Researchers Find How Cancer Cells Shield Themselves From Oxidative Stress

Every cell that uses oxygen performs a dangerous bargain. Mitochondria turn nutrients and oxygen into usable energy, but the machinery leaks chemically reactive forms of oxygen. In controlled amounts, these molecules carry signals. In excess, they attack DNA, proteins and the fatty membranes that hold a cell together.

A fast-growing cancer cell makes the bargain more precarious. Its altered metabolism, oncogenic signaling, iron traffic and relentless production of cellular material tend to raise the background level of reactive oxygen species, or ROS. The tumor benefits from some of that chemistry: ROS can help transmit growth signals and remodel behavior. But one further rise can turn useful flame into lethal fire.

The puzzle is how a cell already living close to that edge survives. A multi-institutional Japanese team led by researchers including Yumiko Fujikawa, Hirotatsu Imai, Shigeo Ohno and Akio Yamashita reports one answer in Signal Transduction and Targeted Therapy. Two large stress-responsive kinases—SMG1 and DNA-dependent protein kinase, or DNA-PK—directly modify the antioxidant regulator NRF2. Their phosphate marks help NRF2 escape its molecular restraint, accumulate and activate a survival program.

A cell’s smoke, spark and signal

“Reactive oxygen species” is a family name, not one chemical. It includes superoxide, hydrogen peroxide and other oxygen-derived oxidants. Superoxide generated by mitochondrial electron transport can be converted by superoxide dismutase into hydrogen peroxide. Catalase, glutathione peroxidases, peroxiredoxins and other systems then contain or remove the peroxide.

Calling all ROS “toxic waste” misses half the biology. Hydrogen peroxide can reversibly modify proteins and transmit information about nutrients, growth and stress. Too little redox signaling can disrupt normal function; too much causes oxidative stress. The relevant variable is not simply presence or absence, but dose, duration, location and the cell’s capacity to repair damage.

Cancer exploits this narrow interval. It can use elevated ROS to promote proliferation and adaptation, while increasing antioxidant capacity enough to avoid destruction. This creates what researchers call a redox vulnerability: the tumor has a stronger defense, but it may also have less distance to fall if that defense is removed.

NRF2, the emergency transcription chief

NRF2 is a transcription factor—a protein that switches groups of genes on. Under quiet conditions, the sensor and adaptor KEAP1 binds NRF2 and presents it for ubiquitination and disposal by the proteasome. The cell continually makes NRF2 and continually removes it, keeping the defense chief on a short leash.

When oxidants modify sensitive cysteines in KEAP1, the disposal process weakens. NRF2 accumulates, moves into the nucleus and binds antioxidant response elements in DNA. It increases programs involved in glutathione synthesis, detoxification, heme metabolism, iron control and resistance to lipid peroxidation. These defenses can protect normal tissue from pollution, inflammation and metabolic stress.

That protective reputation makes NRF2 a biological Janus. Before cancer begins, it can limit DNA damage and inflammation that contribute to tumor initiation. After a tumor has formed, persistent NRF2 activity can help malignant cells tolerate their harsh environment and resist radiation or drugs. The same fire brigade that protects the city can protect a criminal hideout inside it.

Two phosphate marks loosen the handcuff

The study’s molecular advance is unusually specific. Under mild oxidative stress, SMG1 and DNA-PK cooperatively phosphorylated NRF2 at serine 13 and serine 40. Those amino acids sit close to the DLG motif NRF2 uses to interact with KEAP1. Adding the phosphate groups weakened that interaction, stabilized NRF2 and increased antioxidant transcription.

Biochemical assays with purified proteins showed that SMG1 could phosphorylate NRF2 directly rather than merely activating it through a long indirect chain. Genetic knockdown and pharmacological inhibition supported the roles of both kinases in cells. Analysis of newly synthesized RNA showed that blocking either kinase was enough to suppress the NRF2-driven antioxidant program; blocking both produced a modest additional effect.

