UBC Faculty Research and Publications

High-density lipoproteins suppress Aβ-induced PBMC adhesion to human endothelial cells in bioengineered… Robert, Jérôme; Button, Emily B; Stukas, Sophie; Boyce, Guilaine K; Gibbs, Ebrima; Cowan, Catherine M; Gilmour, Megan; Cheng, Wai H; Soo, Sonja K; Yuen, Brian; Bahrabadi, Arvin; Kang, Kevin; Kulic, Iva; Francis, Gordon; Cashman, Neil; Wellington, Cheryl L Aug 22, 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13024_2017_Article_201.pdf [ 1.9MB ]
JSON: 52383-1.0361973.json
JSON-LD: 52383-1.0361973-ld.json
RDF/XML (Pretty): 52383-1.0361973-rdf.xml
RDF/JSON: 52383-1.0361973-rdf.json
Turtle: 52383-1.0361973-turtle.txt
N-Triples: 52383-1.0361973-rdf-ntriples.txt
Original Record: 52383-1.0361973-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessHigh-density lipoproteins suppress Aβ-induced PBMC adhesion to humanendothelial cells in bioengineered vesselsand in monocultureJérôme Robert1,2† , Emily B. Button1,2†, Sophie Stukas1,2, Guilaine K. Boyce1,2, Ebrima Gibbs2,3,Catherine M. Cowan4, Megan Gilmour1,2, Wai Hang Cheng1,2, Sonja K. Soo1,2, Brian Yuen1,2, Arvin Bahrabadi1,2,Kevin Kang1,2, Iva Kulic1,2, Gordon Francis5, Neil Cashman2,3 and Cheryl L. Wellington1,2*AbstractBackground: Alzheimer’s Disease (AD), characterized by accumulation of beta-amyloid (Aβ) plaques in the brain,can be caused by age-related failures to clear Aβ from the brain through pathways that involve thecerebrovasculature. Vascular risk factors are known to increase AD risk, but less is known about potential protectivefactors. We hypothesize that high-density lipoproteins (HDL) may protect against AD, as HDL have vasoprotectiveproperties that are well described for peripheral vessels. Epidemiological studies suggest that HDL is associated withreduced AD risk, and animal model studies support a beneficial role for HDL in selectively reducing cerebrovascularamyloid deposition and neuroinflammation. However, the mechanism by which HDL may protect thecerebrovascular endothelium in the context of AD is not understood.Methods: We used peripheral blood mononuclear cell adhesion assays in both a highly novel three dimensional(3D) biomimetic model of the human vasculature composed of primary human endothelial cells (EC) and smoothmuscle cells cultured under flow conditions, as well as in monolayer cultures of ECs, to study how HDL protects ECsfrom the detrimental effects of Aβ.Results: Following Aβ addition to the abluminal (brain) side of the vessel, we demonstrate that HDL circulatedwithin the lumen attenuates monocyte adhesion to ECs in this biofidelic vascular model. The mechanism by whichHDL suppresses Aβ-mediated monocyte adhesion to ECs was investigated using monotypic EC cultures. We showthat HDL reduces Aβ-induced PBMC adhesion to ECs independent of nitric oxide (NO) production, miR-233 andchanges in adhesion molecule expression. Rather, HDL acts through scavenger receptor (SR)-BI to block Aβ uptakeinto ECs and, in cell-free assays, can maintain Aβ in a soluble state. We confirm the role of SR-BI in ourbioengineered human vessel.Conclusion: Our results define a novel activity of HDL that suppresses Aβ-mediated monocyte adhesion to thecerebrovascular endothelium.Keywords: Alzheimer’s disease, Beta-amyloid, High-density lipoprotein, Engineered vessel, Endothelial cells* Correspondence: wcheryl@mail.ubc.ca†Equal contributors1Department of Pathology and Laboratory Medicine, University of BritishColumbia, Vancouver, BC V6T 1Z3, Canada2Djavad Mowafaghian Centre for Brain Health, University of British Columbia,2215 Wesbrook Mall, Vancouver, BC V6T 1Z3, CanadaFull list of author information is available at the end of the article© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Robert et al. Molecular Neurodegeneration  (2017) 12:60 DOI 10.1186/s13024-017-0201-0BackgroundAlzheimer’s disease (AD) is the leading cause of demen-tia with over 44 million affected persons worldwide andhas no effective treatment [1]. AD is believed to arisefrom the toxic effects of amyloidogenic beta-amyloidpeptides (Aβ), which form higher order structures in-cluding oligomers and fibrils that accumulate as amyloidplaques in the brain parenchyma and cerebral vessels[2]. The role of the cerebrovasculature in AD has gainedincreased recognition in the last decade, as AD risk is in-creased by vascular risk factors, including hypertension,type 2 diabetes mellitus (T2DM), and hypercholesterol-emia [3–5]. Most AD autopsy cases have evidence ofcerebral small vessel disease, including cerebral amyloidangiopathy (CAA) and microvascular degeneration, withmicro- and macro-infarcts reported in over 80% of veryold AD subjects [6]. The cerebrovasculature plays a piv-otal role in clearing Aβ from the brain [7], and activationof cerebrovascular endothelial cells (ECs) has been re-ported in AD brain [8].High-density lipoproteins (HDL) are a heterogeneousmixture of circulating lipoprotein particles that are com-posed of, in addition to apoA-I, cholesterol and phos-pholipids, over 200 lipid species and at least 95 proteinsin normolipidemic plasma [9, 10]. Intriguingly, the ele-vated concentration of plasma high-density lipoproteincholesterol (HDL-C), which is associated with reducedcardiovascular disease risk [11], is also negatively corre-lated with cognitive decline [12]. Preclinical studies alsosupport a beneficial role of HDL on the cerebrovascula-ture. Specifically, transgenic overexpression of humanapoA-I, the major protein component on HDL, from itsendogenous promoter in the liver and intestine, reducesneuroinflammation, improves cognitive function andselectively reduces cerebrovascular amyloid in the APP/PS1 model of amyloidosis, whereas deletion of apoA-I,the major protein component of HDL, has the oppositeeffects [13, 14]. We recently reported that a single intra-venous injection of reconstituted HDL reduces solublebrain Aβ40 and Aβ42 levels after 24 h in symptomaticAPP/PS1 mice [15].Systemically circulating HDL may therefore benefit thebrain, however, the mechanisms by which it may do so areunknown. HDL is well established to possess severalpotent vasoprotective functions in peripheral vascularendothelial cells (ECs) including reducing inflammation,increasing vascular tone through promoting endothelialnitric oxide (NO) synthase activity, and suppressingvascular adhesion molecule expression [16]. Using periph-eral blood mononuclear cell (PBMC) adherence assays,here we show that Aβ induces PMBC adhesion, a classicalmeasure of EC activation, in cerebral microvascular ECsthrough a mechanism distinct from classical inflammatorystimulation by tumor necrosis factor-alpha (TNF-α). Wealso demonstrate that HDL robustly suppresses Aβ-induced monocyte adhesion to ECs both in monolayercultures as well as in a novel 3 dimensional (3D) tissueengineered human vessel.MethodsFabrication of tissue engineered vesselsScaffold-directed human engineered vessels were gener-ated under flow bioreactor conditions as described [17].Briefly, non-woven polyglycolic-acid meshes (PGA;BMS) were coated with polycoprolactone (PCL, Purac)and polyglycolic acid (PLA, Purac) by dipping into a1.75% (w/w/w) solution of PCL/PLA/tetrahydrofuran(Sigma Aldrich). A tubular shape (length 10 mm andinner diameter 2 mm) was obtained by heat welding be-fore external coating with a 10% (w/w) PCL/tetrahydro-furan solution. After sterilization with 70% ethanol for30 min followed by 3 washes with PBS, scaffolds werepre-incubated in advanced DMEM overnight before cellseeding. Human umbilical cord smooth muscle cellswere isolated as described [17] and 2 × 106 cells/cm2were seeded in the inner surface of the vascular scaffoldusing fibrinogen (10 mg/mL clottable protein, Sigma Al-drich) and thrombin (10 mU) as cell carriers [18]. Aftera short static incubation period of 3 days, vascular con-structs were exposed to dynamic conditioning in a flowbioreactor, where the flow of nutrient medium (Ad-vanced DMEM (Invitrogen) with 10% FBS; 0.05%Penicillin/Streptomycin, 1% L-glutamine and 1.5 mM L-ascorbic acid) was directed through the inner lumen ofthe bioreactor circulation loop at 6 mL/min. After14 days of flow conditioning, vascular grafts wereendothelialzed with HUVECs (1.5 × 106 cells/cm2)seeded into the lumen followed by cultivation understatic conditions for 5 days in complete endothelialgrowth medium (EBM-2; Lonza) supplemented with10% foetal bovine serum (FBS) and growth supplementsto form EGM complete media FBS. After the staticphase, vascular grafts were placed back in the bioreactorfor 14 additional days with increasing medium flow (4 to6 mL/min) in complete EGM-2 with 10% FBS.Characterization of engineered vascular tissueFor histological characterization, engineered vessels werefixed in formalin (Fisher), dehydrated through a gradedethanol series using a Sakura Tissue Processor (Sakura),embedded in paraffin and sectioned at 7 μm thickness.Sections were deparaffinised, rehydrated through agraded ethanol series and stained using haematoxylin &eosin (SigmaAldrich) following the manufacture’s in-structions. For immunofluorescence analyses, tissueswere cryopreserved in Cryomatrix (ThermoFisher) andsectioned at 20 μm thickness using a cryotome (Leica).Immunostaining was performed as described [17] usingRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 2 of 19antibodies specific for the EC marker CD31 (clone JC/70A Biolegend, 1:50) and α-smooth muscle actin (clone1A4 SigmaAldrich, 1:200). Sections were counter-stainedwith DAPI and imaged with an inverted microscope(Carl Zeiss). Endothelium barrier integrity was analyzedby injecting Evans blue (Sigma Aldrich) at a final con-centration of 0.5% in the circulation loop of the bioreac-tor for 10 min followed by continuous PBS washing for20 min. Vessels were cut open longitudinally and en