Activation and regulation of cGAS-STING signaling in cancer cells

Micronuclei

Micronuclei are small extranuclear compartments containing chromosomes or chromosome fragments that arise from errors in cell division (Figure 2).^63–65 Micronuclei are indicative of genotoxic stress and are a common feature across multiple cancer types.^63,64 Unlike the primary nucleus, micronuclei often undergo irreversible membrane rupture.^66–68 The causes of micronuclear membrane rupture are not completely understood, but involve defects in micronuclear envelope assembly, protein transport defects, aberrant ATR activity, unrestrained ESCRT-III activity, and oxidative damage resulting from exposure to mitochondria-derived reactive oxygen species.^69–75 Regardless of the precise underlying mechanisms, micronuclear membrane collapse results in a loss of compartmentalization that drives depletion of soluble micronuclear proteins and invasion of cytoplasmic factors.^67,76 Unsurprisingly, micronuclear collapse is associated with broad dysfunction including extensive DNA damage, DNA replication defects, transcriptional silencing and epigenetic aberrations.^67,77–81

One striking consequence of micronuclear membrane collapse is the rapid localization of cGAS to micronuclear chromatin.^25,26,82 cGAS is recruited to micronuclei generated by a range of exogenous insults, including ionizing radiation, RNase H2 loss, and Mps1 inhibition, as well as by intrinsic genomic instability common in many cancer cells.^{25,26,28,29,83–87} However, while cGAS localization to ruptured micronuclei is broadly observed, it is influenced by the type of genotoxic stress, which affects micronuclear chromatin organization.^28,76 Therefore, it is possible that certain genotoxic stressors or chemotherapies may possess enhanced immunostimulatory potential due to their impact on micronuclear chromatin content.

In keeping with its inability to distinguish between genomic and pathogen-derived DNA, cGAS localization to micronuclei is associated with a wide-ranging, pro-inflammatory transcriptional response that closely resembles its function during pathogen infection. Indeed, micronucleated cells show cGAS-STING-dependent increases in IFN-stimulated gene (ISG) expression.^25,26,83 Interestingly, pathway activation appears to peak following a delay and requires ongoing cell cycle progression.^25,88 cGAS-STING-dependent inflammation following micronucleation has been proposed to be a key intermediate in the efficacy of therapies like radiation, DNA repair inhibition, or immune checkpoint blockade.^{17,25,89–92} Genome integrity factors like SMARCAL1 and BLM are thought to promote tumor growth by limiting DNA damage and micronucleation, thereby suppressing cGAS activation and ISG responses.^29,85 In line with this observation, Bloom syndrome, which is driven by mutations in BLM, is associated with heightened ISG levels in the peripheral blood.²⁹ Thus, micronucleation plays critical roles in cancer and human pathogenesis by activating a cGAS-STING-dependent immune response.

The prevailing model that cGAS senses genomic DNA upon micronuclear membrane collapse offers a logical framework, yet emerging evidence reveals shortcomings that underscore gaps in current knowledge. In particular, the discovery of a catalytically silenced nuclear pool of cGAS has raised the question of whether cGAS is catalytically active following recruitment to micronuclear chromatin.^93–95 Current understanding of nuclear cGAS functions is limited, but roles in genome integrity and antiviral DNA sensing have been proposed.^95–102 Regardless of its precise functions, nuclear cGAS is broadly observed across many human and mouse cell lines.^93,94,100 These observations challenged early models that cGAS cytosolic localization prevents autoimmunity by sequestering cGAS away from genomic DNA.^26,103 This puzzle was solved by an elegant series of biochemical and structural studies showing that nuclear cGAS is bound tightly to chromatin and rendered inactive through interactions with the nucleosome acidic patch (Figure 3).^93,104–109 The importance of these inhibitory interactions is underscored by mutations in key residues like R236 and R255, which mediate this binding. Mutating these residues untethers cGAS from chromatin and leads to autoreactivity against self-DNA.^93,110 In addition to inhibition through nucleosome tethering, nuclear cGAS is inhibited through dynamic competition with the dsDNA-binding protein BAF and through targeted degradation by the CRL5-SPSB3 ubiquitin ligase.^110,111

