many 🙌 to Sukanya Panja who bravely took the 🔥 from lab alum Kofi Johnson, big 🙏 to COO-guru Paz Polak for spiritual guidance, NCI for funding, and the broader lung cancer village incl Tuomas Tammela, Carla Kim, @labjacks.bsky.social & Tere Landi for feedback over the years 22/22🧵
While COO is not currently actionable, TOO mut patterns inform CUP diagnosis → @ecuppen.bsky.social www.nature.com/articles/s41.... LUAD vs LUSC histology also guides chemo choice. As oncology WGS becomes routine, it will be possible to link granular COO status with 💊 treatment response 21/🧵
looking forward to see how combinations of single-cell RNA-, ATAC-seq, and other multi-omic atlas-derived features can yield more granular, robust & even clinical-grade links between normal cell epigenomes and tumor somatic mutational topography 20/🧵
while brewing a scATAC version of this approach, we had the honor 👏 of being scooped by a beautiful recent study from (uptown colleague) Alexander Tsankov and Rosa Karlic introducing their method, aptly titled .. SCOOP → www.nature.com/articles/s41... 19/🧵
our methods may generalize to other cancer types and single-cell tissue atlases, in particular chromatin accessibility where classic work from Polak et al have shown strong tissue-of-origin correlations with tumor type somatic mutation density → www.nature.com/articles/nat... 18/🧵
deeper analyses in larger datasets, eg massive lung cancer WGS cohorts in the landmark NCI-SHERLOCK study → www.nature.com/articles/s41... and @genomicsengland.bsky.social, may provide power to study molecular and epidemiological correlates of COO status, particularly in never-smokers 17/🧵
the mapping of some LUADs to cell types associated with chronic lung injury (e.g. AT0, AT2 proliferating) may support the "cancer promoter" model put forth by @charlesswanton.bsky.social to implicate air pollutants as lung carcinogens → www.nature.com/articles/s41... 16/🧵
these and other hypotheses about proximal-distal lineage plasticity and driver genotypes in early LUAD evolution are readily testable in GEMM systems eg those wielded by Gardner, Varmus & Laughney in last year's tour de force → www.science.org/doi/10.1126/... 15/🧵
interestingly, proximal origin LUADs with biallelic TP53 inactivation were substantially more likely to retain their non-distal identity than TP53 wild type tumors, pointing to an unexpected interplay between TP53 status and lineage plasticity 14/🧵
we next asked how plastic are distal vs. non-distal lung cell origins, analyzing both single cell and bulk transcription to identify LUADs with predominantly distal vs. non-distal cell identity 13/🧵
our analyses also linked distal origin LUADs to specific recently-discovered alveolar cell types (AT2 proliferating and AT0 cells) both associated with chronic lung injury 12/🧵
these proximal origin LUADs were strongly enriched in never-smokers and mostly wild type for KRAS 11/🧵
..but this was when lumping all LUADs and LUSCs together. what happened when we looked at tumors one-by-one? while nearly all LUSCs had proximal origins, some LUADs show non-distal origins 10/🧵
our study shows LUAD passenger mutation density is most strongly anti-correlated with AT2 gene expression, similarly LUSC with basal gene expression, consistent with their presumed cell origins 9/🧵
the most common lung cancers (LUAD and LUSC) are thought to arise from specific cell types in the distal (AT2) and proximal (basal) lung, respectively 8/🧵
could this anti-correlation help us trace lung cancers back to the COO? We leveraged recent detailed single cell atlases of normal lung from @fabiantheis.bsky.social to assess these links at various levels of the lung cell type taxonomy → www.nature.com/articles/s41... 7/🧵
but expressed in .. which cell type? since ≥ 50% somatic mutations are thought to happen before normal cells break bad (→ Tomasetti 2013 www.pnas.org/doi/full/10....) this effect should be strongest in normal cells of the lung from which lung cancers arise .. the COOs 6/🧵
how to uncover human lung cancer COOs? It has long been known (→ Pleasance 2010 www.nature.com/articles/nat...) that highly expressed genes have fewer somatic mutations, particularly for certain mutation processes (eg tobacco) 5/🧵
while GEMMs are used "prove" COO, they ultimately show <sufficiency>, ie that cells expressing a given lineage marker (eg SPC+) <can> give rise to tumors of a given histology (eg LUAD) in a driver context (eg Kras G12D mutant, Tp53-/-) → genesdev.cshlp.org/content/34/1... 4/🧵
.. a major confounder is lineage plasticity, both a mechanism of drug resistance in relapsed tumors (→ Rudin, @sawyerslabmskcc.bsky.social www.nature.com/articles/s41...) and a hallmark of histologically heterogenous primary LUADs (→ co-author Bill Travis www.jto.org/article/S155... 3/🧵
Classic approaches for inferring human cancer cell-of-origin (COO) rest on (shaky!) assumptions that cancer cells resemble their COO (eg by histology, transcription, methylation).. → see Visvader 2011 www.nature.com/articles/nat... 👇 2/🧵
hello @bsky.app! let's get 🍊 juices flowing with the latest from the lab → how we used whole genome passenger somatic mutation patterns 🧬 to trace the origins of never-smoker lung adenocarcinoma (LUAD) to cells of the proximal lung 🫁 @NatureGenet www.nature.com/articles/s41... 1/🧵
