Chromatin-bound cGAS is an inhibitor of DNA repair and hence accelerates genome destabilization and cell death

Abstract

DNA repair via homologous recombination (HR) is indispensable for genome integrity and cell survival but if unrestrained can result in undesired chromosomal rearrangements. The regulatory mechanisms of HR are not fully understood. Cyclic GMP-AMP synthase (cGAS) is best known as a cytosolic innate immune sensor critical for the outcome of infections, inflammatory diseases, and cancer. Here, we report that cGAS is primarily a chromatin-bound protein that inhibits DNA repair by HR, thereby accelerating genome destabilization, micronucleus generation, and cell death under conditions of genomic stress. This function is independent of the canonical STING-dependent innate immune activation and is physiologically relevant for irradiation-induced depletion of bone marrow cells in mice. Mechanistically, we demonstrate that inhibition of HR repair by cGAS is linked to its ability to self-oligomerize, causing compaction of bound template dsDNA into a higher-ordered state less amenable to strand invasion by RAD51-coated ssDNA filaments. This previously unknown role of cGAS has implications for understanding its involvement in genome instability-associated disorders including cancer.

Keywords: cGAS; DNA repair; cancer; cell death; chromatin compaction.

Figure 1. cGAS is constantly present in the cytosol and nucleus and is impacted by cell cycle

1. A–C

   Left: Immunofluorescence images of cGAS in the nucleus (DAPI) and cytosol (phalloidin) in BMDMos cultured at low/high density (A), with/without serum (B), or with/without aphidicolin (Aphi) (C). Scale bar: 50 μm. Right: Corresponding quantification of the nuclear cGAS from 6 different fields with n > 50 cells.

2. D–F

   Immunoblot estimation of cGAS in nuclear/cytosolic subcellular fraction of BMDMos cultured under indicated conditions. Lamin B and α‐tubulin are nuclear and cytosolic markers, respectively.

3. G–I

   Flow cytometric analysis of cell cycle of BMDMos depicted in (D–F).

Data information: Data are presented as means ± SEM. Statistical significance was assessed using unpaired Student's t‐test. ****P ≤ 0.0001.Source data are available online for this figure.

Figure EV1. cGAS is constantly present in the cytosol and nucleus

1. Immunoblot estimation of GFP‐hcGAS in nuclear/cytosolic fractions and corresponding flow cytometric analysis of cell cycle of HEK293 cells cultured in low or high density. Lamin B and α‐tubulin are nuclear and cytosolic markers, respectively.

2. Immunoblot estimation of GFP‐hcGAS in nuclear/cytosolic fractions and corresponding flow cytometric analysis of cell cycle of HEK293 cells cultured with or without serum. Lamin B and α‐tubulin are nuclear and cytosolic markers, respectively.

3. Immunoblot estimation of GFP‐hcGAS in nuclear/cytosolic fractions and corresponding flow cytometric analysis of cell cycle of HEK293 cells cultured with or without aphidicolin. Lamin B and α‐tubulin are nuclear and cytosolic markers, respectively.

4. cGAS in nuclear/cytosolic fractions of indicated cell types.

Source data are available online for this figure.

Figure EV2. Localization and retention of cGAS in the nucleus is due to is avid binding to DNA

1. A

   Fluorescence images of GFP‐hcGAS, GFP‐hcGASΔcGAMP, GFP‐hcGASΔDNA, and GFP‐hcGASΔOligo in HEK293 cells cultured with or without aphidicolin. Scale bar: 20 μm.

2. B

   Corresponding quantification of (A). The nuclear cGAS/total cGAS was calculated from 6 different fields with n > 50 cells.

3. C, D

   A nuclear export signaling (NES) is not sufficient to dislodge chromatin‐bound cGAS from the nucleus. (C) Fluorescence images of GFP‐hcGAS, GFP‐hcGAS‐NLS, and GFP‐hcGAS‐NES in HEK293 cells. Scale bar: 10 μm. (D) Immunoblots of subcellular fractions of GFP‐hCGAS‐, GFP‐hCGAS‐NLS‐, and GFP‐hCGAS‐NES‐expressing HEK293 cells.

Data information: Data are presented as means ± SEM. Statistical significance was assessed using one‐way ANOVA followed by Sidak's post‐test. NS: P > 0.05 and ****P ≤ 0.0001.Source data are available online for this figure.