Emerging models of cGAS inhibition through nucleosome tethering raise questions as to how this regulatory mechanism may be circumvented at micronuclei, which are typically composed of chromatin.^63,77,80 Although nucleosome core particles inhibit cGAS activity, the enzyme can be activated by chromatin extracted from cells or reconstituted nucleosome arrays, particularly under conditions that promote higher-order compaction.^26,108 At elevated cGAS concentrations, such as those found in micronuclei, activation can occur even in the absence of array compaction.¹⁰⁸ These findings suggest that cGAS activation within micronuclei may result from locally high enzyme concentrations or distinct compaction states of micronuclear chromatin, enabling activation despite the presence of nucleosomes that inhibit cGAS in the primary nucleus. Another interesting possibility is that micronuclei may possess unique mechanisms to escape nucleosome-imposed regulation. Virology studies provide a precedent for this possibility, as HSV-1 infection untethers nucleosomal cGAS, generating a nuclear pool of catalytically active cGAS that restricts HSV-1 replication.⁹⁶ Similarly, MRE11 displaces cGAS from nucleosomal sequestration independently of its catalytic activity, and loss of MRE11 significantly attenuates cGAS activation upon transfection of ISD or nucleosome core particle.¹¹² Yet, how MRE11 overcomes strong cGAS-nucleosome binding affinity, and how soluble cGAS within the nucleus can sustain its liberated state without reattaching to another nucleosome, remain unknown.

An alternative possibility is that chromatin disorganization or epigenetic changes in micronuclei liberate cGAS from chromatin-imposed inhibition.^28,79,80 Indeed, recent studies show that specific chromatin features, such as histone H3 lysine 79 dimethylation (H3K79me2), are established prior to DNA damage, retained in ruptured micronuclei, and promote cGAS activation in this context.²⁸ In addition, this work showed that cGAS variants with impaired nucleosome binding unexpectedly led to reduced cGAMP production and ISG expression, underscoring the complex interplay between cGAS and nucleosomes.²⁸ Furthermore, disruption of linker histone stoichiometry has been linked to elevated cGAMP levels and inflammatory signaling in Aicardi-Goutières syndrome (AGS) models.¹¹³ Further work will be necessary to understand how micronuclei may evade chromatin-imposed cGAS regulation.

The model of cGAS activation at micronuclei is largely based on population-based measures that query the cGAS-STING pathway at multiple points, such as cGAMP levels or ISG expression, and then correlate these findings with frequencies of cGAS-positive micronuclei. Recent publications have used single-cell reporters, such as GFP-IRF3 nuclear translocation or STING conformational changes, to show that genotoxic agents like hydroxyurea or irradiation can robustly induce cGAS-positive micronuclei while failing to elicit strong downstream pathway activation within early time windows (i.e., <48 hours).^114–116 In contrast, a single-cell reporter based on a split-GFP system, designed to infer cGAS oligomerization through GFP reconstitution, demonstrated that a subset of micronuclei exhibited strong GFP fluorescence, consistent with the presence of catalytically active cGAS within these compartments.¹¹⁷ This upstream readout of cGAS activation complements downstream STING-based sensors, which may be subject to confounding effects from paracrine cGAMP transfer between neighboring cells.^118,119

Longer-term assays have shown that ISG expression becomes detectable at later timepoints, typically 3 to 6 days following genotoxic insult, suggesting a delayed but biologically meaningful cGAS-dependent immune response.^25,88 These differences may, at least in part, reflect variation in the timing and sensitivity of ISG detection across studies, rather than representing a definitive contradiction of the micronucleus activation model. Genotoxic perturbations capable of inducing micronuclei can simultaneously trigger immunostimulatory DNA release via other mechanisms, such as mitochondrial stress or DNA bridge formation, further complicating interpretation.^115,120 Collectively, these findings support a model in which micronuclei can activate cGAS but suggest that kinetics, context, and co-occurring damage responses influence the magnitude and timing of downstream signaling. Additional research is required to evaluate whether currently available single-cell cGAS reporters possess adequate sensitivity to definitively assess the immunogenic capacity of micronuclei. As detailed later, it will also be important to define how key cGAS-STING regulators, including TREX1 and BAF, modulate cGAS activation in this context.

