Antisense oligonucleotides

All ASOs were synthesized (Axolabs GmbH) as 20-mers with the 5 outer nucleosides on either end modified at the 2’ position of the ribose with methoxyethyl groups (2’MOE). The central 10 nucleosides were unmodified deoxynucleosides. All cytosines were methylated, and all internucleoside linkages were phosphorothioate. A total of 278 and 235 ASOs were synthesized to target APOE and TREM2 pre-mRNA, including exonic and intronic regions, respectively. The lyophilized ASOs were reconstituted in phosphate-buffered saline (PBS) (Sigma-Aldrich, D8537), filtered through a 20 µm filter, and their final concentration was determined by measuring their absorbance at 260 nm. Multiple batches were synthesized. The lead ASO sequences are provided in Supplemental Table 1. To determine possible off-targets of the selected lead candidate ASOs, we used Basic Local Alignment Search Tool (blastn) analysis. Lead ASOs targeting APOE and TREM2 did not overlap to any other human transcript (Suppl Table 1).

Cultures of human cell lines

All immortalized cell lines were obtained from ATCC unless listed otherwise. THP-1 cells (TIB-202, CVCL_0006) were cultured in RPMI-1640 medium (Sigma-Aldrich, R0883) supplemented with 2 mM Glutamax (ThermoFisher, 35,050–038), 25 mM HEPES (ThermoFisher, 15,630–056), 10% (v/v) fetal bovine serum (FBS) (Biowest, S1810-500) and 50 µg/mL Gentamycin (ThermoFisher, 15,750–045). K-562 cells (CCL-243, CVCL_0004) were cultured in RPMI-1640 medium supplemented with 10% FBS, 2mM L-Glutamine (Sigma, G8540) and 50 µg/mL Gentamycin. SH-SY5Y cells (CRL-2266, CVCL_0019) were maintained in DMEM/F12 (Sigma, D9785) supplemented with 10% FBS, 0.1 mM NEAA (Gibco, 11,140,050) and 50 µg/mL Gentamycin. SK-N-MC cells (HTB-10, CVCL_0530) were maintained in MEM medium (Gibco, 41,090,028) supplemented with 10% (v/v) FBS, 1 mM Sodium Pyruvate (Gibco, 11,360,070), 17.9mM sodium bicarbonate (Sigma, S-5761), 0.1mM NEEA and 50 µg/mL Gentamycin. Kelly cells were obtained from DSMZ (ACC 355, CVCL_2092) and were cultured in RPMI-1640 medium with 10% (v/v) FBS, 2 mM L-Glutamine and 50 µg/mL Gentamycin. HCT-116 cells (CCL-247, CVCL_0291) were cultured in McCoy's 5A medium (Gibco, 16,600,082) with 10% (v/v) FBS, 100 U/ml Penicillin and 100 µg/ml Streptomycin (ThermoFisher, 15,140–122). Caco2 cells (HTB-37, CVCL_0025) were cultured in MEM medium supplemented with 20% (v/v) FBS, 0.1mM NEAA, 1 mM Sodium Pyruvate, 2 mM L-Glutamine and 50 µg/mL Gentamycin. SK-MEL1 cells were obtained from DSMZ (ACC 303, CVCL_0068) and cultured in DMEM (Gibco, 10,938,025) with 15% (v/v) FBS, 2 mM L-Glutamine and 50 µg/mL Gentamycin. T-46D cells (HTB-133, CVCL_0553) were maintained in RPMI-1640 medium with 10% (v/v) FBS, 2 mM L-Glutamine, 10 µg/mL insulin (Sigma, I9278), 100 U/ml Penicillin and 100 µg/ml Streptomycin. All cell-lines were split bi-weekly and were sub-plated in non-coated 96-well µclear plates (Greiner, 655,090) between 20,000–40,000 cells/well.

Human iPSC-derived microglia (iMGL)

The iPSC lines used in the cellular assays were obtained from the EBiSC consortium (https://ebisc.org) (Suppl Table 2). UKBi011-A (APOE ε4/ε4; CVCL_LE34), UKBi011-A-3 (APOE ε3/3; CVCL_RX83), and UKBi011-A-1 (APOE KO/KO; CVCL_RM82) were used to assess the pharmacology of APOE ASOs in vitro. The human isogenic TREM2 iPSC lines BIONi010-C (parental; CVCL_1E68), BIONi010-C-7 (TREM2 R47H/R47H; CVCL_II86), and BIONi010-C-17 (TREM2 KO/KO; CVCL_RM88) were used to evaluate the activity of TREM2 ASOs in vitro. For off-target analysis using microarray the BIONi010-C (parental; CVCL_1E68) line was used. An additional wild-type iPSC line was included for ASO uptake validation (SIGi001-A; CVCL_EE38).