facepreparations were analysed macroscopically with photodocumentation.Preparation of HDL and PBMCsAll experiments were conducted under an approvedclinical protocol (UBC Clinical Ethics Research BoardH14–03357). Upon receipt of written informed consent,100 mL of fasted blood was collected from normolipide-mic healthy donors into vacutainer tubes. Plasma HDL(1.063–1.21 g/mL) was isolated by sequential potassiumbromide gradient ultracentrifugation as described [19].The purity of the HDL preparations was verified by so-dium dodecyl sulphate-polyacrylamide gel electrophor-esis (SDS-PAGE) followed by Coomassie blue staining toensure no low-density lipoprotein (LDL) or albumincontamination. Eight independent donors were usedacross the experiments, 6 isolated in-house and 2 com-mercially obtained (Leebioscience). Human-derived, lipidfree apoA-I was a kind gift from CSL-Behring. Immor-talized human THP1 monocytes (ATCC) were culturedin RPMI containing 10% FBS, 1% Pen/Strep, 2 mM L-glutamine and 0.1% β-mercaptoethanol. Primary humanPBMC were isolated from healthy donors by centrifuga-tion on a continuous density gradient (Lymphoprep™,Stemcell) following the manufacturer’s instructions.Freshly isolated PBMC were fluorescently labeled with10 μM of Cell-Tracker Red for 30 min (Invitrogen)following the manufacturer’s recommendations.Monocyte adhesion in engineered vesselsVascular grafts were perfused with complete EGM-2with 2% FBS. 1 μM Aβ42 or Aβ40 monomers wereinjected directly into the graft chamber to mimic Aβoriginating from the brain (antelumen) side of the vessel.At time points ranging from 2 to 72 h, THP1 cells werefluorescently labeled with Cell-Tracker Red as describedabove, injected in the graft circulation at a concentrationof 1 × 106 cells/mL and maintained under flow condi-tions for 3 h. For HDL experiments, vascular grafts wereperfused with luminal HDL (200 μg/mL) for 2 h beforeinjecting Aβ in the antelumen side for 8 h. Tissues werelongitudinally cut open, washed extensively with PBSand fixed with 4% PFA. After 20 min, tissues werewashed 3 times with PBS and mounted in Prolong Goldantifade reagent with DAPI. For each independentlyseeded tissue, adherent monocytes and monocytesundergoing diapedesis were counted in 3 randomsquares of 1.23 mm2 using a z-stack covering the wholetissue thickness with a SP8 confocal microscope (Leica),averaged and expressed as percent of vehicle normalizedto 100% for Fig. 1d and f or percent of Aβ normalized to100% for Fig. 1 h and j, and Fig. 8a and b.Static monotypic cell culturehCMEC/D3 (Fisher; passage 27–35), and HUVEC (pas-sage 4–7, isolated as described [17]) cells were culturedusing complete EGM-2 with 2% FBS, ECs were culturedin a humidified incubator at 37 °C at 5% carbon dioxide.For mechanistic experiments, ECs were treated with SR-BI blocking antibody (NB440–113 Novus, 1:500), theSR-BI inhibitor block lipid transport-1 (BLT-1, SigmaAl-drich, 10 μM), the eNOS inhibitor L–NG-nitroarginnemethyl ester (L-NAME, SigmaAldrich, 1 mM), the re-ceptor associated protein (RAP, Oxford biome, 1 μM),the RAGE blocking antibody (176,902, R&D Systems,1:50), heparin (10 mU), heparinase (SigmaAldrich,0.2 mM) or the CD36 blocking antibody (JC63.1,ABCAM, 1:500) for 1 h before HDL priming. FormiRNA experiments, cells were transiently transfectedusing Lipofectamine 2000 (Life Technology) 2 h beforeAβ stimulation with the miR-223 mimetic (Life Technol-ogy, 4,464,066 (MC12301), 100 nM) or miR-223 inhibi-tor (Life Technology, 4,464,084 (MH12301), 100 nM) inEBM-2 containing 0.2% BSA.Monolayer PBMC adhesion assayECs were seeded at 1 × 105 cells/well in 24-well plates andcultured until confluent for 2 to 3 days. On the day of theassay, ECs were primed for 2 h with 100 μg/mL bovineserum albumin as vehicle control or 100 μg/mL HDL be-fore stimulation with 1 ng/mL of TNF-α (Preprotech) orvarious concentrations of monomeric Aβ40 and Aβ42(0.001–1 μM, California Peptide), prepared as describedbelow. After 3 h, 5 × 105 Cell-Tracker Red-labeled PBMCper well were added to ECs for 3 additional hours beforewashing 3 times with PBS to remove non-adherentPBMC. Cells were then fixed with 4% PFA for 15 min be-fore 3 additional PBS washes and 4′,6-diamidino-2-pheny-lindole (DAPI) counterstaining. For each independentexperiment, adherent monocytes were counted in 5 ran-dom squares of 7.84 mm2 using a fluorescent invertedmicroscope (Zeiss), averaged and expressed as percent ofvehicle normalized to 100%. DAPI counterstaining wasused to ensure endothelial cell coverage.Aβ oligomerization/fibrilization and electron microscopyconfirmationRecombinant Aβ40 and Aβ42 peptides (California Peptide)were dissolved in ice-cold hexafluoroisopropanol (HFIP).Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 3 of 19abf ghj k l mic d eFig. 1 (See legend on next page.)Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 4 of 19The HFIP was removed by evaporation overnight. To pre-pare soluble monomers, the peptide film was reconstitutedin DMSO to 5 mM, diluted further to 100 μM in DMEMand used immediately. Oligomers were prepared by dilutingthe 5 mM DMSO peptide solution in phenol red-free F12medium (Life Technologies) to a final concentration of100 μM and incubating for 24 h 4 °C. Fibrils were preparedby diluting the 5 mM peptide solution in 0.1 uM of HCl toa final concentration of 100 μM and incubating for 24 h at37 °C. Aβ monomer, oligomer and fibril preparations werethen either used to stimulate hCMEC/D3 monolayers, asgiven, or were analysed by Transmission Electron Micros-copy (TEM) and dot blot. For TEM, 0.5 μL of 100 μM Aβpreparation was diluted in 2 μL filtered distilled water, spot-ted onto formvar-coated 200-mesh nickel grids (EM Sci-ences) and allowed to dry. Grids were then negativelystained with 0.5% aqueous uranyl acetate for 30 s, andviewed on a FEI Tecnai G2 Spirit Transmission ElectronMicroscope. For dot blot, aliquots of Aβ(0.1 μM) wereadded to PVDF membrane, which were dried and blockedin 3% skimmed milk PBST. After 1 h, blots were incubatedwith β-amyloid 1–16 antibody (6E10 Biolegend 1:500),amyloid A11 oligomeric antibody (AB9234 Millipore,1:1000) or amyloid fibrils OC antibody (AB2286 Millipore,1:1000) in blocking buffer for 16 h, washed extensively inPBST and incubated with anti-mouse or anti-rabbit(1:1000) secondary antibody in blocking buffer. After 1 h,blots were washed as above and developed using enhancedECL and a ChemiDoc MP imager.Measurement of intracellular NOECs were seeded at 5 × 105 cells/well in 6-well platesand cultured for 2 days until confluent in completeEBM-2. Sixteen hours before the assay, ECs were serum-depleted in EBM-2 containing 0.2% FBS. On the day ofthe assay, ECs were incubated with 100 μg/mL of HDLin serum-depleted EBM-2 medium containing 1 μM of4,5-diaminofluorescein diacetate (DAF2, Caymanchem)at 37 °C. After 6 h, ECs were washed with PBS, trypsi-nized, and triazolofluorescein fluorescence was mea-sured (excitation wavelength of 485 nm, emission538 nm), using an Infinite M200Pro plate reader(Tecan). In addition to DAF2 measurement, the phos-phorylation of eNOS at Ser1177 was compared to totaleNOS by immunoblotting (below) in cell lysates har-vested 15 min after HDL treatment.Cell surface biotinylationECs were seeded at 5 × 105 cells/well in 6-well platesand cultured for 2 days until confluent. On the day ofthe assay, ECs were treated with HDL, Aβ or TNF-α asabove. After stimulation, EC monolayers were washedtwice with ice cold PBS (pH 8), cooled on ice for 15 minand biotinylated with 250 μg/mL EZ-link™-sulfo-NHS-biotin (ThermoFisher Scientific) in PBS (pH 8) at 4 °C.After 1 h the reaction was stopped with a 5 min incuba-tion in DMEM with 10% FBS. After two additional PBSwashes, ECs were lysed in RIPA buffer. Following quan-tification of protein concentration using a BCA assay(Thermofisher Scientific), at least 100 μg of protein wereincubated with streptavidin-conjugated sepharose beads(Pierce) at 4 °C overnight. Beads were washed 3 timeswith RIPA buffer and the recovered proteins wereresolved on a SDS-PAGE.Monolayer aβ association, binding and uptakeECs were seeded at 3 × 105 cells/well in 12-well platesand cultured until confluent for 2 to 3 days. On the dayof the assay, ECs were primed for 2 h with 100 μg/mLHDL before stimulating with 0.1 μM of Aβ40 and Aβ42monomers at 37 °C for total association, or at 4 °C forcell surface binding. After 3 h, hCMEC/D3 were washed3 times with PBC and lysed in RIPA buffer containing10 mM Tris pH 7.4, 150 mM NaCl, 1.0% NP-40,1.0````% sodium deoxycholate, 0.1% SDS and cOmpleteprotease inhibitor with EDTA (Roche). Aβ40 (KHB3442,Life Tech) and Aβ42 (KHB3482, Life Tech) were quanti-fied using commercial ELISAs and normalized to totalprotein concentration. For Aβ uptake, hCMEC/D3 wereseeded at 1 × 105 cells/well in 24-well plates and cul-tured to confluence for 2 to 3 days. On the day of theassay endothelial, ECs were primed for 2 h with 1 mg/mLof HDL before stimulating with 1 μM monomeric FITC-Aβ40 and FITC-Aβ42 (Bachem) prepared as describedabove. After 3 h at 37 °C, hCMEC/D3 were washed 3times with PBS and fixed in 4% PFA for 20 min.After one Tris-HCl and two PBS washes, hCMEC/D3were mounted in Prolong antifade reagent.