Research ArticleTHROMBOSIS

Neutrophil macroaggregates promote widespread pulmonary thrombosis after gut ischemia

See allHide authors and affiliations

Science Translational Medicine  27 Sep 2017:
Vol. 9, Issue 409, eaam5861
DOI: 10.1126/scitranslmed.aam5861

Rip n’ roll

Ischemia in critically ill patients can result in thrombosis of unrelated organs, which is partially due to neutrophil recruitment. Yuan et al. combined intravital microscopy of thrombosis after gut ischemia-reperfusion injury with samples from acute respiratory distress syndrome patients. They observed that rolling neutrophils grab and rip fragments from phosphatidylserine-expressing dying platelets, which leads to macroaggregates. These macroaggregates, in turn, induce thrombosis and were not able to be targeted by conventional therapies such as aspirin. Conversely, targeting the necrotic factor cyclophilin D did have beneficial effects. These studies reveal new thrombotic biology and suggest the development of alternatively targeted therapies to prevent distant organ injury.


Gut ischemia is common in critically ill patients, promoting thrombosis and inflammation in distant organs. The mechanisms linking hemodynamic changes in the gut to remote organ thrombosis remain ill-defined. We demonstrate that gut ischemia in the mouse induces a distinct pulmonary thrombotic disorder triggered by neutrophil macroaggregates. These neutrophil aggregates lead to widespread occlusion of pulmonary arteries, veins, and the microvasculature. A similar pulmonary neutrophil-rich thrombotic response occurred in humans with the acute respiratory distress syndrome. Intravital microscopy during gut ischemia-reperfusion injury revealed that rolling neutrophils extract large membrane fragments from remnant dying platelets in multiple organs. These platelet fragments bridge adjacent neutrophils to facilitate macroaggregation. Platelet-specific deletion of cyclophilin D, a mitochondrial regulator of cell necrosis, prevented neutrophil macroaggregation and pulmonary thrombosis. Our studies demonstrate the existence of a distinct pulmonary thrombotic disorder triggered by dying platelets and neutrophil macroaggregates. Therapeutic targeting of platelet death pathways may reduce pulmonary thrombosis in critically ill patients.


The gastrointestinal tract is a major component of the immune system, regulating local and systemic inflammatory responses (14). Prolonged ischemic injury to the intestines, which can occur as a consequence of severe infection (sepsis) (5, 6), major trauma (5), systemic hypoperfusion (1, 4, 6), acute pancreatitis (7), or direct intestinal ischemia-reperfusion (I/R) injury (8), can lead to the development of a systemic inflammatory response that promotes remote organ injury (1, 2, 4, 6, 9). Although the liver (10), kidneys (11), heart (12), lung (13), and intestines are commonly affected by severe intestinal ischemia, acute respiratory distress syndrome (ARDS) is the earliest and arguably most important component of this syndrome (6, 14, 15). Severe lung injury with hypoxemia exacerbates hemodynamic instability (16) and intestinal ischemia (17), which establishes a potentially hazardous cycle of ongoing systemic inflammation and further clinical deterioration (4, 6, 18).

Widespread thrombosis throughout the pulmonary circulation is also a hallmark feature of ARDS, leading to lung hypoperfusion and pulmonary hypertension (16, 19). This thrombotic response is unusual, involving both the macro- and microcirculation of the pulmonary arterial and venous systems (16). Pulmonary thrombosis occurs early in the development of ARDS, impairing gas exchange and increasing pulmonary vascular resistance (16, 20). Persistent thrombosis in the pulmonary vasculature is a critical issue clinically because it is often fatal (>90% mortality) (21) and existing antithrombotic approaches, including anticoagulants and fibrinolytic agents, have limited efficacy at improving lung perfusion (19, 22).

