. Equivalent diameter of particles with the same forward scatter relative intensity as the silica beads was measured according to established methods (71). To characterize CSF particle size, the smaller particle gate was defined based on forward scatter intensity from immediately above instrument noise up to the 0.3-Km silica microbeads, which is equivalent to 0.4-Kmdiameter cell-derived particles (71). The gate for larger particles was set from the middle of a 0.3-Km silica microbeads peak to just above the 0.59-Km silica microbeads; the latter is equivalent to cellular particles approximately 1 Km in diameter (71). For total particle counts, 100 KL CSF stained with specific antibodies was suspended in 300 KL antibody buffer or annexin V binding buffer containing Ca++. Fifty-microliter counting beads were added before the flow assay. Concentration of particles was estimated by measuring particle counts using CountBright absolute counting beads (Invitrogen), according to established methods (79).

CSF AA42 Pull Down Assay One-milliliter human CSF, or 1 mL PBS as negative control, was incubated with 2 Kg mouse antihuman AA(6E10) antibody for 1 hour at 4-C to cross-link the particles binding with AA. Twenty microliters of resuspended volume of protein A/G plus-agarose beads (Santa Cruz Biotechnology, Dallas, TX) were then added and incubated with the CSF for 1 hour at 4-C under continuous gentle agitation, and the sample was then pelleted by centrifugation at 2500 rpm for 5 minutes at 4-C. Supernatants were aspirated and then analyzed using flow cytometry with rabbit anti-AA42-PE antibody as described previously. Pellets were washed 4 times with 1 mL RIPA buffer by centrifugation at 2500 rpm for 5 minutes at 4-C and then boiled in 40 KL of electrophoresis sample buffer for 2 to 3 minutes. The samples were separated by SDS-PAGE using 10% to 20% tris-tricine gel followed by transfer to 0.2-Km nitrocellulose membranes. For immunoblotting, membranes were blocked for 1 hour at room temperature in 2.5% milk in Tris-buffered saline with 0.05% Tween-20 (TBST). Membranes were incubated overnight at 4-C with anti-AA42 antibody (1:1000) and anti-AA (6E10) antibody (1:1000), respectively. After washing 3 times with TBST, membranes were incubated for 90 minutes at room Ó 2015 American Association of Neuropathologists, Inc.



The following antibodies were used: rabbit anti-AA1-42 monoclonal antibody (clone H31L21; Invitrogen, Carlsbad, CA), which recognizes human and mouse AA1-42; cross reactivity to AA1-40 is not observed in sandwich ELISA according to the manufacturer’s instructions; mouse anti-human AA monoclonal antibody (clone 6E10; Covance, Princeton, NJ), mouse antihuman apoE monoclonal antibody (clone 1H4; Abcam, Cambridge, MA); mouse antihuman apoJ monoclonal antibody (clone 3R3/2; Lifespan Biosciences, Seattle, WA); rabbit antimouse apoJ polyclonal antibody (Lifespan Biosciences); fluorescein isothiocyanate (FITC)Yconjugated mouse antihuman apoAI monoclonal antibody (clone APO-1-1; Fitzgerald Industries, Acton, MA); rabbit antimouse apoAI polyclonal antibody (Lifespan Biosciences); FITC-conjugated annexin V (Beckman Coulter, Pasadena, CA); and mouse antihuman tau monoclonal antibody (clone TAU-5; Abcam). Isotype controls were as follows: FITC-conjugated mouse IgG1isotype control (eBioscience, San Diego, CA), mouse IgG1isotype control (eBioscience), and rabbit IgG isotype control (Abcam). Other reagents included Calibration BeadMix (Apogee Flow Systems, Hemel Hempstead, UK), annexin V binding buffer 10 (Beckman Coulter), CountBright absolute counting beads (Invitrogen), Zenon Alexa Fluor 488 mouse IgG1, and phycoerythrin (PE) rabbit IgG labeling kits (Invitrogen).



temperature with antimouse IgG-HRP or antirabbit IgG-HRP in TBST with 2.5% milk. After 3 washes, membranes were incubated in Pierce ECL Western blotting substrate (Thermo Scientific, Waltham, WA) for 5 minutes, exposed to Kodak BioMax MS film (Rochester, NY) for 30 seconds to 5 minutes, and then developed.

Exosome Assay Purified standard exosomes from human pooled serum (System Biosciences, Mountain View, CA; purified exosomes, #EXOP-300A-1) were stained using annexin V-FITC and analyzed using Apogee A50 flow cytometry. Plasma exosomes were prepared by an established technique that uses ultracentrifugation to generate fractions (called layers) enriched in exosomes; they were then analyzed using annexin V staining method used for CSF.

Animals and Primary Cultures

Statistical Analysis Data are presented as mean T SE unless otherwise specified. Group comparisons and correlations were performed with GraphPad Prism software (San Diego, CA) using nonparametric 1-way ANOVA (Kruskal-Wallis test) followed by Dunn’s corrected repeated pair comparisons or 2-way ANOVA followed by Bonferroni-corrected posttests; correlations were estimated by the method of Pearson. Outcomes are displayed graphically as *P G 0.05; **P G 0.01; *** P G 0.001; or **** P G 0.0001.

RESULTS Subjects and Standard CSF Biomarker Results We performed Luminex-based assays for total CSF AA42, total tau, and tau-P181 using a well-established method in our laboratory that has been part of a global quality assurance effort for these measurements (76, 86, 87). As expected, average CSF AA42 concentration was significantly different among the older controls, MCI, AD, and PD (P G 0.0001) with corrected repeat pair comparisons showing that CSF AA42 was significantly lower in AD dementia (201 T 18 pg/mL)

compared with older controls (311 T 29 pg/mL) (Table, Supplemental Digital Content 1, http://links.lww.com/NEN/A741). Cerebrospinal fluid total tau and tau-P181 also were significantly different among these 4 groups (P G 0.0001 for both analytes) with corrected repeat pair comparisons for MCI (115 T 18 pg/mL and 107 T 17 pg/mL) and AD dementia (101 T 7 pg/mL and 69 T 7 pg/mL) significantly greater than older controls (57 T 4 pg/mL and 39 T 4 pg/mL) for both total tau and tau-P181 (Table, Supplemental Digital Content 1, http:// links.lww.com/NEN/A741). Parkinson disease CSF total tau and tau-P181 concentrations (39 T 4 pg/mL and 27 T 2 pg/mL) were lower than older controls, consistent with several previous reports (88Y90); however, neither was significantly different by corrected repeat pair comparisons.

CSF Particle Assay Using Flow Cytometry

Polystyrene fluorescence beads (0.11 and 0.50 Km) and size calibrated nonfluorescent silica beads (0.18, 0.24, 0.30, 0.59, 0.88, and 1.30 Km) were resolved with forward (SALS) and side (LALS) light scatter (Fig. 1A); 0.18 Km was the lower limit for size resolution of nonfluorescent beads. The upper size limit of 0.59 Km for silica beads served to gate for particles of 1.0 Km or less (71) and to minimize contamination by debris (Fig. 1B, left and middle panels). Silica beads of 0.30 Km were used to distinguish between smaller and larger particles (Fig. 1B, middle panel). The gating of CSF particles was derived from the SALS and LALS characteristics of these nonfluorescence silica beads (Fig. 1B, right panel). Anti-apoE-Alexa Fluor 488, anti-apoJ-Alexa Fluor 488 and anti-apoAI-FITC antibodies were used to detect apoEpositive, apoJ-positive and apoAI-positive particles in CSF (Fig. 1C, right panels). Very few background events were detected after filtration (Fig. 1C, left panels) or with fluorescencelabeled isotype antibodies (Fig. 1C, middle panels). Annexin V-positive particles were detected in CSF using annexin VFITC in binding buffer containing Ca2+ (Fig. 1D, right panel); buffer without Ca2+ was used as negative control (Fig. 1D, middle panel). Very few positive events were detected in PBS plus annexin V-FITC (Fig. 1D, left panel). We used 2 approaches to validate our method. First, we analyzed exosomes purified from plasma by ultracentrifugation. Plasma fractions with density expected for exosomes were incubated with annexin V with or without Ca2+ binding buffer and then analyzed using the same method as for CSF particles. Ca2+ binding buffer-dependent annexin V binding was observed in 6.2% and 5.4% particles in plasma fractions with expected density for exosomes (Figure, Supplemental Digital Content 2, http://links.lww.com/NEN/A742, part A, layers 7Y8 and layers 9Y10). Ca2+ binding buffer-independent annexin V binding in these fractions was undetectable and similar to Ca2+ binding buffer-dependent and -independent annexin V binding to plasma particles in fractions not expected to contain exosomes, including supernatant, layer 1, and layer 12. Second, we used purified standard exosomes from human pooled serum and stained them with annexin V-FITC before analysis using our flow cytometry method. We again observed positive annexin V-positive particles (Figure, Supplemental Digital Content 2, http://links.lww.com/NEN/A742, part B). These data show that our method used with CSF can identify EVs from plasma



675


Wild-type C57BL/6 mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Mice with homologoustargeted replacement (TR) of mouse apoE with human APOE D3 (TR APOE3+/+), or APOED4 (TR APOE4+/+) were used for primary cultures (80Y84). Original strains were backcrossed greater than 10 generations to a C57BL/6 genetic background. TR APOE3+/+ or TR APOE4+/+ mice were crossed with apoE knockout (-/-) mice to generate TR APOE3+/- or TR APOE4+/mice, respectively. All mice were housed in a temperaturecontrolled specific pathogen-free facility with a strict 12-hour light/dark cycle and with free access to food and water and used with approval of the University of Washington Animal Care and Use Committee. Primary cultures of mouse cerebral cortical neurons, astrocytes, or microglia were generated as previously described (84, 85). After 1 week in culture, medium was removed, and cells were washed 3 times with PBS and then replaced with serum-free medium for additional 18 hours. This conditioned medium then was collected, and particles were analyzed using flow cytometry as described previously for CSF.


