Human brain tissue and characterization

Brain temporal lobe tissues from 8 cognitive controls and 8 AD patients were obtained from the Brain Bank of the Boston University Alzheimer’s Disease Center (BU ADC). The clinical features and ApoE genotypes are listed in Supplementary Table 1.

Mice and experimental treatments

Human ApoE genetic knock-in mice and ApoE knockout mice were purchased from Taconic Biosciences, Inc. (APOE2: #1547-F, APOE3: #1548-F, APOE4: #1549-F, ApoE−/−: Rensselaer, NY, USA). The human ApoE gene (either the ApoE2, ApoE3 or ApoE4 allele) replaces the endogenous mouse ApoE gene in these mice. C57BL/6 wild-type mice were purchased from Jackson Laboratory (#000664, Bar Harbor, ME, USA) to be used as a control group.

All mice were maintained in microisolator housing in the animal facility at Boston University School of Medicine. Recombinant mCRP was produced as described [18]. Female ApoE mice aged 9–11 months received an intraperitoneally (i.p.) injection with mCRP (200 µg/kg) in three days during 10am–12 pm per week (Monday, Wednesday and Friday) for 6 weeks. This dose was chosen based on previous pharmacokinetic studies [19]. mCRP was dissolved in phosphate buffered saline (PBS) before injection. Vehicle-treated mice were injected with PBS only as a control (n = 15–18 mice in each examined condition). The ApoE4 Protein (#350-04, PEPRO Tech, USA) and ApoE2 protein (#350-12, PEPRO Tech, USA) were IP injected into mice following up with mCRP. All animal procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Boston University Animal Care and Use Committee.

Microvessel isolation

Microvessels from either human or mouse brain tissue were isolated using a method based on dextran gradient centrifugation, combined with sequential cell strainer filtrations as previously described by [20]. Briefly, brains were carefully removed and transferred immediately to cold MCDB131 medium. The cortex/hippocampus tissue was dissected, and all visible white matter was discarded. The isolated tissue was placed to MCDB131 medium and homogenized using a loose fit 7 ml Dounce tissue grinder. The samples were transferred into 14 ml round-bottom centrifuge tube and then centrifuged at 2000 × g for 5 min. The pellets were resuspended in 15% dextran/PBS (70 kDa, Sigma). The samples were thoroughly mixed and centrifuged at 10,000 × g for 15 min, 4 °C. The microvessel -containing pellet located at the bottom of the tubes was collected, and sequentially resuspended using 1 ml PBS. The pellet was transfer onto a 40 μm cell strainer (BD Falcon, San Jose, CA, USA) and wash through < 10 ml PBS. The microvascular remaining on top were collected in MCDB 131 medium containing 0.5% endotoxin-, fatty acid- and protease free BSA and fixed by 4% PFA for further fluorescent staining analysis.

Microvascular profile measurements

To determine the lengths of CD31-positive cerebrovascular profiles, we followed the methodology previously reported by [21]. Fixed brain sections were prepared and subjected to staining as described in the “Immunofluorescence characterization” section. Subsequently, the sections were incubated with anti-CD31 primary antibody (1:200, #550,274, BD Biosciences, San Jose, CA, USA; 1:200 #BBA7, R&D systems, Minneapolis, MN, USA) and Alexa Fluor 594-conjugated secondary antibody.

In mouse brain sections, the CD31+ microvessels (vessels < 6 μm in diameter) were selected, and their lengths were measured from the 3D structures by using the ImageJ in length analysis tool in 5 randomly selected areas in the cortex. Z-stacks were collected at 1-µm steps for a total imaging depth of 16 μm. The average length of CD31+ microvessels was expressed as μm per mm3 of brain tissue.

