Back

SALL1 and neural microenvironmental contextspecify identity of stem cell derived humanmicroglia-like cells

Nikolova, M. T.; Freitag, K.; Okamoto, R.; Catellani, A.; Hinke, J.; Luginbuehl, N. A.; Lin, H.-C.; Kaindl, J.; He, Z.; Wahle, P.; Rohland, J.; Geirsdottir, L.; Camp, J. G.; Winner, B.; Amit, I.; Treutlein, B.

2026-09-28 developmental biology
10.64898/2026.09.25.754357 bioRxiv
Show abstract

Microglia, the immune cells of the brain parenchyma, play a pivotal role in neurodevelopment and neuroprotection. Human stem cell-derived microglia-like cells (MGL) offer a valuable, human-relevant system to study microglia in vitro. However, their cell states and intercellular interactions within complex 3-dimensional (3D) tissue contexts in vitro remain underexplored. Here, we analyze the gene expression profiles of MGL generated by two differentiation protocols in conventional 2D cultures. We show that co-culture with neural organoids promotes a more mature, in vivo-like MGL phenotype but this signature is lost during long-term co-culture. By varying media compositions, we modulate 2D MGL states toward more active or more homeostatic identities, which may enhance the similarity to primary microglia identities upon subsequent co-culture. Moreover, forced expression of the microglial lineage regulator SALL1 shifts MGL away from a general myeloid state toward a more defined pre-microglial fate. SALL1 overexpression within a 3D neural organoid environment further amplifies the shift toward a homeostatic developmental microglia population. Collectively, our transcriptomic analysis delineates human MGL states in 2D and 3D contexts, refines existing differentiation protocols toward more authentic MGL, and establishes a framework to investigate the intricate microglia-neural tissue interplay.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.