DNA damage repair kinase DNA-PK and cGAS synergize to induce cancer-related inflammation in glioblastoma

Abstract

Cytosolic DNA promotes inflammatory responses upon detection by the cyclic GMP-AMP (cGAMP) synthase (cGAS). It has been suggested that cGAS downregulation is an immune escape strategy harnessed by tumor cells. Here, we used glioblastoma cells that show undetectable cGAS levels to address if alternative DNA detection pathways can promote pro-inflammatory signaling. We show that the DNA-PK DNA repair complex (i) drives cGAS-independent IRF3-mediated type I Interferon responses and (ii) that its catalytic activity is required for cGAS-dependent cGAMP production and optimal downstream signaling. We further show that the cooperation between DNA-PK and cGAS favors the expression of chemokines that promote macrophage recruitment in the tumor microenvironment in a glioblastoma model, a process that impairs early tumorigenesis but correlates with poor outcome in glioblastoma patients. Thus, our study supports that cGAS-dependent signaling is acquired during tumorigenesis and that cGAS and DNA-PK activities should be analyzed concertedly to predict the impact of strategies aiming to boost tumor immunogenicity.

Keywords: DNA-PK; cGAS; inflammation; tumor immunogenicity.

Figure 1. DNA‐PK catalytic activity promotes nucleic acid‐dependent type I IFN responses in cGAS‐deficient cells

1. T98G cells were challenged or not with dsDNA or with 2′3′cGAMP for 6 h prior to whole cell extraction and Western Blot (WB) analysis using indicated antibodies.

2. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in A (n = 3 independent experiments).

3. Gli4 cells were treated as in A prior to WB analysis using indicated antibodies.

4. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in C (n = 3 independent experiments).

5. Phorbol 12‐myristate 13‐acetate (PMA)‐treated THP1 cells were incubated for 24 h with conditioned media derived from T98G cells treated or not with dsDNA. IFIT1, IFIT2, MXA, OAS1 mRNA levels were analyzed by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

6. T98G cells were challenged or not with dsDNA for 6 h, in the presence or not of the NU7441 DNA‐PKcs inhibitor, prior WB analysis using indicated antibodies.

7. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in F (n = 4 independent experiments).

8. T98G cells were treated with non‐targeting (CTRL) or DNA‐PKcs‐targeting siRNAs for 72 h prior to 6 h challenge with dsDNA. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

9. T98G cells were treated or not with 0.16 μM camptothecin (CPT) for 48 h prior to immunofluorescence analysis using dsDNA‐ and 53BP1‐specific antibodies, and DAPI nuclear staining (n = 3 independent experiments). Scale bar, 20 μm.

10. Whole cell extracts from T98G cells treated or not for 72 h with 0.16 μM CPT were analyzed by WB using indicated antibodies.

11. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in J (n = 3 independent experiments).

12. T98G cells were treated or not with 0.16 μM CPT for 72 h, in presence or not of NU7441, prior to assessment of IFNB and CXCL10 mRNA levels by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

Data information: All immunoblots are representative experiments (n = 3 independent experiments). All graphs present means ± standard error from the mean (SEM). P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure EV1. DNA‐PKcs promotes type I Interferon responses in cancer cells that do not express cGAS

1. Whole cell extracts from THP‐1^CTRL, THP‐1 ^cGAS−/− and T98G cells were analyzed by WB using indicated antibodies.

2. Gli7 cells were challenged or not with dsDNA for 6 h prior to WB analysis using indicated antibodies.

3. Whole cell extracts from patient derived‐glioblastoma stem‐like cells (GSC 4, 6, 9, 13, 15) were analyzed by WB using indicated antibodies.

4. GSC9 cells were challenged or not with dsDNA for 6 h prior to whole cell extraction and WB analysis using indicated antibodies.

5. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in B (n = 3 independent experiments).

6. GSC9 cells were challenged or not for 6 h with dsDNA, in the presence or not of the NU7441 DNA‐PKcs inhibitor, prior to analysis of IFNB and CXCL10 levels by RT–qPCR (n = 3 independent experiments).

7. As in (F), except that whole cell extracts were analyzed by WB using indicated antibodies.

8. Gli4 cells were challenged or not for 6 h with dsDNA, in the presence or not of the NU7441 DNA‐PKcs inhibitor, prior to analysis of IFNB and CXCL10 levels by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

9. T98G cells were treated with non‐targeting (CTRL) or DNA‐PKcs‐targeting siRNAs prior to whole cell extraction and WB analysis using indicated antibodies.