(See figure on previous page.)Fig. 1 Aβ induces monocyte adhesion in engineered vessels, which is suppressed by HDL. a Schematic representation of bioengineered tissue. bHistological structure of engineered tissue using hematoxalin-eosin staining to reveal a dense tissue formation composed of cells and extracellularmatrix in engineered vessels. c α-smooth muscle actin (α-SMA) confirmed the smooth muscle phenotype of cells in the inner layers and d CD31confirmed an endothelial monolayer. Scale bar represents 200 μm. e Evans blue staining confirmed a tight endothelium. f-g 1 μM of Aβ40 or h-iAβ42 monomers were injected within the tissue chamber (abluminal). Fluorescently labeled human monocytes (THP1, white) were circulated inthe lumen of the engineered vessels. Tissues were counterstained with DAPI (blue) and analyzed using confocal microscopy over time. j-m200 μg/mL of HDL were circulated through the lumen of the grafts for 2 h prior injection of 1 μM of j-k Aβ40 or l-m Aβ42 monomers within thetissue chamber (abluminal) for 8 h prior to circulating fluorescent THP1 in the lumen. Graphs represent means ± SD of adhered monocytesrelative to Aβ treated tissues from at least 3 individual grafts. *p < 0.05, **p < 0.01Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 5 of 19Molecular biologyFor mRNA, cells were lysed in Trizol (Invitrogen) andRNA was extracted and treated with DNase I (Invitrogen)according to the manufacturer’s protocol. cDNA wasgenerated using oligo-dT primers and Taqman reversetranscription reagents (Applied Biosystems). Real-timequantitative PCR was done using FastStart Universal SYBRGreen Master reagent (Roche) on a Light Cycler 96 system(Roche) to quantify gene expression relative to vehicle usingspecific primer against ICAM-1 (fwd: ATGGCAACGACTCCTTCTCG; rev: CGCCGGAAAGCTGTAGATGG)and VCAM-1 (fwd: TGTTTGCAGCTTCTCAAGCTTTT;rev: GATGTGGTCCCCTCATTCGT) and normalizedto GAPDH (fwd: CCTGCACCACCAACTGCTTA; rev:CATGAGTCCTTCCACGATACCA). MiRNA was isolatedusing an miRNeasy mini kit (Qiagen) following the manu-facturer’s instructions, respectively. Reverse transcriptionwas performed using 2 μg of total RNA with specificmiRNA primers (Life Technology). MiR223 and U6 werequantified using specific TaqMan probes (Life Technology).ImmunoblotECs were lysed in RIPA buffer containing eithercOmplete protease inhibitor or Phosphostop (Roche)and quantified using a BCA assay. Equal amounts oftotal protein were separated by SDS-PAGE followed byelectrophoretic transfer to polyvinylidene fluoride(PVDF) membranes (Millipore). After blocking mem-branes for 1 h with 5% skim milk powder in PBST, or5% BSA in TBST for phosphoproteins, ICAM-1,(EP1442Y Abcam, 1:1000), VCAM-1 (EPR5047 Novus,1:1000), peNOS (ser1177 Cellsignalling, 1:1000), eNOS(M221 ABCAM, 1:1000), AnxA1 (D5V2T, Cellsignalling,1:1000), p-Akt (Ser473 D9E, Cellsignalling, 1:2000), Akt(9272, Cellsignalling, 1:1000), p-NF-κB p65 (Ser53693H1, Cellsignalling, 1:1000), NF-κB p65 (D14E12, Cell-signalling, 1:1000), p-SAPK/JNK (Thr183/Tyr185 G9,Cellsignalling, 1:1000), JNK (2C6, Cellsignalling, 1:1000),p-p42/44 MAPK (Erk1/2) (Thr202/Tyr204 D13.14.4E,Cellsignalling, 1:1000), p42/44 MAPK (Erk1/2) (137F5,Cellsignalling, 1:1000), p-Stat3 (Tyr705, Cellsignalling1:1000), Stat3 (124H6, Cellsignalling, 1:1000) andGAPDH (MAB374 1:10,000, Millipore) were immunode-tected by incubating for 16 h in primary antibody inblocking buffers. Membranes were washed extensivelywith PBST or TBST, and incubated with anti-mouse oranti-rabbit (1:1000–10′000, Jackson ImmunoResearch)secondary antibody in blocking buffer. After 1 h, mem-branes were washed as above and developed usingenhanced chemiluminescence (ECL, Amersham) and aChemiDoc MP imager (Biorad). Densitometric imageswere captured with ImageJ and band intensity normalizedto GAPDH as a loading control.Beta-sheet formation assayCell-free Thioflavin-T fibrillization assays were per-formed on an Infinite M2000 Pro plate reader (Tecan)as described (Truran 2015). Briefly, 10 μM monomericAβ40 or Aβ42 were incubated in a buffer consisting of20 mM of Thioflavin-T in 150 mM NaCl and 5 μM ofHEPES at pH 7.4, with and without 1 mg/mL of HDL, at37 °C with 20 s of orbital shaking (3 mm amplitude)every 5 min in a black 96-well plate. Formation of fibril-lar β-amyloid pleated sheets over time was monitored byexcitation at 440 nm and measuring emission intensityat 490 nm every 5 min up to 12 h in total.Human brain protein extraction and ELISAFrozen brain tissues (cortex Brodmann area 9 and cerebel-lum) were provided by the Harvard Brain Tissue ResourceCenter under an approved UBC protocol (C04–0595) andextracted with 8 volumes of ice-cold carbonate buffer(100 mM Na2CO3, 50 mM NaCl, pH 11.5) containingcOmplete protease inhibitor (Roche Applied Science) bymanual homogenization with a tissue probe. After incu-bating on ice for 10 min, lysates were clarified by centrifu-gation at 16,600 rcf for 45 min at 4 °C. The supernatantwas removed and neutralized by adding 1.5-volumes of1 M Tris-HCl pH 6.8 to give a final pH of approximately7.4. Brain tissues from all samples were extracted in anidentical manner, and fractions were aliquoted and imme-diately frozen at −80 °C until analysis. Protein concentra-tions were determined using the Lowry Protein Assay(Biorad). ICAM-1 (ab174445, ABCAM), VCAM-1(ab187393, ABCAM), Aβ40 (KHB3442, Life Tech) andAβ42 (KHB3482, Life Tech) in carbonate extracts werequantified using commercial ELISAs. ELISA data werenormalized to total protein concentration of the extract.For immunofluorescent staining, brains were sectioned at20 μm, rehydrated in PBS and blocked in 4% paraformal-dehyde (PFA). After 20 min, sections were washed oncewith Tris-HCl (0.5 mM, pH 8), twice with PBS, andblocked in blocking buffer (5% goat serum and 1% BSA).After 60 min, section were incubated at 4 °C with anti-bodies against CD31 (WM59 Biolegend, 1:50), ICAM-1(EP1442Y ABCAM 1:200), and VCAM-1 (EPR5047ABCAM 1:200) in blocking buffer overnight. Sectionswere washed 3 times in PBS and incubated at RT withAlexa-488 anti-mouse and Alexa-594 anti-rabbit fluores-cently labeled secondary antibodies (LifeTechnologies,1:600). After 45 min and 3 additional PBS washes, sectionswere mounted in Prolong Gold antifade reagent withDAPI (LifeTechnologies) and imaged on an invertedfluorescent microscope (Zeiss)Statistical analysisAll statistical analyses were performed using SPSS andp-values <0.05 were considered significant. Data wereRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 6 of 19obtained from at least 3 independent experiments andare presented as mean ± SD if not indicated otherwise.Data were first log transformed and analyzed by two-way ANOVA with a blocking factor (experiment) withdirect comparison when comparing two treatments orDunnett’s or Bonferonni multi-comparison tests. Afterstatistical calculations, vehicle data were normalized to100% and represented as a dashed line in the graph, andtested conditions were expressed and graphed as per-centage of vehicle if not otherwise stated.ResultsHDL suppresses aβ-mediated monocyte adhesion to ECsin three-dimensional dynamic engineered human vesselsTo mimic the complexity of native cell-cell and/or cell-matrix interactions observed in native human vessels, weused innovative tissue engineering technology to generatein vitro 3D human vessels composed of primary humanumbilical vein ECs (HUVEC) and primary human smoothmuscle cells (SMC) to maximize translational relevance ofour studies relative to in vitro studies that solely use trad-itional static cell culture models. Our system uses ascaffold-directed dynamic, semi-pulsatile flow bioreactorsystem, on which primary human cells are sequentiallyseeded into 2 mm diameter biodegradable polyglycolic-acid (PGA)/polycaprolactone (PCL) composite matrices.After a short static incubation, vascular constructs areexposed to dynamic flow in a bioreactor, where nutrientmedium is directed through the lumen of the bioreactorcirculation loop to mimic native blood flow. We previ-ously demonstrated that these scaffold-directed engi-neered vessels are useful for studies of endothelialactivation in 3D culture [17]. A schematic of the bioreac-tor system is presented in Fig. 1a. Under our flow bioreac-tor conditions, haematoxylin and eosin staining confirmeda dense tissue formation composed of cells with extracel-lular matrix on the luminal side of the scaffold (Fig. 1b-d).Immunofluorescent staining confirmed a SMC phenotypeof cells in the inner layers and an EC monolayer on theluminal side (Fig. 1c,d). Integrity of the endothelial barrierwas functionally assessed by injecting Evans blue dye intothe bioreactor circulation loop, which was excluded afterEC seeding, demonstrating a functionally tight endothelialbarrier (Fig. 1e).Having established a 3D human vascular model, weinjected 1 μM Aβ40 or Aβ42 directly into the graft cham-ber to mimic Aβ coming from the brain anteluminal side,followed by injection of human THP-1 monocytes intothe lumen chamber to mimic circulating monocytes. BothAβ40 and Aβ42 led to THP1 adhesion to endothelium inthe engineered vessels, with the most robust responseappearing 8 h after Aβ injection (Fig. 1f-h). We then testedthe ability of HDL, isolated from healthy human donorsusing KBr density gradient ultracentrifugation, to suppressAβ-induced monocyte adhesion to ECs in engineered ves-sels by circulating either media alone or 200 μg/mL ofHDL for 2 h prior to injecting 1 μM Aβ into the graftchamber. In this engineered vascular model, HDL robustlysuppressed monocyte adhesion to endothelium by bothAβ40 (5 fold, p = 0.0286) and Aβ42 (10 fold, p = 0.0016)(Fig. 1i-l).HDL attenuates aβ-induced peripheral blood mono-nuclear cell (PBMC) adhesion in monotypic ECsTo define the mechanisms by which HDL attenuatesAβ-induced monocyte binding to ECs, we used staticmonolayer EC cultures and peripheral blood mono-nuclear cells (PBMC) isolated by continuous density gra-dient ultracentrifugation from healthy human donors.We first confirmed that Aβ induces PBMC adhesion inmonotypic HUVEC cultures and that HDL attenuatesthis activity (Fig. 2a,b). As Aβ originates within the brainand the primary pathways by which Aβ is cleared from thebrain involve cerebral vessels [7], we also showed that thehuman brain microvascular endothelial cell line hCMEC/D3, a commonly used cell line for in vitro studies of theblood brain barrier (BBB) [20], also exhibit increased PBMCadhesion upon Aβ treatment and that this too is attenuatedby HDL (Fig. 2c,d). As HUVEC and hCMEC/D3 give nearlyidentical results, we focused on hCMEC/D3 cells for allsubsequent mechanistic experiments.Our first question was to determine if lipid-free apoA-Iwas functionally equivalent to mature HDL in its ability tosuppress Aβ-mediated PBMC binding to hCMEC/D3. Ourresults clearly showed that pre-incubation of hCMEC/D3with lipid-free apoA-I (50 and 100 μg/mL) did not alterAβ-induced PBMC adhesion (Fig. 2e,f), demonstrating thatmature HDL particles are required to attenuate Aβ-induced PBMC adhesion to brain-derived ECs.To establish the experimental conditions to investigatehow HDL suppresses Aβ-mediated PBMC adhesion tohCMEC/D3, we performed PMBC adhesion assays inhCMEC/D3 after a 3 h exposure to increasing concen-trations human Aβ40 and Aβ42 monomers. Comparedto baseline, Aβ40 concentrations of 0.01, 0.1 and 1.0 μMled to significant 175%, 169% and 179% increases inPMBC adhesion (p = 0.033, 0.002 and 0.019 respectively,Additional file 1a), The same concentrations of Aβ42also led to significant 184%, 161% and 223% increases inPMBC adhesion (p = 0.019, 0.0406 and 0.0421, respect-ively) relative to baseline (Additional file 1b). Havingdemonstrated that 0.1 μM of monomeric Aβ40 or Aβ42is sufficient to induce PMBC adhesion in hCMEC/D3,we then tested whether pre-treating hCMEC/D3 for 2 hwith 0 to 400 μg/mL HDL could attenuate Aβ-inducedPBMC adhesion. Significant suppression of PBMCadhesion was observed at concentrations of HDL from100 μg/mL and above. At 100 μg/mL HDL there was anRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 7 of 1980% and 60% reduction of adhered PBMCs in cellsinduced by 0.1 μM Aβ40 (p = 0.017) or 0.1 μM Aβ42(p = 0.018) respectively (Additional file 1c,d).HDL reduces the fibrillization rate of aβ42 and aβ40 incell-free conditionsAs Aβ has been reported to bind HDL-like particles inboth cerebrospinal fluid (CSF) and blood [21], and apoA-Ihas been reported to affect Aβ aggregation [13, 22], wereasoned that one mechanism by which HDL may attenu-ate PBMC adhesion to hCMEC/D3 could be by affectingAβ structure to prevent the formation of toxic higherorder species [23]. A cell-free Thioflavin-T reporter assaywas used to determine whether HDL affects the fibrilliza-tion kinetics of Aβ40 or Aβ42 over a 12 h period, using10 μM Aβ and 10 mg/mL HDL (ratio: 1:1), which is thesame Aβ:HDL ratio used for our cellular PBMC adhesionassays and allows detection of fluorescence signal using aplate reader (Fig. 2c,d). Under these conditions, in theabsence of HDL, the onset of Aβ fibrillization requires atleast 3 h for Aβ42 and at least 10 h for Aβ40. Notably, inthe presence of HDL, we observed delayed onset of fibrilli-zation for both Aβ42 and Aβ40, and a slower rate of Aβ42fibrillization (Fig. 3a). As HDL can suppress PBMCa bc de fFig. 2 Aβ40 and Aβ42 induce PBMC adhesion to ECs, which is suppressed by HDL. HUVEC (a-b) or hCMEC/D3 (c-d) were primed with 100 μg/mL of HDL and stimulated with 0.1 μM a, c Aβ40 (light grey) or b, dAβ42 (dark grey) monomers for 3 h. Fluorescently labelled PBMC wereallowed to adhere to stimulated cells for 3 h followed by washing, fixation, imaging, and counting. hCMEC/D3 were primed with either 100 μg/mL of HDL or 50 μg/mL or 100 μg/mL lipid-free human apoA-I for 2 h followed by stimulation with 0.1 μM e Aβ40 (light grey) or f Aβ42 (darkgrey) monomers for 3 h. Cells were washed, imaged and counted as above. Graphs represent mean ± SD of adhered PBMC relative to vehiclecontrol from at least 3 independent trials where * p < 0.05, **p < 0.01, ***p < 0.001 versus vehicle, § p < 0.05, §§ p < 0.01 versus AβRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 8 of 19adhesion to hCMEC/D3 induced by both Aβ40 and Aβ42within 3 h of Aβ addition, we conclude that suppressionof Aβ fibrillization cannot be the only mechanism bywhich HDL functions to protect ECs from the detrimentaleffects of Aβ.Aβ structure does not affect HDL’s ability to reduce PBMCadhesion to hCMEC/D3Several experiments were then performed to further testwhether Aβ structure affects HDL’s anti-inflammatoryactivity on hCMEC/D3. We varied the structural speciesabdfcegFig. 3 HDL delays beta-sheet formation and attenuates Aβ-induced PBMC adherence independent of Aβ structure. a Representative graph ± SDof technical triplicates of 3 individual experiments where 10 μM Aβ40 or Aβ42 with or without 10 mg/mLl HDL were incubated with 20 μM ofThio-T in 150 mM NaCl and 5 μM of HEPES (pH 7.4) for 12 h at 37 °C. Formation of β-amyloid pleated sheets was monitored every 5 min at exci-tation 440 nm and emission 490 nm. b-c Aβ structures were confirmed using dot blot with antibodies against oligomers (A11) or fibrils (OC) andelectron microscopy (EM). d-e hCMEC/D3 were stimulated for 3 h with 0.1 μM monomeric (m-Aβ, solid bar), oligomeric (o-Aβ, striped bar) orfibrillar-aggregated (f-A Aβ, cross-hatched bar) d Aβ40 or e Aβ42 in the presence or absence of HDL. f-g hCMEC/D3 were pre-incubated for 2 hwith 100 μg/mL HDL prior to Aβ addition (pre, solid bar), co-incubated with 100 μg/mL HDL and Aβ (co, striped bar), or post-incubated by adding100 μg/mL HDL 1 h following Aβ stimulation (post, double striped bar). Graphs represent means ± SD of adhered PBMC relative to vehicle treatedcells from at least 5 independent trials where. *p < 0.05, **p < 0.01, ***p < 0.001 * p < 0.05, **p < 0.01, ***p < 0.001 versus vehicle, § p < 0.05, §§p < 0.01, §§§p < 0.001 versus AβRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 9 of 19of the Aβ preparation added to hCMEC/D3, reasoningthat if HDL acts primarily through maintaining Aβ in asoluble monomeric state, the protective effect of HDLshould be diminished if ECs are stimulated with pre-formed Aβ oligomers or fibrils. Soluble monomeric andoligomeric, and insoluble fibrillar Aβ were prepared asdescribed [24] and structures were confirmed using adot blot assay as well as electron microscopy (Fig. 3b,c).Interestingly, monomeric and oligomeric Aβ40 andAβ42 preparations induced comparably robust PBMCadherence, whereas the insoluble fibrillary form failed tosignificantly increase PBMC adhesion. Importantly, theability of HDL to suppress PMBC adhesion was un-affected by input of either Aβ monomers or oligomers(Fig. 3d-e). These results suggest that delayed Aβ fibrilli-zation may not be the major pathway by which HDLsuppresses PBMC adhesion to ECs under our experi-mental conditions as HDL maintained its protective ef-fect when ECs are treated with pre-formed oligomers.To confirm this, we also varied the timing of HDL andAβ addition to hCMEC/D3 such that, in addition to a2 h HDL pre-incubation as above, HDL was added atthe same time as Aβ (co-incubation) or 1 h after Aβaddition (post-incubation). We reasoned that if HDLacts primarily through maintaining Aβ solubility, pre-incubation and co-incubation designs should giveequivalent and maximal suppression of Aβ-inducedPBMC adhesion as, in both scenarios, Aβ is always inthe presence of HDL and aggregation could be delayed.In contrast, as the post-incubation scenario would allowAβ aggregation to begin prior to addition of HDL, theprotective effect of HDL should be reduced if addedafter the Aβ stimulus if Aβ structure is the key driver ofPBMC adhesion to ECs. We observed that PMBC adhe-sion was significantly and similarly reduced regardless ofwhen HDL was added to hCMEC/D3 relative to Aβ40or Aβ42 providing additional support that HDL’s abilityto attenuate Aβ-mediated PBMC adhesion to hCMEC/D3 is independent of whether Aβ input is monomeric oroligomeric (Fig. 3f,g).These two lines of evidence suggest that, although HDLcan affect Aβ structure in cell-free conditions, theseeffects are neither rapid nor robust enough to fully explainhow HDL suppresses Aβ mediated PBMC adhesion tohCMEC/D3. We next evaluated signalling pathways impli-cated in HDL’s anti-inflammatory effects on ECs.The ability of HDL to suppresses aβ-induced PBMCadhesion to hCMEC/D3 is independent of nitric oxide(NO) production and Annexin A1 (Anx1)Stimulation of NO synthesis by HDL is reported to reducearterial EC activation [25]. We therefore tested whetherHDL induces NO production in hCMEC/D3 using thecell permeable indicator 4,5-Diaminofluorescein diacetate(DAF2), which reacts with NO to produce highly fluores-cent trizaolo-fluorescein. Compared to baseline condi-tions, HDL-treated hCMEC/D3 showed a significant 130%increase (p = 0.01) in NO production (Fig. 4a), as well assignificantly elevated phosphorylated endothelial NO syn-thase (eNOS) on serine 1177 (161%, p = 0.047) (Fig. 4b,c).These results confirm that HDL can induce NO produc-tion in hCMEC/D3, as observed in ECs derived fromother origins.To test whether NO generation is required for HDL tosuppress Aβ-mediated PBMC adhesion to hCMEC/D3,we treated cells with the eNOS inhibitor L-NAME,followed by the addition of HDL, and finally stimulationwith either Aβ40, Aβ42 or TNF-α. We observed thatblocking eNOS with L-NAME had no effect on the abil-ity of HDL to suppress either Aβ- or TNF-α-mediatedPBMC adhesion to hCMEC/D3 (Fig. 4d-f ). Additionalexperiments confirmed the potency of L-NAME to blockeNOS in HDL-stimulated HUVEC (Additional file 2a),yet this also had no effect on the ability of HDL tosuppress either Aβ- or TNF-α- mediated PBMC bindingin HUVEC (Additional file 2b–d). We further ruled outa role for the NO pathway in hCMEC/D3 by measuringthe effect of HDL in the presence of VPC23019, an an-tagonist of sphingosine-1 phosphate receptor 1 (S1P1)and S1P3, receptors that are required for eNOS activity[26], and again found no effect on the ability of HDL tosuppress either Aβ- or TNF-α-mediated PBMC adhesionto hCMEC/D3 (Fig. 4g-i) or HUVEC (Additional file2e–h). Taken together, these