The pathway was strongest under mild stress. When oxidant exposure became excessive, the survival program was overridden and signaling shifted toward ATF4, ATM–CHK2 and JNK/p38 anti-survival responses. The finding therefore describes an intensity-sensitive switch: adapt while the damage remains manageable, die when it passes a threshold.

Finding a chemical wedge

To study SMG1 cleanly, the researchers needed a selective inhibitor. Genetic removal of SMG1 also disturbs nonsense-mediated mRNA decay, the quality-control system that destroys faulty transcripts, and that disturbance can itself pre-activate stress programs. The team screened large chemical collections from RIKEN’s Natural Products Depository, AIST and other sources and identified NPD15008.

The compound bound SMG1 and competed with ATP at the kinase’s active site. It allowed the researchers to switch SMG1 activity down on a short timescale and watch what followed. In PC-3 prostate-cancer cells, NPD15008 raised total ROS, ferrous iron and lipid hydroperoxides. Viability fell markedly, while the effect was substantially weaker in TIG-7 normal human fibroblasts.

Several clues identified the form of death. Caspase inhibition did not restore survival, arguing against apoptosis as the dominant route. A necroptosis inhibitor also failed. Two structurally different ferroptosis inhibitors partially rescued the cells, while oxidized-lipid markers accumulated. An existing DNA-PK inhibitor produced a similar pattern, and combined inhibition intensified parts of the response.

Ferroptosis: when a membrane rusts

In 2012, researchers coined “ferroptosis” for an iron-dependent, non-apoptotic form of regulated cell death triggered by the small molecule erastin. The word joins ferrum, Latin for iron, with death. Its visual metaphor is rust, but the chemistry happens in the polyunsaturated lipids of cellular membranes.

Ferrous iron can amplify radical reactions. If lipid hydroperoxides form faster than systems such as glutathione and GPX4 can neutralize them, oxidation propagates through the membrane. The barrier loses integrity and the cell dies. This is not the orderly dismantling of apoptosis, nor simply accidental poisoning. It is a regulated vulnerability governed by iron, lipids and antioxidant capacity.

Ferroptosis has become attractive in oncology because some drug-resistant, metabolically stressed cancers appear highly dependent on anti-ferroptosis machinery. Yet it is not intrinsically beneficial. The same process contributes to tissue injury in ischemia and is implicated in neurodegenerative disease. A therapy must drive the tumor over the edge without pushing vulnerable normal organs with it.

From Warburg’s flask to the redox switch

Nearly a century ago, Otto Warburg observed that tumors consume glucose and release lactate even when oxygen is present. His interpretation that damaged respiration caused cancer proved too absolute, but the “Warburg effect” opened a durable question: how does malignant metabolism differ, and how does that difference create a weakness?

In 1969, Joe McCord and Irwin Fridovich identified the enzymatic function of superoxide dismutase, demonstrating that cells actively defend themselves against an oxygen radical. In 1994, researchers isolated NRF2. Work that followed established KEAP1 as the restraint that keeps NRF2 low until stress demands a response. In 2012, ferroptosis gave a name to iron-dependent death by lipid oxidation.

The new paper connects branches that developed separately. SMG1 was known chiefly as a guardian of mRNA quality. DNA-PK was a central machine in DNA double-strand-break repair. NRF2 belonged to antioxidant transcription. Ferroptosis belonged to iron and lipid chemistry. The discovery shows that the two stress kinases also meet at NRF2 before oxidative damage becomes overwhelming.

What the experiments actually showed

The evidence spans purified-protein kinase assays, cultured human cell lines, RNA profiling, fluorescent measures of ROS, iron and lipid peroxides, short-term mouse-liver analysis, prostate-cancer xenografts and correlations in The Cancer Genome Atlas. TCGA data linked SMG1 and the DNA-PK gene PRKDC with NRF2 target-gene expression across several tumor types, although patterns varied and genetic mutations in the canonical pathway did not explain everything.

Stable SMG1 knockdown slowed growth of PC-3 tumors implanted in nude mice. The chemical story was less decisive: NPD15008 produced only a modest trend toward slower xenograft growth, and robust, reproducible antitumor effects were not consistently observed. The authors attributed this partly to poor solubility and bioavailability, which limited sustained exposure. That negative result is not a footnote; it defines the distance between a mechanistic probe and a medicine.

