Tight nuclear tethering of cGAS is essential for preventing autoreactivity

Hannah E Volkman; Stephanie Cambier; Elizabeth E Gray; Daniel B Stetson

doi:10.7554/eLife.47491

. 2019 Dec 6;8:e47491. doi: 10.7554/eLife.47491

Tight nuclear tethering of cGAS is essential for preventing autoreactivity

Hannah E Volkman ^1,^†, Stephanie Cambier ^1,^†, Elizabeth E Gray ^1,^‡, Daniel B Stetson ^1,^✉

Editors: Tadatsugu Taniguchi², Tadatsugu Taniguchi³

PMCID: PMC6927687 PMID: 31808743

Abstract

cGAS is an intracellular innate immune sensor that detects double-stranded DNA. The presence of billions of base pairs of genomic DNA in all nucleated cells raises the question of how cGAS is not constitutively activated. A widely accepted explanation for this is the sequestration of cGAS in the cytosol, which is thought to prevent cGAS from accessing nuclear DNA. Here, we demonstrate that endogenous cGAS is predominantly a nuclear protein, regardless of cell cycle phase or cGAS activation status. We show that nuclear cGAS is tethered tightly by a salt-resistant interaction. This tight tethering is independent of the domains required for cGAS activation, and it requires intact nuclear chromatin. We identify the evolutionarily conserved tethering surface on cGAS and we show that mutation of single amino acids within this surface renders cGAS massively and constitutively active against self-DNA. Thus, tight nuclear tethering maintains the resting state of cGAS and prevents autoreactivity.

Research organism: Human, Mouse

Introduction

The cGAS-STING DNA sensing pathway has emerged as a key innate immune response that is important for antiviral immunity (Goubau et al., 2013), contributes to specific autoimmune diseases (Crowl et al., 2017), and mediates important aspects of antitumor immunity (Li and Chen, 2018). cGAS binds to double-stranded DNA and catalyzes the formation of cyclic GMP-AMP (Sun et al., 2013; Wu et al., 2013), a diffusible cyclic dinucleotide that activates the endoplasmic adaptor protein STING (Ishikawa et al., 2009). Activated STING then serves as a platform for the inducible recruitment of the TBK1 kinase, which phosphorylates and activates the transcription factor IRF3, leading to the induction of the type I interferon mediated antiviral response (Liu et al., 2015).

cGAS is important for ‘cytosolic DNA sensing,’ a term that was first proposed in 2006, years before the discovery of STING and cGAS as the essential adaptor and unique sensor of this pathway (Stetson and Medzhitov, 2006). At the time, a key conundrum was how a sequence-independent DNA sensing pathway that was activated by the sugar-phosphate backbone of DNA could avoid constant autoreactivity against genomic DNA that is present in all nucleated cells. We proposed the possibility that the sensor would be sequestered in the cytosol, separated by the nuclear envelope from genomic DNA, and that the inappropriate appearance of DNA in the cytosol would enable detection of foreign DNA while maintaining ‘ignorance’ to self DNA (Stetson and Medzhitov, 2006). The discovery of cGAS and cGAMP, and subsequent structural studies of cGAS binding to DNA, provided an elegant explanation for the sequence independence of the response, the requirement for double-stranded DNA as a ligand, the contribution of the deoxyribose sugar-phosphate backbone of DNA to detection, and the definitive link between DNA sensing and STING (Civril et al., 2013; Li et al., 2013; Sun et al., 2013; Wu et al., 2013). However, the precise location of cGAS prior to its activation has remained largely unexplored, in part because of a lack of tools to track endogenous cGAS. Cytosolic DNA sensing has persisted as the mechanistic framework that guides the field.

There are a number of important problems with the model of cytosolic DNA sensing. First, nearly all DNA viruses (with the exception of the poxviruses) replicate their DNA exclusively in the nucleus. Studies of cGAS-deficient mouse and human cells have revealed that cGAS is important for the IFN-mediated antiviral response to these nuclear-replicating viruses, including herpesviruses (Ma et al., 2015; Wu et al., 2015). Moreover, retroviruses and lentiviruses are detected by cGAS (Gao et al., 2013; Lahaye et al., 2013; Rasaiyaah et al., 2013), but they shield their DNA within a capsid during reverse transcription in the cytosol, releasing this DNA for integration into the genome upon translocation into the nucleus. To fit the concept of cytosolic DNA sensing, current models envision that such viruses ‘leak’ DNA into the cytosol during cellular entry or during exit. Second, as noted in the original description of cytosolic DNA sensing (Stetson and Medzhitov, 2006), cell division results in the breakdown of the nuclear envelope and the mixing of cytosolic and nuclear contents, which challenges a simple model of cytosolic sequestration as the basis for self/non-self discrimination by cGAS. Indeed, recent studies have demonstrated that cGAS can be found associated with mitotic chromosomes (Yang et al., 2017). This association is thought to be mediated by the generic DNA binding properties of cGAS, and it has been proposed that upon resolution of cell division and reformation of the nuclear envelope, cGAS is redistributed to the cytosol (Yang et al., 2017).

Here, we use confocal microscopy and biochemical characterization to determine the resting localization of endogenous cGAS prior to activation. We unexpectedly find that the vast majority of cGAS is in the nucleus, regardless of whether cells are rapidly dividing or post-mitotic. Moreover, we demonstrate that cGAS is tethered tightly in the nucleus by a salt-resistant interaction that rivals that of histones in its strength. Finally, we identify the tethering surface on cGAS and show that tight nuclear tethering prevents cGAS autoreactivity against self-DNA.

Results

Endogenous cGAS is predominantly a nuclear protein

We screened numerous commercially available antibodies to human cGAS for their ability to identify endogenous cGAS unambiguously and specifically using immunofluorescence microscopy. We chose to image HeLa cells, which express endogenous cGAS that is inactive in resting cells and potently activated to produce cGAMP upon transfection of calf thymus DNA (Figure 1—figure supplement 1A). Despite this potent activation of cGAS and production of cGAMP after DNA transfection, HeLa cells fail to activate the type I interferon response because the E7 oncoprotein of human papillomavirus 18 blocks STING-dependent signaling (Lau et al., 2015). To test for specificity of staining, we used lentiCRISPR to generate clonal lines of cGAS-deficient HeLa cells (Figure 1—figure supplement 1B; (Gray et al., 2016). We found that the D1D3G rabbit monoclonal antibody that detects an epitope in the N terminus of human cGAS was suitable for analysis of endogenous cGAS by microscopy. Unexpectedly, endogenous cGAS was localized almost exclusively in the nuclei of all HeLa cells, with little cytosolic staining (Figure 1A, Figure 1—videos 1 and 2). Identically prepared cGAS-deficient HeLa cells had no detectable immunostaining, confirming the specificity of this antibody for endogenous cGAS (Figure 1A). We noted three additional reproducible patterns of cGAS localization in addition to the uniform nuclear staining. First, as observed previously (Yang et al., 2017), we found that cGAS was associated with condensed mitotic chromatin (Figure 1B). Second, we found cGAS in rare, spontaneous, DAPI-positive, micronucleus-like extranuclear structures (Figure 1B). Whereas cGAS localization to micronuclei has been reported recently in a number of studies that primarily visualized overexpressed cGAS (Bartsch et al., 2017; Dou et al., 2017; Glück et al., 2017; Harding et al., 2017; Mackenzie et al., 2017; Yang et al., 2017), we found that all cells with such structures also had extensive endogenous cGAS staining in the main nucleus (Figure 1B). Third, we found endogenous cGAS localized to ‘chromatin bridges’ between adjacent cells (Figure 1B), the origins of which are thought to involve chromosome fusions and incomplete segregation of DNA between daughter cells during mitosis (Maciejowski et al., 2015).

Figure 1. — (A) Clonal lines of HeLa cells were generated using lentiCRISPR encoding either a non-targeting H1 control gRNA (top row) or a cGAS-targeted gRNA (cGAS KO). Cells were fixed with methanol, stained with antibodies to human cGAS and beta-tubulin, counter-stained with DAPI, and visualized by confocal microscopy. (B) We noted three reproducible patterns of cGAS localization in addition to the nucleus: condensed mitotic chromatin (top row), structures resembling micronuclei (middle row), and tendril-like bridges between cells. (C) Mouse *Cgas*+/+ and *Cgas*-/- primary bone marrow-derived macrophages were stained using a mouse antibody to cGAS and processed as in (A).

Figure 1—figure supplement 1. — (A) Clonal lines of HeLa cells were generated using lentiCRISPR encoding either a non-targeting H1 control gRNA (top row) or a cGAS-targeted gRNA (cGAS KO). Cells were fixed with methanol, stained with antibodies to human cGAS and beta-tubulin, counter-stained with DAPI, and visualized by confocal microscopy. (B) We noted three reproducible patterns of cGAS localization in addition to the nucleus: condensed mitotic chromatin (top row), structures resembling micronuclei (middle row), and tendril-like bridges between cells. (C) Mouse *Cgas*+/+ and *Cgas*-/- primary bone marrow-derived macrophages were stained using a mouse antibody to cGAS and processed as in (A).

To extend our findings to primary cells of another species, we searched for antibodies that could identify endogenous mouse cGAS by microscopy. Using primary bone marrow-derived macrophages (BMMs) from wild-type and cGAS-deficient mice and a mouse-specific cGAS antibody, we found nearly exclusive localization of mouse cGAS to the nucleus (Figure 1C). However, even with optimization of blocking conditions and antibody dilutions, we noted that cGAS-deficient mouse macrophages displayed a pattern of nuclear staining that was distinct in its distribution and less abundant than the cGAS staining of wild-type cells (Figure 1C). Despite the imperfect background fluorescence, this was the most sensitive and specific cGAS staining we could identify among the antibodies that we tested.

In our microscopy experiments, we used methanol fixation/extraction because we found that the epitope of the antibody to human cGAS was sensitive to paraformaldehyde (PFA) fixation, which reduced the signal of the specific staining and increased background staining. However, it has been suggested that methanol fixation might also extract a membrane-bound pool of cytosolic cGAS (Barnett et al., 2019), leading to an overestimation of the amount of nuclear cGAS in our images. We found that PFA fixation followed by 0.1% Triton X-100 permeabilization, which would preserve such a pool of cGAS, resulted in a pattern of endogenous cGAS staining that was almost exclusively nuclear and nearly identical to the staining observed in methanol-fixed cells (Figure 1—figure supplement 1C). Finally, to rule out any role for fixation in our interpretation of cGAS localization, we performed Amnis imaging flow cytometry on live, unfixed cGAS KO HeLa cells stably expressing a GFP-cGAS fusion protein. Analysis of thousands of individual cells revealed that the great majority of cGAS colocalized with a fluorescent DNA-intercalating dye that marked the nucleus (Figure 1—figure supplement 1D). Together, these data reveal that, contrary to expectation, cGAS is primarily a nuclear protein in both human and mouse cells.

cGAS is tethered tightly in the nucleus

We sought to reconcile the nuclear localization of endogenous cGAS in Figure 1 with the widely accepted notion that cGAS is primarily a cytosolic protein. To track endogenous cGAS localization thoroughly, we modified a protocol for salt-based elution of histones from purified nuclei (Shechter et al., 2007). We prepared extracts separating cytosol from nuclei using a solution containing 0.2% NP-40 detergent followed by low speed centrifugation. After washing the pellets with detergent-free lysis buffer, we lysed the nuclei in a solution of 3 mM EDTA and 0.3 mM EGTA in water. Following this zero salt nuclear lysis, the pellets remaining after centrifugation were treated with stepwise increases of NaCl in a buffer containing 50 mM Tris-HCl pH 8.0 and 0.05% NP-40. We tracked endogenous cGAS throughout this sequential extraction and elution protocol using six different cell lines from humans and mice, sampling primary cells, immortalized cells, and tumor cells. These included cells that were actively dividing (HeLa, SiHa, mouse fibroblasts, human fibroblasts) as well as primary mouse macrophages that are largely non-dividing (Luo et al., 2005). We monitored the specificity of the extractions using the cytosolic protein tubulin, the nuclear zero/low salt elution-enriched protein LSD1, and nuclear pellet-localized histones H2B and H3. In all of these cells, we found that the vast majority of cGAS was not only in the nuclear fractions, but it was remarkably resistant to salt-based elution (Figure 2A). In most of these cells, a NaCl concentration of 0.75 M or higher was required to solubilize the majority of cGAS, similar to the amount of salt required to initiate the liberation of histones (Figure 2A). Importantly, the salt elutions reflect sequential treatments of the same nuclear extracts such that the sum of all the cGAS signals in these fractions can be compared to the cytosolic extracts in order to determine the relative amounts of cGAS in the cytosol and nucleus. These comparisons, calculated by densitometry analysis (Figure 2B), corroborate the microscopy studies in Figure 1 and reveal that the great majority of endogenous cGAS is in the nucleus, not in the cytosol. Moreover, our findings demonstrate that the conventional nuclear washes of ~420 mM NaCl that are typically used to isolate nuclear proteins are insufficient to liberate cGAS from the nucleus (Sun et al., 2013). Such tight tethering of cGAS in the nucleus cannot be explained by its low intrinsic affinity for DNA, the dissociation constant of which has been estimated at 1–2 μM (Civril et al., 2013; Li et al., 2013).

Figure 2. — (A) Mouse and human cell lines were separated into cytosolic and nuclear fractions, followed by sequential stepwise elutions of nuclear pellets with the indicated concentrations of NaCl. cGAS and the indicated control proteins (shown for HeLa cells) were monitored throughout the elution by western blot. (B) Densitometry measurements quantitating the relative amounts of endogenous cGAS protein in the cytosol, the 0 salt nuclear lysis, and the cumulative nuclear pellet.

cGAS is nuclear regardless of cell cycle phase or activation state

One potential explanation for the nuclear localization of cGAS, particularly in dividing cells like HeLa cells, is that this reflects the previously observed association of cGAS with condensed mitotic chromatin. Thus, recently divided cells might still retain cGAS in the nucleus before its redistribution to the cytosol via mechanisms that remain unexplained. Importantly, the fact that almost 95% of cGAS is nuclear in largely post-mitotic, differentiated primary mouse macrophages argues against this possibility (Figures 1C and 2). We further tested this by tracking the localization of endogenous cGAS throughout controlled cell cycles in HeLa cells. Based on our observation that nuclear cGAS is resistant to salt elution up to 0.75 M NaCl (Figure 2), we used a widely available commercial extraction kit to separate cytosol from the nuclear proteins that elute in ~420 mM NaCl (nuclear supernatant, NS), and we additionally examined the residual pellets to visualize the entire pool of tightly tethered cGAS (nuclear pellet, NP). We used double-thymidine block to synchronize HeLa cells at the G1/S boundary (Bootsma et al., 1964), followed by release that resulted in a uniform progression through a single cell cycle. At 4 and 8 hr post release, PI staining confirmed uniform populations of cells in S and G2/M phases, respectively (Figure 3A). By 24 hr, the cells had become asynchronous again (Figure 3A). At all of these time points, we found that the localization of the majority of cGAS to the nuclear pellet did not change (Figure 3B). Thus, endogenous cGAS is a tightly tethered nuclear protein, regardless of cell cycle phase.

Figure 3. — (A) HeLa cells were arrested at the G1/S border using double thymidine (DT) block, followed by release and harvest at the indicated time points for measurement of DNA content. (B) Cells from (A) were fractionated and cGAS localization was determined by western blot. (C) HeLa cells were transfected with Lipofectamine alone (Lipo) or with CT-DNA for 4 hr, followed by extraction, salt elution, and western blot for endogenous cGAS.