10. Whole cell extracts from T98G cells, treated or not with camptothecin (CPT) for 48 h, in presence or not of NU7441 inhibitor (24 h), were analyzed by WB using indicated antibodies.

11. T98G cells were treated or not with CPT for 48 h, in presence or not of NU7441 inhibitor (24 h), prior to immunofluorescence analysis using dsDNA‐ and 53BP1‐specific antibodies and DAPI nuclear staining (n = 3 independent experiments). Scale bar, 20 μm.

12. T98G cells were treated or not with 25 μM etoposide (ETO) for 72 h, in presence or not of NU7441, prior to assessment of IFNB and CXCL10 mRNA levels by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

Data information: All immunoblots show representative experiments (n = 3 independent experiments). All graphs present means ± standard error from the mean (SEM). P‐values were determined by Student's t‐test. *P < 0.05, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure 2. DNA‐PK‐dependent detection of cytosolic dsDNA drives STING‐independent IRF3‐dependent type I IFN responses in cancer cells lacking cGAS

1. T98G cells were treated or not with dsDNA for 6 h prior to immunofluorescence analysis using DNA‐PKcs‐ and pDNA‐PKcs‐specific antibodies, and DAPI nuclear staining (n = 3 independent experiments). Scale bar, 20 μm.

2. Whole cell extracts from T98G cells were incubated with 80 nt‐long biotinylated ssDNA or dsDNA prior to pull‐down using streptavidin‐affinity beads. Input and eluates were analyzed by WB using indicated antibodies.

3. T98G cells were treated with non‐targeting (CTRL) or KU70‐targeting siRNAs prior to whole cell extract preparation, and pull‐down as in (B). Inputs and eluates were analyzed by WB using indicated antibodies.

4. T98G cells were treated with CTRL or KU70‐targeting siRNAs prior to challenge with dsDNA for 6 h. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

5. CTRL or STING ^−/− T98G cells were transfected or not with dsDNA for 8 h prior to analysis of IFNB and CXCL10 mRNA levels (n = 3 independent experiments).

6. As in (E), except that CTRL or IRF3 ^−/− T98G cells were transfected or not with dsDNA for 6 h prior to analysis (n = 3 independent experiments).

7. CTRL or STING ^−/− T98G cells were treated or not with CPT for 72 h prior to analyses of IFNB and CXCL10 mRNA levels (n = 3 independent experiments).

8. As in (G), except that CTRL and IRF3 ^−/− T98G cells were used. Graphs present a representative biological triplicate (n = 3 independent experiments).

9. Schematic representation of cytosolic dsDNA‐dependent type I IFN induction in T98G cells.

Data information: All immunoblots are representative experiments (n = 3 independent experiments). All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure EV2. DNA‐PK‐dependent detection of cytosolic dsDNA drives STING‐independent IRF3‐dependent type I IFN responses in cancer cells lacking cGAS

1. Gli7 cells were treated or not with dsDNA for 6 h prior to immunofluorescence analysis using pDNA‐PKcs‐specific antibody and DAPI nuclear staining (n = 3 independent experiments). Scale bar, 20 μm.

2. THP‐1 cells were transfected or not with 80 nt‐long biotinylated dsDNA prior to pull‐down using streptavidin‐affinity beads. Input and eluates were analyzed by WB using indicated antibodies.

3. Whole cell extracts from control HCT116 (CTRL) and HCT116 ^PRKDC−/− were used in pull‐down experiments using biotinylated dsDNA and streptavidin affinity beads. Inputs and eluates were analyzed by WB using indicated antibodies.

4. T98G cells were treated with non‐targeting (CTRL) or KU70‐targeting siRNAs prior to whole cell extraction and analysis by WB using indicated antibodies.

5. Whole cell extracts from CTRL or STING ^−/− T98G cells transfected or not with 2′3′cGAMP were analyzed by WB using indicated antibodies.

6. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in (E). Graphs present a representative biological triplicate (n = 3 independent experiments).

7. Whole cell extracts from CTRL or STING ^−/− T98G cells transfected or not with dsDNA were analyzed by WB using indicated antibodies.

8. Whole cell extracts from CTRL or IRF3 ^−/− T98G cells transfected or not with dsDNA were analyzed by WB using indicated antibodies.

9. Whole cell extracts from CTRL or STING ^−/− T98G cells treated or not with CPT for 72 h were analyzed by WB using indicated antibodies.