Numerous hypotheses have been advanced to explain the link between ischemic gut injury and systemic inflammation (1, 23); however, there is currently limited understanding of mechanisms whereby ischemic gut injury triggers thrombosis in distant organs. Endothelial cell dysfunction in the intestinal microvasculature, as a direct result of ischemic endothelial injury, is considered a central initiating event for systemic inflammatory responses (1, 2, 4). Ischemic endothelial cells are highly reactive to leukocytes and platelets (2325) and are also potent inducers of blood coagulation, leading to fibrin deposition throughout the intestinal microvasculature (24). Adhesive interactions between leukocytes, platelets, and endothelial cells can generate a broad range of prothrombotic (2629) and proinflammatory (30) molecules. Proinflammatory molecules released from the gut perturb endothelial cells in other organs, potentially inducing widespread activation of the innate immune and hemostatic systems (1, 31). Consistent with this, depletion of neutrophils (32) or inhibition of blood coagulation (33, 34) can reduce remote organ injury after intestinal I/R injury, and recent studies suggest that platelets can also play an important role in promoting pulmonary thrombosis after gut ischemia (35, 36). Here, we have identified a previously unrecognized thrombotic mechanism triggered by dying platelets and neutrophil macroaggregates that leads to the development of a mixed arterial-venous thrombosis disorder in the lung.


Gut ischemia induces a neutrophil-dependent thrombotic response in mouse lungs

To investigate the impact of gut ischemia on vascular perfusion defects in the lung, we subjected mice to intestinal (gut) I/R injury (60-min ischemia followed by 120-min reperfusion), and we harvested the lungs for histological analysis. Extensive fibrin-rich thrombi were present throughout the pulmonary vasculature (Fig. 1, A and B, and fig. S1A). A notable feature of these thrombi was the presence of large intravascular leukocyte clusters (macroaggregates) containing up to 20 leukocytes in a single cross section (Fig. 1A). The leukocyte macroaggregates were common in both the pulmonary venous and arterial systems and were typically associated with fibrin (fig. S1, A and B). To examine whether the presence of leukocyte macroaggregates correlated with pulmonary vascular obstruction, we developed a lung casting technique using Microfil that enabled selective perfusion of pulmonary arteries or veins. In sham-treated animals, the entire pulmonary arterial and venous system remained patent down to the level of capillaries (Fig. 1, C and D). However, in mice subjected to intestinal I/R injury, marked perfusion defects were observed in both the arterial and venous systems of the lung, with occluded vessels ranging from small arterioles and venules (10 to 40 μm in diameter) to larger arteries and veins (~100 μm in diameter) (Fig. 1, C and D). Confocal imaging of the Microfil-perfused lung vasculature revealed that larger neutrophil macroaggregates were located at sites of vascular obstruction (Fig. 1, E and F). These leukocytes were myeloperoxidase (fig. S2A) and Gr-1–positive (Gr-1+) (Fig. 1, E and G), indicating that they predominantly consisted of neutrophils. Consistent with a critical role for neutrophils in promoting vascular plugging, depletion of neutrophils using an anti–Gr-1 antibody before ischemia eliminated thrombosis in the lungs after gut I/R injury (Fig. 1, H and I). These findings indicate that gut I/R injury in the mouse triggers a neutrophil-rich thrombotic response that causes widespread occlusion of the pulmonary arterial and venous systems.

Fig. 1. Gut I/R injury induces the formation of occlusive neutrophil-rich thrombi in the pulmonary vasculature.

Mice were subjected to gut I/R injury or sham operation. (A) Representative hematoxylin and eosin (H&E) staining comparing intravascular leukocyte aggregates and fibrin in lungs of I/R-injured and sham-operated mice. (B) Quantification of the number of pulmonary intravascular leukocyte aggregates, normalized for the surface area of lung sections (I/R, n = 14; sham, n = 8). (C and D) Lungs were flushed and perfused with Microfil to identify defective vascular perfusion. (C) Representative photographs of the arterial and venous vasculature of the left lung lobe or (D) phase contrast images of the indicated lung vessel branches depicting vascular perfusion (vessel branches are color-coded with corresponding vessel sizes indicated) (I/R, n = 7; sham, n = 3). (E and F) Mice were administered DyLight 647–anti–Gr-1 antibody, before gut I/R injury and lung Microfil perfusion, to identify colocalization of neutrophil aggregates (red) and defective Microfil perfusion (blue) in the pulmonary arteries and veins. Confocal images (E) are from one representative experiment (vessels, yellow dotted lines), and the percentage (%) of blocked vessels with associated neutrophil aggregates was quantified (F) (I/R-artery, n = 5; I/R-vein, n = 3). (G) Representative confocal image depicting neutrophil aggregates (red) in the pulmonary circulation (green, collagen autofluorescence) after gut I/R injury. (H and I) H&E staining (H) and quantification (I) of pulmonary intravascular fibrin formation in I/R-injured mice with or without preneutrophil depletion (anti–Gr-1 and RB6-8C5). Scale bars, 50 μm (A), 1000 μm (C), 100 μm (D), 20 μm (E), 10 μm (G), and 50 μm (H). Error bars represent means ± SEM. *P < 0.05.

