Human iPS cell lines and culture
The human iPS cell lines used for ATAC–seq (Supplementary Table 1) were derived at Rutgers University Cell and DNA Repository (RUCDR). The human iPS cell lines were generated using the Sendai virus method to ensure that they are integration-free and underwent the following quality-control procedures: immunofluorescence staining for pluripotency, mycoplasma contamination test, in-house RNA-seq-based pluripotency test (Pluritest) and eSNP-karyotyping20,21 or G-band karyotyping at RUCDR. All donors were of European ancestry and the samples were previously used for schizophrenia GWAS studies59,60. All donors were also analysed for copy-number variants, and none had large copy-number variants (>100 kb)61. There are 29 schizophrenia cases and 33 controls, of which 37 are male with an average age of 49.5 years (schizophrenia case–control status or age does not affect ASoC mapping20,21). Two control-donor human iPS cell lines homozygous for APOE3, CD04 and CD09 (abbreviated from the full cell line IDs: CD0000004 and CD0000009) were used in CRISPR–Cas9 editing. Human iPS cells were cultured using a feeder-free method on Matrigel-coated (Thermo Fisher Scientific) plates in mTeSR plus medium (100-0276, StemCell). The media were changed every other day, and cells were passaged as clumps every 4–6 days using ReLeSR (100-0483, StemCell). All cell cultures were confirmed to be mycoplasma-free using the PCR detection kit (ab289834, Abcam). Human iPS cell lines were obtained from the RUCDR NIMH Stem Cell Center (www.nimhgenetics.org). These iPS cell lines were generated from cryopreserved lymphoblasts deposited by the Molecular Genetics of Schizophrenia (MGS) consortium59,60, which had collected the biomaterials with informed consent as approved by the Endeavor Health (formerly NorthShore University HealthSystem) institutional review board (IRB), which also approved the current study.
PICALM RNA expression in human post-mortem brains
Frozen human brain samples (frontal cortex BA10 region) were received through the NIH biobank from Harvard Brain Tissue Resource Center and the University of Miami Brain Endowment Bank. For the PICALM expression assay, the grey matter was dissected from brain blocks on dry ice. RNA was isolated using the Direct-zol RNA MiniPrep Kit (Zymo), and reverse-transcribed into complementary DNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s instructions. Quantitative PCR (qPCR) reactions were set up using PowerUp SYBR Green Master Mix for qPCR (Applied Biosystems) and run on the QuantStudio Real-Time PCR System (Applied Biosystems). The data were analysed using the \({2}^{-\Delta \Delta {C}_{t}}\) method and normalized to COTL1. Primer sequences were as follows: PICALM-exon-1-F, TCTGCCGTATCCAAGACAGT; PICALM-exon-2-R, AAGACCACCACCCAACTACT; COTL1-F, CCAAGATCGACAAAGAGGCTT; COTL1-R, CGATGGTGGAGCCGTCATATTT.
PICALM staining of hMGs
Paraffin sections (6 μm) from post-mortem human brain tissue were obtained from UCLA or RUSH Alzheimer’s Disease Center (RADC) repositories. The RADC brains came from the Religious Orders Study and Rush Memory and Aging Project (ROSMAP)62. All ROSMAP participants enrolled without known dementia and agreed to detailed clinical evaluation and brain donation at death. Both studies were approved by the Rush University Medical Center IRB. The sections were incubated at 60 °C for 60 min, deparaffinized in xylene and rehydrated through a series of increasingly dilute ethanol solutions. Epitope retrieval was performed using decloaking buffer (Biocare Medical) at 95 °C for 30 min. Subsequently, freshly prepared 0.1% sodium borohydride was added to the sections and incubated at 4 °C for 30 min. Peroxidase blocking (3% H2O2) was carried out at room temperature for 30 min. The sections were then treated with permeabilization buffer (0.2% Triton X-100 in 1× TBS) at room temperature for 45 min. Non-specific epitopes were blocked by incubating the sections for 60 min in blocking buffer-I (10% donkey serum, 1% BSA, 0.3 M glycine, 0.1% Triton X-100) followed by a 60 min incubation with background punisher (Biocare Medical). Antibody staining was performed at room temperature using the intelliPATH FLX system and reagents provided by the manufacturer (Biocare Medical). Rabbit anti-PICALM antibodies (Sigma-Aldrich, HPA019061, 1:200) and goat anti-IBA1 (Abcam, ab5076, 1:1,000) were incubated with the sections for 48 h at 4 °C. After washing, secondary antibodies were applied and incubated overnight at 4 °C. Autofluorescence was quenched using 0.1% Sudan Black B in 70% ethanol, and nuclei were stained with Hoechst 33342. Finally, coverslips were mounted using VectorShield mounting medium. Images were acquired on an automated Nikon Eclipse Ti2 microscope fitted with the Yokogawa spinning-disk field-scanning confocal system and Photometrics PRIME 95B sCMOS camera, using a ×20 objective. z stack images were deconvolved using Nikon NIS-Elements AR5.20.01 software and processed with Fiji/ImageJ (1.54f, 64 bit). For quantification of PICALM in IBA1+ microglia, all processing was conducted by a researcher blinded to individual diagnoses. Images were acquired on the SLIDEVIEW VS200 slide scanner (Olympus) using ×20 magnification. Individual microglia were cropped using OlyVIA software (Olympus). All further processing was conducted in FiJi/ImageJ (1.54f, 64 bit). Background fluorescence was measured from non-specific tissue staining and normalized by subtracting 50% of the measured values. Binary masks of microglia were generated by thresholding the IBA1 signal, and manually inspected to remove non-specific signals or adjacent cells. PICALM signal was multiplied by the microglial masks, and fluorescence intensities (integrated densities) were measured. Data were plotted and analysed using Prism 10 (GraphPad).
