Circulating mitochondrial DNA is a proinflammatory DAMP in sickle cell disease

Abstract

The pathophysiology of sickle cell disease (SCD) is driven by chronic inflammation fueled by damage associated molecular patterns (DAMPs). We show that elevated cell-free DNA (cfDNA) in patients with SCD is not just a prognostic biomarker, it also contributes to the pathological inflammation. Within the elevated cfDNA, patients with SCD had a significantly higher ratio of cell-free mitochondrial DNA (cf-mtDNA)/cell-free nuclear DNA compared with healthy controls. Additionally, mitochondrial DNA in patient samples showed significantly disproportionately increased hypomethylation compared with healthy controls, and it was increased further in crises compared with steady-state. Using flow cytometry, structured illumination microscopy, and electron microscopy, we showed that circulating SCD red blood cells abnormally retained their mitochondria and, thus, are likely to be the source of the elevated cf-mtDNA in patients with SCD. Patient plasma containing high levels of cf-mtDNA triggered the formation of neutrophil extracellular traps (NETs) that was substantially reduced by inhibition of TANK-binding kinase 1, implicating activation of the cGAS-STING pathway. cf-mtDNA is an erythrocytic DAMP, highlighting an underappreciated role for mitochondria in sickle pathology. These trials were registered at www.clinicaltrials.gov as #NCT00081523, #NCT03049475, and #NCT00047996.

Conflict of interest statement

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Figures

Graphical abstract

Figure 1.

cfDNA is increased in patients with SCD. (A) Quantification of cf-nDNA in the cross-sectional cohort (upper panel) and paired samples (lower panel) by real-time qPCR assays of human genomic targets GAPDH and TERT. Upper panels: HCs, n = 13; SCD-Baseline, n = 72, and SCD-Crisis, n = 20. Data are sample mean ± SD. *P < .05, nonparametric Kruskal-Wallis test with Dunn’s multiple-comparison test. Lower panels: SCD-Baseline and SCD-Crisis pairs, n = 18. Circles represent the mean of each sample. *P < .05, **P < .01, nonparametric Wilcoxon matched-pairs signed-rank test. (B) qPCR assays were performed for human mitochondrial targets (MT-ND1 and MT-ND6), as well as for nuclear targets (GAPDH and TERT), and the cf-mtDNA/cf-nDNA ratio was calculated by normalizing the cycle threshold (ct) values with the mean ct value of GAPDH from HCs. Quantitation of cf-mtDNA/cf-nDNA ratio using the following mitochondrial and nuclear targets: MT-ND1/GAPDH, MT-ND1/TERT, MT-ND6/GAPDH, and MT-ND6/TERT. Error bars represent the sample mean ± SD. *P < .05, ***P < .0005, ****P < .0001, nonparametric Kruskal-Wallis test with Dunn’s multiple comparison test.

Figure 2.

Global mitochondrial hypomethylation in cfDNA samples. Bisulfite sequencing (BS-Seq) graphical output of mitochondrial DMRs in the cfDNA of patients with SCD. cfDNA from 7 longitudinal pairs (7 SCD-Baseline and 7 SCD-Crisis) of patients with SCD and an HC was extracted and treated with bisulfite, and the whole genome was sequenced for the BS-Seq analysis to analyze the levels of cytosine guanine dinucleotide (CpG) across the samples. DMRs with ≥3 CpGs and a mean methylation difference between groups of samples ≥10% are considered significant. Blue and red lines indicate mean percentage of methylation of SCD-Baseline and SCD-Crisis, respectively, across 7 pairs. The green line represents the percentage of mitochondrial methylation in the HC.

Figure 3.

Quantitation of mtDNA in RBCs of HCs, ASs, and patients with SCD. (A) Graph shows a significantly higher copy number of mtDNA in freshly acquired RBCs from patients who were SCD-Baseline (n = 2) compared with HCs and ASs (n = 3 each). (B) Confirmation of the findings from panel A with a separate set of HCs (n = 4) and SCD-Baseline subjects (n = 4). Whole blood from HCs, ASs, and SCD-Baseline subjects were stained with pan-leukocyte (CD45) and erythroid (CD71, CD235a) markers, followed by FACS. Errors bars represent the sample mean ± SD. *P < .05, 1-way analysis of variance with Tukey’s multiple-comparisons test (A), nonparametric Student t test (B).

