Adaptive inhibition of CGAS signaling by TREX1

Abstract

Mammalian cells react to the accumulation of double-stranded DNA in the cytosol by secreting antiviral and pro-inflammatory cytokines, notably type I interferon. Recent data demonstrate that overactivation of this pathway is prevented by an adaptive feedback mechanism elicited by type I interferon receptors and executed by the exonuclease TREX1.

Mammalian cells detect cytosolic double-stranded DNA (dsDNA) accumulation as promoted (among other sources) by rupturing micronuclei through cyclic GMP-AMP synthase (CGAS), resulting in the elicitation of a stimulator of interferon response cGAMP interactor 1 (STING1)-dependent signal transduction cascade [1]. STING1 activation by CGAS not only results in the execution of NF-κB- and interferon regulatory factor 3 (IRF3)-mediated transcriptional programs coupled with the secretion of pro-inflammatory cytokines including type I interferon (IFN), but also in the activation of specific cell death programs (at least in some cells) [2]. In line with this notion, the CGAS-dependent detection of cytosolic dsDNA species aberrantly localized in the cytosol mediates critical cell-extrinsic (through the engagement of the immune system) and cell-intrinsic (through the direct elimination of infected or potentially transformed cells) antimicrobial and oncosuppressive effects [1,2]. Excessive CGAS signaling, however, may also be detrimental, as illustrated by the so-called Aicardi–Goutières syndrome (AGS) [3]. AGS is indeed an inheritable inflammatory disorder associated with mutations in three prime repair exonuclease 1 (TREX1), which encodes an exonuclease that efficiently degrades cytosolic dsDNA species, hence effectively shutting down CGAS signaling [4]. Recent data demonstrate that TREX1 levels increase in response to accrued type I IFN signaling, delineating a novel feedback loop through which malignant cells keep CGAS activation at bay in support of tumor progression and immune evasion [5].

Tani and collaborators set to understand the transcriptional programs associated with micronucleation as elicited by pulsed pharmacological TTK protein kinase (TTK, best known as MPS1) inhibition in human non-small cell lung carcinoma (NSCLC) H1944 cells bearing KRAS and STK11 mutations, demonstrating robust CGAS activation and consequent upregulation of multiple interferon-stimulated genes (ISGs) as well as TREX1. Immunoblotting experiments confirmed that pulsed MPS1 inhibition increases TREX1 abundance in H1944 cells and showed that this is accompanied by the phosphorylation of signal transducer and activator of transcription 1 (STAT1), which is known to participate in type I IFN signaling [6]. STAT1 deletion abrogated TREX1 upregulation as elicited by pulsed MPS1 inhibition in H1944 cells. Conversely, TREX1 deletion elicited STAT1 phosphorylation coupled with ISG induction and secretion of pro-inflammatory cytokines even at baseline, an effect that could be rescued by the transgene-enforced expression of wild-type, but no catalytically inactive, TREX1. Collectively, these findings demonstrate that TREX1 is upregulated as an adaptive response to type I IFN signaling as elicited by micronucleation, which TREX1 limits via its exonuclease activity [5]. Next, Tani and co-workers aimed at understanding the consequences of TREX1 deletion in multiple human NSCLC cell lines with different CGAS and STING1 competence. They found that both CGAS and STING1 expression is required for the loss of TREX1 to elicit STAT1 phosphorylation and the secretion of ISG-encoded cytokines including interferon beta 1 (IFNB1). These findings were validated by deleting CGAS and STING1 in H1944 cells, overall confirming that TREX1 suppresses an interferogenic signal transduction cascade that relies on CGAS and STING1 [5].

As type I IFN activates STAT1 [6], and the latter turned out to be required for adaptive TREX1 induction in the context of micronucleation, Tani and colleagues opted to examine the role of autocrine/paracrine IFNB1. Both IFNB1 and the ISG-encoded chemokine C-X-C motif chemokine ligand 10 (CXCL10) were enriched in the supernatant of TREX1^−/− H1944 cells. Moreover, recombinant IFNB1 administration promoted STING1 and TREX1 expression in a dose-dependent manner and through a mechanism that depended on both STAT1 and interferon alpha and beta receptor subunit 1 (IFNAR1), not only in STING1-competent H1944 cells, but also in NSCLC cell lines with epigenetic STING1 suppression, such as A549 cells. These findings suggested that phosphorylated STAT1 can compete with epigenetic modification that silence STING1 expression. In line with this possibility, A549 cells treated with epigenetic modifiers exhibited an accrued sensitivity to recombinant IFNB1 with respect to STING1 and TREX1 expression. Moreover, IFNAR1 deletion prevented the upregulation of STING1 as elicited by the loss of TREX1 in H1944 cells, collectively demonstrating that autocrine/paracrine type I IFN signaling co-regulate STING1 and TREX1 expression in the context of a feedback inhibitory loop that can be disrupted by TREX1 inhibition [5].

