How the immune system can sometimes aid tumor growth

In a discovery that may finally explain one of the most confounding paradoxes in cancer biology, researchers at the Salk Institute for Biological Studies have identified the molecular mechanism through which the immune system’s own first-line defenses against cancer can be hijacked to promote tumor growth. The study, published in the journal Science on September 10, 2026, reveals that chronic exposure to interferon II—a signaling protein family that normally mobilizes immune cells to destroy emerging tumors—triggers a cascade of mitochondrial dysfunction inside cancer cells that ultimately suppresses the very immune response meant to eliminate them. Most strikingly, the researchers demonstrated that blocking a single downstream metabolic signal in this pathway reversed immunotherapy resistance in a mouse model of melanoma, sending previously untreatable tumors into complete and lasting regression.

Interferons have long occupied a privileged place in immunology. These pro-inflammatory signaling proteins act as molecular alarms, rushing to the site of a nascent cancer and recruiting specialized immune cells such as T cells and B cells to seek out and destroy malignant tissue. This early interferon response is widely regarded as a critical and powerful component of the body’s natural cancer surveillance. Yet clinicians and researchers have repeatedly observed a troubling pattern: in established tumors, interferon signaling that persists too long often correlates with worse outcomes, immune evasion, and resistance to checkpoint blockade therapies such as anti-PD1. Why a protein that begins as an anti-cancer warrior becomes a tumor’s accomplice has remained one of the field’s most stubborn open questions.

The Salk team, led by senior author Gerald Shadel, professor and holder of the Audrey Geisel Chair in Biomedical Science, approached the mystery from an unusual angle—the mitochondria. Shadel’s laboratory has spent years investigating how these cellular powerhouses communicate with the immune system, most notably through their discovery that mitochondria can provoke interferon responses by releasing their genetic material, mitochondrial DNA, into the rest of the cell, where it is perceived as a foreign invader. For the new study, the researchers deliberately inverted this question. Rather than asking how mitochondria shape interferon signaling, they asked how sustained interferon exposure reshapes mitochondrial function—and whether that remodeling could hold the key to the paradox of pro-tumor interferon activity.

To find out, the team exposed melanoma cells in the laboratory to type I or type II interferons for either brief, acute periods or extended, chronic periods. The differences were dramatic. Acute interferon exposure left the mitochondria largely unscathed, but chronic exposure produced measurable deterioration in the organelles’ energetic function. When the researchers then transferred these chronically exposed melanoma cells into a mouse model, they made an unexpected and consequential observation: chronic interferon II exposure did not slow tumor growth—it enhanced it. The finding provided a clean experimental demonstration that the duration of interferon signaling, not merely its presence, determines whether it helps or harms the host in the fight against cancer.

Digging into the cellular machinery behind this switch, the researchers uncovered a previously unknown pathway that begins deep inside the mitochondria. Chronic type II interferon exposure causes mitochondrial RNA, or mtRNA, to escape the confines of the organelle and leak into the main body of the cell. Sensing this misplaced genetic material, the cell’s innate immune surveillance systems interpret it as evidence of viral invasion and respond by producing type I interferon—a second, distinct interferon family. Type I and type II interferons then act in concert to drive up the levels of an enzyme called cyclooxygenase 2, which in turn increases the synthesis of a bioactive lipid known as prostaglandin E2. This lipid messenger, long associated with inflammation and immunosuppression in tumors, effectively flips the tumor microenvironment from a state of immune attack to one of immune tolerance.

The identification of prostaglandin E2 as the linchpin of this pathway immediately suggested a therapeutic test. If this lipid signal was the agent responsible for damping down the immune response, the researchers reasoned, then preventing melanoma cells from manufacturing it might restore the immune system’s ability to see and attack the cancer. The stakes of this question extend well beyond basic biology. Anti-PD1 immunotherapies—among the most widely used cancer treatments in the world today—work by blocking a signal that cancer cells deploy to keep immune cells at bay. But tumors frequently deploy additional, independent immunosuppressive pathways, allowing them to continue growing even in patients receiving anti-PD1 treatment. Understanding and disabling those alternative escape routes is one of the most urgent challenges in modern oncology.