results provide strongsupport that the eNOS pathway is not involved in theability of HDL to suppress PBMC adhesion to Aβ- orTNF-α-activated hCMEC/D3 and HUVEC.In contrast to a recent report suggesting that HDL up-regulates Anx1, which affects HDL’s ability to suppressECs activation [27], we observed no significant changein Anx1 protein levels in hCMEC/D3 after either HDLor Aβ treatment (Additional file 2i).HDL-mediated suppression of aβ-induced PBMC adhesionto hCMEC/D3 is independent of miR-233In addition to numerous proteins and lipid species, HDLparticles carry microRNAs (miRNAs) that can correlatewith vascular disease risk [28]. One of the most abundantmiRNAs in HDL is miR-233, which has recently beenshown to affect gene expression in human coronary arteryECs including suppressing ICAM-1 expression [29]. Threelines of evidence suggest that HDL-mediated suppressionof Aβ-mediated PBMC binding to hCMEC/D3 is independ-ent of miR-233. First, direct transfection of a miR-223mimetic did not prevent Aβ40 or TNF-α-mediated PBMCadhesion to hCMEC/D3, although a statistically significantreduction with Aβ42 was observed (Additional file 3a–c).Second, priming hCMEC/D3 with HDL in the presence ofRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 10 of 19a specific miR-223 inhibitor did not diminish the ability ofHDL to suppress PMBC adhesion by Aβ40 , Aβ42 or TNF-α (Additional file 3d–f). Third, in the 5 h period wherehCMEC/D3 were exposed to HDL (2 h pre-treatment then3 h with Aβ), no significant transfer of miR-223 wasdetected (see Additional file 3g), consistent with a previousreport that after 6 h, miR-223 was not transferred fromHDL to several types of ECs including HUVEC [30]. Theseresults suggest that, although the miR-233 mimetic cansuppress Aβ42-mediated PBMC binding to hCMEC/D3when added directly, miR-233 is unlikely to underlie theability of HDL to suppress PBMC adhesion to hCMEC/D3triggered by Aβ40, Aβ42 or TNF-α.HDL-mediated suppression of aβ-induced PBMC adhesionto hCMEC/D3 is independent of ICAM-1 and VCAM-1.ICAM-1 and VCAM-1 are canonical markers of EC activa-tion that mediate PBMC adhesion by classical inflammatorystimuli [31]. Several studies have reported elevated micro-vascular inflammatory markers in the AD brain [32, 33].Using ELISA, we confirmed that protein levels of intercellu-lar adhesion molecule 1 (ICAM-1) but not vascular celladhesion molecule 1 (VCAM-1) are increased in the cortexof AD patients compared to either cerebellum from thesame individual or the cortex from non-cognitively impaired(NCI) control subjects (Additional file 4). Increased ICAM-1 levels were further confirmed using immunofluorescencea b cd e fg h iFig. 4 HDL suppression of Aβ-induced inflammation is independent of eNOS and S1P. a Intracellular NO production was measured by treatinghCMEC/D3 with 100 μg/mL HDL in the presence of 1 μM DAF-2 for 6 h. Fluorescence was measured at 485 nm. b Phosphorylation of eNOS wasmeasured by treating hCMEC/D3 with 100 μg/mL HDL for 15 min before immunoblotting for phosphorylated eNOS (p-eNOS) or total eNOS.Representative immunoblots are shown in (c). hCMEC/D3 were pretreated for 1 h with the eNOS inhibitor L-NAME (d-f) or the S1P1 and S1P3 inhibitorVPC23019 (g-i) followed by 100 μg/mL HDL for 2 h. Cells were then stimulated with 0.1 μM Aβ40 (d,g) Aβ42 monomers, (e,h) or 1 ng/mL of TNF-α(f,i) for 3 h before testing PBMC adherence. Graphs represent means ± SD of adhered PBMC relative to vehicle treated cells for at least 5 independenttrials. *p < 0.05, **p < 0.01, ***p < 0.001 * p < 0.05, **p < 0.01, ***p < 0.001 versus vehicle, § p < 0.05, §§ p < 0.01, §§§p < 0.001 versus Aβ or TNF-αRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 11 of 19revealing both vascular and glial ICAM-1 reactivity whereasVCAM-1 remained unchanged (Additional file 5a–b). Asexpected, soluble Aβ40 and Aβ42 levels were elevated in thecortex of AD patients compared to their correspondingcerebellum or to the cortex of NCI subjects (Additional file4). Because ICAM-1 is a classical marker of EC activationand HDL can attenuate PBMC adherence by reducingICAM-1 levels in aortic derived ECs [34], we also evaluatedthe roles of ICAM-1 and VCAM-1 in hCMEC/D3 activa-tion. Using Western blotting, we observed that pretreatmentof monotypic hCMEC/D3 cultures with HDL significantlysuppressed TNF-α-mediated induction of ICAM-1 (from1343% to 920%, p = 0.041) and VCAM-1 (from 2838% to1928% p = 0.012), relative to untreated cells (Additional file6a–c), results that are consistent with HDL’s anti-inflammatory effects on peripherally derived ECs. By con-trast, Aβ treatment did not affect either mRNA levels ortotal protein levels of VCAM-1 and ICAM-1 in hCMEC/D3(Additional file 6d–m). Although total ICAM-1 levelsshowed no significant change in response to either Aβ40 orAβ42 treatment, biotinylation assays revealed significantlyincreased cell surface ICAM-1 levels, but not VCAM-1levels, in response to 0.1 μM of Aβ40 (p = 0.009) and astrong trend in response to Aβ42 (p = 0.06) (Additional file6 h–j). Unexpectedly, however, we observed that cell surfaceICAM-1 levels were not reduced by a 2 h exposure to HDL(Additional file 6 k–m), showing that cell surface ICAM-1levels do not correlate with PMBC adhesion in Aβ-activatedhCMEC/D3. We next analysed the effects of Aβ on thephosphorylation of multifunctional serine/threonine proteinkinases that are involved in EC activation, survival,apoptosis, proliferation and migration. TNF-α activatedNFκB p-P65 (p = 0.002), SAPK/JNK (p = 0.005), MAPK/ERK (p = 0.01), but not Akt or STAT2 pathways, whereasAβ did not alter any of these pathways in hCMEC/D3(Additional file 7). Taken together, these results suggest thatAβ does not activate any of classical inflammatory pathwaysknown to increase PBMC adhesion through ICAM-1 orVCAM-1.Suppressing aβ uptake into hCMEC/D3 blocks PBMCadherenceAs our results show that Aβ activates hCMEC/D3through a mechanism that is independent of canonicalEC activation intracellular signalling pathways, wehypothesised that binding or uptake of Aβ to ECs mightinfluence PBMC adhesion. To test this hypothesis, wefirst used temperature modulation experiments to inves-tigate how HDL affects the interactions of Aβ withhCMEC/D3 and observed that HDL pre-treatment sig-nificantly reduced total association measured at 37 °C(Fig. 5a,b), cell surface binding measured at 4 °C mea-sured by ELISA (Fig. 5c,d) and intracellular uptake offluorescently labeled Aβ40 and Aβ42 (Fig. 5e). Second,we tested whether blocking RAGE and LRP1, known re-ceptors that modulate Aβ binding and uptake in ECs[35], might reduce PBMC adhesion. Blocking RAGEwith a specific antibody abrogated PBMC adhesion toboth Aβ40 and Aβ42 but not TNF-α stimulatedhCMEC/D3(Fig. 5f-h). Similarly, blocking LRP1 with re-ceptor associated protein (RAP) abolished Aβ40 andAβ42 but not TNF-α effects (Fig. 5i-k). Finally, as hep-arin sulphate proteoglycans (HSPG) are also involved inbinding and uptake of Aβ in neuron [36], we then satu-rated or removed HSPG by treating hCMEC/D3 withheparin or heparinase III, respectively. Both treatmentsabolished the ability of Aβ40 and Aβ42 to induce PBMCbinding whereas TNF-α induced-PBMC adhesionremained significant (Fig. 5l-n). Together, these observa-tions show that, unlike TNF-α, Aβ induces PBMC adhe-sion to ECs through a pathway that requires interactionswith or internalization through Aβ receptors on the ECsurface. We also observed that HDL does not decreasethe expression of RAGE or LRP1 (Additional file 8), sug-gesting that down-regulation of these receptors cannotexplain the protective effect of HDL on Aβ-inducedPBMC binding to hCMEC/D3.Scavenger receptor-BI (SR-BI) is necessary for HDL to sup-press aβ-induced PBMC adhesion to ECsScavenger receptor BI (SR-BI) is the principal HDL recep-tor on ECs and activates several HDL signalling pathwaysin addition to mediating selective cholesterol uptake uponHDL docking [37]. We observed that inhibiting SR-BIbinding with a specific blocking antibody abolished theability of HDL to suppress both Aβ- and TNF-α-mediatedPBMC adhesion to hCMEC/D3 (Fig. 6a-c). Selectivity toSR-BI was confirmed by demonstrating that blockingscavenger receptor CD36 with a specific antibody had noeffect on the ability of HDL to suppress either Aβ-or TNF-α-mediated PBMC adhesion to hCMEC/D3(Fig. 6d-f). Intriguingly, inhibiting SR-BI-mediated selectivecholesterol uptake with block lipid transfer 1 (BLT1) didnot affect the ability of HDL to suppress either Aβ- orTNF-α-mediated PBMC adhesion to hCMEC/D3(Fig. 6g-i). We further observed that inhibiting SR-BIwith a specific blocking antibody abolished the ability ofHDL to suppress Aβ uptake in hCMEC/D3 (Fig. 6j). Theseresults demonstrate that HDL requires SR-BI-mediatedsignalling to inhibit Aβ uptake and subsequently suppressPBMC adhesion to hCMEC/D3 in response to Aβ- orTNF-α in a manner independent of lipid uptake.Having used monotypic static cell cultures to demon-strate that HDL suppresses Aβ-induced PBMC adhesion toendothelial cells through a mechanism that requires SR-BIand is independent of ICAM-1 expression, we confirmedthat this mechanism also explains the results in our 3Dbioengineered vessels. We observed that inhibiting SR-BIRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 12 of 19a bc def g hi j kl m nFig. 5 (See legend on next page.)Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 13 of 19with a specific antibody circulated through the vessel lumenabolished the ability of HDL to suppress Aβ-mediatedmonocyte adhesion to the vascular endothelium (Fig. 7a-b)without reducing ICAM-1 expression (Fig. 7c-d).DiscussionFar from being an inert vascular lining, ECs form ametabolically active and specialized interface betweenblood and underlying tissues. In response to an inflam-matory stimulus, ECs undergo activation, which is clas-sically defined by increased interaction with bloodleukocytes. EC dysfunction plays a central role in periph-eral vascular disease, stroke, heart disease, diabetes, insu-lin resistance, chronic kidney failure, tumor growth,metastasis, venous thrombosis, and severe viral infec-tious diseases [38]. The richly vascularized human braincontains ECs that form the nearly impenetrable BBB,which regulates the flux of substances in and out of theCNS. The contribution of cerebrovascular dysfunction indementia and cognitive decline is well acknowledged[39, 40], yet pathways that mediate cerebrovascular ECdysfunction and protection are not well understood.Using classical PMBC adhesion assays, we demon-strate that Aβ can act as an inflammatory stimulus to in-crease PBMC adhesion to brain microvascular ECs thatcan be suppressed by HDL. Our results show that theability of HDL to attenuate Aβ-induced PBMC adhesionto ECs is independent of Aβ structure, NO production,miR-223 and, unexpectedly, of ICAM-1 and VCAM-1expression. Rather, HDL requires SR-BI to suppress Aβ-mediated PBMC adhesion to ECs, which can also beattenuated by inhibiting internalization of Aβ by block-ing RAGE or LRP1, or by inhibiting Aβ-HSPG interac-tions. We also show that HDL can also prevent Aβfibrillization, suggesting that HDL may also help tomaintain Aβ solubility as it transits across the BBB.However, as this function of HDL is relatively slow, it isunlikely to account for HDL’s ability to suppress Aβ-mediated PBMC adhesion to ECs. Our observations areconsistent with a previous study that demonstrated thatdeficiency of SR-BI increases cerebrovascular amyloiddeposition in the J20 mouse model of AD [41].As ICAM-1 levels are clearly elevated in AD cortex,and previous studies have reported induced adhesionmolecule expression after Aβ stimulation [42], we weresurprised to observe that Aβ had negligible effects onadhesion molecule expression and kinase activationunder our experimental conditions, whereas TNF-α ro-bustly induces ICAM-1 and VCAM-1 expression as wellas NFκB, MAPK/ERK, and SAPK/JNK signalling. Thecellular processes triggered by internalized Aβ that pro-mote PBMC adhesion remain to be determined.In cultured aortic ECs, HDL reduces adhesion mol-ecule expression and subsequent monocyte binding via amechanism involving SR-BI signalling through eNOS,which requires the S1P receptor [25]. Our observationsthat HDL’s ability to suppress Aβ‑ and TNF-α-mediatedPBMC adhesion to hCMEC/D3 is independent of NOproduction and S1P activity suggests that brain ECs mayhave pathways distinct from ECs derived from arterialsources, at least under the culture conditions used here.Reduced eNOS expression in the AD brain has beenhypothesized to increase Aβ deposition and promoteproduction of reactive oxygen species production, whichreduces vasomotor regulation of penetrating arterioles[43, 44]. HDL clearly induces NO production inhCMEC/D3 and HUVEC, and although NO productionis not required to protect hCMEC/D3 from Aβ- orTNF-α mediated PBMC adhesion per se, HDL-stimulated NO production may provide beneficial effectson other cerebrovascular EC functions.A wealth of studies have explored in vitro associationsof Aβ with lipoprotein components, primarily focussingon apoE as genetic variants of apoE have established ef-fects on Aβ metabolism in AD [45]. As Aβ is a hydro-phobic peptide, it is not surprising that it associates withlipoprotein particles, and also to amphipathic apolipo-proteins. Elucidating the structure-function relationshipsof Aβ with lipoproteins remains an active area of re-search. While Aβ might associate with plasma lipopro-teins [46], the effect of HDL on Aβ structure remainspoorly understood. Interestingly, HDL constituents ontheir own have been reported to both interact with Aβand diminish its toxic effects. Specifically, small lipo-somes (<50 μm) accelerate Aβ40 fibrillization and theamounts of amorphous aggregates become larger asliposome increase (>50 μm) [47]. By contrast, nanolipo-somes (100 nm) prevent Aβ42 fibrillization and reducesAβ42-induced EC dysfunctions [48]. In addition, lipid-free apoA-I induces Aβ aggregation and generation of(See figure on previous page.)Fig. 5 HDL reduces Aβ association, binding and uptake to hCMEC/D3 whereas blocking Aβ binding or uptake reduces PBMC adhesion tohCMEC/D3. a-d hCMEC/D3 were pre-treated with 100 μg/mL of HDL and 0.1 μM a,c Aβ40 or b,d Aβ42 monomers as described in Fig. 2 at either37 °C (association a,b) or at 4 °C (binding c,d). Cells were lysed in RIPA buffer and Aβ were measured using commercial ELISA. e hCMEC/D3 werepre-treated with HDL (1 mg/mL) for 2 h before stimulating with 1 mM of fluorescently labelled Aβ40 or Aβ42 monomers. Scale bar represents10 μm. f-n hCMEC/D3 were pre-treated with (f-h) RAGE blocking antibody, i-k RAP or l-n heparin or heparinase III 60 min before stimulation withAβ40 or Aβ42 monomers or TNF-α for 3 h. PBMC adhesion assays were conducted as described in Fig. 2. Graphs represent means ± SD relativeto vehicle treated cells for at least 3 independent trials. *p < 0.05, **p < 0.01 versus vehicleRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 14 of 19a b cd e fgjh iFig. 6 (See legend on next page.)Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 15 of 19amorphous complexes, both of which reducing amyloidtoxicity [22]. Our results clearly show that mature HDLdelays Aβ fibrillization in a cell-independent assay withsubstantially slower kinetics than HDL’s ability to attenu-ate Aβ-induced PBMC adhesion to ECs. As HDL alsoreduces Aβ-induced PBMC adhesion to ECs independ-ent of monomeric vs. oligomeric input structures, we be-lieve that HDL’s ability to delay Aβ fibrillization is adistinct function compared to reduce Aβ-inflammation,which relies exclusively to a cellular process involvingSR-BI.As traditional static monotypic cell cultures do notreproduce the physiological complexity of the native vas-cular bed, we also used a 3D model of the human vascula-ture in our studies. These engineered vessels mimic anative vessel with a luminal monolayer of human ECs sur-rounded by several layers of human SMCs, through whichhuman mononuclear cells can be circulated under flowconditions. This model has been previously used to studyatherosclerosis in peripheral arteries [17]. Here, weinjected Aβ on the anteluminal side to mimic brain-produced Aβ and circulated HDL and monocytes throughthe lumen. We confirmed that Aβ induces monocyte ad-hesion to ECs under native-like flow conditions in engi-neered human vessels, which can be suppressed by HDLthrough a mechanism that requires SR-BI and that is inde-pendent of changes in ICAM-1 levels.ConclusionNumerous studies show that HDL has several propertiesthat are associated with improved vascular function [49].While some lines of evidence also suggest a beneficialrole for HDL in protecting from cognitive decline, thenature of this association remains largely unknown.Whereas lipid-free apoA-I can be transported into theCNS [50], there is thus far no evidence that matureHDL might cross the BBB, supporting the view thatHDL might act predominately from the blood compart-ment and would therefore primarily affect ECs. Our datasuggest at least three mechanisms by which HDL couldhave beneficial effects on cerebral vessels. First, HDLprevents Aβ-induced PBMC adhesion to brain micro-vascular ECs through a mechanism that requires SR-BIand suppresses Aβ uptake. The cellular pathways bywhich internalized Aβ promotes PBMC adhesion to ECsremains to be defined, and appears to be distinct fromthose induced by classical inflammatory stimuli. Second,similar to its effects in peripheral EC, HDL induces NOsecretion in hCMEC/D3. Although we ruled out a rolefor eNOS in Aβ-induced PBMC adhesion, HDL-stimulated NO production may still help to attenuatevasomotor dysfunction as observed in the aging brain[44]. Third, although HDL can protect hCMEC/D3 fromPBMC adhesion independent of Aβ structure, the abilityof HDL to maintain Aβ in a soluble state may facilitateefficient Aβ clearance out of the brain.(See figure on previous page.)Fig. 6 HDL suppression of Aβ-induced inflammation requires SR-BI. hCMEC/D3 were pre-treated for 1 h with a-c SR-BI or d-f CD36 blockingantibodies or g-i BLT1 followed by 100 μg/mL HDL for 2 h. Cells were then stimulated with 0.1 μM monomeric a,d,g Aβ40 or b,e,h Aβ42 or(c,f,i) 1 ng/mL of TNF-α for 3 h before evaluating PBMC adherence. j hCMEC/D3 were pre-treated for 1 h with SR-BI blocking antibody followedby HDL (1 mg/mL) for 2 h before stimulating with 1 mM of fluorescently labelled Aβ40 or Aβ42 monomers. Scale bar represents 10 μm. Graphsrepresent means ± SD of adhered PBMC relative to vehicle treated cells from at least 4 independent trials. *p < 0.05, **p < 0.01, ***p < 0.001 *p < 0.05, **p < 0.01, ***p < 0.001 versus vehicle, § p < 0.05 versus Aβ or TNF-α.a bc dFig. 7 HDL suppresses Aβ-induced monocyte adhesion in engineeredvessels via SR-BI. SR-BI blocking antibody was circulated through thelumen of engineered vessels prepared using HUVEC for 1 h priortreatment with 200 μg/mL of HDL for 2 h followed by injection of1 μM of monomeric a Aβ40 or b Aβ42 within the tissue chamber(abluminal) for 8 h prior to injecting circulating fluorescent THP1 intothe lumen. c-d ICAM-1 protein was measured in tissue lysates preparedin RIPA buffer by commercial ELISA. Graphs represent means ± SD ofadhered monocytes relative to Aβ treated tissues from at least 4individual grafts. * and # p < 0.05Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 16 of 19Additional filesAdditional file 1: Aβ induced dose dependent PBMC adhesion and HDLattenuate dose dependent Aβ-induced PBMC adhesion to hCMEC/D3. In allconditions, hCMEC/D3 were stimulated with 0–1 μM Aβ40 (light grey) orAβ42 (dark grey) monomers for 3 h. Fluorescently labelled PBMC wereallowed to adhere to (a) Aβ40- or (b) Aβ42- stimulated cells for 3 additionalhours. Cells were washed, fixed, and imaged to count adhered PBMC.hCMEC/D3 were primed with increasing doses (25–400 μg/mL) of HDL for2 h and stimulated with 0.1 μM (c) Aβ40 (light grey) or (d) Aβ42 (dark grey)for 3 h. Fluorescently labelled PBMC were allowed to adhere to stimulatedcells for 3 h followed by washing, fixation, imaging, and counting. Graphsrepresent mean ± SD of adhered PBMC relative to vehicle control from atleast 3 independent trials where * p < 0.05, **p < 0.01, ***p < 0.001 versusvehicle, § p < 0.05, §§ p < 0.01 versus Aβ (PDF 422 kb)Additional file 2 HDL suppression of Aβ-induced inflammation isindependent of eNOS and S1P in HUVEC. (a) L-NAME and (e) VPC23019potency was tested by measuring intracellular NO production in HUVECafter incubating with 100 μg/mL HDL and 1 μM DAF-2 for 6 h. Fluorescencewas measured at 485 nm. (b-d, f-h) In all conditions, HUVEC or hCMEC/D3were stimulated with 0.1 μM monomeric Aβ40 or Aβ42 or 1 ng/mL ofTNF-α for 3 h prior to measuring PBMC adherence. HUVEC were pre-treatedfor 1 h with (b-d) the eNOS inhibitor L-NAME or (f-h) the S1P1 and S1P3inhibitor VPC23019 followed by 100 μg/mL HDL for 2 h. (i) hCMEC/D3 werepre-treated with 100 μg/mL of HDL for 2 h before simulating with Aβ40 orAβ42. Total cellular expression of Annexin-1 (Anx1) was analysed byimmunoblotting and compared to GAPDH. Graphs represent means ± SDfrom at least 3 independent trials. * p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001 versus vehicle, § p < 0.05, §§ p < 0.01, §§§p < 0.001, §§§§p < 0.0001 versus Aβ or TNF-α (PDF 1059 kb)Additional file 3: HDL does not signal through miR-223 to reduce Aβ-induced inflammation in hCMEC/D3. (a-c) hCMEC/D3 were pre-treated with100 μg/mL of HDL as described in Fig. 2 with or without 10 nM of miR-223mimetic nucleotides or (d-f) in the absence or presence of a specific miR223inhibitor for 2 h before stimulation with (a,d) Aβ40, (b,e) Aβ42 or (c, f) TNF-αbefore testing PBMC adherence to ECs. (g) Intracellular levels of mature miR-223in hCMEC/D3 were quantified by real-time PCR and normalized to U6 after a5 h treatment with 100 μg/mL of HDL. Graphs represent means ± SD of ad-hered PBMC relative to vehicle treated cells for at least 5 independent trials.*p < 0.05, **p < 0.01, ***p < 0.001 * p < 0.05, **p < 0.01, ***p < 0.001 versusvehicle, § p < 0.05, §§ p < 0.01, §§§p < 0.001 versus Aβ or TNF-α (PDF 464 kb)Additional file 4: Demographic data, Aβ40 and Aβ42, and adhesionmolecule quantification in Alzheimer’s disease patients and non-cognitiveimpaired controls (PDF 431 kb)Additional file 5: Cortical ICAM-1 expression is increased in AD. Cryopre-served cortex and cerebellum of AD or NCI patients were cut at 20 μm. AfterPFA fixation sections were washed with PBS and stained against (a) ICAM-Ior (b) VCAM-1 and CD31 as a vascular marker and imaged using an invertedfluorescent microscope. Arrow demonstrates colocalization of ICAM-1 andCD31. Bar represents 50 μm (PDF 62676 kb)Additional file 6: Adhesion molecules are enhanced by TNF-α but notAβ.(a-c) hCMEC/D3 were primed with 100 μg/mL HDL for 2 h followedby stimulation with 1 ng/mL TNF-α for 3 h. Cell lysates were prepared inRIPA and protein levels of (a) ICAM-1 and (b) VCAM-1 were measured bydenaturing immunoblotting (c). (d-g) hCMEC/D3 were stimulated withmonomeric (d,f) Aβ40 or (e,g) Aβ42 at the indicated concentrations and(d,e) ICAM-1 and (f,g) VCAM-1 mRNA levels were measured by real-timePCR. (h-j) Following Aβ stimulation, cell surface proteins were biotinylatedand isolated by immunoprecipitation. Protein levels of cell surface andtotal (h) ICAM-1 and (i) VCAM-1 were measured by denaturingimmunoblotting (j). (k-m) hCMEC/D3 were pre-treated with 100 μg/mLof HDL for 2 h followed by stimulation with Aβ. After 3 h total and cellsurface ICAM-1 expression were measured as above. Graphs representmeans ± SD from at least 3 independent trials. ***p < 0.001 versusvehicle, § p < 0.05, versus Aβ or TNF-α (PDF 2282 kb)Additional file 7: Aβ does not activate phosphorylation ofmultifunctional serine/threonine protein kinases. hCMEC/D3 werestimulated with 0.1 mM of monomeric Aβ or 1 ng/mL of TNF-αfor15 min before lysing cells in RIPA containing phosphostop. Phosphoryl-ation of (a-b) p65, (c) STAT3, (d) Akt, (e) SAPK/JNK and (f) p42/44 MAPKwere analysed by immunoblotting and compared to respective total p65,STAT3, Akt, SAPK/JNK and p42/44 MAPK respectively. (b) Nuclear trans-location of p65 was analysed by immunofluorescence 15 min after Aβstimulation. Graphs represent means ± SD relative to vehicle treated cellsin 4 trials. *p < 0.05, **p < 0.01 (PDF 9645 kb)Additional file 8: HDL treatment does not alter LRP1 or RAGE proteinlevels in hCMEC/D3. (a-c) hCMEC/D3 were treated with HDL (0–400 μg/mL) for 5 h before lysing in RIPA. (a) LRP1 and (b) RAGE protein levelswere quantified by immunoblotting (c). Graphs represent means ± SDrelative to vehicle treated cells in 3 trials (PDF 1114 kb)AbbreviationsAD: Alzheimer Disease; Aβ: Beta-amyloid peptide; BLT-1: Block lipid transport1; CAA: Cerebral amyloid angiopathy; CNS: Central nervous system; EBM-2: Endothelial growth basal medium; EC: Endothelial cell; eNOS: Endothelialnitric oxide synthase; FBS: Fetal bovine serum; FITC: Fluoresceinisothiocyanate; hBMEC: Human brain microvascular endothelial cell; hCMEC/D3: Human cortical microvascular endothelial cell line D3; HDL: High densitylipoprotein; hSMC: Human smooth muscle cell; HSPG: Heparan sulphateproteoglycan; ICAM-1: Inducible cellular adhesion molecule-1; L-NAME: L–NG-nitroarginne methyl ester; LRP1: Low density lipoprotein receptor relatedprotein-1; NO: Nitric oxide; PBMC: Peripheral blood mononuclear cell;RAGE: Receptor of advanced glycation endproducts; SR-BI: Scavengerreceptor BI; VCAM-1: Vascular cell adhesion molecule −1AcknowledgementsNot applicableFundingThis work was supported by operating grants from the Canadian Institutes ofHealth Research (CIHR), a Canadian Consortium of Neurodegeneration andAging (CCNA), a Djavad Mowafaghian Centre for Brain Health Catalyst grant,philanthropic funding from the Jack Brown and Family Alzheimer’s ResearchFoundation, the Y.P. Heung Foundation, and a private BC-based foundation toCLW, and a Weston Brain Institute Rapid Response grant to JR. JR was sup-ported by a BrightFocus postdoctoral fellowship and a Swiss National ScienceFoundation Early.postdoctoral fellowship. EB was supported by a CIHR DoctoralScholarship. GKB was supported by a UBC four years doctoral Fellowship. IK wassupported in part by a Weston Brain Institute Rapid Response grant. NC wassupported by CIHR and Brain Canada and GF was supported by a CIHR ProjectGrant 148,883 (GAF).Availability of supporting data:All data generated or analysed during this study are included in thispublished article and its supplementary information files.Author contributionsJR, EB and CLW conceived the project. JR and EB made equivalent contributionsto this study and performed most of the experiments with assistance from SS,GKB, EG, CC, WHC, SKS, BY, MG, AB, KK and IK. Data analysis was primarilyperformed by JR. EB, and CLW, who together wrote the manuscript withassistance from EG, GF and NC. The authors declare no competing financialinterests. All authors read and approved the final manuscript.Authors’ informationNot applicableEthical approval and consent to participateAll experiments were conducted under an approved clinical protocol fromthe UBC Clinical Ethics Research Board after obtaining written informed consentfrom all subjects (H13–02719: isolation of vascular cells and H14–03357:isolation of lipoprotein and PBMC).Consent for publicationNot applicableRobert et al. Molecular Neurodegeneration  (2017) 12:60 Page 17 of 19Competing interests:The authors declare that they have no competing interests.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Department of Pathology and Laboratory Medicine, University of BritishColumbia, Vancouver, BC V6T 1Z3, Canada. 2Djavad Mowafaghian Centre forBrain Health, University of British Columbia, 2215 Wesbrook Mall, Vancouver,BC V6T 1Z3, Canada. 3Department of Neurology, University of BritishColumbia, Vancouver, BC V6T 2B5, Canada. 4Department of Zoology,University of British Columbia, Vancouver, BC V6T 2B5, Canada. 5Departmentof Medicine, University of British Columbia, Vancouver, BC V6Z 1Y6, Canada.Received: 4 May 2017 Accepted: 7 August 2017References1. Alzheimer’s Association. 2015 Alzheimer’s disease facts and figures.Alzheimer’s & dementia : the journal of the Alzheimer’s Association.2015;11(3):332–84.2. Perl DP. Neuropathology of Alzheimer’s disease. The Mount Sinai journal ofmedicine, New York [Internet]. 2010;77(1):32–42.3. Debette S. Vascular risk factors and cognitive disorders. Rev Neurol. 2013;169(10):757–64.4. Justin BN, Turek M, Hakim AM. Heart disease as a risk factor for dementia.Clin Epidemiol. 2013;5:135–45.5. Craft S. The role of metabolic disorders in Alzheimer disease and vasculardementia: two roads converged. Arch Neurol. 2009;66(3):300–5.6. Schneider JA, Arvanitakis Z, Bang W, Bennett DA. Mixed brain pathologiesaccount for most dementia cases in community-dwelling older persons.Neurology. 2007;69(24):2197–204.7. Tarasoff-Conway JM, Carare RO, Osorio RS, Glodzik L, Butler T, Fieremans E,Axel L, Rusinek H, Nicholson C, Zlokovic BV, Frangione B, Blennow K,Ménard J, Zetterberg H, Wisniewski T, de Leon MJ. Clearance systems in thebrain-implications for Alzheimer disease. Nat Rev Neurol. 