Genomic instability and innate immunity

We established that ER-anchored TREX1 degrades cytosol-exposed DNA at micronuclei and chromosome bridges, and that it limits cGAS activation at these sites. An unexpected feature of this regulation is that TREX1 requires its association with the ER in order to reach and resect micronuclear DNA, even though that DNA lies free in the cytoplasm once the micronuclear envelope has ruptured.

TREX1 must also contend with the physical environment cGAS itself creates, since cGAS-DNA phase separation restricts access to cytosolic DNA.

The requirement for ER tethering was paradoxical, since ruptured micronuclei are open to the cytoplasm. We resolved it by identifying barrier-to-autointegration factor, BAF, as a regulator of TREX1 activity at micronuclei.

BAF accumulates rapidly on ruptured micronuclei through direct binding to cytosol-exposed DNA and concentrates additional factors there, including membrane-associated LEM-domain proteins. Rather than promoting degradation, BAF shields micronuclear DNA. Recombinant BAF inhibited TREX1 nuclease activity in vitro in a manner requiring BAF DNA binding, since a DNA-binding-deficient mutant failed to inhibit.

A soluble TREX1 lacking its transmembrane domain failed to accumulate at ruptured micronuclei and was defective for resection. Depleting BAF restored resection by the soluble enzyme without restoring its localization, indicating that ER tethering allows TREX1 to bypass the barrier rather than to localize through it. BAF likewise competes with cGAS for micronuclear DNA, and BAF depletion increased cGAS accumulation and raised cGAMP levels.

The barrier therefore limits the diffusive entry of DNA-binding proteins at ruptured micronuclei, accounts for the dependence of TREX1 on ER tethering, and identifies BAF as a gatekeeper that simultaneously admits TREX1 and restrains cGAS engagement with immunostimulatory DNA.

Chromosomally unstable tumors resolve the tension between instability and immune visibility by inducing TREX1. cGAS-STING signaling drives TREX1 upregulation, the induced enzyme digests cytosolic DNA, and the response is dampened. The result is an adaptive negative feedback loop that enables immune evasion.

To test whether this makes TREX1 a therapeutic liability, we generated Trex1 knockout derivatives of the genomically unstable, triple-negative breast cancer line EO771.LMB. TREX1 loss markedly increased IFN-β secretion in a cGAS-dependent manner. In syngeneic hosts, Trex1 knockout slowed tumor growth, potentiated the response to anti-PD-1 checkpoint blockade, prolonged host survival, and increased intratumoral CD8 T-cell infiltration, again requiring cGAS. These effects were selective for tumors that retained the ability to mount a type I interferon response downstream of STING.

TREX1 induction is therefore an innate immune checkpoint that shields chromosomally unstable tumors from surveillance, and its removal is a strategy for restoring antitumor immunity in tumors poised to respond. Concordant reports from several groups appeared alongside ours, and subsequent studies have shown that acute TREX1 inactivation is well tolerated and effective in combination with checkpoint blockade. Together this work has made a strong case that TREX1 is a tractable target in genomically unstable cancers, and developing that into a therapeutic strategy is a central effort of the lab.