iMG differentiation from human iPS cells
iMGs were generated from human iPS cell lines as described in our previous study63 using the Brownjohn’s method64. At least 2 days after passaging, when human iPS cells reached about 80% confluency, they were dissociated with Accutase (07920, StemCell) and plated at 10,000 cells per well in 96-well round-bottom ultra-low-attachment plates (7007, Corning) in 100 µl embryoid body (EB) medium (complete mTeSR with 50 ng ml−1 BMP-4 (120-05, PeproTech), 20 ng ml−1 SCF (300-07, PeproTech), 50 ng ml−1 VEGF-121 (100-20 A, PeproTech) and ROCK inhibitor (1254/1, R&D Systems)). Haematopoietic medium was prepared by adding to the X-VIVO 15 (BE08-879H, Lonza), 1% GlutaMax (35050061, Thermo Fisher Scientific), 1% penicillin–streptomycin (10378016, Thermo Fisher Scientific), 55 µM β-mercaptoethanol (21985023, Thermo Fisher Scientific), 100 ng ml−1 M-CSF (300-25, PeproTech) and 25 ng ml−1 IL-3 (200-03, PeproTech). After 5 days of culturing EBs in haematopoietic medium, primitive macrophage progenitors (PMPs) started appearing in the suspension and were produced continuously in suspension for 34 days. After 10 days of culturing EBs, PMPs were collected from suspension and plated in RPMI 1640 medium (21870076, Thermo Fisher Scientific) at 180,000 cells per cm2 in 6- or 12-well plates. Complete iMG medium was as follows: RPMI 1640 with 10% FBS (S11150H, R&D Systems), 1% penicillin–streptomycin, 1% GlutaMax, 100 ng ml−1 IL-34 (200-34, PeptroTech) and 10 ng ml−1 GM-CSF (300-03, PeproTech). The final differentiation of PMPs into iMGs occurred over 25 days.
iAst cell differentiation from NPCs
iPS-cell-derived neural progenitor cells (NPCs) were prepared using PSC neural induction medium (A1647801, Thermo Fisher Scientific). NPCs were differentiated to astrocytes by seeding dissociated single cells at 15,000 cells per cm2 density on Matrigel-coated plates in astrocyte medium (1801, ScienCell: astrocyte medium, 2% FBS (0010), astrocyte growth supplement (1852) and 10 U ml−1 penicillin–streptomycin solution (0503)). The initial NPC seeding density and single-cell dissociation are critical, particularly during the first 30 days of differentiation, to efficiently generate a homogenous population of astrocytes. On day −1, NPCs were pipetted with a p1000 pipette 3–5 times to yield a single-cell suspension and inhibit cell death. The NPC medium was switched to the astrocyte medium on day 0. From day 2, cells were fed every 48 h for 20–30 days. After 30 days of differentiation, astrocytes were split 1:3 weekly with Accutase and expanded for up to 120 days (15–17 passages) in the astrocyte medium. The final differentiation of iAst cells occurred over 30 days.
Differentiation of glutamatergic neurons
We followed an established protocol to differentiate iPS cells into glutamatergic neurons (iN-Glut)65. In brief, iPS cells were dissociated into single cells using Accutase (07920, StemCell) and replated at 7.5 × 105 cells per well in a six-well plate in mTeSR plus medium (100-0276, StemCell) with 5 μM ROCK inhibitor (1254/1, R&D Systems) on day −1. On day 0, cells were infected with 200 μl per well lentivirus cocktail containing 100 μl NGN2 virus and 100 μl rtTA virus65. After a two-day puromycin selection, iGlut cells on day 5 were dissociated with Accutase and plated as a 100 μl blob on glass coverslips (GG-12-15-Pre, Neuvitro). From day 6 onwards, 500 μl of neuronal culture medium was added into each well with a half-volume medium change every 3 days. Doxycycline was withdrawn on day 21 of differentiation. The final differentiation of iN cells occurred over 30 days.
Differentiation of dopaminergic neurons
The protocol for the differentiation of dopaminergic neurons (iDNs) was adapted from a previous study66. In brief, dopaminergic priming medium was added to the cells at 50% confluence on day 0. On day 7, the cells were replated onto six-well plates coated with Matrigel at 5 × 105 cells per well and switched to dopaminergic differentiation medium. The medium was changed every other day. On day 30, dopaminergic neurons were collected using Accutase (07920, StemCell) for ATAC–seq.
Differentiation of GABAergic neurons
We generated GABAergic neurons (iN-GA) from NPCs using the protocol described previously67 but with NPCs as the source cells. NPCs were seeded at 200,000 cells per cm2 on day 0. Virus neural expansion medium cocktail was added on day 1 with ASCL1-puro and DLX2-hygro virus and replaced by the expansion medium cocktail containing 2 µg ml−1 doxycycline (D9891, Sigma-Aldrich) the same day. Puromycin and hygromycin selection were conducted between day 2 and day 6. On day 7, we switched the medium to conditioned NeuralbasalPlus medium and changed the medium every 3 days. Doxycycline was withdrawn on day 16, and 50 nM Ara-C was included in the medium if non-neuronal cells were observed. On day 28, neurons were collected using Accutase for ATAC–seq.
Immunofluorescence staining of iMG and iAst cells
For characterizing iMG and iAst cells, cells were fixed in 4% PFA (P6148, Sigma-Aldrich) for 10 min at room temperature, followed by incubation with primary antibodies at 4 °C overnight in 3% BSA containing 0.3% Triton X-100. Cells were washed three times in PBS, and then incubated with secondary antibodies at room temperature for 1 h in 3% BSA containing 0.3 % Triton X-100. Next, the cells were washed another three times with PBS and incubated in 0.5 μg ml−1 4′,6-diamidino-2-phenylindole (DAPI) at room temperature for 10 min. Images were acquired using the Nikon ECLIPSE TE2000-U microscope.
