A pathogenic subpopulation of human glioma associated macrophages linked to glioma progression
bioRxiv. 2025. DOI 10.1101/2025.02.12.637857. Preprint.
How to cite
AMA
Hei Yu KK, Abou-Mrad Z, Törkenczy K, Schulze I, Gantchev J, Baquer G, et al. A pathogenic subpopulation of human glioma associated macrophages linked to glioma progression. bioRxiv. 2025. doi:10.1101/2025.02.12.637857APA
Hei Yu, K. K., Abou-Mrad, Z., Törkenczy, K., Schulze, I., Gantchev, J., Baquer, G., et al. (2025). A pathogenic subpopulation of human glioma associated macrophages linked to glioma progression. bioRxiv. https://doi.org/10.1101/2025.02.12.637857BibTeX
@article{heiyu2025pathogenic,
title = {A pathogenic subpopulation of human glioma associated macrophages linked to glioma progression},
author = {Hei Yu, Kenny Kwok and Abou-Mrad, Zaki and Törkenczy, Kristof and Schulze, Isabell and Gantchev, Jennifer and Baquer, Gerard and others},
journal = {bioRxiv},
year = {2025},
doi = {10.1101/2025.02.12.637857}
}Aggressive gliomas and slower-growing tumors that later transform both converge on the same dangerous behavior, and the immune cells packed into these tumors — glioma-associated macrophages — have long been suspected of driving that shift. The trouble is that these macrophages are a mixed crowd, so no one could say which of them does the damage.
The team combined single-cell RNA and ATAC sequencing to trace the regulatory wiring of these cells and pulled out one subset, governed by the transcription factor FOSL2, that concentrates in high-grade tumors. Two surface markers, ANXA1 and HMOX1, let them physically isolate these cells and test them.
The isolated cells promoted invasion and new blood vessels, nudged nearby T cells toward a suppressive state, and clustered in the tumor's oxygen-starved regions — marking them as a possible target in high-grade glioma.
Key findings
- A FOSL2-centered gene-regulatory network defined a macrophage subset (mGAMs) enriched in high-grade glioma. The network was inferred from parallel single-cell RNA and ATAC sequencing of IDH-wildtype (n=7), IDH-mutant (n=8) and normal-brain (n=2) samples, and low-input ChIP-seq confirmed differential FOSL2 binding at a 38.9% regulon overlap (odds ratio 6.63; p < 2.2e-16).
- ANXA1 and HMOX1 served as surface markers for prospective isolation of mGAMs. Sorted mGAMs showed higher phagocytic capacity than their counterparts (n=8) and, under hypoxia, skewed co-cultured CD4+ T cells toward a FOXP3+ regulatory phenotype.
- Cyclic immunofluorescence placed mGAMs in hypoxic niches within high-grade tumors. t-CyCIF of IDH-wildtype high-grade (n=9), IDH-mutant (n=9) and normal-brain (n=1) specimens found ANXA1+HMOX1+IBA1+ mGAMs aggregating in clusters in glioblastoma and high-grade astrocytomas, co-localized with HIF1α and co-registered into MALDI mass-spectrometry metabolic maps.
- mGAMs shared somatic mitochondrial mutations with peripheral-blood monocytes. Mitochondrial single-cell ATAC-seq (n=2; 650 high-confidence variants) found a higher proportion of monocyte-derived mtDNA variants in mGAMs (37.3%) than in non-mGAMs (34.5%).
CyteFinder II in the methods
“Images were acquired on a CyteFinder II slide scanning fluorescence microscope (RareCyte Inc., USA) with CyteFinder software (v3.11.024). Stitching and registration of tiles and cycles were done in MCMICRO using ASHLAR (v1.10.2) module. Single-cell segmentation was performed using the cell detection feature in QuPath.”
— Hei Yu et al., bioRxiv (2025), Methods, “Cyclic Immunofluorescence”
Why it matters for CyteFinder II users
If you are weighing the CyteFinder II for a tissue study, look at the one job its imaging was asked to do here. The paper's core biology — the FOSL2 regulatory network, the ANXA1/HMOX1 marker pair, the functional and mitochondrial-lineage assays — came from single-cell sequencing, flow sorting, and epigenomics, not from the microscope. What the CyteFinder II carried was the spatial question: once mGAMs were defined by dissociated single-cell data, where did they actually sit inside intact glioma tissue? The instrument acquired multi-cycle t-CyCIF images of ANXA1, HMOX1, IBA1, FOSL2, CD44, SOX2, and HIF1α across fresh-frozen surgical specimens, one staining round at a time, then handed the mosaics to MCMICRO and ASHLAR for stitching and to QuPath for single-cell segmentation. Those maps placed ANXA1+HMOX1+IBA1+ mGAMs in HIF1α+ hypoxic niches and were co-registered with mass-spectrometry images of the same sections. That is what the CyteFinder II adds to a study like this: it returns a dissociated cell signature to its spatial context in situ, and it leaves the stitching and segmentation steps open for you to choose.