Figure 2. cGAS promotes irradiation‐induced micronucleus generation and cell death

1. A, B

   Micronuclei (indicated by arrowhead) in GFP‐NLS‐ or GFP‐hcGAS‐expressing HEK293 cells before (0 h) or 24 h after γ‐irradiation (IR; 10 Gy). Scale bar: 10 μm (A). (B) The average MNs/cell. Graphs show mean ± SEM (n = 3 independent experiments) representing six different microscopic fields with over 200 cells.

2. C

   IFNB1 response in HEK293 cells stimulated with transfected plasmid DNA. Mean ± SEM of n = 3 independent experiments.

3. D

   Experimental outline for micronucleus generation and cell death after γ‐irradiation.

4. E

   Micronucleus (indicated by arrowhead) and cGAS staining in WT and cGAS^−/− BMDMos exposed to γ‐irradiation (10 Gy). Scale bar: 10 μm.

5. F

   Average MNs/cell in BMDMos. MN graphs show mean ± SEM (n = 3 independent experiments) representing eight different microscopic fields with over 200 cells.

6. G

   Cell death in WT and cGAS^−/− BMDMos that were first synchronized at G2/M, then γ‐irradiated (10 Gy) followed by release and analysis at indicated time points. Mean ± SD, x biological triplicates (n = 3) per treatment group are shown.

Data information: Statistical significance in (B), (C), and (F) was assessed using unpaired two‐tailed Student's t‐test. ***P ≤ 0.001 and ****P ≤ 0.0001. Statistical significance in (G) was assessed using two‐way ANOVA test, ****P < 0.0001.Source data are available online for this figure.

Figure 3. cGAS inhibits DNA damage repair and promotes micronucleus generation and cell death independently of STING

1. A, B

   BMDMos from cGAS^−/− mice exhibit enhanced DNA repair efficiency than those from WT and Sting^−/− mice. (A) Representative comet tails of WT, cGAS^−/−, and Sting^−/−BMDMos exposed to γ‐irradiation (IR: 10 Gy) on ice, then incubated at 37°C for indicated duration. (B) Corresponding quantification of the comet tail moments from 20 different fields with n > 200 comets of three independent experiments.

2. C, D

   cGAS promotes micronucleus generation in BMDMos independently of STING. (C) Confocal microscopic visualization of micronucleus (indicated by arrowhead) and cGAS staining in WT, cGAS^−/−, and Sting^−/− BMDMos exposed to γ‐irradiation (10 Gy). Scale bar: 10 μm. (D) Average MNs/cell in corresponding representative images. Bar graphs show mean values from eight different microscopic fields with over 200 cells.

3. E, F

   γ‐Irradiation‐induced cell death in WT and cGAS^−/− BMDMos (E). γ‐Irradiation‐induced cell death in Sting^−/− and cGAS^−/− Sting^−/− BMDMos (F).

Data information: Data are presented as mean ± SEM of n = 3 independent experiments. Statistical significance was assessed using two‐way ANOVA in (B) and one‐way ANOVA in (D), (E), and (F) followed by Sidak's post‐test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.Source data are available online for this figure.

Figure EV3. STING signaling is dispensable for inhibition of DNA repair by cGAS

1. A

   Pulsed‐field gel electrophoresis analysis of γ‐irradiated (10 Gy) WT and cGAS^−/− BMDMos.

2. B, C

   Comet assay in GFP‐NLS‐ and GFP‐hcGAS‐expressing HEK293T cells γ‐irradiated (IR: 10 Gy) for 15 min (B). RT–PCR analysis of IFNB1 response in GFP‐NLS‐ or GFP‐hcGAS‐expressing HEK293T cells stimulated with transfected DNA for 6 h (C).

3. D, E

   Comet assay of HEK293 cells stimulated with 10 μg/ml cGAMP for indicate periods, then γ‐irradiated and incubated at 37°C for indicated duration (D). (E) Immunoblots of IRF3 phosphorylation in HEK293 cells treated as in (D).

4. F–H

   Images (F) and quantifications (G) of comet tails 15 min after irradiation of GFP‐NLS‐, GFP‐hcGAS‐, and GFP‐hcGASΔcGAMP‐expressing HEK293 cells. RT–PCR analysis of IFNB1 response in GFP‐NLS‐ or GFP‐hcGAS‐expressing HEK293 cells stimulated with transfected 23 DNA for 6 h (H).