Human iPSC-derived microglia (iMGL) were differentiated using a method described by Cowley et al. [54, 55]. Briefly, iPSCs were dissociated with TrypLE select (ThermoFisher, 12,563,011) into single cell suspension and embryoid bodies (EBs) were produced in AggreWell800 (Stem Cell Technologies, 34,811) in mTeSR medium (Stemcell Technologgies, 85,850) supplemented with 50 ng/ml BMP-4 (Invitrogen, PHC9534), 50 ng/ml VEGF (PeproTech, 100–20) and 20 ng/ml SCF (Miltenyi, 130–096-693). The medium was refreshed daily by 75% for 3 consecutive days. Next, EBs were transferred to factory plates to create precursor factories by placing 10–20 EBs per well in X-VIVO 15 medium (Lonza, BE02-060F) containing 100 U/ml Penicillin and 100 µg/ml Streptomycin (ThermoFisher, 15,140–122), 2 mM Glutamax supplement, 0.05 mM 2-mercaptoethanol (ThermoFisher, 31,350–010), 0.1 µg/ml M-CSF (ThermoFisher, PHC9501) and 0.025 µg/ml IL-3 (ThermoFisher, PHC0031). The factories were cultured for 6–8 weeks (with weekly 50% medium change) before the cells were harvested for final differentiation into microglia. Microglia precursors were harvested for no more than once a week. At harvest, microglia precursors were plated at a density of 15,000 cells/well in non-coated 96-well µclear plates (Greiner, 655,090) and cultured in advanced DMEM/F12 medium (ThermoFisher, 12,634–010) containing 100 U/ml Penicillin and 100 µg/ml Streptomycin, 2 mM Glutamax supplement, 0.05 mM 2-mercaptoethanol, 10 ng/ml GM-CSF (ThermoFisher, PHC2015) and 100 ng/ml IL-34 (Peprotech, 200–34) for maximally 14 days (with 50% medium change every 2–3 days).

RNA isolation and real-time quantitative PCR

Total cellular RNA was isolated using the RNeasy96 kit (Qiagen, 74,182) according to manufacturer’s protocol. Briefly, cells were lysed with RLT buffer at RT and an equal volume of 70% (v/v) ethanol was added. The mixture was pipetted 5 times and transferred to columns in which the RNA was bound to the filter by a vacuum system (Qiagen). The RNA was washed sequentially with the RW1 and RPE buffers provided in the kit. After the final RPE washing step, a centrifugation was applied to make sure any residual buffer was eliminated (6 min at 5,600 × g at room temperature (RT). The RNA was then eluted with RNase-free water by centrifugation at 5,600 × g at RT for 4 min. The RNA concentration was determined by spectroscopy using a Nanodrop (ThermoFisher, ND-8000).

Equal amounts of RNA were reverse transcribed using the high-capacity cDNA reverse transcription kit (ThermoFisher, 4,368,813) according to manufacturer’s instructions on a thermocycler using the following incubation steps: 10 min at 25 °C, 2 h at 37 °C, and 85 °C for 5 min.

Realtime-quantitative PCR (RT-qPCR) was performed on diluted cDNA mixed with PowerUp SYBR Green master mix (ThermoFisher, A25743) or SensiFast Sybr No Rox Mix (Meridan Bioscience, bio-98020) and primers at a final concentration of 500 nM. Exceptionally, human MALAT1 and SMN levels were measured with a Taqman assay mixed with PrimeTime master mix (IDT DNA Technologies, 1,072,102). Primers amplifying regions of various human housekeeping genes were used to quantify and normalize the mRNA expression levels. We designed 4 different primer assays to quantify APOE and TREM2 mRNA, each spanning an exon-exon boundary along the mature transcript (Suppl Fig. 1A,B, Suppl Table 3). All primers were purchased from IDT DNA Technologies are listed in Supplemental Table 3. Realtime detection of the mRNA quantities was done on a QuantStudio 12k Flex (ThermoFisher) or Quantstudio 7Pro (ThermoFisher) in 384 well plates using standard settings and with the following cycling parameters: 2 min at 50°C and 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and 1 min at 60 °C. For SYBR green primers, a melt curve was included at the end of the run for 15 s at 95 °C, 1 min at 60 °C and 15 s at 95 °C. Alternatively, detection was done on a LightCycler 480 (Roche) with following parameters: initial denaturation (95°C, 10 min), 40 amplification cycles of denaturation (95°C, 10 s) followed by annealing and extension (60°C, 30 s), and a cooling step (40°C, 15 s).

RT-qPCR data were analyzed in qBase + (Biogazelle, version 3.2). First, at least 2 out of 8 most stable housekeeping genes were identified by a GeNorm analysis [56]. Next, target gene expression data were normalized to the geometric mean of the 2 most stable housekeeping genes and calibrated to a control condition. Expression data are expressed as calibrated normalized relative quantities (CNRQ) except when mentioned otherwise.

