Glioblastoma (GBM) is a fast-growing and currently incurable brain cancer, and the most common malignant brain tumor in adults. Standard treatment includes surgery to remove the tumor followed by radiochemotherapy to target residual and infiltrating tumor cells, yet even with aggressive therapy, recurrence is nearly universal. This resistance has been linked to a population of tumor cells that enter quiescence—a stem-like state that allows them to survive toxic treatments and later re-initiate tumor growth.
Most cancer cells are characterized by rapid and uncontrolled division, making them vulnerable to therapies that target proliferating cells, however in GBM, residual stem-like cells often occupy this quiescent state, often referred to as G0, thereby evading treatment. It’s also important to remember that unlike senescent cells, which irreversibly enter a non-dividing state, quiescent cells can return to active cell division, making them capable of seeding new tumors.
You can think of these quiescent, G0 tumor cells as embers hidden beneath the ashes of a dying fire. After aggressive treatment extinguishes most of the tumor, these embers remain dormant—silent and unseen—but still capable of reigniting the disease. Understanding how these “embers” survive and what wakes them up is key to preventing tumor recurrence.
To better characterize these elusive cell populations, a new study in Nature Communications from the Paddison lab integrates single-cell RNA sequencing with cell cycle state classification and functional genomics. Their work, led by Dr. Anca Mihalas and Dr. Patrick Paddison, builds on the idea that G0 is not just a passive default but an actively regulated, plastic state influenced by developmental programs—and that targeting the mechanisms controlling these transitions could open new therapeutic avenues.
They then performed a genome-wide CRISPR knockout screen, dubbed the “G0-trap” to find genes whose loss pushes cells into this quiescent state. This screen identified 75 genes enriched and 37 depleted in quiescent cells with high p27/CDT1 signals.
Prioritizing targets, the team focused on genes whose knockout induced G0 without broadly impairing viability, leading them to KAT5—a histone acetyltransferase that modifies both histone and non-histone proteins to enhance DNA accessibility. Follow-up assays confirmed that knocking out KAT5 indeed traps glioblastoma stem-like cells in G0.
Using single-cell RNA sequencing, the authors compared cell cycle states in control versus KAT5-knockout (KAT5-KO) glioblastoma stem-like cell (GSC) cultures. KAT5-KO cells largely disappeared from proliferative clusters and instead formed distinct groups enriched for a “Neural G0” gene signature—a profile lacking most cycling genes but enriched in quiescence-associated genes.
“In this work we identified the histone acetyltransferase KAT5 as a crucial regulator of glioblastoma cell states,” says Mihalas. “High KAT5 activity correlates with proliferation and high-grade gliomas in patients, while low KAT5 activity induces G0-like programs and genes associated with neurodevelopment.”
To explore KAT5’s role in vivo, they implanted patient-derived glioma stem-like cells engineered with doxycycline-inducible KAT5 expression into the brain of recipient mice. All mice were initially kept on doxycycline, however after the tumors could be seen via MRI, mice were split into two groups and either kept on doxycycline (KAT5-on), or had their doxycycline withdrawn (KAT5-off), which allowed direct comparison of tumors with and without KAT5 activity.
Using CUT&Tag, a chromatin profiling method pioneered in the Henikoff lab, they found KAT5 enriched near transcription start sites of genes essential for cell cycle progression and tumor survival, including MYC and E2F targets. Strikingly, single-cell RNA sequencing showed KAT5-off tumors had a surge in quiescent, neural G0-like populations—particularly oligodendrocyte precursor–like and radial glia–like clusters—alongside reduced proliferative and mesenchymal gene expression. This indicates KAT5 maintains proliferative states in glioblastoma, and its loss shifts cells into a dormant, therapy-resistant state.
Next, the team evaluated KAT5 activity in primary glioma cells by examining its relationship with protein synthesis and cell state across tumor grades. Since quiescent G0-like cells synthesize less protein, they used protein synthesis rate as a functional readout of cell state and a surrogate for KAT5 activity. They found that KAT5 activity, measured by histone H4 acetylation, tightly correlated with protein synthesis rates. Low KAT5 activity linked to reduced rRNA levels, decreased DNA replication, and lower protein synthesis rates—hallmarks of quiescence.
Across a panel of primary tumors, lower-grade gliomas (LGGs) generally showed lower KAT5 activity and translation compared to higher-grade gliomas (HGGs). Tumors that recurred as HGGs exhibited upregulation of both KAT5 and protein synthesis. In culture, MYC alone could induce KAT5 activity and translation in human neural stem cells (hNSCs), whereas oncogenic EGFR/AKT stimulated translation independently of KAT5. Notably, NSCs and GSCs with high translation rates were more sensitive to the translation inhibitor homoharringtonine, suggesting that protein synthesis—and by extension KAT5 activity—may serve as a biomarker of therapeutic vulnerability.
Reflecting on these findings, Paddison stated, “KAT5 plays key roles in facilitating transitions from G0-like states in GBM tumors by promoting cell cycle gene expression and protein translation rates while suppressing neurodevelopmental gene expression, rendering cells more sensitive to the translation elongation inhibitor homoharringtonine.”
Speaking about potential clinical applications, Mihalas added, “KAT5 inhibition leads to a significant improvement of survival in preclinical animal models and confers additional survival probability in combination with standard of care therapy. These findings make KAT5 an attractive target for combination therapy in glioblastoma,” with Paddison predicting that “KAT5/H4-Ac high tumors will be more susceptible to treatment with homoharringtonine or biomimetic compounds.”
By uncovering how KAT5 keeps glioblastoma cells cycling—and what happens when that control slips—the Paddison lab opens the door to a new way of thinking about tumor recurrence. Instead of chasing only the dividing cells, their work suggests targeting the silent embers—the G0 dwellers—could be key to preventing relapse. And if protein synthesis rates can flag which tumors are primed to respond, treatments like homoharringtonine might finally hit glioblastoma where it hurts.
Fred Hutch/University of Washington/Seattle Children’s Cancer Consortium Members Drs. Toshio Tsukiyama and Patrick Paddison contributed to this research.
The spotlighted research was funded by the he National Institutes of Health, the Burroughs Wellcome Career Award for Medical Scientists, and the Kuni Foundation.
Mihalas AB, Arora S, O’Connor SA, Feldman HM, Cucinotta CE, Mitchell K, Bassett J, Kim D, Jin K, Hoellerbauer P, Delegard J, Ling M, Jenkins W, Kufeld M, Corrin P, Carter L, Tsukiyama T, Aronow B, Plaisier CL, Patel AP, Paddison PJ. 2025. KAT5 regulates neurodevelopmental states associated with G0-like populations in glioblastoma. Nature Communications. doi: 10.1038/s41467-025-59503-w.
Science Spotlight writer Jenny Waters is a postdoctoral research fellow in the Hsieh lab at Fred Hutch. She studies how mRNA translation coordinates bladder cancer transformation and metastasis by post-transcriptionally regulating expression of oncogenic proteins. Outside of the lab, Jenny enjoys spending time with her dogs, convincing her husband to join her on trail runs, and pretending every steep hill is just a "gentle incline."
For the Media