Protein digestion

LC–MS grade solvent including acetonitrile, water, methanol and formic acid were obtained from ThermoFisher (USA). Micro-vessels (pellet) were resuspended in lysis buffer (100 uL for mouse sample, 200 uL for human brain sample, respectively) containing 8 M urea, 5 mM dithiothreitol (DTT), 50 mM ammonium bicarbonate (NH4HCO3) with EDTA free protease inhibitor (Sigma) and phosphatase inhibitor (PhosSTOP, Roche). After brief sonication using a Branson probe sonicator with 10% power (40 kHz) for 10 s on ice to shear DNA, the samples were reduced by DTT to a final concentration of 5 mM for 60 min at room temperature (RT) and alkylated by the addition of iodoacetamide (15 mM) and incubation at RT for 30 min in the dark. The extra iodoacetamide in solution was quenched by adding DTT to a final concentration of 10 mM and incubating for 20 min. Proteins were diluted with 50 mM NH4HCO3 to bring urea concentration down to below 1 M. After quantification with a BCA kit (ThermoFisher), proteins were digested with LC/MS-grade trypsin (1:50 enzyme to protein ratio, w/w) at 37 °C overnight with shaking followed by the addition of formic acid (FA) to 1% in solution to terminate the trypsinization. The resulting peptides were desalted using a C18 solid phase extraction Sep-Pak column (SPE, Waters) as per the manufacturer’s instructions. Briefly, after SPE columns were activated using 90% methanol and pre-conditioned using 0.1% FA, the peptide digests were loaded, washed with buffer of 0.1% FA for 2 times and eluted with 0.1% FA-60% acetonitrile (ACN). The desalted peptides were dried under vacuum at 45 °C and kept at − 20 °C prior to Tandem Mass Tag (TMT) labeling.

Tandem mass tag peptide labeling

Prior to TMT labeling, peptide quantification was performed by Pierce quantitative colorimetric assay (ThermoFisher). Each sample comprising 90 μg peptides was resuspended in 0.1 M triethylammonium bicarbonate (TEAB) and incubated with the.

TMTPro 16plex reagents (ThermoFisher). The ratio of TMT reagent to peptide was 4:1 (w/w). Reaction was carried out for 1 h at room temperature. To quench the reaction, 5% hydroxylamine was added to each sample and incubated for 15 min. Equal amounts of each sample were combined in a new tube and a speed vac was used to dry the labelled peptide sample. The labelled peptides were desalted using a C18 SPE column (Waters). Ten percent of the peptide was aliquoted and dried for global proteomics while 90% of the peptide was saved for phosphopeptide enrichment.

Phosphopeptide enrichment

Phosphopeptides were enriched by using Fe-NTA magnetic beads (CubeBiotech). Briefly, the Fe-NTA beads were washed three times with binding buffer of 0.1% TFA-80% ACN. The peptides were dried and resuspended in binding buffer then incubated with the Fe-NTA bead slurry for 30 min on a shaker. Bound phosphopeptides were washed 3 times with binding buffer, then serially eluted twice by adding 200 μL of 5% ammonium hydroxide-50% ACN to the beads. Eluted phosphopeptides were combined and dried in a speed vac prior to LC/MS.

LC–MS analysis

LC–MS analysis was performed using an Orbitrap Exploris 480 mass spectrometer equipped with high field asymmetric waveform ion mobility spectrometry (FAIMS) and interfaced to an Easy nanoLC1200 system (ThermoFisher Scientific). (Phospho)peptides were loaded onto a C18 pre-column (75 mm i.d. × 2 cm, 3 μm, 100 Å, ThermoFisher Scientific) then separated on a reverse-phase nano-spray column (75 mm i.d. × 50 cm, 2 μm, 100 Å, ThermoFisher Scientific) using gradient elution. Samples (2 μL) were injected and separated over 180 min gradient. The mobile phase A was consisted of 0.1% FA-2% ACN, and mobile phase B was consisted of 0.1% FA-80% ACN. The gradient consisted of 6% to 40% mobile phase B over 155 min, was increased to 95% mobile phase B over 4 min and maintained at 95% mobile phase B for 3 min at a flow rate of 250 nL/min. The MS instrument was operated in positive ion mode over a full mass scan range of m/z 350−1400 at a resolution of 60,000 with a normalized AGC target of 300% and maximum ion injection time of 25 ms for peptide and an auto mode was set for phoshopeptides respectively. The source ion transfer tube temperature was set at 275 °C and a spray voltage set to 2.5 kv. Data was acquired on a data dependent mode with FAIMS running 3 compensation voltages at − 50 v, − 57 v and − 64 v. MS2 scans were performed at 45,000 resolution with a maximum injection time of 80 ms and 200 ms for peptides and phosphopeptides respectively at normalized collision energy 34. Dynamic exclusion was enabled using a time window of 90 s.

