RPA and Rad51 constitute a cell intrinsic mechanism to protect the cytosol from self DNA

Abstract

Immune recognition of cytosolic DNA represents a central antiviral defence mechanism. Within the host, short single-stranded DNA (ssDNA) continuously arises during the repair of DNA damage induced by endogenous and environmental genotoxic stress. Here we show that short ssDNA traverses the nuclear membrane, but is drawn into the nucleus by binding to the DNA replication and repair factors RPA and Rad51. Knockdown of RPA and Rad51 enhances cytosolic leakage of ssDNA resulting in cGAS-dependent type I IFN activation. Mutations in the exonuclease TREX1 cause type I IFN-dependent autoinflammation and autoimmunity. We demonstrate that TREX1 is anchored within the outer nuclear membrane to ensure immediate degradation of ssDNA leaking into the cytosol. In TREX1-deficient fibroblasts, accumulating ssDNA causes exhaustion of RPA and Rad51 resulting in replication stress and activation of p53 and type I IFN. Thus, the ssDNA-binding capacity of RPA and Rad51 constitutes a cell intrinsic mechanism to protect the cytosol from self DNA.

Figure 1. The ssDNA-binding of RPA and Rad51 prevents type I IFN activation caused by leakage of nuclear ssDNA into the cytosol.

(a,b) The cytoplasm (a) or the nucleus (b) of HEK293T cells was microinjected with ssDNA647N (ssDNA, red) and FITC-dextran (dextran, green) as tracer. Images were taken at the indicated time points. Scale bar, 10 μM. (c) Representative image of a binuclear HEK cell. Fluorescence of ssDNA647N (ssDNA, red) microinjected into one nucleus, marked by FITC-dextran (dextran, green), is taken up by the non-injected adjacent nucleus. Scale bar, 10 μM. (d) ssDNA647N (ssDNA) microinjected into nuclei of HeLa cells with knockdown of RPA70 and Rad51 (siRPA+siRad51) leaks into the cytosol within 15 min, but remains nuclear in cells transfected with negative control siRNA (siCtrl). Co-injected FITC-dextran (green) marks the injected compartment. Scale bar, 10 μM. (e,f) Phosphorylation of IRF3 and p53 (e) as well as IFN-β induction (f) after knockdown of RPA70 and Rad51 (siRPA, siRad51) is suppressed by additional knockdown of cGAS. siCtrl, negative control siRNA. (f) Shown are the means of one experiment run in triplicates out of two independent experiments. Error bars, s.d. ***P<0.001 versus siCtrl. ^###P<0.001 versus knockdown of RPA70, Rad51 or RPA70 and Rad51, respectively, by analysis of variance followed by Tukey' post hoc test.

Figure 2. TREX1-deficient cells accumulate short ssDNA in the cytosol and nucleus and exhibit enhanced ssDNA-binding of RPA and Rad51.

(a) Accumulation of ssDNA in the cytoplasm and nucleus of patient (AGS1) compared with wild-type (WT) cells visualized by immunostaining (anti-ssDNA, green, left). AGS1 cells pretreated with S1 nuclease (AGS1+S1) were stained as control. Nuclei are counterstained with DAPI (blue). Scale bar, 100 μm. Mean fluorescence intensities (MFI) are depicted on the right. Data are from at least three independent experiments. Error bars, s.e.m. ***P<0.001. (b) Perinuclear halo formation in TREX1-deficient (LE, AGS) and WT fibroblast nuclei. Indicated are the diameters of at 50% (magenta) and 10% (cyan) of the maximum intensities (left). Quantification of halo size at 10% maximum intensity (right). Shown are the means of two independent experiments for each patient (LE1, LE2, AGS1 and AGS2) and WT controls (n=2). Error bars, s.e.m. ***P<0.001. (c) Nuclear RPA foci formation without and after detergent extraction (ext). ***P<0.05 versus WT; ^#P<0.001 versus non-extracted nuclei. (d) Total nuclear intensity (TNI) of Rad51. (e) Nuclear Rad51 foci in TREX1-deficient cells distribute more evenly throughout the cell cycle as shown by DNA (DAPI) content analysis. (c–e) WT controls (n=2). Box plots indicate the interquartile range (25–75%) from at least two independent experiments. Solid lines, median. Triangles, mean. Whiskers, 10–90th percentiles. *P<0.05; ***P<0.001 versus WT, Kruskal–Wallis test.

Figure 3. TREX1 is a tail-anchored protein.

(a) Schematic of fluorescently tagged TREX1 constructs used for fluorescence protease protection assay. (b) Confocal images of living HeLa cells co-expressing GFP-TREX1 and mCherry-TREX1. Proteinase K treatment after selective plasma membrane permeabilization leads to rapid dissipation of cytosolic GFP fluorescence, while mCherry resists proteolysis. Scale bar, 20 μm. (c) Schematic of TREX1 constructs containing a C-terminal bovine opsin fragment with two N-linked glycosylation sites (glyc) used for glycosylation reporter assay. (d) In vitro translated WT FLAG-TREX1-wt-glyc, but not mutant FLAG-TREX1-D272fs-glyc lacking the transmembrane domain, is glycosylated in the presence of microsomal membranes (memb.) as visualized by anti-FLAG immunoblotting. Glycosylation of WT TREX1 is removed by endoglycosidase H (endo H). (e) Electron microscopy of immunogold-labelled GFP-TREX1 in HeLa cells shows expression of TREX1 (black dots marked by arrows) along the outer nuclear membrane and the cytosolic site of the ER membrane. White arrowheads indicate the inner nuclear membrane. Scale bar, 200 nm. (f) Formation of organized smooth ER consisting of lamellar stacks, branching tubules or whorls in cells overexpressing GFP-TREX1. Scale bar, 200 nm. (g) Magnification of the inset in (f) shows an even distribution of GFP-TREX1 at the extraluminal site of the ER. Scale bar, 100 nm. co, no lysate control; cy, cytoplasm; nu, nucleus.

