TREX1 enzyme shields tumors by destroying immune-activating DNA fragments

An enzyme that tumor cells use to stay invisible to the immune system has been caught in the act of destroying a very specific kind of molecular evidence. In a study published in EMBO Reports, researchers report that the exonuclease TREX1 keeps murine cancer cells immunologically quiet by degrading short fragments of cytosolic DNA, and that these fragments arise not from the micronuclei long suspected of fueling inflammation, but from transcriptionally active regions of the genome processed by the nuclease MRE11. When TREX1 is removed, the fragments pile up, interferon signaling ignites, and tumors in mice become dramatically more responsive to immunotherapy.

TREX1 is a dimeric 3-prime to 5-prime exonuclease anchored to the endoplasmic reticulum with its catalytic domain facing the cytoplasm. It chews up a variety of DNA substrates, including single-stranded DNA, double-stranded DNA, and RNA:DNA hybrids, preventing their recognition by cGAS, the cytosolic sensor that activates STING and triggers type I interferon signaling. High TREX1 activity has been associated with poor clinical outcomes in several cancers, and previous work in mouse models has shown that TREX1 loss can suppress tumor growth and improve responses to immune checkpoint blockade. What remained unclear was exactly which DNA molecules TREX1 degrades to produce these effects.

To find out, the team, led by Chun-Yen Yang and Yea-Lih Lin, used CRISPR-Cas9 to knock out the Trex1 gene in two murine cancer cell lines: Clem2 lung carcinoma cells and B16F10 melanoma cells. They isolated multiple independent knockout clones, confirmed the deletions by sequencing, and selected clones with normal cGAS and STING expression so that the results would not be confounded by altered expression of the signaling machinery itself. They then quantified cytosolic DNA two ways: by immunofluorescence microscopy with antibodies against single- and double-stranded DNA, and by a sensitive microfluidic capillary electrophoresis platform called BIABooster, which can separate DNA fragments by size directly from unpurified samples with a detection limit down to 10 femtograms per microliter.

The two methods told a consistent but quantitatively different story. Immunofluorescence showed a significant increase in cytosolic ssDNA and dsDNA in Trex1-knockout cells, roughly a twofold rise. BIABooster, which measures total DNA regardless of strand composition, revealed a much larger effect: cytosolic DNA increased approximately sevenfold in Clem2 cells and sixfold in B16F10 cells. The fragments ranged from about 75 to 400 base pairs, with a predominant peak between 150 and 200 base pairs, the length of DNA wrapped around a single nucleosome. In absolute terms, the researchers estimated cytosolic DNA levels of roughly 0.2 to 1.2 femtograms per cell in Clem2 cells, equivalent to about 0.003 to 0.02 percent of the mouse genome, or approximately one megabase of DNA per knockout cell. Digestion with Duplex DNase, which preferentially cuts double-stranded DNA, reduced the signal in a dose-dependent manner, indicating that the pool is a mixture with double-stranded DNA the predominant detected species.

A key question was where this DNA came from. Micronuclei, small membrane-bound structures containing chromatin fragments from chromosome mis-segregation, are widely considered a major source of cytosolic DNA in cancer cells. But the data argued against them here. Basal micronuclei levels were actually lower in the murine cancer lines than in human U2OS cells used as a positive control, and Trex1 knockout did not raise micronuclei frequency at all, even under replication stress induced by topotecan. More decisively, when the researchers blocked mitotic entry with the CDK1 inhibitor RO-3306, preventing micronuclei formation entirely, the abundance of short cytosolic DNA was unchanged. In the larger fragment range of 800 to 20,000 base pairs, where DNA from ruptured micronuclei would be expected, cytosolic DNA was scarce and showed no consistent dependence on genotype or treatment.

Instead, the nuclease MRE11 emerged as the source. Treating Trex1-knockout Clem2 cells with Mirin, an MRE11 inhibitor, markedly reduced the abundance of short cytosolic DNA fragments. MRE11 is known to resect DNA ends at double-strand breaks and stalled replication forks, and the team had previously shown that it promotes cytosolic DNA release in SAMHD1-depleted cells under replication stress. To map the genomic origin of the fragments, the researchers sequenced cytosolic DNA from wild-type and knockout cells aligned to the mouse genome. The DNA was overwhelmingly nuclear in origin, with mitochondrial reads below 0.2 percent in all samples, and it displayed a markedly non-random chromosomal distribution: several chromosomes were over-represented and others, including the gene-poor Y chromosome, strongly depleted. Genes classified as transcriptionally active by a median split of matched RNA-seq data carried significantly higher cytosolic DNA signal than silent genes in both genotypes, with a P value below 0.0001. Repeat-element composition matched genomic expectation, and TREX1 loss increased abundance without changing the overall composition of the pool.

