APOE deficiency impacts neural differentiation and cholesterol biosynthesis in human iPSC-derived cerebral organoids

Background The apolipoprotein E (APOE) gene is the strongest genetic risk factor for Alzheimer’s disease (AD); however, how it modulates brain homeostasis is not clear. The apoE protein is a major lipid carrier in the brain transporting lipids such as cholesterol among different brain cell types. Methods We generated three-dimensional (3-D) cerebral organoids from human parental iPSC lines and its isogenic APOE-deficient (APOE−/−) iPSC line. To elucidate the cell-type-specific effects of APOE deficiency in the cerebral organoids, we performed scRNA-seq in the parental and APOE−/− cerebral organoids at Day 90. Results We show that APOE deficiency in human iPSC-derived cerebral organoids impacts brain lipid homeostasis by modulating multiple cellular and molecular pathways. Molecular profiling through single-cell RNA sequencing revealed that APOE deficiency leads to changes in cellular composition of isogenic cerebral organoids likely by modulating the eukaryotic initiation factor 2 (EIF2) signaling pathway as these events were alleviated by the treatment of an integrated stress response inhibitor (ISRIB). APOE deletion also leads to activation of the Wnt/β-catenin signaling pathway with concomitant decrease of secreted frizzled-related protein 1 (SFRP1) expression in glia cells. Importantly, the critical role of apoE in cell-type-specific lipid homeostasis was observed upon APOE deletion in cerebral organoids with a specific upregulation of cholesterol biosynthesis in excitatory neurons and excessive lipid accumulation in astrocytes. Relevant to human AD, APOE4 cerebral organoids show altered neurogenesis and cholesterol metabolism compared to those with APOE3. Conclusions Our work demonstrates critical roles of apoE in brain homeostasis and offers critical insights into the APOE4-related pathogenic mechanisms. Supplementary Information The online version contains supplementary material available at 10.1186/s13287-023-03444-y.

With the development of iPSC technologies, the emergence of three-dimensional cerebral organoid model system with distinct cell diversity provides an optimal tool to define the cell-type-specific effects of disease associated genes in their native environment [27][28][29].
In our previous study, we have successful detected multiple AD-related pathologies in cerebral organoids from AD patients, where we found APOE4 exacerbates ADrelated pathologies and induces lipid droplet accumulation in cerebral organoids compared to APOE3 [30,31].Using chimeric cerebral organoid system, Huang et al. further revealed the differential effects of neuronal and astrocytic APOE4 on lipid metabolism and AD pathologies, supporting the importance in defining apoE effects in different cell types [32].Thus, to dissect the roles of apoE in different brain cell types, we conducted singlecell RNA sequencing (scRNA-seq) analysis and pathway validation in the cerebral organoids from APOE-deficient (APOE −/− ) and isogenic parental iPSC lines.We found that APOE deficiency impacts cellular composition of the iPSC-derived cerebral organoids by modulating the eukaryotic initiation factor 2 (EIF2) pathway and cholesterol biosynthesis pathway.Some of our observations were further extended to APOE4 cerebral organoids.Our findings provide new insights into the cellular and molecular mechanisms through which apoE modulates brain homeostasis in a cell-type-specific manner.

APOE-deficient and genotype-specific isogenic iPSC lines
Two different sets of human parental iPSC lines and their isogenic APOE −/− iPSC lines were used in the study.One set of the parental (Con) and isogenic APOE −/− iPSC lines were obtained from the Xcell Science.Detailed information about these lines can be found in the Xcell Science website (http:// www.xcell scien ce.com/ produ cts/ ipsc).For the MC0192 iPSC line set, isogenic APOE −/− iPSC line was obtained via CRISPR/Cas9 knockout of the APOE gene.APOE deletion for both isogenic sets have been confirmed in our previous publication [31].The APOE4 and APOE3 isogenic lines were kind gifts from Dr. Yadong Huang [15].

Cerebral organoid culture
Cerebral organoids were generated using STEMdiff ™ Cerebral Organoid Kit (Stemcell Technologies) following manufacturer's instructions.Human iPSC colonies were dissociated into single-cell suspension with Accutase at Day 0. Cells were seeded into a U-bottom ultra-low-attachment 96-well plate (15,000 cells/well) in Medium A with 10 μM Y-27632.Additional 100 μL of medium A was added into each well on Day 2 and Day 4, respectively.EBs were moved to 48-well low attachment plates in Medium B on Day 5 and left for an additional 3-5 days.EBs were further embedded into 20 μL of Matrigel and cultured in Medium C + D in 6-well low attachment plates for 3 days.In the final stage, organoids were moved to an orbital shaker in 10-cm dishes and cultured in Medium E, which was replaced with neuronal maturation medium after 4 weeks.Cerebral organoids were harvested at Day 90 for scRNA-seq and validation.For the integrated stress response inhibitor (ISRIB) treatment, cerebral organoids at Day 60 were treated with neuronal maturation medium containing 100 nM ISRIB or vesicle DMSO for 30 days.Medium was changed every three days, and cerebral organoids at Day 90 were harvested for further analysis.Cerebral organoids in abnormal size (< 1 mm diameter) were considered unhealthy and excluded from further experiments.