Yang et al


676




FIGURE 1. Measurement of cerebrospinal fluid (CSF) particles using Apogee A50 flow cytometry. (A) Calibration Beadmix was used to assess light scatter and fluorescence. Shown are typical data from the BeadMix analyzed on the A50-Micro flow cytometer. The left cytogram shows that all populations included, 2 green fluorescent (488 nm laser) polystyrene beads with diameters of 0.11 and 0.5 Km, and 6 silica beads with diameters of 0.18, 0.24, 0.30, 0.59, 0.88, and 1.30 Km, were resolved from each other and from instrument noise. The right cytogram was gated to exclude the fluorescent latex beads so that the nonfluorescent beads can be evaluated more clearly. (B) Histograms show side scattering (left) and forward scattering (middle) from 0.18, 0.24, 0.30, 0.59, 0.88, and 1.3 Km silica microspheres (blue peaks) and the noise threshold limit (dotted line). Representative dot plot (right) of a CSF sample showing ‘‘smaller’’ particle gate based on forward angle light scatter from 0.30 Km silica microspheres and ‘‘larger’’ particle gate based on forward angle light scatter from 0.30 to 0.59 Km silica microspheres. (C) Flow cytometric analysis of CSF apolipoprotein-bearing particles in human CSF. Fluorescence analysis of CSF samples labeled with apoEAlexa Fluor 488, apoJ-Alexa Fluor 488, or apoAI-FITC (right panels). Density plots show background from phosphate-buffered saline (PBS) labeled with the same antibodies (left panels) used as background control. Isotype control of corresponding species of IgG was used to label the same CSF samples (middle panels) as negative controls and results used to set gating for positive events. (D) Fluorescence analysis of a CSF sample labeled with annexin V-FITC with binding buffer containing Ca2+ (right). The same labeling procedure but using buffer without Ca2+ was used as negative control (middle). PBS labeled with annexin V-FITC was used as background control (left). (E) Flow cytometric analysis of A-amyloid (AA42)-positive particles in human CSF. PBS (upper, left) and isotype control (upper, right) was used as controls. A representative fluorescence plot shows AA42-positive events in CSF (lower, left) that were nearly completely removed following pull down of AA42 with 6E10 antibody coupled to agarose beads (lower, right). (F) The protein retrieved from the 6E10 pull down pellet was analyzed by Western blot using 6E10 (right) or AA42 (left) antibody to reveal expected smear of AA42-immunoreactive molecular masses (and immunoglobulin bands). PBS was substituted for CSF as the negative control.



FIGURE 3. Flow cytometric analysis of the association of Aamyloid (AA42) with other cerebrospinal fluid (CSF) particles. (A) Dual fluorescence analysis of a representative CSF sample showing AA42 colabeled with apoE-positive (upper left) or apoJpositive (upper right) particles but not apoAI-positive (lower left) or annexin V-positive (lower right) particles. (B) Scatter plot with best-fit line for apoE associated AA42 particles versus apoJ associated AA42 particles for all 131 participants.

FIGURE 2. Cerebrospinal fluid (CSF) particles were classified by the presence of apoE, apoJ, apoAI, or A-amyloid (AA42), as detected by labeled antibodies or by binding of annexin V, and stratified by size into larger (A, 300Y590 nm) or smaller (B, G300 nm) particles for each sample. One-way ANOVA for each molecularlydefined particle concentration (events/KL) in these 2 size ranges each had P G 0.0001. Corrected multiple comparisons for CSF particle concentration had P G 0.01 for all paired comparisons except for apoE-positive versus apoJ-positive, which were not significantly different for either size range.

Each CSF sample was analyzed 4 times. To estimate the precision of measurements by this approach, we included detection of AA42 in each run, meaning that AA42-positive CSF particles were measured 4 times for each individual: once each with apoE-positive (run 1), apoJ-positive (run 2), apoAIpositive (run 3), and annexin V-positive (run 4) particles. Concentration (events/KL) of AA42-positive particles correlated well between runs (Figure, Supplemental Digital Content 3, http://links.lww.com/NEN/A743): run 1 versus run 2 (R2 = 0.85, P G 0.0001) versus run 3 (R2 = 0.91, P G 0.0001) or versus run 4 (R2 = 0.85, P G 0.0001). Overall, the coefficient of variance for flow cytometric detection of CSF AA42-positive particles was 9.4% for these 131 samples assayed in quadruplicate.

Overview of CSF Particles We used our flow cytometry assays to evaluate the molecular composition, concentration, and size distribution of CSF



677


with the density of exosomes and suggests that at least some of the EVs analyzed in CSF likely are exosomes. Anti-AA42-PE antibody was used to detect AA42-positive CSF particles (Fig. 1E, lower left panel). Again, we used PBS (Fig. 1E, upper left panel) and isotype negative controls (Fig. 1E, upper right panel). Furthermore, pull down of CSF AA peptides by 6E10 antibody showed nearly completely removal of AA42positive events from CSF (Fig. 1E, lower right panel). The pull down pellet was then examined by Western blot using 6E10 (Fig. 1F, right panel) and anti-AA42 antibodies (Fig. 1F, left panel) and showed the expected widely distributed molecular masses for AA species (91) (Fig. 1F, right panel). Anti-tau-Alexa Fluor 488 was used to assay particle associated tau. Unlike the other 5 probes that yielded clear signal in human CSF, particle associated tau was minimal, averaging (SD) 24 T 29 events/KL (n = 5 different CSF samples). Taken together, these results demonstrate specificity and minimal nonspecific signal for apoEpositive, apoJ-positive, apoAI-positive, annexin-V-positive, and AA42-positive particle assays in human CSF.

Yang et al

particles. Figure 2 presents a scatterplot with mean T 95% confidence interval of data from the 131 samples. Cerebrospinal fluid particles were classified by the presence of apoE, apoJ, apoAI, or AA42 (as detected by labeled antibodies) or by binding of annexin V and stratified by size into larger (90.30 Km) or smaller (G0.30 Km) particles. One-way ANOVA for the concentration of the 5 molecularly defined particles had P G 0.0001 in each size range (Fig. 2A, B). Of all possible pairs, only apoEpositive and apoJ-positive particle concentrations were not


significantly different in the 2 size ranges, indicating that apoEpositive and apoJ-positive CSF particles largely overlap in concentration and size. Given these results, we determined colabeling of particles with apoE and apoJ in a subset of 5 CSF samples for which we had sufficient residual material (Figure, Supplemental Digital Content 4, http://links.lww.com/NEN/A744). ApoEpositive and apoJ-positive co-labeled particles comprised 38% T 5% of all apoE-positive particles and 59% T 7% of all apoJpositive particles.


678




ApoE-Positive and ApoJ-Positive Particles We compared concentration and size of apoE-positive and apoJ-positive CSF particles in different age and diagnostic groups. Older controls (gray) were used as the age-appropriate comparison group for MCI or AD dementia (black), and PD (striped). The average concentration of CSF apoE-positive particles was significantly different among the 3 age groups (Fig. 4A, P G 0.01) with significant difference for young versus middle-aged controls (P G 0.05) and young versus older controls (P G 0.01) but not for middle-aged versus older controls. Cerebrospinal fluid apoE-positive particle concentration was not significantly different among older controls, MCI, AD dementia, and PD. These data suggest an age-related reduction of CSF

apoE-positive particle concentration in mid-life but not in association with either of these neurodegenerative diseases. Next, we investigated associations of APOE genotype with CSF apoE-positive particle concentration. We stratified our results into those from individuals homozygous for the D3 allele (APOE D3/D3, n = 64); heterozygous for D3 and D4 (APOE D3/D4, n = 54) or homozygous for the D4 allele (APOE D4/D4, n = 13), regardless of age or diagnosis. Cerebrospinal fluid apoE-positive particle concentration was 2711 T 283 events/KL in APOE D3/D3, 2471 T 338 events/KL in APOE D3/D4, and 336 T 54 events/KL in APOE D4/D4 (P G 0.0001 overall with P G 0.001 for APOE D4/D4 vs. either APOE D3/D3 or APOE D3/D4). To determine whether the genotype effect was dependent on age, we performed the same analysis restricted to individuals older than 65 years of age and observed a similar outcome: CSF apoE-positive particle concentration was 2663 T 384 events/KL in APOE D3/D3, 2842 T 723 events/KL in APOE D3/D4, and 314 T 76 events/KL in APOE D4/D4 (P G 0.001 overall with P G 0.01 for APOE D4/D4 vs. either APOE D3/D3 or APOE D3/D4). We further stratified by APOE genotype and by control (CTL), diagnosis of MCI or AD dementia (M/A), or diagnosis of PD (Fig. 4B, P G 0.01 overall with corrected repeated pair comparisons significant for APOE D3/D3 vs. APOE D4/D4 in CTL and M/A groups, P G 0.05). Moreover, comparison across the 3 diagnostic groups (CTL, M/A, and PD) for individuals with APOE D3/D3 or with APOE D3/D4 was not significant for either genotype. Given the apparent influence of age and APOE on CSF apoE-positive particle concentration, we plotted these variables in Figure 4C. This graph suggests a loss of homeostasis for CSF apoE-positive particle concentration around age 60 years that is most prominent in those individuals who are APOE D4/D4. We next analyzed CSF apoE-positive particle size by APOE genotype. APOE D3/D3 had 35% T 2%, APOE D3/D4 had 38% T 2%, and APOE D4/D4 had 60% T 3% larger particles among total apoE-positive particles (P G 0.0001 overall with corrected repeated pair comparisons showing P G 0.001 for APOE D4/D4 vs. APOE D3/D4 or vs. APOE D3/D3). Again restricting analysis to individuals 965 years of age, we observed virtually identical outcomes: APOE D3/D3 had 33% T 2%, APOE D3/D4 had 38% T 2%, and APOE D4/D4 had 61% T 3% larger particles among total apoE-positive particles. Within diagnostic groups (Fig. 4D), 2-way ANOVA had P G 0.0001 for APOE genotype in CTL and M/A groups, with