Figure 4. In situ biochemistry of TREX1-deficient cells.

(a) Interaction of YFP-TREX1 with endogenous cytosolic nucleic acids in HeLa cells analysed by FLIM-FRET. The fluorescence lifetime of YFP-TREX1 (right), but not of YFP alone (left), decreases significantly in the presence of Sytox Orange (+SO). The lifetime reduction of YFP-TREX1 is abolished in cells pretreated with DNase I (+SO+D). At least ten cells were measured per experiment. Means and s.d. of three independent experiments. ***P<0.001. (b) FCCS of HeLa cells microinjected with dsDNA488_647N. Mean cross-correlation and s.e.m. of three independent experiments measured at t<60 min. (WT, n=2). *P<0.05; **P<0.01; ***P<0.001. (c) Representative FCCS images. ATTO488 (A488, green), ATTO647N (A647N, red). Rhodamine B-dextran (yellow) marks the microinjected cytosol. Scale bar, 10 μm.

Figure 5. Constitutive DNA damage and type I interferon activation in TREX1-deficient cells.

(a) Alkaline single-cell electrophoresis depicting DNA damage in patient fibroblasts (LE, AGS) as shown by formation of longer comets with deformed nuclei compared with wild type cells (WT). Scale bar, 100 μm (left). The Olive tail moment (OTM) corresponds to the amount of DNA fragmentation (right). Means and s.e.m. of two independent experiments for each patient (LE1, LE2, AGS1, AGS2) and WT control cell lines (WT, n=2). ***P<0.001. (b) Increased pan-nuclear γH2AX-staining in TREX1-deficient fibroblasts determined by flow cytometry. Means and s.d. of at least three independent experiments for each patient (LE1, LE2, AGS1, AGS2) and WT controls (WT, n=5). **P<0.01; ***P<0.001. (c) Growth curves of TREX1-deficient fibroblasts (LE1, LE2, AGS1, AGS2) and WT cells (n=5). *P<0.05 (LE1) and **P<0.01 (LE2, AGS1, AGS2) versus WT at day 7. (d) Immunoblot analysis of unstressed patient fibroblasts (LE1, LE2, AGS1, AGS2) showing activation of p53 (p-p53; Ser15), p16, Chk1 (p-Chk1; Ser345) as well as IRF3 (p-IRF3). β-actin was probed as a loading control. (e) Senescent phenotype of patient fibroblasts (LE1, LE2, AGS1, AGS2) as shown by an increased percentage of β-galactosidase-positive cells (WT, n=4; left) and representative images of senescent cells (right). Scale bar, 100 μm. Means and s.d. of at least three independent experiments. *P<0.05; **P<0.01; ***P<0.001. (f) IFN-β secretion over 24 h. (WT, n=4). Means and s.e.m. of five independent experiments. *P<0.05; **P<0.01; ***P<0.001 versus WT.

Figure 6. TREX1-deficient cells are DNA repair-proficient, but more sensitive to type I interferon-dependent stimuli.

(a,b) Knockdown of TREX1 (siTREX1) in HeLa cells causes phosphorylation of p53 and IRF3 along with induction of IFN-β. This is abrogated in cells with additional knockdown of STING (siTREX+siSTING). siCtrl, negative control siRNA. ***P<0.001 versus siCtrl; ^###P<0.001 versus siTREX1. (c) Constitutive activation of IFN-β in TREX1-deficient fibroblasts (SLE, AGS) is further enhanced by siRNA-induced knockdown of RPA70 (siRPA) or Rad51 (siRad51) or both (siRPA+siRad51). This is suppressed by additional knockdown of cGAS (sicGAS). Shown are the means of one experiment run in triplicates out of two independent experiments. Error bars, s.d. *P<0.001 versus siCtrl. ^#P<0.001 versus knockdown of RPA70, Rad51 or RPA70 and Rad51, respectively, by analysis of variance followed by Tukey's post hoc test. (d) Repair kinetics of CPDs (left) and DSBs (right) in response to solar-simulated radiation (SSR). Means and s.e.m. of at least three independent experiments for each patient and wild-type control cell lines (WT, n=5). *P<0.05 (LE1, LE2) and **P<0.01 (AGS1, AGS2) immediately after ultraviolet exposure for CPDs, *P<0.05 (LE1, LE2) and **P<0.01 (AGS1, AGS2) at 6 h for DSBs. Although TREX1-deficient cells react more sensitive to SSR than WT cells, both DNA lesions are efficiently repaired within 24 h post irradiation. (e) IFNB upregulation in patient fibroblasts challenged with poly(I:C) and ultraviolet C irradiation. (WT, n=2). Means and s.e.m. of five independent experiments. *P<0.05; **P<0.01; ***P<0.001 versus WT. *P<0.05 (AGS2); **P<0.01 (LE1); ***P<0.001 (LE2, AGS1) for poly(I:C) versus poly(I:C) plus ultraviolet C. Mann–Whitney U-test.