Neutrophil-rich thrombi occur in the lungs of ARDS patients and in the splanchnic circulation of mice with gut I/R injury

Intestinal ischemia is common in patients with severe ARDS (14, 17, 37), and the degree of gut hypoperfusion correlated with the extent of lung injury and pulmonary thrombosis (16, 20, 38). Thrombi in the pulmonary circulation of ARDS patients are present in both the macro- and microcirculation of the lungs (15). To gain insight into the cellular composition of thrombi in ARDS patients, we performed histological analysis on the pulmonary microvasculature of 12 postmortem lung specimens (table S1). These studies confirmed extensive fibrin-rich thrombi throughout the pulmonary vasculature (Fig. 2A and fig. S2B). These thrombi contained prominent neutrophil macroaggregates with up to 30 neutrophils per aggregate (Fig. 2A). Aggregates were widespread, occurring in 20 to 60% of pulmonary vessels examined (Fig. 2B). In contrast, neutrophil macroaggregates were uncommon in postmortem lung specimens from humans with acute pulmonary edema or in explants from patients with emphysema (Fig. 2, A and B, fig. S2B, and table S2). Similar to our findings in the mouse, neutrophil macroaggregates were present in medium-sized pulmonary arteries and veins (200 to 300 μm in diameter) (Fig. 2C). These studies indicate that a mixed arterial-venous thrombotic response involving extensive neutrophil aggregation can also develop in the human lungs.

Fig. 2. Neutrophil-rich mixed arterial-venous thrombi in the lungs of ARDS patients and splanchnic circulation of ischemic mice.

(A to C) Postmortem lung specimens from patients with ARDS or acute pulmonary edema (APE) or from explanted lungs from emphysema (EMP) patients. (A) H&E and Carstair’s staining of ARDS and EMP lung specimens to detect intravascular neutrophil aggregates (right, H&E, arrows) and associated fibrin formation (left, Carstair’s). (B) The number of pulmonary vessels (%) containing neutrophil aggregates in ARDS, APE, or EMP patients was quantified (ARDS, n = 12; APE, n = 11; EMP, n = 10). (C) Carstair’s staining depicting neutrophil-rich thrombi in both arteries and veins of ARDS specimens. ns, not significant. RBCs, red blood cells. (D to G) Mice were administered phycoerythrin (PE)–Gr-1 antibody before gut I/R injury. (D) Fluorescence images depicting neutrophil aggregates (red) in mesenteric veins (dotted line) after I/R injury. (E) Fluorescence images depicting neutrophil aggregates (red) in the systemic arterial (left ventricle of the heart) and venous (IVC) blood after sham operation or gut I/R injury. (F) Number of aggregated versus single neutrophils in mesenteric veins 30 to 90 min after I/R injury (n = 6). (G) The impact of ischemia time on neutrophil aggregate formation in mesenteric veins (n = 3 to 4). *P < 0.05; **P < 0.01; ***P < 0.005.

To investigate whether neutrophil aggregation induced by gut ischemia occurred in the splanchnic circulation, we performed confocal and multiphoton intravital microscopy on the intestinal, mesenteric, and portal veins of mice. Intravascular neutrophil aggregation induced by gut I/R injury was widespread, occurring throughout the venous circulation of the intestines, mesentery (Fig. 2D and fig. S3A), portal vein (fig. S3B), and inferior vena cava (Fig. 2E). Neutrophil macroaggregates were also present in the left ventricle of the heart, suggesting that these aggregates also form in the pulmonary venous circulation of mice (Fig. 2E). Notably, up to 50% of all Gr-1+ leukocytes in the mesenteric veins participated in the formation of rolling aggregates (Fig. 2, F and G), with aggregate size varying between 2 and >10 cells (Fig. 2, D and E, and fig. S3, A and B). Real-time analysis of neutrophil aggregation revealed that these aggregates formed within minutes of blood reperfusion and persisted throughout the entire reperfusion period (up to 2 hours). The number and size of neutrophil aggregates correlated closely with the duration of ischemia (Fig. 2G) but was not associated with increases in the circulating neutrophil count (fig. S3C). Consistent with previous reports (39, 40), we confirmed no significant increase in circulating endotoxin levels after gut I/R injury. Moreover, pretreating mice with broad-spectrum antibiotics to promote gut sterilization did not prevent the formation neutrophil macroaggregates (fig. S3, D to F), indicating that these aggregates were unlikely to be generated by gut-derived bacterial endotoxin.

