Cardiolipin-mediated procoagulant activity of mitochondria contributes to traumatic brain injury-associated coagulopathy in mice

Abstract

Cardiolipin (CL) is an anionic phospholipid located exclusively in the mitochondrial inner membrane. Its presence in blood indicates mitochondrial damage and release from injured cells. Here, we report the detection of CL-exposed brain-derived mitochondrial microparticles (mtMPs) at 17 547 ± 2677/μL in the peripheral blood of mice subjected to fluid percussion injury to the brain. These mtMPs accounted for 55.2% ± 12.6% of all plasma annexin V-binding microparticles found in the acute phase of injury. They were also released from cultured neuronal and glial cells undergoing apoptosis. The mtMPs synergized with platelets to facilitate vascular leakage by disrupting the endothelial barrier. The disrupted endothelial barrier allowed the release of mtMPs into the systemic circulation to promote coagulation in both traumatically injured and mtMP- or CL-injected mice, leading to enhanced fibrinolysis, vascular fibrin deposition, and thrombosis. This mtMP-induced coagulation was mediated by CL transported from the inner to the outer mitochondrial membrane and was blocked by the scavenging molecule lactadherin. The mtMP-bound CL was ∼1600 times as active as purified CL in promoting coagulation. This study uncovered a novel procoagulant activity of CL and CL-exposed mitochondria that may contribute to traumatic brain injury-associated coagulopathy and identified potential pathways to block this activity.

© 2016 by The American Society of Hematology.

Figures

Figure 1

Antibrain and anti-CL antibodies were detected in FPI mice. (A) Immunoreactivity to MBH of 1/1000 diluted plasma from FPI mice at different postinjury times (n = 6/point; repeated measures ANOVA, *P < .01 vs baseline before FPI). (B) Binding of IgGs purified from sham or FPI mice to MBH (n = 6/dose; paired Student t test, *P < .01). (C) Anti-brain IgGs (0.5 μg/mL) from FPI mice were incubated with increasing concentrations of CL for 30 minutes, and then with MBH for 30 minutes at RT (n = 5-7/dose; repeated measures ANOVA, *P < .01 vs IgG at 0 CL). (D, top) CL blotted to nitrocellulose membrane was probed with IgGs (0.25 μg/mL) from sham and FPI mice (10 days post-FPI) in the presence of increasing amounts of MBH. (Bottom) Phosphatidylcholine (PC), PS, and CL blotted onto nitrocellulose membrane (18 μg/spot) were probed with the CL antibody purified from plasma collected 10-14 days after FPI from TBI mice.

Figure 2

CL was procoagulant and microparticle-bound. (A) A structural comparison between CL and PS. (B) The time to clot of phospholipid-depleted plasma in the presence of purified CL, PS, or PC (n = 6; 1-way ANOVA, *P < .01 vs PC). (C) Clotting times of plasma collected at different points from C57BL/6J mice subjected to FPI or sham surgery (n = 6; 1-way ANOVA, *P < .01 vs sham mice). (D) Clotting time of plasma collected from non-injured C57CL/6J mice 30 minutes after CL injection (n = 6/group; 1-way ANOVA). (E) hematoxylin and eosin (HE, top) and phosphotungstic acid-haematoxylin (PTAH, bottom) stains of the lungs show perivascular accumulation of erythrocytes resulted from vascular leakage (top left) and intravascular fibrin deposition (arrow, bottom left) in mice injected with CL, but not those injected with PBS (right panels, 5-10 evaluations/treatment). (F) Anti-CL antibody detected CL in microparticles, not microparticle-free plasma (MPFP) collected from mice 6 hours after FPI (MPFP was prepared as previously described).