Short-term administration of NPD15008 increased ROS, ferrous iron and lipid-hydroperoxide signals in mouse liver without the apoptosis marker examined. That confirms the pathway can operate beyond a culture dish, but it also raises the safety question. A systemic inhibitor will encounter normal tissues that rely on SMG1, DNA-PK and NRF2.

The therapeutic opportunity—and the traps

A future strategy could inhibit SMG1, DNA-PK or their action on NRF2 in tumors already close to oxidative collapse. It might be paired with radiation, chemotherapy or metabolism-directed treatment that raises ROS, or with a ferroptosis inducer that attacks a second defense. Biomarkers could identify tumors with high NRF2 output and dependence on the new axis.

But each part of that sentence hides a development problem. SMG1 protects the integrity of mRNA surveillance. DNA-PK repairs dangerous DNA breaks and supports immune-system gene rearrangement. NRF2 protects liver, lung, kidney and other tissues from toxic injury. Broadly disabling them could create genomic damage, inflammation or organ toxicity. DNA-PK inhibitors are already being explored in oncology, but the new pathway supplies both a possible rationale and an additional safety mechanism to watch.

Tumors are heterogeneous. Some have KEAP1 or NRF2 mutations that keep NRF2 active; others may depend on glutathione, thioredoxin, GPX4, FSP1 or parallel stress programs. Blocking one gate can cause traffic to reroute. The useful therapeutic question is unlikely to be “Can NRF2 be turned off everywhere?” It is “Which tumors are addicted to this particular route, during which treatment window, and how can inhibition be delivered locally enough?”

Antioxidants are not a simple answer

The study should not be read as advice for patients to take or avoid antioxidant supplements. ROS biology is contextual, and dietary antioxidants do not reproduce precise genetic or pharmacological manipulation of SMG1, DNA-PK or NRF2 inside a tumor. Some antioxidants protect healthy tissue; in some experimental settings, antioxidant support can also help malignant cells. Treatment decisions belong with oncology teams and clinical evidence, not an extrapolation from a cell-culture pathway.

Likewise, “oxidative stress kills cancer” is too crude. Many therapies generate oxidative damage, but normal cells can be harmed and tumors can adapt. The promise of the 2026 discovery is precision: it identifies a molecular hand on the antioxidant switch, potentially allowing researchers to remove defense where a cancer is unusually dependent on it.

A national collaboration around a tiny mark

The author list spans the University of the Ryukyus, Juntendo University, RIKEN, Yokohama City University, Kindai University, AIST, Tohoku University, the University of Tokyo and other institutions. The project combined chemical libraries, antibody development, cancer models, RNA analysis and kinase biochemistry. Its physical centerpiece is microscopic: phosphate groups attached to two serines on one protein.

The paper also discloses that Yamashita and Ohno are co-inventors on Japanese and international patent applications related to SMG1 inhibitors; the remaining authors reported no competing interests. Funding included public Japanese programs and private foundations and industry programs. Those disclosures do not invalidate the findings, but they belong in assessing a target that may move toward commercial development.

The shield and the threshold

Cancer is often described as uncontrolled growth, but survival is just as important. A malignant cell must endure low oxygen, scarce nutrients, immune attack, DNA damage and the chemical exhaust of its own metabolism. Its success depends on sensing stress early enough to adapt.

The SMG1/DNA-PK–NRF2 axis supplies a new view of that judgment. Mild oxidative stress activates the shield. Excessive stress overwhelms it. Inhibition moves the threshold, allowing iron and oxidized lipids to turn manageable damage into ferroptotic death.

That is a compelling target, not yet a therapy. The next chapters require better compounds, tumor-selective delivery, biomarkers, toxicology, combination studies and eventually carefully designed human trials. For now, the discovery has done something fundamental: it has shown where two old guardians of RNA and DNA place their hands on cancer’s fire alarm—and how removing those hands can make the alarm become the blaze.