Primary antibodies used for microglial immunofluorescence and their dilutions for incubation were as follows: rat anti-TREM2 (MABN755, Sigma-Aldrich, 1:100), rabbit anti-CD45 (SAB4502541, Sigma-Aldrich, 1:200), mouse anti-PU1 (89136, Cell Signalling, 1:100), mouse anti-IBA1 (MA5-27726, Thermo Fisher Scientific, 1:100), rabbit anti-P2Y12 (702516, Invitrogen, 1:200), rabbit anti-TMEM119 (AB209064, Abcam, 1:100), rabbit anti-PLIN2 (15294-1-AP, Proteintech, 1:200), rabbit anti-ATP6AP2 (SAB2702080, Sigma-Aldrich, 1:100), rabbit anti-VAMP1 (702787, Thermo Fisher Scientific, 1:100), rabbit anti-HMGCR (SAB4200528, Sigma-Aldrich, 1:100) and mouse anti-CD74 (14-0747-82, Thermo Fisher Scientific, 1:100). Secondary antibodies were Alexa 488 donkey anti-rat (A21208, Invitrogen, 1:1,000), Alexa 594 anti-rabbit (A21207, Invitrogen, 1:1,000), Alexa 647 anti-mouse (A32787, Invitrogen, 1:1,000), Alexa donkey 594 anti-mouse (A21203, Invitrogen, 1:1,000) and Alexa donkey 647 anti-rabbit (A32795, Invitrogen, 1:1,000). Primary antibodies used for iAst immunofluorescence and their dilutions for incubation were rabbit anti-vimentin (3932, Cell Signaling, 1:200), mouse anti-GFAP (G3893, Sigma-Aldrich, 1:100) and mouse anti-s100β (S2532, Sigma-Aldrich, 1:100). Secondary antibodies were Alexa 488 donkey anti-rabbit (A21206, Invitrogen, 1:1,000) and Alexa 594 anti-mouse (A21203, Invitrogen, 1:1,000).
RNA isolation and sequencing
Cells from human iPS, iAst and iMG cell cultures were dissociated using Accutase (07920, StemCell). Total RNAs were extracted using the RNeasy Plus Kit (74134, Qiagen). cDNAs were reverse transcribed from RNAs using a high-capacity cDNA reverse transcription kit (4368814, Applied Biosystems). RNA-seq was performed by Novogene on the Illumina NovaSeq 2000 platform with targeted 30 million paired-end reads (2 × 150 bp) per sample.
RNA-seq data and differential expression analyses
Raw FASTQ files were aligned to the human hg38 genome GRCh38.p14 using STAR v.2.7.2 and counted according to GENCODE annotation release version 35 on the fly. The ComBat-seq function from the R package sva was applied to remove batch effects between the two experiments (that is, rounds of iMG differentiation). We subsequently used the R package EdgeR (v.4.0.16) to calculate counts per million values from sva-corrected read counts for PCA analysis and plotting. Differential gene expression analysis was performed by applying general linear models and F tests between different groups (risk versus non-risk) (glmQLFit and glmQLFtest functions). When constructing the generalized linear models, we considered cell line ID (CD04 and CD09) as a coefficient to further remove line-specific effects. DEGs were defined as their Benjamini–Hochberg-adjusted P values (FDR) < 0.05.
ATAC–seq
ATAC–seq sample preparation was performed as previously described20,21. In brief, 75,000 viable cells were used for each transposition mixture reaction. The samples were then incubated at 37 °C for 30 min on a thermomixer at 1,000 rpm. The eluted DNA was shipped to the University of Minnesota Genomic Center for library preparation and ATAC–seq.
ATAC–seq data analysis and peaking calling
All raw sequence reads generated by Illumina NextSeq were demultiplexed at the University of Minnesota Genomics Center and provided as 2 × 75 bp paired-end FASTQ files (targeting 60 million reads per sample). Only paired-end reads that survived Trimmomatic processing v.0.39 (ILLUMINACLIP:NexteraPE-PE.fa:2:30:7, SLIDINGWINDOW:3:18, MINLENGTH:26) were retained. The FASTQ files were individually mapped against the human genome reference file including decoy sequences (GRCh38p7.13/hg38, 1000 Genome Project) using bowtie2 (-x 2000, -mm –qc-filter –met 1 –sensitive –no-mixed -t) and subsequently merged and sorted as BAM-formatted files using samtools v.1.14, with only uniquely high-quality mapped reads (MAPQ > 30, SAM flags 0×1, 0×2) retained. Picard tools MarkDuplicate was then used to remove all PCR and optical duplicated reads from the BAM file.
To further eliminate allelic bias towards reference alleles during the alignment step, we performed WASP (v.0.3.4) calibration on the generated raw BAM files68. In brief, we first called the VCF file profiles on all SNP variants per sample individually using GATK HaplotypeCaller to generate cell-line-specific VCF files. The cell-line-specific VCFs were used as the basis of WASP calibration and realignment, and new WASP-calibrated BAM file sets were collected as the final output for the ATAC–seq peak calling and ASoC SNP calling (see below). All analysed ATAC–seq samples passed standard quality control based on the characteristic nucleosomal periodicity of the insert fragment size distribution and high signal-to-noise ratio around transcription start sites (TSSs). For peak calling, MACS269 was used to generate peak files (narrowPeak format) with the recommended settings at FDR = 0.05 (-f BAMPE, –nomodel, –call-summits –keep-dup-all -B). Peaks that fell within the ENCODE blacklisted regions were removed. We also removed peaks within chromosomes X and Y, mitochondrial genome and decoy regions.
ASoC mapping
The ASoC approach20,21 was used to identify putative functional variants that showed differential chromatin accessibility between the two alleles of a SNP in ATAC–seq samples heterozygous for the tested SNP, with the assumption that a functional common GWAS SNP does not display monoallelic chromatin accessibility. In brief, GATK (v.4.1.8.1) was used for SNP calling as recommended by the GATK Best Practices (https://gatk.broadinstitute.org/hc/en-us/sections/360007226651-Best-Practices-Workflows)70. As noted above, WASP-calibrated BAM files were used as input and variants were called against the human GRCh38.p14 (hg38) reference genome and the corresponding dbSNP version 154, and only reads with MAPQ score ≥ 30 were used (-stand_call_conf 30). Subsequently, recalibration of SNPs and indels was performed in tandem using the VariantRecalibrator function (-an DP -an QD -an FS -an SOR -an MQ -an ReadPosRankSum -mode SNP -tranche 100.0 -tranche 99.5 -tranche 95.0 -tranche 90.0) and scores were recalibrated using reference database including HapMap v.3.3 (priority = 15), 1000G_omni v2.5 (priority = 12), Broad Institute 1000G high-confidence SNP list phase 1 (priority = 10), Mills 1000G golden standard INDEL list (priority = 12) and dbSNP v154 (priority = 2). Heterozygous SNP sites with tranche level >99.5% were extracted. To reduce bias introduced by any acquired (or de novo) mutations during cell growth, only SNPs with corresponding rs# records found in dbSNP v154 were retained. Biallelic SNP sites (GT: 0/1) with minimum read depth count (DP) ≥ 20 and minimum reference or alternative allele count ≥ 2 were retained. The binomial P values (non-hyperbolic) were calculated using binom.test(x, n, P = 0.5, alternative = “two.sided”, conf.level = 0.95) from the R package, and Benjamini–Hochberg correction was applied to all qualified SNPs as the correcting factor of the R function p.adjust(x, method = “fdr”). We set the threshold of ASoC SNP at an FDR value of 0.05. To ascertain whether rs10792832 also exhibits ASoC in human brain MG, we first converted the snATAC–seq data of hMG24 into pseudo-bulk ATAC–seq data for each sample. We then used GATK for SNP calling to identify individuals heterozygous (A/G) for rs10792832 in hMG, followed by ASoC testing as described above.