Single-cell suspension, library preparation and sequencing
The parental and isogenic APOE -/-iPSC lines from the Xcell Science were differentiated into cerebral organoids and subjected to single-cell RNA-sequencing (scRNAseq) analysis at Day 90.Cerebral organoids were dissociated into single-cell suspension using the Worthington Papain Dissociation System kit (Worthington Biochemical, LK003150) following manufacturer's instruction [33].Cerebral organoids were gently transferred to an individual 60-mm dish with 2.5 mL of Papain + DNase solution.Three organoids were pooled as one sample.Cerebral organoids were minced into small pieces (< 1 mm), and the plates were transferred to an orbital shaker at 45 rpm inside a humidified tissue culture incubator at 37 °C and 5% CO 2 .After 15 min, minced pieces were gently dissociated using a 1 mL pipette and returned to the incubator for additional 15 min.Dissociated tissues were gently pipetted up and down 10 times and transferred to an empty 15-mL conical tube for the debris to settle (1-3 min).Cell suspension (avoid debris) was transferred to the prepared stop solution and centrifuged at 300 g for 5 min.Dissociated cells were resuspended in ice-cold PBS containing 0.05% BSA and passed through 30 µm cell strainer.In all samples, more than 90% of the cells were viable based on trypan blue examination.Approximately 6000 isolated cells from one sample were loaded into each sample well on a chip and combined with Gel Beads containing barcoded oligonucleotides using a 10 × Chromium Controller (10 × Genomics).Single-cell libraries were constructed according to the manufacturer's instructions and were sequenced by an Illumina HiSeq 4000 Sequencing Systems.

Single-cell transcriptome analysis
The single-cell sequence preprocessing was performed using the standard 10 × Genomics Cell Ranger Single Cell Software Suite (v5.0.0).Briefly, raw sequencing data were demultiplexed, aligned to the Human reference genome, GRCh38 2020-A (GENCODE v32/Ensembl 98), and the reads aligned to each gene were counted.Cell filtration, normalization, clustering, and differential expression analyses were performed with R (v4.0.3) and the singlecell analysis package Seurat v4 [34].For quality control, cells with greater than 25,000 unique molecular identities (UMI) counts, less than 5000 counts, less than 200 features, and mitochondria percentage greater than 12% were discarded.The resulting count matrix was 36,601 genes by 27,199 cells.For each line (parental or isogenic), SCTransform was applied setting vars.to.regress to mitochondria percentage [35].Each organoid was treated as a separate batch and integrated with IntegrateData, setting normalization.method to 'SCT' , and setting references to organoid 1 from each line.Following batch correction, FindNeighbors was run on the first 20 Principal Components (PCs), FindClusters was run with resolution = 0.5.Clusters were visualized with t-SNE plot run on the first 30 PCs.For computing marker genes of each cluster, FindMarkers was run using the MAST method, and setting latent.vars to the cellular detection rate, computed by scaling the number of features per cell [36].For computing differentially expressed genes between conditions, cells were subset either to each cell type or cell cluster, and DEGs were computed again with Find-Markers using MAST method, setting latent.vars to the cellular detection rate.DEGs were filtered for Bonferroni-corrected p values < 0.05 and absolute value of log2 fold change greater than 0.25.Pathway analyses were performed with Ingenuity Pathway Analysis (Content version: 62,089,861).For pseudotime trajectory analysis, Monocle3 was run with default settings, using the previously computed t-SNE plot dimensionality reduction as input [37].For computing the trajectory graph, the root node was manually selected to correspond to cells within cluster 4 radial glia.Cell types that were not part of the trajectory were removed for visualization.For region specification analysis, pseudotime trajectory analysis for cortical development was performed and plotted with region specific signature genes score (cortical signature gene list: TNNC1, MYL4, CALB2, RELN, LHX1, LHX5, EBF3, SATB2, DOK5, EMX1, TIAM2, TBR1, FOXG1; hippocampal signature gene list: ZBTB20, GLIS3, DCX, NRP2, GRIA1, PROX1, NEUROD1, C1QL2, ELAVL2, ELAVL4, DKK3, POU3F1, and SPOCK1) [38,39].