10. Whole cell extracts from CTRL or IRF3 ^−/− T98G cells treated or not with CPT for 72 h were analyzed by WB using indicated antibodies.

11. THP‐1^CTRL, THP‐1 ^cGAS−/− and THP‐1 ^STING−/− were challenged or not with dsDNA for 6 h, prior to WB analysis using indicated antibodies.

12. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in (K). Graphs present a representative biological triplicate (n = 3 independent experiments).

Data information: All immunoblots are representative experiments (n = 3 independent experiments). All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure 3. cGAS and DNA‐PKcs cooperate for optimal dsDNA‐induced type I Interferon responses

1. T98G cells stably expressing cGAS (T98G^cGAS) or not (T98G^Empty) were transfected or not with dsDNA for 6 h prior to whole cell extraction and WB analysis using indicated antibodies.

2. IFNB and CXCL10 mRNA levels were analyzed by RT–qPCR in samples treated as in (A). Graphs present a representative biological triplicate (n = 5 independent experiments).

3. T98G^Empty and T98G^cGAS were transfected or not with biotinylated dsDNA prior to whole cell extraction and pull‐down using streptavidin‐affinity beads. Inputs and eluates were analyzed by WB using indicated antibodies.

4. T98G^Empty and T98G^cGAS were transfected or not with dsDNA for 6 h in the presence or not of NU7441 prior to IFNB and CXCL10 expression analysis. Graphs present a representative biological triplicate (n = 5 independent experiments).

5. T98G^Empty and T98G^cGAS were treated or not with CPT for 72 h in combination or not with NU7441 (48 h) prior to IFNB and CXCL10 expression analysis. Graphs represent a biological triplicate (n = 4 independent experiments).

6. THP‐1^CTRL and THP‐1 ^cGAS−/− were transfected or not with dsDNA for 6 h in presence or not of NU7441 prior to IFNB and CXCL10 expression analysis (n = 3 independent experiments).

7. Intracellular cGAMP levels were analyzed in samples treated as in (E) by ELISA (n = 3 independent experiments).

8. Intracellular cGAMP levels were analyzed in samples treated as in (F) by ELISA. Graphs present a representative biological triplicate (n = 2 independent experiments).

9. T98G expressing a catalytic dead cGAS allele (T98G^cGAS‐CD) and T98G were treated as in (A) prior to IFNB and CXCL10 levels analysis. Graphs present a representative biological triplicate (n = 3 independent experiments).

10. T98G^cGAS were transfected or not with dsDNA in presence or not of NU7441, prior to WB analysis using indicated antibodies.

11. THP‐1 were transfected or not with dsDNA in presence or not of NU7441, prior to WB analysis using indicated antibodies.

12. Schematic representation of the molecular mechanisms involved in the cooperation between DNA‐PKcs and cGAS for type I IFN induction.

Data information: All immunoblots are representative experiments (n = 3 independent experiments). All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure EV3. DNA‐PKcs and cGAS synergize for the activation of type I Interferon responses

1. T98G cells expressing a cGAS‐encoding vector (T98G^cGAS) were treated or not with dsDNA for 6 h prior to IF analysis using pDNA‐PKcs‐specific antibody and DAPI nuclear staining (n = 3 independent experiments). Scale bar, 20 μm.

2. THP‐1 and THP‐1 ^{cGAS −/−} were transfected or not with 80‐nt long biotinylated dsDNA prior to whole cell extraction and pull‐down using streptavidin‐affinity beads. Inputs and eluates were analyzed by WB using indicated antibodies.

3. T98G cells expressing an empty (T98G^Empty) or T98G^cGAS were transfected or not with 5, 10 or 20 μg of biotinylated dsDNA prior to whole cell extraction and pull‐down using streptavidin‐affinity beads. Inputs and eluates were analyzed by WB using indicated antibodies.

4. T98G^Empty or T98G^cGAS were transfected or not with dsDNA for 6 h in the presence or not of the NU7441 DNA‐PKcs inhibitor. Whole cell extracts were analyzed by WB using indicated antibodies.

5. THP‐1^CTRL and THP‐1 ^{cGAS −/−} were transfected or not with dsDNA for 6 h in presence or not of the NU7441 DNA‐PKcs inhibitor prior to analysis of protein expression by WB using indicated antibodies.

6. CFPAC and CFPAC ^{cGAS −/−} were transfected or not with dsDNA for 6 h in presence or not of the NU7441 DNA‐PKcs inhibitor prior to WB analysis using indicated antibodies.

