Animals. The Rosa26-Stopfl/fl-hCD33M/m mice were generated on a C57BL/6 genetic background and described previously [15, 19, 21]. The CX3CR1Cre (B6J.B6N(Cg)-Cx3cr1tm1.1(cre)Jung/J) and 5XFAD (B6.Cg-Tg(APPSwFlLon,PSEN1*M146L*L286V) 6799Vas/Mmjax) mice were obtained from the Jackson Laboratory. To generate 5XFAD mice expressing either hCD33M (Cx3cr1Cre+/+-hCD33M+/−) or hCD33m (Cx3cr1Cre+/+-hCD33m+/−), we crossed (Cx3cr1Cre+/+-hCD33M−/+ or Cx3cr1Cre+/+-hCD33m−/+ mice with 5XFAD−/+ mice and confirmed the progeny containing hCD33 and 5XFAD with designated primers [15, 19, 21]. Breeders containing either of the hCD33 transgenes were homozygous for CX3CR1Cre so that all the mice contained a single copy of CX3CR1Cre. All animals were maintained in ventilated racks (Tecniplast, Green Line) and cage environmental enrichment comprising 5 cm diameter plastic tubes and nesting material (“Nestlets”, Ancare Inc.). Animals were fed irradiated chow (LabDiets, 5053) and were housed with a 12 h/12 h light/dark cycle. All protocols were in accordance with the Canadian Council on Animal Care (CCAC) and were approved by the Animal Care and Use Committee at the University of Alberta.
Immunofluorescence (IF) staining. Half brain sections were fixed in 4% PFA at 4°C for 24 hr, followed by incubation with 30% sucrose (4°C) for a minimum of 72 hr. The tissue was then embedded in embedding medium for frozen tissue specimens (OCT, Thermo Scientific) and stored at -80°C until being further processed by cryostat (Thermo Scientific). Coronal sections (20 µm) within the hippocampus region were collected and a minimum of seven sections per sample were mounted onto Superfrost Plus microscope slides (Thermo Scientific) and stored at -80 ̊C.
For IF staining, slides were removed from − 80°C, allowed to adjust to room temperature for 20 min and washed with PBS prior to antigen retrieval. Slides were then incubated with PBS containing 5% goat serum and 0.1% Triton X-100 for 10 min at room temperature. Slides were then further treated with blocking solution containing 5% goat serum in PBS-T (PBS containing 0.2% Tween-20), followed by incubation with 500 µl of 5% goat serum containing primary antibodies overnight at 4°C. The following primary antibodies were used in our experiments: anti-Iba-1 (rabbit monoclonal, FUJIFILM Wako Chemicals, 1:600 dilution), and anti-Aβ (MOAB-2, abcam, 1:1000), anti-Ki67 (B56, abcam, 1:50) and anti-LAMP1 (1D4B, abcam, 1:200). The slides were washed three times in PBS-T the following day and incubated with the secondary antibodies (AF568-conjugated anti-rabbit, and AF647-conjugated anti-mouse, all used at 1:500 dilution) for 1 hr, followed by three more washes in PBS-T. To minimize the fluorescent background, autofluorescence quenching kit (TrueVIEW) was used as per the manufacturer’s protocol. Lastly, the slides were incubated with Hoechst (1:2000 dilution of 10 mg/ml stock solution) for 15 min and cover-slipped with permanent mounting medium (TrueVIEW).
Thioflavin-S (Thio-S) staining of amyloid aggregates was performed as described previously [22]. Briefly, sections were stained with 150 µM Thioflavin-S (Sigma) solution in 40% ethanol for 10 min at room temperature. Slides were then washed with 50% ethanol, followed by two washes in PBS prior to mounting coverslips.
Microscopy. Fluorescence microscopy was performed with the LSM 700 laser scanning confocal microscope (ZEISS), equipped with Axiocam 702 mono camera (ZEISS) and the images were captured at 10X magnification. A minimum of five brain sections from each animal were assessed for analysis. Confocal microscopy images were captured with the same microscope in confocal mode (software: Zen2.6 Black edition, ZEISS) and at 63X magnification oil immersion objective (N.A. 1.4) at 1024 x 1024 pixel resolution. Sample identity was blinded for all analyses and images were processed with Zen2.6 Blue edition software (ZEISS).