DNA Bridges

DNA bridges are aberrant chromatin structures that form when a single chromosome is physically tethered between two daughter nuclei following cell division (Figure 2). These aberrations are often triggered by dicentric chromosomes, or chromosomes with two centromeres, that can form following depletion of the telomere reserve during telomere crisis or mis-repair of DNA double-strand breaks.^7,8,121 Pulling forces generated by migrating daughter cells can stretch DNA bridges into cell-connecting fibers that can stretch beyond 200 μm in length in cell culture systems.^7,122 Although initially compartmentalized, cellular pulling forces disrupt normal nuclear lamina protections at DNA bridges, resulting in apparent membrane collapse and exposure of genomic DNA to the cytosol.⁷ Similar to micronuclei, cGAS accumulates on cytosol-exposed DNA bridges, where its localization correlates with increased cGAMP production and ISG expression.⁸³ DNA bridges are proposed to serve as more potent activators of cGAS than micronuclei.¹¹⁵ Examination of a panel of genotoxic stressors revealed that optimal cGAS activation most strongly correlated with identification of cGAS-coated DNA bridges rather than micronucleation rates.¹¹⁵ Mechanistically, cellular pulling forces may be sufficient to disrupt chromatin architecture to an extent that alleviates nucleosome-mediated cGAS inhibition. In support, forces of 20 pN are sufficient to mechanically remove histone cores from native chromatin.¹²³ Alternatively, transcription of telomeric-repeat-containing RNA (TERRA) from dysfunctional telomeres is proposed to amplify innate immune responses during telomere crisis through a MAVS-dependent RNA sensing mechanism.¹²⁴ During telomere crisis, the buildup of DNA bridges and other cytosolic DNA species triggers autophagy, promoting cell death as a mechanism to limit chromosomal instability.¹²⁵

While these findings support a model in which DNA bridges are robust triggers of cGAS-STING signaling, conflicting results across experimental systems point to important contextual variables. For example, one study employing enforced telomere fusions via TRF2ΔBΔM expression^7,126 generated persistent, elongated chromatin bridges without detectable cGAS activation or micronucleation, suggesting that bridge formation alone may be insufficient to trigger immune sensing.⁸³ Discrepancies with other systems may stem from differences in cell type, rupture dynamics, or the biochemical features of exposed DNA. Further studies are needed to define the molecular and biophysical conditions that lead to cGAS activation at DNA bridges and to determine whether DNA bridges activate cGAS more efficiently than other cytosolic DNA substrates on a per-unit basis.

Extrachromosomal DNA

Extrachromosomal DNA (ecDNA) are acentric, circular DNA molecules that amplify oncogenes in >60% of cancer types (Figure 2).¹²⁷ ecDNA form during catastrophic genomic events, such as chromothripsis, which generate DNA fragments that are subsequently circularized.^12,128–132 In contrast to chromosomal DNA, which is evenly distributed during mitosis via centromere-mediated attachment to the spindle, ecDNA lack centromeres and thus segregate randomly.^133–135 The unequal partitioning of ecDNA during cell division leads to differential oncogene copy numbers across daughter cells, driving both intratumoral and intercellular heterogeneity.^{133,134,136–138} This uneven inheritance allows cancer cells to rapidly adapt to environmental pressures, including therapeutic interventions, by altering ecDNA copy number.¹³⁹ During mitosis, ecDNA remain closely associated with chromosomal DNA, thus providing a means for acentric nuclear retention.^133,140,141 Although ecDNA-chromosome tethering during mitosis helps promote nuclear retention of ecDNA, this mechanism is imperfect and ecDNA have frequently been observed in nuclear atypia, such as micronuclei.^128,142,143

Observations of ecDNA entrapment within micronuclei raises the possibility that ecDNA mis-segregation drives cGAS-STING activation. The open chromatin state of ecDNA facilitates enhanced engagement with the transcriptional machinery, resulting in rampant transcription not observed in linear oncogene amplifications.¹⁴⁴ These findings suggest that the high accessibility of ecDNA chromatin could enhance its immunostimulatory potential by reducing nucleosome-based cGAS inhibition. In support of this, tumors containing ecDNA frequently downregulate components of innate immune pathways, including STING1, implying selective pressure to suppress cGAS activation.¹⁴⁵ It is important to distinguish ecDNA from extrachromosomal circular DNA elements (eccDNAs). While ecDNAs are typically large, often megabase-scale, and frequently harbor oncogenes that drive tumor evolution, eccDNAs are much smaller, ranging from hundreds of base pairs to a few kilobases, and generally arise as byproducts of DNA repair, replication stress, or apoptosis. Unlike ecDNA, which are predominantly found in tumor cells and during early neoplastic stages, eccDNA are detected in virtually all cell lines and tissues.¹⁴⁶ Recent studies have shown that eccDNA can possess potent STING-dependent immunostimulant activity, particularly when released extracellularly and sensed by neighboring cells.¹⁴⁷ Further work will be necessary to assess the immunostimulatory potential of both cancer-derived ecDNA and eccDNA.