Neuron and astrocyte isolation by FACS
Cerebral organoids were dissociated into single cells following the same cell isolation protocol described above.All steps were performed using sterile techniques.Dissociated cells were resuspended in ice-cold PBS containing 0.05% BSA and passed through 30 µm cell strainer.Cell suspension was then centrifuged at 300 g for 5 min at 4 °C.Cell pellet was resuspended in basal medium plus 0.5% BSA (basal medium: phenol-red-free DMEM/ F12 + Neurobasal + 1 × N2 + 1 × B27) and incubated on ice for 15 min.CD90 and GLAST were used to sort neurons and astrocytes from cerebral organoids, respectively [40,41].Cells were washed with basal medium and resuspended at a concentration of 1 × 10 7 cells/mL in basal medium plus 0.05% BSA in a 5 mL round bottom tube.Human anti-CD90-PE antibody (Miltenyi Biotec, 130-117-388) and anti-GLAST-APC antibody (Miltenyi Biotec, 130-123-641) were added according to manufacturers' recommendations and incubated on ice for 30 min.Stained cells were washed with basal medium for three times, centrifuged at 300 g for 5 min, and resuspended with basal medium plus 0.05% BSA at 2.5 × 10 6 cells/ml for FACS sorting (BD FACS Aria).Sytox Blue (1:1000 dilution, ThermoFisher S34857) was used as the viability dye.Cells were sorted on a FACS Aria II (BD Biosciences) equipped with 405 nm, 561 nm, and 633 nm lasers, using the 100 micron nozzle with 20 psi sort pressure.In preliminary and antibody titration experiments, FMO and single-stain controls were found to be equivalent, therefore sort gates were drawn based on single-stained controls.Sort targets were determined by identifying cells on FSC-A versus SSC-A, then performing doublet discrimination with hierarchical SSC-H versus SSC-W and FSC-H versus FSC-W plots, then identifying lives cells on a Sytox Blue-BV421-A versus FSC-A plot, and finally interrogating GLAST-APC-A versus CD90-PE-A for CD90 single positive and GLAST single positive cells.CD90 + neurons were harvested for RT-QPCR analysis or seeded on PLO/Laminin coated coverslips for downstream assays.GLAST + astrocytes were plated in Matrigel-coated wells in astrocyte media (ScienCell) and expanded for downstream assays.The purity of sorted cells was confirmed by immunostaining of cell-type-specific markers.

Radial glia/neural stem cell differentiation
Radial glia/neural stem cell was generated using STEMdiff ™ SMADi Neural Induction Kit (Stemcell Technologies, 08581) following manufacturer's instruction.Human iPSCs were dissociated into single cells and seeded on AggreWell 800 plates in EB formation medium (Stemcell Technologies, 05893) to initiate EB formation.After 24 h, EB formation medium was exchanged to neural induction medium with daily half medium change for 3-4 days.Next, EBs were collected and replated onto Matrigel-coated dishes and cultured in neural induction medium for another 5-7 days to induce neural rosette formation.Neural rosettes were isolated as a single-cell suspension and replated onto Matrigel-coated dishes in neural induction medium for another 2-3 days.Radial glia/neural stem cells were maintained and amplified in neural progenitor cell medium (Stemcell Technologies, 05833) for further experiments.

Tissue processing
Cerebral organoids or cells were lysed with RIPA Lysis and Extraction Buffer supplemented with Protease and Phosphatase Inhibitor Cocktails for Cell Lysis (Roche).Samples were kept on ice for 60 min after sonication, and then centrifuged in an ultracentrifuge (Beckman-Coulter) at 100,000 g, for 1 h at 4 °C.Supernatants were collected for Western blotting analysis.Total protein concentration in the soluble fraction was determined using a Pierce BCA Protein Assay Kit.

Immunofluorescence staining
Cerebral organoids and specific cell types were harvested and fixed in 4% paraformaldehyde for 30 min then washed with PBS three times.After fixation, cerebral organoids were dehydrated with 30% sucrose in PBS at 4 °C.Optical cutting temperature (OCT) compound (VWR) was used to embed cerebral organoids and frozen on dry ice.Tissue was sectioned at 30 μm and collected on glass slides and stored at − 20 °C.For immunostaining, tissue sections or fixed cells were permeabilized in 0.25% Triton X-100 and blocked with blocking buffer containing 4% normal donkey serum, 2% BSA and 1 M glycine in PBS.Sections were then incubated with primary antibodies in blocking buffer overnight at 4 °C.Primary antibodies and their dilutions used in this study are as follows: Tuj1 (Sigma, T2200, 1:1000), GFAP (Millipore, MAB360, 1:300), S100β (Abcam, ab52642, 1:100), Nestin (Abcam, ab18102, 1:500), Pax6 (Biolegend, 901,302, 1:300), and CTIP2 (Abcam, ab18465, 1:100).On the following day, sections were washed three times with PBS, and then incubated with secondary antibodies for 2 h at room temperature.Finally, sections were washed three times with PBS before mounting with the glass coverslip.After washing three times with PBS, samples were incubated with fluorescently conjugated secondary antibodies (Alexa Fluor 488 and 594 conjugates, Invitrogen, 1:500) for 2 h at room temperature and washed three times with PBS before mounting with the glass coverslip.Sections were washed three times in PBS and mounted with Vectashield (H-1000, Vector Laboratories).Fluorescent signals were detected by Keyence fluorescence microscopy (model BZ-X, Keyence) and quantified using ImageJ software.The S100β + or CTIP2 + areas were quantified with fixed threshold and then normalized to the area of DAPI in the same image.