The read pileup statistics proximal to SNP sites were generated using samtools mpileup function, and differential allele-specific reads was performed using the SNPsplit Perl package (v.0.3.2) (www.bioinformatics.babraham.ac.uk/projects/SNPsplit). The final readouts from both read pileup and SNP-specific reads were visualized using the R package Gviz. Moreover, when comparing the changes in chromatin accessibility caused by genotypes across samples or between different cell types, read counts were scaled and normalized using the deepTools package (v.2.0) bamCoverage function and re-scaled to reads per genomic content as the base unit71. We confirmed there was no obvious mapping bias to reference alleles by visualizing the volcano plots that graph the allelic read-depth ratios against −log2[P] values in scatter plots.
sLDSC analysis of GWAS enrichment
Stratified linkage disequilibrium score regression (sLDSC)72 analysis was performed using the hg38 version of European genotype data (SNPs) from 1000 Genomes Phase 3 and v.2.2 baseline linkage disequilibrium/weights as previously described20. In brief, linkage disequilibrium score estimations were precalculated from the hg38 version of the 1000 Genomes EUR file set (w_hm3_no_hla.snplist), with a window size of 1 cM (ld-wind-cm 1). We used the GWAS summary statistics of major psychiatric disorders and non-psychiatric diseases (Supplementary Table 9) for partition heritability, with several datasets lifted over from hg19 to hg38 when necessary. Disease-specific regressions were performed using hm3 SNP weights against each disease independently for cell-type-specific analysis.
Torus GWAS enrichment analysis
Bayesian hierarchical model (TORUS) was applied to perform an SNP-based enrichment analysis to explore whether ASoC SNPs are enriched in any of the diseases73 as previously described20. For the GWAS enrichment test, ASoC SNPs derived from each cell type were applied independently. The annotations are encoded as Boolean (true if an SNP has an annotation). The GWAS datasets used for enrichment/TORUS analysis were consistent with the diseases analysed in sLDSC. A univariate analysis was performed to assess the enrichment of ASoC SNPs in each GWAS dataset.
CRISPR–Cas9 editing of human iPS cells
CRISPR guide RNA (gRNA) sequences were designed as described74, and we selected the gRNAs with the highest scores for specificity (Supplementary Table 19). The gRNAs were cloned into the pSpCas9(BB)-2A-Puro vector (Addgene, 62988) for co-expression with Cas9 based on an established protocol75. For transfection, 3 μg of CRISPR–Cas9–gRNA construct was combined with 3 μg of ssODNs (1:1 ratio) in Opti-MEM medium (31985062, Thermo Fisher Scientific) and Lipofectamine stem reagent (STEM00001, Thermo Fisher Scientific) was used for transfection. Puromycin selection was performed to eliminate untransfected cells and was withdrawn after 72 h of transfection. Resistant colonies were collected 14 days after transfection and a small amount of DNA from each colony was used for Sanger sequencing to verify editing. The purity of the selected clones was confirmed for on-target editing and the absence of off-target editing (see below).
Quality control of the CRISPR-edited iPS cell lines
Primers were designed to amplify regions corresponding to the four top-ranking predicted off-targets to check on-target and off-target editing. A list of all primer and oligo sequences is provided in Supplementary Table 19. To confirm the pluripotency of CRISPR–Cas9-edited human iPS cell lines, the cells were stained with pluripotency marker antibodies: rabbit anti-OCT4 (ab181557, Abcam, 1:250), goat anti-NANOG (AF1997-SP, R&D Systems, 1:50) and mouse anti-SSEA4 (ab16287, Abcam, 1:250). Images were taken using a Nikon ECLIPSE TE2000-U microscope.
eSNP-karyotyping was performed for all cell lines used to eliminate potential chromosomal abnormalities, as previously described20,21. RNA-seq data were processed using the eSNP Karyotyping package76 rewritten for GATK v.4 and R v.4.2, using raw FASTQ files as the input. Alignment to the human hg38 genome was performed by Bowtie2 v.2.5.1, and only common SNPs (MAF > 0.05) from dbSNP 154 were retained for zygosity block analysis. The plotted zygosity block size was 1.0 Mb.
CRISPRoff epigenome editing of human iPS cells
CRISPRoff77 was used to repress PICALM expression in iMGs. The gRNA sequences were designed using Benchling (https://benchling.com; Supplementary Table 19), and cloned into the CROPseq-Guide-Puro vector (Addgene, 86708) for co-expression with CRISPRoff v.2.1 (Addgene, 167981) as described previously77. After 72 h of drug selection, transduced cells were sorted using a BD FACSAria II, and the sorted cells were passaged twice and then differentiated into iMGs.