Next, we asked whether cGAS localization is dependent on its activation state. We transfected HeLa cells with calf thymus DNA and harvested them four hours later, at a time when they were making large amounts of cGAMP (Figure 1—figure supplement 1A). We performed sequential extractions and salt elutions, comparing cGAS distribution in control and stimulated cells. We did not observe any concerted relocalization of cGAS into the cytosol, despite its robust activation at this time point (Figure 3C). These data demonstrate that activation of cGAS by foreign DNA does not result in a dramatic redistribution to the cytosol.

cGAS nuclear tethering and cGAS activation are governed by separate mechanisms

We next determined the domains of cGAS that contribute to its tight tethering in the nucleus. The core of human cGAS is comprised of a bilobed nucleotidyltransferase (NTase) structure bridged by an alpha-helical spine (Figure 4A) (Civril et al., 2013; Li et al., 2013). In addition, the N-terminal ~150 residues of cGAS form an unstructured domain that is positively charged and refractory to crystallization. Interestingly, this N terminus of cGAS was recently demonstrated to be essential for its activation by DNA through a process of phase condensation that assembles cGAS on long double-stranded DNA (Du and Chen, 2018). We reconstituted cGAS-deficient HeLa cells with GFP-cGAS fusions of full-length human cGAS and several truncation mutants corresponding to the structural domains of cGAS. To do this, we cloned a GFP-cGAS expression construct into a doxycycline-regulated lentiviral vector that enabled transduction of cGAS-deficient HeLa cells followed by selection for these transduced cells in the absence of cGAS expression. Induction of GFP-cGAS expression with doxycycline (Dox) allowed us to examine its localization in the absence of exogenous DNA stimulation, unlike standard transient transfections in which the plasmid DNA encoding cGAS also serves as a potent activating ligand. Our panel of truncation mutants revealed a number of important features of its nuclear tethering. First, we found that the GFP-cGAS (161-522) truncation mutant lacking the N terminus remained tethered in the nuclear pellet, and that the isolated N terminus of cGAS (1-161) localized to the cytosol and nuclear supernatant, with very little signal in the nuclear pellet (Figure 4B). Second, we found that removal of either the alpha-helical spine (161-213) or the C-terminal lobe of cGAS (383-522) resulted in a protein that was mislocalized and, in the case of the 213–522 mutant, also unstable (Figure 4B). Thus, the intact core of cGAS is required for its nuclear tethering.

Figure 4. — (A) Structure of human cGAS with domains colorized. (B) TERT-immortalized human foreskin fibroblasts were reconstituted with the indicated Dox-inducible GFP-cGAS lentivirus constructs, treated with 0.1 μg/ml Dox for 24 hr, and then separated into cytosolic (Cyt), nuclear supernatant (NS), and nuclear pellet (NP) fractions. FL: Full-Length. (C) cGAS-deficient HeLa cells reconstituted with the indicated GFP-cGAS constructs were induced for 24 hr with three doses of Dox. Whole cell lysates that recover all cGAS were prepared and blotted with anti-GFP antibody. (D) Cells from (C) were transfected with CT-DNA for four hours, followed by cGAMP measurement in lysates by modified ELISA. (E) Cells described in C-D were treated with 0.1 μg/ml Dox for 24 hr to induce GFP-cGAS expression, then harvested and used for sequential fractionation and salt elution as in Figure 2. (F) Dox-inducible, full-length mouse cGAS constructs were introduced into hTERT-immortalized human fibroblasts. Cells were treated with 0.1 μg/ml Dox for 24 hr followed by stimulation for 4 hr and measurement of cGAMP in cell lysates. (G) Unstimulated cells from (F) were fractionated and blotted for cGAS.

Figure 4—figure supplement 1. — (A) Structure of human cGAS with domains colorized. (B) TERT-immortalized human foreskin fibroblasts were reconstituted with the indicated Dox-inducible GFP-cGAS lentivirus constructs, treated with 0.1 μg/ml Dox for 24 hr, and then separated into cytosolic (Cyt), nuclear supernatant (NS), and nuclear pellet (NP) fractions. FL: Full-Length. (C) cGAS-deficient HeLa cells reconstituted with the indicated GFP-cGAS constructs were induced for 24 hr with three doses of Dox. Whole cell lysates that recover all cGAS were prepared and blotted with anti-GFP antibody. (D) Cells from (C) were transfected with CT-DNA for four hours, followed by cGAMP measurement in lysates by modified ELISA. (E) Cells described in C-D were treated with 0.1 μg/ml Dox for 24 hr to induce GFP-cGAS expression, then harvested and used for sequential fractionation and salt elution as in Figure 2. (F) Dox-inducible, full-length mouse cGAS constructs were introduced into hTERT-immortalized human fibroblasts. Cells were treated with 0.1 μg/ml Dox for 24 hr followed by stimulation for 4 hr and measurement of cGAMP in cell lysates. (G) Unstimulated cells from (F) were fractionated and blotted for cGAS.

We compared the full-length and N-terminal deletion mutant of cGAS in more detail. We tested a 100-fold range of dox concentrations that induced varying levels of the GFP-cGAS fusion constructs, from robust to nearly undetectable (Figure 4C). Consistent with the recent definition of the requirement of the N terminus for cGAS condensation onto DNA (Du and Chen, 2018), we found that the mutant lacking the N terminus of cGAS was severely compromised for DNA-activated cGAMP production at all levels of expression when compared to full-length cGAS (Figure 4D). However, sequential salt elution of the two forms of cGAS revealed nearly identical distribution and similarly tight tethering in the nucleus (Figure 4E).

We analyzed six additional mutants in the context of full-length murine cGAS to further explore the relationships between nuclear localization, nuclear tethering, DNA binding, DNA-mediated dimerization/oligomerization, and catalytic activity. The K335E mutant is severely defective for DNA binding and DNA-dependent activation (Li et al., 2013). The K395M/K399M mutation corresponds to the K407A/K411A mutation in the DNA-binding platform of human cGAS that results in reduced DNA binding and defective DNA-mediated activation (Civril et al., 2013). The ‘zinc thumb-less’ mutant lacks amino acids 378–393 in mouse cGAS (390–405 in human), which form a protrusion that interacts with double-stranded DNA and is essential for cGAS activation (Civril et al., 2013). The K382A mutation results in decreased DNA binding and defective DNA-mediated dimerization, whereas the E386A mutation binds to DNA but fails to dimerize (Li et al., 2013). Finally, the E211A/D213A mutant disrupts cGAS catalytic activity (Li et al., 2013). We cloned each of these cGAS constructs into the doxycycline-controlled lentiviral vector, introduced them into TERT-immortalized human fibroblasts, and then induced expression with doxycycline. We confirmed prior studies demonstrating that all six mutants are severely impaired for DNA-activated cGAMP production (Figure 4F; Civril et al., 2013; Li et al., 2013). Importantly, all six of these mutants remained predominantly nuclear and tethered (Figure 4G). Together, our data reveal two important points about the requirements for cGAS localization versus its activation. First, the nuclear tethering of cGAS can be uncoupled from robust DNA binding, from DNA-activated dimerization, from DNA-activated condensation, and from catalytic activity, revealing separate mechanistic processes that govern the resting and activated states of cGAS. Second, it has been argued that our conditions of cGAS extraction and salt elution might result in the unnatural oligomerization and condensation of cytosolic cGAS onto DNA that might be liberated during the extraction, which could then cause such condensed cGAS to co-sediment with nuclei during the low speed centrifugation step (Barnett et al., 2019). Our findings argue against this possibility because the domains and specific amino acids that are essential for DNA-induced condensation are all dispensable for its nuclear localization and tethering.

We tested two additional potential cGAS tethering mechanisms. The nuclear protein NONO was recently found to bind to HIV capsid and mediate cGAS detection of HIV cDNA in the nucleus of human dendritic cells (Lahaye et al., 2018). Interestingly, NONO was responsible for the nuclear localization of a pool of cGAS that was extractable by ~400 mM salt in these cells (Lahaye et al., 2018). To test whether NONO is also essential for the tight tethering of the majority of nuclear cGAS, we generated four independent clonal lines of NONO-deficient HeLa cells (Figure 4—figure supplement 1A). We found that NONO-deficient cells produced normal amounts of cGAMP after DNA transfection (Figure 4—figure supplement 1B), consistent with the prior report (Lahaye et al., 2018). However, the tight nuclear tethering of cGAS was unaffected in NONO-deficient HeLa cells (Figure 4—figure supplement 1A). NONO may act as a ‘bridge’ to enable cGAS detection of virus-encapsidated DNA, as demonstrated in the prior study (Lahaye et al., 2018), but it is not the primary tether of cGAS. Lastly, we used a validated lentiCRISPR approach to disrupt the gene encoding IFI16 (Gray et al., 2016), which has been proposed to interact with cGAS and contribute to its activation (Orzalli et al., 2015). We found that IFI16 was extracted by ~420 mM salt into the nuclear supernatant and that IFI16 disruption resulted in no change in cGAS protein levels or its tight tethering in the nuclear pellet (Figure 4—figure supplement 1C).

Intact chromatin is required for cGAS tethering

cGAS binds to DNA in a sequence-independent fashion, and its association with chromatin is thought to be generic, limited to mitosis, and mediated by its intrinsic affinity for DNA. To broadly assess the requirement for chromatin in the tethering of nuclear cGAS, we treated nuclei from both THP1 and HeLa cells with two broad-spectrum nucleases that digest both DNA and RNA. First, we digested the nuclear extracts after the zero salt lysis step for 30 min at 37° C with micrococcal nuclease, which was sufficient to convert the vast majority of chromatin into nucleosome-protected DNA fragments under 200 bp in length (Figure 5A). Second, we added salt-active nuclease (SAN) to the 250 mM salt elution step, which eliminated all detectable DNA from the samples (Figure 5A). In both cases, the nuclease digestions resulted in a collapsed salt elution profile for histones and cGAS, with the majority of cGAS released from the pellets at 250 mM and 500 mM salt (Figure 5B). These data demonstrate that intact chromatin is required for the organization of cGAS nuclear tethering.

Figure 5. — (A) THP-1 or HeLa cell nuclear extracts were left untreated (NoTx), treated after the 0 salt wash step with micrococcal nuclease (MN), or treated with at the 0.25 M NaCl elution step with Salt Active Nuclease (SAN). Supernatants (supe) and pellets were collected, and the untreated pellet was sonicated to shear large genomic DNA. DNA was extracted, run on an agarose gel, and visualized with SYBR Safe. (B) Extracts treated as described above were used for sequential salt elution, followed by blotting for Histone H2B or cGAS.

Identification of the evolutionarily conserved cGAS tethering surface

In the course of our studies of the relationship between DNA binding and nuclear tethering, we generated the R222E mutant of mouse cGAS (Li et al., 2013). This mutant, purified in recombinant form lacking the cGAS N terminus, was previously shown to be defective for binding to short (45 bp) double-stranded DNA oligonucleotides in vitro, with reduced DNA-activated cGAMP production (Li et al., 2013). However, in a transient transfection-based IFN-luciferase assay, R222E cGAS activated STING at levels comparable to WT cGAS (Li et al., 2013). We introduced the R222E mutation into full-length murine cGAS and transduced cGAS-deficient HeLa cells using our dox-inducible lentivirus system. Remarkably, we found that dox-induced expression alone, which did not activate WT cGAS, resulted in massive cGAMP production by the R222E mutant (Figure 6A). This constitutive cGAMP production by R222E cGAS was similar to the amount made by WT cGAS upon DNA transfection (Figure 6A). Importantly, DNA transfection did not further activate R222E, demonstrating that it was maximally active in the absence of exogenous DNA (Figure 6A).

Figure 6. — (A) cGAS KO HeLa cells were reconstituted with the indicated Dox-inducible murine cGAS lentivirus constructs, treated with 0.1 μg /ml Dox for 24 hr, then transfected for four hours with lipofectamine alone or lipofectamine + CT DNA, followed by measurement of cGAMP in cell lysates. (B) Salt elution profiles of WT and R222E mouse cGAS; the top two panels are in HeLa cells and the bottom panel is in TERT-immortalized human fibroblasts. (C) Crystal structures of mouse cGAS (green) assembled on DNA, modeled from PDBID 5N6I. DNA is orange, R222 is highlighted in purple spheres, and the locations of the active sites are noted with red asterisks. (D) Overlay of mouse cGAS (green) and human OASL (cyan; PDBID 4XQ7). The unique loop in cGAS is colored yellow and indicated with the red arrow. (E) Alignments of cGAS across vertebrate phylogeny, with conserved, positively charged residues highlighted in red and the central loop amino acids indicated by the yellow bar. (F) The positively charged residues near R222 are highlighted in purple spheres. (G) Salt elution profiles of cGAS KO HeLa cells transduced with the indicated murine cGAS constructs after induction with 0.1 μg/ml Dox for 24 hr. (H) Cells from (G) were treated with 0.1 μg/ml Dox for 24 hr, then transfected for four hours with lipofectamine alone or lipofectamine + CT-DNA, followed by measurement of cGAMP in cell lysates. (I) Salt elution profiles of cGAS KO HeLa cells transduced with the indicated human cGAS constructs after induction with 0.1 μg/ml Dox for 24 hr. (J) Cells from (I) were treated as in (H), followed by measurement of cGAMP in cell lysates. Statistical comparisons were made between resting WT cGAS and each mutant, using one-way ANOVA of log-transformed biological replicates. *p=0.0155; **p=0.0018; ***p<0.0001; ns: not significant.

One potential explanation for this striking result is that the R222E mutation changed the conformation of the cGAS active site such that it was no longer dependent on the DNA-triggered structural rearrangements that are normally required for activation of cGAMP production. To test this directly, we made compound mutants of R222E with three of the mutants described in Figure 4 to test the contributions of DNA binding (K335E; K395M/K399M) and DNA-activated dimerization (K382A). We found that all three of these compound mutants were completely defective for cGAMP production, in the absence or presence of exogenous DNA (Figure 6A). Thus, the massive constitutive activation of the cGAS R222E mutant is DNA-dependent.

We evaluated the salt elution profiles of the murine R222E cGAS mutant in both cGAS KO HeLa cells and TERT-HFFs. In both cases, we found that the R222E mutant remained predominantly nuclear, but it eluted at significantly lower salt concentrations (Figure 6B), similar to the elution profile of cGAS after digestion of nuclear extracts with nucleases (Figure 5). Taken together, these data demonstrate that the cGAS R222E mutant is untethered and constitutively active against self-DNA.

We mapped the location of murine cGAS R222 onto the crystal structure of DNA-bound murine cGAS (Figure 6C) (Andreeva et al., 2017), compared it to related proteins, and noted several important features for further study. R222 is found within the core NTase domain of cGAS, distant from the active site and located proximal to DNA in the structure, consistent with its original description as a DNA-binding residue (Figure 6C). We found that R222 and its surrounding surface define a module that is unique to cGAS and not conserved in OAS proteins (Figure 6D; Ibsen et al., 2015). This module includes a protruding loop that is poorly resolved among numerous cGAS structures, suggesting dynamic flexibility (Figure 6D). Phylogenetic analysis of this surface reveals that R222 and R241 of murine cGAS are completely conserved across ~360 million years of vertebrate evolution, despite poor overall sequence conservation in this region of the protein (Figure 6E). Notably, R241, which is immediately adjacent to R222 in the cGAS structure, projects away from DNA and does not participate in DNA binding (Figure 6F). Finally, we found two additional positively charged residues within this surface that are largely - but not completely - conserved, corresponding to K240 and R244 of murine cGAS (Figure 6E–F).

We generated the K240E, R241E, and R244E single point mutants of murine cGAS, along with a ‘Δ-Loop’ mutant in which the central amino acids of the protruding loop (IPRGNP) were replaced with a flexible SGSGSG sequence. We evaluated the effects of these mutations on cGAS nuclear tethering and on constitutive and DNA-activated cGAMP production. We found that the K240E mutant was untethered and significantly constitutively active upon expression alone (Figure 6G–H), but it was also impaired by 90% for DNA-activated cGAMP production compared to WT cGAS, suggesting that K240 plays a dual role in both nuclear tethering and in DNA-dependent activation of cGAS. Remarkably, the R241E mutant was untethered, even more constitutively active than R222E, and not further activated by DNA transfection (Figure 6G–H). The amount of cGAMP produced constitutively by R241E was 300 times more than resting WT cGAS and nearly triple the amount produced by DNA-activated WT cGAS. Finally, the R244E mutant and the Δ-Loop mutant were mildly less tethered and mildly constitutively active, but they remained strongly inducible by DNA transfection (Figure 6G–H). Thus, specific amino acids are essential for both tight tethering of nuclear cGAS and for prevention of cGAS activation by self-DNA.

To test whether tethering and negative regulation of cGAS are evolutionarily conserved, we generated mutants in human cGAS (hcGAS) corresponding to murine R222 (hR236E) and murine R241 (hR255E). We found that both hR236E cGAS and R255E hcGAS mutants were untethered and constitutively active, with R255E hcGAS constitutively producing over 100 times more cGAMP than WT hcGAS (Figure 6I–J). The amounts of constitutive cGAMP produced by the human cGAS mutants were lower than the amounts made by the corresponding murine mutants, likely reflecting the fact that the human cGAS enzyme is intrinsically less active than murine cGAS (Zhou et al., 2018). Moreover, these mutants remained further inducible by DNA (Figure 6I–J), suggesting subtle differences in the contributions of these specific amino acids to regulation of mouse and human cGAS. Taken together, these findings identify the tethering surface of cGAS and demonstrate that tight nuclear tethering is a fundamental and evolutionarily conserved mechanism that prevents cGAS activation by self-DNA.

