Semaglutide partly protected human forebrain organoids against the structural and metabolic effects of cholesterol stress. The response was more limited in older cultures, where some calcium activity measures improved without restoring tissue size or several disease-associated markers.
Researchers at Tsinghua University used human forebrain organoids to investigate whether the glucagon-like peptide-1 (GLP-1) receptor agonist semaglutide could lessen cholesterol-induced injury. Their study in Alzheimer’s & Dementia compares responses at different stages of maturation, combining structural, metabolic and functional measurements.
The organoids were differentiated from human induced pluripotent stem cells (hiPSCs). By day 30, they contained neuroepithelial rosettes, inner neural progenitor populations and neurons expressing MAP2 in outer regions. The team confirmed GLP-1 receptor expression and used a complementary two-dimensional neural progenitor cell model for viability testing.
The main organoid experiments used 100 nM semaglutide, cholesterol overload and matched controls. Readouts included cell survival, apoptosis, lipid droplets, oxidative stress, gene and protein expression, and calcium activity. A GLP-1 receptor antagonist and pathway inhibitors were used to investigate mechanism. Early exposure under unstressed conditions covered days 14–30; the later model was assessed over days 65–80.
Without cholesterol stress, early semaglutide treatment did not significantly change organoid radius, perimeter, proliferation or apoptosis. Some neurogenesis-related expression increased, while synaptic-transmission pathways decreased, suggesting that developmental responses varied by endpoint.
Cholesterol overload caused lipid accumulation, loss of rosette structure and smaller organoids, together with stronger apoptotic and stress-related signals. Semaglutide co-treatment partly preserved survival, size and architecture. Cleaved caspase-3 decreased and the anti-apoptotic protein Bcl-2 increased. SOX2-positive progenitor and Ki67 proliferation signals were better maintained, while the MAP2-positive neuronal area recovered only partly.
The GLP-1 receptor antagonist compound 5d, or C5d, abolished the protective response measured by live/dead staining. In the reported early-stage comparisons, size measurements used 13 organoids per group, viability assays used six independent experiments and protein measurements used four.
Cholesterol disrupted mitochondrial networks and increased reactive oxygen species. The semaglutide group retained more connected mitochondrial structures and showed lower oxidative stress and lipid-droplet accumulation. Expression and phosphorylation findings implicated cAMP–PKA and PI3K–AKT–mTOR signaling.
Pharmacological inhibition weakened parts of the signaling response, supporting involvement of both pathways. The results suggest interactions between them, but do not establish which pathway acts upstream of the other.
Single-cell RNA sequencing identified eight major populations in day-30 organoids. Cholesterol altered some progenitor populations and lipid- and synapse-related expression; semaglutide partly reversed these changes. CellChat analysis also inferred partial recovery of cell–cell communication patterns, including midkine-associated signaling. These communication changes were inferred from transcriptomic data and were not direct functional measurements of signaling between cells.
In organoids treated from days 65 to 80, cholesterol again reduced tissue area. Semaglutide did not restore overall size or significantly lower the increases in amyloid precursor protein and phosphorylated tau at threonine 217. Some metabolic and stress-related gene changes improved, and fragmented MAP2-positive branches decreased, but total MAP2-positive area did not recover.
Calcium imaging showed more frequent activity events and lower synchrony after cholesterol exposure. Semaglutide brought event frequency toward control values, while recovery of synchrony remained limited. The measurements used Fluo-4 AM in neuron-enriched regions; because the dye is not neuron-specific, they describe cellular calcium activity rather than direct electrical recordings.
The contrast between early structural protection and the narrower response in older cultures suggests that the response depends on the state of the tissue. The authors interpret the effect mainly as stabilization of metabolic and stress responses in this model, with limited reversal of established molecular changes.
The organoid transcriptomes overlapped with selected Alzheimer’s disease signatures, but also with signatures associated with Huntington’s and Parkinson’s diseases. Cholesterol stress therefore reproduced responses shared across neurodegenerative conditions rather than a disease-specific course.
The system lacks aging-related features, microglia, functional vasculature and Alzheimer’s-associated genetic backgrounds. Exogenous cholesterol also bypasses the systemic processes that transport lipids in vivo. The concentrations used in these mechanistic experiments, together with uncertainty about human brain exposure, limit direct extrapolation to treatment effects.
The authors propose testing disease-relevant genetic backgrounds and matched controls, and incorporating vascular or immune components. Such models could help establish when the observed protection persists and which tissue features determine the response.
Researchers exposed human hippocampal organoids to valproic acid for four days, then profiled mRNA and miRNA. They found 19 altered miRNAs and tested selected target sequences in HEK-293T reporter cells, with limited replication from one iPSC line.
Researchers exposed human cortical organoids to acetaminophen for five days and followed structural, transcriptomic and network outcomes for several months. Small early transcriptional changes did not translate into detectable differences in later cellular composition or the measured functional endpoints.