Homogeneous Time-Resolved Fluorescence (HTRF) analysis of APOE and TREM2 protein levels

APOE and TREM2 levels were measured using an HTRF assay (Cisbio, 63ADK004PEH and 63ADK099PEH, respectively) following manufacturer’s instructions. Briefly, conditioned medium was collected prior to the ASO treatment and 48h-96h-7d post-treatment. The cells were lysed 7 days post treatment in 50 µl/well RIPA buffer (Sigma, R0278) supplemented with phosphatase inhibitors (PhosSTOP; Roche, 4,906,837,001) and protease inhibitors (Complete mini EDTA-free; Roche, 11,836,170,001) according to manufacturer’s instructions. Medium samples were diluted 1:4 for APOE and used undiluted for TREM2, where the cell lysates were diluted 1:2. Samples were incubated for 3 h at RT (APOE) or overnight at 4°C (TREM2) with a mixture of Europium Cryptate (donor) and d2 (acceptor) labelled antibodies following manufacturer’s instructions. Fluorescence was measured on an Envision (Perkin Elmer) at 665 nm and 615 nm (cycle 16,600 µs, delay 50 µs, # of flashes 50) and the HTRF ratio was calculated by multiplying the ratio of the signal at 665 nm over the signal at 615 nm by 10,000. The HTRF Δ ratio was calculated by the ratio of the samples minus the ratio of the background signal (unconditioned medium negative control). Fluorescence signal obtained from serially diluted recombinant APOE and TREM2 was used to interpolate the concentration of APOE and TREM2, respectively, and normalized for the dilution factor. The final data are expressed as ng/mL APOE and pg/mL TREM2, respectively.

Fluorescence microscopy

For live-cell imaging using an IncuCyte Zoom (Essenbioscience/Satorius, software version 2016B), we treated cells with an Alexa Fluor-555 labelled MALAT1 ASO and imaging was started approximately 5 min after addition of the ASOs using a 10 × or 20 × objective at 37 °C and 5% CO2. Images were obtained at frequent time points over time. Images/videos were used in a qualitative manner.

For live-cell imaging using an Opera Phenix High-Content Screening System (Perkin Elmer, HH14000000), cells were incubated with a CellMask™ Deep Red Plasma membrane dye (1:10.000, Thermofisher, C10046), LysoTracker™ Green dye (1:20.000, Thermofisher, L7526) and Hoechst 33,342 (1:10.000, Thermofisher, H3570). After 15 min the fluorescently labelled MALAT1 ASO (custom made ASOs (IDT DNA) with Alexa 594 label on the 3 or 5 end) was added. Immediately after the addition of ASO, imaging was initiated using a 20 × or 40 × water immersion objective at 37 °C and 5% CO2. At least 11 fields per well with a Z-stack of at least 4 planes were imaged per condition. Images were taken every 20 min for the first 2 h, followed by image acquisition every 2 h for (until 24 h), and finally one image on day 2 and on day 3.

For immunocytochemical analysis, cells were fixed using 4% (w/v) paraformaldehyde (Pierce, 28,908) dissolved in PBS (Sigma, #D8537-500ML) for 15 min at RT. Cells were blocked and permeabilized with PBS supplemented with 0.1% (v/v) Triton X-100 (Sigma, 1,086,031,000) (PBS-T) and 5% (v/v) normal goat serum (Sigma Aldrich, G9023) for 1.5 h at RT. Washing steps were performed using PBS-T. Primary and secondary antibodies were diluted in PBS-T. A list of primary and secondary antibodies used can be found in Supplemental Table 4 and 5, respectively. Imaging was performed using an Opera Phenix High-Content Screening System (Perkin Elmer, HH14000000) using a 40 × water immersion objective, imaging at least 9 fields per well with a Z-stack consisting of at least 4 planes. If images were used in a quantitative manner, the analysis was performed using the Opera Phenix Harmony software (Perkin Elmer, version 4.9) using the maximum intensity projections. Fluorescence data were normalized to the number of nuclei.

Hybridization ELISA

Cells and tissues were lysed in RIPA buffer (Sigma, R0278) supplemented with phosphatase (PhosSTOP; Roche, 4,906,837,001) and protease (Complete mini EDTA-free; Roche, 11,836,170,001) inhibitors. Cells were lysed at a ratio of 1 µl per 300 cells while tissues were lysed in 1 ml per 250 mg wet tissue weight. Tissue was homogenized in Lysing Matrix D (MP Biomedicals, 6913–500) using a FastPrep-24 5G (MP Biomedicals) with 3 cycles of 20 s at a speed of 6.0 m/s with 5 min between each cycle on ice to avoid sample heating. Lysates were then centrifuged for 5 min at 20,000 × g and 4 °C to remove debris.

