Finding. An androgen response from a normal adrenal cell can mark cancers that need their mitochondria. Miao and Baba treat mouse zona fasciculata cells with dihydrotestosterone, peel global RNA-output suppression off the relative transcriptional remodeling, and keep a cross-sex signature. Projected onto human orthologs it becomes an androgen-suppressed transcriptional state (ASTS) you can score in cancer cell lines. High ASTS lines shrug at EGFR drugs and show weaker EGFR CRISPR dependency. They also show stronger dependency on mitochondrial translation and on respiratory electron transport. The pattern is not one lineage's private habit. It survives adjustment for canonical drivers and cell-state covariates.
Why this paper matters
Most mitochondrial-vulnerability stories start inside the tumor: a mutation, a Complex I loss, a mtDNA genotype. This one starts in a non-malignant endocrine perturbation and asks whether the resulting state still means something in cancer. The answer, in public pharmacogenomic and CRISPR matrices, is yes: the same state that came from androgen suppression points at mitoribosomes and the electron-transport chain, and away from EGFR. That is a methodological claim as much as a biological one. Transcriptional states borrowed from physiology can rank organelle dependencies.
EGFR resistance is the clinical earworm. Mitochondrial translation is the organelle one. Keep both. The paper does not say androgen therapy will kill those mitochondria. It says the state is a map.
What they actually measured
A careful split of DHT effects (global RNA drop versus relative remodeling) in mouse adrenal cells. A humanized signature. Then correlation and covariate-adjusted association with drug-response and CRISPR-dependency catalogs, highlighting EGFR and mitochondrial translation / respiratory ETC. Proof-of-concept language is theirs, and it is the right volume.
How to read the score
High 80s. Cross-context mitochondrial dependency with a named opposing axis (EGFR). Confidence is medium because it is a projected signature on cell lines, not a closed mechanistic loop in a mouse tumor.
Caveats
No patient DHT series. No single mitochondrial gene named as the dependency. Association after covariates is not causation. Adrenal zona fasciculata is not a generic androgen target tissue; the authors are using it as a state factory.
What to do with it
If you score cell-line states against CRISPR mitochondria gene sets, compute ASTS and look at the EGFR-low, mitoribosome-high quadrant. If you develop mitochondrial-translation inhibitors, this is a possible enrichment signature. Do not add an androgen to an EGFR-refractory protocol from this brief.