Fig 2: DNA-flexibility controls DNA-potential to bind and activate cGAS.a DNA looping assay schematic and fluorescence images of single DNA in donor and acceptor channels after 1 M NaCl treatment for noted time. Scale bar = 10 μm. b Loop fraction and time of damaged ISDs. In this study, data are presented as mean values ± standard deviations (SDs), and error bars indicate SDs (Mean ± SD). n = 3 independent experiments. c Coomassie staining of purified human cGAS (hcGAS) protein. M: Marker; kD: kilodaltons. d Left: EMSA of hcGAS binding to noted ISD using HT-DNA as control. Right: EMSA bound fraction of 50 nM ISDs binding to serial dilutions of hcGAS protein. Mean ± SD. n = 3 independent experiments. e Representative images of hcGAS–DNA condensates after mixing 2 μM hcGAS with 2 μM ISD. Red: Cy3-labelled DNA; Green: Alexa Fluor- 488 (AF-488)-labelled hcGAS; BF: bright field. Scale bar = 20 μm. f Looping fraction and times of noted ISDs. Mean ± SD. n = 3 independent experiments. g Left: Representative images of hcGAS–DNA condensates after mixing 2 μM hcGAS with noted ISD. Red: Cy3-labelled DNA; Green: AF-488-labelled hcGAS; BF: bright field. Scale bar = 20 μm. Middle: cGAMP produced by mixing 100 nM cGAS with 100 nM ISD for 2 h. Right: Ifnb mRNA levels in the MEF cells transfected with 100 ng/mL ISD for 2 h. Mean ± SD. n = 6 biologically independent samples. h EMSA analyses of 50 nM ISD binding to hcGAS. Mean ± SD. n = 3 independent experiments. i Left: Representative images of cGAS–DNA condensate formation after mixing hcGAS with noted ISD. Red: Cy3-labelled DNA; Green: AF-488-labelled hcGAS; BF: bright field. Scale bar = 20 μm. Middle: cGAMP produced by mixing hcGAS and ISD. Right: Ifnb mRNA levels in the MEF cells transfected with ISD. Mean ± SD. n = 6 biologically independent samples. Unless specifically noted, P values were from unpaired two-tailed Student’s t test (***p < 0.001, **p < 0.01, *p < 0.05) in this study. Source data are provided as a Source Data file.

Fig 3: DNA flexibility is associated with DNA potential to bind and activate cGAS.a Looping time of seq1–10 ISDs. Mean ± SD. n = 3 independent experiments. b Spearman analyses of the correlation between the ISD looping time and noted cGAS-binding or -activating potential. P and R values were from two-sided spearman analyses. c Pearson analyses of the correlation between the DNA looping time and noted cGAS-binding or -activating potential. P and R values were from two-sided Pearson analyses. Red dashed lines are the best fit lines from the linear regression. Mean ± SD. Looping time: n = 3 independent experiments; EMSA, MST, MARQTFQ cGAMP and Ifnb mRNA: n = 6 biologically independent samples. Source data are provided as a Source Data file.

Fig 4: Loss of cGAS-mediated daily AIS impairs the radiotherapy efficiency.a Ifnb mRNA (left) and protein (right) levels in the tumours at 4 h after single focal 8 Gy IR. Mean ± SD. MRNA: n = 8 tumours from 4 mice. Protein: n = 4 tumours from 4 mice. b Cxcl10 mRNA levels in the tumours in (a). Mean ± SD. n = 8 tumours from 4 mice. c Ifnb mRNA (left) and protein (right) levels in the inhibitor-pretreated mice at 4 h after single IR. Before IR, mice were injected with 5 mg/kg RU.521 for 3 h or treated by oral gavage with 4 mg/kg of niraparib for 3 h. Mean ± SD. MRNA: n = 8 tumours from 4 mice. Protein: n = 4 tumours from 4 mice. d Cxcl10 mRNA levels in the tumours in (c). Mean ± SD. n = 4 tumours from 4 mice. e Schematic showing right flank implanted tumour treated by three focal fractions of 8 Gy IR on D10, D11 and D12. f, g Tumour volumes (f) and survival percentages (g) in the RU.521-pretreated mice described in (e). P values in (f) were from unpaired two-sided Student’s t-test without adjustment for multiple comparisons. P values in (g) were from log-rank test. Mean ± SD. n = 8 mice. h Primary (right flank) and abscopal (left flank) tumour volumes in the mice treated with RU.521 prior to IR. Implanted tumours were treated by three focal IR fractions on the primary tumour, followed by CTLA-4 mAb treatment. Mean ± SD. n = 10 mice. i, CD8 + T cells in the abscopal tumours described in (h). Mean ± SD. n = 10 mice (tumours). j Schematic showing IS of DNA-flexibility. cGAS might apply R222 to detect the DNA-flexibility via DNA-curving. k The differences between flexsurveillance and DNA indiscriminate models. P values in (g) were from log-rank test, while all other P values were from unpaired two-tailed Student’s t test without adjustment for multiple comparisons (***p < 0.001, **p < 0.01, *p < 0.05). Source data are provided as a Source Data file.

Fig 5: Human R236 substitution relaxes the DNA-length specificity of cGAS.a EMSA analyses of 50 nM 20 bp ISDs binding the noted mouse mcGAS proteins. Mean ± SD. n = 3 independent experiments. b cGAMP produced by mixing 100 nM noted mcGAS and 100 nM 20 bp noted ISD for 2 h. Unless specifically noted, P values were from unpaired two-tailed Student’s t test without adjustment for multiple comparisons (***p < 0.001, **p < 0.01, *p < 0.05, N.S. = not statistically significant (p > 0.05)) in this study. Mean ± SD. n = 6 biologically independent samples. c Ifnb mRNA levels in the human THP-1 cells transfected with 100 ng/mL 20 bp noted ISDs for 2 h. Mean ± SD. n = 3 biologically independent samples. d Left: cGAMP produced by 100 nM human hcGAS in serial dilutions of noted ISD for 2 h. Right: cGAMP produced by serial dilutions of human hcGAS in 100 nM noted ISD for 2 h. Mean ± SD. n = 3 biologically independent samples. e EMSA analyses of 50 nM 20 bp ISDs binding to the noted human hcGAS proteins. Mean ± SD. n = 3 biologically independent samples. f cGAMP produced by mixing 100 nM noted human hcGAS and100 nM 20 bp noted ISD for 2 h. Mean ± SD. n = 6 biologically independent samples. g cGAMP produced by 100 nM noted mouse mcGAS (left) or human hcGAS (right) in the presence of 100 nM 28- or 20 bp ISD for 2 h. Mean ± SD. n = 6 biologically independent samples. h cGAMP produced by 100 nM noted mouse mcGAS (left) or human hcGAS (right) in the presence of 100 nM 44- or 20 bp ISD for 2 h. Mean ± SD. n = 6 biologically independent samples. i Sequence alignment to show that the human K185 and L195 residues are non-conserved between human and mouse. j cGAMP produced by 100 nM noted human hcGAS in the presence of 100 nM 44- or 20 bp ISD for 2 h. Mean ± SD. n = 6 biologically independent samples. Source data are provided as a Source Data file.