7. IFNB and CXCL10 mRNA levels were assessed by RT–qPCR in CFPAC^CTRL and CFPAC ^{cGAS −/−} treated as in D. Graphs present a representative biological triplicate (n = 3 independent experiments).

8. As in (F), except that MEF and MEF ^{cGas −/−} were transfected.

9. As in (G), except that MEF and MEF ^{cGas −/−} were transfected. Graphs present a representative biological triplicate (n = 3 independent experiments).

10. GL261^CTRL and GL261 ^{cGAS −/−} were transfected or not with dsDNA for 6 h in presence or not of the NU7441 DNA‐PKcs inhibitor. Whole cell extracts were analyzed by WB using indicated antibodies. IFNB and CXCL10 mRNA levels were assessed by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

11. cGAS activity upon treatment with 2.5 μM of the CU76 cGAS inhibitor and 2, 4, 6, or 8 μM of NU7441 was measured by ELISA (n = 3 independent experiments).

12. T98G^Empty, T98G^cGAS and T98G^cGAS‐CD were transfected or not with dsDNA for 6 h prior to analysis of protein expression by WB using indicated antibodies.

Data information: All graphs present means ± SEM. All immunoblots show representative experiments (n = 3 independent experiments). P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Source data are available online for this figure.

Figure 4. cGAS re‐expression in glioblastoma cancer cells impairs tumorigenesis

1. The proliferation of T98G^Empty and T98G^cGAS was monitored in 2D cultures over 4 days (n = 3 independent experiments).

2. The volume of spheroids formed by T98G^Empty and T98G^cGAS was monitored over 8 days (n = 3 independent experiments).

3. Representative images of T98G^Empty and T98G^cGAS spheroids measured in (B), at day 1 and day 8. Scale bar, 250 μm.

4. The volume of subcutaneous T98G^Empty and T98G^cGAS tumors in nude mice was measured every 3–4 days by caliper (n = 6 mice per group).

5. Representative pictures of T98G^Empty and T98G^cGAS tumors from D, at day 90 post subcutaneous engraftment.

6. T98G^Empty and T98G^cGAS stably expressing a GFP reporter (T98G‐GFP^Empty and T98G‐GFP^cGAS, respectively) were xenotransplanted into the head of tg(mfap4:RFP) zebrafish line at 3 days post fertilization (dpf). Zebrafish embryos were imaged daily over 3 days. The graph represents the mean (± SEM) percentage of tumor growth normalized by the area on the day of transplantation (n = 21 T98G‐GFP^Empty and n = 29 T98G‐GFP^cGAS embryos). Scale bar, 200 μm.

Data information: All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ****P < 0.0001. One‐way Anova with Tukey's multiple comparisons test was used for mice analyses. Mann–Withney test was performed to analyze tumor growth in zebrafish. Source data are available online for this figure.

Figure EV4. Re‐expression of cGAS in glioblastoma cells decrease invasiveness in vivo

1. The number of invasive cells per embryo transplanted in Fig 4G was manually counted at D0, D1 and D2 post transplantation. Data represents mean (± SEM) of n = 21 (T98G‐GFP^Empty) or 29 (T98G‐GFP^cGAS) embryos. A Mann–Whitney test was performed to assess the significance *P < 0.05.

2. Representative images of elongated T98G cells counted in (A). Scale bar, 100 μm. Arrows indicate elongated pseudopodia.

Figure 5. cGAS re‐expression in glioblastoma cancer cells promotes macrophage recruitment

1. Graph represents the mean (± SEM) fold migration of THP‐1 cells through a 3 μm transwell insert when conditioned media from T98G^Empty or T98G^cGAS was applied to lower chamber for 6 h. CCL2 was used as positive control (n = 8 biological replicates).

2. CXCL10, CCL2 and CCL5 protein levels in conditioned media from T98G^Empty and T98G^cGAS cells were assessed using proteome profiler. Proteins that were found to be the most upregulated in T98G^cGAS are shown. Representative immunoblots (n = 3 independent experiments).

3. T98G^Empty and T98G^cGAS were transfected or not with dsDNA prior to analyses of CCL2 and CCL5 by RT–qPCR (n = 4 independent experiments).

4. Zebrafish embryos injected with T98G‐GFP^Empty, T98G‐GFP^cGAS, or PBS at 3 dpf (Fig 4E) were imaged at 24 h post transplantation. Representative images of macrophage recruitment (purple) in the head (n = 21 T98G‐GFP^Empty and n = 29 T98G‐GFP^cGAS embryos). Scale bar, 200 μm.