FIGURE 4. Cerebrospinal fluid (CSF) apoE-positive particle analysis using flow cytometry. (A) Older controls were used as part of the age series (gray) and also as the control group for mild cognitive impairment (MCI), Alzheimer disease (AD) dementia (black), and Parkinson disease (PD) (striped). Horizontal lines indicate groups compared by ANOVA (3 age groups, or older controls, MCI, AD dementia, and PD). One-way ANOVA had **P G 0.01 for 3 age groups. Corrected posttests had * P G 0.05 for young versus middle-aged or ** P G 0.01 for young versus older controls. (B) Comparison of apoE-positive particle concentration stratified by APOE genotype and by diagnostic category as control (CTL, gray), MCI and AD dementia (M/A, black), or PD (striped). Two-way ANOVA had ** P G 0.01 for APOE genotype among CTL and M/A groups, and corrected posttests had * P G 0.05 for APOE D4/D4 versus APOE D3/D3 in both CLT and M/A groups. There was no significant difference between CTL and PD groups. (C) ApoE-positive particle concentrations were plotted against age for CTL and M/A groups stratified by the 3 APOE genotypes, and for PD groups stratified by the 2 APOE genotypes. (D) Percent of larger apoE-positive particles stratified by APOE genotype and by diagnostic category as control (CTL, gray), MCI and AD dementia (M/A, black), or PD (striped). Two-way ANOVA had ** P G 0.0001 for APOE genotype among CTL and M/A groups, and corrected posttests had ** P G 0.01 for APOE D4/D4 versus APOE D3/D3 or * P G 0.05 for APOE D3/D4 versus APOE D3/D3 in both CLT and M/A groups. There was no significant difference between CTL and PD groups. (E) Scatter plot of CSF apoE-positive particle size versus concentration for each sample shows that APOE D4/D4 (black) is associated with fewer, larger CSF apoE-positive particles. Ó 2015 American Association of Neuropathologists, Inc.


679


We also determined the concentration of CSF particles colabeled for AA42 and each of the other analytes. Only apoE and apoJ were detectably colabeled with AA42-positive; no detected AA42-positive particles colabeled with apoAI or annexin V (Fig. 3A). For each individual, we added the concentration of apoE-positive/AA42-positive (average T SD = 88 T 6 events/KL) and apoJ-positive/AA42-positive (average T SD = 95 T 7 events/ KL) colabeled particles and then subtracted this sum from the total concentration of AA42-positive particles averaged from the 4 runs for each of the 131 individuals. The average (TSD) difference in concentration between total AA42-positive particle concentration and the sum of apoE-positive/AA42-positive and apoJ-positive/AA42-positive colabeled particles was 23 T 36 events/KL. By this calculation, the majority, and potentially all, of AA42-positive particles were colabeled for apoE and/or apoJ; others have shown that these 2 apolipoproteins can be present on the same CSF lipoprotein (92). Finally, we observed a highly significant correlation between apoE-positive/AA42-positive colabeled particles and apoJ-positive/AA42-positive co-labeled particles (Fig. 3B, slope = 0.99 T 0.05, R2 = 0.78, P G 0.0001). Together, these data show that there are several types of CSF particles that can be identified with these molecular probes, that apoE-positive and apoJ-positive CSF particles are similar in concentration and size, and that most, if not all, AApositive particles are labeled for apoE, apoJ, or both. Furthermore, our data show that approximately one-third to onehalf of apoE-positive and apoJ-positive particles labeled for both proteins, and the AA42 distribution between apoE-positive and apoJ-positive particles was approximately equal.


Yang et al

D3/D4 (P G 0.05 for both) and D4 homozygotes (P G 0.01 for both) showing a progressively greater fraction of larger apoE-positive particles than D3/D3. Figure 4E plots apoEpositive particle size vs. concentration for each individual and shows that APOE D4/D4 individuals have fewer, larger CSF apoE-positive particles. Together these data indicate that APOE D4/D4 participants had CSF apoE-positive particles that are larger but at a lower level on average than individuals with APOE D3/D3 or APOE D3/D4. One possibility is that APOE D4 homozygotes are somehow generally deficient in the ability to elaborate normal amounts of appropriately sized CSF lipoproteins. One test of


this hypothesis would be to compare the relationship between APOE and secretion of another CSF particle; for this we selected the closely related apoJ-positive particles. Another possibility suggested by results of others is compensatory induction of apoJ in brain of APOE D4/D4 AD individuals with low brain levels of apoE (93). To test these possibilities and examine the relationship between the levels of apoE-positive particles and apoJ-positive particles as a function of APOE genotype, we determined the CSF concentration and size of apoJ-positive particles, which, like apoE-positive particles, are produced by CNS cells (94). There was no difference in apoJ-positive particle concentration when stratified by age, by diagnosis (not shown), or by


FIGURE 5. Murine primary cultures of cerebral cortical astrocytes or microglia were prepared from homozygous (D3+/+ or D4+/+) or hemizygous (D3+/j or D4+/j) targeted replacement (TR) mice with human APOE. After 24 hours in culture, the medium was replaced with serum-free medium. After another 24 hours in culture, conditioned medium was assayed for apoE-positive particle concentration and size distribution using flow cytometry. (A) Two-way ANOVA had significant genotype-dependent (***P G 0.001) and gene dose-dependent (***P G 0.001) differences in apoE-positive particle concentration from astrocyte conditioned medium. Corrected posttests had ** P G 0.01 for D3+/+ versus D4+/+. (B) Two-way ANOVA of microglia apoE-positive particles had ** P G 0.01 for gene dose but was not significant for genotype. (C) There was neither genotype- nor gene dose-dependent effects on astrocyte conditioned medium apoE-positive particle size distribution. (D) In contrast, two-way ANOVA of microglia conditioned medium particle size distribution had corrected posttests significant for homozygous (***P G 0.001 for D3+/+ vs. D4+/+) and hemizygous microglia (***P G 0.001 for D3+/- vs. D4+/-).

680




FIGURE 7. Cerebrospinal fluid (CSF) annexin V-positive particle concentration in age and clinical diagnostic groups. One-way ANOVA had ** P G 0.01 for older controls, mild cognitive impairment (MCI), Alzheimer disease (AD) dementia, and Parkinson disease (PD) groups; corrected posttests had ** P G 0.01 for older controls versus AD dementia and # P G 0.01 for MCI versus AD dementia.

greater percentage of larger apoE-positive particles than did TR APOE3 microglia from both homozygous (P G 0.001) and hemizygous (P G 0.001) mice (Fig. 5D). These data demonstrated that the APOE genotype-dependent differences observed in apoE-positive particle concentration and size for human CSF were reproduced experimentally in primary cultures of mouse glia where there is no influence by processes of MCI or AD dementia. Unlike apoE, apoJ is widely synthesized by neurons and glia in brain (94). Using the same conditioned medium as described previously, apoJ-positive particles were detected from primary cultures of all 3 cell types with comparable concentrations between astrocytes and neurons, and microglia about half of these other cells; like CSF, there was no difference in apoJ-positive particles between TR APOE 3/3 and TR APOE4/4 for each of the 3 cell types (Figure, Supplemental Digital Content 5, http://links.lww.com/NEN/A745).

ApoAI-Positive CSF Particles

FIGURE 6. Cerebrospinal fluid (CSF) apoAI-positive particle concentration in age and diagnostic groups. One-way ANOVA had * P G 0.05 for the 3 age groups and *** P G 0.001 for older controls, mild cognitive impairment (MCI), Alzheimer disease (AD) dementia, and Parkinson disease (PD) groups; corrected posttests had ** P G 0.01 for older controls vs. MCI, *** P G 0.001 older controls versus AD dementia, and * P G 0.05 for older controls versus PD.