Filipin III and LipidTOX staining
Cholesterol Assay Kit (Cell-Based) (Abcam, ab133116) and HCS LipidTOX ™ Red Neutral Lipid Stain (Invitrogen, H34476) were used for cholesterol and lipid droplet detection, respectively.Cells were fixed using 4% PFA for 30 min, then incubated with Filipin III or LipidTOX solution prepared according to manufacturer's recommendations for 60 min and examined immediately on a confocal laser scanning fluorescent microscopy (model LSM880 Invert, Carl Zeiss) using 40 × oil (EC Plan-Neofluar 40x/1.30Oil DIC M27) or 63 × oil (Plan-Apochromat 63x/1.40Oil DIC M27) objectives with numerical aperture of 1.0.For visualization, excitation of Filipin was at 405 nm and LipidTOX was excited at 594 nm.The Filipin average intensities were quantified using Fiji software.Lipid droplets were analyzed with custom scripts written in MATLAB (r2020b).Briefly, images were thresholded with constant thresholds for each individual channel.Droplets were detected by binarizing images, applying area filter, applying MATLAB's imfindcircles function, and then applying watershed transform on selected circular features.

RT-qPCR
Trizol/chloroform method was used to extract RNA from organoids and different cell types, followed by DNase and cleanup using the RNase-Free DNase Set and the RNeasy Mini Kit (QIAGEN).The cDNA was prepared with the iScript cDNA synthesis kit (Bio-Rad).Real-time qPCR was conducted with Universal SYBR Green Supermix (Bio-Rad) using an iCycler thermocycler (Bio-Rad).Relative gene expression was normalized to ACTB gene coding β-actin and assessed using the 2 − ∆∆CT method.Primers used to amplify target genes by RT-qPCR are as Additional file 1: Table S1.

Statistical analyses
For cerebral organoid comparison between two groups, the student's t test was performed to determine the significance using GraphPad Prism.All statistical tests were two-sided.Data were presented as Mean ± SEM.A p value of < 0.05 was considered statistically significant.Specific statistical methods, the number of the experiments, and the significance levels for each analysis are described in the legends of individual figures.
In the brain, apoE is manly produced by astrocytes, activated microglia, vascular mural cells and choroid plexus cells, and to a lesser extent by stressed neurons [6].To evaluate the cell-type-specific APOE expression in the cerebral organoids, APOE gene expression levels in different cell clusters were plotted based on the scRNA-seq data (Additional file 1: Fig. S2).In the parental control cerebral organoids, APOE is predominantly expressed in radial glia and astrocyte populations, with subtle expression in the neuronal population (Additional file 1: Fig. S2a), while apoE immunostaining showed relative diffuse distribution (Additional file 1: Fig. S2b).These results indicate that glia cells are the main source of apoE production in the iPSC organoids.APOE expression was not detected in any cluster of APOE −/− cerebral organoids, immunostaining showed background apoE expression in APOE −/− cerebral organoids.

APOE deficiency alters neural differentiation fate in cerebral organoids
The cerebral organoids differentiated from iPSCs have been found to resemble the developing brain and exhibit different cerebral layers expressing specific markers [44,45].While investigating the effects of APOE deficiency on the cellular composition of cerebral organoids, we found dramatic differences between parental and APOE −/− cerebral organoids (Fig. 1c).Specifically, excitatory neuron population was significantly reduced by APOE deficiency, whereas radial glia, astrocyte and inhibitory neuron populations were increased (Fig. 1d).To validate the cellular composition changes in the APOE −/− cerebral organoids, the expression levels of cell-type-specific markers were evaluated via RT-qPCR and immunostaining.Consistent with the scRNA-seq data, the mRNA levels of excitatory neuronal marker (SLC17A7), deep layer neuron markers (TBR1 and BCL11B/CTIP2), and upper layer neuron marker (SATB2) decreased in the APOE −/− cerebral organoids (Fig. 1e), indicating that APOE deficiency inhibits neuronal layer formation in cerebral organoids.On the other hand, the marker for ventricular zone (SOX2) and astrocytic markers S100B and GFAP increased significantly (Fig. 1f-g, Additional file 1: Fig. S3a).Immunostaining also showed a significant increase in the ratio of S100β-positive astrocytes in the APOE −/− cerebral organoids (Fig. 1f ).Pseudotime trajectory analysis also revealed distinct differentiation trajectories between parental and APOE −/− cerebral organoids (Additional file 1: Fig. S3b).Region specific analysis showed low hippocampal identity in both control and APOE −/− cerebral organoids, whereas the cortical identity increased in cerebral organoids along with the maturation of neurons (Additional file 1: Fig. S3c).Together, these results suggest an important role of apoE in neural fate regulation in the iPSC-derived cerebral organoids.

Impaired neural differentiation in APOE-deficient cerebral organoids is associated with activated EIF2 signaling pathway
To investigate how APOE deficiency affects gene expression and pathways in different cell types within the cerebral organoids, the differentially expressed genes (DEGs) in the main cell clusters were identified and subjected to the pathway analysis (Fig. 2a-h, Additional file 1: Fig. S4).The two major cell populations, excitatory neuron and radial glia clusters, showed an overwhelming upregulation of cellular stress-related pathways, including "EIF2 signaling pathway" and "mTOR signaling pathway" (Fig. 2b, f ).Consistently, Western blotting confirmed the significant increase of eukaryotic initiation factor 2α (eIF2α) phosphorylation in APOE −/− cerebral organoids compared to the controls (Fig. 2i, Additional file 1: Fig. S7 for raw data).To further investigate the contribution of EIF2 signaling pathway to the phenotypic changes, APOE −/− cerebral organoids were treated with the ISRIB which blocks the phosphorylation of eIF2α in the integrated stress response [46] from Day 60 for 30 days.While Western blotting showed a significant reduction of eIF2α phosphorylation by ISRIB in the cerebral organoids (Fig. 2j, Additional file 1: Fig. S8 for raw data), the ISRIB administration restored the altered mRNA levels of neuronal markers (DCX, SLC17A7 and TBR1) and astrocytic markers (S100B and GFAP) in APOE −/− cerebral organoids (Fig. 2k).Immunostaining confirmed that ISRIB administration increased CTIP2-positive neurons in the cerebral organoids (Fig. 2l).Together, these results indicate that APOE deficiency leads to altered neuronal differentiation in the cerebral organoids by activating EIF2 signaling pathway.