Discussion

Put simply, the cytosolic DNA sensing model holds that cGAS is a cytosolic protein that is activated by binding to double-stranded DNA that appears in the cytosol (Stetson and Medzhitov, 2006). Our findings demonstrate that DNA detection by cGAS is more complex than this simple model, and they warrant further study of the relationship between resting and activated cGAS, together with the spatial and biochemical transitions that accompany its activation.

We show, using microscopy and biochemical fractionation, that the great majority of endogenous cGAS is nuclear prior to its activation, in all cells tested, in all phases of the cell cycle. Many recent studies in cell lines, most using overexpressed, tagged cGAS, have shown images of cGAS localized to the cytosol and absent from the nucleus, although nuclear cGAS has been observed in mouse fibroblasts expressing GFP-cGAS (Yang et al., 2017) and in mouse hematopoietic stem cells (Xia et al., 2018). Others have recently reported regulated translocation of a fraction of overexpressed cGAS from the cytosol to the nucleus in response to DNA damage (Liu et al., 2018). Finally, studies of cGAS localization to micronuclei envision the recruitment of cytosolic cGAS to cytosolic micronuclei upon rupture of the membrane surrounding these structures (Bartsch et al., 2017; Dou et al., 2017; Glück et al., 2017; Harding et al., 2017; Mackenzie et al., 2017; Yang et al., 2017). We cannot explain these disparate findings, but our data demonstrate that endogenous cGAS is in the nucleus prior to its activation, in the absence of exogenous DNA damaging agents, and prior to the appearance of micronuclei that require mitosis for their formation.

We find that endogenous cGAS is tethered tightly in the nucleus by a force that is remarkably resistant to salt extraction, which explains why this pool of cGAS has been missed when conventional cytosolic and nuclear extracts have been reported in prior studies. We demonstrate that cGAS nuclear localization and nuclear tethering do not require the specific amino acids in cGAS that are essential for robust DNA binding, for DNA-induced oligomerization, for DNA-induced condensation into phase-separated liquid droplets, or for catalytic activity. We show that cGAS nuclear tethering requires intact chromatin. Finally, we identify the tethering surface of cGAS, which resides within NTase domain, and we show that failure to tether cGAS results in its constitutive activation by self-DNA.

By identifying and mutating the tethering surface of cGAS, we uncover a number of important features that govern cGAS localization and its disposition prior to activation. First, the location of the tethering surface corroborates the domain mapping and mutational analysis presented in Figure 4, and it demonstrates that cGAS nuclear tethering and DNA-based activation are two distinct processes. Second, our findings are incompatible with a number of recent studies that have suggested an important role for the N terminus of cGAS in localization to either the cytosolic plasma membrane (Barnett et al., 2019) or to the nucleus (Gentili et al., 2019), or a primary role for cGAS DNA binding in its association with chromatin (Jiang et al., 2019). In particular, our data offer an alternative explanation for the observation that cGAS co-sediments with plasma membranes upon fractionation (Figure 1 of Barnett et al., 2019): in order to reveal this pool of cGAS, cellular extracts needed to be treated prior to fractionation with nuclease, which liberates tethered nuclear cGAS from chromatin (Figure 5).

What is the nature of the tether that tightly binds cGAS to chromatin? A recent study found that recombinant, truncated cGAS lacking its N terminus binds more tightly to purified, recombinant nucleosomes than to naked DNA (Zierhut et al., 2019), which is consistent with the requirement for chromatin in cGAS nuclear tethering (Figure 5). However, the relevance of these in vitro studies to the tight nuclear tethering of cGAS that we define in cells remains unclear. Interestingly, the residues that are important for nuclear tethering overlap with, but are distinct from, one of the DNA-binding surfaces of cGAS (Figure 6). Because of this, we propose that interaction between cGAS and the tethering factor(s) would not be compatible with assembly of cGAS on DNA, which would explain why resting nuclear cGAS is not activated by the billions of base pairs of genomic DNA that reside in the same compartment. If cGAS were tethered to nuclear chromatin through DNA binding, it would be difficult to explain how it is not constitutively activated. Moreover, if resting cGAS were saturated with self-DNA (in an undefined manner that would not activate cGAS), then its activation would require the titration of cGAS off ‘inert’ self-DNA and onto ‘activating’ foreign DNA. How could DNA tether resting cGAS and then be replaced with the identical chemical structure to activate it? There is no biological precedent for such a model, and the fact that loss of tethering results in massive cGAS activation by self-DNA strongly suggests that tethered nuclear cGAS is physically sequestered from DNA. Based on these considerations, we speculate that tethered nuclear cGAS does not directly interact with genomic DNA, and that the tethering mechanism maintains resting cGAS in a state that is competent to detect foreign DNA. More broadly, we propose that analogous tethers might serve as platforms to regulate transactions between other DNA-binding proteins and DNA in the nucleus, serving as a mechanism to prevent inappropriate activation of enzymes that require and/or act on DNA.

We show that untethered cGAS is constitutively active against self-DNA. Thus, cGAS is not ‘inert’ prior to its encounter with foreign DNA, as has been envisioned in the cytosolic DNA sensing model. Instead, tight nuclear tethering is an active process that is essential for preventing autoreactivity, a finding that has a number of implications. First, there must be a regulated step prior to assembly of cGAS onto DNA that allows its activation. We propose a model of ‘regulated desequestration’ of nuclear cGAS and DNA that is required for its full activation, which is then followed by the phase separation and condensation of cGAS onto DNA that was recently demonstrated (Du and Chen, 2018). Our model explains the fact that resting cGAS is not active, and it also allows for nuclear cGAS, together with the factors that regulate tethering, to distinguish self DNA from foreign DNA within the same compartment. We note that the production of the diffusible second messenger cGAMP by cGAS provides a simple explanation for how activated nuclear cGAS could trigger the cytosolic signaling complex of STING, TBK1, and IRF3. Second, there may be conditions in which perturbation of nuclear tethering alone would result in activation of cGAS by self-DNA. Such conditions might include genetic polymorphisms in cGAS or the tether that govern the strength of cGAS tethering, or environmental changes that relax cGAS tethering in response to stress or damage. In these cases, cGAS autoreactivity would not require the production of a distinct DNA ligand that is erroneously sensed as ‘foreign,’ nor would it require that such DNA leave the nucleus. Finally, because untethering alone is sufficient to activate cGAS without the need for foreign DNA, it is possible that self-DNA contributes to ligand-dependent cGAS activation in the context of viral infection. In other words, upon ‘regulated desequestration,’ cGAS assembly on either self-DNA or foreign DNA would achieve the identical result of cGAMP production and antiviral response.

In conclusion, we have found that two separate processes govern the resting and activated states of cGAS, and we propose that a deeper understanding of these states will illuminate new aspects of cGAS biology, with implications for mechanisms of self/non-self discrimination by the innate immune system.

Materials and methods

Key resources table.

Reagent type (species) or resource	Designation	Source or reference	Identifiers
Cell line (human)	HeLa	ATCC	CCL-2
Cell line (human)	SiHa	ATCC	HTB-35
Cell line (human)	THP1	ATCC	TIB-202
Cell line (human)	HEK 293T	ATCC	CRL-3216
Cell line (human)	Primary human foreskin fibroblasts	Millipore	Cat # SCC-058
Cell line (mouse)	mouse bone marrow macrophages, C57BL/6J	Gray EE et al, J Immunol 2015 195:1939 PMID: 27496731
Cell line (human)	HeLa H1 LentiCRISPR control clonal lines	This Paper
Cell line (human)	HeLa cGAS LentiCRISPR clonal lines	This Paper
Cell line (human)	Hela cGAS KO pSLIK GFP	This Paper
Cell line (human)	Hela cGAS KO pSLIK GFP FL mcGAS	This Paper
Cell line (human)	HeLa cGAS KO pSLIK GFP mcGAS 161–522	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS WT	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R222E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R222E/K335E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R222E/K335E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R222E/K395M/K399M	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R222E/K382A	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS K240E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R241E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS R244E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK mcGAS ΔLoop	This Paper
Cell line (human)	HeLa cGAS KO pSLIK hcGAS WT	This Paper
Cell line (human)	HeLa cGAS KO pSLIK hcGAS R236E	This Paper
Cell line (human)	HeLa cGAS KO pSLIK hcGAS R255E	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS full length (FL)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS (161-522)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS (1-161)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS (213-522)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS (161-213)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS (161-382)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK GFP-mcGAS (213-382)	This Paper
Cell line (human)	hTERT human fibroblast pSLIK mcGAS K335E	This Paper
Cell line (human)	hTERT human fibroblast pSLIK mcGAS K395M/K399M	This Paper
Cell line (human)	hTERT human fibroblast pSLIK mcGAS Zinc thumbless	This Paper
Cell line (human)	hTERT human fibroblast pSLIK mcGAS K382A	This Paper
Cell line (human)	hTERT human fibroblast pSLIK mcGAS E386A	This Paper
Cell line (human)	hTERT human fibroblast pSLIK mcGAS E211A/D213A	This Paper
Cell line (human)	hTERT human fibroblast H1 lentiCRISPR control	Gray EE et al, Immunity 2016 45:255 PMID: 27496731
Cell line (human)	hTERT human fibroblast IFI16 lentiCRISPR	Gray EE et al, Immunity 2016 45:255 PMID: 27496731
Cell line (mouse)	mouse bone marrow macrophages, Cgas-/-	Gray EE et al, J Immunol 2015 195:1939 PMID: 26223655
Cell line (mouse)	WT MEF (mouse embryonic fibroblasts)	Gray EE et al, J Immunol 2015 195:1939 PMID: 26223655
Transfected construct	pSLIK-Neo	Addgene	Plasmid #25735
Transfected construct	pSLIK-Blasticidin	This Paper
Antibody	rabbit monoclonal anti-cgas D1D3G	Cell Signaling	cat # 15102S
Antibody	rabbit monoclonal anti-cgas, mouse specific	Cell Signaling	cat # 31659
Antibody	rabbit monoclonal anti-LSD1 C69G12	Cell Signaling	cat # 2184S
Antibody	rabbit polyclonal anti-histone H3	abcam	cat # ab1791
Antibody	rabbit polyclonal anti-histone H2B	abcam	cat # ab1790
Antibody	rabbit polyclonal anti-GFP	abcam	cat # 6556
Antibody	mouse monoclonal anti-Ifi16	abcam	cat # 55328
Antibody	rabbit polyclonal anti-NONO	Sigma Aldrich	cat # N8789-200UL
Antibody	rabbit polyclonal anti- α/β tubulin	Cell Signaling	cat # 2148
Antibody	mouse monoclonal anti-β actin	SIGMA	cat # A5441
Antibody	rabbit IgG control	Fisher	cat # 10500C
Antibody	rabbit polyclonal anti-cgas	SIGMA	cat # HPA031700
Antibody	mouse monoclonal anti-β-tubulin	Cell Signaling	cat # 86298
Commercial assay or kit	NE-PER nuclear and cytoplasmic extraction reagents	Thermo Fisher Scientific	cat # PI78835
Commercial assay or kit	propidium iodide	SIGMA	cat # P4170
Commercial assay or kit	Luciferase Assay System	Promega	cat # E4550
Commercial assay or kit	2',3'-Cyclic GAMP Direct EIA Kit	Arbor Assays	cat # K067-H1
Chemical compound, drug	thymidine	VWR	cat # 80058–750
Chemical compound, drug	SAN (salt active nuclease)	SIGMA	cat # SRE0015
Chemical compound, drug	calf thymus DNA	SIGMA	cat # D4764
Chemical compound, drug	micrococcal nuclease	New England Biolabs	cat # M0247S
Chemical compound, drug	DAPI	SIGMA	cat # D9542
Chemical compound, drug	SYBR safe	Apex Bio	cat # A8743
Chemical compound, drug	DRAQ5	Thermo Fisher	cat # 62251
Chemical compound, drug	NP-40 substitute	SIGMA	cat # 74385
Chemical compound, drug	protease inhibitor tablet	Pierce	cat # PIA32955
Chemical compound, drug	Prolong Gold Antifade mountant	Thermo Fisher	cat # P36930
Chemical compound, drug	Roche Block	Sigma	cat # 11921673001
Other	Immobilon-FL PVDF, 0.45 µm western blot membrane	SIGMA	cat # IPFL00010
Other	Immobilon-PSQ PVDF,0.2 µm western blot membrane	SIGMA	cat # ISEQ00010

Open in a new tab

Cell lines and mice

The following human cell lines were purchased from ATCC. Some of these lines are on the list of commonly misidentified cell lines maintained by the International Cell Line Authentication Committee; we used STR profiling from the University of Arizona Genetics Core to confirm their identity. We also tested for Mycoplasma contamination using a commercially available kit (ABM, cat # G238). All cell lines used in this study tested negative for Mycoplasma contamination.

HeLa: ATCC CCL-2
SiHa: ATCC HTB-35
THP1: ATCC TIB-202

Primary human foreskin fibroblasts were purchased from Millipore, cat # SCC058, and immortalized with retrovirus expressing h-TERT for this study.

C57BL/6J mice were purchased from Jackson Laboratories. Cgas-/- mice were generated previously, and bone marrow macrophages and MEFs were made as described (Gray et al., 2015).

Immunofluorescence microscopy

HeLa cells or primary mouse bone marrow-derived macrophages (BMMs) were seeded onto glass coverslips overnight, then fixed and permeabilized with ice cold methanol at −20 degrees C for 10 min. Cells were then washed in PBS and blocked at room temperature for 2 hr (HeLa cells in Roche block in PBS; BMM in 5% normal goat serum in PBS). Cells were then incubated with primary antibody in block overnight at 4° C. For HeLa cells, we used anti-cGAS CST D1D3G Ab at 1:50 and anti-β-tubulin CST D3U1W Ab at 1:100. For mouse BMMs, we used anti-cGAS CST D3080 Ab at 1:250. Cells were washed in PBS and incubated with secondary Ab (goat anti-rabbit Alexafluor 488, goat anti-mouse Alexafluor 546, Invitrogen) at 1:500 for 1 hr at room temperature. Cells were then washed with PBS, stained with DAPI and mounted on glass slides with ProLong Gold Antifade Mountant (Thermo Fisher). For PFA fixation, cells were fixed in 4% PFA for 10 min, washed in PBS, then permeabilized in 0.1% Triton X-100 in PBS for 10 min. Cells were then washed in PBS, blocked for one hour at room temperature in Roche block in PBS, then stained as stated above. Images were captured with a Nikon C2RSi Scanning Laser Microscope, using a Plan Apo VC 60 × Oil DIC N2 objective in the 405, 488, and 562 dichroic channels and z-steps of 0.125 µm with NISElements software, and then pseudocolored using Fiji open source software. Z-series and 3D volume views were created with NISElements software.

Generation of cGAS and NONO knockout HeLa cells

LentiCRISPR vector generation and lentiviral transductions were done as described previously (Gray et al., 2016). Clonal lines of HeLa cells were generated by limiting dilution and then assessed for targeting by Sanger sequencing, western blot analysis, and functional assays for cGAMP production. The guide RNAs used were:

H1 off-target control: 5’-(G)ACGGAGGCTAAGCGTCGCAA (Sanjana et al., 2014), where the (G) denotes a nucleotide added to enable robust transcription from the U6 promoter; cGAS: 5’-GGCGCCCCTGGCATTCCGTGCGG, where the underlined sequence denotes the Protospacer Adjacent Motif (PAM); NONO: 5’-CTGGACAATATGCCACTCCGTGG.

Amnis imagestream analysis

cGAS KO HeLa cells transduced with pSLIK GFP-mcGAS were treated with 1 μg/ml dox for 24 hr. Cells were washed in PBS and then rested in complete media for 24 hr. Cells were then released from the plate with trypsin, washed in PBS, and stained with 3.125 μM DRAQ5 in PBS before running on an Amnis Imagestream X Mark II imaging cytometer. Data were analyzed with Ideas software (version 6.2).