Cytosolic DNA Export

Genomically unstable tumors rely on DNA repair pathways for survival; however, essential nuclear processes like DNA repair can provoke an innate immune response through the release of nucleic acid fragments to the cytosol (Figure 2). The mechanisms by which immunostimulatory dsDNAs are generated in the nucleus and how they translocate to the cytosol are unclear. Here, we provide a summary of nuclear processes operative in genomically unstable tumors that have been connected to cytosolic DNA release and cGAS-STING activation.

Acquired DNA repair defects are one major source of cytosolic DNA release in cancer. Mismatch repair deficiency results in the accumulation of aberrant DNA repair intermediates that are ultimately released into the cytosol.¹⁴⁸ Analysis of mouse models reveals that mismatch repair-deficient tumors exhibit sensitivity to immune checkpoint blockade and potent T cell priming. Inactivation of cGAS or STING in tumors suppresses T cell infiltration and undermines the efficacy of checkpoint blockade, underscoring the importance of cytosolic DNA detection in mismatch repair-deficient tumor immunogenicity.¹⁴⁹ Excision of UV photoproducts from DNA during nucleotide excision repair has also been connected to cytosolic DNA release.^149,150 Finally, induction of extrachromosomal telomere repeat DNA through the alternative lengthening of telomeres pathway has also been connected to a cGAS-STING-dependent IFN response.¹⁵¹

R-loops, arising from a nascent RNA re-hybridizing to the template DNA strand, are a common consequence of transcription (Figure 2). R-loop formation is a common occurrence in cancers resulting from heightened oncogene transcription that can drive genomic instability by promoting transcription-replication conflicts.¹⁵² The endonucleases XPG and XPF can process R-loops to release RNA:DNA hybrids that trigger cGAS signaling.¹⁵³ Although a direct binding of cGAS to RNA:DNA hybrids has not been demonstrated, in vitro analyses suggest that coincubation of recombinant cGAS and synthetic RNA:DNA hybrids can induce cGAMP generation, albeit at a significantly lower level than dsDNA.¹⁵⁴ Whether this interaction happens in cells, and whether the low-grade cGAMP induction by RNA:DNA hybrids is sufficient to provoke an inflammatory disease phenotype, remain unknown. Also, how cancers evade R-loop-triggered immunity is a mystery, although upregulation of R-loop regulators, such as RNase H, represents one conceivable mechanism.

Generation of ssDNA fragments, e.g., DNA resection during homology-directed repair or replication fork reversal, is a common intermediate step in the cellular response to DNA damage, facilitating both repair pathway engagement and the activation of DNA damage signaling cascades such as ATR-CHK1.¹⁵⁵ In the absence of the dNTPase SAMHD1, MRE11 and RECQ1 mediate the release of ssDNA from stalled replication forks, ultimately leading to a chronic IFN response.¹⁵⁶ Subsequent work revealed that SAMHD1 deficiency leads to accumulation of immunostimulatory dsRNA species, resulting in activation of the MDA5-MAVS signaling axis. Notably, cGAS deletion attenuates this response by reducing basal MDA5 expression, underscoring a key role for the cGAS-STING axis in priming innate immune responses in the context of SAMHD1 deficiency. Similarly, ARID1A loss activates type I IFN signaling through a distinct mechanism involving R-loop–mediated accumulation of cytosolic ssDNA.¹⁵⁷ DNA-damaging chemotherapies used during cancer therapy can also release ssDNA fragments into the cytosol to engage the innate immune response.¹⁵⁸ Despite the evidence for ssDNA generation, how ssDNA activates cGAS remains unclear. Anisotropy analysis confirms that cGAS preferentially binds dsDNA over ssDNA, with a ~17-fold difference in K_d values.⁵⁷ While a sequence-dependent, direct sensing of ssDNA has been described,¹⁵⁹ such effect has not been generalized outside of the context of HIV and relies on a specific conformation of DNA. One possibility is that cGAS is activated by duplex DNA formed through the annealing of complementary ssDNA at repetitive regions. Future work will be necessary to understand how ssDNA release can activate cGAS-STING signaling.