Image analyses and quantification. All the global analyses were performed on widefield images captured from the hemi-brains. This included Aβ plaque burden, number of Aβ clusters, percentage of ThioS+ clusters, and IBA-1 density in the whole brain as well as region-specific analysis of overall Aβ levels. In this regard, the total area of the detected signal was quantified and normalized to the total area of the brain frame. For the analyses performed on individual plaques, images captured in the confocal were used. The details of analyses performed on plaques are as follows:
Aβ and Thioflavin-S area for individual plaques. For measuring total Aβ area within each cluster, the plaque area was selected within a frame and Aβ or ThioS fluorescent signal was distinguished by thresholding. The total area of Aβ/ThioS within the frame was then measured and recorded per plaque. Classification of Thio-S positive plaques were done by scrolling through the z-stack and judging manually. Plaque compaction was measured through dividing the Thioflavin-S area by the total Aβ area. This was calculated both globally (in high plaque density regions) as well as for individual plaque clusters.
Plaque-associated microglia. The area of each plaque cluster was selected by a frame and the total number of plaque-associated microglia was calculated by counting the total number of IBA-1 positive nuclei (stained with Hoechst) within the allocated frame region. To measure plaque-associated microglia density, this number was divided by the area of plaque within the frame region.
Microglia-plaque interface. The area of each ThioS positive plaque cluster was selected by a frame and the thresholding for the ThioS channel was adjusted in a way that only the perimeter of the core would be quantified. The IBA-1 signal overlapping with ThioS within the perimeter was then measured by the software and the overlap was quantified as a percentage of total perimeter.
Percentage of internalized Aβ by microglia. Microglia interacting with Aβ clusters were selected and a barrier mask for the cells was defined based on the IBA-1 signal. The total area of Aβ within the mask was measured, quantified and normalized to the total area of IBA-1.
Percentage of proliferative microglia (Ki67 staining). The IBA-1+ cell bodies surrounding plaques (PAM) were counted and the number of Ki67+ cells was recorded. The proliferative population was then calculated as the percentage of total PAM that were Ki67+.
Quantification of LAMP1 + in dystrophic neurites (DNs) area in dorsal subiculum. The total area of LAMP1+ spheroids within the dorsal subiculum was measured and quantified as the percentage of the total brain frame.
Quantification of DN area in individual neuritic plaques (µm 2 ). The total area of spheroid within each neuritic plaque was measured and recorded. A total of 400 plaques from 10 mice per cohort were used for this analysis.
Extraction of soluble and insoluble Aβ from mouse brain. Half brain samples were individually homogenized in sterile PBS containing protease (cOmplete, Roche) and phosphatase inhibitor cocktails (Thermo Fisher Scientific). Homogenization was done using ceramic magnetic beads (2.8-mm ceramic beads; Bertin Technologies SAS) in an Omni bead Ruptor system (3.2 M/s shake speed, 10s rupture, 10 s break, three repeats) to obtain 10% (w/v) homogenates.
Extraction of soluble and insoluble Aβ was done as described previously with slight modifications [23]. Briefly, 130 µL of homogenate was thawed on ice and mixed with equal amount of 2% Triton X-100, to achieve a final concentration of 1% in the homogenate. Samples were then incubated on ice for 15 min, while being vortexed every 5 min, followed by ultracentrifugation at 100,000 rcf (4°C for 15 min). The supernatant was extracted and tested for total protein concentration with BCA assay. The pellet (insoluble fraction) was either resuspended in PBS and sonicated for 1 min for proteinase K (PK) treatment, or monomerized by addition of formic acid (final concentration of 70% (v/v). The volume of PBS used for resuspension of the pellet was adjusted based on the protein concentration of the soluble fraction.
For PK digestion, the insoluble fraction was aliquoted and treated with PK for 1 hr at 37 ºC. The remaining material were then monomerized by addition of formic acid at a final concentration of 70% (v/v) and sonicated for 1 min. The solutions were neutralized (1:20) with neutralization buffer (1M of Tris base, 0.5 M of Na2HPO4, 0.05% NaN3 (w/v)) prior to measurements. Aβ levels were quantified by an electrochemiluminescence-linked immunoassay (Meso Scale Discovery (MSD), Assay 2) as per manufacturer’s protocol. The plates were read on the SECTOR Imager 6000 and data analysis was performed using the MSD DISCOVERY WORKBENCH software v.2.0.