5. I, J

   Images (I) and quantifications (J) of micronuclei in GFP‐NLS‐ and GFP‐hcGASΔcGAMP‐expressing HEK293 cells 24 h after γ‐irradiation (IR; 10 Gy). DAPI (DNA). Scale bar: 10 μm. Each data set bar comet graph was calculated from six different microscopic fields with over 200 cells.

6. K

   Quantifications of comet tails 15 min after irradiation (10 Gy) of GFP‐NLS‐, GFP‐hcGAS‐, or GFP‐mcGAS‐expressing HEK293 cells. Each data set bar comet graph was calculated from six different microscopic fields with over 200 cells.

Data information: Statistical significance was assessed using one‐way ANOVA followed by Sidak's post‐test. NS P > 0.05, ***P ≤ 0.001, and ****P ≤ 0.0001. Mean ± SEM of n = 3 independent experiments.Source data are available online for this figure.

Figure 4. cGAS accelerates γ‐irradiation‐induced depletion of bone marrow cells independently of STING

1. A–D

   Kinetics of in vivo depletion of indicated bone marrow cells in WT mice (n = 3) after γ‐irradiation (9 Gy).

2. E–H

   Bone marrow cells in WT (n = 3) and cGAS^−/− (n = 3) mice 10 h post‐γ‐irradiation.

3. I–L

   Bone marrow cells in Sting^−/− (n = 3) and cGAS^−/− Sting^−/− (n = 3) mice 10 h post‐γ‐irradiation.

Data information: Data in this figure are presented as mean ± SD. One‐way ANOVA followed by Sidak's post‐test. NS: P > 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.

Figure 5. DNA binding and subsequent oligomerization of cGAS are essential for inhibition of HR‐DNA repair

1. A, B

   Schematics of HR and NHEJ reporter assays.

2. C, D

   Obtained results showing enhanced HR efficiency upon knockdown of endogenous cGAS in U2OS cells. Immunoblot inserts depict knockdown efficiency of cGAS and histone H1.2.

3. E, F

   Results showing the effect of hcGAS, hcGAS▵cGAMP, hcGAS▵DNA, or hcGAS▵Oligo on HR (E) or NHEJ (F) in HEK293 cells. Corresponding immunoblot inserts depict cGAS expression.

Data information: Data are means with SEM, n = 3. Statistical significance was assessed using one‐way ANOVA followed by Sidak's post‐test. NS: P > 0.05, *P ≤ 0.05, ****P ≤ 0.0001.Source data are available online for this figure.

Figure EV4. cGAS suppresses DNA repair without inhibiting ATM activation

1. A

   Reporter assays showing the effect of NLS and NES on cGAS‐mediated inhibition of DNA repair.

2. B

   Both full‐length hcGAS and hcGAS^cat (161–522aa) inhibit HR repair.

3. C–E

   cGAS does not impede ATM activation. ATM phosphorylation in γ‐irradiated (10 Gy) GFP‐NLS‐ and GFP‐hcGAS‐expressing HEK293T cells (C), GFP‐NLS‐, GFP‐hcGAS‐, and GFP‐hcGASΔcGAMP‐expressing HEK293 cells (D), or γ‐irradiated (2.5 Gy) WT, cGAS^−/−, and Sting^−/− BMDMos (E).

Data information: Data are means ± SD, n = 3. Statistical significance was assessed using one‐way ANOVA followed by Sidak's post‐test. ***P < 0.001 ****P < 0.0001, NS: P > 0.05.Source data are available online for this figure.

Figure 6. Nuclear cGAS is constantly bound to the chromatin and not specifically recruited to DSB sites

1. A

   cGAS is not recruited to DSB sites: Confocal microscopic images of GFP‐NLS‐ or GFP‐hcGAS‐expressing U2OS‐DSB reporter cells incubated (or not) with Shield‐1 and 4‐OHT to induce the expression and translocation of mCherry‐LacI‐FokI (red) to specific DSB sites. Scale bar: 10 μm. The arrowheads indicate DSB sites.

2. B

   cGAS does not co‐localize with γ‐H2AX at DSB sites: GFP‐NLS‐ or GFP‐hcGAS‐expressing HEK293 cells exposed (or not) to γ‐irradiation (IR: 10 Gy), then stained for γ‐H2AX. Scale bar: 10 μm.

3. C, D

   Nuclear cGAS is mainly chromatin‐bound and remains unaltered upon γ‐irradiation. (C) Cytosolic (cyto) and nuclear fractions of γ‐irradiated (10 Gy, 30 min) BMDMos analyzed for cGAS and indicated molecules. (D) Cytosolic, soluble nuclear, and chromatin fractions from BMDMos were immunoblotted for cGAS and indicated proteins.