Samples were hybridized with custom HPLC-purified probes (IDT DNA) consisting of unmodified 20-mer DNA sequence complementary to the ASO sequence. The probes contained a 5’ end digoxigenin label and a 3’ end tetraethylene glycol (TEG) spacer and biotin label. Each sample was hybridized with 2 nM probe diluted in hybridization buffer containing SSPE (Thermofisher, 15,591,043) and 0.4% (v/v) Tween-20 (Sigma, P7949). Hybridization was performed at 75 °C for 1 h. Serially diluted ASO with known amounts were used to define a standard curve.

The hybridized material was then spotted on neutravidin-coated plates (Thermofisher, 15,217) for 1h at RT. Plates were washed 2 times with Tris-buffered saline supplemented with 0.05% (v/v) Tween-20 (TBS-T, Sigma-Aldrich, T9039). Non-hybridized probe or any mismatched probe: ASO pair was destroyed by treating samples for 2 h with 300 U/ml S1 nuclease (Invitrogen, 18,001–016) diluted in S1 nuclease solution consisting of 300 mM sodium acetate, 10 mM Zn (CH3COO)2*2H2O, 50% (v/v) glycerol, pH 4.6. Each sample was then washed with TBS-T two times to remove non-hybridized material. Next, samples were incubated for 30 min at RT with an anti-digoxigenin antibody conjugated to alkaline phosphatase (Roche, 11,093,274,910) diluted in Super block buffer (Thermofisher, 37,536) containing 1 M sodium chloride (Promega, V4221). After 2 washes with TBS-T, ATToPhos substrate (Promega, S1000) was added to each sample and incubated at RT for 10 min. The reaction was stopped by adding 1 M sodium phosphate (J.T. Baker, 3824–01). After 15 min at RT, plates were read on an Envision (PerkinElmer) at an excitation of 423 nm and emission of 568 nm. Raw fluorescence data were exported and samples with unknown ASO levels were extrapolated based on the relative fluorescence units (RFU) and the known ASO concentration from the standard curve.

Transcriptome analysis by microarray

RNA was extracted in a similar manner as described above. The RNA concentration was determined by spectroscopy using a Nanodrop (ThermoFisher, ND-8000). Amplification and labelling of total RNA were performed using the GeneChip® PICO Reagent Kit following the manufacturer’s protocol (ThermoFisher 2016, P/N 902790). Biotin-labeled target samples were hybridized to the Clariom™ GO Screen containing probes for 19,409 protein-coding genes. Target hybridization was processed on the GeneTitan® Instrument according to manufacturer’s instructions provided for Expression Array Plates (P/N 952361). Images were analyzed using the GeneChip® Command Console Software (GCC) (ThermoFisher). Microarray data were processed using the statistical computing R-program package limma (Version 3.42.0) and Bioconductor tools [57]. The gene expression values were normalized using Robust Multi-array Average (RMA) [58]. Individual probes were grouped into gene-specific probe sets based on Entrez Gene using the metadata package goscreenhuhsentrezg (version 25.0.0) [59].

Electrophysiology

All experimental protocols were approved by the ethical committee of Janssen Pharmaceutica N.V. Primary neurons were freshly dissociated from embryonic E18-19 rat cortices as described previously [60] and plated onto 48-well MEA plates (Maestro Pro-167, Axion Biosystems). One day before plating the cells, each 48-well MEA plate was pre-coated with a polyethyleneimine (PEI) (0.1%) solution (Sigma, P3143), washed four times with sterile distilled water, and then allowed to dry overnight. On the day of plating, Laminin (20 μg/ml) (Sigma, L2020) was added to each 48-well plate which was then incubated for 1 h at 37 °C. Thereafter the neurons were cultured at 37 °C, 5% CO2 in Neurobasal medium (Thermofisher, 21,103–049) supplemented with 0.5 mM L-Glutamine (Thermofisher, 25,030,149) and 2% B27 (Thermofisher, 17,504,044).

Data analysis was performed using AxIs software (version 3.6.1.14, Axion Biosystems Inc.). The threshold for the spike detection was ≥ 5.2 × the standard deviation of the RMS (root mean square) noise. Statistical analysis consisted of expressing the treatment ratio of exposed wells (percentage change between the baseline and the treatment) normalized to the treatment ratio in control experiments. Normalized treatment ratios of n = 8 wells were averaged per condition. Each well of the MEA served as its control, and the changes in electrical activity elicited by the treatments were expressed as a percent of that control activity and normalized to the wells treated with the vehicle control PBS. The final concentration of PBS added to each well was 0.1% (1 μl/ml), which did not alter the pH or the ionic concentration of the medium.

