Extracellular vesicles package dsDNA to aggravate Crohn's disease by activating the STING pathway

Abstract

Crohn's disease (CD) is an intestinal immune-dysfunctional disease. Extracellular vesicles (EVs) are membrane-enclosed particles full of functional molecules, e.g., nuclear acids. Recently, EVs have been shown to participate in the development of CD by realizing intercellular communication among intestinal cells. However, the role of EVs carrying double-strand DNA (dsDNA) shed from sites of intestinal inflammation in CD has not been investigated. Here we isolated EVs from the plasma or colon lavage of murine colitis and CD patients. The level of exosomal dsDNA, including mtDNA and nDNA, significantly increased in murine colitis and active human CD, and was positively correlated with the disease activity. Moreover, the activation of the STING pathway was verified in CD. EVs from the plasma of active human CD triggered STING activation in macrophages in vitro. EVs from LPS-damaged colon epithelial cells were also shown to raise inflammation in macrophages via activating the STING pathway, but the effect disappeared after the removal of exosomal dsDNA. These findings were further confirmed in STING-deficient mice and macrophages. STING deficiency significantly ameliorated colitis. Besides, potential therapeutic effects of GW4869, an inhibitor of EVs release were assessed. The application of GW4869 successfully ameliorated murine colitis by inhibiting STING activation. In conclusion, exosomal dsDNA was found to promote intestinal inflammation via activating the STING pathway in macrophages and act as a potential mechanistic biomarker and therapeutic target of CD.

Conflict of interest statement

The authors declare no competing interests.

© 2021. The Author(s).

Figures

Fig. 1. Abnormally elevated exosomal dsDNA was positively correlated with the disease activity in murine colitis and active human CD.

A–C Classification of EVs in colon lavage of murine colitis. A Western blot of specific EVs markers: The simultaneous expression of three classic positive markers of EVs including Alix, CD63, and CD81, along with the absence of Calnexin, together identified EVs. The whole-cell lysate of CT26, a murine cell line, was used as a positive control. B Nanoparticle-tracking analysis of diluted EVs fractions was determined by ZetaView PMX 110 (Particle Metrix, Meerbusch, Germany). The distribution of EVs particle concentration (y axis) by size (x axis) was shown and 126.2-nm particles accounted for the highest proportion (98.5%). C Representative transmission electron microscope images of EVs fractions. The yellow arrow pointed to examples of EVs, which were cup-shape membrane-enclosed particles with diameters of 30–150 nm. D Absolute quantification of nDNA and mtDNA within EVs extracted from plasma and colon lavage of murine colitis. Hist1h3F gene and mtCOI gene were applied to represent nDNA and mtDNA respectively. Exosomal mtDNA and nDNA were significantly higher in the plasma and colon lavage of murine colitis. Murine colitis treated with GW4869 were also examined. n = 5–10/group. In the illustrations, healthy wild-type model, murine colitis, and murine colitis treated with GW4869 were abbreviated as WT, WT + DSS, and WT + DSS (GW4869 IP) respectively. Colon lavage was abbreviated as CL. E Classification of EVs isolated from plasma of active human CD, including western blot, nanoparticle-tracking analysis, and representative transmission electron microscope images. In the nanoparticle-tracking analysis, 117.7-nm particles accounted for the highest proportion (98.6%). In the representative transmission electron microscope image, the yellow arrow pointed to EVs. F Levels of exosomal nDNA and mtDNA from the plasma of CD patients were measured (n = 5/group). Both were significantly higher in patients with active CD. H3 clustered histone 7 gene and mtCOI gene were applied to represent nDNA and mtDNA respectively. G Correlations between the disease activity and levels of exosomal dsDNA in plasma in murine colitis and CD patients (Spearman’s rank correlation analysis). Disease activity index, Crohn’s Disease Activity Index, exosomal nDNA, and exosomal mtDNA were abbreviated as DAI, CDAI, exo-nDNA, and exo-mtDNA in the illustrations respectively. Data were displayed as mean values ± SD at least three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 2. EVs from damaged intestinal epithelial cells were internalized by macrophages in vitro.

A EVs presence in colons of human CD, exhibited by immunohistochemistry of CD63 (brown particles pointed by black arrows in the illustration), a transmembrane protein of EVs. The nuclei (blue) at the bottom of intestinal epithelial cells were pointed by black arrowheads. More EVs were stained in active CD and gathered in the intestinal epithelium. Colonoscopy images of corresponding patients were shown. B Transmission electron microscopy images of mitochondrial damages in colonic epithelia of murine colitis. Mitochondrial vesicular swelling and peripheral lucent zones were observed in the murine colitis model (abbreviated as WT + DSS in the illustration), compared to the wild-type healthy control. C Classification of EVs from LPS-damaged CT26 cells, including western blot, transmission electron microscopy, and nanoparticle-tracking analysis. The whole-cell lysate of CT26 was set as a positive control in the western blot experiment. In the nanoparticle-tracking analysis, 104.1-nm particles accounted for the highest proportion (98.5%). In the representative transmission electron microscope image, the yellow arrow pointed to EVs. D Labeled EVs were shown to be internalized by bone marrow-derived macrophages (BMDMs) after 15 h incubation. EVs from LPS-damaged CT26 cells were isolated and labeled with PKH26 (red). EVs from normal CT26 cells were used as control. BMDMs were treated with labeled EVs for 15 h and then observed under fluorescence confocal microscopy. F-actin (green) and nuclei (blue) of BMDMs were stained before the observation. EVs from damaged colon epithelial cells are abbreviated as c-EV in the illustration. All results were representative of at least three independent experiments.

