Genomic instability is a hallmark of neoplastic transformation that leads to the accumulation of mutations, and generates a state of replicative stress in neoplastic cells associated with dysregulated DNA damage repair (DDR) responses. The importance of increasing mutations in driving cancer progression is well established, whereas relatively little attention has been devoted to the DNA displaced to the cytosol of cancer cells, a byproduct of genomic instability and of the ensuing DDR response. The presence of DNA in the cytosol promotes the activation of viral defense pathways in all cells, leading to activation of innate and adaptive immune responses. In fact, the improper accumulation of cytosolic DNA in normal cells is known to drive severe autoimmune pathology. Thus, cancer cells must evade cytoplasmic DNA detection pathways to avoid immune-mediated destruction. The main sensor for cytoplasmic DNA is the cyclic GMP–AMP synthase, cGAS. Upon activation by cytosolic DNA, cGAS catalyzes the formation of the second messenger cGAMP, which activates STING (stimulator of IFN genes), leading to the production of type I interferon (IFN-I). IFN-I is a critical effector of cell-mediated antiviral and antitumor immunity, and its production by cancer cells can be subverted by several mechanisms. However, the key upstream regulator of cytosolic DNA-mediated immune stimulation is the DNA exonuclease 3′-repair exonuclease 1 (TREX1). Here, we will discuss evidence in support of a role of TREX1 as an immune checkpoint that, when up-regulated, hinders the development of antitumor immune responses.
Immune activation directed to nucleic acids is required to maintain genome integrity and is an evolutionarily conserved response found in the earliest forms of life . In eukaryotes, self-DNA is spatially separated from the cytosol by the nuclear membrane and several DNA sensors are present in the endosomal and cytosolic compartments to detect DNA of invading viruses and bacteria. Interestingly, while some sensors bind pathogen-specific DNA motifs (e.g. unmethylated CpG), others do not discriminate between self and non-self DNA and are activated following recognition of double-stranded (ds)DNA that ‘leaks’ from the nucleus to the cytosol . Intriguingly, endogenous DNA can accumulate in the cytosol under homeostatic conditions and trigger autoimmune pathology that is driven by type I interferon (IFN-I) in the absence of the cytoplasmic DNase III, better known as 3′-repair exonuclease 1, TREX1 [3,4]. Thus, leakage of DNA from the nucleus to the cytoplasm occurs under physiologic conditions. While the mechanisms whereby this occurs remain incompletely understood, DNA replication byproducts and endogenous retroviruses were shown to accumulate in the cytosol of TREX1-deficient cells [3,4]. Another condition leading to accumulation of IFN-stimulatory cytosolic DNA in normal cells is ataxia-telangiectasia mutated (ATM) deficiency. Interestingly, in ATM-deficient cells, this occurs without the exposure to genotoxic agents, indicating that a functional DNA damage repair (DDR) response is required for the daily metabolic activities of nontransformed cells . Overall, these data suggest that leakage of endogenous DNA into the cytosol is not an accidental event, but a physiological process that is tightly regulated by DDR effector molecules and by TREX1. Recent data by Härtlova et al.  support the interpretation that this process is fundamental for the constitutive production of basal levels of IFN-I that maintain the innate immune system readiness to fight infectious agents.
In neoplastic cells, down-regulation of DDR mechanisms is common and may be required to allow the proliferation of genetically unstable cells . Thus, cancer cells’ progression from a pre-cancerous to a cancerous state may be intimately linked to the accumulation of cytosolic IFN-stimulatory DNA. Activation of IFN-I at this critical step in carcinogenesis also provides key signals to the immune system to drive immunosurveillance and immunoediting [7–9]. To survive and expand, cancer cells must develop one of these three properties: (1) prevent innate immune sensing of cytosolic DNA, (2) subvert the molecular programs driven by IFN-I to induce pro-tumorigenic gene expression programs and (3) develop several immune suppressive mechanisms to block ongoing antitumor immune responses. These mechanisms are not mutually exclusive and often coexist. Here, we will focus on the regulation of cytosolic DNA by TREX1, the enzyme chiefly responsible for ‘cleaning up’ the cytosol from the byproducts of DNA damage, and repair processes to maintain host innate immune tolerance to self-DNA, in the steady state and during genotoxic stress.