Human whole blood sampling

Human whole blood samples were collected into Sodium-Heparin BD Vacutainer of 4 healthy donors through voluntary blood donation approved by the Commission of Medical ethics of the ZNA according to European Union guidelines. The donors signed an informed consent and the collected samples were anonymized, and no further personal data was collected. The samples were diluted 1/10 in RPMI-1640 medium (Sigma-Aldrich, R0883) with 10% (v/v) FBS (Biowest, S1810-500), 100 U/ml Penicillin and 100 µg/ml Streptomycin (ThermoFisher, 15,140–122) and plated in non-coated 96-well µclear plates (Greiner, 655,090). The ASOs and positive controls (Toll-like receptor (TLR)-7 and TLR-8 agonists; CAS ID: 1,642,857–69-9 and 144,875–48-9) were added to the blood immediately after plating at 3 different concentrations (1, 2.5 and 10 µM). One day later, samples were collected by a centrifugation at 100 × g for 3 min. Supernatants were transferred to a 96-well PCR plate (ThermoFisher, AB0600), centrifuged at 2,500 × g for 2 min and transferred to a 96-well ½ Area OptiPlate (Perkin Elmer, 6,002,290) for storage at -80 °C until further use.

Cytokine and chemokine profiling

Levels of interleukins (IL) IL-1β, IL-6, IL-8, IL-10, IL-12p70, interferon gamma (IFN-γ) and tumor necrosis factor alpha (TNF-α) in culture medium and human whole blood were determined simultaneously using a commercially available quantitative electrochemiluminescence assay (Meso Scale Diagnostics, K15008B-2). A commercially available GAPDH electrochemiluminescent assay (Meso Scale Diagnostics, K151PWD-2) was used to quantify GAPDH protein levels in cell lysates as loading control. The assays were performed following manufacturer’s recommendations. Briefly, medium or blood samples were diluted 1:2 in unconditioned medium or dilution buffer respectively and cell lysates were diluted 1:10. MSD plates were blocked with blocking buffer (MSD kit) for 1 h at RT and washed 3 × with washing buffer (PBS + 0.05% Tween-20). For the 7-plex cytokine assay a serially diluted calibrator was used to interpolate unknown values. Samples and calibrators were added and incubated for 2 h at RT after the detection antibody was added and incubated for 2 h at RT. The plates were washed 3 × with washing buffer and 2X read buffer T was added. For the GAPDH assay, samples were added and incubated for 1 h at RT. The plates were washed 3 × with washing buffer and the detection antibody was added for 1 h at RT. After washing 3 × with washing buffer 1 × read buffer T was added. Immediately after the read buffer was added, plates were measured using an MSD plate reader (Meso Scale Diagnostics, Sector Imager 6000). iMGL treated with Lipopolysaccharide (LPS) (Thermofisher, 00–4976-03) at 100 ng/mL (w/v) or whole blood treated with TLR-7 and TLR-8 agonists at 10 µM were used as a positive control to induce pro-inflammatory cytokine release. Results were analyzed using MSD DISCOVERY WORKBENCH software (Meso Scale Diagnostics, version 4.0.12). For the cytokine release analysis, the unknown samples were interpolated to the standard curve. GAPDH values were normalized to the untreated samples and further used as a normalization factor for the cytokine release measures.

Transgenic mice

Homozygous AppNL−G−F[61] mice on the C57BL/6J background [62] were crossed with Rag2−/− IL2rγ−/− hCSF1KI mice purchased from The Jackson Laboratory (strain #017708) to generate AppNL−G−F mice on a background allowing transplantation and survival of human microglia [48, 53, 63]. Mice were bred and housed in local facilities at KU Leuven under a 14h light and 10h night cycle with food and water ad libitum. All experiments were approved by Ethical Committee of Laboratory Animals of the KU Leuven according to local (ECD project number P177/2017) and European Union guidelines.

Xenotransplantation of human microglia

Generation of microglia from human embryonic stem cells (hESC) was based on the previously described MIGRATE protocol [53]. First, the hESC H9 cell line (WA09 WiCell Research Institute, CVCL 9773) was seeded and induced towards EB formation, by using mTeSR1 media (Stemcell Technologies, 85850) supplemented with 50ng/ml BMP4 (PeproTech, 125–05), 50ng/ml VEGF (PeproTech, 100–20) and 20ng/ml SCF (PeproTech 300–07). Next, the generation of hematopoietic cells was initiated by culturing EBs in X-VIVO 15 media supplemented with 50 ng/ml SCF (PeproTech 300–07), 50 ng/ml M-CSF (PeproTech, 300–25), 50 ng/ml IL-3 (PeproTech, 200–03), 50 ng/ml Flt3-ligand (PeproTech, 300–19) and 5 ng/ml TPO (PeproTech, 300–18). Lastly, EBs were cultured in X-VIVO 15 media supplemented with 50 ng/ml Flt-3-ligand (PeproTech, 300–19), 50 ng/ml M-CSF (PeproTech, 300–25) and 25 ng/ml GM-CSF (PeproTech, 300–03) for the induction of the final myeloid lineage producing progenitor cells.