ApoAI in CSF exists on a subclass of lipoproteins that are largely distinct from apoE-positive CSF lipoproteins (1). Concentration of apoAI-positive particles was progressively increased across the 3 age groups (P G 0.05), corrected paired comparisons significant for young versus older controls (P G 0.05) (Fig. 6). In contrast, CSF apoAI-positive particle concentration was significantly lower in both MCI and AD dementia versus older controls (P G 0.001, corrected paired comparisons had P G 0.01 for older controls vs. MCI and P G 0.001 for older controls vs. AD dementia), but there was no significant difference between MCI and AD dementia. Parkinson disease participants also had lower CSF apoAI-positive particle concentration than older controls after correction for repeated pair comparisons (P G 0.05). The concentration of apoAI-positive



681


APOE genotype (apoJ-positive events/KL): 1585 T 284 in APOE D3/D3 (n = 64); 1237 T 170 in APOE D3/D4 (n = 54); and 1934 T 445 (n = 13) in APOE D4/D4. Moreover, there was no difference in the percentage of larger apoJ-positive particles across these 3 genotypes: 43% T 2% for APOE D3/D3, 42% T 2% for APOE D3/D4, and 46% T 4% for APOE D4/D4. These data indicate that APOE D4/D4 individuals are not inherently deficient in their ability to produce appropriate amounts of different sized apoJ-positive particles in the CNS and contribute these to CSF. Our results suggest a relatively specific association of lower level but larger apoE-positive particles in CSF from APOE D4/D4 participants regardless of diagnosis. We tested this further using primary neuron or glial cultures from homozygous or hemizygous TR APOE3 and TR APOE4 mice. Primary cerebral cortical neuron cultures did not elaborate detectable apoE-positive particles into culture medium over 24 hours. Medium conditioned by primary cerebral cortical astrocyte cultures for 24 hours revealed a significant genotype-dependent (Fig. 5A, P G 0.001) and gene dose-dependent difference in apoE-positive particle concentration (Fig. 5A, P G 0.001). Corrected posttests showed that TR APOE4+/+ astrocyte conditioned medium had significantly lower apoE-positive particle concentration than TR APOE3+/+ astrocyte medium (P G 0.01). Results from microglia cultures were significant for gene dose but not allele type with respect to apoE-positive particle concentration (Fig. 5B, P G 0.01). We next determined apoE-positive particle size in conditioned medium of primary cultures from TR mice. There was neither allele type nor dose effect on astrocyte conditioned medium particle size distribution (Fig. 5C). In contrast, TR APOE4 microglia conditioned medium had a significantly


Yang et al

particles did not vary by APOE genotype in the 131 samples (events/KL): 276 T 20 for APOE D3/D3, 303 T 36 for APOE D3/D4, and 267 T 33 for APOE D4/D4. Distribution by size showed that the proportion of larger apoAI-positive particles did not vary significantly across the 6 diagnostic groups (not shown) or across APOE genotype: 37% T 2% for APOE D3/D3, 38% T 2% for APOE D3/D4, and 40% T 3% for APOE D4/D4. These data indicate an increase in CSF apoAI-positive particle concentration with advancing age that is reversed in the early stages of AD and in PD. As expected from previous work showing no apoAI expression by CNS parenchymal cells (58, 95), we did not detect apoAI-positive particles in conditioned medium from mouse neurons, astrocytes, or microglia, a further demonstration of the specificity of our apoAI CSF particle assay.

Annexin V-Positive CSF Particles

CSF, mostly exosomes and microvesicles, and not the usually larger apoptotic bodies.

AA42-Positive CSF Particles AA42-positive CSF particle concentration was significantly different across the 3 age groups (P G 0.05) with P G 0.05 for young controls versus each of the 2 other age groups (Fig. 8). There was no significant difference across the 3 APOE genotypes with respect to AA42-positive particle concentration or size distribution (not shown). No AA42-positive particle was detected in conditioned medium from mouse primary cerebral cortical neurons, astrocytes, or microglia. Interestingly, the concentration of AA42-positive particles (events/KL) was not significantly associated with the concentration of CSF AA42 as determined by immunoassay and reported above (r = 0.12, n = 131). We interpret these data as evidence of different pools of AA42 in CSF measured by these 2 different methods. Because we had determined AA42 concentration (pg/mL) by immunoassay and AA42-positive particle concentration (events/Kl), we calculated AA42 events/pg, which averaged (SD) 609 T 62 events/pg across all CSF samples, but neither varied significantly by diagnostic group nor correlated with other CSF measurements (not shown); as expected, AA42 events/pg were 77% greater in APOE D4/D4 than the other 2 APOE genotypes (P G 0.01). The average proportion of AA42-positive particles that were larger ranged from 41% to 51% across the 6 diagnostic groups, and 46% T 2% for APOE D3/D3, 45% T 1% for APOE D3/D4, and 50% T 3% for APOE D4/D4; neither was significant. This particle size distribution is consistent with our earlier finding that AA42 is present largely on apoE and/or apoJ containing particles in CSF.

Exploratory Analyses for Correlations Among CSF Particle Concentrations We sought insight into potential biological mechanisms by performing exploratory analysis of correlations among CSF particle concentrations, setting > = 0.01 to minimize false-positive results. We found that annexin V-positive and apoAI-positive particle concentrations were strongly positively correlated with each other for the entire cohort (r = 0.54, P G 0.0001). No other significant correlation of CSF particle concentrations was observed. The correlation between levels of annexin V-positive and apoAI-positive particles was especially strong for PD (r = 0.82, P G 0.0001). Investigation of this intriguing finding will require experimental models different from primary cell culture.

DISCUSSION

FIGURE 8. Cerebrospinal fluid (CSF) A-amyloid (AA42)Ypositive particle levels in age and diagnostic groups. One-way ANOVA had * P G 0.05 for 3 age groups, and * P G 0.05 for older controls, mild cognitive impairment (MCI), and Alzheimer disease (AD) dementia groups. Corrected posttests had * P G 0.05 for young versus middle-aged or versus older controls. There was no significance for paired comparisons in older controls, MCI and AD dementia groups, or between older controls and PD.

682

Here, we describe a flow cytometric assay to detect annexin V-positive, apoE-positive, apoAI-positive, apoJ-positive, and AA42-positive particles ranging from approximately 0.2 Km, the lower limit of size detection, to 1 Km in human CSF; we selected a particle size cutoff of 1 Km to limit the contribution of larger EVs, mostly apoptotic bodies (45). Because 0.2 Km is the cutoff for identifying particles accurately with the Apogee A50 flow cytometer, apolipoprotein, or AA42 containing particles smaller than 0.2 Km or their free fractions could not be detected using this method. We demonstrated that apoE-positive Ó 2015 American Association of Neuropathologists, Inc.



We observed significant variation in annexin V-positive particle levels (events/KL) among older controls, MCI, AD dementia, and PD groups (Fig. 7, P G 0.01) but not among the 3 age groups. Annexin V-positive particle concentration was significantly lower in AD dementia than in older controls (P G 0.01) but not for MCI versus older controls; however, AD dementia was significantly different from MCI (P G 0.01). There was no difference in CSF annexin V-positive particle concentration among APOE D3/D3 (2649 T 226 events/KL, n = 64), APOE D3/D4 (2929 T 268 events/KL, n = 54), and APOE D4/D4 (2310 T 274 events/KL, n = 13) individuals. The proportion of annexin V-positive particles that were larger did not vary significantly across the 6 diagnostic groups (not shown), or with APOE: 16% T 1% for APOE D3/D3, 15% T 1% for APOE D3/D4, and 18% T 1% for APOE D4/D4. This size distribution suggests that annexin V detected smaller subclasses of EVs in



APOE and CSF Particles Peripheral apoE is generated mostly by liver and endothelial cells; CNS apoE is generated mostly by glia (96). The APOE genotype is associated reproducibly with apoE protein levels in plasma and serum, with APOE D4 predicting lower apoE protein levels in humans (7, 11, 97Y99) and in animal models (100) because of preferential isoform binding to different lipoproteins that are cleared at different rates (98, 101, 102). ApoE4 has increased affinity for low-density lipoprotein receptor with more rapid clearance of apoEcontaining remnant lipoproteins. Carriers of apoE4 have increased low-density lipoprotein cholesterol levels that contribute to increased risk for cardiovascular disease and may linked to subsequent neurodegenerative disease and dementia (103, 104). In contrast, the effect of APOE D4 on CNS apoE protein levels is not clear for brain (13, 100, 105Y107) or CSF (107). One potential limitation of previous CSF studies is not distinguishing particle-bound from total apoE. Our data show a highly significant association between lower concentration of CSF apoE-positive particles in APOE D4/D4 CSF than in APOE D3/D3 or APOE D3/D4 CSF. Importantly, no significant association was observed between APOE genotype and particle concentration or size for any other CSF particle assayed, including similar apoJ-positive particles. Although limited by the small number of subjects, our findings that APOE D4/ D4 subjects have apoE-positive particles that are larger but at a lower level on average than in APOE D3/D3 or APOE D3/D4 may provide insight into the marked acceleration of AD risk for APOE D4/D4. Indeed, it has been suggested that differential direct interaction of apoE isoforms with AA influences AA clearance and/or aggregation in the CNS. ApoE4 increases AA aggregation and impairs clearance relative to other apoE isoforms, thus providing a mechanistic explanation for how apoE isoforms influence the risk of AD pathogenesis (17, 108, 109). Further mechanistic studies are needed to characterize the role of these particles in relation to neurodegeneration. We modeled the influence of APOE on generators of CNS apoE by using murine primary cerebral cultures of neurons,