Activated Wnt/β-catenin signaling in astrocytes from APOE-deficient cerebral organoids
In astrocyte clusters, "Axonal guidance signaling" and "Synaptogenesis signaling" were identified as the top ranked pathways enriched by DEGs (Fig. 2h), both excitatory neuron and astrocyte clusters showed changes in "Sirtuin signaling", indicating the important role of astrocyte-derived apoE in neuronal development and maintenance.In addition, the Wnt/β-catenin signaling pathway was also predominantly affected by APOE deficiency in radial glia and astrocyte populations (Fig. 2f,  h).To validate the scRNA-seq results, astrocytes were isolated from the dissociated cerebral organoids by FACS via GLAST1-mediated sorting (Fig. 3a).SFRP1 encodes a Wnt signaling modulator protein.It was one of the most significantly changed genes by APOE deletion in radial glia (Fig. 2e) and astrocyte clusters (Fig. 2g), we confirmed that APOE deficiency reduced SFRP1 expression in the isolated astrocytes by RT-qPCR (Fig. 3b).Western blotting also showed the decreased levels of SFRP1 as well as increase of β-catenin and Wnt7b in the isolated APOE −/− astrocytes compared to the controls (Fig. 3c-f, Additional file 1: Fig. S9 for raw data).Since SFRP1 is secreted by glia cells and has been shown to reduce neurogenesis by inhibiting Wnt signaling [47,48], these results suggest that APOE deficiency may also impact cellular composition in the iPSC-derived cerebral organoids through Wnt/β-catenin signaling pathway by suppressing astrocytic SFRP1 production.

Cell-type-specific modification of cholesterol biosynthesis pathway in APOE-deficient cerebral organoids
ABCA1 and ABCG1 genes belong to the ATP-binding cassette family and encode the cholesterol transporter proteins that play key roles in cholesterol and phospholipid homeostasis, Cholesterol efflux via ABCA1 and ABCG1 is a mechanism for cells to eliminate excess cholesterol and prevent cellular cholesterol accumulation [7,49].Our scRNA-seq analysis found increased expression of ABCA1 and ABCG1 in the astrocyte clusters of APOE −/− cerebral organoids (Fig. 2g).Western blotting showed significant increases in the protein levels of ABCA1 and ABCG1 in astrocytes isolated from APOE −/− cerebral organoids (Fig. 3g-l Additional file 1: Fig. S10 for raw data).However, greater cholesterol accumulation (Fig. 3j) and increased lipid droplet number/size (Fig. 3k) were observed in the APOE −/− astrocytes compared to controls when stained for Filipin III and LipidTOX, respectively.Thus, increases of ABCA1 and ABCG1 gene expression in APOE-deficient astrocytes may be induced as a compensatory mechanism against excess intracellular lipid accumulation.
On the other hand, we observed an enrichment of lipid metabolism and cholesterol biosynthesis pathways in the 5 clusters of excitatory neurons, especially in cluster 3, with key cholesterol synthesis-related genes significantly upregulated (Fig. 2c-d, Additional file 1: Fig. S4).To validate these scRNA-seq results, excitatory neurons were isolated from the dissociated cerebral organoids by FACS via CD90-mediated sorting, where the purity of isolated neurons was confirmed by immunostaining of neuronal marker (Tuj1) (Fig. 4a).Consistent with the results from scRNA-seq, RT-qPCR showed significant increases in the mRNA levels of multiple enzymes involved in the cholesterol biosynthesis pathway in isolated neurons from APOE −/− cerebral organoids (Fig. 4b).Nonetheless, when intracellular cholesterol was stained with Filipin III [50] Fig. 3 Cholesterol and lipid droplet accumulation are increased in astrocytes isolated from APOE-deficient iPSC-derived cerebral organoids.a Astrocytes were isolated from the cerebral organoids through FACS using the cell surface marker GLAST1.Representative fluorescence microscopy images of the isolated astrocytes stained with antibodies against S100β and GFAP.Scale bar: 20 μm.b SFRP1 mRNA expression in the isolated astrocytes was quantified by RT-qPCR (n = 6 wells/genotype).c-f Protein levels of SFRP1 (d), β-catenin (e) and Wnt7b (f) in the isolated astrocytes were quantified by Western blotting (n = 6 wells/genotype).g-i Protein levels of ABCA1 (h) and ABCG1 (i) in the isolated astrocytes were quantified by Western blotting (n = 4 wells/genotype).j The isolated astrocytes were plated on coverslips and stained with Filipin III for cholesterol.Filipin III intensities were quantified in 3 fields of each coverslip and averaged (n = 4 coverslips/genotype), Scale bars, 20 μm.k The isolated astrocytes were plated on coverslips and stained with LipidTox for lipid droplets.The lipid droplet number and size per cell were quantified in 3 fields of each coverslip and averaged (n = 5 coverslips/genotype). Scale bars: 10 μm.Experiments were repeated in two independently differentiated batches.All data are expressed as mean ± SEM.Student's t tests were performed to determine statistical significance.**p < 0.01, **** p < 0.0001.For raw data see Additional file 1: Figs.S9 and S10 in the isolated neurons, there were no significant changes between control and APOE −/− neurons (Fig. 4c).
In contrast to the excitatory neuron population, we found that genes related to cholesterol biosynthesis were rather downregulated in both radial glia and astrocyte population (Additional file 1: Fig. S5).Since radial glial cells have been identified as adult neural stem cells in the subventricular zone, which is the major source of neurons and astrocytes during development [51], we differentiated the iPSCs into radial glia/neural stem cells to validate the scRNA-seq results.Immunostaining of specific markers (Nestin and PAX6) showed the successful radial glial differentiation (Fig. 4d).Consistent with the scRNA-seq results, mRNA levels of multiple cholesterol synthesis genes decreased in APOE −/− radial glia/neural stem cells when analyzed by RT-qPCR (Fig. 4e), although Filipin III staining did not detect significant differences between control and APOE −/− iPSC-derived glia/neural stem cells (Fig. 4f ).Taken together, these results suggest that APOE deficiency differently impacts cholesterol metabolism depending on cell types, which may influence neuronal differentiation in the iPSC-derived cerebral organoids.Repeated experiments using another set of parental and APOE −/− isogenic lines confirmed the major findings including neurogenesis deficits and lipid metabolism dysregulation due to APOE deficiency (Additional file 1: Fig. S6).