CRISPRa to overexpress PICALM
We used CRISPR-ERA78 to design PICALM activation gRNA. A total of four gRNA candidates close to the TSS of PICALM with low predicted off targets and E scores were selected and cloned into the lentiviral gRNA vector lentiGuide-Hygro-mTagBFP2 (Addgene, 99374) through Gibson assembly. We then co-transfected gRNA plasmids with lenti-EF1a-dCas9-VPR-Puro (Addgene, 99373)79 into HEK293T cells using Fugene HD (Promega) to evaluate the PICALM activation efficiency of each CRISPRa gRNA and selected one gRNA with the highest activation efficiency for further establishing inducible CRISPRa iPS cell line that can overexpress PICALM. To establish inducible CRISPRa iPS cell line on risk allele background, we first transduced lentivirus for the lentiGuide-Hygro-mTagBFP2 carrying the selected PICALM gRNA and sorted BFP-positive cells to enrich the cells with high expression of gRNAs. After sorting, iPS cells were maintained in 100 µg ml−1 hygromycin to enrich cell stably expressing the gRNAs. Then cells were transduced with the lentivirus of inducible dxCas9-VPR expressing vector pLenti-tetON-dxCas9(3.7)-VPR-EF1a-TagRFP-2A-tet3G (Addgene, 167937). BFP- and RFP-positive cells were sorted to enrich iPS cells with high expression of dxCas9-VPR. The sorted iPS cells were maintained with hygromycin and expanded for iMGs differentiation, followed by CRISPRa induction of PICALM expression by treating cells with 2 µg ml−1 doxycycline for 25 days before collecting iMGs for analyses. The four tested gRNAs are gRNA 3-GAGTTCCATCACGTAACGCG, gRNA 4-GCCTCAGGCGACCTGTTGGC, gRNA 6-GCAGTGTCAACGTCTTTCCA, and gRNA 7-GGGCGGGCGTCGAAGAGGAA (the best-performing one used in CRISPRa in iMGs).
Gene expression analysis by qPCR
For qPCR, reverse transcription was performed using the Thermo Fisher Scientific High-capacity RNA-to-cDNA reverse transcription kit (4366596, Applied Biosystems) with random hexamers according to the manufacturer’s protocol. qPCR was performed using the TaqMan Universal PCR Master Mix (4364338, Applied Biosystems) on the Roche 480 II instrument (with Roche LightCycler 480 1.5.1), using gene-specific FAM-labelled TaqMan probes or custom-designed probes from IDT (Supplementary Table 19). GAPDH was used as the control.
Myelin isolation from mouse brains for phagocytosis
Myelin was isolated from mouse brains by homogenization in 0.32 M sucrose buffer (0.32 M sucrose and 2 mM EGTA). The samples were then further homogenized using a Dounce homogenizer, layered on top of 0.85 M sucrose buffer (0.85 M sucrose and 2 mM EGTA), and centrifuged at 75,000g at 4 °C for 30 min. Crude myelin was collected from the interface, resuspended in Tris-Cl Buffer (0.2 M Tris-HCl, pH 7.5), and homogenized using a Dounce homogenizer. The samples were centrifuged at 75,000g at 4 °C for 15 min. The pellets were resuspended in Tris-HCl solution (20 mM Tris-HCl, 2 mM EDTA, 1 mM DTT, pH 7.5) and homogenized using a Dounce homogenizer. The samples were then centrifuged at 12,000g at 4 °C for 15 min, and the pellets were resuspended in Tris-Cl Solution. The samples were then centrifuged at 12,000g at 4 °C for 10 min. Pellets were resuspended in 0.32 M sucrose buffer, layered on top of 0.85 M sucrose buffer and centrifuged at 75,000g at 4 °C for 30 min. Purified myelin was collected from the interface, resuspended in Tris-HCl buffer, and homogenized using a Dounce homogenizer. The samples were next centrifuged at 75,000g at 4 °C for 15 min, and the pellets were resuspended in Tris-HCl solution and homogenized using a Dounce homogenizer. The samples were then centrifuged at 12,000g at 4 °C for 15 min, and the pellets resuspended in Tris-HCl solution and centrifuged at 12,000g at 4 °C for 10 min. The pellets were resuspended in Tris-HCl Solution, aliquoted and stored at −80 °C. The protein content of isolated myelin was determined using the BCA protein assay kit.
Phagocytosis assay for iMGs
iMGs were grown on MatTek 96-well plates with a glass bottom (NC1844174, Thermo Fisher Scientific) until day 25. For Aβ phagocytosis, the β-amyloid (1-42) aggregation kit was used (A-1170-025, rPeptide). The peptide was resuspended in 5 mM Tris at 1 mg ml−1 concentration. Myelin and Aβ peptides were labelled using the pHrodo Red Microscale Labelling Kit (P35363, Thermo Fisher Scientific) and pHrodo Deep Red Labelling Kit (P35355, Thermo Fisher Scientific) according to the vendor’s protocol.
For pHrodo phagocytosis experiments, pHrodo-labelled myelin or Aβ was diluted to 15 µg ml−1 in RPMI 1640 medium (21870076, Thermo Fisher Scientific), bath sonicated for 1 min and added to the iMGs along with CellMask Green Plasma Membrane Stain (C37608, Thermo Fisher Scientific, 1:1,000) and NucBlue Live ReadyProbes Reagent (R37605, Thermo Fisher Scientific, 2 drops per ml), mixed gently and incubated at 5% CO2 and 37 °C for 30 min. As a negative control, 10 µM cytochalasin D (PHZ1063, Thermo Fisher Scientific) was added to cells along with pHrodo-labelled protein and retained throughout uptake assays. Live imaging (5% CO2, 37 °C) was performed for 3 h using the Nikon ECLIPSE TE2000-U microscope at 45 min intervals.
For the pHrodo phagocytosis experiment that included LD staining, the iMGs were treated with 1 µM TrC (10007448, Cayman Chemical) in complete iMG medium for 18 h. Next, BODIPY 493/503 (D3922, Thermo Fisher Scientific, 1:1,000) and CD45 antibodies (14-0451-82, Thermo Fisher Scientific, 1:500) were added to cells with and without TrC and incubated for 30 min and quickly washed twice with RPMI 1640. Then pHrodo-labelled myelin or Aβ was diluted to 15 µg ml−1 in RPMI 1640 medium, bath sonicated for 1 min and added to the iMGs along with NucBlue Live ReadyProbes Reagent and incubated at 5% CO2, 37 °C for 30 min. As a negative control, 10 µM cytochalasin D was added to cells along with pHrodo-labelled protein and retained throughout uptake assays. Live Imaging (5% CO2, 37 °C) was performed for a total of 3 h using Nikon ECLIPSE TE2000-U microscope at 45 min intervals. Fiji/ImageJ (v.1.54f, 64 bit) software was used to quantify pHrodo fluorescence intensity (https://fiji.sc).
