The sensing of foreign nucleic acids is a major strategy by which the innate immune system senses viral infection. Distinct receptor systems for viral RNA and viral DNA trigger antiviral responses, including the production of type I interferons. In the cytosol, members of the family of RIG-I-like helicases detect viral RNA, whereas the presence of cytosolic double-stranded DNA is sensed by cGAS. To induce the expression of type I interferons, both sensing systems rely on the coordinated activation of transcription factors, including NF-κB and IRF3. However, the cGAS pathway is unusual in that its most upstream components are not linked through classic protein-protein interactions. Instead, cGAS is an enzyme that, after binding DNA, catalyzes the production of a second-messenger molecule, 2′3′-cGAMP, which engages the endoplasmic reticulum adaptor STING1. The cGAS-STING cascade is not restricted to signaling in a cell-intrinsic fashion but instead, by means of incorporation into viral particles or through transfer via gap junctions, also functions in a paracrine and systemic manner2,3,4. Thus, tiny amounts of DNA can be sufficient to induce strong systemic immune responses. Given this amplificatory nature of the cGAS–STING–type I interferon pathway, it seems reasonable that regulatory mechanisms would exist to limit excessive activation of this signaling cascade. In this issue of Nature Immunology, a report by Fan and colleagues proposes just such a mechanism by which cGAS is regulated5.

An interesting outcome of research on the role of DNA sensing in controlling pathogen infection has been the appreciation that in addition to viral (or pathogen-derived) DNA, host DNA can also activate cGAS and upregulate cytokine production. Indeed, once naturally occurring turnover of DNA is dysregulated and self DNA accumulates in the cytosol, it can contribute to the pathogenesis of autoimmune diseases6. Although such studies have emphasized the important role of DNA-degrading enzymes in the prevention of chronic inflammatory responses, little is known about the regulatory mechanisms that operate directly on cGAS. Controlling the expression of cGAS appears to be one mechanism at play, and indeed cGAS expression undergoes substantial induction after viral infection or stimulation with type I interferons. Cytosolic DNA has also been found to trigger autophagy, and the recruitment of cGAS into autophagosomes and its consequent degradation might constitute another way of limiting its activity7. Another means of controlling protein function in a rapid and reversible manner is post-translational modification (PTM). Phosphorylation is a widely applied PTM that controls diverse signal-transduction pathways, and a published study has highlighted the phosphorylation of cGAS by the kinase AKT8. Fan and colleagues now propose an alternative mechanism for the regulation of DNA-triggered antiviral responses by identifying glutamylation as a mechanism of controlling cGAS activity5.

Glutamylation is a reversible PTM in which glutamate side chains of variable length are formed at specific sites on a target protein. The presence and length of the glutamate side chains are regulated in a dynamic fashion by the tubulin-tyrosine ligase–like (TTLL) family of enzymes, which catalyze glutamylation, and by members of the cytosolic carboxypeptidase (CCP) family, which remove the modification. Although it was initially discovered on tubulins, it is now clear that glutamylation is a regulator of diverse cellular processes; however, its involvement in innate immunological signaling pathways has not yet been explored.

On the basis of published work demonstrating that glutamylation regulates megakaryopoiesis9, Fan and colleagues investigate its possible involvement in signaling via the innate immune system5. In an attempt to systematically explore this, the authors start with mice that have individual deletion of each of the six CCP-encoding genes and infect these mice with the DNA virus herpes simplex virus type 1 or vaccinia virus. They observe that absence of CCP5 or CCP6 results in enhanced susceptibility to infection with either of the two viruses, which is paralleled by decreased production of type I interferons. These findings are also evident on a cellular level, with compromised production of type I interferons in bone marrow–derived macrophages or dendritic cells deficient in either CCP5 or CCP6 that are either challenged with DNA virus or stimulated with synthetic double-stranded DNA. In vivo and in vitro responses to RNA viruses or to the synthetic RNA duplex poly(I:C) are not affected by the absence of either CCP5 or CCP6, which indicates that both enzymes are specifically involved in regulating DNA-sensing pathways but not in regulating RNA-sensing pathways.

Fan and colleagues next focus their attention on identifying the deglutamylation substrate(s) of CCP5 and CCP6. A combination of immunoprecipitation with antibody to glutamate side chains and mass spectrometry results in the identification of cGAS as a glutamylated target of CCP5. In a series of experiments, the authors next characterize two glutamylation sites within cGAS (Glu272 and Glu302 in mice, and Glu286 and Glu372 in humans) and assign specific roles for deglutamylation of these residues by CCP5 and CCP6, with the latter being responsible for removal of polyglutamate chains from Glu272 (human Glu286) and the former being in charge of hydrolyzing the monoglutamate chain at Glu302 (human Glu372).

