Researchers now report that many cancers do not just turn on embryonic genes again; they also reactivate the molecular tools that rewrite RNA, helping tumours grow fast and adapt aggressively.
How cancer rewinds cell identity
Modern oncology no longer sees cancer as pure genetic chaos. Many tumours behave more like a distorted development program that has been rolled back and misused. Instead of inventing new tricks, malignant cells dig into very old ones: the same strategies that once drove the embryo’s explosive growth.
The new work, led by the Centre for Genomic Regulation (CRG) in Barcelona with collaborators at ETH Zurich and published in Nucleic Acids Research, strengthens this view. The team shows that cancer cells revive a specific part of early development: a set of RNA “editing” proteins called splicing factors that largely switch off after birth.
Cancer cells do not only reactivate embryonic genes; they restart the machinery that edits those genes’ messages, shifting the whole logic of growth.
During early embryogenesis, cells divide at high speed yet remain plastic, able to mature into nearly any tissue. That flexibility demands a dynamic way to manage genes: one DNA sequence must give rise to many possible protein versions, depending on context. Cancers appear to copy that logic. When they fall back into an embryonic-like state, they recover both the genes and the editing tools that made early development so flexible.
Splicing factors: the RNA editors turned rogue
At the centre of this story lies RNA splicing, a step that many people skip when thinking about genetics. Textbooks often present the flow as DNA to RNA to protein, but the reality is messier. Once a gene is copied into RNA, the cell cuts and stitches that RNA before sending it to the protein factories.
From one gene to many messages
Most human genes come split into segments called exons and introns. Splicing factors, a large family of specialised proteins, decide which segments stay and which go. Different decisions yield different RNA transcripts from the same gene, and those transcripts can produce proteins with distinct shapes and functions.
- Keep all exons: full-length protein, often seen in stable, differentiated cells.
- Skip specific exons: shorter or altered protein, which may change how a cell divides or moves.
- Use alternative start or end exons: version tuned to stress, development stage, or tissue type.
Healthy tissues keep this system under tight control. Signals from the environment, hormones, or developmental cues adjust splicing patterns gradually. The new study shows that cancer cells break that gentle tuning. They switch back on a set of splicing factors usually active only in embryos, pushing the cell into a radically different splicing landscape.
Altering splicing factors does not touch the DNA; it changes the “software layer” that interprets the genome, protein by protein.
➡️ No vinegar and no baking soda: pour half a glass and the drain cleans itself
➡️ Microwaving a lemon : A simple kitchen trick you’ll keep using
➡️ A High-Fat Diet Can Harm Memory, Mouse Study Finds
➡️ You keep getting up at night to pee? Specialists say this warning sign should never be ignored
➡️ It’s the perfect time to take fig tree cuttings: how to do it in October
Once these embryonic splicing factors return, they tilt the balance towards RNA variants that favour rapid division, survival under stress and easier migration through tissues. A tumour can then better cope with poor oxygen supply, scarce nutrients or chemotherapy, because its protein toolkit shifts on demand.
A molecular domino effect
One striking result of the CRG–ETH team: you do not need to change dozens of factors at once. Tweaking a relatively small group of “initiator” splicing regulators appears enough to disturb the complete network. Splicing factors control not only structural genes but often each other, forming densely connected circuits.
In this dense web, activating a handful of nodes triggers a domino effect. A few altered proteins reverberate through the network, rewiring the splicing of hundreds of RNAs. Genes that once held growth in check produce weaker or inactive protein forms, while pro-growth and pro-migration variants flood the cell.
MYC, the familiar oncogene with a new angle
The study also ties this splicing upheaval to one of oncology’s usual suspects: the MYC oncogene. MYC is a transcription factor, a protein that binds DNA and switches other genes on or off. Abnormally high MYC activity turns up in a long list of cancers, from lymphomas to lung tumours and many solid malignancies.