astrocytes, or microglia from TR APOE mice. We demonstrated that glia were the major source of particle-associated apoE, and validated our findings from human CSF, although there was a somewhat complex relationship among glial cell type, APOE gene and dose, and apoE-positive particles. The concentration of apoE-positive particles depended strongly on gene dose in astrocytes but not in microglia. Importantly, one of our major observations, i.e. that APOE D4/D4 associated with significantly lower apoE-positive particle concentration in human CSF, was replicated in homozygous astrocyte cultures; although the same trend was observed in hemizygous mice, the difference here was not significant. Consistent with this, others have shown that astrocytes preferentially degrade apoE4, leading to reduced apoE secretion and concentration (100). The other major observation from human CSF, that is, apoE-positive particles from APOE D4/D4 CSF are larger, was not replicated in astrocytes but was replicated with microglia cultures homozygous or hemizygous for TR APOE4. Again, these results are consistent with the work of others that showed astrocytesecreted particles are the same size whether containing apoE4 or apoE3 (110) and that microglia-produced apoE particles differ from astrocytes in size and shape (111). Interestingly, microglia-conditioned medium contained a greater concentration of apoE-positive particles than conditioned medium from astrocytes, consistent with our previous demonstration that on a per cell basis, murine primary cerebral microglia synthesize more apoE than astrocytes (112). We cannot exclude the potential contribution of neuron-derived apoE-positive-containing particles in vivo, although none were detected in neuron primary cultures under resting conditions (113). Indeed, CNS neurons can express apoE (114Y116) but at lower levels than glia. In addition, others have shown neuron-derived apoE3 and apoE4 differ in their susceptibility to proteolysis, with more fragments being generated from apoE4 than from apoE3 (117). Genetic variants of both APOE and CLU, the gene that encodes apoJ, are associated with AD risk. Others have shown that apolipoprotein-containing particles interact with AA peptides and may influence AA metabolism and clearance (118Y123). Consistent with this, we observed that apoE-positive and apoJpositive CSF particles co-labeled with AA42 in roughly equal proportion, and that this accounted within experimental error for all AA42-positive particles. Although we cannot determine relative activity with respect to AA metabolism from these data, our results from human CSF indicate quantitatively similar binding of AA42 by apoE-positive and apoJ-positive particles, perhaps even the subset of particles co-labeled with apoE and apoJ.

Aging and CSF Particles We observed 3 different relationships between advancing age and CSF particle concentration; there was no significant relationship between age and size distribution. Cerebrospinal fluid apoE-positive particle concentration was reduced by about half and AA42-positive particle concentration was reduced by about one-quarter in middle-aged adults compared to young adults; this reduction persisted in older adults. Interestingly, although similar in size distribution and concentration to apoEpositive particles, CSF apoJ-positive particles did not show significant differences across age groups. Annexin V-positive particles also showed no change in concentration across age groups.



683


and apoJ-positive particles are largely indistinguishable and partially colabeled, that the vast majority if not all particleassociated AA42 is on apoE-positive and/or apoJ-positive particles, and that annexin V-positive particles partially overlapped but are on average smaller than particles bearing apolipoproteins. Our analyses focused on associations with APOE, aging, or the disease states MCI, AD dementia, or PD. Our data suggested that APOE D4/D4 participants specifically had CSF apoE-positive particles that were larger but at a lower level on average than in APOE D3/D3 or APOE D3/D4 participants. This finding was reproduced experimentally in conditioned medium from TR APOE mouse primary glia cultures. Changes related to age included reduction in CSF apoE-positive and AA42-positive particle concentrations in middle age that persisted into older age and progressive age-related increase in apoAI-positive particle concentration. Both apoAI-positive and annexin V-positive CSF particle levels were significantly lower in human CSF of MCI and/or AD dementia than in agematched controls.


Yang et al

Only apoAI-positive particles increased in concentration from young to middle to older age groups, approximately doubling in average concentration across the adult human lifespan.

CSF Particles in MCI, AD, or PD

684

expected, the conditioned medium from murine primary cerebral cortical neurons, astrocyte, or microglia had no detectable apoAI-positive particles. Further in vivo investigation will be needed to understand the mechanistic relationships and significance of apoAI and annexin V binding particles in CNS. There is no consistent association of total CSF apoE or apoJ protein concentration and the diagnosis of MCI, AD dementia, or PD (7Y10, 12Y16, 27, 37, 38, 61, 129). Here, neither concentration nor size distribution of apoE-positive or apoJ-positive particles in CSF was associated with a diagnosis of MCI or AD dementia, or with PD. However, for these and other results, especially those stratified by disease group and genotype, some of the sample sizes are small, and some of the results from our novel approach should be viewed as requiring confirmation from larger sample sets. In summary, the aim of this work was to develop and apply flow cytometry to the quantification, molecular composition, and size distribution of CSF particles and determine the relationship of these data to brain aging and common neurodegenerative diseases. In the absence of a validation cohort, and because cases and controls were highly selected, we did not formally analyze diagnostic sensitivity and specificity. Given the age and diagnostic group differences identified in this novel approach to CSF particle analysis, we expect that our data may be used as a platform to develop future biomarkers. Our major findings were that APOE D4/D4 is associated with fewer, larger particles; that apoE-positive and apoJ-positive particles each account for approximately half of particle-bound AA42, that total AA42 concentration is not correlated with AA42-positive particle concentration; and that apoAI-positive particles increase in concentration with advancing age and this is reversed in patients with MCI, AD dementia, and PD. Although further observation and experimental work is needed to understand the significance of CSF particles and their relationships to exosomes, other EVs, or CSF lipoproteins (58, 110, 130), our unique approach opens a new window into translational research for neurodegenerative diseases. REFERENCES 1. Montine T, Montine K, Swift L. Central nervous system lipoproteins in Alzheimer’s disease. Am J Pathol 1997;151:1571Y5 2. Koch S, Donarski N, Goetze K, et al. Characterization of four lipoprotein classes in human cerebrospinal fluid. J Lipid Res 2001;42:1143Y51 3. Borghini I, Barja F, Pometta D, et al. Characterization of subpopulations of lipoprotein particles isolated from human cerebrospinal fluid. Biochim Biophys Acta 1995;1255:192Y200 4. Pitas RE, Boyles JK, Lee SH, et al. Lipoproteins and their receptors in the central nervous system. Characterization of the lipoproteins in cerebrospinal fluid and identification of apolipoprotein B,E(LDL) receptors in the brain. J Biol Chem 1987;262:14352Y60 5. Roheim PS, Carey M, Forte T, et al. Apolipoproteins in human cerebrospinal fluid. Proc Natl Acad Sci U S A 1979;76:4646Y9 6. Cruchaga C, Kauwe JS, Nowotny P, et al. Cerebrospinal fluid APOE levels: an endophenotype for genetic studies for Alzheimer’s disease. Hum Mol Genet 2012;21:4558Y71 7. Lehtimaki T, Pirttila T, Mehta PD, et al. Apolipoprotein E (apoE) polymorphism and its influence on ApoE concentrations in the cerebrospinal fluid in Finnish patients with Alzheimer’s disease. Hum Genet 1995;95:39Y42 8. Landen M, Hesse C, Fredman P, et al. Apolipoprotein E in cerebrospinal fluid from patients with Alzheimer’s disease and other forms of dementia is reduced but without any correlation to the apoE4 isoform. Dementia 1996;7:273Y8




Perhaps our most intriguing observation was a lack of correlation between CSF concentration of AA42 and AA42positive particles across all groups as well in subanalyses of MCI and AD dementia or PD groups. The failure of correlation may reflect different fractions of AA42 targeted by different methods; for example, particles smaller than 0.2 Km or particlefree AA42 would not be detected by our flow cytometric method. As mentioned, AA peptides are known to interact with lipid particles, and it is entirely possible that that our novel flow cytometry approach detects a subset of particle-bound AA42 that has a metabolism different from particle-free AA42. We did not detect AA42 in murine primary cultures from TR APOE mice and will need to pursue this potentially important finding in different models. Our most statistically robust association with disease state was CSF apoAI-positive particle concentration; we observed that CSF concentration of apoAI-positive particles are at a lower level in MCI, AD dementia, and PD compared with the older controls; reversing the age-related increase in CSF apoAI-positive particle concentration discussed previously. In sharp contrast to apoE, apoAI is not synthesized in the CNS, and its presence in CSF indicates transit across the blood-brain barrier (124). Polymorphisms in the gene that encodes apoAI have been correlated with AD dementia or all-cause dementia (19, 20). Plasma and serum apoAI levels are decreased in patients with AD dementia (22Y24), and serum levels of apoAI have been associated inversely with risk of AD dementia (21). Furthermore, recent studies have shown the lack of apoAI increases cerebral amyloid angiopathy and memory deficits of AD dementia (125), and increasing plasma apoAI levels provides cognitive and neuropathological benefits in animal models of AD dementia (126). Finally, others have observed that lower plasma levels of apoAI correlate with increased risk and earlier onset of PD (26). One interpretation of our results and those from other studies is that reduction in the amount of functional apoAI in the CNS is permissive for the initiation or progression of neurodegeneration through a variety of mechanisms or for clinical expression. This raises the possibility of increasing CNS apoAI activity as a therapeutic target for AD and PD. Interestingly, we observed a strong correlation of the level of apoAI-positive and annexin V-positive CSF particle concentrations across all samples. Others have demonstrated an interaction between apoAI and membrane lipids with PS exofacial flopping, suggesting that apoAI and annexin V may share the ability to bind to cell surface PS (127, 128); this would readily account for the observed association. Like apoAIpositive particles, we did observe a strong association between CSF annexin V-positive particle concentration and AD dementia. In contrast to apoAI-positive particles, there was no association between annexin V-positive particle concentration and MCI or PD, suggesting that this change in CSF annexin V binding is relatively specific to later stage of AD, perhaps a reflection of large-scale cerebral neurodegeneration. We were unable to pursue this idea further in our primary culture models because, as



33. 34. 35. 36. 37. 38. 39. 40.