Innate immune sensing of DNA in cancer
Cytosolic dsDNA is sensed by the cyclic GMP–AMP synthase (cGAS, also known as Mab-21 domain containing 1 — MB21D1) [10–12]. Cyclic guanosine monophosphate–adenosine monophosphate (cGAMP) produced by cGAS upon activation by dsDNA binds to and activates the endoplasmic reticulum-localized protein STING (stimulatory IFN genes; also known as transmembrane protein 173 — TMEM173) to induce IFN-I production through the engagement of the TANK-binding kinase 1 (TBK1) and the following activation of transcription factors IRF3 (interferon regulatory factor 3) and NF-κB (nuclear factor kappa-B) .
cGAS has been thought to bind DNA without a sequence specificity or length restriction. However, recent evidence indicates that both length and structure of the DNA may be important. cGAS was shown to bind long dsDNA or short dsDNA with unpaired open ends containing guanosines (Y-form DNA), which is present in highly structured single-stranded DNA of retroviruses and some endogenous retroelements . In another report, Andreeva et al. demonstrated a DNA length-dependency of cGAS activation. Importantly, they also showed that the histone-like mitochondrial nucleoid proteins HU and the mitochondrial transcription factor A (TFAM), as well as high-mobility group box 1 protein (HMGB1), stimulated long DNA sensing by cGAS, supporting the preferential binding of cGAS to nucleoid-like structures or bent DNA .
The nature of the DNA that accumulates in the cytosol of cancer cells and the exact mechanisms regulating its levels remain incompletely understood, but genotoxic stress induced by treatments such as radiotherapy can induce a rapid increase in cytosolic dsDNA . Micronuclei form as a result of chromosome damage and are frequently observed in cells undergoing transformation, representing a potential biomarker of cancer risk, as well as a mechanism of mutagenesis [17,18]. New evidence by Mackenzie et al.  shows that breakdown of the micronuclear membrane during chromothripsis leads to exposure of the DNA contained within micronuclei to cGAS and downstream activation of IFN-I, thus possibly representing a cell-intrinsic immune surveillance mechanism. Interestingly, micronuclei can also form following treatment with DNA-damaging agents such as ionizing radiation in cancer cells that undergo division without completing DNA repair. DNA released from these micronuclei then activates cGAS explaining cancer cell-intrinsic IFN-I production that occurs few days following radiation, but not the earlier increase in cytosolic dsDNA and IFN-I activation detectable within 24 h [16,20]. Thus, multiple mechanisms contribute to DNA translocation to the cytosol of cancer cells. It remains to be established how mutations in different DDR components may affect this process and the pathways by which it occurs.
Role of TREX1 as a gatekeeper of cGAS-mediated IFN-I activation
Although beneficial in jumpstarting antitumor immunity, inappropriate activation of intracellular DNA-sensing pathways is responsible for the pathogenesis of autoimmune and inflammatory disorders . Under normal physiology, improper activation of IFN-I by endogenous DNA is prevented by the coordinate action of intracellular and extracellular DNA nucleases that continuously dispose of self-DNA . Among them, as mentioned above, TREX1 is a critical regulator of the cytosolic DNA capable of activating cGAS.
Originally designated as DNase III, TREX1 is the most abundant 3′ → 5′ DNA exonuclease in cells [23,24]. A similar amino acid sequence to E. coli proofreading exonucleases and the ability to remove mismatched 3′-terminal bases first suggested that TREX1 might edit DNA during its replication [23,24]. However, Trex1-deficient mice did not show increased mutations or cancer frequency, but experienced a profound immune disorder with a severe reduction of survival due to excessive IFN-I-driven inflammation . Importantly, the phenotype of Trex1-deficient mice is similar to that of patients diagnosed with the Aicardi–Goutières Syndrome (AGS), a genetic disorder due to mutations of Trex1 [3,26,27].
TREX1 deficiency in mice and patients leads to impaired clearance of endogeneous retroelements and/or byproducts of defective DNA replication resulting in aberrant accumulation of cytosolic DNA species [3,4,26]. Importantly, the pathology elicited by Trex1 deficiency in mice required expression of STING, TBK1 and IRF3, providing evidence that it is mediated by the accumulation of IFN-stimulatory cytosolic DNA [3,26]. Similar findings were reported in Trex1-deficient mice lacking cGAS, demonstrating that cGAS is the upstream sensor of self-DNA that drives the subsequent activation of IFN-I via the STING–TBK1–IRF3 axis [28–30].