Figure 4.

Differential mitochondrial mass, membrane potential, and superoxide production in RBCs from HCs and patients who were SCD-Baseline (n = 4 each). (A) HCs display a slight increase in the percentage of mature RBCs compared with patients with SCD, but the difference is not significant. (B) Immature RBCs (reticulocytes) were significantly higher in SCD patients. Mature RBCs from SCD patients showed a significant increase in mitochondrial mass (quantified as the fluorescent intensity of MTG) (C), the amount of mitochondrial superoxide (estimated with MitoSOX Red) (E), and mitochondrial membrane potential (quantified with MTDR) (G) compared with mature RBCs from HCs but there was no significant difference in mitochondrial mass (D), superoxide levels (F), or mitochondrial membrane potential (H) between SCD and HC reticulocytes. Errors bars represent the sample mean ± SD. *P < .05, Student t test.

Figure 5.

Sickle cell RBCs retain mitochondria. Confocal imaging of blood cells from a patient who was SCD-Baseline stained with mitochondrial (MTG; green), and nuclear (Hoechst; blue) markers. The stained live cells were plated onto 35-mm No. 1.5 poly-d-lysine–coated glass-bottom Petri dishes, and superresolution imaging was performed on the instant SIM. Note that the majority of cells that stained with MTG are anucleate (Hoechst negative). A 3-dimensional superresolution imaging of this figure showing mitochondrial retention in red cells is shown in supplemental Video 1 containing cells that are stained with mitochondrial markers MTG (green), and TMRM (red), markers that stain inactive (depolorized) and active (polarized) mitochondria respectively, and nuclear marker Hoechst (blue).

Figure 6.

Electron microscopy analysis of SCD RBCs. Ultrastructural analysis of RBCs using electron microscopy. RBCs from patient who was SCD-Baseline were processed using FIB-SEM. (A) Electron micrograph of peripheral blood erythrocytes from a patient with SCD showing mitochondria. Scale bars, 2 μM. (B) Three-dimensional rendering of mitochondria inside the sickle erythrocyte. Image segmented mitochondria are red. Scale bars, 10 μM. (C) Series of image frames from the FIB-SEM image stack. Slice thickness, 9 nm. Isotropic voxel size, 9 nm. (D) Outward morphology of sickle erythrocyte. Cell periphery, green. Scale bars, 10 μm.

Figure 7.

Induction of NET formation by gDNA and mtDNA and SCD plasma, with and without BX795. (A) Fluorescence microscopy images from a representative experiment showing NET formation following 7 hours of treatment with a similar concentration of gDNA purified from whole blood or mtDNA purified from platelets or 10% plasma with high cf-mtDNA from a patient with SCD. As a control, neutrophils were left untreated in RPMI 1640 or in RPMI 1640 containing AE Buffer (the buffer in which gDNA and mtDNA were eluted following purification with a QIAGEN kit). Original magnification ×20. (B) NET production (average number of NETs ± SD) induced with AE Buffer (n = 2; 1.67 ± 2.92), gDNA (n = 2; 0.80 ± 1.60), purified mtDNA (n = 5; 14.00 ± 9.70), or cf-mtDNA SCD plasma (n = 5; 18.90 ± 8.80). Representative fluorescence microscopy images (C; original magnification ×20) and NET counts (D) showing that a 4-hour pretreatment with 10 mM BX-795 decreased NET production after a 5 hour-induction with 500 mM purified mtDNA (n = 3; ***P = .0004) or cf-mtDNA in SCD plasma (n = 5; ****P = .0001), but it did not affect the basal NETosis with healthy plasma (H-Plasma; n = 3; P = .1170). n.s., not significant.