Powering on cGAMP mini factories

Cyclic GMP–AMP (cGAMP) synthase (cGAS) is an essential innate immune sensor. Remarkably, in addition to its role in the early detection of pathogenic DNA molecules, cGAS also monitors cellular health through the sensing of nuclear and mitochondrial DNA aberrantly localised to the cell cytoplasm. This central position of cGAS requires tight molecular controls which are only starting to be understood. In this issue of EMBO Reports, Zhao and colleagues (Zhao et al, 2021) describe a novel mechanism switching on DNA sensing, relying on the formation of primary condensates of cGAS and GTPase‐activating protein‐(SH3 domain)‐binding protein 1 (G3BP1).

cGAS is an essential immune sensor that detects pathogenic or aberrant endogenous DNA in the cytoplasm. A study in this issue reports that activation of DNA sensing depends on the formation of primary condensates of cGAS and G3BP1.

CGAS detection of cytosolic DNA results in the production of cGAMP (Sun et al, ²⁰¹³), which acts as an essential secondary messenger through its interaction with stimulator of interferon genes (STING). This culminates in the production of interferon (IFN)‐β and pro‐inflammatory cytokines to orchestrate antiviral activities including the clearance of infected or damaged cells. Unlike other nucleic acid sensors that are restricted to specific cell subtypes and intracellular localisation or are triggered by pathogen‐associated molecular patterns (PAMPs) specific to viruses and bacteria, cGAS’s capacity to detect endogenous DNA requires mechanisms limiting its aberrant engagement. For example, nucleosomes and barrier‐to‐autointegration factor (BAF) directly prevent cGAS’s activity on nuclear DNA (Decout et al, ²⁰²¹). Moreover, cytosolic DNA is rapidly targeted for degradation by the three‐prime repair exonuclease 1 (TREX1).

In addition to these protective mechanisms, it was recently proposed that the activation of cGAS was conditioned by its capacity to form membrane‐less like organelles, also known as liquid‐like phase separation (LLPS), where the concentration of multiple cGAS proteins, its substrates and DNA constitute cGAMP producing mini factories (Du & Chen, 2018; Xie et al, ²⁰¹⁹). This argument originally stemmed from a series of in vitro studies elegantly demonstrating that incubation of cGAS in the presence of adenosine triphosphate/guanosine triphosphate (ATP/GTP) and DNA rapidly leads to the formation of liquid droplets, dynamically fusing together over time to form larger droplets—the size of the droplet being correlated with the yield of cGAMP produced (Du & Chen, 2018; Xie et al, ²⁰¹⁹). While additional experiments gave support to the formation of intracellular LLPS through the fusion of DNA/cGAS puncta (Du & Chen, 2018), a detailed understanding of the LLPS formation and its impact on cGAS sensing in cells is currently lacking. Nonetheless, growing evidence from pathogen products that can modulate cGAS–LLPS does support a key role for phase condensation in cGAS’s sensing, which may also help protect DNA from TREX1 degradation (Xiao et al, ²⁰²¹).

In this issue of EMBO Reports, Zhao and colleagues discovered another important piece of the cGAS–LLPS puzzle (Zhao et al, ²⁰²¹). The authors demonstrated that in vitro incubation of GTPase‐activating protein‐(SH3 domain)‐binding protein 1 (G3BP1) together with cGAS resulted in liquid droplets similar to those observed when cGAS was incubated with 60 bp DNA. The liquid droplets formed by G3BP1–cGAS increased cGAMP synthesis in the presence of DNA, which was most noticeable when low levels of DNA were used. This indicated that G3BP1 could raise the threshold of sensitivity of cGAS to DNA. Unlike DNA‐induced cGAS–LLPS, the G3BP1–cGAS liquid droplets did not fuse with each other and represented a more static interaction. This indicated that G3BP1–cGAS condensation rather primed cGAS sensing by promoting primary protein condensation. Critically, Zhao and colleagues demonstrated that upon DNA addition to such G3BP1–cGAS condensates, the DNA quickly displaced G3BP1 to condense with cGAS—something which was not seen with RNA molecules (Zhao et al, ²⁰²¹). Zinc ions also induced cGAS primary condensates in the absence of DNA, which acted cooperatively with G3BP1 in promoting LLPS response to DNA.