Fig. 3. EVs activated the STING pathway and triggered macrophages to be pro-inflammatory in a dsDNA-dependent manner in CD.

A Trauma patients with no history of CD and no gastrointestinal symptoms were used as controls. Expressions of p-STING, STING, p-IRF3, IRF3, p-p65, and p65 in colon tissues of active CD patients and controls were determined by western blot. The activation of the STING pathway was shown in active CD. B Immunofluorescence co-staining of CD68, a macrophage marker, and phosphorylated STING (p-STING) was performed in the colonic mucosa of active human CD. Nuclei were counterstained with DAPI. Trauma patients with no history of CD and no gastrointestinal symptoms were used as controls. Scale bar, 20 μm. C–E EVs from LPS-damaged CT26 cells were incubated with wild type (WT) or STING−/− BMDMs. EVs were shown to trigger inflammation in macrophages by activating STING pathway. C Activation of STING pathway in BMDMs was determined by western blot. D, E mRNA levels of inflammatory cytokines, Arg1 and IL12p40 in BMDMs. The increased mRNA levels of IL-6, TNF-α, IFN-β, IL12p40 and decreased mRNA level of Arg1 showed that inflammation in macrophages was triggered by EVs from LPS-damaged CT26 cells. EVs from LPS-damaged CT26 cells are abbreviated as c-EV in the illustrations. F, G EVs from the equal number of damaged CT26 cells were assigned to two groups. One was lysed by sonication and digested with dsDNase so that exosomal dsDNA was digested thoroughly. The other group was directly treated with dsDNase to digest free dsDNA outside EVs, therefore exosomal dsDNA stayed intact. WT and STING−/− BMDMs were treat with the two groups of EVs. Exosomal dsDNA was abbreviated as exoDNA in the illustrations. F Activation of STING pathway in BMDMs stimulated by the two groups of EVs. The EVs treated with dsDNase without sonication still activated the STING pathway in BMDMs while EVs treated with dsDNase plus sonication failed to. G, H mRNA levels of inflammatory cytokines, Arg1 and IL12p40 in BMDMs. EVs treated with dsDNase plus sonication failed to trigger inflammation in BMDMs. Data were displayed as mean values ± SD at least three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 4. Blockade of EVs release by GW4869 inhibited intestinal inflammatory responses via inhibiting the activation of the STING pathway.

A–D The expressions of p-STING, STING, p-IRF3, IRF3, p-p65, and p65 in colon tissues from Wild type (WT), STING−/−, WT murine colitis, WT murine colitis treated with GW4869, and STING−/− murine colitis models were determined by western blot. WT murine colitis was abbreviated as WT + DSS in the illustrations. WT murine colitis treated with GW4869 was abbreviated as WT + DSS (GW4869 IP). STING−/− murine colitis was abbreviated as WT + DSS. The activation of the STING pathway was observed in the WT murine colitis model and was inhibited in WT murine colitis treated with GW4869. E, F The proportion of TUNEL-positive cells was quantified to estimate intestinal apoptotic level in WT, WT murine colitis, WT murine colitis treated with GW4869, and STING−/− murine colitis models. The level of apoptosis was higher in the WT murine colitis model and lower in WT murine colitis treated with GW4869, and STING−/− murine colitis models. Representative images were shown. Scale bar = 50 µm. G The expression of inflammatory cytokines including IL-6, TNF-α, and IFN-β in the homogenates of murine colon tissues was measured by ELISA (n = 5–9/group). All results were representative of at least three independent experiments. Data were displayed as mean values ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 5. Either GW4869 administration to inhibit EVs release or STING deficiency improved the disease prognosis of murine colitis.

A–E Comparison of disease prognosis among six experimental murine models (n = 11–20/group). WT (GW4869 IP): to examine possible side effects of GW4869, GW4869 was administrated in normal wild-type mice. WT + DSS (GW4869 IP): Murine colitis models treated with GW4869. STING−/− + DSS: STING knockout (STING−/−) mice were employed and induced by DSS to establish colitis. Indicators of disease activity and severity were examined including A Disease activity index (DAI), B percent changes in body weight, C survival rate, D lengths and representative images of colons, E histological score of colons and representative colon H&E images. Data were displayed as mean values ± SD at least three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 6. Model of exosomal dsDNA-mediated intestinal inflammation via activation of STING pathway in CD.

sEVs full of dsDNA, which could be secreted by damaged intestinal epithelial cells, were taken up by macrophages to activate the STING pathway and trigger inflammation. Inflammatory cytokines released by macrophages, such as IL-6, TNF-α and IFN-β, further damaged intestinal epithelial cells.