Proteomic data analysis

MS2 spectra were processed and searched by MaxQuant (version 1.6) against a database containing native (forward) human or mouse protein sequences (UniProt) and reversed (decoy) sequences for protein identification. The search allowed for two missed trypsin cleavage sites, variable modifications of methionine oxidation, and N-terminal acetylation. The carbamidomethylation of cysteine residues was set as a fixed modification. For phosphopeptides, serine, threonine and tyrosine phosphorylation were set as variable modification. Ion tolerances of 20 and 4.5 ppm were set for the first and second searches, respectively. The candidate peptide identifications were filtered assuming a 1% false discovery rate threshold based on searching the reverse sequence database. Quantification was performed using the TMT reporter on MS2 (TMT pro16-plex).

Correction for multiple hypothesis testing was accomplished using the Benjamini–Hochberg false discovery rate (FDR). Human homologs of mouse genes were identified using HomoloGene (version 68). All analyses were performed using the R environment for statistical computing (version 3.6.0).

Bioinformatic analysis was performed in the R (version 3.6.1) statistical computing environment [22]. Gene Set Enrichment Analysis (GSEA) (version 2.2.1) [23] was used to identify biological terms, pathways and processes that were coordinately up- or downregulated within each pairwise comparison [24]. The Entrez Gene identifiers of the human homologs of all genes in the Ensembl Gene annotation were ranked by the Wald statistic computed between mCRP and PBS within each ApoE genotype group. Ensembl human genes matching multiple mouse Entrez Gene identifiers and mouse genes with multiple human homologs (or vice versa) were excluded prior to ranking so that the ranked list represents only those human Entrez Gene IDs that match exactly one mouse Ensembl Gene. Each ranked gene list was then used to perform preranked GSEA analyses (default parameters with random seed 1234) using the Entrez Gene versions of the Hallmark, BioCarta, KEGG, Reactome, PID, transcription factor motif, microRNA motif, and Gene Ontology (GO) gene sets in the Molecular Signatures Database (MSigDB) version 7.1 [25]. All the gene sets were used for analysis to facilitate full reproducibility of results.

Isolation, culture, and characterization of CD31 + brain endothelial cells (BECs)

Brain tissue samples were gently dissociated into single-cell suspensions using the Adult Brain Dissociation kit (#130,107,677, Miltenyi Biotec, Auburn, CA, USA). Single-cell isolation and characterization of CD31 + cells were conducted as previously described with a minor modification [26], (Zhang et al., 2021). Briefly, the brains containing the cortex and hippocampus regions were carefully removed and dissected using forceps under sterile conditions. The dissected tissues were then cut into eight slices using a razor blade. To dissociate the cells, the tissues underwent enzymatic digestion using specific components in the gentleMACS™ Octo Heat Dissociator, which facilitates mechanical dissociation. Following dissociation, a Debris Removal Solution was used to eliminate myelin and cell debris. Additionally, the Red Blood Cell Removal Solution was employed to remove erythrocytes from the cell suspension. These steps were conducted to prepare a purified cell suspension suitable for CD31 microbead-based isolation and culture of BECs.

After removal of CD45+ cells, CD31+ BECs were resuspended in fresh EBM-2 basal medium with all supplements (#CC-3202, EGM™-2-MV BulletKit™, Lonza, Portsmouth, NH, USA). Then, the BECs were seeded in 96-well plates coated with collagen type I (5 µg/mL, #354,231, BD Bioscience, San Jose, CA, USA) at a density of 104 cells per well. The medium was changed every 2 days.