Altered neurogenesis and cholesterol metabolism in APOE4 cerebral organoids
APOE4 has been associated with cell-type-specific functional abnormalities in AD brain, including neurogenesis deficits, impaired synaptic function, neuronal degeneration, cholesterol lipid metabolism dysfunction, and inflammatory response [9,31,50,52].Thus, we explored the APOE4 effects using the iPSC-derived cerebral organoids from an AD patient carrying APOE ε4/ ε4 (APOE4) and the corresponding APOE ε3/ε3 (APOE3) isogenic line at Day 90.While evaluating the cellular composition differences between APOE3 and APOE4 organoids, RT-qPCR found decreased neuronal lineage marker mRNAs and higher astrocytic marker mRNAs in APOE4 organoids compared to those in APOE3 organoids (Fig. 5a).In addition, Western blotting showed enhanced eIF2α phosphorylation levels in the APOE4 organoids (Fig. 5b, Additional file 1: Fig. S11 for raw data).When neurons and astrocytes were isolated from the organoids via FACS sorting, we found that the expression of selected genes related to cholesterol biosynthesis was increased in neurons (Fig. 5c) but decreased in astrocytes (Fig. 5d) from APOE4 organoids compared to those from APOE3 organoids by RT-qPCR.Filipin III staining showed higher cellular cholesterol levels in both neurons (Fig. 5e) and astrocytes (Fig. 5f ) isolated from APOE4 organoids compared to those from APOE3 organoids.Taken together, these results indicate that APOE4 impairs neurogenesis and cholesterol metabolism in the iPSC cerebral organoids.The phenotypic deficits in APOE4 organoids can largely be reversed by genome editing APOE4 to APOE3.