ChIP
The ChIP–qPCR assay was performed by combining two protocols from the Simple ChIP Enzymatic Chromatin IP kit (91820, Cell Signaling) and the Magna ChIP A/G Chromatin Immunoprecipitation kit (17-10085, Sigma-Aldrich). In total, 107 cells were used for each reaction with 1% formaldehyde (28908, Thermo Fisher Scientific) cross-linking in 20 ml of cell suspension. The Cell Signaling IP kit was used for nuclei preparation and subsequent recovery reactions according to the vendor’s protocols. For chromatin digestion, 1.25 µl of micrococcal nuclease was used and incubated for 20 min at 37 °C to digest DNA to the length of approximately 150–900 bp. To break the nuclear membrane, lysate was sonicated for three sets of 20 s pulses with a 1/8-inch probe.
The Sigma-Aldrich ChIP Assay kit was used for the reaction according to the vendor’s instructions. Normal rabbit IgG (2729, Cell Signaling) was used as negative control. In total, 1 µl of proteinase K was used for reverse cross-linking of protein–DNA complexes to free DNA at 62 °C for 2 h with shaking, followed by incubation at 95 °C for 10 min. For each reaction, DNA was eluted into 50 µl of elution buffer C. qPCR was performed using TaqMan Universal PCR Master Mix (4364338, Applied Biosystems) on the Roche 480 II instrument, using IDT custom probe for detection PICALM/PU.1 ratio (Supplementary Table 19). The ChIP DNA product for the heterozygous site rs10792832 was also subjected to Sanger sequencing (Thermo Fisher Scientific, Sequencing Analysis Software 7.0.0). Genomics DNAs for the heterozygous site were sequenced by using primers used for on-target CRISPR–Cas9 editing of rs10792832.
Fatty acid (Red-C12) transfer assay
iMG and iAst cells were grown on coverslips until day 25. Neurons (iNs) were grown on coverslips until day 30. Cells were incubated with 8 μM BODIPY 558/568 (Red-C12, D3835, Thermo Fisher Scientific) for 16 h in neuronal growth medium, washed twice with warm PBS and incubated with fresh medium for 1 h. Red-C12 labelled neurons and unlabelled astrocytes/microglia were washed twice with warm PBS, and the RedC12 intensity was examined using fluorescence microscopy.
LD staining with BODIPY for iMGs
iMGs were grown on glass coverslips until day 25. Cells were then fixed for 30 min at room temperature with 4% PFA (P6148, Sigma-Aldrich) in PBS, briefly washed in PBS twice and incubated in PBS with BODIPY 493/503 (D3922, Thermo Fisher Scientific, 1:1,000 from 1 mg ml−1 stock solution in DMSO) and DAPI for 10 min at room temperature. The BODIPY intensity was examined by fluorescence microscopy.
LD staining with LipidTOX for iMGs
iMGs were grown on glass coverslips until day 25. Cells were then fixed for 30 min at room temperature with 4% PFA (P6148, Sigma-Aldrich) in PBS, briefly washed in PBS twice and incubated in PBS with LipidTOX (H34476, Thermo Fisher Scientific, 1:1,000) and DAPI for 1 h at room temperature. The LipidTOX intensity was examined using fluorescence microscopy.
ROS staining
iMGs were grown on glass coverslips until day 25. Cells were treated with 1 µM TrC (10007448, Cayman Chemical) in complete iMG medium for 18 h. Cells were subsequently incubated in complete iMG medium with CellROX Deep Red (C10422, Invitrogen, 1:500) for 30 min at 37 °C. Next, the cells were stained with BODIPY to detect LDs. The CellROX intensity was examined by fluorescence microscopy.
Lipid peroxidation assay using BODIPY C11
iMGs were grown on glass coverslips until day 25. Cells were treated with 1 µM TrC as described in ROS staining. Next, cells were incubated in complete iMG medium with BODIPY 581/591 C11 (D3861, Thermo Fisher Scientific, 1:1,000) for 15 min at 37 °C, then fixed for antibody staining by rat anti-TREM-2 (MABN755, Sigma-Aldrich, 1:100). Five hundred and sixty-eight nm excitation wavelength was applied to excite reduced BODIPYC11 and 488 nm excited oxidized BODIPYC11. Fiji/ImageJ (v.1.54f 64 bit) software was used to quantify BODIPYC11 fluorescence intensity (https://fiji.sc).
Filipin staining for iMGs
iMGs were grown on glass coverslips until day 25. Cells were then fixed for 10 min at room temperature with 4% PFA (P6148, Sigma-Aldrich) in PBS. Cells were incubated with a solution of filipin (0.1 mg ml−1, F‐9765, Sigma-Aldrich) for 30 min. After staining with were rat anti-TREM-2 (MABN755, Sigma-Aldrich, 1:100), cells were washed and counterstained with propidium iodide (0.35 μg ml−1; P4170, Sigma-Aldrich) for 10 min at room temperature. Four hundred and five nm excitation wavelength was used to excite filipin. For each field of view (FOV), filipin fluorescence intensity was calculated by dividing the number of blue puncta by the number of microglia. The values were then normalized to the filipin fluorescence intensity value of the risk allele and used for statistical analysis. The Fiji software was used to quantify filipin fluorescence intensity (https://fiji.sc).
Lysosomal staining for iMGs
iMGs were grown on glass coverslips until day 25. Cells were incubated with LysoTracker Red DND-99 (L7528, Invitrogen, 100 nM) in complete iMG medium for 30 min at 37 °C. Cells were then fixed for 30 min at room temperature with 4% methanol-free PFA (28906, Pierce) for 10 min at room temperature. The cells were then stained with BODIPY as described above (LD staining with BODIPY) to detect LDs. The images were taken on a Nikon ECLIPSE TE2000-U microscope.