CD31+ brain endothelial cells (BECs) were isolated from wild-type (WT) mice and cultured for 5 days. Subsequently, the in vitro cultured BECs were treated with mCRP at concentrations of 10 or 20 μg/ml, or with mCRP at a concentration of 10 μg/ml along with recombinant ApoE2 (1.0 μM) or ApoE4 (1.0 μM) protein. After a 24-h incubation period, the expression levels of various proteins of interest were assessed using specific antibodies. The cell nuclei were stained with DAPI.

Immunofluorescence characterization

Immunofluorescence was used to characterize the postmortem human- and mouse brains. Mouse brains were collected after PBS perfusion, post-fixed in 4% paraformaldehyde for 48 h, and changed to 30% sucrose in PBS at 4 ℃. Coronal cryosections (30 μm in thickness) were used for the free-floating staining method. For frozen human postmortem brain, the sample was embedded in OCT compound, cut into 16 µm thick cryosections and mounted on gelatin-coated histological slides. The sections were allowed to air dry for 30 min and immediately fixed in ice-cold fixation buffer for 15 min. Brain slides were preincubated in blocking solution with 5% [vol/vol] horse serum (Sigma-Aldrich, St Louis, MO, USA) in 1 Tris-buffered saline for 2 h at room temperature. The slides were incubated individually with primary antibodies overnight. Brain slides were then stained with secondary antibodies conjugated with Alexa Fluor 488, 594, 549, or 647 (1:500, Thermo Fisher Scientific, Carlsbad, CA, USA) for 1 h at room temperature. The sections were mounted with ProLong Gold antifade reagent with DAPI for nuclear staining (#P36935, Thermo Fisher Scientific, Carlsbad, CA, USA). The stained slides were observed under fluorescence microscopy (Carl Zeiss, Germany).

The following primary antibodies were used: (1) 3H12 antibody (1:50) against mCRP [27]; (2) anti-CD31 antibodies (1:200, #550,274, BD Biosciences, San Jose, CA, USA; 1:200 #ab28364, Abcam, Cambridge, MA, USA; 1:200 #BBA7, R&D systems, Minneapolis, MN, USA); (3) anti-phosphorylated CD31 at tyrosine 702 (pCD31) antibody (1:200, #ab62169, Abcam, Cambridge, MA, USA); (4) anti-ApoE antibody (1:500, #701,241, Invitrogen, Carlsbad, CA, USA); (5) anti-phosphorylated tau (pTau) PHF1 antibody (1:200, from Dr. Benjamin Wolozin Lab); (6) anti-NeuN antibody (1:1000, #ab177487, Abcam, Cambridge, MA, USA) to identify neurons; (7) anti-GFAP antibody (1:1000, #14-9892-82, Fisher Scientific, Hampton, NH, USA) for an astrocyte biomarker; (8) anti-CD68 (1:500, #MCA1957GA, Bio-Rad Laboratories, Hercules, CA, USA) and anti-Iba-1 (1:500, #019-19741, Wako Chemicals, Osaka, Japan) antibodies for microglia biomarkers; (9) anti-Von Willebrand Factor antibody (1:200, #ab11713, Abcam, Cambridge, MA, USA) to identify vascular damage; (10) anti-CD3 (1:500, #ab16669, Abcam, Cambridge, MA, USA; #MAB4841, R&D systems, Minneapolis, MN, USA) and anti-CD8 (1:500, #NBP1-49045SS, Novus Biologicals, Littleton, CO, USA) antibodies to identify T lymphocytes; (11) anti-CD19 antibody (1:500, #NBP2-25196SS, Novus Biologicals, Littleton, CO, USA) to identify B lymphocytes; and (12) anti-CD14 antibody (1:200, #11-0141-82, Invitrogen, Carlsbad, CA, USA) to detect monocytes; (13) anti-Lectin antibody (1:500, #DL-1174, Vector, Burlingame, CA, USA) and anti-CD144 antibody (1:200, #14–1441-82 Invitrogen, Carlsbad, CA, USA) to label vascular and endothelia; (14) anti-Lima1 (1:250, # NBP187947, Novus, Centennial, CO) and anti-Lima1 (1:100, #sc-136399, Santa Cluz, Dallas, Texas, USA); (15) anti-VE-cadherin (1:200, #PA5-19,612, Thermo Fisher Scientific, Waltham, Massachusetts, USA).