Role of TREX1 as an immune checkpoint in cancer
Little information is available about TREX1 regulation and function in tumors. The microRNA-103 (miR-103) has been recently identified as a regulator of TREX1 expression in endothelial cells following genotoxic treatment. Vascular-targeted delivery of miR-103 not only inhibited angiogenesis in mouse tumor models but also increased the production of pro-inflammatory cytokines due to TREX1 down-regulation, shifting the tumor microenvironment from immune suppression toward immune activation . Although the role of miR-103 in regulating the expression of Trex1 in cancer cells remains to be investigated, these data support a role for TREX1 as an immune checkpoint that hinders antitumor immune responses.
More compelling evidence for such a role of TREX1 comes from our recent studies, showing that TREX1 up-regulation in cancer cells abrogates tumor response to immune checkpoint inhibitors . In mouse breast and colorectal carcinomas resistant to anti-cytotoxic T-lymphocyte associated protein-4 (CTLA-4) or anti-program cell death-1 (PD-1), local tumor radiotherapy induced the production of IFN-I by the cancer cells. This effect was mediated by a marked and rapid increase in dsDNA in the cytosol of the irradiated cancer cells and was abrogated by knockdown of cGAS or STING and by forced up-regulation of TREX1 in the cancer cells. The production of IFN-I by the cancer cells elicited via cGAS/STING pathway activation was required to attract to the tumor and activate basic Leucine Zipper ATF-like transcription factor 3 (BATF3)-dependent dendritic cells (DCs), which are specialized in cross-presenting tumor antigens to CD8 T-cells [16,32,33]. Thus, TREX1, by controlling the accumulation of IFN-stimulatory DNA, effectively inhibited the development of systemic antitumor T cells required to mediate abscopal effects (i.e. rejection of tumors outside of the radiation field) .
Importantly, cancer cell-derived DNA has been shown to stimulate IFN-I production via the cGAS/STING pathway in DCs infiltrating the tumor by gaining access to their cytoplasm via incompletely defined mechanisms [34,35]. Thus, it is likely that TREX1 not only works as a break for cancer cell-intrinsic activation of cGAS/STING but also limits the tumor DNA available to activate this pathway in DCs. In support of this hypothesis, we have data indicating that TREX1 expressed by cancer cells limits STING-dependent activation of DCs exposed to irradiated cancer cell-derived material (Diamond et al. submitted).
The activation of IFN-I pathway can lead, in some cases, to the expression of interferon-stimulated genes associated with immune tolerance and resistance to therapy [36,37]. In such cases, TREX1 seems to play a beneficial role by preventing the induction of IFN-I by cytosolic DNA, as shown recently by Erdal et al.  in breast cancer cells.
Overall, the emerging evidence indicates that TREX1 is an important regulator of tumor immunogenicity, and supports additional investigations to further dissect TREX1 as a potentially druggable target in cancer.
TREX1 modulation by cancer therapy
In homeostatic conditions, basal TREX1 expression is sufficient to keep the cytosol clean from IFN-stimulatory DNA. In cancer, a chronic imbalance between TREX1 and cytosolic DNA levels could foster IFN-driven chronic inflammation that supports tumor progression .
The relationship between TREX1 and cytosolic DNA is more complex following genotoxic cancer therapy. In several human and mouse carcinoma cells, cytosolic DNA increased within 24 h after a single radiation dose above 2–4 Gy, reaching a plateau at doses between 8 and 10 Gy. However, there was a dose size threshold, varying between 12 and 15 Gy in most carcinoma cells tested, above which cytosolic DNA plunged due to the up-regulation of TREX1 . Thus, there is a limited range of radiation doses that stimulate this rapid DNA accumulation in the cytosol of cancer cells and lead to activation of cGAS/STING pathway and antitumor immunity. The radiation dose-dependent mechanisms that induce Trex1 up-regulation remain to be defined and are under active investigation. Nevertheless, these findings have great implications for the use of radiation in combination with immunotherapy [40,41].