Proteomics.
Tissue homogenization and sample preparation. For whole proteome analysis of each mouse genotype, individual half brains (n = 5 male mice per group) were homogenized using a glass homogenizer in urea-based lysis buffer (8 M urea, 100 mM Tris pH 8.5, 1% SDS, 5 mM EDTA, 1 mM AEBSF, 1 mM PMSF and 4 mM IAM) to obtain 10% w/v homogenates. A round of probe sonication (30% amp, 2s on/ 2s off, 2 min) was applied, followed by incubation in the dark on ice for 15 min. Lysates were clarified by centrifugation and total protein concentration of each sample was determined in the recovered supernatant by the BCA assay.
A total of 50 µg of protein from each brain lysate were processed using ProTrap XG cartridges [24] (Proteoform Scientific inc.) following the manufacturer’s protocol with some modifications. Briefly, 100 mM NaCl was added into each sample and the volume was adjusted to 100 µL. Proteins were precipitated in acetone (1:4 ratio) directly on the filtration cartridge at R.T. for 30 min. The cartridge was spun down at 2500 rcf for 2 min and protein pellet was washed once with acetone. The pellet was resuspended in 100 µL of 8 M urea by vortexing for 30 s, bath sonication for 10 min, and incubation at R.T. for 30 min. The urea in the samples was diluted by addition of 400 µL of 100 mM Tris buffer (pH 8). Proteins were reduced (10 mM DTT) and alkylated (25 mM iodoacetamide) at 37°C for 30 min, then 25 mM DTT was added. Digestion was initiated by addition of trypsin at a 50:1 (protein:enzyme) mass ratio. Samples were incubated at R.T. overnight. The reaction was then quenched by addition of trifluoroacetic acid (TFA) (final 2.5%). Peptides were desalted using a SPE column. The cartridge was primed (300 µL ACN), equilibrated (300 µL of 0.1% TFA in water), loaded twice, and washed (300 µL 5% ACN, 0.1% TFA in water). Peptides were eluted (300 µL of 50% ACN, 0.1% TFA in water) into a new tube and dried down using a speedvac and stored at -20°C until LC-MS/MS analysis.
Mass spectrometry and data analysis. The samples were analyzed using a nanoflow-HPLC (Thermo Scientific EASY-nLC 1200 System) coupled to an Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Fisher Scientific inc.) in data independent acquisition mode. Digested peptides were recovered in buffer A (3.9% ACN, 0.1% formic acid in water). Reverse phase separation of the peptides was done with an Aurora Ultimate™ analytical column (25 cm x 75 µm ID with 1.7 µm media, IonOpticks). Peptides were eluted with a solvent B gradient (0.1% FA in 80% ACN) for 120 min. The gradient was run at 400 nL/min with analytical column temperature set at 45°C. DIA analysis was done as reported by Mehta et al. 2022 with some modifications [25]. Full scan MS1 spectra (350–1400 m/z) were acquired with a resolution of 120,000 at 200 m/z with a normalized AGC Target of 200% and a maximum injection time of 20 ms. MS2 was acquired in the linear ion trap, ACG target value for fragment spectra was set to 2000%. Twenty-eight 38.5 m/z windows were used with an overlap of 1 m/z. Resolution was set to 30,000 using a dynamic maximum injection time and a minimum number of desired points across each peak set to 6.