“Chronic interferon exposure is a major factor in immunotherapy resistance,” said Melissa Johnson, a graduate student researcher in Shadel’s laboratory and first author of the study. “We wondered whether cancer cells that have become resistant to anti-PD1 therapy were upregulating the immunosuppressive mitochondria-centered pathway we identified, and whether that pathway is a viable target for combating immunotherapy resistance.” The team’s experimental results answered that question with unusual force. When the researchers blocked prostaglandin E2 synthesis in mouse melanoma cells, the immune system’s capacity to recognize and destroy the cancer was restored. More remarkably, blocking the lipid signal reversed resistance to anti-PD1 therapy itself: in nine out of ten mice, tumors that had previously shrugged off immunotherapy regressed completely and did not return.

The completeness and durability of those responses set the findings apart from typical incremental advances in tumor immunology. Rather than merely slowing tumor growth, eliminating the prostaglandin E2 signal appeared to reawaken a sustained, effective anti-tumor immune attack—suggesting that the mitochondria-to-prostaglandin pathway is not simply one suppressive mechanism among many, but a genuine molecular switch governing whether the tumor microenvironment invites or repels immune destruction. In the melanoma model, flipping that switch off was sufficient to transform an immunologically “cold,” therapy-resistant tumor back into one the immune system could eradicate.

The implications for future cancer treatment are considerable. The study points to a potential strategy for sustaining the immune system’s assault on tumors by intervening in the mitochondrial signaling axis rather than, or in addition to, checkpoint blockade. Because the pathway is cell-autonomous—operating within the tumor cells themselves—it offers a target that may complement existing immunotherapies and provide an option for patients whose cancers have exhausted the benefits of anti-PD1 agents. While the work remains at the preclinical stage in mouse models, the researchers emphasize that the pathway’s components, from mtRNA release to cyclooxygenase 2 activity to prostaglandin E2 production, represent a series of druggable nodes that could each be targeted therapeutically. Existing drugs that inhibit cyclooxygenase enzymes, for example, raise the possibility of repurposing well-characterized compounds to disrupt the pathway at its enzymatic core.

Beyond its immediate translational promise, the work carries a broader conceptual message for cancer biology: mitochondrial signaling cannot be treated as a background housekeeping function when modeling how tumors interact with the immune system. Shadel’s team demonstrated that a signaling molecule long classified purely as an immune modulator exerts direct, physical effects on mitochondrial integrity within cancer cells, and that those effects feed forward into an inflammatory lipid axis with decisive immunological consequences. The finding enriches the field’s understanding of how the immune system attacks cancer cells but can also be stymied by other factors in the tumor environment, and it makes a compelling case for integrating mitochondrial signaling functions into the design and interpretation of cancer studies.

For a question that has puzzled the field for decades—why the immune system sometimes helps tumors grow—the Salk team has delivered not just an explanation but a roadmap. Chronic interferon II drives mitochondrial RNA out of its organelle home, ignites a type I interferon response inside the tumor cell itself, amplifies cyclooxygenase 2, floods the microenvironment with prostaglandin E2, and thereby blindfolds the immune system at the very moment it should be striking. Cut the pathway at its lipid endpoint, and the blindfold falls away. The research was funded by the National Institutes of Health alongside private philanthropic support from the Glenn Foundation for Medical Research, the Cancer Research Institute, the NOMIS Foundation, and others, and the team included contributors spanning immunology, metabolism, and epigenetics—an interdisciplinary breadth that mirrors the pathway’s own reach across cellular compartments. As immunotherapy resistance continues to challenge clinicians worldwide, this mitochondria-centered mechanism offers a durable answer to a longstanding mystery and, potentially, a powerful new lever to pull in the clinic.