For in vitro experiments, progenitor cells were collected on day 25 and differentiated into microglia in differentiation medium described in Abud et al [64] (TIC; DMEM/F12, glutamine (2 mM) (ThermoFisher, 10,565,018), N-acetylcysteine (5 μg/ml) (Sigma-Aldrich, A7250), insulin (1:2 000) (Sigma-Aldrich, I9278), Apo-Transferrin (100 μg/ml) (Sigma-Aldrich, T1147), sodium selenite (100 ng/ml) (Sigma-Aldrich, S9133), cholesterol (1.5 μg/ml) (Sigma-Aldrich, C4951) and heparan sulfate (1 μg/ml) (Ams Bio, GAG-HS01) supplemented with 50 ng/ml IL-34 (PeproTech, 200–34), 50 ng/ml M-CSF (PeproTech, 300–25), 10 ng/ml CX3CL1 (Peprotech, 300–31) and 25 ng/ml TGF-β (PeproTech, 100–21)) for 7 days, with half of the media changed every other day. Cells were cultured in a humidified incubator at 37 °C and 5% CO2.

For xenotransplantation experiments, progenitor cells were harvested on day 18. Prior to transplantation, newborn mice were intraperitoneally injected with CSF1R inhibitor BLZ945 (Seleck, S7725) at a dose of 200 mg/kg body weight for 2 consecutive days to deplete endogenous mouse microglia. Subsequently, progenitor cells were collected at a final concentration of 250,000 cells/µl in PBS. Four days old pups were anesthetized by hypothermia and injected bilaterally with 1 µl of cell suspension using the following coordinates from bregma: anteroposterior, − 1 mm; mediolateral, ± 1 mm. Mice were allowed to recover on a heating pad at 37°C and were then transferred back to their cage.

Intracerebroventricular delivery of ASOs

ASO stock solution (50 mg/ml) was diluted in PBS under sterile conditions and administered via intracerebroventricular (icv) injection as previously described [65]. Briefly, mice were anesthetized by continuous isoflurane inhalation and injected bilaterally with 2.5 µl of vehicle (PBS) or ASO (dose ranging from 3 to 125 µg) using the following coordinates from bregma: anteroposterior, − 0.22 mm; mediolateral, ± 1 mm; dorsoventral, -2.75 mm. The injection was done over 2 min and the needle was held in place for 5 min after the infusion. Mice were allowed to recover on a heating pad at 37°C and were then transferred back to the cage. Mice were monitored daily and were sacrificed 1 or 4 weeks after the injection.

Tissue harvest and isolation of human and mouse microglia from mouse brain

Animals were sacrificed by overdose with sodium pentobarbital (Dolethal, Vetoquinal) and transcardially perfused with heparinized ice-cold PBS (5 U.I./ml heparin (LEO, 030710) in PBS). Liver, kidney, and spinal cord were immediately collected and snap frozen in liquid nitrogen. After removing cerebellum and brainstem, brains were dissected into three pieces; one was snap frozen in liquid nitrogen for bio-analysis and HTRF assays, a second one was fixed in 4% PFA (Sigma Aldrich, 1.00496.5000) for immunohistochemistry (IHC), and the third one was collected in FACS buffer (PBS, 2% FBS (Life Technologies, 10,270,106), 2 mM EDTA (Sigma-Aldrich, E7889) for further tissue homogenization and microglia isolation.

For the isolation of human and mouse microglia, the brains were collected as described above and were enzymatically dissociated using the Neuronal Dissociation Kit (P) (Miltenyi, 130–092-628) following manufacturer’s instructions. Actinomycin D (Sigma-Aldrich, A1410) was added to solutions used from tissue collection to myelin removal step, at a final concentration of 5uM. Dissociated samples were filtered through a cell strainer (70-µm mesh) and centrifuged at 300g for 15 min at 4°C. Pellets were resuspended in a 30% Percoll solution (Merck, 17–5445-02) in FACS buffer and centrifuged at 300 × g for 15 min at 4 °C. Myelin layer and debri were carefully removed and the supernatant was discarded. Then, the cell pellet enriched in microglia was washed with FACS buffer and incubated with FcR blocking solution (Miltenyi, 130–092-575) according to manufacturer’s instructions. After the blocking step, cells were washed with FACS buffer and incubated with the following antibodies for 30 min at 4°C: anti-CD11b (1:50, Milteny, 130–113-806), anti-hCD45 (1:50, BD Bioscience, 555485), anti-mCD45 (1:200, BD Bioscience, 563890). E780 (1:1000, Thermo Fisher, 65–0865-14) was used as viability marker. Human and mouse microglia were sorted as CD11b + hCD45 + and CD11b + mCD45 + cells, respectively, using the Miltenyi MACS Quant Tyto cell sorter (Suppl Fig. 2). Sorted cells were pelleted and stored at -80°C.