DIA data analysis was performed in the software Spectronaut (v17) using direct DIA analysis workflow using default settings. The database for the searches was the Uniprot mouse proteome (2021, 55,336 sequences) with the hAPP and hCD33 sequence added for the Tg lines. Trypsin/P was selected as the digestion enzyme with a maximum of two missed tryptic cleavages and the search was performed with a maximum false discovery rate of 1% for peptides. Carbamidomethylation (C) was added as fixed modification and, deamidation at N/Q, and M oxidation were set as variable modifications. Before pairwise comparison of the populations under study, protein abundance variability was corrected by normalizing abundance using a global approach on the abundance average. Ratios for the comparisons, CD33M versus Control, CD33m versus Control, CD33M versus CD33m, were generated and a t-test was applied, the list of candidates was generated for significant proteins with p-value ≤ 0.05 and fold change ≥ 1.5 for each comparison. GO analysis for the selected candidates was performed directly in Spectronaut, over and underrepresented GO entries with p-value ≤ 0.05 were considered significant. Volcano plots between paired groups (CD33M versus Control, CD33m versus Control and CD33M versus CD33m) were generated on Prism (v9) using the list of proteins on each individual comparison. The triplot was generated using excel by initially graphing the log2(ratio) of each comparison on each side of a triangle. The intersection of lines between the specific fold change value and the vertex at the opposite end of the triangle corresponds to the proteomic correlation between the populations compared. The location of each circle on the graph indicates the association of that protein to the specific comparison. Identifications closer to the vertex, correspond to proteins with higher association to that genotype, while identifications closer to the center of the triangle indicate similar levels of that protein in all comparisons. From the lists of candidates generated, proteins with increased abundance to the CD33m and CD33M populations were extracted and highlighted in the triplot (blue and red, respectively).
Isolation of adult mouse microglia. Isolation of microglia was performed as described previously [26]. Briefly, mice were perfused with 15 mL of ice-cold HBSS buffer containing Actinomycin D (5 µg/mL) and Triptolide (10 µM). Brain was then extracted and stored in 10 mL storage buffer containing Actinomycin D (5 µg/mL), Anisomycin (27.1 µg/mL) and Triptolide (10 µM) at 4°C. Minced brains were homogenized in ice cold storage buffer with a 5 ml syringe plunger through a 40 µm filter (Corning) under sterile conditions. The cell suspension was then transferred to a 15 mL tube and centrifuged for 5 min at 4°C (300 rcf). After centrifugation, the pellet was collected and resuspended in 10 mL of ice cold 40% Percoll (Sigma) diluted in 1x (final) HBSS and centrifuged (30 min, 500 rcf, 4°C). After carefully removing the Percoll layer containing myelin debris, microglia were pelleted. After centrifuging (5 min, 300 rcf, 4°C), the collected cell pellet was washed with 10 mL of ice-cold 1x HBSS buffer. Samples were resuspended in 50 µl of ice-cold flow buffer (0.5% BSA, 1 mM EDTA, in 1x PBS, Sterile Filtered) containing antibodies targeting Cd11b (BV510, clone ICRF44, Biolegend), CD45 (APC/Cy7, clone 30-F11, Biolegend), and Cx3cr1 (PerCP/Cy5.5, clone SA011F11, Biolegend) from Biolegend all at 1:200 dilution for 20 min on ice. Following incubation, the samples were washed with ice cold flow buffer, centrifuged (5 min, 300 rcf, 4°C), and resuspended in 700 µl of ice-cold cell sorting buffer (1x HBSS containing 10% FBS and 1 mM EDTA) in preparation for cell sorting. An estimated 70,000 microglia (CD11b+, CD45+, Cx3cr1+) were sorted using the 100 µm nozzle at a sorting speed of approximately 3500 events/sec. We collected sorted samples in 1.7 ml Eppendorf tubes. Sorted cells were centrifuged (300 rcf, 5 min, 4°C) and the supernatant was removed. Pelleted cells were resuspended in 100 µL PBS + 0.1% BSA and 10µL of the sample immediately counted using a Neubauer chamber using 0.4% Trypan blue solution (Thermo Fisher). For experiments, samples with a viability of over 95% were used. Cells were resuspended in PBS in an appropriate volume to achieve a concentration of 1000 cells/mL. This cell suspension was used to generate the gel-beads + cell emulsion by the 10X Chromium Controller (PN-1000202) using the Chromium Next GEM Single Cell 3′ GEM, Library & Gel Bead Kit v3.1 (PN-1000121), Chromium Next GEM Chip G Single Cell Kit (PN-1000120) and Single Index Kit T Set A, (PN-1000213). Reverse transcription, cDNA amplification, library preparation, and sample barcoding were performed following the available manufacturer’s protocol. Finally, sample libraries were pooled and sequenced in Illumina HiSeq P150 (Sequencing type: Paired-end, single indexing) to an average depth of ~ 50,000 reads per cell.
Single-cell RNA sequencing.