How MYC lights the fuse
Using large datasets of gene expression, the team showed that MYC does more than speed up the cell cycle. When MYC rises, it selectively targets a narrow set of splicing factors – the same “initiators” that sit high in the regulatory hierarchy. Once MYC boosts those initiators, they push the rest of the splicing network out of balance.
| Step | What happens in the cell |
|---|---|
| 1. MYC activation | MYC levels rise due to mutation, amplification or upstream signals. |
| 2. Splicing initiators induced | MYC directly increases transcription of key splicing factor genes. |
| 3. Network cascade | Initiators alter splicing of many RNAs, including other regulators. |
| 4. Tumour-friendly isoforms | Protein variants that promote proliferation and survival dominate. |
This sequence shifts the cell into a new functional state without inserting dramatic new mutations into DNA. The genomic “text” stays broadly similar, but the punctuation and paragraph breaks change, so the cell reads a different story. That nuance matters: drug developers can target changeable regulators without attacking the hard-coded genome.
AI maps a hidden layer of cancer biology
Splicing is complex to measure. Each gene can produce dozens of RNA variants, and each variant might rise or fall in a tissue-specific way. Tracking that complexity across thousands of tumours quickly becomes overwhelming.
To handle that scale, the researchers trained an artificial intelligence model on vast transcriptomic datasets. Rather than measuring every spliced molecule, the algorithm infers which splicing factors must be active based on the global pattern of gene expression. It reads the “fingerprint” that an active splicing regulator leaves across many targets.
Artificial intelligence here acts like a decoder ring: it reconstructs the hidden activity of splicing regulators from noisy expression patterns in bulk tumour samples.
This AI-based approach brings two major benefits. First, it reveals recurring splicing programs shared across cancer types, suggesting common therapeutic targets. Second, it might help detect cancers earlier. Subtle shifts in splicing activity often appear before a mass shows up on a scan or causes symptoms, so splicing fingerprints could act as early blood-based warning signs.
Potential new routes for diagnosis and treatment
The study hints at a new kind of biomarker. Instead of hunting for single faulty genes, clinicians could track combinations of splicing changes that signal a cell has slipped into an embryonic-like, MYC-driven state. Such signatures might improve risk stratification in conditions like early lung nodules or ambiguous breast lesions.
For therapy, the highly connected splicing network may look intimidating at first glance. Yet that same connectivity creates leverage points. If a few initiator factors hold together the malignant configuration, drugs or RNA-based therapies that dampen those factors might reset the system. An “inverse domino” could nudge cancer cells back toward a more stable program that limits growth.
This strategy differs from classic chemotherapy, which attacks all rapidly dividing cells and often harms bone marrow, gut and hair follicles. A splicing-focused treatment would instead aim at the decision layer that controls which protein versions a cell makes, leaving the DNA intact. That could reduce some long-term toxicities and delay the rise of resistant clones that sidestep direct DNA damage.
What this means beyond this single study
The work sits within a broader shift in cancer research, where scientists pay much closer attention to RNA-level regulation. For years, focus fell mainly on mutations and chromosomal rearrangements. Now, alternative splicing joins epigenetic changes and 3D genome folding as major contributors to how a tumour behaves, independent of its raw mutation load.
For patients and clinicians, this raises practical questions. Could regular blood tests one day track splicing-based signatures as people at high risk age, much like cholesterol or PSA today? Might paediatric cancers, which often show fewer DNA mutations, rely even more on developmental splicing programs and therefore respond strongly to therapies targeting those programs?
At the same time, tinkering with splicing carries risks. Many tissues depend on finely tuned splicing decisions, especially the brain and immune system. Broadly blocking a factor that tumours abuse might also disturb neurons or T cells. Future drugs will likely need high precision, perhaps by disrupting only the interaction between a splicing factor and selected cancer-critical RNAs rather than silencing the factor everywhere.
For now, the message is that cancer behaves less like a random glitch and more like a misused developmental script. By tracking how tumours revive embryonic splicing tools and bend them to their own ends, researchers gain a new layer of targets and a more nuanced way to think about early detection, relapse and long-term control.