Renzo Mancuso, Johanna Van Den Daele, Nicola Fattorelli, Bart De Strooper
6 min
Microglia are central to Alzheimer's disease (AD) pathology, yet modeling them in mice is challenging because many human AD-associated genes lack direct mouse orthologs or show functional divergence. This study investigates whether human embryonic stem cell (ESC)-derived microglia can be transplanted into the mouse brain to create a more accurate, humanized model for studying microglial function in disease.
The researchers differentiated human H9 ESCs into microglia-like cells and transplanted them into the brains of immunodeficient (Rag2-/- Il2rγ-/-) mice. To facilitate engraftment, the host mice were pre-treated with a CSF1R inhibitor to deplete endogenous mouse microglia. The researchers then used single-cell RNA sequencing to compare the transcriptomic profiles of these transplanted human cells against primary human microglia and host mouse microglia, both under homeostatic conditions and following an acute challenge with oligomeric amyloid-beta.
The transplanted human microglia successfully integrated into the mouse brain, displaying a ramified morphology and expressing homeostatic markers (TMEM119, P2RY12) similar to primary human microglia. Transcriptomic analysis confirmed that the mouse central nervous system environment effectively drives these cells from an artificial in vitro state toward a natural, homeostatic phenotype. When challenged with oligomeric amyloid-beta, the human microglia exhibited a distinct 'cytokine response' (CRM) state. Crucially, this response differed from that of the host mouse microglia, with many human-specific genes—including several AD risk genes—showing divergent expression patterns. This demonstrates that the humanized model captures aspects of the human microglial response to AD-related pathology that are missed in traditional mouse models.
This study provides a robust proof-of-concept for using humanized chimeric models to study microglial biology. By overcoming the limitations of mouse-human genetic divergence, this approach allows researchers to investigate the function of human-specific AD risk genes in an in vivo environment. This model serves as a powerful tool for dissecting the cellular mechanisms of neurological diseases and could significantly improve the translational relevance of future preclinical studies.
Alex: [analytical, probing] That raises the obvious referee question. How do you know the mouse environment isn't suppressing the very phenotypes you want to study? [[RP_SECTION:species-specific-inflammatory-responses|Species Specific Inflammatory Responses]]
Sam: [grounded, precise] The authors try to address it by comparing the response to amyloid-beta in human and mouse microglia. The correlation is significant but limited. Over two hundred genes respond in opposite directions between the species, including Alzheimer's risk genes such as ABI3 and PICALM. <break time="0.6s" /> That indicates the mouse microglia are not simply a stand-in for the human cells. It does not fully settle the suppression question, though. The human cells sit in a mouse niche, so some residual influence from the host can't be excluded.
Alex: [slower, processing] So where the two species diverge, a mouse-only model could give a misleading read of the mechanism.
Sam: [thoughtful, building the point] That is the working interpretation. The human microglia showed particularly strong upregulation of IL1B and CCL2, cytokines relevant to Alzheimer's pathology, and these did not follow the same trajectory in mouse cells. The value of the model is that the baseline is already human. Forcing the cells into homeostasis through the niche means the amyloid-beta response can be read without the mouse's own immune signalling as a confound. The authors describe this as a bridge, which seems the right word.
Alex: [analytical, curious] What limits the bridge? Presumably a single stem cell line, and the missing adaptive immune system? [[RP_SECTION:model-limitations-and-future|Model Limitations and Future]]
Sam: [grounded, precise] Both are real constraints. The host mice lack an adaptive immune system, which is thought to contribute to neuroinflammation in Alzheimer's. Grafting efficiency also varies across stem cell lines, which adds technical noise. So generalising from one line needs care, and the authors present the model as a bridge, not a replica.
Alex: [probing] And the direction they point to from here?
Sam: [thoughtful, measured] Using CRISPR to introduce patient-specific mutations into the donor cells, then testing therapeutic interventions in a humanised microglial context in vivo. That is prospective, not something this paper demonstrates.
Alex: [reflective, summarizing] So the shift is from modelling the disease generally to asking how human genetic architecture shapes the response.
Sam: [steady, matter-of-fact] Yes, with the caveat that the niche changes the cells and the host changes the readout. The evidence supports a more human-relevant system, not a neutral one.
Alex: [calm, professional] If you want the figures and the method choices we skipped, you can generate a deep dive of this paper. The paper has the rest either way.
Sam: [warm, professional] Thanks for listening.