FACS sorting of iMGs
iMGs were grown on six-well plates until day 25. iMGs were then dissociated with Accutase (07920, StemCell) and stained with CD45-PE (12-0451-82, Invitrogen, 1:300) for 30 min at 37 °C and washed with RPMI medium three times by centrifugation at 300g for 5 min. Next the iMGs were stained with BODIPY 493/503 (D3922, Thermo Fisher Scientific, 1:1,000 from 1 mg ml−1 stock solution in DMSO) for 10 min at 37 °C with after washing twice with RPMI medium by centrifugation at 300g for 5 min. Cells were resuspended in PBS and sorted using BD FACsAria Fusion. Data were collected using BD FACSDiva software (v.9.0.1) and analysed using FlowJo (v.11.0).
PICALM KO by CRISPR–Cas9 editing C20 cells
C20 cells were maintained in DMEM/F12 medium containing 10% FBS and 1% penicillin–streptomycin. PICALM exon 1 was targeted using the following oligonucleotide sequences—sgRNA F, CACCGGCCGGTGACACTGTGCTGGG; and R, AAACCCCAGCACAGTGTCACCGGCC. Control oligo sequences were generated using sequences not specific to the human genome. Recombinant lentiviruses were generated in HEK293T cells using MISSION Lentiviral Packaging Mix (Sigma-Aldrich, SHP001). C20 cells were transduced with filtered virus-containing medium, and stable pools were selected in blasticidin (20 µg ml−1).
Immunoblots for iMG and C20 cells
For iMG, cells were lysed in phosphosafe buffer (71296, EMD Millipore) with protease inhibitor (04693132001, Roche) and PhosphoSTOP tablet (4906845001, Roche). Next, 50 μg of protein lysates were run on TGX Stain-Free Precast Gel (4–15%) (4561083, Bio-Rad) in Tris/glycine/SDS running buffer (1610732, Bio-Rad). The gel was then transferred onto a 0.45-μm PVDF membrane (88585, Thermo Fisher Scientific). The membrane was blocked with 5% skimmed milk and probed with rabbit anti-PICALM (HPA019061, Sigma-Aldrich, 1:1,000) or mouse anti-β-actin (A5316, Millipore, 1:10,000) at 4 °C overnight. Secondary antibodies, anti-rabbit-HRP (7074, Cell Signaling, 1:20,000) and anti-mouse-HRP (7076, Cell Signaling, 1:20,000), were then added. The blots were imaged and quantified using Bio-Rad Image Lab v.6.1.0.
For C20 line, cells were lysed in RIPA buffer (50 mM Tris, 50 mM NaCl, 0.5% sodium deoxycholate, 1% Triton X-100, 0.1% SDS, 5 mM EDTA, pH 8) containing complete protease inhibitors (Roche), sonicated and lysates cleared by centrifugation (21,000g for 2 min). Fifty μg of lysate was used per lane on 4–20% SDS PAGE gels and transferred onto nitrocellulose membranes. Non-specific sites were blocked with PBS containing 1% BSA and 1% fish gelatin at room temperature for 1 h. Membranes were incubated with primary antibodies rabbit anti-PICALM (Sigma-Aldrich, HPA019061, 1:500) and mouse anti-β-actin (Proteintech, 66009-1-lg, 1:50,000) in at 4 °C for 16 h. Secondary antibodies IRDye 680RD donkey anti-rabbit IgG and IRDye 800CW donkey anti-mouse IgG (Li-COR) were incubated at room temperature for 2 h. Blots were imaged and quantified on a Li-COR Odyssey infrared imaging system.
Myelin phagocytosis for C20 cells
Myelin isolated from the mouse brain (as described above) was conjugated to pHrodo-Green (Thermo Fisher Scientific, 35369) or pHrodo-Red (Thermo Fisher Scientific, P36600). C20 cells were incubated with pHrodo-conjugated myelin (20 µg ml−1) at 37 °C (5% CO2) for 1 h. Negative controls were treated with cytochalasin D (10 nM, Invitrogen). After fixation, cells were stained with rabbit anti-BIN1 antibody (Proteintech, 14647-1-AP, 1:500) for 2 h. Images were acquired on a Nikon Eclipse Ti2 (Yokogawa spinning-disk field-scanning confocal) microscope at ×20 magnification and captured using a Photometrics PRIME 95B sCMOS camera. Single-plane images were processed using Fiji/ImageJ (v.1.54f, 64 bit) software to threshold whole-cell masks (created from BIN1 staining). The integrated density of pHrodo-myelin within each cell was measured from five random fields of view (per biological replicates) and the median values from each replicate were used for statistical analysis.
LD staining for C20 cells
Live C20 cells were labelled with BODIPY (2 µM in PBS) at 37 °C (5% CO2) for 15 min. At room temperature, cells were fixed with 4% PFA for 30 min, and nuclei were labelled with Hoechst 33342. The volumes of BODIPY+ droplets and the number of nuclei was quantified from deconvolved image stacks using ImageJ (v.1.54f, 64 bit)/Fiji software. Five random fields of view (per biological replicate) were acquired for quantification, and the median values from each biological replicate were used for statistical analysis.
Organelle marker staining for C20 cells
For transferrin endocytosis, C20 cells were washed and incubated in uptake medium (DMEM containing 25 mM HEPES) for 1 h. Cells were treated with 10 µg ml−1 transferrin Alexa Fluor 555 conjugate (Invitrogen, T35352) in DMEM/HEPES containing 1 mg ml−1 BSA at 37 °C for 30 min. Cells were then chilled to 4 °C, and non-internalized transferrin was washed from cell surfaces with acid wash (0.5 M NaCl, 0.2 M acetic acid, pH 2.8) before fixing the cells in PFA. For cholera toxin internalization, cells were washed with labelling medium (1 mg ml−1 BSA in serum-free DMEM containing 25 mM HEPES), chilled to 4 °C and treated with 100 nM cholera toxin subunit B Alexa Fluor 647 conjugate (Invitrogen, C34778) in labelling medium at 4 °C for 10 min. Cells were washed with ice-cold PBS and fixed with 4% PFA (in PBS) at room temperature for 15 min. Nuclei were labelled with Hoechst 33342.
