Generated editorial illustration of the study's perturbation-to-myelination workflow. It is not microscopy, a data figure, or a treatment claim. Credit: CellXperience.
A genome-wide screen has placed Wnt signaling between APOE4 and an oligodendrocyte filled with excess lipid.
The study, published online in Cell Stem Cell on October 15, began with human iPSC-derived oligodendrocytes and moved through chromatin accessibility, genetic perturbation, three-dimensional myelinoids, and a tau-transgenic mouse model. Its main intervention was GSK3B inhibition. Across those systems, the research compound CHIR99021 reduced lipid-droplet measurements and increased myelin-associated staining.
The evidence builds a mechanistic chain, rather than a treatment case. No person received the compound, the mouse experiments did not measure cognition, and GSK3B regulates many processes beyond the pathway highlighted here.
A screen built around lipid burden
The researchers differentiated an APOE4-homozygous iPSC line into oligodendrocytes, then infected 90 million cells with a genome-wide CRISPR knockout library targeting protein-coding genes. After ten days, live cells were stained with BODIPY and sorted into the highest and lowest 15 percent of lipid-droplet fluorescence. Sequencing the guide RNAs enriched in each fraction produced hundreds of candidate regulators.
Wnt genes appeared in a coherent direction. Knocking out negative regulators including DKK1 and NKD1 was associated with the low-lipid fraction, while disruption of canonical pathway components such as FZD10 and WNT16 appeared in the high-lipid fraction. Independent perturbations then supported the screen: DKK1 knockdown and beta-catenin overexpression each lowered lipid-droplet counts in APOE4 oligodendrocytes.
The signal extended beyond the knockout library. Bulk RNA sequencing of unperturbed cells sorted by lipid content found Wnt-related expression differences. Isogenic APOE3 and APOE4 oligodendrocytes entered an ATAC-sequencing comparison, using two replicates per genotype. The analysis identified 47,999 differentially accessible regions mapped to 9,671 protein-coding genes. Regions around Wnt and myelination programs were less accessible in APOE4 cells.
Two replicates per genotype leave the chromatin estimate vulnerable to line- and batch-specific effects. The team added an external human dataset: single-nucleus ATAC sequencing from prefrontal cortex of 93 aged donors. Wnt-related motifs were enriched among regions with lower accessibility in APOE4 carriers, particularly in oligodendrocyte-lineage cells and neurons. That convergence makes a purely culture-specific explanation less likely, though postmortem association cannot establish causality.
GSK3B offers a pharmacologic handle
Beta-catenin protein was lower in APOE4 oligodendrocytes, while GSK3B immunoreactivity was higher. Because GSK3B promotes beta-catenin degradation, the authors used the kinase as an intervention point.
Activating GSK3B with Dif3 increased lipid droplets in APOE3 oligodendrocytes. Treating APOE4 cells with 1 micromolar CHIR99021 for one week lowered BODIPY-positive droplets. Stimulated Raman scattering microscopy also found less phospholipid and cholesterol signal within individual droplets.
RNA sequencing after GSK3B inhibition showed lower expression of several cholesterol-synthesis genes and higher expression of oxidative-phosphorylation, lysosomal-lipid, and autophagy-related programs. Live imaging found more contact between lipid droplets and mitochondria, without a corresponding increase in droplet-lysosome colocalization. A fluorescent phospholipase substrate produced less signal after treatment.
Those measurements suggest several routes through which the compound could alter stored lipids. They do not yet identify the indispensable route. CHIR99021 is a selective and widely used research inhibitor; the experiment did not isolate a single downstream mechanism that explains every lipid and myelin result.
Myelinoids connect lipid biology to a tissue feature
The team generated oligodendrocyte-containing brain organoids from parental APOE3 KOLF2.1J iPSCs and a CRISPR-edited, isogenic APOE4 line. At 20 weeks, APOE4 myelinoids carried less myelin-basic-protein immunoreactivity than APOE3 controls.
Treatment began at week 18. APOE4 myelinoids received 2 micromolar CHIR99021 for two weeks. Myelin-basic-protein staining increased, while the number of Olig2-positive cells did not change. Neurofilament staining also rose. The readout is compatible with greater myelin production from existing oligodendrocytes and better preservation or growth of axons.
Immunoreactivity is an informative tissue marker, though it does not show compact myelin ultrastructure, conduction velocity, or circuit function. The myelinoid experiment also centers on one parental background and its edited counterpart. Replication across unrelated APOE4 donors would test whether the response travels beyond this genetic context.
Two mouse experiments, two endpoints
APOE4;PS19 mice combine a humanized APOE4 background with a tau-transgenic model. The researchers used separate cohorts for lipid and myelin questions.
Five-month-old male mice received 5 milligrams per kilogram of CHIR99021 by intraperitoneal injection twice weekly for two weeks. Hippocampal lipid-droplet staining fell relative to vehicle. Eight-month-old female mice received the same dose schedule for six weeks. Myelin-basic-protein immunoreactivity increased, without a detected change in the number of Olig2-positive cells.
The longer experiment did not change phosphorylated tau at threonine 181 or serine 396. Measures of GFAP, microglial number, and microglial morphology also showed no significant treatment difference. The result is therefore narrower than a general improvement in Alzheimer-like pathology.
Age, sex, duration, and endpoint changed together between the two cohorts. Lipid reduction and myelin staining cannot be read as a single longitudinal sequence. The manuscript describes blinded tissue analysis but does not establish whether altered staining produced behavioral or electrophysiological benefit.
The therapeutic gap remains wide
GSK3B participates in glycogen metabolism, development, inflammation, synaptic biology, and tau phosphorylation. Systemic inhibition could produce effects far outside oligodendrocytes. The study used intermittent dosing over weeks in mice, while a neurodegenerative therapy might require long exposure. Target engagement, brain distribution, therapeutic window, and cell-type specificity will need careful work.
The target is also upstream of the desired tissue result. A development program would have to connect compound exposure to oligodendrocyte lipid handling, compact myelin, axonal physiology, and behavior. It would need replication in additional human iPSC backgrounds, including heterozygous APOE3/4 lines that reflect many carriers.
The paper's contribution is a tractable bridge from a human-cell genetic screen to tissue-scale assays. Wnt/GSK3B signaling now sits inside an experimentally supported model of APOE4-associated oligodendrocyte dysfunction. The work identifies a direction for target validation; it does not show that GSK3B inhibition can treat Alzheimer's disease.
Source and disclosures
The current event is the peer-reviewed Cell Stem Cell article “Genome-wide CRISPR screen reveals that Wnt signaling regulates lipid accumulation in APOE4 oligodendrocytes”, available online October 15, 2026. The journal article was behind publisher access, so experimental details were checked against the authors' complete matching bioRxiv manuscript, posted October 16, 2025. Claims here are limited to results present in that manuscript and the final abstract. The preprint discloses a patent application by Leyla Anne Akay and Li-Huei Tsai based on the screen; Tsai also reports scientific-advisory roles at 4M Therapeutics and TAC Therapeutics. The hero is an original generated editorial illustration, not primary evidence.