Discussion
The iPSC-derived cerebral organoids provide a unique model system in which human cell-type specification, self-organization, and heterogeneous intercellular communication occur simultaneously [28,29].In this study, we investigated how APOE deficiency influences the molecular pathways and cellular functions in the organoids by scRNA-seq combined with multiple validation approaches.While several cell types were identified in cerebral organoids consistent with previous findings [29,43], we found that APOE deficiency dramatically modulates the cellular compositions, accompanied by enhanced cellular stress and lipid dyshomeostasis.Our scRNA-seq data revealed an increase of radial glia and astrocyte populations, but a decrease of excitatory neurons in isogenic APOE −/− cerebral organoids.Furthermore, we found a significant increase of inhibitory neurons in APOE −/− cerebral organoids.Interesting, such developmental excitation-inhibition imbalance has been observed in several other disease-related iPSC cerebral organoid model systems [53][54][55].Indeed, apoE has been identified as an important regulator balancing brain progenitor cell fate and regulating neurogenesis in the dentate gyrus [56][57][58].At the early developmental stage, apoE is essential for inhibiting cell proliferation, maintaining neural precursor characteristics and promoting neurogenesis [56].Apoe knockout mice also show the excessive proliferation of neural stem cells and a shift of neural differentiation from neurogenesis to astrogenesis [56,58].Thus, it would be interesting to further define the molecular mechanisms by which apoE regulates neuronal fate, providing insight into the roles of apoE isoforms in AD and other neurological diseases.
Although the role of apoE in lipid metabolism has been intensively investigated, there is still a lack of comprehensive assessment on how apoE regulates lipid homeostasis in a cell-type-specific manner.As apoE predominantly mediates lipid efflux, our results showed that APOE deficiency results in excess cholesterol accumulation and lipid droplet formation in astrocytes isolated form iPSC-derived cerebral organoids.The increased levels of ABCA1 and ABCG1 in the APOE −/− astrocytes are likely due to a compensative reaction to excessive intracellular lipid accumulation.Intriguingly, cholesterol has been shown to activate canonical Wnt/β-catenin signaling [59], consistent with our finding in astrocyte population through scRNA-seq.Thus, the increase of astrocyte proportion in APOE −/− cerebral organoids may be induced at least partially by the activated Wnt/β-catenin pathway in astrocytes or radial glia as Wnt/β-catenin signaling has been shown to be involved in the differentiation process of reactive astrocytes upon brain injury [60].In addition, the lipid overload status is presumably induced in radial glia/neural stem cells when apoE-mediated lipid efflux is suppressed, since the reduction of cholesterol synthesis-related genes was observed in radial glia as well as astrocyte populations in the APOE −/− organoids.While lipid droplets are highly abundant in neural stem cells under physiological conditions [61] which correlates with increased proliferative ability [62], the excess lipid accumulation in neural stem cells may steer their differentiation fate into astrocytic lineage although further studies are needed.
Since neurons rely on astrocytic apoE for cholesterol supplies to maintain their functions and homeostasis [5,63], neuronal populations may be vulnerable to the relatively stressful condition of the organoid models in the absence of apoE.Indeed, APOE −/− cerebral organoids showed upregulation of cellular stress-related pathways, including EIF2 signaling and mTOR signaling.We also found significant increase of cholesterol biosynthesis genes in excitatory neuron population in APOE −/− cerebral organoids, which likely reflects a compensatory reaction to the insufficient lipid supplies due to the lack of astrocytic apoE.Consistent with our finding, astrocytic apoE has been shown to specifically silence genes involved in neuronal cholesterol biosynthesis [5].As cholesterol biosynthesis requires multiple enzymatic steps that involve high energy consumption, the emergent condition may also exacerbate ER stress and autophagy in neurons, thereby influencing the cellular composition in APOE −/− cerebral organoids.Of note, we demonstrated that the administration with eIF2 phosphorylation inhibitor ISRIB restores the altered cellular composition in the APOE −/− cerebral organoids by facilitating neurogenesis and suppressing astrogliosis.Since eIF2α phosphorylation is the central event in the integrated stress response (ISR) [64], our findings indicate the predominant involvement of ISR cause by APOE deficiency in defining neuronal differentiation.
ApoE4 has been shown to mediate cholesterol efflux in astrocytes less efficiently than APOE3, resulting in the exacerbated accumulation of cholesterol and lipid droplets [22,50,65].We also found that the expression of cholesterol synthesis gene is lower in neurons and higher in astrocytes isolated from APOE4 cerebral organoids than those with APOE3, which is accompanied by the increase of intracellular cholesterol accumulation.Since APOE4 organoids showed similar phenotypes with APOE −/− cerebral organoids in cholesterol metabolism, our results suggest that APOE4 may cause lipid dyshomeostasis through loss-of-function effects compared to APOE3.In addition, APOE4 cerebral organoids also possessed diminished neuronal differentiation similar to the APOE −/− organoids.Consistently, studies in mouse models have found that both Apoe deficiency and human APOE4 isoform impair the hippocampal neurogenesis [57,58].Of note, our group recently described a downregulation of Wnt/β-catenin pathway in human iPSC-derived astrocytes carrying APOE4 compared to those carrying APOE3 [66], which is opposite to the effects of APOE deficiency.The discrepancy in the effects of APOE4 and APOE deficiency on brain cell differentiation requires further investigation in future.Importantly, since the significant increase of the phosphorylation level of eIF2α was observed in APOE4 cerebral organoids, our results suggest that ISR inhibition can be a potential therapeutic target for APOE4-mediated pathogenesis in AD.Indeed, several recent studies show that ISRIB could restore hippocampal protein synthesis and delay cognitive decline in aged mice and AD mouse models [67,68].

Conclusions
Our study demonstrates the essential role of apoE in brain homeostasis using the iPSC-derived cerebral organoids and offers critical insights into the underlying mechanisms of APOE4-related phenotypes.APOE deficiency and APOE4 influence lipid metabolism and neuronal differentiation by activating ISR pathway in a cell-type-specific manner.One limitation of our study is the ex vivo nature of iPSC-derived cerebral organoids lacking systemic effects compared to in vivo environment.Another limitation of the cerebral organoid system is the lack of several important brain cell types, such as microglia and vascular cells [29,33].Thus, future studies should be directed to validating our findings in human brains and optimized iPSC-derived cerebral organoid systems with the incorporation of additional brain cell types.Together, our findings will provide novel insights on the underlying mechanisms in APOE4-associated disease pathogenesis and guide the development of apoEtargeted therapy.S1.Primer information for RT-qPCR.
• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year

•
At BMC, research is always in progress.