Salt extractions

We modified a published protocol for histone extraction (Shechter et al., 2007). Cells were pelleted, washed in PBS, resuspended in 1 mL extraction buffer (10 mM Hepes pH 7.9, 10 mM KCl, 1.5 mM MgCl₂, 0.34 M sucrose, 10% glycerol, 0.2% NP-40, and Pierce protease inhibitors), and incubated on ice for 10 min with occasional vortexing. Nuclei were spun at 6500 x g for 5 min at 4° C. The cytosolic fraction (supernatant) was collected for further analysis. Nuclei were then washed for 1 min on ice in extraction buffer without NP-40 and spun at 6500 x g for 5 min at 4°C. Pelleted nuclei were then resuspended in 1 mL zero salt buffer (3 mM EDTA, 0.2 mM EGTA, and protease inhbitors), and vortexed intermittently for 1 min (10 s on, 10 s off). Nuclei were then incubated on ice for 30 min, vortexing for 15 s every 10 min. Lysates were then spun at 6500 x g for 5 min at 4° C. The zero salt supernatant was collected for further analysis. The remaining pellets were then resuspended in first salt buffer (50 mM Tris-HCl, pH 8.0, 0.05% NP-40, 250 mM NaCl), incubated on ice for 15 min with vortexing for 15 s every 5 min. Lysates were spun at full speed (15,000 rpm) at 4°C for 5 min. Supernatants were collected for further analysis. Subsequent salt extractions were performed on the pellet with sequential increases in NaCl concentration (500 mM, 750 mM, 1 M, 1.25 M, 1.5 M, 1.75 M, and 2 M). Samples in each salt wash were incubated on ice for 15 min with vortexing for 15 s every 5 min. Supernatants following each salt condition were collected for further analysis. The final pellet was then resuspended in salt buffer with 2M NaCl and sonicated with a Covaris M220 focused ultrasonicator at 5% ChIP (factory setting), or digested with Salt Active Nuclease (SAN) where the buffer was supplemented with 20 mM MgCl₂. All samples were supplemented with denaturing SDS-PAGE sample buffer, separated on acrylamide gels, transferred to membranes for western blot (0.2 μM pore size for histone blots, 0.45 μM pore size for all other blots), and blotted with the indicated primary and secondary antibodies using standard approaches. Western blot images were acquired and densitometry analysis was performed using a BioRad Chemidoc and associated software.

NE-PERS kit modification

The NE-PERS kit instructions (Thermo Fisher) were followed completely, with the following modification: after spinning the pellet out of the NER buffer, the supernatant was removed and saved as ‘nuclear supernatant (NS)'. The remaining pellet was resuspended in a volume of NER buffer equal to the first, and either sonicated (using Covaris M220 5% ChIP factory setting), or digested with SAN in NER buffer supplemented with 20 mM MgCl₂. This was then saved as ‘nuclear pellet (NP)'.

Double Thymidine block

Cells were seeded onto plates to achieve 40% confluency. The next day cells were treated with 2 mM thymidine in complete media for 19 hr. Cells were then washed in warm PBS and rested in complete media for 9 hr. Cells were then treated again with 2 mM thymidine in complete media for 16 hr. Cells were then either harvested for analysis by western blot or flow cytometry, or washed and returned to complete media for harvest at post-release time points. Flow cytometry analysis was performed as described above.

cGAMP quantitation assay

Cells were plated at 100,000 cells/well in a 24 well tissue culture dish. 24 hr later, cells were transfected with either 10 μg/ml CT-DNA in lipofectamine 2000 (Invitrogen; ratio of 1 μL lipofectamine per 1 μg CT-DNA) (Stetson and Medzhitov, 2006), or with an identical volume of lipofectamine 2000 alone. 4 hr later, cells were harvested and lysates were prepared using cGAMP EIA assay protocol provided by manufacturer (Arbor Assays), in a volume of 200 μL sample suspension buffer.

Constructs

The pSLIK-Neo doxycycline-inducible lentiviral vector was obtained from Addgene and modified to replace the Neo cassette with a blasticidin resistance cassette. GFP fusions to the murine cGAS open reading frame were generated by PCR mutagenesis and designed to incorporate a four-glycine flexible linker between the last amino acid of GFP and the first amino acid of cGAS. Lentivirus production and blasticidin selection were done using standard techniques.

cGAS-deficient HeLa cells were reconstituted with Dox-inducible lentiviruses encoding GFP, the indicated GFP-human cGAS fusions, or the indicated GFP-mouse cGAS constructs. Cells were plated at 50,000 cells per well in a 24 well plate for quantitation of cGAMP, or 250,000 cells per well in a 6-well plate for salt extractions. 24 hr later, cells were treated with doxycycline for 24 hr. Then, cells were harvested directly for anti-GFP western blot from the 6-well dishes, and the 24-well dishes were transfected with either 10 μg/mL CT-DNA complexed with lipofectamine 2000, or with lipofectamine 2000 alone. 4 hr later, lysates were prepared and analyzed for cGAMP content as described above (whole cells lysed in 200 μL sample suspension buffer).

Nuclease digestions and salt elutions

Salt extractions were performed as described above with the following modifications. 1 × 10⁶ cells were used for each condition. Following the zero salt wash, all samples were resuspended in digestion buffer (50 mM Tris pH 8.0, 0.05% NP-40, 1 mM MgCl₂, 5 mM CaCl₂). For the MNase digestion, MNase was added at 20,000 gel units per sample. For the SAN digestion, 50 units SAN nuclease plus 20 mM MgCl₂ were added at the 250 mM NaCl step. All samples were incubated at 37°C for 10 min. Samples were then spun down and supernatants and pellets were separated and then processed for western blots using the salt elution protocol. For analysis of DNA content, the MNase-digested supernatants and pellets were analyzed after digestion and before commencement of salt elution. For the SAN digestion, the pellet was collected after the 500 mM NaCl elution for assessment of DNA content. DNA was run on an agarose gel, stained using SYBR-Safe reagent (Apex Bio), and visualized using a BioRad Chemidoc.

Experimental replicates and reproducibility

All data presented in this paper are representative of 2–4 independent experiments with comparable results.

Acknowledgements

We are grateful to Quinton Dowling, Neil King, and Emily Schutsky for their help with cGAS structural analysis, to Andrew Oberst and Naeha Subramanian for the pSLIK-Neo vector, to Michael Gale, Jr for use of the confocal microscope, and to the entire Stetson lab for helpful discussions.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Daniel B Stetson, Email: stetson@uw.edu.

Tadatsugu Taniguchi, Institute of Industrial Science, The University of Tokyo, Japan.

Funding Information

This paper was supported by the following grants:

National Institutes of Health AI084914 to Daniel B Stetson.
Jane Coffin Childs Memorial Fund for Medical Research to Hannah E Volkman.
Burroughs Wellcome Fund 1013540 to Daniel B Stetson.
Howard Hughes Medical Institute 55108572 to Daniel B Stetson.
Bill and Melinda Gates Foundation OPP1156262 to Daniel B Stetson.
Cancer Research Institute A84161 to Elizabeth E Gray.

Additional information

Competing interests

No competing interests declared.

Author contributions

Data curation, Formal analysis, Investigation, Methodology.

Investigation, Methodology.

Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition.

Additional files

Transparent reporting form

elife-47491-transrepform.docx^{(248.4KB, docx)}

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files.

References

Andreeva L, Hiller B, Kostrewa D, Lässig C, de Oliveira Mann CC, Jan Drexler D, Maiser A, Gaidt M, Leonhardt H, Hornung V, Hopfner KP. cGAS senses long and HMGB/TFAM-bound U-turn DNA by forming protein-DNA ladders. Nature. 2017;549:394–398. doi: 10.1038/nature23890. [DOI] [PubMed] [Google Scholar]
Barnett KC, Coronas-Serna JM, Zhou W, Ernandes MJ, Cao A, Kranzusch PJ, Kagan JC. Phosphoinositide interactions position cGAS at the plasma membrane to ensure efficient distinction between self- and viral DNA. Cell. 2019;176:1432–1446. doi: 10.1016/j.cell.2019.01.049. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bartsch K, Knittler K, Borowski C, Rudnik S, Damme M, Aden K, Spehlmann ME, Frey N, Saftig P, Chalaris A, Rabe B. Absence of RNase H2 triggers generation of immunogenic micronuclei removed by autophagy. Human Molecular Genetics. 2017;26:3960–3972. doi: 10.1093/hmg/ddx283. [DOI] [PubMed] [Google Scholar]
Bootsma D, Budke L, Vos O. Studies on synchronous division of tissue culture cells initiated by excess thymidine. Experimental Cell Research. 1964;33:301–309. doi: 10.1016/S0014-4827(64)81035-1. [DOI] [PubMed] [Google Scholar]
Civril F, Deimling T, de Oliveira Mann CC, Ablasser A, Moldt M, Witte G, Hornung V, Hopfner KP. Structural mechanism of cytosolic DNA sensing by cGAS. Nature. 2013;498:332–337. doi: 10.1038/nature12305. [DOI] [PMC free article] [PubMed] [Google Scholar]
Crowl JT, Gray EE, Pestal K, Volkman HE, Stetson DB. Intracellular nucleic acid detection in autoimmunity. Annual Review of Immunology. 2017;35:313–336. doi: 10.1146/annurev-immunol-051116-052331. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dou Z, Ghosh K, Vizioli MG, Zhu J, Sen P, Wangensteen KJ, Simithy J, Lan Y, Lin Y, Zhou Z, Capell BC, Xu C, Xu M, Kieckhaefer JE, Jiang T, Shoshkes-Carmel M, Tanim K, Barber GN, Seykora JT, Millar SE, Kaestner KH, Garcia BA, Adams PD, Berger SL. Cytoplasmic chromatin triggers inflammation in senescence and Cancer. Nature. 2017;550:402–406. doi: 10.1038/nature24050. [DOI] [PMC free article] [PubMed] [Google Scholar]
Du M, Chen ZJ. DNA-induced liquid phase condensation of cGAS activates innate immune signaling. Science. 2018;361:704–709. doi: 10.1126/science.aat1022. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gao D, Wu J, Wu YT, Du F, Aroh C, Yan N, Sun L, Chen ZJ. Cyclic GMP-AMP synthase is an innate immune sensor of HIV and other retroviruses. Science. 2013;341:903–906. doi: 10.1126/science.1240933. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gentili M, Lahaye X, Nadalin F, Nader GPF, Lombardi EP, Herve S, De Silva NS, Rookhuizen DC, Zueva E, Goudot C, Maurin M, Bochnakian A, Amigorena S, Piel M, Fachinetti D, Londoño-Vallejo A, Manel N. The N-Terminal domain of cGAS determines preferential association with centromeric DNA and innate immune activation in the nucleus. Cell Reports. 2019;26:3798. doi: 10.1016/j.celrep.2019.03.049. [DOI] [PMC free article] [PubMed] [Google Scholar]
Glück S, Guey B, Gulen MF, Wolter K, Kang TW, Schmacke NA, Bridgeman A, Rehwinkel J, Zender L, Ablasser A. Innate immune sensing of cytosolic chromatin fragments through cGAS promotes senescence. Nature Cell Biology. 2017;19:1061–1070. doi: 10.1038/ncb3586. [DOI] [PMC free article] [PubMed] [Google Scholar]
Goubau D, Deddouche S, Reis e Sousa C. Cytosolic sensing of viruses. Immunity. 2013;38:855–869. doi: 10.1016/j.immuni.2013.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gray EE, Treuting PM, Woodward JJ, Stetson DB. Cutting edge: cgas is required for lethal autoimmune disease in the Trex1-Deficient mouse model of Aicardi-Goutières syndrome. The Journal of Immunology. 2015;195:1939–1943. doi: 10.4049/jimmunol.1500969. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gray EE, Winship D, Snyder JM, Child SJ, Geballe AP, Stetson DB. The AIM2-like receptors are dispensable for the interferon response to intracellular DNA. Immunity. 2016;45:255–266. doi: 10.1016/j.immuni.2016.06.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
Harding SM, Benci JL, Irianto J, Discher DE, Minn AJ, Greenberg RA. Mitotic progression following DNA damage enables pattern recognition within micronuclei. Nature. 2017;548:466–470. doi: 10.1038/nature23470. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ibsen MS, Gad HH, Andersen LL, Hornung V, Julkunen I, Sarkar SN, Hartmann R. Structural and functional analysis reveals that human OASL binds dsRNA to enhance RIG-I signaling. Nucleic Acids Research. 2015;43:5236–5248. doi: 10.1093/nar/gkv389. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ishikawa H, Ma Z, Barber GN. STING regulates intracellular DNA-mediated, type I interferon-dependent innate immunity. Nature. 2009;461:788–792. doi: 10.1038/nature08476. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jiang H, Xue X, Panda S, Kawale A, Hooy RM, Liang F, Sohn J, Sung P, Gekara NO. Chromatin-bound cGAS is an inhibitor of DNA repair and hence accelerates genome destabilization and cell death. The EMBO Journal. 2019;38:e102718. doi: 10.15252/embj.2019102718. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lahaye X, Satoh T, Gentili M, Cerboni S, Conrad C, Hurbain I, El Marjou A, Lacabaratz C, Lelièvre J-D, Manel N. The capsids of HIV-1 and HIV-2 determine immune detection of the viral cDNA by the innate sensor cGAS in dendritic cells. Immunity. 2013;39:1132–1142. doi: 10.1016/j.immuni.2013.11.002. [DOI] [PubMed] [Google Scholar]
Lahaye X, Gentili M, Silvin A, Conrad C, Picard L, Jouve M, Zueva E, Maurin M, Nadalin F, Knott GJ, Zhao B, Du F, Rio M, Amiel J, Fox AH, Li P, Etienne L, Bond CS, Colleaux L, Manel N. NONO detects the nuclear HIV capsid to promote cGAS-Mediated innate immune activation. Cell. 2018;175:488–501. doi: 10.1016/j.cell.2018.08.062. [DOI] [PubMed] [Google Scholar]
Lau L, Gray EE, Brunette RL, Stetson DB. DNA tumor virus oncogenes antagonize the cGAS-STING DNA-sensing pathway. Science. 2015;350:568–571. doi: 10.1126/science.aab3291. [DOI] [PubMed] [Google Scholar]
Li X, Shu C, Yi G, Chaton CT, Shelton CL, Diao J, Zuo X, Kao CC, Herr AB, Li P. Cyclic GMP-AMP synthase is activated by double-stranded DNA-induced oligomerization. Immunity. 2013;39:1019–1031. doi: 10.1016/j.immuni.2013.10.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
Li T, Chen ZJ. The cGAS-cGAMP-STING pathway connects DNA damage to inflammation, senescence, and Cancer. The Journal of Experimental Medicine. 2018;215:1287–1299. doi: 10.1084/jem.20180139. [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu S, Cai X, Wu J, Cong Q, Chen X, Li T, Du F, Ren J, Wu YT, Grishin NV, Chen ZJ. Phosphorylation of innate immune adaptor proteins MAVS, STING, and TRIF induces IRF3 activation. Science. 2015;347:aaa2630. doi: 10.1126/science.aaa2630. [DOI] [PubMed] [Google Scholar]
Liu H, Zhang H, Wu X, Ma D, Wu J, Wang L, Jiang Y, Fei Y, Zhu C, Tan R, Jungblut P, Pei G, Dorhoi A, Yan Q, Zhang F, Zheng R, Liu S, Liang H, Liu Z, Yang H, Chen J, Wang P, Tang T, Peng W, Hu Z, Xu Z, Huang X, Wang J, Li H, Zhou Y, Liu F, Yan D, Kaufmann SHE, Chen C, Mao Z, Ge B. Nuclear cGAS suppresses DNA repair and promotes tumorigenesis. Nature. 2018;563:131–136. doi: 10.1038/s41586-018-0629-6. [DOI] [PubMed] [Google Scholar]
Luo Y, Tucker SC, Casadevall A. Fc- and complement-receptor activation stimulates cell cycle progression of macrophage cells from G1 to S. The Journal of Immunology. 2005;174:7226–7233. doi: 10.4049/jimmunol.174.11.7226. [DOI] [PubMed] [Google Scholar]
Ma Z, Jacobs SR, West JA, Stopford C, Zhang Z, Davis Z, Barber GN, Glaunsinger BA, Dittmer DP, Damania B. Modulation of the cGAS-STING DNA sensing pathway by gammaherpesviruses. PNAS. 2015;112:E4306–E4315. doi: 10.1073/pnas.1503831112. [DOI] [PMC free article] [PubMed] [Google Scholar]
Maciejowski J, Li Y, Bosco N, Campbell PJ, de Lange T. Chromothripsis and Kataegis induced by telomere crisis. Cell. 2015;163:1641–1654. doi: 10.1016/j.cell.2015.11.054. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mackenzie KJ, Carroll P, Martin CA, Murina O, Fluteau A, Simpson DJ, Olova N, Sutcliffe H, Rainger JK, Leitch A, Osborn RT, Wheeler AP, Nowotny M, Gilbert N, Chandra T, Reijns MAM, Jackson AP. cGAS surveillance of micronuclei links genome instability to innate immunity. Nature. 2017;548:461–465. doi: 10.1038/nature23449. [DOI] [PMC free article] [PubMed] [Google Scholar]
Orzalli MH, Broekema NM, Diner BA, Hancks DC, Elde NC, Cristea IM, Knipe DM. cGAS-mediated stabilization of IFI16 promotes innate signaling during herpes simplex virus infection. PNAS. 2015;112:E1773–E1781. doi: 10.1073/pnas.1424637112. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rasaiyaah J, Tan CP, Fletcher AJ, Price AJ, Blondeau C, Hilditch L, Jacques DA, Selwood DL, James LC, Noursadeghi M, Towers GJ. HIV-1 evades innate immune recognition through specific cofactor recruitment. Nature. 2013;503:402–405. doi: 10.1038/nature12769. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sanjana NE, Shalem O, Zhang F. Improved vectors and genome-wide libraries for CRISPR screening. Nature Methods. 2014;11:783–784. doi: 10.1038/nmeth.3047. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shechter D, Dormann HL, Allis CD, Hake SB. Extraction, purification and analysis of histones. Nature Protocols. 2007;2:1445–1457. doi: 10.1038/nprot.2007.202. [DOI] [PubMed] [Google Scholar]
Stetson DB, Medzhitov R. Recognition of cytosolic DNA activates an IRF3-dependent innate immune response. Immunity. 2006;24:93–103. doi: 10.1016/j.immuni.2005.12.003. [DOI] [PubMed] [Google Scholar]
Sun L, Wu J, Du F, Chen X, Chen ZJ. Cyclic GMP-AMP synthase is a cytosolic DNA sensor that activates the type I interferon pathway. Science. 2013;339:786–791. doi: 10.1126/science.1232458. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wu J, Sun L, Chen X, Du F, Shi H, Chen C, Chen ZJ. Cyclic GMP-AMP is an endogenous second messenger in innate immune signaling by cytosolic DNA. Science. 2013;339:826–830. doi: 10.1126/science.1229963. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wu JJ, Li W, Shao Y, Avey D, Fu B, Gillen J, Hand T, Ma S, Liu X, Miley W, Konrad A, Neipel F, Stürzl M, Whitby D, Li H, Zhu F. Inhibition of cGAS DNA sensing by a herpesvirus virion protein. Cell Host & Microbe. 2015;18:333–344. doi: 10.1016/j.chom.2015.07.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
Xia P, Wang S, Ye B, Du Y, Li C, Xiong Z, Qu Y, Fan Z. A circular RNA protects dormant hematopoietic stem cells from DNA sensor cGAS-Mediated exhaustion. Immunity. 2018;48:688–701. doi: 10.1016/j.immuni.2018.03.016. [DOI] [PubMed] [Google Scholar]
Yang H, Wang H, Ren J, Chen Q, Chen ZJ. cGAS is essential for cellular senescence. PNAS. 2017;114:E4612–E4620. doi: 10.1073/pnas.1705499114. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhou W, Whiteley AT, de Oliveira Mann CC, Morehouse BR, Nowak RP, Fischer ES, Gray NS, Mekalanos JJ, Kranzusch PJ. Structure of the human cGAS-DNA complex reveals enhanced control of immune surveillance. Cell. 2018;174:300–311. doi: 10.1016/j.cell.2018.06.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zierhut C, Yamaguchi N, Paredes M, Luo J-D, Carroll T, Funabiki H. The cytoplasmic DNA sensor cGAS promotes mitotic cell death. Cell. 2019;178:302–315. doi: 10.1016/j.cell.2019.05.035. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.47491.sa1

