Brain-derived microparticles induce systemic coagulation in a murine model of traumatic brain injury

Abstract

Traumatic brain injury (TBI) is associated with coagulopathy, although it often lacks 2 key risk factors: severe bleeding and significant fluid resuscitation associated with hemorrhagic shock. The pathogenesis of TBI-associated coagulopathy remains poorly understood. We tested the hypothesis that brain-derived microparticles (BDMPs) released from an injured brain induce a hypercoagulable state that rapidly turns into consumptive coagulopathy. Here, we report that mice subjected to fluid percussion injury (1.9 ± 0.1 atm) developed a BDMP-dependent hypercoagulable state, with peak levels of plasma glial cell and neuronal BDMPs reaching 17 496 ± 4833/μL and 18 388 ± 3657/μL 3 hours after TBI, respectively. Uninjured mice injected with BDMPs developed a dose-dependent hyper-turned hypocoagulable state measured by a progressively prolonged clotting time, fibrinogen depletion, and microvascular fibrin deposition in multiple organs. The BDMPs were 50 to 300 nm with intact membranes, expressing neuronal or glial cell markers and procoagulant phosphatidylserine and tissue factor. Their procoagulant activity was greater than platelet microparticles and was dose-dependently blocked by lactadherin. Microparticles were produced from injured hippocampal cells, transmigrated through the disrupted endothelial barrier in a platelet-dependent manner, and activated platelets. These data define a novel mechanism of TBI-associated coagulopathy in mice, identify early predictive markers, and provide alternative therapeutic targets.

© 2015 by The American Society of Hematology.

Figures

Figure 1

FPI-induced and BDMP-dependent coagulation. (A) Cerebral injury is visible in the left parietal lobe of a mouse 3 hours after FPI compared with the brain from a mouse subjected to sham surgery. (B) The clotting time of PPP collected from mice 3 hours after FPI or sham surgery was measured in a PS-dependent assay (n = 6, paired t test). (C) Clotting times were compared between PPP and homologous MPFP collected from mice subjected to FPI and sham surgery (n = 9, paired t test). (D-E) GFAP+ and NSE+ microparticles were detected by flow cytometry over time in blood samples from TBI and sham mice (n = 6 at each time point), and the values are after the subtraction of values from isotype IgGs (1-way ANOVA, n = 48, *P < .01, **P < .05). (F) Fractions of GFAP+ and NSE+ microparticles that express PS as measured by annexin V binding (a representative of 16 separate experiments).

Figure 2

Procoagulant activity of BDMPs from freeze-thawing injury. (A) (Upper) TEM image of BDMPs (bar = 200 nm). (Inset) A microparticle stained with an immune-gold-labeled GFAP antibody (bar = 100 nm). (Lower) TEM image of BDMPs with negative staining to enhance the view of membrane structures (bar = 200 nm). (B) BDMPs and platelet microparticles solubilized in an SDS lysis buffer were probed for NSE, GFAP, and von Willebrand factor (T, whole brain lysate; Pts, platelet lysate as control). (C) The expression of TF, PS, and both was measured for BDMPs (represents 20 separate measurements). (D) PS-dependent clotting time was measured in the presence of increasing numbers of BDMPs (n = 6, 1-way ANOVA, P < .001) and compared with those induced by 1.6 μg/μL of purified brain PS and PC, respectively. (E) Plasma clotting times induced by 25 000/μL BDMPs were measured in the presence of increasing concentrations of bovine lactadherin. (F) TF-dependent thrombin generation was measured in reaction containing either 1 pM TF or 25 000 BDMPs (representative of 3 separate experiments).

Figure 3

Coagulopathy induced by infusion of BDMPs into uninjured mice. (A) Blood samples were collected from mice infused with BDMPs or PBS and tested for PS-dependent clot formation (n = 12, 1-way ANOVA, **P = .036, *P < .001 compared with PBS). (B) Levels of fibrinogen were measured in the plasma samples collected for studies in A (n = 12, 1-way ANOVA, *P < .001 compared with PBS). Sections of formaldehyde-fixed lungs, kidney, and heart from mice injected with (C-E) BDMPs or (F-H) PBS were stained with phosphotungstic acid hematoxylin (dark blue; the dye also stains muscle cells). The left panel shows fibrin deposition in the microvasculature of the (C) lung, (D) kidney, and (E) stromatic vessels of the heart from mice injected with BDMPs (arrows). Arrowheads indicate intact walls of fibrin deposited vessels (bar = 25 μm). (F-H) Fibrin deposition was not detected in organs collected from mice receiving PBS (bar = 25 μm). Images are representative of 6 mice injected with BDMPs and 5 with PBS.

Figure 4

Production and transmigration of BDMPs. (A) Evans blue leaked from the vasculature of a mouse subjected to FPI, but significantly less from the one with sham surgery (representatives of 3 pairs of mice). (B) A representative image of hippocampal cells that were cultured for 7 days. (C) Cultured hippocampal cells were stimulated with 50 μM of the calcium ionophore A23187. The media collected (upper left) before and (upper right and lower left) after stimulation were analyzed for BDMPs by flow cytometry (n = 6, paired t test, P < .001). (D) HUVECs grown to confluence (left) become retracted and granulated (arrows) after being stimulated with 25 μM histamine (right; bar = 20 μm). (E) PKH26-labeled BDMPs detected in the bottom chamber by flow cytometry 3 hours after BDMPs were incubated with resting and histamine-activated (His) HUVECs in the presence and absence of 3 × 105/μL human live (Pts) or lyophilized (Lyo. Pts) platelets (n = 6/group, 1-way ANOVA, *P < .005 vs baseline). Serotonin (5 μM) also promotes microparticle transmigration in the absence of live platelets. (F) SEM images showing BDMPs (arrowhead) on the opposite side of the membrane from histamine-activated HUVECs that were incubated with BDMPs in the (a) presence and (b) absence of live platelets or (c) resting HUVECs incubated with BDMPs (bar = 5 μm).

Figure 5

Effect of BDMPs on platelets. (A) Human platelets incubated with PKH26-labeled BDMPs and a CD61 antibody (10 minutes at 37°C) were gated on CD61 positivity and analyzed for PKH26 fluorescence by flow cytometry (1-way ANOVA, n = 12, *P < .01 compared with baseline). (B) Scanning electron microscopic images of platelets immobilized onto fibrinogen and incubated with either (left) PBS (bar = 1 μm) or (right) BDMPs (bar = 1 μm). (C) Calcium influx in platelets incubated with 25 000/μL BDMPs for the indicated times (n = 6, repeated-measures ANOVA, *P < .001, the values are after background subtraction). (D) CD62p expression was measured on platelets incubated with 25 000/μL BDMPs for 30 minutes at 37°C in the presence and absence of type I collagen (n = 6, paired t test). (E) Annexin V binding to platelets (CD61+) treated with BDMPs was measured by flow cytometry (n = 6, 1-way ANOVA, *P < .001 compared with baseline). (F) Human platelets were monitored for aggregation in an optical aggregometer after 25 000/μL BDMPs were added to 100 μL PRP with and without collagen (n = 6, paired t test).