Reanalysis of Mancuso et al. data

Data from Mancuso et al., 2022 [22] was used to compare the transcriptomic impact of the ASO treatments to the profile of TREM2 and APOE-KO human microglia in vivo. Single-cell data of xenografted microglia in 6-month-old Apphu/hu and AppNL−G−F mice were extracted from the dataset. The iPSC lines used were: UKBi011-A-3 (APOE ε3/3; CVCL_RX83), UKBi011-A-1 (APOE KO/KO; CVCL_RM82), BIONi010-C (APOE ε3/KO; CVCL_1E68) and BIONi010-C-3 (APOE KO/KO; CVCL_II82) from the EBiSC consortium (https://ebisc.org), and hESC H9 (WA09 WiCell Research Institute, CVCL 9773) and H9-TREM2 KO [66]. Mice engrafted with hESC H9 wild-type-derived cells were used to produce Suppl Fig. 10A-B. Mice engrafted with TREM2-KO or APOE-KO cells (and respective isogenic controls) were used for differential expression analysis to produce Fig. 5. Cell states annotations and marker genes were assigned according to the original manuscript. Differentially expressed genes were found by applying the FindMarkers() function from the Seurat R package for side-by-side comparisons of transplanted cell lines. The comparisons were performed with the following parameters: assay = "SCT", test.use = "wilcox", min.pct = 0, logfc.threshold = 0. We used the Wilcoxon rank sum test to calculate P-values. We performed DE on the "SCT" assay calculated from SCTransform(), since Pearson residuals resulting from regularized negative binomial regression effectively mitigate depth-dependent differences in differential expression, as described by [67]. We tested all genes without applying thresholds on average fold-change between conditions or minimum fraction of cells expressing the genes. Only genes with their adjusted P value < 0.05 (post-hoc, Bonferroni correction) were considered as significant.

RNA seq data analysis

Total RNA extraction from sorted human microglia followed a methodology similar to the one detailed earlier. The study encompassed six distinct experimental groups. For each treatment and vehicle group, sorted human microglia were gathered from four animals. Consequently, the RNAseq analysis involved a total of 24 animals distributed across six experimental categories: 1-week treatment of APOE ASO-1, 4-week treatment of APOE ASO-1, 1-week treatment of TREM2 ASO-171, 4-week treatment of TREM2 ASO-171, and the respective vehicle groups for both 1 week and 4 weeks. RNA samples were then sent to BGI (bgi.com) for library preparation and sequencing. Samples were sequenced on their DNBSEQ-G400 platform. Sequencing data were mapped using STAR version = 2.7.10 against the joined reference library by combining the Mus musculus reference genome (mm10/GRCm38) and the human genome (GRCh38). The full count matrix was produced by FeatureCount v1.6.3 from the Subread package [68], using reads mapped to the human genome. We conducted differential expression analysis comparing the treatment and vehicle groups using the DEseq package [69], which streamlines standard differential expression analysis procedures in a single function DEseq. In this analysis, a design matrix was constructed for each treatment group, with the corresponding duration-matched vehicle group as the reference group. We then employed the Wald test to ascertain the log2fold changes and associated p-values for the comparison. To control for multiple testing, we applied the false discovery rate (FDR) adjustment using DEseq2's default Benjamini–Hochberg procedure.

Microglial subtype gene set enrichment analysis

Subtype-signature genes are selected from the top 50 most significantly upregulated genes (adjusted p-value < 0.05) compared to other sub-types by Mancuso et al. 2022 [22]. Enrichment scores are calculated by using the Gene Set Enrichment Analysis Preranked module [69] for these gene sets and tested among all genes detected in isolated microglial cells, sorted from up to downregulated genes based on the log-fold changes calculated from differential expression analysis between treatment and vehicle groups (described in previous section and Suppl Fig. 12A-D). Significant GSEA enrichment was denoted by FWER < 0.05.