Decision letter

Editor: Tadatsugu Taniguchi¹

Reviewed by: Andrea Ablasser²

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "cGAS is predominantly a nuclear protein" for consideration by eLife. Your article has been reviewed by two peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

Although the reviewers felt the manuscript dealt with an issue that has not received much attention, i.e., the localization of cGAS to the nucleus, the consensus was that this issue had been previously described. As such, the observation is not entirely novel. Other key aspects of the current manuscript, including the functional impact of cGAS tethering to chromatin, remain unaddressed in a convincing manner. Other conclusions are also not clearly supported by the presented experimental data. Thus, the reviewers and editors decided that the paper is unacceptable for publication at eLife in its current form.

The complete reviews are given below, in hopes that you will find them useful for revising the manuscript for submission elsewhere. You will note that there was some disagreement on the novelty of the findings and other aspects of the paper, as is expected. However, the consensus statement above was produced after extensive discussions between the reviewers and the editors.

Reviewer #1:

This manuscript shows that endogenous cGAS, a nucleic acid sensor, is predominately localized to the nucleus. This observation suggests that current models of "cytosolic nucleic acid sensing" need modification.

The authors show endogenous cGAS is localized to the nucleus of HeLa cells by use of a monoclonal antibody in confocal micrographs. Specificity was verified by use of CRISPR'ed HeLa cells. This was extended to mouse cells using a different mAb, with specificity control being cGAS KO cells. Localization was verified with western blot analysis of nuclear and cytosolic extracts. Nuclear localization was independent of cGAS activation or cell cycle. Additional data support strong association that is salt dependent and the N-terminus is dispensable.

The data for nuclear localization are clear with excellent specificity controls. While it cannot be ruled out that there is a small amount of cGAS in the cytosol that mediates activation, the work certainly indicates that current models of cytosolic nucleic acid sensing are either incomplete and missing a step before cGAS becomes localized to the cytosol, or perhaps more likely, that it mediates sensing while in the cytosol itself. In either case, the paper raises important issues, even though it does not clearly address these latter points.

Reviewer #2:

Cyclic GMP-AMP synthase (cGAS), an innate immune sensor of double-stranded DNA was initially described as a cytosolic protein. However, increasing number of studies in the last 4 years indicate that this protein is also present in the nucleus. One of the key question in the field is how constant activation of nuclear cGAS by genomic DNA is voided to prevent immunopathology. In the present manuscript, Volkman et al. demonstrate that cGAS is present in the nucleus and is tightly tethered to chromatin. Based on this hypothesize that inactivation of nuclear cGAS is due to this tight tethering to chromatin. Further, they claim that this tethering is independent of the functional domains involved in cGAS activation. Their data showing the presence of cGAS in the nucleus looks convincing and expected given what has already been reported across several publication. The other key messages of the manuscript are however inconclusive and highly speculative. Overall, the observations reported in this manuscript are very preliminary. Most of their claims are at this time point of their analysis still speculative not supported by data. Many of the key experiments are also not properly designed and lack controls. Here below are some suggestions for improvement. I hope they find them useful.

Specific points:

1) Conceptual advancement over previous studies: The main message of this study i.e. the presence of cGAS in the nucleus has already been established by several previous studies (e.g. Orzalli et al., 2015;, Yang et al., 2017; Lahaye et al., 2018).

2) The proposal that the inactivity of nuclear cGAS is due to the tight tethering chromatin is interesting. However, this idea was not experimentally tested in this study. Besides, this idea is really not new and has been tested more rigorously in studies already in the public domain (Zierhut C and Funabiki H, bioRxiv, 2017, doi: 10.1101/168070).

3) The authors claim that tethering of cGAS to chromatin is independent of cGAS features important for its activation is not convincing. The functional features of cGAS required for its activation are mainly its DNA binding, enzymatic and dimerization activities. The functional behaviour of the cGAS truncations mutants cannot be assumed to be similar to that within the functional full-length protein. The structural-functional features of cGAS are well defined and a more convincing approach would be for the authors use cGAS point mutants defective in these features. It is really difficult to make any firm conclusion from the truncation mutants since essential controls are lacking. For example, what is the affinity of the different truncation mutants to DNA?

4) The K395M/K399M mutant of mouse cGAS used in Figure 4F is an inactive mutant. Is this because of defect in DNA binding mutant or dimerization? If author want to use enzyme inactive mutant of cGAS, the well characterized 225A/D227A mutant of human cGAS which is more appropriate.

5) The authors speculate that DNA is not the primary tether of cGAS but instead propose some protein component(s) of higher order chromatin. What is the experimental basis for this this proposal? In fact, their preliminary findings with the nuclease digestions suggest implicate DNA.

6) The authors claim that nuclear localization of cGAS is via active translocation rather than via passive mechanisms. While interesting, this proposal is also preliminary and require rigorous experimental controls. At the very minimums they should test whether blocking inhibitors of protein nuclear import prevent nuclear localization of cGAS.

7) Many of the experiments lack essential controls. For example, without loading controls for nuclear and cytosolic markers in the immunoblots in Figure 3, it difficult to interrogate the veracity of their claims.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Endogenous cGAS is predominantly a nuclear protein" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Tadatsugu Taniguchi as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Andrea Ablasser (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. Although nuclear localization of cGAS has already been established by published literature and that the current work leaves unclear the mechanistic basis and implications of cGAS’s nuclear localization, the consensus was that this manuscript is still relevant for the current discussion about the localization of cGAS. The reviewers have identified the following key points that need to be addressed before it can be considered further.

Summary:

cGAS, the main innate immune sensor of DNA inside cells is essential for immune defense against numerous pathogens, but has also been implicated in many inflammatory diseases. cGAS was initially assumed to be sequestered in the cytosol away from the genomic DNA in the nuclear compartments. However, an increasing number of studies indicate that cGAS is not only present in the cytosol but also in the nucleus and plasma membrane, raising the question, how nuclear cGAS is prevented from eliciting an immune reaction to self-DNA. Many of the previous investigations have mainly looked at overexpressed cGAS. In this study Volkman et al. have combined subcellular fractionation and microscopy to study the localization of endogenous cGAS. They show that endogenous cGAS is mainly in the nucleus where it is tightly tethered to chromatin. They propose that the inactivity of nuclear cGAS to genomic DNA is due to its tethering to chromatin. The points in this article are of general interested to researchers in the field of innate immunity.

Essential revisions:

1) A limitation of the paper is that the fractionation experiments are potentially susceptible to an artifact, which is that cGAS might associate with some dense structure that co-sediments with the nucleus, but is not actually itself nuclear. The potential for the fractionation experiments to be misleading means that the complementary immunofluorescence experiments (shown in Figure 1) are of critical importance. One main worry about the data in Figure 1 is that because the nucleus is thicker than the rest of the cell, it may appear that cGAS is concentrated in the nucleus simply because the confocal slices sectioned primarily through the nucleus, and missed the cytosol at the periphery of the cell; or, if a z-series projection of all slices is shown, the fact that there would be more slices through the nucleus might improve the signal for this part of the cell as compared to the cytosol. It is hard to tell exactly what was done because the Materials and methods and legends don't fully describe the imaging and analysis. This part of the paper should therefore be strengthened with some additional methodological description and controls. For example, was a Z-series collected? What step size? Are projections shown? Can a 3D reconstruction (x-z projection) of the cell be shown to confirm that the cGAS is truly nuclear? What is the volume of the nucleus versus the volume of the cytosol and is the apparent localization in the nucleus merely because the nucleus contains most of the cell volume? Can some kind of quantitation be performed on the images in Figure 1 (cytosol vs. nucleus)? Can a control cytosolic protein be shown to convince the reader that were cGAS cytosolic it would be detected? Is it surprising that the DAPI and cGAS images appear so similar? Why doesn't the so-called micronucleus in Figure 1B stain with DAPI (maybe it is not a micronucleus)?

2) One of the major conclusions of this study is that nuclear cGAS localization is not via passive association with chromatin (as reported by others e.g. Yang et al., 2017, Gentili et al., 2019) but via a more specific mechanism. The subcellular fractionation and mutant experiments are not sufficient to justify this conclusion. For example, in Figure 3F, in the absence of mitotic nuclear membrane dissolution, how can the presence of control GFP in their nuclear soluble (NS) and nuclear pellet (NP) fraction be explained? Due to passive or active import? Or could it be post-lysis contamination of such fractions with cytosolic components? Of course, GFP-cGAS is more abundant that control GFP in the NP fraction. However, this could simply be due to the ability of GFP-cGAS and not the GFP control to bind to DNA hence its retention therein. Therefore, similar to the above point, can the authors induce GFP-cGAS expression in cells arrested at G0 or G1 and visualize its localization by life confocal imaging to see distinctly that freshly synthesized cGAS does indeed translocate into the nucleus independent of mitotic nuclear membrane dissolution? In view of these caveats, the authors’ own admission that the nuclear import inhibitors tested did not impede nuclear localization (surprisingly not mentioned this in the text), the claim that nuclear cGAS is via active translocation remains very weak and is potentially misleading. Moreover, given the sensitivity of nuclear pellet-associated cGAS to nucleases (Figure 5) and since none of the cGAS mutants tested completely abrogates DNA binding, it is in simply not possible to discount the role genomic DNA in nuclear localization. Therefore, in the absence of direct evidence the authors should mellow down on this claim.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for submitting your revised work "Endogenous cGAS is predominantly a nuclear protein". The Reviewing Editor has now carefully gone through the manuscript and consulted with the Senior Editor. We are basically in favour of accepting the paper for eLife. Indeed, we appreciate your inclusion of the requested images to show more clearly the subcellular localization of cGAS. We also recognize the technical challenge of ruling out completely the role of DNA binding in cGAS nuclear localization and are willing to accept your explanation highlighted in the Discussion. However, for reasons described below, for us to consider this manuscript further the remaining experiment to shed more light into how cGAS localizes into the nucleus is still required.

As you are aware, cGAS nuclear localization has been reported across many publications already. Moreover, avid binding to chromatin as a possible mechanism for preventing cGAS-mediated immune response to nuclear DNA – one of the main points of the current manuscript, has also been described (Zierhut et al., Cell, 2019, formerly a preprint on bioRxiv). Therefore, in view of the current state of the field, the potentially new aspect of this manuscript is how cGAS is localized to the nucleus. Your theory that cGAS nuclear localization is via active transport from the cytosol advances an interesting concept with major implications in the field since this could provide new avenues for manipulating cGAS-mediated biological processes. On the other hand, if incorrect, this idea can mislead the field. As you have admitted in response to our earlier request, your nuclear import inhibitors experiments so far do not support your theory (these are relevant data and you may consider to include them in the manuscript). Although you have toned down on your initial conclusion, it remains unclear whether nuclear localization is indeed via active or non-specific mechanisms. Therefore at a minimum, the suggested live confocal imaging comparing the localization of inducible GFP-cGAS is still essential. This experiment is feasible and requires standard techniques that should be available in your institution or can be arranged through collaboration.

We previously brought to your attention the fact that GFP control alone does localize to the nucleus (Figure 3D). We have noted your explanation that such localization is likely because of passive translocation of GFP due to its size (27 kD). In this regard, it is equally important to note that passive diffusion across the nuclear envelope does not have fixed molecular mass threshold; proteins of molecular mass of up to 200 kD can diffuse across the nuclear envelope with varying kinetics (Timney et al., J. Cell Biol., 2016). Therefore, an alternative explanation perhaps worth considering is that nuclear cGAS localization is simply via passive diffusion. That is why live confocal imaging comparing the localization of inducible GFP-cGAS and a GFP-tagged control protein with similar molecular mass as cGAS would help to shed some light onto whether cGAS nuclear localization is indeed due to active or passive mechanisms. Even if the conclusion from this experiment were that localization is via passive mechanism, this would still be an important contribution and would provide more clarity to the field.

[Editors’ note: this article was then rejected after discussions between the reviewers, but the article was accepted after an appeal against the decision.]

Thank you for choosing to send your work entitled “Tight nuclear tethering of cGAS is essential for preventing autoreactivity" for consideration at eLife.

Your article and your letter of appeal have been considered by a Senior Editor, and we regret to inform you that we are upholding our original decision. Detailed comments by the Reviewing Editor are described below. I feel very sorry that we cannot be more positive at this stage and sincerely hope that the comments are valuable to your study.

This work was originally submitted under the title "Endogenous cGAS is predominantly a nuclear protein". The initial conclusion of the study were that: (1) cGAS is predominantly a nuclear protein, (2) that nuclear localization of cGAS is independent of DNA binding or cell cycle as previously reported by others, and (3) that this localization is due to active importation from the cytosol. The outstanding concerns throughout the previous rounds of revisions were that their data were inconsistent with the conclusion regarding mechanisms of nuclear localization and association with the chromatin.

In the previous decision letter, we requested one control experiment which in my opinion was feasible and hence the quickest path to publication. Specifically, to verify their claim that nuclear localization was via active mechanisms, we requested for life-microscopic images of their inducible cGAS. The authors did not however provide this controls experiment but have left out the data that we previously pointed out to be inconsistent with their conclusion (e.g. previously in Figure 3D). They have also included new data (not asked for) and have rewritten the manuscript in such a manner that the main message is now very different from the original submission.

In this version, the authors propose that at resting state, cGAS is kept inactive through tight binding (which the authors call tethering) to chromatin and that this tethering is via yet to be identified tether and not DNA. Implicit in this model is that for cGAS activation to occur, cGAS has to undergo untethering to enable it to bind DNA. They claim to have identified a conserved surface on cGAS responsible for this tethering and conclude that this surface is distinct from that for DNA binding. In my assessment the authors have misinterpreted their data and these fresh claims are misleading. In fact the new data support the opposite of their conclusion: that attachment of cGAS to the chromatin in via DNA.