Learn more biomedcentral.com/submissions
Ready to submit your research Ready to submit your research ?Choose BMC and benefit from: ? Choose BMC and benefit from:

Fig. 1
Fig. 1 APOE deficiency leads to cellular composition changes in iPSC-derived cerebral organoids.a-d Parental control and isogenic APOE −/− iPSCs were differentiated into cerebral organoids and dissociated into single cells at Day 90 for scRNA-seq analysis.Three cerebral organoids were pooled and analyzed as one sample (n = 3 samples/genotype).a t-SNE plot and cluster identification of scRNA-seq data from all samples.Cell clusters were defined based on the expression of cluster marker genes and known marker genes.ExN, excitatory neuron; RG, radial glia; InN, inhibitory neuron; IPC, intermediate progenitor cell; Astro, astrocyte; UD, undecided.b Dot plot of canonical genes to classify t-SNE clusters.Cluster identities are labeled on the left and canonical marker genes are indicated on the bottom.c Percentage of cell clusters in each cerebral organoid sample.d Comparison of the percentage of different cell types between control and APOE −/− cerebral organoids.e-g The mRNA levels of cerebral layer markers (e; SLC17A7, TBR1, BCL11B, and SATB2), neural stem cell marker (f; SOX2) and astrocytic markers (g; S100B and GFAP) were quantified by RT-qPCR.Three cerebral organoids were pooled and analyzed as one sample (n = 6-9 samples/genotype).h Astrocyte differentiation in cerebral organoids were evaluated by S100β immunostaining (n = 4 organoids/genotype).Scale bar: 50 μm.Experiments were repeated in two independently differentiated batches (e-h).All data are expressed as mean ± SEM.Student's t tests were performed to determine statistical significance.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fig. 2
Fig. 2 APOE deficiency alters neural differentiation in cerebral organoids by activating the EIF2 signaling pathway.a-h DEGs in cell clusters were identified through the scRNA-seq and subjected to the Ingenuity pathway analysis.Volcano plot and gene ontology analyses for DEGs (APOE −/− vs. Con) in total excitatory neuron clusters (a, b), excitatory neuron cluster 3 (c, d), radial glia clusters (e, f), and astrocyte clusters (g, h) are shown.Genes significant at the P value ≤ 0.05 and fold change ≥ 1.2 are denoted in purple.Selective DEGs are labeled in the volcano plots.i Phosphorylation levels of eIF2α in the control and APOE −/− cerebral organoids at Day 90 were quantified by Western blotting.Three cerebral organoids were pooled and analyzed as one sample (n = 3 samples/genotype).j APOE −/− cerebral organoids at Day 60 were treated with DMSO or ISRIB (100 nM) for 30 days.Phosphorylation level of eIF2α in the cerebral organoids was quantified by Western blotting.Three cerebral organoids were pooled and analyzed as one sample (n = 6 samples/genotype).k The mRNA levels of cerebral layer markers (DCX, SLC17A7, TBR1 and CTIP2) and astrocytic markers (S100B and GFAP) were quantified by RT-qPCR.Three cerebral organoids were pooled and analyzed as one sample.All data are expressed as mean ± SEM (n = 4-5 samples/genotype).l Representative microscopy images of cerebral organoids stained with neuronal layer V marker CTIP2 and DAPI.The immunoreactivity of CTIP2 was quantified and normalized by DAPI fluorescent intensity (n = 6 organoids/genotype).Scale bar: 50 μm.Experiments were repeated in two independently differentiated batches (i-l).All data are expressed as mean ± SEM.Student's t tests were performed to determine statistical significance.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.For raw data see Additional file 1: Figs.S7 and S8

Fig. 4
Fig. 4 Differential effects of APOE deficiency on gene expression related to cholesterol biosynthesis in excitatory neurons and radial glia.a Neurons were isolated from the cerebral organoids through FACS using the cell surface marker CD90.Representative fluorescence microscopy images of the isolated neurons stained with antibodies against Tuj1.Scale bar: 20 μm.b The mRNA levels of selective cholesterol biosynthesis genes in the isolated neurons were quantified by RT-qPCR (n = 4 wells/genotype).c The neurons isolated from cerebral organoids were plated on coverslips and stained with Filipin III for cholesterol.Filipin III intensities were quantified in three fields of each coverslip and averaged (n = 4 coverslips/genotype), Scale bars, 20 μm.d Parental and APOE −/− isogenic iPSCs were differentiated into radial glia/neural stem cells.Representative fluorescence microscopy images of the differentiated radial glia/neural stem cell stained with antibodies against Nestin and PAX6.Scale bar: 20 μm. e The mRNA levels of selective cholesterol biosynthesis genes in the radial glia/neural stem cell were quantified by RT-qPCR (n = 6 wells/genotype).f The radial glia/neural stem cells were plated on coverslips and stained with Filipin III for cholesterol.Filipin III intensities were quantified in three fields of each coverslip and averaged (n = 4 coverslips/genotype). Scale bars: 20 μm.Experiments were repeated in two independently differentiated batches.All data are expressed as mean ± SEM.Student's t tests were performed to determine statistical significance.**p < 0.01, ***p < 0.001, n.s., not significant