But which enzymes are responsible for the glutamylation of cGAS? To address this, Fan and colleagues analyze nine members of the TTLL family, well known for their ability to modify and mark proteins by glutamylation, assessing their interaction with cGAS. This approach yields the monoglutamylase TTLL4 and the polyglutamylase TTLL6, but not other TTLL enzymes, as cGAS-interaction partners. The authors show that TTLL4 and TTLL6 have non-redundant roles in the glutamylation of cGAS, with TTLL4 mediating the monoglutamylation of cGAS at Glu302, and TTLL6 adding polyglutamylation chains on Glu272. Thus, it seems that specific TTLL enzymes and CCPs work in opposition to each other in modifying the glutamylation status of cGAS.

The authors next seek to determine how glutamylation affects the activity of cGAS by focusing on its two core functions: DNA binding and catalysis of cGAMP. Through complementary experiments, they conclude that the two TTLL-CCP modules—TTLL4–Glu302-CCP5 and TTLL6–Glu272-CCP6—regulate both functions of cGAS independently of each other (Fig. 1). Thus, polyglutamylation at Glu272 blocks the ability of cGAS to bind to double-stranded DNA, whereas monoglutamylation at Glu302 interferes with its catalytic function. Accordingly, preventing polyglutamylation at Glu272 (via mutant cGAS with the E272A substitution or knockout of TTLL6) increases the binding ability of cGAS, whereas interfering with the monoglutamylation of Glu302 (via mutant cGAS with the E302A substitution or knockout of TTLL4) augments the production of cGAMP. Impairment of glutamylation of cGAS by knockout of TTLL4 or TTLL6 'translates' into greater production of type I interferons and more efficient blocks an the replication of herpes simplex virus type 1 relative to that of wild-type mice, and this effect is even greater in mice with double knockout of both TTLL4 and TTLL6. Conversely, mice with double knockout of both CCP5 and CCP6 display lower production of type I interferons and diminished resistance to infection with a DNA virus relative to that of wild-type mice.

Figure 1: Glutamylation regulates the activity of cGAS.
figure 1

Katie Vicari/Nature Publishing Group

The carboxypeptidases CCP5 and CCP6 promote cytosolic DNA sensing by removing glutamate side chains from cGAS, thereby enabling DNA binding and cGAMP synthesis. This process is antagonized by the enzymes TTLL4 and TTLL6, which mediate mono- and polyglutmatylation of cGAS. Upon infection with herpes simplex virus, deglutamylated cGAS binds to cytosolic DNA and produces cGAMP, which leads to antiviral gene expression via STING and IRF3. ER, endoplasmic reticulum.

On the basis of those findings, Fan and colleagues propose that the glutamylation status of cGAS is a critical determinant in controlling its functional capability and in balancing the host response to DNA5. The catalog of post-translational alterations that affect immunological signaling pathways is a complex one. However, the idea of 'fine-tuning' the cGAS-STING pathway in a rapid and reversible manner is of special attractiveness, given that by itself cGAS cannot reliably distinguish between self (or harmless) triggers or situations and those that are non-self (or dangerous). Enzymes that control the PTM could thus function as molecular 'switches' that integrate important environmental cues and alter the functional ability of cGAS, adjusted to the needs and requirements of the host. Such a system would allow rapid recruitment of 'active' cGAS in the setting of pathogen infection. On the other hand, it could transiently turn off the DNA-sensing pathway during mitosis or apoptosis, scenarios in which cytosolic DNA should be made 'invisible' to cGAS.

These findings raise the next question: what controls the enzymes that control cGAS? Fan and colleagues offer a partially resolved solution5. They observe that in resting cells, specific TTLL enzymes and CCPs are in competition to modify cGAS activity, with the result being the co-existence of both glutamylated (inactive) cGAS and deglutamylated (active) cGAS. Stimulation of cells leads to downregulation of TTLL expression and thereby passively generates enrichment for 'active versions' of cGAS. Thus, it seems that the ratio between TTLL enzymes and CCPs dynamically controls the pool of DNA-binding, competent and enzymatically active cGAS. How the expression and functionality of these enzymes is regulated remains an open issue. Along these lines, given the substantial effect of glutamylation on the outcome of infection with a DNA virus, it is very likely that viruses have developed countermeasures against this regulatory step. Thus, one could speculate about the existence of virus-encoded TTLL enzymes or factors that intercept with CCPs. The discovery of this role of cGAS marks an important breakthrough in the understanding of DNA sensing. Although the cGAS product cGAMP holds great 'translational' potential as an adjuvant, it is also a promising target for pharmacological intervention in autoinflammatory syndromes. Although the relevance of glutamylation of cGAS still needs to be validated in other experimental settings, the study by Fan and colleagues5 opens the door to more research on non-canonical PTMs. Thus, there might be a plethora of non-canonical PTMs that could be equally important in the regulating the sensing of nucleic acids. In the future, it will be important to determine whether additional members of the DNA-sensing cascade are substrates of these PTMs and to also find ways of modulating these activities for creating novel therapeutic opportunities.