Specific comments:

1) What the authors conclude as tethering surface lie within the DNA binding surface of cGAS. The R222E, R240E, R241E, R244E mutants that the authors report as tethering defective mutants are in fact DNA binding mutants. This is well established in the field (e.g. Figure 5C and Figure 6 of Li et al., Immunity, 2013). Moreover, the other mutants, for example the K335E, K382A, E386A, K395M, K399M which the authors used to conclude that tethering of cGAS to chromatin is independent of DNA binding have also DNA binding mutants. The main difference these sets of mutants is that the suggested "tethering mutants" (R222E, R240E, R241E, R244E) are severely defective in DNA binding compared to the K335E, K382A, E386A, K395M, K399M mutants which retain substantial DNA binding (Figure 6 of Li et al., Immunity, 2013). Therefore in my view, the correct interpretation is that interaction of cGAS with chromatin involves some form of interaction with DNA binding. This is consistent with the authors' data showing that cGAS-chromatin interaction is highly sensitive to DNases (Figure 5).

2) Related to the above point, in my view, the more plausible explanation for the spontaneous activity of the R222E, R240E, R241E, R244E mutants is that mutations in these amino acids likely triggers a conformational change lowering the threshold of activation by DNA. And, of course the compound mutations in R222E together with either K335E, K382A or K395M is expected to result in an inactive mutant (Figure 6A, Figure 4F), since the latter mutations are in themselves inactivating mutation (Li et al., 2013).

3) The manuscript contains statements that are potentially misleading to the readers. For example in the Introduction the authors state "Here, we use confocal microscopy and biochemical characterization to determine the resting localization of endogenous cGAS prior to activation". In the Discussion, the authors go on to conclude, "We show, using microscopy and biochemical fractionation, that the great majority of endogenous cGAS is nuclear prior to its activation." Inherent in this statement is that in resting state cGAS is kept inactive through interaction with chromatin but then undergoes spatial redistribution upon activation. There is no evidence that this is the case.

4) The authors proposed tethering models and the Discussion largely assume that all/or most of the cGAS activation occurs in the nucleus, how do the authors explain cGAS activation following DNA transfection or following some bacterial infections – that such sensing also occurs in the nucleus or that this involves redistribution of cGAS to the cytosol? A shortcoming of the proposed model is that it does factor the constant presence of a chromatin-free cGAS and that it is this pool that most likely becomes activated by foreign or misplaced self-DNA?

In brief, this revised manuscript has not addressed previous concerns. Whereas their data showing that cGAS is abundantly nuclear is convincing and consistent with those reported by recent studies, as exemplified above, many of the key conclusions, especially the new ones, are in my assessment not supported by the data. Therefore I unfortunately I cannot recommend the manuscript for publication. My suggestion to the authors would be to include the suggested controls and refocus the manuscript to the original message and instead describe/characterize the new autoactive mutants more coherently in separate manuscript.

eLife. 2019 Dec 6;8:e47491. doi: 10.7554/eLife.47491.sa2

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

This manuscript shows that endogenous cGAS, a nucleic acid sensor, is predominately localized to the nucleus. This observation suggests that current models of "cytosolic nucleic acid sensing" need modification.

The authors show endogenous cGAS is localized to the nucleus of HeLa cells by use of a monoclonal antibody in confocal micrographs. Specificity was verified by use of CRISPR'ed HeLa cells. This was extended to mouse cells using a different mAb, with specificity control being cGAS KO cells. Localization was verified with western blot analysis of nuclear and cytosolic extracts. Nuclear localization was independent of cGAS activation or cell cycle. Additional data support strong association that is salt dependent and the N-terminus is dispensable.

The data for nuclear localization are clear with excellent specificity controls. While it cannot be ruled out that there is a small amount of cGAS in the cytosol that mediates activation, the work certainly indicates that current models of cytosolic nucleic acid sensing are either incomplete and missing a step before cGAS becomes localized to the cytosol, or perhaps more likely, that it mediates sensing while in the cytosol itself. In either case, the paper raises important issues, even though it does not clearly address these latter points.

We thank the reviewer for the assessment of the manuscript, and we agree that the resting localization of endogenous cGAS needs to be taken into account when considering its mechanism of activation. Our study provides clear evidence that the cytosolic DNA sensing model is insufficient to explain the transition between resting and activated cGAS.

Reviewer #2:

[…]

Specific points:

1) Conceptual advancement over previous studies: The main message of this study i.e. the presence of cGAS in the nucleus has already been established by several previous studies (e.g. Orzalli et al., 2015;, Yang et al., 2017; Lahaye et al., 2018).

We thank the reviewer for highlighting these papers that we have already cited in our submitted manuscript, together with the Funabiki bioRxiv preprint that we did not cite because it is not yet a peer-reviewed publication. None of these papers, nor any of the published studies on cGAS and micronuclei, suggest that cGAS is predominantly and constitutively a nuclear protein as we show here. To summarize their findings relative to what we show in our manuscript.

Orzalli MH et al:

* Importantly, in data that were in Figure 4H of the original manuscript and are now included as supplemental data, we show that IFI16 plays no role in cGAS tethering.

Yang et al:

*Our data clearly demonstrate that the great majority of endogenous cGAS is nuclear, in all cells tested, and independent of mitosis. If the model that cGAS redistributes to the cytosol during interphase were true, then arresting cells in interphase should cause cGAS relocalization. In Figure 3 of our manuscript, we show that cells arrested for three days in interphase do not relocalize endogenous cGAS to the cytosol.

Lahaye et al:

*In this study, a fraction of cGAS was found to be in the nucleus, although the authors only sampled the low salt nuclear extract and did not identify the salt-resistant pool of endogenous cGAS that comprises the majority of nuclear cGAS across nine different cell types. The authors also presented evidence that NONO is required for the nuclear localization of this low salt-extractable cGAS. We now present data in a new supplemental figure showing endogenous cGAS localization in four independent clonal lines of NONO-deficient HeLa cells. NONO deficiency does not influence the tight tethering of cGAS in the nucleus. We acknowledge in the manuscript that differences in cell types could influence the importance of NONO in controlling cGAS nuclear localization, but we emphasize that NONO is not required for nuclear tethering of cGAS.

Zierhut and Funabiki:

*This manuscript follows the Yang et al. proposal that cGAS is cytosolic, associates with mitotic chromatin, and then redistributes to the cytosol during the next interphase. We elaborate on the data in this manuscript in point 2 below.

2) The proposal that the inactivity of nuclear cGAS is due to the tight tethering chromatin is interesting. However, this idea was not experimentally tested in this study. Besides, this idea is really not new and has been tested more rigorously in studies already in the public domain (Zierhut and Funabiki, 2017, bioRxiv, doi: 10.1101/168070).

In the Funabiki manuscript, the authors incubate recombinant cGAS with purified nucleosomes. They show, interestingly, that recombinant cGAS binds to these purified nucleosomes slightly more tightly than it binds to naked DNA. However, none of these studies were done with endogenous cGAS in actual cells. Instead, they were done in a test tube, with salt and solute concentrations that do not equate to those inside live cells. We propose that our analyses of endogenous cGAS in eight different cell lines are relevant, they are rigorous, and they reveal new and interesting biology. Our future goals are to define the nature of the tether, perturb it, and examine the consequences for cGAS localization and activation. However, these experiments are outside the scope of our current manuscript that describes cGAS nuclear tethering for the first time.

3) The authors claim that tethering of cGAS to chromatin is independent of cGAS features important for its activation is not convincing. The functional features of cGAS required for its activation are mainly its DNA binding, enzymatic and dimerization activities. The functional behaviour of the cGAS truncations mutants cannot be assumed to be similar to that within the functional full-length protein. The structural-functional features of cGAS are well defined and a more convincing approach would be for the authors use cGAS point mutants defective in these features. It is really difficult to make any firm conclusion from the truncation mutants since essential controls are lacking. For example, what is the affinity of the different truncation mutants to DNA?

4) The K395M/K399M mutant of mouse cGAS used in Figure 4F is an inactive mutant. Is this because of defect in DNA binding mutant or dimerization? If author want to use enzyme inactive mutant of cGAS, the well characterized 225A/D227A mutant of human cGAS which is more appropriate.

There are a number of important points here and we welcome the opportunity to address them with new data.

First, defined truncation mutants are a well-established approach to explore isolated regions and domains of proteins required for a specific function or interaction. We used this approach to define the region(s) of cGAS that are required to mediate its tight tethering in the nucleus. Notably, we clearly show that the unstructured N-terminus, which is essential for DNA-induced condensation and cGAS activation, is completely dispensable for cGAS nuclear localization and tethering.

Second, the K395M/K399M murine cGAS is the corresponding mutant to the K407A/411A mutant of human cGAS described in Civril et al., 2013. This mutant disrupts key lysines in the DNA-binding platform, its DNA binding is reduced but not absent, and it is completely defective for cGAMP production, as shown in Figure 4F of our revised manuscript and in their original description of the mutant.

Third, we now include five additional point mutants in Figure 4 of the revised manuscript to test DNA binding and DNA-induced dimerization/oligomerization. We cloned and validated these mutants, transudced them into human fibroblasts using the dox-inducible lentivirus system, induced with dox, measured resting and activated cGAMP production, and determined nuclear localization and tethering. These experiments with defined mutants provide compelling evidence that the mechanism of cGAS activation by DNA is distinct from the mechanism of cGAS tethering. We cannot claim that cGAS tethering is completely independent of DNA binding; such a claim would be impossible to test because proving that a particular mutant of cGAS is completely inert for DNA binding is not currently possible.

5) The authors speculate that DNA is not the primary tether of cGAS but instead propose some protein component(s) of higher order chromatin. What is the experimental basis for this this proposal? In fact, their preliminary findings with the nuclease digestions suggest implicate DNA.

In the Discussion section of the manuscript, we attempt to reconcile the known weak affinity of cGAS for DNA with the remarkably tight tethering of nuclear cGAS. The new mutants suggested by the reviewer and discussed in point 4 above make the question more relevant since we use defined, biochemically characterized mutants that are defective for DNA binding and DNA-induced oligomerization and show that cGAS remains tethered. We suggest that DNA binding alone cannot explain the tight tethering. Moreover, we offer the logical consideration that if nuclear cGAS were directly bound to DNA, two problems arise. First, how would DNA-bound cGAS remain inactive? Second, how would cGAS that is saturated with self DNA be competent to respond to foreign DNA? Such a model would require titration of cGAS off of self DNA and onto foreign DNA, at which point it would become active. We speculate that non-DNA components of higher order chromatin are essential for cGAS tethering. We are not attempting to prove that with the data in our paper; instead, we are proposing a model that can explain both the known properties of cGAS and our new findings.

6) The authors claim that nuclear localization of cGAS is via active translocation rather than via passive mechanisms. While interesting, this proposal is also preliminary and require rigorous experimental controls. At the very minimums they should test whether blocking inhibitors of protein nuclear import prevent nuclear localization of cGAS.

This is a good suggestion. We attempted to use two different inhibitors of nuclear import: ivermectin, which inhibits importin α/β-dependent nuclear import (Wagstaff et al., Biochem J, 2012), as well as importazole, which inhibits importin β-dependent nuclear import (Soderholm et al., ACS Chem Biol, 2011). These inhibitors only block a fraction of nuclear import mechanisms, and there are numerous pathways of nuclear import for which there are no adequate inhibitors. We found that neither of these drugs, alone or in combination, prevented cGAS nuclear localization in arrested cells, and that they were both extremely toxic to the cells. We modify the description of these data to state only that cGAS nuclear localization is more complicated than simple passive association with chromatin during mitosis, and we remove the statement suggesting active import. We emphasize that our cell cycle analyses are the first to rigorously test the prevailing model that cGAS is cytosolic during interphase.

7) Many of the experiments lack essential controls. For example, without loading controls for nuclear and cytosolic markers in the immunoblots in Figure 3, it difficult to interrogate the veracity of their claims.

We apologize for not including these loading controls in the original figure, and this is an important point because we switch from the sequential salt washes in Figure 2 to the three-fraction separation in Figure 3. We now include loading controls monitoring cytosol, nuclear supernatant, and nuclear pellets, the same proteins that are followed as controls for Figure 2; this allows more specific comparison of the two approaches.

[Editors' note: the author responses to the re-review follow.]

Essential revisions:

1) A limitation of the paper is that the fractionation experiments are potentially susceptible to an artifact, which is that cGAS might associate with some dense structure that co-sediments with the nucleus, but is not actually itself nuclear. The potential for the fractionation experiments to be misleading means that the complementary immunofluorescence experiments (shown in Figure 1) are of critical importance.

We emphasize that among the recently published studies evaluating the cellular distribution of cGAS, our study is the only one in which the microscopy of endogenous cGAS agrees perfectly with the biochemical fractionations. We have presented microscopy data for endogenous cGAS in two cell types with two distinct fixation methods. We have performed biochemical analysis of endogenous cGAS in six different cell types from mouse and human, and we have employed seven defined cGAS mutants to test whether nuclear localization and tight tethering require robust DNA binding, DNA-activated dimerization, ‘phase separation,’ and catalytic activity.

One main worry about the data in Figure 1 is that because the nucleus is thicker than the rest of the cell, it may appear that cGAS is concentrated in the nucleus simply because the confocal slices sectioned primarily through the nucleus, and missed the cytosol at the periphery of the cell; or, if a z-series projection of all slices is shown, the fact that there would be more slices through the nucleus might improve the signal for this part of the cell as compared to the cytosol. It is hard to tell exactly what was done because the Materials and methods and legends don't fully describe the imaging and analysis. This part of the paper should therefore be strengthened with some additional methodological description and controls. For example, was a Z-series collected? What step size? Are projections shown? Can a 3D reconstruction (x-z projection) of the cell be shown to confirm that the cGAS is truly nuclear?

We have updated the Materials and methods to describe in more detail the confocal microscopy and the z-series step size. Additionally, we have added two videos to Figure 1: a combined rendering of all the z-stacks through an image, and a 3D rendering of the merged z-stacks. Together, these data clearly demonstrate that the great majority of endogenous cGAS is in the nucleus.

What is the volume of the nucleus versus the volume of the cytosol and is the apparent localization in the nucleus merely because the nucleus contains most of the cell volume? Can some kind of quantitation be performed on the images in Figure 1 (cytosol vs. nucleus)?

Volumetric measurements of the entire nucleus and cytosol are outside the scope of this manuscript and require more sophisticated tools and analysis that we do not currently use. Importantly, the biochemical analysis of cGAS and direct quantitation of cGAS abundance by densitometry in Figure 2B offer a quantitative comparison of the amount of cGAS in the cytosol versus the nucleus. Again, we emphasize that the microscopy of endogenous cGAS agrees perfectly with the biochemistry.

Can a control cytosolic protein be shown to convince the reader that were cGAS cytosolic it would be detected?

We have now included a three-color image of endogenous cGAS, tubulin, and DAPI in Figure 1A, together with z-series and 3D renderings as supplemental movies. This demonstrates that an abundant cytosolic protein is preserved and detected with our fixation and staining methods.

Is it surprising that the DAPI and cGAS images appear so similar? Why doesn't the so-called micronucleus in Figure 1B stain with DAPI (maybe it is not a micronucleus)?

We realized that the images presented in Figure 1 were dimmer than ideal in the rendered pdf, so we re-analyzed the images to uniformly brighten the cGAS and DAPI fluorescence. The micronucleus-like structure in Figure 1B is indeed DAPI positive, and the brightened images now show this clearly.

2) One of the major conclusion of this study is that nuclear cGAS localization is not via passive association with chromatin (as reported by others e.g. Yang et al., 2017, Gentili et al., 2019) but via a more specific mechanism. The subcellular fractionation and mutant experiments are not sufficient to justify this conclusion. For example, in Figure 3F, in the absence of mitotic nuclear membrane dissolution, how can the presence of control GFP in their nuclear soluble (NS) and nuclear pellet (NP) fraction be explained? Due to passive or active import? Or could it be post-lysis contamination of such fractions with cytosolic components? Of course, GFP-cGAS is more abundant that control GFP in the NP fraction. However, this could simply be due to the ability of GFP-cGAS and not the GFP control to bind to DNA hence its retention therein.

