Multiomics immune profiling of a patient-relevant orthotopic lung cancer model using SEPARATE-Seq.

Relevant pre-clinical models are essential for driving progress in cancer therapy research. Here, we develop a pre-clinical study framework using an injectable orthotopic lung adenocarcinoma (LUAD) model (ORTHO) that replicates key features of human LUAD patients and is dissectible into tumoural and non-tumoural adjacent tissue, in analogy with patient samples. We also present SEPARATE-Seq, a broadly applicable technique enabling the partitioning of vascular and intratissue immune cells along with scRNA-Seq. By applying both SEPARATE-Seq and spatial transcriptomics to our dissectible ORTHO model, we confirm that our model replicates key immune features of human LUAD patients. Similarly to these patients, we observe NK-cell dysfunction and neutrophil dichotomy, and show that these are affected by their vascular/intratissue or tumour/adjacent location, highlighting the need for these spatial distinctions. Additionally, we show that several immune populations are restricted to specialised, local niches within the tumour, including a ring of lipid-associated TAMs lining the tumour edge and hubs of interferon-stimulated cells. Overall, our resource, available through an interactive tool, provides a comprehensive multiomics immune characterisation of a reproducible pre-clinical LUAD mouse model.