41. 42. 43. 44. 45. 46.

47. 48.

49. 50. 51.

52. 53. 54. 55. 56. 57.

(apolipoprotein J), an inhibitor of the complement membrane-attack complex. Biochem J 1993;293:27Y30 Boggs LN, Fuson KS, Baez M, et al. Clusterin (Apo J) protects against in vitro amyloid-beta (1Y40) neurotoxicity. J Neurochem 1996;67:1324Y7 Matsubara E, Frangione B, Ghiso J. Characterization of apolipoprotein J-Alzheimer’s A beta interaction. J Biol Chem 1995;270:7563Y7 DeMattos RB, Cirrito JR, Parsadanian M, et al. ApoE and clusterin cooperatively suppress Abeta levels and deposition: evidence that ApoE regulates extracellular Abeta metabolism in vivo. Neuron 2004;41:193Y202 Desikan RS, Thompson WK, Holland D, et al. The role of clusterin in amyloid-beta-associated neurodegeneration. JAMA Neurol 2014;71: 180Y7 Prikrylova Vranova H, Mares J, Nevrly M, et al. CSF markers of neurodegeneration in Parkinson’s disease. J Neural Transm 2010;117:1177Y81 van Dijk KD, Jongbloed W, Heijst JA, et al. Cerebrospinal fluid and plasma clusterin levels in Parkinson’s disease. Parkinsonism Relat Disord 2013;19:1079Y83 Braccioli L, van Velthoven C, Heijnen CJ. Exosomes: a new weapon to treat the central nervous system. Mol Neurobiol 2014;49:113Y9 Heijnen HF, Schiel AE, Fijnheer R, et al. Activated platelets release two types of membrane vesicles: microvesicles by surface shedding and exosomes derived from exocytosis of multivesicular bodies and alphagranules. Blood 1999;94:3791Y9 Vader P, Breakefield XO, Wood MJ. Extracellular vesicles: emerging targets for cancer therapy. Trends Mol Med 2014;20:385Y93 Gyorgy B, Szabo TG, Pasztoi M, et al. Membrane vesicles, current stateof-the-art: emerging role of extracellular vesicles. Cell Mol Life Sci 2011; 68:2667Y88 Thery C, Ostrowski M, Segura E. Membrane vesicles as conveyors of immune responses. Nat Rev Immunol 2009;9:581Y93 Raposo G, Stoorvogel W. Extracellular vesicles: exosomes, microvesicles, and friends. J Cell Biol 2013;200:373Y83 Burger D, Schock S, Thompson CS, et al. Microparticles: biomarkers and beyond. Clin Sci (Lond) 2013;124:423Y41 Dachary-Prigent J, Freyssinet JM, Pasquet JM, et al. Annexin V as a probe of aminophospholipid exposure and platelet membrane vesiculation: a flow cytometry study showing a role for free sulfhydryl groups. Blood 1993;81:2554Y65 Vella LJ, Greenwood DL, Cappai R, et al. Enrichment of prion protein in exosomes derived from ovine cerebral spinal fluid. Vet Immunol Immunopathol 2008;124:385Y93 Wang S, Cesca F, Loers G, et al. Synapsin I is an oligomannose-carrying glycoprotein, acts as an oligomannose-binding lectin, and promotes neurite outgrowth and neuronal survival when released via glia-derived exosomes. J Neurosci 2011;31:7275Y90 Xin H, Li Y, Buller B, et al. Exosome-mediated transfer of miR-133b from multipotent mesenchymal stromal cells to neural cells contributes to neurite outgrowth. Stem Cells 2012;30:1556Y64 Morel L, Regan M, Higashimori H, et al. Neuronal exosomal miRNAdependent translational regulation of astroglial glutamate transporter GLT1. J Biol Chem 2013;288:7105Y16 Zachau AC, Landen M, Mobarrez F, et al. Leukocyte-derived microparticles and scanning electron microscopic structures in two fractions of fresh cerebrospinal fluid in amyotrophic lateral sclerosis: a case report. J Med Case Rep 2012;6:274 Rajendran L, Honsho M, Zahn TR, et al. Alzheimer’s disease betaamyloid peptides are released in association with exosomes. Proc Natl Acad Sci U S A 2006;103:11172Y7 Sharples RA, Vella LJ, Nisbet RM, et al. Inhibition of gamma-secretase causes increased secretion of amyloid precursor protein C-terminal fragments in association with exosomes. FASEB J 2008;22:1469Y78 Bulloj A, Leal MC, Xu H, et al. Insulin-degrading enzyme sorting in exosomes: a secretory pathway for a key brain amyloid-beta degrading protease. J Alzheimers Dis 2010;19:79Y95 Saman S, Kim W, Raya M, et al. Exosome-associated tau is secreted in tauopathy models and is selectively phosphorylated in cerebrospinal fluid in early Alzheimer disease. J Biol Chem 2012;287:3842Y9 Emmanouilidou E, Melachroinou K, Roumeliotis T, et al. Cell-produced alpha-synuclein is secreted in a calcium-dependent manner by exosomes and impacts neuronal survival. J Neurosci 2010;30:6838Y51 Perez-Gonzalez R, Gauthier SA, Kumar A, et al. The exosome secretory pathway transports amyloid precursor protein carboxyl-terminal



685


9. Mulder M, Ravid R, Swaab DF, et al. Reduced levels of cholesterol, phospholipids, and fatty acids in cerebrospinal fluid of Alzheimer disease patients are not related to apolipoprotein E4. Alzheimer Dis Assoc Disord 1998;12:198Y203 10. Lefranc D, Vermersch P, Dallongeville J, et al. Relevance of the quantification of apolipoprotein E in the cerebrospinal fluid in Alzheimer’s disease. Neurosci Lett 1996;212:91Y4 11. Schiele F, De Bacquer D, Vincent-Viry M, et al. Apolipoprotein E serum concentration and polymorphism in six European countries: the ApoEurope Project. Atherosclerosis 2000;152:475Y88 12. Blennow K, Hesse C, Fredman P. Cerebrospinal fluid apolipoprotein E is reduced in Alzheimer’s disease. Neuroreport 1994;5:2534Y6 13. Beffert U, Cohn JS, Petit-Turcotte C, et al. Apolipoprotein E and betaamyloid levels in the hippocampus and frontal cortex of Alzheimer’s disease subjects are disease-related and apolipoprotein E genotype dependent. Brain Res 1999;843:87Y94 14. Yamauchi K, Tozuka M, Nakabayashi T, et al. Apolipoprotein E in cerebrospinal fluid: relation to phenotype and plasma apolipoprotein E concentrations. Clin Chem 1999;45:497Y504 15. Montine KS, Bassett CN, Ou JJ, et al. Apolipoprotein E allelic influence on human cerebrospinal fluid apolipoproteins. J Lipid Res 1998;39:2443Y51 16. Merched A, Blain H, Visvikis S, et al. Cerebrospinal fluid apolipoprotein E level is increased in late-onset Alzheimer’s disease. J Neurol Sci 1997;145:33Y9 17. Verghese PB, Castellano JM, Garai K, et al. ApoE influences amyloidbeta (Abeta) clearance despite minimal apoE/Abeta association in physiological conditions. Proc Natl Acad Sci U S A 2013;110:E1807Y16 18. LaDu MJ, Munson GW, Jungbauer L, et al. Preferential interactions between ApoE-containing lipoproteins and Abeta revealed by a detection method that combines size exclusion chromatography with nonreducing gel-shift. Biochim Biophys Acta 2012;1821:295Y302 19. Vollbach H, Heun R, Morris CM, et al. APOA1 polymorphism influences risk for early-onset nonfamiliar AD. Ann Neurol 2005;58: 436Y41 20. Helbecque N, Codron V, Cottel D, et al. An apolipoprotein A-I gene promoter polymorphism associated with cognitive decline, but not with Alzheimer’s disease. Dement Geriatr Cogn Disord 2008;25:97Y102 21. Saczynski JS, White L, Peila RL, et al. The relation between apolipoprotein AYI and dementia: the Honolulu-Asia aging study. Am J Epidemiol 2007;165:985Y92 22. Kawano M, Kawakami M, Otsuka M, et al. Marked decrease of plasma apolipoprotein AI and AII in Japanese patients with late-onset nonfamilial Alzheimer’s disease. Clin Chim Acta 1995;239:209Y11 23. Merched A, Xia Y, Visvikis S, et al. Decreased high-density lipoprotein cholesterol and serum apolipoprotein AI concentrations are highly correlated with the severity of Alzheimer’s disease. Neurobiol Aging 2000; 21:27Y30 24. Liu HC, Hu CJ, Chang JG, et al. Proteomic identification of lower apolipoprotein A-I in Alzheimer’s disease. Dement Geriatr Cogn Disord 2006;21:155Y61 25. Wang ES, Sun Y, Guo JG, et al. Tetranectin and apolipoprotein AYI in cerebrospinal fluid as potential biomarkers for Parkinson’s disease. Acta Neurol Scand 2010;122:350Y9 26. Qiang JK, Wong YC, Siderowf A, et al. Plasma apolipoprotein A1 as a biomarker for Parkinson disease. Ann Neurol 2013;74:119Y27 27. Maarouf CL, Beach TG, Adler CH, et al. Cerebrospinal fluid biomarkers of neuropathologically diagnosed Parkinson’s disease subjects. Neurol Res 2012;34:669Y76 28. Paula-Lima AC, Tricerri MA, Brito-Moreira J, et al. Human apolipoprotein A-I binds amyloid-beta and prevents Abeta-induced neurotoxicity. Int J Biochem Cell Biol 2009;41:1361Y70 29. Koldamova RP, Lefterov IM, Lefterova MI, et al. Apolipoprotein AYI directly interacts with amyloid precursor protein and inhibits A beta aggregation and toxicity. Biochemistry 2001;40:3553Y60 30. Lambert JC, Heath S, Even G, et al. Genome-wide association study identifies variants at CLU and CR1 associated with Alzheimer’s disease. Nat Genet 2009;41:1094Y9 31. Harold D, Abraham R, Hollingworth P, et al. Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer’s disease. Nat Genet 2009;41:1088Y93 32. Ghiso J, Matsubara E, Koudinov A, et al. The cerebrospinal-fluid soluble form of Alzheimer’s amyloid beta is complexed to SP-40,40


Yang et al

58. 59. 60. 61. 62. 63. 64. 65.