Despite a central role in cGAS-STING activation, mechanisms behind cytosolic release of self-DNA are poorly understood. Due to a limited number of mechanistic studies focused on cytosolic DNA release, it has been widely assumed that DNA fragments passively leak into the cytosol. However, there is evidence that the nuclear transport receptor exportin-1 transports excised RNA:DNA hybrids into the cytosol.¹⁵³ In addition, nuclear factors, such as RPA and RAD51, have been proposed to counteract cytosolic DNA leakage via binding of ssDNA within the nucleus.¹⁶⁰ Future studies will be necessary to elucidate the mechanisms underlying nuclear retention and release of DNA fragments.

Autophagy

Autophagy is a cellular process that delivers proteins and organelles to the lysosome for degradation and recycling.^201–204 Recent advances implicate autophagy as a closely integrated feature of STING signaling (Figure 3). Independently of its scaffolding function for TBK1 and IRF3 engagement, activated STING triggers noncanonical autophagy that bypasses the autophagy initiation machinery, including the ULK1 and PI3K complexes.⁴³ Following TBK1 recruitment and activation at the trans-Golgi network, STING is trafficked to post-Golgi vesicles.²⁰⁵ There, STING’s novel proton channel activity mediates the de-acidification of the vesicular interior, which then recruits V-ATPase and ATG16L1 to initiate a noncanonical LC3 lipidation onto single-layer membranes.^45,46 This process is termed conjugation of ATG8 to single membranes (CASM) and is a defining aspect of noncanonical autophagy. Loss of ATG9A does not abolish LC3 lipidation, indicating that the canonical autophagy machinery is dispensable for STING-triggered autophagy.⁴³ STING-dependent CASM has been proposed to aid in the clearance of cytosolic nucleic acids and pathogens.²⁰⁶ This signaling axis also activates TFEB to upregulate lysosome biogenesis, contributing to endolysosomal homeostasis.^206–208 Separately from CASM, post-Golgi vesicular STING also promotes M1 ubiquitin linearization that triggers NF-κB signaling.²⁰⁹ Thus, activation of autophagy downstream of STING is closely coupled with immune signaling and maintenance of cellular homeostasis, suggesting that autophagy itself may play important roles in antitumor immunity.

Concomitant with noncanonical autophagy initiation, activated STING itself is rapidly degraded,^205,210 effectively forming a negative feedback loop (Figure 3). STING degradation requires V-ATPase activity, consistent with lysosomal degradation.^205,210 Several groups have described an autophagy-independent mode of STING degradation,^43,205 where activated STING is selectively sorted to the endolysosomal compartment via recognition of its phosphorylated C-terminal tail and delivery via clathrin-coated vesicles.^43,163 Yet, STING has been shown to be phosphorylated by the autophagy initiator ULK1 and shortly thereafter degraded,^205,210 as well as selectively delivered to the autophagosome by the autophagy receptor p62.²¹¹ ESCRT-mediated direct delivery to the lysosome has also been described.²¹² Further studies of STING degradation will clarify the relative contributions of these proposed mechanisms and ultimately improve our understanding regarding the dynamics of cGAS-STING signaling termination.

Despite our incomplete understanding, the autophagy / cGAS-STING crosstalk opens up exciting possibilities in the context of cancer. As autophagy promotes cellular homeostasis, cancer cells may rely on STING-dependent autophagy for survival. Blocking the proton channel activity of STING using C53 or similar drugs may represent a novel therapy strategy, as this would selectively inhibit autophagy while leaving the IFN pathway untouched.⁴⁶ This strategy may be effective even in late cancers that have co-opted STING signaling by silencing the IRF3/IFN axis. The exact relationship between STING-triggered autophagy and its oncological significance should be clarified in future studies.