4. E–G

   cGAS co‐isolates with DNA repair proteins because of bound chromatin bridges. (E) Nuclease digestion abrogates the co‐isolation of cGAS and DNA repair proteins: Lysates of control (−IR) and γ‐irradiated (+IR, 10 Gy, 30 min) GFP‐hcGAS‐expressing HEK293 cells were treated (or not) with benzonase before cGAS immunoprecipitation and analysis for indicated proteins. (F) Agarose gel analysis of DNA in corresponding cell lysates in (E). (G) Co‐isolation of cGAS and DNA repair proteins depends on its binding to DNA: cGAS pulldowns along with lysate inputs of control and γ‐irradiated HEK293 cells expressing GFP‐hcGAS or GFP‐hcGASΔDNA probed for indicated proteins.

Source data are available online for this figure.

Figure 7. cGAS inhibits the HR‐DNA repair by impeding RAD51‐mediated strand invasion

1. Confocal images of γ‐irradiated GFP‐NLS‐ and GFP‐hcGAS‐expressing HEK293 cells stained for RAD51 (red) with or without γ‐irradiation. Scale bar: 10 μm.

2. Schematics of the D‐loop formation assay, including pre‐incubation of template dsDNA with cGAS^cat (i) or with cGAS^cat being added after RAD51 was bound to dsDNA (ii).

3. Pre‐incubation of dsDNA with mcGAS^cat prevents D‐loop formation by human RAD51, but does not affect the RAD1 activity once RAD51 filaments are bound to dsDNA. The percentage of D‐loop formed in each reaction (left) was graphed as the average of triplicates ± SD.

4. Schematics of the D‐loop assay.

5. Pre‐incubation of template dsDNA with hcGAS^cat blocks subsequent D‐loop formation. The percentage of D‐loop formation (below) was graphed as the average of triplicates ± SD.

Data information: Unpaired two‐tailed Student's t‐test was used for statistical analyses. NS P > 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.Source data are available online for this figure.

Figure EV5. cGAS inhibits RAD51‐mediated DNA strand exchange and D‐loop formation

1. A

   Coomassie Blue staining of purified hcGAS^cat, mcGAS^cat, Rad51, HOP2, and MND1.

2. B

   Schematics of the D‐loop assay.

3. C

   Pre‐incubation of template dsDNA with cGAS blocks subsequent D‐loop formation regardless of the presence of cGAMP precursors (ATP+GTP).

4. D

   Schematics of the strand exchange reaction.

5. E–H

   Pre‐incubation of dsDNA with cGAS protein inhibited the DNA strand exchange activity of human RAD51 (E, F) and yeast Rad51 (G, H) regardless of the presence of precursors (ATP+GTP) of cGAMP. The percentage of DNA strand exchange in each reaction was graphed as the average of triplicates ± SD.

Data information: Statistical significance was assessed using one‐way ANOVA followed by Sidak's post‐test. NS: P > 0.05, ****P ≤ 0.0001.Source data are available online for this figure.

Figure 8. cGAS compacts dsDNA and inhibits D‐loop formation via oligomerization

1. Negative‐stain electron micrographs of cGAS‐dsDNA complexes following incubation of dsDNA with indicated cGAS variants. Scale bar: 100 nm.

2. Effect of indicated hcGAS variants on D‐loop formation when pre‐incubated with dsDNA.

3. Percentage of D‐loop formed in each reaction (left) graphed as the average of triplicates ± SD.

4. Overview of a single 1:1 hcGAS‐DNA complex depicting the location of the Y215 within the cGAS‐dsDNA interface.

5. DR‐GFP assay showing that hcGASΔDNA‐Y215E is impaired in HR inhibition.

6. hcGASΔDNA‐Y215E but not hcGASΔcGAMP has a decreased affinity to dsDNA24.

7. hcGASΔDNA‐Y215E and hcGASΔcGAMP are defective in synthase activity.

8. Negative‐stain electron micrographs showing that hcGAScat‐ΔDNA‐Y215E is defective in inducing cGAS‐dsDNA complexes. Scale bar: 100 nm.

9. Effect of indicated hcGAS variants on D‐loop formation.

Data information: Data are means ± SD, n = 3. Unpaired Student's t‐test was used for statistical analyses: NS P > 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.Source data are available online for this figure.