Library preparation. The FACS-isolated cells were processed on the 10x chromium controller following the 10x Genomics Next GEM Single Cell 3’ GEM, library, and v3.1 Gel Bead kit (10x Genomics; Cat. No. 1000121) and sequenced the samples on the Illumina HiSeq P150 sequencer at Novogene corporation inc. The samples were paired-end, single index sequenced at an average read depth of 50,000 reads per cell. The resulting BCL files were demultiplexed into FASTQ files and aligned to a custom Mus Musculus 10 (MM10) reference genome, adjusted to include Clec7a, a polymorphic pseudogene. The Clec7a annotation was added to the 10x Genomics pre-built MM10 genome (2020-A) GTF file from the 10x Genomics parent GTF file (gencode.vM23.primary_assembly.annotation.gtf.gz). The final custom genome was created by combining the MM10 genome (2020-A) FASTA file with the Clec7a modified GTF file using the cellranger mkref function in the 10x cell ranger pipeline (v3.0.0). Finally, the samples were aligned the custom genome using the cellranger count function to generate barcoded and sparse matrices, both raw and filtered, along with BAM files for downstream analyses.
Quality control, dimensionality reduction and clustering. Quality control, dimensionality reduction and initial clustering was performed in the R statistical environment (v4.1.2) using Seurat (v4.0; https://github.com/satijalab/seurat). A Seurat object was created per dataset to include genes expressed in a minimum of 3 cells and cells expressing a minimum of 200 genes using the CreateSeuratObject() function. The object was further refined to remove doublets and multiplets by removing cells with high gene counts (> 3000 genes) and dead cell by removing cells high percentages of mitochondrial genes (> 10%). All datasets were merged using the merge() function and normalized using the SCTransform() function according to the binomial regression model (Highly variable features = 3000, nCount and mitochondrial genes regressed).
Dimensionality reduction was performed using RunPCA(), FindNeighbors() (Dimensions = 15) and FindClusters() functions. 25 PCs were used for downstream analyses as determined by the PCA elbow plot. The FindClusters() function was run at multiple resolutions, ranging from 0 to 1, separated by 0.1. All clustering resolutions (0 to 1) were plotted on a tree generated by the Clustree package using the clustree() function. The 0.5 resolution was chosen for clustering based on the most stable level as identified by clustree().
The final clustering of the dataset was performed in a Jupyter notebook (v6.0.3) running a python environment (Python 3.8.3) and using Single Cell Clustering Assessment Framework (SCCAF) package (v0.0.10, https://github.com/SCCAF/sccaf). The SeuratObject was converted to an h5ad file using SeuratDisk by first converted the RDS file to an h5Seurat file using the Saveh5Seurat() and converting the h5Seurat file to an h5ad file using the convert() function (Destination = h5ad). The h5ad file was read into the python environment using Scanpy (v1.6.0)[27]. The clustering was refined using the SCCAF_optimize_all() function (minimum accuracy = 95%, iterations = 150). The machine learning algorithm iteratively clustered the dataset until a 95% self-projection accuracy was reached. The final clustering iteration was projected onto a UMAP using the sc.pl.umap() function.
Open field assay. All mice were habituated in the behavior suite for 1 hr prior to testing. White noise was played via a sound device during habituation and during the experiment. Two white cube-shaped (40 cm x 40 cm) arenas, placed side-by-side were used for this experiment, enabling recording of two mice at a time. One mouse was placed in each arena for testing. Each mouse’s behavior and movement in a 15 min period were recorded and tracked via a camera connected to EthoVision 17 (Noldus, Wageningen, the Netherlands). Tracking was dependent on the mouse’s center-point as detected by EthoVision 17, with the nose- and tail-points being defined too. The center zone was defined in EthoVision 17 as a 30 cm x 30 cm area in the middle of the arena. After testing, the mouse was placed back into its home cage, and the arenas were thoroughly cleaned with 70% ethanol prior to testing the next mouse. Differences between the groups were tested with one-way ANOVA, followed by Tukey’s Test.
Statistical analyses. Data represented as mean ± SD. The D'Agostino-Pearson normality test was used to test for Gaussian distribution of datasets. Differences between the groups were evaluated with one-way Anova followed by the Holm–Sidak Test. A probability of P < 0.05 was considered indicative of significant differences between groups.