Tumors do not just mutate. They borrow, steal, and run the wrong program.

Stealing the neighbor's machinery

In a preprint on glioblastoma, researchers report that tumor cells hijack ribosomes from astrocytes, delivered through tunneling nanotubes. The transferred ribosomes are translationally active and reprogram the tumor's translatome. A protein called INTS12 is a critical mediator of the transfer, and knocking it down with an AAV-mediated approach extended survival in orthotopic glioblastoma models. m:…1477

The picture is not a tumor growing in isolation. It is a tumor borrowing the protein-making machinery of the healthy brain cells around it, then using that machinery to rewrite what it produces. Cut the transfer route, and the models lived longer, which makes the route itself a target, not just the tumor's own mutations. m:…1477

The genome's loose pages

A second preprint looks at extrachromosomal DNA, the free-floating circles of DNA that carry cancer genes outside the chromosomes. A method called FLAGx lets researchers resolve, quantitatively and in place, the transcriptional activity and nuclear position of individual ecDNA genes. Gene context, which genes sit on the circle, influences where the ecDNA sits in the nucleus and how its transcription goes wrong. m:…3489

The same study found that ecDNA clustering, cell cycle regulation, and drug responses depend on gene context, while transcriptional shutdown in mitosis is a persistent feature across ecDNA species. What an ecDNA circle does depends on what it carries and where it parks, not just on the genes it contains. m:…3489

Tumors running the wrong program

A third preprint, on fusion-driven pediatric sarcomas, found that these tumors express neuronal marker genes in the wrong place, an ectopic expression of a developmental program. That misplaced program also puts two surface proteins, STEAP1 and NPTXR, together on Ewing sarcoma cells, offering a target pair for AND-gated therapies designed to act only when both markers are present. m:…9558

A fourth preprint, on pancreatic ductal adenocarcinoma, found that malignant epithelial cells positive for keratin 17 are enriched for gene programs linked to the epithelial-mesenchymal transition, a migratory cell state. Tumors co-expressing K17 and LAMC2 had significantly worse overall survival, with a hazard ratio of 2.53, compared with double-negative tumors. In 3D pancreatic spheroids, K17 promoted migratory outgrowth without changing proliferation rates: movement, not faster growth. m:…4342

The common thread

Across the four preprints, the recurring move is not a new mutation but a borrowed or misplaced program: ribosomes taken from neighbors, circles of DNA whose behavior depends on their cargo and position, a neuronal program switched on in a sarcoma, a migratory program switched on in pancreatic cancer. That reframes the therapeutic question from which gene is broken to which program is running, and where it came from. m:…1477 m:…3489 m:…9558 m:…4342

The practical openings named in the preprints are correspondingly specific: block a transfer mediator such as INTS12, gate a therapy on two surface markers such as STEAP1 and NPTXR, or read a tumor's migratory state from markers such as K17 rather than from growth rate alone. m:…1477 m:…9558 m:…4342

None of this replaces mutation testing, but it changes what a complete tumor profile should include: not only which genes are altered, but which programs are switched on, which neighbors are being tapped, and how the tumor's DNA is organized outside the chromosomes. A profile that misses those dimensions will misread why the tumor behaves as it does. m:…1477 m:…3489 m:…9558 m:…4342