67. 68. 69. 70. 71. 72.

73.

74.

75. 76.

77. 78. 79. 80.

686

81. Sullivan PM, Mezdour H, Aratani Y, et al. Targeted replacement of the mouse apolipoprotein E gene with the common human APOE3 allele enhances diet-induced hypercholesterolemia and atherosclerosis. J Biol Chem 1997;272:17972Y80 82. Maezawa I, Maeda N, Montine TJ, et al. Apolipoprotein E-specific innate immune response in astrocytes from targeted replacement mice. J Neuroinflammation 2006;3:10 83. Cudaback E, Li X, Montine KS, et al. Apolipoprotein E isoform-dependent microglia migration. FASEB J 2011;25:2082Y91 84. Cudaback E, Li X, Yang Y, et al. Apolipoprotein C-I is an APOE genotype-dependent suppressor of glial activation. J Neuroinflammation 2012;9:192 85. Li X, Rose SE, Montine KS, et al. Antagonism of neuronal prostaglandin E(2) receptor subtype 1 mitigates amyloid beta neurotoxicity in vitro. J Neuroimmune Pharmacol 2013;8:87Y93 86. Carrillo MC, Blennow K, Soares H, et al. Global standardization measurement of cerebral spinal fluid for Alzheimer’s disease: an update from the Alzheimer’s Association Global Biomarkers Consortium. Alzheimers Dement 2013;9:137Y40 87. Li G, Sokal I, Quinn JF, et al. CSF tau/Abeta42 ratio for increased risk of mild cognitive impairment: a follow-up study. Neurology 2007;69:631Y9 88. Montine TJ, Shi M, Quinn JF, et al. CSF Abeta(42) and tau in Parkinson’s disease with cognitive impairment. Mov Disord 2010;25:2682Y5 89. Shi M, Bradner J, Hancock AM, et al. Cerebrospinal fluid biomarkers for Parkinson disease diagnosis and progression. Ann Neurol 2011;69: 570Y80 90. Kang JH, Irwin DJ, Chen-Plotkin AS, et al. Association of cerebrospinal fluid beta-amyloid 1-42, T-tau, P-tau181, and alpha-synuclein levels with clinical features of drug-naive patients with early Parkinson disease. JAMA Neurol 2013;70:1277Y87 91. Woltjer RL, Sonnen JA, Sokal I, et al. Quantitation and mapping of cerebral detergent-insoluble proteins in the elderly. Brain Pathol 2009; 19:365Y74 92. Suzuki T, Tozuka M, Kazuyoshi Y, et al. Predominant apolipoprotein J exists as lipid-poor mixtures in cerebrospinal fluid. Ann Clin Lab Sci 2002;32:369Y76 93. Bertrand P, Poirier J, Oda T, et al. Association of apolipoprotein E genotype with brain levels of apolipoprotein E and apolipoprotein J (clusterin) in Alzheimer disease. Brain Res Mol Brain Res 1995;33:174Y8 94. Page KJ, Hollister RD, Hyman BT. Dissociation of apolipoprotein and apolipoprotein receptor response to lesion in the rat brain: an in situ hybridization study. Neuroscience 1998;85:1161Y71 95. Elshourbagy NA, Liao WS, Mahley RW, et al. Apolipoprotein E mRNA is abundant in the brain and adrenals, as well as in the liver, and is present in other peripheral tissues of rats and marmosets. Proc Natl Acad Sci U S A 1985;82:203Y7 96. Mahley RW, Weisgraber KH, Huang Y. Apolipoprotein E: structure determines function, from atherosclerosis to Alzheimer’s disease to AIDS. J Lipid Res 2009;50(Suppl):S183Y8 97. Eto M, Watanabe K, Ishii K. Reciprocal effects of apolipoprotein E alleles (epsilon 2 and epsilon 4) on plasma lipid levels in normolipidemic subjects. Clin Genet 1986;29:477Y84 98. Gregg RE, Zech LA, Schaefer EJ, et al. Abnormal in vivo metabolism of apolipoprotein E4 in humans. J Clin Invest 1986;78:815Y21 99. Hanis CL, Hewett-Emmett D, Douglas TC, et al. Effects of the apolipoprotein E polymorphism on levels of lipids, lipoproteins, and apolipoproteins among Mexican-Americans in Starr County, Texas. Arterioscler Thromb 1991;11:362Y70 100. Riddell DR, Zhou H, Atchison K, et al. Impact of apolipoprotein E (ApoE) polymorphism on brain ApoE levels. J Neurosci 2008;28:11445Y53 101. Weisgraber KH. Apolipoprotein E distribution among human plasma lipoproteins: role of the cysteine-arginine interchange at residue 112. J Lipid Res 1990;31:1503Y11 102. Steinmetz A, Jakobs C, Motzny S, et al. Differential distribution of apolipoprotein E isoforms in human plasma lipoproteins. Arteriosclerosis 1989;9:405Y11 103. Reilly M, Rader DJ. Apolipoprotein E and coronary disease: a puzzling paradox. PLoS Med 2006;3:e258 104. Stampfer MJ. Cardiovascular disease and Alzheimer’s disease: common links. J Intern Med 2006;260:211Y23 105. Ramaswamy G, Xu Q, Huang Y, et al. Effect of domain interaction on apolipoprotein E levels in mouse brain. J Neurosci 2005;25:10658Y63




66.

fragments from the cell into the brain extracellular space. J Biol Chem 2012;287:43108Y15 LaDu MJ, Gilligan SM, Lukens JR, et al. Nascent astrocyte particles differ from lipoproteins in CSF. J Neurochem 1998;70:2070Y81 Bassett CN, Neely MD, Sidell KR, et al. Cerebrospinal fluid lipoproteins are more vulnerable to oxidation in Alzheimer’s disease and are neurotoxic when oxidized ex vivo. Lipids 1999;34:1273Y80 Hesse C, Larsson H, Fredman P, et al. Measurement of apolipoprotein E (apoE) in cerebrospinal fluid. Neurochem Res 2000;25:511Y7 Wahrle SE, Shah AR, Fagan AM, et al. Apolipoprotein E levels in cerebrospinal fluid and the effects of ABCA1 polymorphisms. Mol Neurodegener 2007;2:7 Chandler WL. Microparticle counts in platelet-rich and platelet-free plasma, effect of centrifugation and sample-processing protocols. Blood Coagul Fibrinolysis 2013;24:125Y32 Lacroix R, Judicone C, Poncelet P, et al. Impact of pre-analytical parameters on the measurement of circulating microparticles: towards standardization of protocol. J Thromb Haemost 2012;10:437Y46 Lacroix R, Robert S, Poncelet P, et al. Overcoming limitations of microparticle measurement by flow cytometry. Semin Thromb Hemost 2010; 36:807Y18 Orozco AF, Lewis DE. Flow cytometric analysis of circulating microparticles in plasma. Cytometry A 2010;77:502Y14 Sokolow S, Henkins KM, Bilousova T, et al. Pre-synaptic C-terminal truncated tau is released from cortical synapses in Alzheimer’s disease. J Neurochem 2014 Henkins KM, Sokolow S, Miller CA, et al. Extensive p-tau pathology and SDS-stable p-tau oligomers in Alzheimer’s cortical synapses. Brain Pathol 2012;22:826Y33 Sokolow S, Henkins KM, Bilousova T, et al. AD synapses contain abundant Abeta monomer and multiple soluble oligomers, including a 56-kDa assembly. Neurobiol Aging 2012;33:1545Y55 Postupna NO, Keene CD, Latimer C, et al. Flow cytometry analysis of synaptosomes from post-mortem human brain reveals changes specific to Lewy body and Alzheimer’s disease. Lab Invest 2014;94:1161Y72 Sokolow S, Luu SH, Nandy K, et al. Preferential accumulation of amyloid-beta in presynaptic glutamatergic terminals (VGluT1 and VGluT2) in Alzheimer’s disease cortex. Neurobiol Dis 2012;45:381Y7 Chandler WL, Yeung W, Tait JF. A new microparticle size calibration standard for use in measuring smaller microparticles using a new flow cytometer. J Thromb Haemost 2011;9:1216Y24 Robert S, Poncelet P, Lacroix R, et al. Standardization of platelet-derived microparticle counting using calibrated beads and a Cytomics FC500 routine flow cytometer: a first step towards multicenter studies? J Thromb Haemost 2009;7:190Y7 McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:263Y9 Albert MS, DeKosky ST, Dickson D, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011; 7:270Y9 Gibb WR, Lees AJ. The relevance of the Lewy body to the pathogenesis of idiopathic Parkinson’s disease. J Neurol Neurosurg Psychiatry 1988; 51:745Y52 Li G, Millard SP, Peskind ER, et al. Cross-sectional and longitudinal relationships between cerebrospinal fluid biomarkers and cognitive function in people without cognitive impairment from across the adult life span. JAMA Neurol 2014;71:742Y51 Cholerton BA, Zabetian CP, Quinn JF, et al. Pacific Northwest Udall Center of excellence clinical consortium: study design and baseline cohort characteristics. J Parkinsons Dis 2013;3:205Y14 Peskind ER, Riekse R, Quinn JF, et al. Safety and acceptability of the research lumbar puncture. Alzheimer Dis Assoc Disord 2005;19:220Y5 Nielsen MH, Beck-Nielsen H, Andersen MN, et al. A flow cytometric method for characterization of circulating cell-derived microparticles in plasma. J Extracell Vesicles 2014;3:doi: 10.3402/jev.v3.20795 Knouff C, Hinsdale ME, Mezdour H, et al. Apo E structure determines VLDL clearance and atherosclerosis risk in mice. J Clin Invest 1999; 103:1579Y86