To evaluate immunostaining results, ImageJ was used to measure total intensity after adjusting the threshold. The data obtained from two independent researchers who were blinded to the treatment groups were pooled and averaged.

Proximity ligation assay

To examine protein–protein interactions, we employed a proximity ligation assay (PLA) on both frozen brain sections and primary endothelial cells as following [28]. Briefly, the samples were incubated with blocking buffer at 37 °C for 1 h. To detect LIMA1 and pCD31 interaction, samples were incubated overnight at 4 °C with mouse anti-LIMA1 antibody (1:100, #sc-136399, Santa Cluz, Dallas, Texas, USA) and rabbit anti-pCD31 antibody (1:200, # PA5-104,813, Thermo Fisher Scientific, Waltham, Massachusetts, USA). To detect LIMA1 and VE-cadherin interaction, samples were incubated overnight at 4 °C with mouse anti-Lima1 antibody (1:100, #sc-136399, Santa Cluz, Dallas, Texas, USA) and rabbit anti- VE-cadherin antibody ((1:200, #PA5-19,612, Thermo Fisher Scientific, Waltham, Massachusetts, USA). The samples were similarly incubated with mouse IgG and goat IgG or rabbit IgG and rat IgG as control groups, respectively. Proximity ligation was then conducted in situ as described by the manufacturer’s instructions (#DUO92007, Sigma-Aldrich, St Louis, MO, USA). We used the Duolink PLA probes anti-mouse PLUS and anti-goat MINUS to visualize mCRP/CD31 interactions and the Duolink PLA probes anti-mouse PLUS and anti-rabbit MINUS to visualize LIMA1/pCD31 and LIMA1/VE-cadherin interactions by Duolink in Situ Detection Reagents Orange. The samples were further incubated with Alexa Fluor 488-and 647-conjugated secondary antibodies (Invitrogen, Carlsbad, CA, USA) for 1 h at room temperature. Following serial washes, the samples were stained with DAPI (1:104) and observed with a fluorescence microscope (ZEISS Axio Observer).

Detection of oxidative stress

Oxidative stress level in brain derived endothelial cells was assayed using CellROX Orange Reagent (Cat# C10443, Thermofisher Scientific) according to the manufacturer’s instructions. Briefly, WT brain endothelial cells were treated in vitro with 10 μg/ml concentrations of mCRP or vehicle control and incubated or different periods of time up to 24 h (h) on 96-well plate. The CellROX orange dye was added to the cells at a final concentration of 5 μM for 30 min before being washed three times with PBS. The cells were proceeded to imaging by using Zeiss LSM880 Confocal Microscope. The fluorescence density (O.C.U per μm2) was measured by ImageJ software (NIH). Three independent experiments were performed for the comparisons.

Quantification and statistical analysis

Experimenters were blinded to the genotypes during testing. All data are presented as the mean ± standard error of the mean (SEM). Statistical analyses were performed using GraphPad Prism (version 8.0). For mouse data, datasets were analyzed for significance using Welch’s two samples t-test. Comparisons between three or more groups were conducted using one-way or two-way analysis of variance (ANOVA). Post hoc multiple comparisons were carried out using Tukey’s test. For human data, we performed univariate analyses stratified by control vs. AD brains. Mean ± SD was determined, and ANOVA tests were conducted on variables. The correlation analysis was conducted by Pearson or Spearman test according to the distribution pattern of the data. Statistical significance was defined as p value < 0.05.