Histological analysis

Brains were fixed in 4% PFA solution overnight. Serial coronal sections (35 um thick) were cut with a vibrating microtome and collected in free-floating conditions. Sections were kept at -20°C in cryoprotectant solution (PBS supplemented with 30% ethylene glycol (MP, 151,089) and 30% glycerol (Sigma, G5516)). For IHC, sections were permeabilized with 0.2% Triton X-100 (Sigma-Aldrich, X100) in PBS for 15min, and subsequently blocked for 1h with 2% normal donkey serum in PBS-0.2% Tween 20 (VWR, 0777). After washes with PBS-0.1% Tween 20, sections were incubated with primary antibodies at 4°C overnight. After washes with PBS-0.1% Tween 20, sections were incubated for 1h at room temperature with appropriate secondary antibodies. For the IHC with hCD9, hP2RY12, Iba1, and X-34 in Fig. 4 and Supplemental Fig. 11, the protocol was modified as follows: sections were permeabilized with 0.1% Tween 20 (Sigma-Aldrich, P1379) in PBS for 15min at room temperature and blocked with 5% normal donkey serum in PBS-0.1% Tween 20 for 1h at room temperature. Washing steps were performed PBS 0.1% Tween 20. A list of primary and secondary antibodies used can be found in Supplemental Table 4 and 5. DAPI (ThermoFisher, 62,248) was used to stain nuclei, and sections were mounted with Mowiol–DABCO (Sigma-Aldrich, 81,381) mixture or FluorSave (Calbiochem, 345,789). X-34 was used to stain amyloid plaques. Briefly, after permeabilization and before blocking step, sections were incubated for 20 min at room temperature with X-34 (1:1000, Sigma-Aldrich, SML1954) in 40% (v/v) ethanol solution supplemented with 20 mM NaOH and then washed with 40% (v/v) ethanol in PBS. Images were acquired at 20 × or 60 × magnification using Nikon A1R Eclipse confocal system or Nikon Ti2 Widefield microscope. Distribution of ASO in the brain was analyzed using Fiji Image J software. Threshold was set using brain sections from vehicle-treated mice. After setting the threshold, integrated density was measured and normalized to surface area. To measure microglial activation at the site of Aβ plaques in relation to the rest of the tissue, the center of the X-34 labeled plaques were selected and signal intensities of microglial markers (hCD9, hP2RY12) were extracted at rings (annuli) of incrementing diameter surrounding the plaque center. The measurements were executed via a macro in Fiji software. Only plaques surrounded by human cells (i.e., hP2RY12 positive cells) were considered for the analysis. The acquisition of the images and the analysis was carried out blinded to avoid introducing unconscious bias. To quantify Aβ plaques, 3 sections per mouse were acquired at 4 × magnification on Nikon A1R Eclipse confocal system. Plaques were identified as X-34 positive structures; and plaque number, size and volume were analysed using NIS software. Plaque number and volume was recorded for the whole brain section, plaque size was expressed as average size of all plaques measured in one section.

RNAscope

RNAscope analysis was performed according to the instructions by the manufacturer on the Manual Fluorescent Multiplex kit v2 (ACDbio, 323,100) on PFA fixed tissue sections. Sections were air dried and incubated with Protease IV for tissue digestion and then incubated with the following probes for 2h at 40°C: Hs-APOE-C3 (ACDbio, 433091), Hs-TREM2-C1 (ACDbio, 420491) as part of the hybridization step. After the amplification step, sections were incubated with the TSA Plus fluorophores (TSA Cyanine Plus Evaluation Kit, Perkin Elmer, NEL744E001KT) for 30min at 40°C. Immediately after all steps were performed, sections were blocked and standard IHC was performed. Images were acquired with the Nikon A1R Eclipse confocal system.

Aβ extraction and ELISA detection

Snap frozen brain tissue, collected as described above, was homogenizes with PBS in the presence of protease inhibitor using the FastPrep beads. Homogenates were centrifuged at 5000 g for 5 min and supernatants were collected and further centrifuged for 1 h at 100,000 g. Supernatant were recovered as PBS-soluble fraction. The insoluble material was then solubilized in 6 M guanidine buffer, sonicated and centrifuged at 70,000 rpm for 20 min, and labeled as guanidine-soluble fraction. The levels of Aβ40 and Aβ42 were measured by ELISA. First, ELISA Meso Scale Discovery (MSD) 96-well plates were coated with human antibodies for Aβ40 and Aβ42 at 4 °C overnight and blocked with 0.1% casein in PBS for 1 h. Next, standards and samples were loaded on to the plates together with the sulfo-tag labeled JRF/AβN/25 antibody. Plates were recorded on the MSD microplate reader.

Statistical analyses

Statistical analyses were performed with Graphpad Prism 7 or 8 (LaJolla, Ca, USA). We compared experimental groups using Mann–Whitney test, one-way or two-way ANOVA (repeated measures) using Tukey’s multiple comparisons test as post hoc test. The differences from the MEA recordings were determined using one-way ANOVA with Dunnett's correction. Dose–response experiments were analyzed using a non-linear regression with a 95% confidence interval to calculate IC50 values with log-transformed ASO concentrations. The statistically significance levels were set at p-value < 0.05 (*), < 0.01 (**), < 0.001 (***), < 0.0001 (****) with a confidence interval of 95%. Data sets were plotted as mean ± standard deviation unless stated otherwise.