Figure 3

mtMPs were detected in plasma from FPI mice and contained surface-exposed CL. Transmission electron microscope images. (A) Intact (left) and membrane-disrupted (middle, arrow) mitochondria and mitochondria-embedded BDMP (right; asterisk, mitochondria) detected in plasma of FPI mice. (B) Mitochondria (asterisk), membrane microparticles (arrowhead), and (C) mitochondria-embedded membrane microparticles (*) from purified BDMPs. (D) A section of an uninjured mouse brain shows a dense perinuclear distribution of mitochondria (left, arrowhead: nuclear membrane). A locally enlarged image further shows mitochondria (asterisk) and endoplasmic reticulum (ER) and membrane-bound/free polyribosomes (arrow). (E) Total mtMPs and (F) mtDNA detected in plasma samples from sham and FPI mice. (G) Annexin V bound to more than 80% of mtMPs purified by TOM22 antibody (left), and the binding was abolished by EDTA (right). (H) Anti-CL antibody purified from FPI mice, but not a control IgG, bound purified mtMPs. (I, top) CL and PS captured to the nitrocellulose membrane were incubated with human annexin V (PBS and annexin V as controls). Annexin V bound to the phospholipids was detected by a polyclonal annexin V antibody. (Bottom) Allophycocyanin-conjugated annexin V bound to CL and PS coated on microbeads detected by flow cytometry (n = 3-6/group; paired Student t test).

Figure 4

mtMPs were procoagulant. (A) A dose-dependent acceleration of plasma clotting time induced by mtMPs and BDMPs (n = 3/point; repeated measures ANOVA, *P < .001 vs no microparticles). (B) This mtMP-induced coagulation was blocked by 200 nM lactadherin (n = 3/group; paired t test). (C) Plasma clotting induced by 25 μg/mL CL in the presence of increasing concentrations of lactadherin (n = 3; repeated measures ANOVA, * P < .01 vs no CL). Uninjured mice were injected with PBS, mtMPs, or BDMPs (both at 2.5 × 104/μL). Plasma samples were collected 30 minutes after injection to measure clotting time (D; n = 6/group; repeated measures ANOVA, *P < .01 vs PBS) and D-dimer (E; n = 6/group; repeated measures ANOVA, P < .01 vs PBS). (F) HE stain of the lungs from mtMP-injected mice shows erythrocyte accumulation in the extravascular space (*, top left) compared with PBS-injected mice (top right), and extensive fibrin deposition in the dilated pulmonary interstitial vessels (arrowheads, bottom left) compared with PBS-injected mice (top right). Microvascular fibrin-rich thrombosis is shown in the pulmonary vasculature (arrow, bottom right) of mtMP-injected mice. Images are representative of 6 mice/group.

Figure 5

mtMPs were released from cultured neurons and glial cells. (A) Human glioblastoma T98G (A) and neuroblastoma SH-SY5Y (B) cells before (left) and after (middle) stimulated with the calcium ionophore A23187 or LPS (right) (bar = 50 μm). (A-B, top) Morphological changes; (bottom) levels of mtMPs detected in conditioned media from these cells before and after stimulations. (C) MitoTracker Green+ glial and neuronal cell microparticles that were annexin V+ detected by flow cytometry. The relative levels of annexin V+/MitoTracker Green+ T98G cells and SH-SY5Y cells that were also positive for GFAP and NSE, respectively (n = 3-5; 1-way ANOVA, *P < .011 and **P < .005 vs resting cells). (D) Levels of mitochondrial DNA in the supernatants of T98G and SH-SY5Y cells at resting and after stimulation with either A23187 or LPS (n = 3-5; 1-way ANOVA, *P < .001 vs resting cells).

Figure 6

CL and mtMP activated ECs and disrupted their barrier function. HUVECs before (A,C) and after stimulation with CL (B, 25 μg/mL) or mtMP (D, 2.5 × 104/μL) for 2 hours at 37°C were stained for CD31 (PECAM1, green), VWF (red), and 4′,6-diamidino-2-phenylindole (blue) to mark cell–cell junction, endothelial cells, and the nucleus, respectively. Resting cells stained with VWF (E) and 4′,6-diamidino-2-phenylindole (F) served as controls (representative of 3-5 experiments; bar = 20 μm). (G) The conditioned media from cultured ECs before and after treated with CL or mtMPs were analyzed for VWF by immunoblot (top) and ELISA (bottom). (H) Confluent HUVECs on collagen-coated PET membrane were stimulated with CL or mtMPs in the presence and absence of fresh platelets for 3 hours at 37°C, followed by incubation with FITC-Dextran for 30 minutes at 37°C. Fluorescent dextran was detected in the bottom chambers (n = 3-4; 1-way ANOVA).