GFP, because of its size (27 kD), is very well known to be able to enter the nucleus through passive diffusion through the nuclear pore, without any requirement for energy-dependent nuclear import (Timney et al., J. Cell Biol., 2016). The distribution of GFP alone compared GFP-cGAS in Figure 3D is dramatically different. While there is a very small amount of GFP remaining in the nuclear pellet, the localization of the great majority of GFP-cGAS to the pellet and its absence from the low salt nuclear supernatant are the most important points here. Again, the fractionation of GFP-cGAS agrees with the fractionation of endogenous cGAS.

Therefore, similar to the above point, can the authors induce GFP-cGAS expression in cells arrested at G0 or G1 and visualize its localization by life confocal imaging to see distinctly that freshly synthesized cGAS does indeed translocate into the nucleus independent of mitotic nuclear membrane dissolution?

Live confocal imaging of GFP-cGAS requires tools and analyses that we do not currently have, and we do not think that they would add substantively to the data presented throughout the manuscript. In Figure 3C-D, we arrest cells at the G1/S border, induce cGAS with dox, and show that it localizes to the nucleus identically to the localization in cycling cells. Importantly, this experiment definitively demonstrates that mitosis is not required for cGAS nuclear localization. The next question of precisely how cGAS enters the nucleus is one that will require more work, but it is not relevant for the conclusion that we draw based on the data in the manuscript. We have changed the last sentence of the paragraph describing Figure 3 to focus on this important point and to not speculate further.

Original sentence: “These data demonstrate that cGAS nuclear localization is not mediated by its passive association with chromatin during mitosis, and they suggest a more specific mechanism by which cGAS enters the nucleus.”

Revised sentence: “These data demonstrate that mitosis is not required for cGAS nuclear localization.”

In view of these caveats, the authors’ own admission that the nuclear import inhibitors tested did not impede nuclear localization (surprisingly not mentioned this in the text), the claim that nuclear cGAS is via active translocation remains very weak and is potentially misleading.

The nuclear import inhibitors that are commercially available only block a fraction of specific modes of nuclear import, whereas some forms of nuclear import cannot be chemically inhibited.

Moreover, given the sensitivity of nuclear pellet-associated cGAS to nucleases (Figure 5) and since none of the cGAS mutants tested completely abrogates DNA binding, it is in simply not possible to discount the role genomic DNA in nuclear localization. Therefore, in the absence of direct evidence the authors should mellow down on this claim.

We tested the contributions of robust DNA binding, DNA-activated dimerization, and DNA-activated ‘’phase-separation” using structure-guided mutants that all dramatically reduce the DNA-mediated activation of full-length cGAS in live cells. All of these mutants remain nuclear and remain tethered. We agree that none of the cGAS mutants tested (nor any that have ever been made) completely abrogate DNA binding. Importantly, the ability of cGAS or its mutants to bind DNA has always been measured with truncated, recombinant cGAS lacking its N terminus, under salt and solute conditions that do not adequately resemble the cellular environment, and with short double-stranded synthetic DNAs that do not fully recapitulate the size and disposition of DNA within cells. We feel that our data, with defined mutants, support the conclusions that we draw and our interpretation of the findings.

In the Discussion section of the manuscript, we attempt to rationalize our findings using seven defined cGAS mutants, together with the fact that chromatin is required for cGAS tethering. We interpret our data and offer a proposal based on our novel findings, with a very important logical consideration that has not been discussed in the literature. We feel that this is an extremely important point, that we have not over-interpreted our data, that it is appropriate for inclusion in the Discussion section, and that it is worth presenting because it offers a new model for cGAS regulation. To emphasize that it is impossible to completely rule out DNA binding, we have re-worded this section of the Discussion:

“While we cannot completely rule out a role for DNA binding in cGAS nuclear tethering because no known cGAS mutant is completely devoid of DNA binding, we speculate that DNA itself is not the primary tether of cGAS. […] Based on these considerations, we speculate that nuclear cGAS does not directly interact with genomic DNA, and that the tethering mechanism maintains resting cGAS in a state that is competent to detect foreign DNA.”

[Editors' note: further revisions were requested prior to acceptance, as described below.]

As you are aware, cGAS nuclear localization has been reported across many publications already. Moreover, avid binding to chromatin as a possible mechanism for preventing cGAS-mediated immune response to nuclear DNA – one of the main points of the current manuscript, has also been described (Zierhut et al., Cell, 2019, formerly a preprint on bioRxiv). Therefore, in view of the current state of the field, the potentially new aspect of this manuscript is how cGAS is localized to the nucleus. Your theory that cGAS nuclear localization is via active transport from the cytosol advances an interesting concept with major implications in the field since this could provide new avenues for manipulating cGAS-mediated biological processes. On the other hand, if incorrect, this idea can mislead the field. As you have admitted in response to our earlier request, your nuclear import inhibitors experiments so far do not support your theory (these are relevant data and you may consider to include them in the manuscript). Although you have toned down on your initial conclusion, it remains unclear whether nuclear localization is indeed via active or non-specific mechanisms. Therefore, at a minimum, the suggested live confocal imaging comparing the localization of inducible GFP-cGAS is still essential. This experiment is feasible and requires standard techniques that should be available in your institution or can be arranged through collaboration.

We previously brought to your attention the fact that GFP control alone does localize to the nucleus (Figure 3D). We have noted your explanation that such localization is likely because of passive translocation of GFP due to its size (27 kD). In this regard, it is equally important to note that passive diffusion across the nuclear envelope does not have fixed molecular mass threshold; proteins of molecular mass of up to 200 kD can diffuse across the nuclear envelope with varying kinetics (Timney et al., J. Cell Biol., 2016). Therefore, an alternative explanation perhaps worth considering is that nuclear cGAS localization is simply via passive diffusion. That is why live confocal imaging comparing the localization of inducible GFP-cGAS and a GFP-tagged control protein with similar molecular mass as cGAS would help to shed some light onto whether cGAS nuclear localization is indeed due to active or passive mechanisms. Even if the conclusion from this experiment were that localization is via passive mechanism, this would still be an important contribution and would provide more clarity to the field.

We thank the reviewers and the editors for their comments. At issue throughout this revision and resubmission process is a potential “mechanism” for whether, how, and when cGAS enters the nucleus. These experiments are asked for in the context of questions over whether our findings are real or an elaborate artifact, because cytosolic DNA sensing is the foundational framework of the field. It has been suggested that our fixation conditions for microscopy might result in the loss of a relevant pool of cytosolic cGAS and an overestimate of the amount of nuclear cGAS (Barnett et al., 2019). Moreover, it has also been suggested that our extraction conditions might cause cytosolic cGAS to artifactually condense on DNA liberated during extraction, which would make it co-sediment with the nuclei during centrifugation that separates cytosol from nucleus (Barnett et al., 2019). We had addressed both of these concerns with experiments in our prior revised manuscript, but they still remain. Finally, although we were the first to describe tight nuclear tethering of cGAS, its relevance remained unknown and was only speculated on in the Discussion section of our last submission.

Since the original submission of our paper to eLife in December 2018, four new reports have been published that address cGAS localization and its relevance, each arriving at distinct conclusions.

1) Barnett et al., Cell 2019 (PMID 30827685) suggest that cGAS is predominantly a cytosolic protein anchored to the plasma membrane via interactions between its N-terminus and PIP2.

2) Zierhut et al., Cell 2019 (PMID 31299200) suggest that cGAS is cytosolic but associates with mitotic chromatin after nuclear envelope breakdown. They suggest that purified nucleosomes mildly inhibit activation of truncated cGAS (lacking its N-terminus) in a test tube but they do not demonstrate anything about tethering in cells.

3) Gentili et al., Cell Reports 2019 (PMID 30917330) suggest that cGAS localizes to specific regions of the genome, and that its N terminus is important for nuclear localization.

4) Jiang et al., EMBO 2019 (PMID 31544964) suggest that cGAS is bound to chromatin through its DNA-binding residues, and that cGAS condensation onto genomic DNA inhibits DNA repair.

In new data presented in a new Figure 6, we have identified the surface of cGAS that is responsible for its tight nuclear tethering. This surface, and key amino acids within it, are conserved in every known vertebrate cGAS protein spanning 350 million years of evolution. We mutated single conserved amino acids of cGAS within this surface. This results in untethering and elution of cGAS in low salt, but the mutant cGAS proteins remain predominantly nuclear. Remarkably, these single amino acid mutations cause massive, constitutive, DNA-dependent activation of cGAS. Some of these mutant cGAS enzymes are so active upon expression alone that they cannot produce any more cGAMP in response to DNA transfection. Thus, untethered cGAS assembles on and is saturated by self-DNA, resulting in maximal activation. Finally, the location of this tethering surface, which is in the NTase domain (not the N-terminus), is such that binding of cGAS to a tethering protein would be incompatible with DNA binding. Thus, tethering and DNA-based activation of cGAS are mutually exclusive states.

By identifying the tethering surface of cGAS, disrupting it, and observing the consequences of untethering, we:

– prove that cGAS is indeed predominantly a nuclear protein that is tightly tethered to chromatin (this has been questioned as an artifact of fixation or lysis conditions).

– corroborate all of our mutant data in Figure 4 showing that DNA binding, oligomerization, condensation, and catalysis are all dispensable for nuclear localization and tethering.

– demonstrate that other recent studies suggesting that the N-terminus of cGAS is required for cGAS localization to either the plasma membrane (Kagan) or nucleus (Manel) are incorrect.

– reveal that tethering physically sequesters cGAS from genomic DNA in the nucleus, distinct from recently proposed models that invoke DNA binding as the mechanism of nuclear localization (e.g. Gekara, Funabiki, Manel).

– show that tethering is essential for preventing cGAS assembly on self-DNA.

– reveal that in order for cGAS to be activated, tethering must be overcome. Thus, cGAS is carefully regulated prior to its encounter with DNA, and there must be a regulated step to permit cGAS activation by foreign DNA.

– identify a fundamental and evolutionarily conserved mechanism that underlies self/non-self discrimination by cGAS.

We feel that these new and exciting findings not only corroborate the existing data in our paper, but they also reveal a fundamental new mechanism for preventing nuclear cGAS from responding to nuclear genomic DNA. As such, we think that these data also obviate any lingering concerns on the part of the reviewers and editors over whether nuclear tethering is real and relevant.

We include several additional pieces of data in order to address any residual concerns over our microscopy images:

1) New microscopy images in which we compare methanol fixation to paraformaldehyde fixation for visualization of endogenous nuclear cGAS.

2) Imaging flow cytometry analysis in which we quantitate nuclear versus cytosolic GFP-cGAS in thousands of live, unfixed cells.

[Editors’ note: the author responses to the final round of editors’ comments follow.]

This work was originally submitted under the title "Endogenous cGAS is predominantly a nuclear protein". The initial conclusion of the study were that: (1) cGAS is predominantly a nuclear protein, (2) that nuclear localization of cGAS is independent of DNA binding or cell cycle as previously reported by others, and (3) that this localization is due to active importation from the cytosol.

We are unaware of any papers, published prior to our submission on December 3, 2018, that have demonstrated that cGAS nuclear localization is independent of DNA binding or cell cycle, but would welcome any references we may have overlooked.

The outstanding concerns throughout the previous rounds of revisions were that their data were inconsistent with the conclusion regarding mechanisms of nuclear localization and association with the chromatin.

In the previous decision letter, we requested one essential control experiment which in my opinion was feasible and hence the quickest part to publication. Specifically, to verify their claim that nuclear localization was via active mechanisms, we requested for life-microscopic images of their inducible cGAS. The authors did not however provide this control experiment but have left out the data that we previously pointed out to be inconsistent with their conclusion (e.g. previously in Figure 3D). They have also included new data and have rewritten the manuscript in such a manner that the main message is now very different from the original submission.

Two important points here. First, the main point of our manuscript is, and always has been, that cGAS is a tightly tethered nuclear protein. When we submitted these findings to eLife on December 3, 2018, this was absolutely a novel finding. In our subsequent revisions, we included fifteen additional cGAS mutants to support this, all of which are completely congruent with our main point.

Second, we agreed with the prior reviewer's assessment that our claim that cGAS import was an active process was not fully supported by our data. We removed this from the revised manuscript because the experimentation required to adequately test it would not be incisive or quantitative. Importantly, chemical inhibitors that block all forms of active nuclear import do not exist.

Indeed, the question of how cGAS gets into the nucleus is tangential to the important and novel observation that cGAS is in fact predominantly nuclear, and that it is tightly tethered in the nucleus. For this reason, we decided to include the mechanistic data that unequivocally define the tethering surface of cGAS, together with the first description of the consequences of untethering. These data absolutely support the original message of the manuscript and do not change the original point that cGAS is a tightly tethered nuclear protein. In fact, the new data demonstrate the actual importance of tight nuclear tethering. We believe that proving the "mechanism" of cGAS nuclear import is beyond the scope of this manuscript, and we instead defined the cGAS nuclear tethering surface and its essential role in preventing cGAS autoreactivity. In addition, we believe our new data in Figure 6 supports, enriches, and extends the main point of our original manuscript.

In this version, the authors propose that at resting state, cGAS is kept inactive through tight binding (which the authors call tethering) to chromatin and that this tethering is via yet to be identified tether and not DNA. Implicit in this model is that for cGAS activation to occur, cGAS has to undergo untethering to enable it to bind DNA. They claim to have identified a conserved surface on cGAS responsible for this tethering and conclude that this surface is distinct from that for DNA binding. In my assessment the authors have misinterpreted their data and these fresh claims are misleading. In fact, the new data support the opposite of their conclusion: that attachment of cGAS to the chromatin in via DNA.

Specific comments:

1) What the authors conclude as tethering surface lie within the DNA binding surface of cGAS. The R222E, R240E, R241E, R244E mutants that the authors report as tethering defective mutants are in fact DNA binding mutants. This is well established in the field (e.g. Figure 5C and Figure 6 of Li et al., Immunity, 2013). Moreover, the other mutants, for example the K335E, K382A, E386A, K395M, K399M which the authors used to conclude that tethering of cGAS to chromatin is independent of DNA binding have also DNA binding mutants. The main difference these sets of mutants is that the suggested "tethering mutants" (R222E, R240E, R241E, R244E) are severely defective in DNA binding compared to the K335E, K382A, E386A, K395M, K399M mutants which retain substantial DNA binding (Figure 6 of Li et al., Immunity, 2013). Therefore, in my view, the correct interpretation is that interaction of cGAS with chromatin involves some form of interaction with DNA binding. This is consistent with the authors' data showing that cGAS-chromatin interaction is highly sensitive to DNases (Figure 5).

There are some factual errors here:

R222E was defined as a severe DNA binding mutant in the Li et al. paper, but it was also shown to be fully active upon transient transfection in which the plasmid DNA also serves as a potent activating ligand. We explain this clearly and cite the Li paper extensively.

K240E (not R240E) was also defined as a DNA binding mutant in the Li paper but was also paradoxically fully active upon transient overexpression.

R241 is actually not a DNA binding residue. It points away from DNA in the crystal structure, and it was in fact never tested in the Li paper. Thus, R241 is not part of the DNA-binding surface of cGAS.

R2444 is not a "DNA-binding" residue. It is actually unresolved in the crystal structure, and it was also not tested in the Li paper. Perhaps the editor is accidentally referring to the R342E mutant, which is 98 amino acids away from R244.

As such, to say that all the mutants that we "report as tethering defective mutants are in fact DNA binding mutants" is incorrect, and we hope the editor will reconsider this point upon revisiting Li et al.

2) Related to the above point, in my view, the more plausible explanation for the spontaneous activity of the R222E, R240E, R241E, R244E mutants is that mutations in these amino acids likely triggers a conformational change lowering the threshold of activation by DNA. And, of course the compound mutations in R222E together with either K335E, K382A or K395M is expected to result in an inactive mutant (Figure 6A, Figure 4F), since the latter mutations are in themselves inactivating mutation (Li et al., 2013).

R222E and K240E were defined as DNA binding mutants in a test tube, with recombinant cGAS lacking its N terminus, under salt conditions that do not resemble those in cells. If these are severe DNA binding mutants, why are they massively and constitutively activated in a DNA-dependent manner? Instead, R222E, K240E, R241E and the additional mutants that we define are constitutively active. The well-defined DNA binding residues that compromise cGAS function when mutated show that the constitutive activation of R222E is DNA-dependent. Importantly, as shown in Figure 4, K335E, K395M/399M, zinc thumbless, K382A, E386A are all still tethered in the nucleus. This is amply supported by the seventeen distinct mutants we describe, all of which have been added in our revised submissions to put our model to a rigorous test.

3) The manuscript contains statements that are potentially misleading to the readers. For example, in the Introduction the authors state "Here, we use confocal microscopy and biochemical characterization to determine the resting localization of endogenous cGAS prior to activation". In the Discussion the authors go on to conclude, "We show, using microscopy and biochemical fractionation, that the great majority of endogenous cGAS is nuclear prior to its activation." Inherent in this statement is that in resting state cGAS is kept inactive through interaction with chromatin but then undergoes spatial redistribution upon activation. There is no evidence that this is the case.