In line with the potent immunostimulatory effects of type I IFN, TREX1^−/− H1944 cells secreted a broad array of NF-κB- and IRF3-dependent cytokines as compared to their control counterparts. Accordingly, TREX1^−/− H1944 spheroids were more proficient than their wild-type counterparts at recruiting immune cell effectors, notably CD8⁺ T cells and natural killer (NK) cells, in microfluidic 3D co-culture experiments. Moreover, the loss of TREX1 promoted the upregulation of the NK cell activating ligand MHC class I polypeptide-related sequence A (MICA) on the surface of H1944 cells, correlating with increased cancer cell lysis by NK cells in co-culture experiments as a consequence of (1) accrued interferon gamma (IFNG) secretion, and (2) superior cancer cell sensitivity to IFNG itself, at least in part as mediated by accrued STAT1 signaling as driven by IFNG [5,7].

Next, Tani and co-workers interrogated publicly available transcriptomic datasets to show that TREX1 levels are increased in malignant tissues as compared to their normal counterparts across a panel of human malignancies. Moreover, in patients with melanoma, TREX1 expression correlated with STING1 abundance as well as with a transcriptional signature of type I IFN signaling. Consistent with a broader role for the ability of TREX1 to adaptively suppress type I IFN signaling, TREX1 deletion increased IFNB1 secretion and (in some instances) promoted cell death in a large panel of cancer cell lines of diverse histological derivation. Moreover, immune checkpoint inhibitor (ICI)-refractory mouse melanomas established in immunocompetent host with Trex1^−/− B16F10 cells not only displayed a significant increase in NK cell infiltration and transcriptional signs of immune activation at baseline, but also responded to an ICI specific for programmed cell death 1 (PDCD1, best known as PD-1), an increased sensitivity correlating with an even more pronounced infiltration by immune effector cells encompassing NK cells and CD3⁺ T cells. Similar findings were obtained with mouse CT26 colorectal cancers and mouse 393P-KL lung cancers established subcutaneously in immunocompetent, syngeneic hosts. Along similar lines, the loss of Trex1 sensitized B16F10 melanomas to both MPS1 inhibitors and STING1 agonists [5].

In summary, these findings confirm and extend previous observations pointing to TREX1 as a promising target for the development of novel combinatorial partners for ICIs and potentially other therapeutic agents such as radiation therapy [4,8]. Alongside they formalize the existence of a feedback loop directly linking type I IFN signaling to the inhibition of nucleic acid sensing (Fig. 1). Whether TREX1 also degrades cytosolic mitochondrial DNA (mtDNA), another potent driver of type I IFN secretion in stressed cells [9,10], remains to be formally demonstrated. Despite these and other unresolved questions, Tani and colleagues elegantly delineated yet another homeostatic mechanism that mammals employ to limit potentially pathogenic inflammatory reactions, with potential therapeutic applications in cancer and inflammatory disorders. To the best of our knowledge, however, additional work is needed for pharmacological TREX1 modulators to reach clinical development.

Figure 1. Adaptive inhibition of type I IFN signaling by TREX1.

Genomic instability may be associated with the accumulation of micronuclei, structurally impaired portions of the nucleus that can elicit cyclic GMP-AMP synthase (CGAS) signaling and consequent stimulator of interferon response cGAMP interactor 1 (STING1) activation, which is coupled with the recruitment of TANK binding kinase 1 (TBK1) to the endoplasmic reticulum (ER) surface. Active TBK phosphorylates interferon regulatory factor 3 (IRF3) to promote its translocation to the nucleus, where is transactivate multiple genes including interferon beta 1 (IFNB1). This results in the secretion of active IFNB1, which signals through a heterodimeric receptor that consists of interferon alpha and beta receptor subunit 1 (IFNAR1) and IFNAR2 and operates via signal transducer and activator of transcription 1 (STAT1). Recent data demonstrate that three prime repair exonuclease 1 (TREX1) one of the genes adaptively transactivated by active STAT1, resulting in the expression of a cytosolic nuclease that actively degrade interferogenic DNA species and hence suppresses CGAS activation. cGAMP, 2’3’-cyclic GMP-AMP; P, inorganic phosphate.