118. Strittmatter WJ, Weisgraber KH, Huang DY, et al. Binding of human apolipoprotein E to synthetic amyloid beta peptide: isoform-specific effects and implications for late-onset Alzheimer disease. Proc Natl Acad Sci U S A 1993;90:8098Y102 119. Strittmatter WJ, Roses AD. Apolipoprotein E and Alzheimer’s disease. Annu Rev Neurosci 1996;19:53Y77 120. LaDu MJ, Falduto MT, Manelli AM, et al. Isoform-specific binding of apolipoprotein E to beta-amyloid. J Biol Chem 1994;269:23403Y6 121. Bales KR, Verina T, Dodel RC, et al. Lack of apolipoprotein E dramatically reduces amyloid beta-peptide deposition. Nat Genet 1997;17:263Y4 122. Permanne B, Perez C, Soto C, et al. Detection of apolipoprotein E/dimeric soluble amyloid beta complexes in Alzheimer’s disease brain supernatants. Biochem Biophys Res Commun 1997;240:715Y20 123. Holtzman DM, Bales KR, Wu S, et al. Expression of human apolipoprotein E reduces amyloid-beta deposition in a mouse model of Alzheimer’s disease. J Clin Invest 1999;103:R15Yr21 124. Yu C, Youmans KL, LaDu MJ. Proposed mechanism for lipoprotein remodelling in the brain. Biochim Biophys Acta 2010;1801:819Y23 125. Lefterov I, Fitz NF, Cronican AA, et al. Apolipoprotein AYI deficiency increases cerebral amyloid angiopathy and cognitive deficits in APP/ PS1DeltaE9 mice. J Biol Chem 2010;285:36945Y57 126. Lewis TL, Cao D, Lu H, et al. Overexpression of human apolipoprotein AYI preserves cognitive function and attenuates neuroinflammation and cerebral amyloid angiopathy in a mouse model of Alzheimer disease. J Biol Chem 2010;285:36958Y68 127. Chambenoit O, Hamon Y, Marguet D, et al. Specific docking of apolipoprotein AYI at the cell surface requires a functional ABCA1 transporter. J Biol Chem 2001;276:9955Y60 128. Smith JD, Waelde C, Horwitz A, et al. Evaluation of the role of phosphatidylserine translocase activity in ABCA1-mediated lipid efflux. J Biol Chem 2002;277:17797Y803 129. Lindh M, Blomberg M, Jensen M, et al. Cerebrospinal fluid apolipoprotein E (apoE) levels in Alzheimer’s disease patients are increased at follow up and show a correlation with levels of tau protein. Neurosci Lett 1997;229:85Y8 130. Fagan AM, Holtzman DM, Munson G, et al. Unique lipoproteins secreted by primary astrocytes from wild type, apoE (-/-), and human apoE transgenic mice. J Biol Chem 1999;274:30001Y7



687


106. Sullivan PM, Mace BE, Maeda N, et al. Marked regional differences of brain human apolipoprotein E expression in targeted replacement mice. Neuroscience 2004;124:725Y33 107. Fryer JD, Simmons K, Parsadanian M, et al. Human apolipoprotein E4 alters the amyloid-beta 40:42 ratio and promotes the formation of cerebral amyloid angiopathy in an amyloid precursor protein transgenic model. J Neurosci 2005;25:2803Y10 108. Castellano JM, Kim J, Stewart FR, et al. Human apoE isoforms differentially regulate brain amyloid-beta peptide clearance. Sci Transl Med 2011;3:89ra57 109. Rebeck GW, Reiter JS, Strickland DK, et al. Apolipoprotein E in sporadic Alzheimer’s disease: allelic variation and receptor interactions. Neuron 1993;11:575Y80 110. DeMattos RB, Brendza RP, Heuser JE, et al. Purification and characterization of astrocyte-secreted apolipoprotein E and J-containing lipoproteins from wild-type and human apoE transgenic mice. Neurochem Int 2001;39:415Y25 111. Xu Q, Li Y, Cyras C, et al. Isolation and characterization of apolipoproteins from murine microglia. Identification of a low density lipoproteinlike apolipoprotein J-rich but E-poor spherical particle. J Biol Chem 2000; 275:31770Y7 112. Yang Y, Cudaback E, Jorstad NL, et al. APOE3, but not APOE4, bone marrow transplantation mitigates behavioral and pathological changes in a mouse model of Alzheimer disease. Am J Pathol 2013;183:905Y17 113. Chen HK, Liu Z, Meyer-Franke A, et al. Small molecule structure correctors abolish detrimental effects of apolipoprotein E4 in cultured neurons. J Biol Chem 2012;287:5253Y66 114. Han SH, Einstein G, Weisgraber KH, et al. Apolipoprotein E is localized to the cytoplasm of human cortical neurons: a light and electron microscopic study. J Neuropathol Exp Neurol 1994;53:535Y44 115. Metzger RE, LaDu MJ, Pan JB, et al. Neurons of the human frontal cortex display apolipoprotein E immunoreactivity: implications for Alzheimer’s disease. J Neuropathol Exp Neurol 1996;55:372Y80 116. Xu PT, Gilbert JR, Qiu HL, et al. Specific regional transcription of apolipoprotein E in human brain neurons. Am J Pathol 1999;154:601Y11 117. Brecht WJ, Harris FM, Chang S, et al. Neuron-specific apolipoprotein e4 proteolysis is associated with increased tau phosphorylation in brains of transgenic mice. J Neurosci 2004;24:2527Y34


Altered microRNA profiles in cerebrospinal fluid exosome in Parkinson disease and Alzheimer disease.

Preanalytical variables and Alzheimer disease biomarker concentrations in cerebrospinal fluid.

Imaging and cerebrospinal fluid biomarkers in early preclinical alzheimer disease.

Cerebrospinal Fluid Biomarker for Alzheimer Disease Predicts Postoperative Cognitive Dysfunction.

Retromer in Alzheimer disease, Parkinson disease and other neurological disorders.

Associations between cerebral small-vessel disease and Alzheimer disease pathology as measured by cerebrospinal fluid biomarkers.

Inosine to increase serum and cerebrospinal fluid urate in Parkinson disease: a randomized clinical trial.

Importance and impact of preanalytical variables on Alzheimer disease biomarker concentrations in cerebrospinal fluid.

Cerebrospinal Fluid Markers of Neurodegeneration and Rates of Brain Atrophy in Early Alzheimer Disease.

Cerebrospinal fluid β-amyloid and phospho-tau biomarker interactions affecting brain structure in preclinical Alzheimer disease.

Cerebrospinal fluid biomarkers for Alzheimer disease and subcortical axonal damage in 5,542 clinical samples.

Cerebrospinal fluid α-synuclein predicts cognitive decline in Parkinson disease progression in the DATATOP cohort.

Cerebrospinal fluid peptides as potential Parkinson disease biomarkers: a staged pipeline for discovery and validation.

Apolipoprotein E genotype and the diagnostic accuracy of cerebrospinal fluid biomarkers for Alzheimer disease.

Effect of Cognitive Reserve on Age-Related Changes in Cerebrospinal Fluid Biomarkers of Alzheimer Disease.

Decreased levels of soluble amyloid beta-protein precursor in cerebrospinal fluid of live Alzheimer disease patients.

Update on the core and developing cerebrospinal fluid biomarkers for Alzheimer disease.

Neurogranin as a Cerebrospinal Fluid Biomarker for Synaptic Loss in Symptomatic Alzheimer Disease.

Association of cerebrospinal fluid prion protein levels and the distinction between Alzheimer disease and Creutzfeldt-Jakob disease.

Cerebrospinal fluid taurine in Alzheimer's disease.

Overlap between Parkinson disease and Alzheimer disease in ABCA7 functional variants.

Testosterone as the missing link between pesticides, Alzheimer disease, and Parkinson disease.

Cerebrospinal fluid biomarkers of Alzheimer's disease.

Testosterone as the missing link between pesticides, Alzheimer disease, and Parkinson disease--reply.