Perhaps, one of the most important aspects of this work is that it cements further the biological relevance of LLPS for cGAS sensing of cytosolic DNA. On one hand, loss of G3BP1 expression markedly decreased intracellular cGAS aggregates and suppressed DNA‐induced puncta formation along with type‐I IFN production—aligning with previous studies (Liu et al, ²⁰¹⁹; Zhao et al, ²⁰²¹). On the other, pharmacological inhibition of the G3BP1–cGAS interaction with epigallocatechin gallate (EGCG) directly impacted DNA‐induced cGAS–LLPS in vitro, which was matched by a reduced DNA‐dependent cGAS activation in cells in a G3BP1‐dependent manner.

These results collectively indicate a three‐step model for cGAS intracellular activation (Fig 1). In steady state, cytoplasmic cGAS and G3BP1 interact to form primary condensates, also potentiated by zinc ions. When DNA larger than 45 bp or ladder‐like DNA structures accumulate in the cytoplasm, the DNA actively displaces G3BP1 in the cGAS–G3BP1 condensates to form LLPS with cGAS/zinc/ATP/GTP; this then promotes cGAMP synthesis. Critically, as established in extensive prior studies, G3BP1 deficiency strongly impaired cGAS‐driven type‐I IFN response to transfected DNA (Liu et al, ²⁰¹⁹). Since low levels of type‐I IFNs were still produced and residual cGAS condensates were still observed in G3BP1‐deficient cells, this model indicates that G3BP1 “powers on” the formation and activation of cGAS–LLPS, which is otherwise much less efficient. Given that G3BP1 is also involved in the formation of RNA‐stress granules with retinoic acid‐inducible gene 1 (RIG‐I) (Xiao et al, ²⁰²¹), it will be interesting to define whether its recruitment to facilitate the formation of cGAS–LLPS is competed against by RNA‐stress granules, thereby flicking off the cGAS switch upon RNA virus infections. Another point to consider in future studies is the intracellular stoichiometry of the cGAS–G3BP1 duo, the tissue variation and inducibility of which have the capacity to directly impact cGAS function.

Figure 1. GTPase‐activating protein‐(SH3 domain)‐binding protein 1 (G3BP1) enhances the formation of cyclic GMP–AMP (cGAMP) synthase–DNA–liquid‐like phase separation (cGAS–DNA–LLPS).

(1) cGAS and G3BP1 spontaneously form primary condensates in the cell cytoplasm in the absence of DNA. (2) Upon accumulation of large cytoplasmic DNA, for instance from a DNA virus, DNA molecules displace G3BP1 moieties from the cGAS–G3BP1 condensates. (3) The resulting cGAS–DNA–LLPS synthesise large levels of cGAMP and fuse with each other to form large cytoplasmic puncta. The resulting cyclic GMP–AMP (cGAMP) can then bind and activate endoplasmic reticulum–stimulator of interferon genes (ER–STING) to induce downstream signalling. Zinc potentiates further the activity of G3BP1 on cGAS sensing.

As illustrated in these studies with G3BP1 and prior reports with 25 bp DNA and double‐stranded RNA (dsRNA), the formation of primary liquid condensates of cGAS does not guarantee its activation and cGAMP production (Du & Chen, 2018; Zhao et al, ²⁰²¹). The findings obtained with G3BP1, however, suggest that low cytosolic levels of short DNA and dsRNA that can interact with cGAS may in fact facilitate sensing of longer DNA through the formation of primary condensates that are therefore primed for DNA sensing. While warranting further investigations, it is noteworthy that two independent recent studies supported a potentiation of cGAS sensing with low doses of 2’O‐methyl‐DNA oligonucleotides or transfer RNAs (tRNAs) (preprint: Chen et al, ²⁰²⁰; Valentin et al, ²⁰²¹). With a rapidly growing number of therapeutic DNA/RNA oligonucleotides getting approved as treatments, it will be important to consider the effect these can have on cGAS primary condensates and priming of its aberrant engagement. Finally, pharmacological targeting of G3BP1 or the formation of its complex with cGAS, as illustrated here with EGCG, may represent novel opportunities to control aberrant cGAS engagement in diseases such as type‐I interferonopathies driven by TREX1 deficiency (Cai et al, ²⁰²¹).

Conflict of interest

Michael Gantier is a named inventor of a provisional patent relating to cGAS inhibitors.

EMBO reports (2022) 23: e54231.