2015;11(8):457–70.8. Grammas P. Neurovascular dysfunction, inflammation and endothelial activation:implications for the pathogenesis of Alzheimer’s disease. J Neuroinflammation.2011;8:26.9. Kontush A, Lhomme M, Chapman MJ. Unraveling the complexities of theHDL lipidome. J Lipid Res. 2013;54(11):2950–63.10. Shah AS, Tan L, Lu Long J, Davidson WS. The proteomic diversity of highdensity lipoproteins: our emerging understanding of its importance in lipidtransport and beyond. J Lipid Res. 2013 Feb;2411. Mahdy Ali K, Wonnerth A, Huber K, Wojta J. Cardiovascular disease riskreduction by raising HDL cholesterol–current therapies and futureopportunities. Br J Pharmacol. 2012;167(6):1177–94.12. Zuliani G, Cavalieri M, Galvani M, Volpato S, Cherubini A, Bandinelli S, Corsi AM,Lauretani F, Guralnik JM, Fellin R, Ferrucci L. Relationship between low levels ofhigh-density lipoprotein cholesterol and dementia in the elderly. The InChiantistudy. J Gerontol A Biol Sci Med Sci. 2010;65(5):559–64.13. Lefterov I, Fitz NF, A CA, Fogg A, Lefterov P, Kodali R, Wetzel R, Koldamova R.Apolipoprotein A-I deficiency increases cerebral amyloid angiopathy andcognitive deficits in APP/PS1DeltaE9 mice. J Biol Chem. 2010;285(47):36945–57.14. Lewis TL, Cao D, Lu H, Mans RA, Su YR, Jungbauer L, Linton MF, Fazio S,Mj LD, Li L. Overexpression of human apolipoprotein A-I preserves cognitivefunction and attenuates neuroinflammation and cerebral amyloidangiopathy in a mouse model of Alzheimer disease. J Biol Chem. 2010;285(47):36958–68.15. Robert J, Stukas S, Button E, Cheng WH, Lee M, Fan J, Wilkinson A, Kulic I,Wright SD, Wellington CL. Reconstituted high-density lipoproteins acutelyreduce soluble brain Aβ levels in symptomatic APP/PS1 mice. BiochimBiophys Acta. 2015;1862(5):1027–36.16. Riwanto M, Landmesser U. High-density lipoproteins and endothelialfunctions: mechanistic insights and alterations in cardiovascular disease.J Lipid Res. 2013;54(12):3227–43.17. Robert J, Weber B, Frese L, Emmert MY, Schmidt D, von Eckardstein A,Rohrer L, Hoerstrup SP. A three-dimensional engineered artery model for invitro atherosclerosis research. PLoS One. 2013;8(11):e79821.18. Hoerstrup SP, Cummings Mrcs I, Lachat M, Schoen FJ, Jenni R, Leschka S,Neuenschwander S, Schmidt D, Mol A, Günter C, Gössi M, Genoni M, Zund G.Functional growth in tissue-engineered living, vascular grafts: follow-up at100 weeks in a large animal model. Circulation. 2006;114:I159–66.19. Robert J, Lehner M, Frank S, Perisa D, von Eckardstein A, Rohrer L. Interleukin 6stimulates endothelial binding and transport of high-density lipoproteinthrough induction of endothelial lipase. Arterioscler Thromb Vasc Biol.2013;33(12):2699–706.20. Czupalla CJ, Liebner S, Devraj K. In vitro models of the blood-brain barrier.Methods in molecular biology (Clifton, NJ). 2014;1135:415–37.21. Koudinov AR, Berezov TT, Koudinova NV. The levels of soluble amyloid betain different high density lipoprotein subfractions distinguish Alzheimer’s andnormal aging cerebrospinal fluid: implication for brain cholesterolpathology. Neurosci Lett. 2001;314(3):115–8.22. Paula-Lima AC, Tricerri MA, Brito-Moreira J, Bomfim TR, Oliveira FF,Magdesian MH, Grinberg LT, Panizzutti R, Ferreira ST. Human apolipoproteinA-I binds amyloid-beta and prevents Abeta-induced neurotoxicity.Int J Biochem Cell Biol. 2009;41(6):1361–70.23. Lee SJC, Nam E, Lee HJ, Savelieff MG, Lim MH. Towards an understanding ofamyloid-β oligomers: characterization, toxicity mechanisms, and inhibitors.Chem Soc Rev. 2016;46(2):310–23.24. Stine WB, Jungbauer L, Yu C, LaDu MJ. Preparing synthetic Aβ in differentaggregation states. Methods in molecular biology (Clifton, NJ). 2011;670:13–32.25. Besler C, Heinrich K, Rohrer L, Doerries C, Riwanto M, Shih DM, Chroni A,Yonekawa K, Stein S, Schaefer N, Mueller M, Akhmedov A, Daniil G, Manes C,Templin C, Wyss C, Maier W, Tanner FC, Matter CM, Corti R, Furlong C, Lusis AJ,von Eckardstein A, Fogelman AM, Luscher TF, Landmesser U. Mechanismsunderlying adverse effects of HDL on eNOS-activating pathways in patientswith coronary artery disease. J Clin Invest. 2011;121(7):2693–708.26. Tölle M, Klöckl L, Wiedon A, Zidek W, van der Giet M, Schuchardt M. Regulationof endothelial nitric oxide synthase activation in endothelial cells by S1P1 andS1P3. Biochem Biophys Res Commun. 2016;476(4):627–34.27. Pan B, Kong J, Jin J, Kong J, He Y, Dong S, Ji L, Liu D, He D, Kong L, Jin DK,Willard B, Pennathur S, Zheng L. A novel anti-inflammatory mechanism ofhigh density lipoprotein through up-regulating annexin A1 in vascularendothelial cells. Biochim Biophys Acta. 2016;1861(6):501–12.28. Vickers KC, Palmisano BT, Shoucri BM, Shamburek RD, Remaley AT.MicroRNAs are transported in plasma and delivered to recipient cells byhigh-density lipoproteins. Nat Cell Biol. 2011;13(4):423–33.29. Tabet F, Vickers KC, Cuesta Torres LF, Wiese CB, Shoucri BM, Lambert G,Catherinet C, Prado-Lourenco L, Levin MG, Thacker S, Sethupathy P, Barter PJ,Remaley AT, Rye K-A. HDL-transferred microRNA-223 regulates ICAM-1expression in endothelial cells. Nat Commun. 2014;5:3292.30. Wagner J, Riwanto M, Besler C, Knau A, Fichtlscherer S, Röxe T, Zeiher AM,Landmesser U, Dimmeler S. Characterization of levels and cellular transfer ofcirculating lipoprotein-bound microRNAs. Arterioscler Thromb Vasc Biol.2013;33(6):1392–400.31. Lawson C, Wolf S. ICAM-1 signaling in endothelial cells. Pharmacol Rep.2009;61(1):22–32.32. Canobbio I, Abubaker AA, Visconte C, Torti M, Pula G. Role of amyloidpeptides in vascular dysfunction and platelet dysregulation in Alzheimer’sdisease. Front Cell Neurosci. 2015;9:65.33. Tchalla AE, Wellenius GA, Sorond FA, Gagnon M, Iloputaife I, Travison TG,Dantoine T, Lipsitz LA. Elevated soluble vascular cell adhesion molecule-1 isassociated with cerebrovascular resistance and cognitive function. J Gerontol ABiol Sci Med Sci. 2016;72(4):560–6.34. Calabresi L, Gomaraschi M, Villa B, Omoboni L, Dmitrieff C, Franceschini G.Elevated soluble cellular adhesion molecules in subjects with low HDL-cholesterol. Arterioscler Thromb Vasc Biol. 2002;22(4):656–61.35. Qosa H, Abuasal BS, Romero IA, Weksler B, Couraud P-O, Keller JN,Kaddoumi A. Differences in amyloid-β clearance across mouse and humanblood-brain barrier models: kinetic analysis and mechanistic modeling.Neuropharmacology. 2014;79:668–78.36. Liu C-C, Zhao N, Yamaguchi Y, Cirrito JR, Kanekiyo T, Holtzman DM, Bu G.Neuronal heparan sulfates promote amyloid pathology by modulating brainamyloid-β clearance and aggregation in Alzheimer’s disease. Sci Transl Med.2016;8(332):332ra44.37. Rohrer L, Ohnsorg PM, Lehner M, Landolt F, Rinninger F, von Eckardstein A.High-density lipoprotein transport through aortic endothelial cells involvesscavenger receptor BI and ATP-binding cassette transporter G1. Circ Res.2009;104(10):1142–50.Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 18 of 1938. Rajendran P, Rengarajan T, Thangavel J, Nishigaki Y, Sakthisekaran D, Sethi G,Nishigaki I. The vascular endothelium and human diseases. Int J Biol Sci.2013;9(10):1057–69.39. Snyder HM, Corriveau RA, Craft S, Faber JE, Greenberg SM, Knopman D,Lamb BT, Montine TJ, Nedergaard M, Schaffer CB, Schneider JA, Wellington C,Wilcock DM, Zipfel GJ, Zlokovic B, Bain LJ, Bosetti F, Galis ZS, Koroshetz W,Carrillo MC. Vascular contributions to cognitive impairment and dementiaincluding Alzheimer’s disease. Alzheimers Dement. 2015;11(6):710–7.40. Raz L, Knoefel J, Bhaskar K. The neuropathology and cerebrovascularmechanisms of dementia. J Cereb Blood Flow Metab. 2016;36(1):172–86.41. Thanopoulou K, Fragkouli A, Stylianopoulou F, Georgopoulos S.Scavenger receptor class B type I (SR-BI) regulates perivascular macrophagesand modifies amyloid pathology in an Alzheimer mouse model.Proc Natl Acad Sci U S A. 2010;107(48):20816–21.42. Giri R, Selvaraj S, Miller CA, Hofman F, Yan SD, Stern D, Zlokovic BV, Kalra VK.Effect of endothelial cell polarity on beta-amyloid-induced migration ofmonocytes across normal and AD endothelium. Am J Physiol Cell Physiol.2002;283(3):C895–904.43. Lamoke F, Mazzone V, Persichini T, Maraschi A, Harris MB, Venema RC,Colasanti M, Gliozzi M, Muscoli C, Bartoli M, Mollace V. Amyloid β peptide-induced inhibition of endothelial nitric oxide production involves oxidativestress-mediated constitutive eNOS/HSP90 interaction and disruption ofagonist-mediated Akt activation. J Neuroinflammation. 2015;12:84.44. Katusic ZS, Austin SA. Endothelial nitric oxide: protector of a healthy mind.Eur Heart J. 2014;35(14):888–94.45. Kanekiyo T, Xu H, Bu G. ApoE and Abeta in Alzheimer’s disease: accidentalencounters or partners. Neuron. 2014;81(4):740–54.46. Wilson LM, Pham CLL, Jenkins AJ, Wade JD, Hill AF, Perugini MA, Howlett GJ.High density lipoproteins bind Abeta and apolipoprotein C-II amyloid fibrils.J Lipid Res. 2006;47(4):755–60.47. Terakawa MS, Yagi H, Adachi M, Lee Y-H, Goto Y. Small liposomes acceleratethe fibrillation of amyloid β (1-40). J Biol Chem. 2015;290(2):815–26.48. Truran S, Weissig V, Madine J, Davies HA, Guzman-Villanueva D, Franco DA,Karamanova N, Burciu C, Serrano G, Beach TG, Migrino RQ. Nanoliposomesprotect against human arteriole endothelial dysfunction induced by β-amyloid peptide. J Cereb Blood Flow Metab. 2015;36(2):405–12.49. Di Angelantonio E, Sarwar N, Perry P, Kaptoge S, Ray KK, Thompson A,Wood AM, Lewington S, Sattar N, Packard CJ, Collins R, Thompson SG,Danesh J. major Lipids, apolipoproteins, and risk of vascular disease. JAMA.2009;302(18):1993–2000.50. Stukas S, Robert J, Lee M, Kulic I, Carr M, Tourigny K, Fan J, Namjoshi D,Lemke K, DeValle N, Chan J, Wilson T, Wilkinson A, Chapanian R,Kizhakkedathu JN, Cirrito JR, Oda MN, Wellington CL. Intravenously injectedhuman apolipoprotein A-I rapidly enters the central nervous system via thechoroid plexus. J Am Heart Assoc. 2014;3(6):e001156.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:Robert et al. Molecular Neurodegeneration  (2017) 12:60 Page 19 of 19


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items