5. Graph presents the quantification of macrophages recruited at tumor site 24 h post xenotransplantation in (D) (n = 21 T98G‐GFP^Empty and n = 29 T98G‐GFP^cGAS embryos).

6. Heads of zebrafish treated as in (D) were isolated prior to RNA extraction and analysis of M1 (tnfa and cxcl11) or M2 (arg2 and il10) polarization markers. Each value in the graph is the mean of 25 embryos.

Data information: All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ****P < 0.0001. Mann–Withney test was performed to analyze macrophage recruitment in zebrafish. Source data are available online for this figure.

Figure EV5. Re‐expression of cGAS in glioblastoma tumors promotes the secretion of chemokines that enhance macrophage recruitment

1. THP‐1 monocytes were incubated for 24 h with conditioned media (CM) derived from T98G^Empty and T98G^cGAS cells, prior to analyses of CD197, CD80, CXCL10, CD209 and Fibronectin gene expression, by RT–qPCR (n = 4 independent experiments).

2. CD14⁺ monocytes derived from healthy donors were incubated for 72 h with conditioned media derived from T98G^Empty and T98G^cGAS cells, prior to flow cytometry analysis of M1 (CD86, CD38) and M2 (CD206, CD209) polarization markers (n = 5 donors).

3. T98G^Empty, T98G^cGAS and T98G^cGAS‐CD cells were transfected or not with dsDNA for 6 h, prior to analyses of CCL2 and CCL5 mRNA levels by RT–qPCR. Graphs present a representative biological triplicate (n = 3 independent experiments).

4. T98G^Empty and T98G^cGAS were transfected or not with dsDNA for 6 h in the presence or not of NU7441 prior to CCL2 and CCL5 expression analysis. Graphs present a representative biological triplicate (n = 3 independent experiments).

5. As in (D), except that IFNL2/3, CCL3, IL6 expressions were analyzed (n = 3 independent experiments).

Data information: All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 6. cGAS and DNA‐PKcs levels increase with tumor grade

1. Correlation plot between PRKDC and MD21B1 expression in glioblastoma patients. Three distinct populations can be visualized: PRKDC ^low /MD21B1 ^low ; PRKDC ^high /MD21B1 ^low and PRKDC ^high /MD21B1 ^high (total patients n = 224).

2. Violin plots present chemokine gene expression (CXCL10, CCL2 and CCL5) in glioblastoma samples from (A).

3. Violin plots present macrophage gene expression (CD68 and ITGAM) in glioblastoma samples from (A).

4. Violin plots present pro‐inflammatory M1 macrophage gene expression (CD86 and CD80) in glioblastoma samples from (A).

5. Violin plots present anti‐inflammatory M2 macrophages gene expression (CD200R1 and CD209) in glioblastoma samples from (A).

6. Violin plots present M1/M2 gene expression ratio, calculated using the mean expression of the genes indicated in (D) and (E).

7. Graph presents the survival rate of glioblastoma patients from A that present PRKDC ^low /MB21D1 ^low , PRKDC ^high /MB21D1 ^low and PRKDC ^high /MB21D1 ^high expression.

8. Violin plots present the expression of PRKDC and MB21D1 in datasets analyzed in A, based on tumor grade (II to IV).

9. Representative images of immunohistochemical analysis of DNA‐PKcs and cGAS proteins in human brain tumor samples. Scale bar, 75 μm.

10. Schematic representation of the molecular mechanisms involved in the cooperation between DNA‐PKcs and cGAS for type I IFN responses and chemokine secretion. In cells with undetectable cGAS levels, interaction with exogenous (1) or endogenous (2) cytosolic dsDNA leads to DNA‐PK activation (3) and promotes IRF3‐dependent type I IFN responses (4). In cells where both DNA‐PK and cGAS are expressed, cytosolic DNA is detected by both DNA‐PK and cGAS (5). DNA‐PK induces cytokine and chemokine secretion through IRF3 activation and enables cGAS phosphorylation, thus promoting the activation of the cGAS‐STING pathway. This cooperation fuels cancer‐associated inflammation.

Data information: All graphs present means ± SEM. P‐values were determined by Student's t‐test. ns: not significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. One‐way Anova with Tukey's multiple comparisons test was used to compare gene expression among populations in glioblastoma dataset. Source data are available online for this figure.