For immunofluorescence staining with antibodies, C20 cells grown on glass coverslips were fixed with 4% PFA and blocked (3% BSA, 50 mM NH4Cl, 10 mM glycine, PBS) at room temperature for 30 min. The cells were incubated with rabbit anti-Giantin (Covance, PRB-114C, 1:5,000), mouse anti-EEA1 (BD Transduction Laboratories, 610457, 1:500) or mouse anti-AP-4ε (BD Transduction, 612018, 1:100) antibodies at room temperature for 2 h. All C20 cell images were acquired on a Nikon Eclipse Ti2 spinning-disk field-scanning confocal microscope at ×60 magnification. z stacks (100 nm step size) were processed to generate maximum-intensity projections using Fiji/ImageJ (1.54f, 64 bit).
Lipid extraction from iMGs
iMGs carrying the PICALM risk allele, non-risk allele or risk-CRISPRa were grown in six-well plates until day 25. iMGs were dissociated with Accutase (07920, StemCell), collected and washed with PBS three times by centrifugation at 300g for 5 min. Lipids were extracted from iMGs as described previously80,81. In brief, a deuterated lipid standard mixture (330820, LM6003, 860739, 330727, 860657, 860658, Avanti Research) was added to iMGs equally and organic solvent (chloroform:methanol, 1:2 (v/v)) was added to iMG pellets and subjected to tip sonication, followed by vortexing and centrifugation to extract neutral lipid. The organic layer was collected and saved for later use and acidic extraction was performed on the remaining pellet by a mixture of chloroform:methanol:37% 1 M HCl (40:80:1, v/v/v). After the addition of chloroform and 0.1 M HCl to induce phase separation, the samples were vortexed and centrifuged. The bottom organic layer was combined with the previously collected organic layer and dried using vacuum centrifuge. The dried lipid extract was reconstituted in methanol:chloroform (1:3, v/v) for further analysis.
Lipidomics analysis of iMGs
Lipid extracts were subjected to global lipidomics analysis using liquid chromatography–tandem mass spectrometry (LC–MS/MS). LC–MS-grade methanol (A456, Thermo Fisher Scientific), acetonitrile (A955, Thermo Fisher Scientific), isopropanol (A461, Thermo Fisher Scientific), water (W71, Thermo Fisher Scientific), formic acid (A117, Thermo Fisher Scientific) and ammonium formate (78314, Millipore Sigma) were used. The samples from two different iMG differentiations were analysed, with lipidomics data acquired in two technical replicates for each sample. Reconstituted lipid extracts were separated on a Hypersil GOLD Vanquish C18 UHPLC column (15 cm × 2.1 mm, C18 1.9 μm and 175 Å) using an Orbitrap Tribrid IQ-X mass spectrometer (Thermo Fisher Scientific) coupled to Vanquish Horizon UHPLC (Thermo Fisher Scientific). A binary gradient was used at a flow rate of 300 µl min−1, using mobile phase A (water:acetonitrile, 6:4, with 10 mM ammonium formate and 0.1% formic acid) and organic phase (isopropanol:methanol:acetonitrile, 8:1:1, with 10 mM ammonium formate and 0.1% formic acid). In brief, mobile phase B increased from 20% to 95% over 17 min and was maintained at 95% for 5 min for washing. It was ramped down to 20% over 0.1 min and equilibrated for 5 min for the next injection. The analytical column was maintained at 50 °C. A full-scan MS was performed using the Orbitrap with a resolution of 60,000 at m/z 200. MS/MS spectra were acquired at a resolution of 15,000 at m/z 200. MS/MS scans were acquired for 1.5 s, followed by an MS scan. In positive-ion mode, MS/MS fragmentation using a stepped collision energy of 30%, 35% and 40% in higher-energy collisional dissociation (HCD) was performed with a spray voltage of 3.5 kV. In negative-ion mode, a spray voltage of 3 kV and a stepped collision energy of 30%, 35% and 40% in MS/MS acquisition using HCD were used. The acquired tandem mass spectra were processed using LipidSearch v.5.1.6 (Thermo Fisher Scientific) for lipid annotation and quantification. Lipids were identified based on precursor ion masses and their corresponding MS/MS spectra were matched against the database. Annotated lipids were quantified by calculating their peak areas using LipidSearch. All lipids were normalized to the peak area of an internal standard with the same head group as the target lipids. Total lipid levels were determined by summing the peak areas of individual lipid species sharing the same head group.
Imaging quantification and statistical analyses
For image analyses, we assayed fluorescence intensity, puncta density, puncta area and/or puncta size using the number of samples specified in the corresponding figures. For each FOV, the fluorescence intensity was calculated by dividing the number of fluorescent puncta by the number of microglia. The puncta number and the total area of puncta were acquired by applying a threshold to the respective fluorescent areas and performing the Analyze particles function. The number of microglia was calculated by applying a threshold to the DAPI fluorescence signal and measuring the number of nuclei through the same function. The images were acquired in a manner that the operator was blinded to the sample and genotype identification. For testing statistical differences between two groups, we used a two-tailed unpaired Student’s t-test when the experimental design did not involve multiple batches (for example, different rounds of experiments); otherwise, we used the R packages lme4 and lmerTest to fit data into an LMM to account for potential random effects from different experimental rounds and clones. For statistical tests involving more than two groups, we applied a one-way ANOVA with the appropriate post hoc test (Dunnett or Tukey), while accounting for any possible random effects from experimental rounds when applicable. The data analysis was performed using R v.4.3.2, GraphPad Prism 9 and Microsoft Excel. The results were considered to be significant if P < 0.05. All data with error bars were presented as the mean ± s.e.m. For ATAC–seq analysis of ASoC SNPs, the two-sided binomial test in R package was used to test allelic bias, and the Benjamini–Hochberg correction was applied to all qualified ASoC SNPs. For RNA-seq DEG analysis, limma (v.3.58.1)/EdgeR (v.4.0.16) with a generalized linear model (GLM) F test was used, and the Benjamini–Hochberg procedure was used to adjust P values accounting for multiple testing.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Source link