Up-regulation of Trex1 has also been described in glioma and melanoma cells after treatment with cross-linking agents (i.e. nimustine, carmustine, fotemustine and the topoisomerase inhibitor topotecan) and ultraviolet light. At the cancer cell level, it seems to be part of a survival response evoked by genotoxic stress [42,43]. However, an equally important implication is the tolerogenic effect of removing the adjuvant activity of cytosolic DNA. Interestingly, ultraviolet light caused oxidative modifications in the DNA that become resistant to degradation by TREX1, while retaining the ability to bind to cGAS, indicating that in some situations cytosolic DNA may escape the control by TREX1 .
Thus, the limited information available supports the possibility that induction of TREX1 is a common response to cancer treatment that may control the immunogenicity of therapy-induced cell death .
This is a critical time in cancer immunology and anticancer immunotherapy. The success of therapies targeting immune checkpoints that inhibit antitumor immune responses has enabled the study of responders and non-responders, fostering a tremendous progress in our understanding of previously unappreciated cross-talks between neoplastic and immune cells [46–48]. Targeting inhibitory receptors expressed by T cells, such as CTLA-4 and PD-1, is a very effective treatment in some patients, but clearly is not sufficient in the majority of patients. Thus, new strategies are required, based on the emerging evidence that tumor immune resistance is a complex phenomenon with multiple levels of regulation.
In this context, improved understanding of the mechanisms that regulate self-tolerance and response to invading pathogens is critical since neoplastic cells have features of both, self and non-self, as best exemplified by the role of neoantigens generated by mutations in tumor rejection .
Innate immune sensing of cytosolic DNA is a process that is central to self-tolerance, antiviral immunity and genomic instability associated with neoplastic transformation (Figure 1). As the main regulator of cytosolic DNA levels, TREX1 can play a fundamental role in cancer immune resistance. Importantly, given the induction of TREX1 by high radiation doses and by many genotoxic chemotherapy agents, interventions to block TREX1 activity within the tumor during treatment may be required to enhance the occurrence of immunogenic cell death and the ability of these treatments to synergize with immunotherapy. With a growing number of combinations of cytotoxic treatments being tested for the ability to enhance immunotherapy responses in clinical trials [50–52], more investigations into the role of TREX1 and of other regulators of nucleic acids sensing by the immune system in response to treatment are urgently needed.
TREX1 is a checkpoint for cGAS-mediated IFN-I activation by cytosolic DNA.
Dysregulation of DNA damage repair responses in cancer cells leads to the accumulation of DNA in the cytosol that can be exacerbated by genotoxic therapy.
Cytosolic DNA is sensed by cGAS which activates viral defense pathways via the adaptor protein STING, culminating in the production of interferon type I and up-regulation of interferon-stimulated genes.
Interferon type I activates antigen-presenting cells and antitumor immune responses.
Cancer cells escape immune-mediated control by interfering with the activation of immune effectors at multiple levels.
The exonuclease TREX1 is responsible for degrading cytosolic DNA to prevent autoimmunity.
In cancer cells, TREX1 degrades the interferon stimulatory DNA that accumulates spontaneously and is enhanced by treatment, precluding antitumor immune activation.
TREX1 up-regulation has been described in cancer cells in response to genotoxic agents including radiotherapy.
Targeting of TREX1 should be explored to enhance the efficacy of combinations of genotoxic agents with immunotherapy.
basic Leucine Zipper ATF-like transcription factor 3
cyclic GMP–AMP synthase
cytotoxic T-lymphocyte associated protein-4
DNA damage repair
type I interferon
interferon regulatory factor 3
program cell death-1
stimulator of IFN genes
TANK-binding kinase 1
3′-repair exonuclease 1
S.D. is supported by NIH [R01CA201246 and R01CA198533], the Breast Cancer Research Foundation [BCRF-16-054] and the Chemotherapy Foundation. C.V.-B. is supported, in part, by a post-doctoral fellowship from the DOD BRCP [W81XWH-13-1-0012] and by the 2017 Kellen Junior Faculty Fellowship from the Anna-Maria and Stephen Kellen Foundation.
S.D. is the recipient of a research grants from Lytix Biopharma and has received honorarium for consultant/advisory services from Eisai, Inc., EMD Serono, Lytix Biopharma, Nanobiotix and StemImmune.