With five mouse cGAS mutants and two human cGAS mutants, we show that there is a dramatic redistribution of constitutively active cGAS from high salt fractions to low salt fractions. The DNA-dependent activation of these constitutive mutants is accompanied by a redistribution into lower salt fractions. If DNA is primarily responsible for tightly tethering inactive, resting cGAS, it seems unlikely that the highly constitutive, DNA-dependent, autoreactive cGAS would elute at much lower salt concentrations.

4) The authors proposed tethering models and the Discussion largely assume that all/or most of the cGAS activation occurs in the nucleus, how do the authors explain cGAS activation following DNA transfection or following some bacterial infections – that such sensing also occurs in the nucleus or that this involves redistribution of cGAS to the cytosol? A shortcoming of the proposed model is that it does factor the constant presence of a chromatin-free cGAS and that it is this pool that most likely becomes activated by foreign or misplaced self-DNA?

As Figure 6 demonstrates that the untethered mutants of cGAS remain nuclear and are activated by self-DNA, we hope it is clear that the cytosol is not the only place that cGAS can be activated by DNA.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Transparent reporting form

elife-47491-transrepform.docx^{(248.4KB, docx)}

Data Availability Statement

All data generated or analyzed during this study are included in the manuscript and supporting files.

[bib1] Andreeva L, Hiller B, Kostrewa D, Lässig C, de Oliveira Mann CC, Jan Drexler D, Maiser A, Gaidt M, Leonhardt H, Hornung V, Hopfner KP. cGAS senses long and HMGB/TFAM-bound U-turn DNA by forming protein-DNA ladders. Nature. 2017;549:394–398. doi: 10.1038/nature23890. [DOI] [PubMed] [Google Scholar]

[bib2] Barnett KC, Coronas-Serna JM, Zhou W, Ernandes MJ, Cao A, Kranzusch PJ, Kagan JC. Phosphoinositide interactions position cGAS at the plasma membrane to ensure efficient distinction between self- and viral DNA. Cell. 2019;176:1432–1446. doi: 10.1016/j.cell.2019.01.049. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib3] Bartsch K, Knittler K, Borowski C, Rudnik S, Damme M, Aden K, Spehlmann ME, Frey N, Saftig P, Chalaris A, Rabe B. Absence of RNase H2 triggers generation of immunogenic micronuclei removed by autophagy. Human Molecular Genetics. 2017;26:3960–3972. doi: 10.1093/hmg/ddx283. [DOI] [PubMed] [Google Scholar]

[bib4] Bootsma D, Budke L, Vos O. Studies on synchronous division of tissue culture cells initiated by excess thymidine. Experimental Cell Research. 1964;33:301–309. doi: 10.1016/S0014-4827(64)81035-1. [DOI] [PubMed] [Google Scholar]

[bib5] Civril F, Deimling T, de Oliveira Mann CC, Ablasser A, Moldt M, Witte G, Hornung V, Hopfner KP. Structural mechanism of cytosolic DNA sensing by cGAS. Nature. 2013;498:332–337. doi: 10.1038/nature12305. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib6] Crowl JT, Gray EE, Pestal K, Volkman HE, Stetson DB. Intracellular nucleic acid detection in autoimmunity. Annual Review of Immunology. 2017;35:313–336. doi: 10.1146/annurev-immunol-051116-052331. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] Dou Z, Ghosh K, Vizioli MG, Zhu J, Sen P, Wangensteen KJ, Simithy J, Lan Y, Lin Y, Zhou Z, Capell BC, Xu C, Xu M, Kieckhaefer JE, Jiang T, Shoshkes-Carmel M, Tanim K, Barber GN, Seykora JT, Millar SE, Kaestner KH, Garcia BA, Adams PD, Berger SL. Cytoplasmic chromatin triggers inflammation in senescence and Cancer. Nature. 2017;550:402–406. doi: 10.1038/nature24050. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib8] Du M, Chen ZJ. DNA-induced liquid phase condensation of cGAS activates innate immune signaling. Science. 2018;361:704–709. doi: 10.1126/science.aat1022. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Gao D, Wu J, Wu YT, Du F, Aroh C, Yan N, Sun L, Chen ZJ. Cyclic GMP-AMP synthase is an innate immune sensor of HIV and other retroviruses. Science. 2013;341:903–906. doi: 10.1126/science.1240933. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib10] Gentili M, Lahaye X, Nadalin F, Nader GPF, Lombardi EP, Herve S, De Silva NS, Rookhuizen DC, Zueva E, Goudot C, Maurin M, Bochnakian A, Amigorena S, Piel M, Fachinetti D, Londoño-Vallejo A, Manel N. The N-Terminal domain of cGAS determines preferential association with centromeric DNA and innate immune activation in the nucleus. Cell Reports. 2019;26:3798. doi: 10.1016/j.celrep.2019.03.049. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] Glück S, Guey B, Gulen MF, Wolter K, Kang TW, Schmacke NA, Bridgeman A, Rehwinkel J, Zender L, Ablasser A. Innate immune sensing of cytosolic chromatin fragments through cGAS promotes senescence. Nature Cell Biology. 2017;19:1061–1070. doi: 10.1038/ncb3586. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] Goubau D, Deddouche S, Reis e Sousa C. Cytosolic sensing of viruses. Immunity. 2013;38:855–869. doi: 10.1016/j.immuni.2013.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] Gray EE, Treuting PM, Woodward JJ, Stetson DB. Cutting edge: cgas is required for lethal autoimmune disease in the Trex1-Deficient mouse model of Aicardi-Goutières syndrome. The Journal of Immunology. 2015;195:1939–1943. doi: 10.4049/jimmunol.1500969. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Gray EE, Winship D, Snyder JM, Child SJ, Geballe AP, Stetson DB. The AIM2-like receptors are dispensable for the interferon response to intracellular DNA. Immunity. 2016;45:255–266. doi: 10.1016/j.immuni.2016.06.015. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Harding SM, Benci JL, Irianto J, Discher DE, Minn AJ, Greenberg RA. Mitotic progression following DNA damage enables pattern recognition within micronuclei. Nature. 2017;548:466–470. doi: 10.1038/nature23470. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Ibsen MS, Gad HH, Andersen LL, Hornung V, Julkunen I, Sarkar SN, Hartmann R. Structural and functional analysis reveals that human OASL binds dsRNA to enhance RIG-I signaling. Nucleic Acids Research. 2015;43:5236–5248. doi: 10.1093/nar/gkv389. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib17] Ishikawa H, Ma Z, Barber GN. STING regulates intracellular DNA-mediated, type I interferon-dependent innate immunity. Nature. 2009;461:788–792. doi: 10.1038/nature08476. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Jiang H, Xue X, Panda S, Kawale A, Hooy RM, Liang F, Sohn J, Sung P, Gekara NO. Chromatin-bound cGAS is an inhibitor of DNA repair and hence accelerates genome destabilization and cell death. The EMBO Journal. 2019;38:e102718. doi: 10.15252/embj.2019102718. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] Lahaye X, Satoh T, Gentili M, Cerboni S, Conrad C, Hurbain I, El Marjou A, Lacabaratz C, Lelièvre J-D, Manel N. The capsids of HIV-1 and HIV-2 determine immune detection of the viral cDNA by the innate sensor cGAS in dendritic cells. Immunity. 2013;39:1132–1142. doi: 10.1016/j.immuni.2013.11.002. [DOI] [PubMed] [Google Scholar]

[bib20] Lahaye X, Gentili M, Silvin A, Conrad C, Picard L, Jouve M, Zueva E, Maurin M, Nadalin F, Knott GJ, Zhao B, Du F, Rio M, Amiel J, Fox AH, Li P, Etienne L, Bond CS, Colleaux L, Manel N. NONO detects the nuclear HIV capsid to promote cGAS-Mediated innate immune activation. Cell. 2018;175:488–501. doi: 10.1016/j.cell.2018.08.062. [DOI] [PubMed] [Google Scholar]

[bib21] Lau L, Gray EE, Brunette RL, Stetson DB. DNA tumor virus oncogenes antagonize the cGAS-STING DNA-sensing pathway. Science. 2015;350:568–571. doi: 10.1126/science.aab3291. [DOI] [PubMed] [Google Scholar]

[bib22] Li X, Shu C, Yi G, Chaton CT, Shelton CL, Diao J, Zuo X, Kao CC, Herr AB, Li P. Cyclic GMP-AMP synthase is activated by double-stranded DNA-induced oligomerization. Immunity. 2013;39:1019–1031. doi: 10.1016/j.immuni.2013.10.019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Li T, Chen ZJ. The cGAS-cGAMP-STING pathway connects DNA damage to inflammation, senescence, and Cancer. The Journal of Experimental Medicine. 2018;215:1287–1299. doi: 10.1084/jem.20180139. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib24] Liu S, Cai X, Wu J, Cong Q, Chen X, Li T, Du F, Ren J, Wu YT, Grishin NV, Chen ZJ. Phosphorylation of innate immune adaptor proteins MAVS, STING, and TRIF induces IRF3 activation. Science. 2015;347:aaa2630. doi: 10.1126/science.aaa2630. [DOI] [PubMed] [Google Scholar]

[bib25] Liu H, Zhang H, Wu X, Ma D, Wu J, Wang L, Jiang Y, Fei Y, Zhu C, Tan R, Jungblut P, Pei G, Dorhoi A, Yan Q, Zhang F, Zheng R, Liu S, Liang H, Liu Z, Yang H, Chen J, Wang P, Tang T, Peng W, Hu Z, Xu Z, Huang X, Wang J, Li H, Zhou Y, Liu F, Yan D, Kaufmann SHE, Chen C, Mao Z, Ge B. Nuclear cGAS suppresses DNA repair and promotes tumorigenesis. Nature. 2018;563:131–136. doi: 10.1038/s41586-018-0629-6. [DOI] [PubMed] [Google Scholar]

[bib26] Luo Y, Tucker SC, Casadevall A. Fc- and complement-receptor activation stimulates cell cycle progression of macrophage cells from G1 to S. The Journal of Immunology. 2005;174:7226–7233. doi: 10.4049/jimmunol.174.11.7226. [DOI] [PubMed] [Google Scholar]

[bib27] Ma Z, Jacobs SR, West JA, Stopford C, Zhang Z, Davis Z, Barber GN, Glaunsinger BA, Dittmer DP, Damania B. Modulation of the cGAS-STING DNA sensing pathway by gammaherpesviruses. PNAS. 2015;112:E4306–E4315. doi: 10.1073/pnas.1503831112. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] Maciejowski J, Li Y, Bosco N, Campbell PJ, de Lange T. Chromothripsis and Kataegis induced by telomere crisis. Cell. 2015;163:1641–1654. doi: 10.1016/j.cell.2015.11.054. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Mackenzie KJ, Carroll P, Martin CA, Murina O, Fluteau A, Simpson DJ, Olova N, Sutcliffe H, Rainger JK, Leitch A, Osborn RT, Wheeler AP, Nowotny M, Gilbert N, Chandra T, Reijns MAM, Jackson AP. cGAS surveillance of micronuclei links genome instability to innate immunity. Nature. 2017;548:461–465. doi: 10.1038/nature23449. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] Orzalli MH, Broekema NM, Diner BA, Hancks DC, Elde NC, Cristea IM, Knipe DM. cGAS-mediated stabilization of IFI16 promotes innate signaling during herpes simplex virus infection. PNAS. 2015;112:E1773–E1781. doi: 10.1073/pnas.1424637112. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] Rasaiyaah J, Tan CP, Fletcher AJ, Price AJ, Blondeau C, Hilditch L, Jacques DA, Selwood DL, James LC, Noursadeghi M, Towers GJ. HIV-1 evades innate immune recognition through specific cofactor recruitment. Nature. 2013;503:402–405. doi: 10.1038/nature12769. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] Sanjana NE, Shalem O, Zhang F. Improved vectors and genome-wide libraries for CRISPR screening. Nature Methods. 2014;11:783–784. doi: 10.1038/nmeth.3047. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib33] Shechter D, Dormann HL, Allis CD, Hake SB. Extraction, purification and analysis of histones. Nature Protocols. 2007;2:1445–1457. doi: 10.1038/nprot.2007.202. [DOI] [PubMed] [Google Scholar]

[bib34] Stetson DB, Medzhitov R. Recognition of cytosolic DNA activates an IRF3-dependent innate immune response. Immunity. 2006;24:93–103. doi: 10.1016/j.immuni.2005.12.003. [DOI] [PubMed] [Google Scholar]

[bib35] Sun L, Wu J, Du F, Chen X, Chen ZJ. Cyclic GMP-AMP synthase is a cytosolic DNA sensor that activates the type I interferon pathway. Science. 2013;339:786–791. doi: 10.1126/science.1232458. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Wu J, Sun L, Chen X, Du F, Shi H, Chen C, Chen ZJ. Cyclic GMP-AMP is an endogenous second messenger in innate immune signaling by cytosolic DNA. Science. 2013;339:826–830. doi: 10.1126/science.1229963. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] Wu JJ, Li W, Shao Y, Avey D, Fu B, Gillen J, Hand T, Ma S, Liu X, Miley W, Konrad A, Neipel F, Stürzl M, Whitby D, Li H, Zhu F. Inhibition of cGAS DNA sensing by a herpesvirus virion protein. Cell Host & Microbe. 2015;18:333–344. doi: 10.1016/j.chom.2015.07.015. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib38] Xia P, Wang S, Ye B, Du Y, Li C, Xiong Z, Qu Y, Fan Z. A circular RNA protects dormant hematopoietic stem cells from DNA sensor cGAS-Mediated exhaustion. Immunity. 2018;48:688–701. doi: 10.1016/j.immuni.2018.03.016. [DOI] [PubMed] [Google Scholar]

[bib39] Yang H, Wang H, Ren J, Chen Q, Chen ZJ. cGAS is essential for cellular senescence. PNAS. 2017;114:E4612–E4620. doi: 10.1073/pnas.1705499114. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] Zhou W, Whiteley AT, de Oliveira Mann CC, Morehouse BR, Nowak RP, Fischer ES, Gray NS, Mekalanos JJ, Kranzusch PJ. Structure of the human cGAS-DNA complex reveals enhanced control of immune surveillance. Cell. 2018;174:300–311. doi: 10.1016/j.cell.2018.06.026. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] Zierhut C, Yamaguchi N, Paredes M, Luo J-D, Carroll T, Funabiki H. The cytoplasmic DNA sensor cGAS promotes mitotic cell death. Cell. 2019;178:302–315. doi: 10.1016/j.cell.2019.05.035. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Tight nuclear tethering of cGAS is essential for preventing autoreactivity

Hannah E Volkman

Stephanie Cambier

Elizabeth E Gray

Daniel B Stetson

Roles

Abstract

Introduction

Results

Endogenous cGAS is predominantly a nuclear protein

Figure 1. Endogenous cGAS is predominantly a nuclear protein.

Figure 1—figure supplement 1. Characterization of clonal cGAS KO HeLa cells and microscopy conditions for cGAS visualization.

Figure 1—video 1. Compiled z-series confocal images of the HeLa cells visualized in Figure 1A.

Figure 1—video 2. Rendered z-series of HeLa cells rotated around the y-axis.

cGAS is tethered tightly in the nucleus

Figure 2. cGAS is tightly tethered in the nucleus.

cGAS is nuclear regardless of cell cycle phase or activation state

Figure 3. cGAS nuclear localization is independent of cell cycle phase or activation status.

cGAS nuclear tethering and cGAS activation are governed by separate mechanisms

Figure 4. cGAS nuclear localization and tethering are independent of robust DNA binding, dimerization, condensation, and catalytic activity.

Figure 4—figure supplement 1. NONO and IFI16 are dispensable for cGAS nuclear localization and tethering.

Intact chromatin is required for cGAS tethering

Figure 5. Intact chromatin is required for cGAS tethering.

Identification of the evolutionarily conserved cGAS tethering surface

Figure 6. The cGAS tethering surface prevents autoreactivity.

Discussion

Materials and methods

Key resources table.

Cell lines and mice

Immunofluorescence microscopy

Generation of cGAS and NONO knockout HeLa cells

Amnis imagestream analysis

Salt extractions

NE-PERS kit modification

Double Thymidine block

cGAMP quantitation assay

Constructs

Nuclease digestions and salt elutions

Experimental replicates and reproducibility

Acknowledgements

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Additional files

Data availability

References

Decision letter

Roles

Author response

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases