Back

Differential Assembly of Mouse and Human Tumor Microenvironments

Courau, T.; Jaszczak, R. G.; Samad, B.; Flynn, E.; Reeder, G. C.; Chew, N. W.; Tsui, J.; Ray, A.; Wismer, H.; Bunis, D.; Lupin-Jimenez, L.; Gavil, N. V.; Masopust, D.; Graham, J. P.; Skelly, D. A.; Vesco, X.; Liu, E. T.; Fragiadakis, G. K.; Combes, A. J.; Krummel, M. F.

2025-07-05 immunology
10.1101/2025.07.01.661217 bioRxiv
Show abstract

Mouse models are frequently used to develop treatments for human cancer. Yet, we lack a comprehensive understanding of the comparative organization of mouse and human tumor microenvironments (mu/huTMEs). Through immunoprofiling of commonly used mouse models, we found that the immune composition of most muTMEs resemble poorly infiltrated human tumors extensively biased toward high macrophages densities. Relatedly, we discover species-specific biases of chemokine expression networks, factors which drive TMEs assembly. Further, assessing coarse cellular networks, we find conserved correlations between some immune cell frequencies, while other relationships only appear conserved in the huTMEs matching muTME profiles. Despite this variable alignment, we define robust cell type-specific gene expression programs conserved in TMEs across species and cohorts and identify ones that are coordinated between cell populations in both species. Together, we isolate and offer methods to study the multiple areas of hazard and opportunities for using mice to model human cancer. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=192 HEIGHT=200 SRC="FIGDIR/small/661217v1_ufig1.gif" ALT="Figure 1"> View larger version (71K): org.highwire.dtl.DTLVardef@22bcf7org.highwire.dtl.DTLVardef@a410b7org.highwire.dtl.DTLVardef@1455eb5org.highwire.dtl.DTLVardef@14cc6f5_HPS_FORMAT_FIGEXP M_FIG C_FIG

Published in Nature Immunology (predicted rank #2) · training set

Matching journals

The top 6 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.