The immunological consequences followed directly. Trex1 knockout induced phosphorylation of IRF3, the canonical effector downstream of cGAS-STING, and upregulated interferon-stimulated genes including Ccl5, Cxcl10, Isg15, and Ifit1, with an average 7.7-fold increase in Clem2 cells. Reintroducing wild-type TREX1 into knockout cells, but not a nuclease-dead D18N mutant, brought interferon gene expression back to baseline, confirming that the effect depends on TREX1’s catalytic activity. Genotoxic drugs such as topotecan, etoposide, and the ATR inhibitor AZD6738 raised interferon signaling only modestly by comparison, roughly twofold in wild-type cells and about 40 percent more in knockouts. Conditioned medium from Trex1-knockout cells strongly induced pro-inflammatory M1 markers, including Cd80, Cd86, and Tnfa, in RAW264.7 macrophages, while failing to induce the anti-inflammatory M2 markers Cd163, Cd206, and Tgfb, pointing to a diffusible signal that skews innate immunity toward an anti-cancer phenotype.

In vivo, the effects were striking. When Trex1-knockout and wild-type Clem2 cells were grafted subcutaneously into immunocompetent C57BL/6J mice, knockout tumors grew about fivefold smaller by day 28, and most remained below 500 cubic millimeters at day 40 while all wild-type tumors had reached the maximum allowed size. Median survival extended from 35 days with wild-type tumors to 61 days with knockouts. Notably, the knockout did not slow Clem2 proliferation in two-dimensional culture or three-dimensional spheroids, and it did not alter sensitivity to topotecan, etoposide, AZD6738, or bleomycin, suggesting the in vivo effect arises from the tumor immune microenvironment rather than intrinsic growth defects. Trex1 loss also did not impair replication fork resection or DNA end resection at double-strand breaks, arguing that TREX1 inhibition is unlikely to exacerbate genomic instability.

Therapy experiments sharpened the picture. Topotecan, which produced only a modest and non-significant response in wild-type tumors, markedly increased sensitivity of knockout tumors, extending median survival to 67 days compared with 39 days for treated wild-type tumors, though none of the mice were cured. Anti-PD-1 checkpoint blockade was far more dramatic: while treated wild-type tumors responded with all mice staying below 500 cubic millimeters at day 28, knockout tumors showed complete regression in 90 percent of mice, with 90 percent still tumor-free at day 95. When nine of these protected mice were later rechallengeed with wild-type Clem2 cells without further treatment, none developed tumors, and anti-Clem2 antibodies appeared in their sera, evidence of durable humoral immune memory. In athymic nude and NSG immunodeficient mice, by contrast, knockout and wild-type tumors grew identically and survival did not differ, confirming that tumor suppression requires an intact host immune system.

The authors are careful about the limits of their model. The work was done in two murine cell lines, and B16F10 knockout cells did show reduced proliferation in vitro, meaning that tumor-growth effects reported for that line in earlier studies may partly reflect intrinsic proliferation defects rather than immunity alone. The researchers also note that they cannot formally rule out contributions from other TREX1-dependent effects, and that establishing a causal link between transcriptional activity, local DNA processing, and cytosolic DNA production will require further investigation. Still, by directly measuring the DNA species involved, the study provides a mechanistic framework for TREX1 inhibitors now in development as cancer immunotherapies: the target substrate is not micronuclear DNA but short, nucleosome-sized, transcription-derived fragments that, when allowed to accumulate, turn a cold tumor into one the immune system can remember.

Subject of Research: How the TREX1 exonuclease degrades short cytosolic DNA fragments to suppress anti-tumor immunity in murine cancer models

Article Title: TREX1 limits anti-tumor immunity by degrading short cytosolic DNA molecules

Article References: Yang, C.-Y., Chami, B., Ursino, C., Johnson, S. M., Faget, J., Mettling, C., Gros, L., Pasero, P., & Lin, Y.-L. (2026). TREX1 limits anti-tumor immunity by degrading short cytosolic DNA molecules. EMBO Reports. https://doi.org/10.1038/s44319-026-00961-8

Image Credits: AI Generated

DOI: 10.1038/s44319-026-00961-8

Keywords: TREX1, cytosolic DNA, cGAS-STING, MRE11, micronuclei, interferon signaling, macrophage differentiation, anti-PD-1, immune memory, cancer immunotherapy, tumor immunogenicity, murine cancer models