Neutrophil aggregation is platelet-dependent

Intestinal I/R injury is associated with platelet activation and formation of microvascular thrombi in the intestinal microcirculation (24, 25, 36). Depletion of platelets using an anti-GPIbβ antibody eliminated neutrophil aggregation induced by gut ischemia (Fig. 3A). Furthermore, costaining for platelet-specific markers revealed the presence of platelets within the neutrophil aggregates in both the mesenteric (Fig. 3B) and pulmonary (Fig. 3C) vasculature. Similarly, all neutrophil aggregates in the pulmonary circulation of patients with ARDS costained for the presence of platelets (Fig. 3D), indicating that a similar phenomenon may occur in humans. P-selectin–PSGL-1 (P-selectin glycoprotein ligand-1) bonds initiate adhesive interactions between platelets and neutrophils. Consistent with this, platelets adherent to intestinal arterioles and venules after gut I/R injury, as well as platelets contained within neutrophil aggregates in the mesentery and pulmonary circulation, expressed surface P-selectin (fig. S4, A and B). Moreover, neutrophil aggregates in the mesentery were eliminated in P-selectin–deficient (P-sel−/−) mice and in bone marrow–transplanted mice that lacked platelet P-selectin (P-selPlt−/−) (Fig. 3A).

Fig. 3. Neutrophil-rich thrombus formation is platelet-dependent.

(A to C) C57Bl/6 mice or the indicated genotype (A) were administered PE–Gr-1 and DyLight 647–GPIbβ antibodies before gut I/R injury. (A) Quantification of the impact of platelet depletion (C57BL/6JPlt-depleted), P-selectin deficiency (P-sel−/−), and hematopoietic P-selectin deficiency (P-selPlt−/−) on neutrophil aggregate formation in mesenteric veins during I/R injury (n = 3). (B and C) Confocal images of heterotypic platelet-neutrophil aggregates in mesenteric veins (B) and pulmonary vasculature (C) after I/R injury. (D) Histological immunostaining of ARDS specimens depicting platelets (integrin αIIb, brown, yellow arrowheads) within intravascular neutrophil aggregates (blue) in the lung vasculature (marked with yellow dotted lines). Scale bars, 50 μm (B), 10 μm (C), 200 μm (D, left), 20 μm (D, right). Error bars represent means ± SEM. **P < 0.01.

Neutrophil aggregation is induced by phosphatidylserine-expressing platelets

To investigate whether platelets were sufficient to induce neutrophil aggregation, we perfused human neutrophils over preformed human platelet thrombi ex vivo. Platelet thrombi were ineffective at supporting neutrophil macroaggregate formation [fig. S5, A and B (left)]; however, when costimulated with potent platelet agonists [thrombin (Thr) and collagen-related peptide (CRP)], these thrombi supported the formation of large neutrophil aggregates (10 to 40 cells) on >50% of the thrombi [fig. S5, A and B (left and right)]. A similar phenomenon was observed in vivo because platelet thrombi stimulated with locally injected Thr/CRP, but not the weak agonist adenosine diphosphate (ADP), supported neutrophil aggregation (Fig. 4A and movie S1). This is despite comparable levels of P-selectin expression and neutrophil recruitment under both experimental conditions (fig. S5, C and D). Notably, the induction of neutrophil aggregation by Thr/CRP was considerably delayed (maximal 30′ to 40′ after agonist injection) (fig. S5E), relative to the rapid increase in neutrophil recruitment (fig. S5D). Similar findings were apparent when thrombi were stimulated with the PAR4 (protease-activated receptor 4) agonist peptide and CRP, indicating that this phenomenon was unlikely to be related to fibrin generation by microinjected Thr. We investigated the possibility that neutrophil aggregation relies on a late potent platelet activation event, such as the surface exposure of phosphatidylserine (PS). Injection of Thr/CRP, but not of ADP, induced a high level of platelet PS exposure on the surface of thrombi (Fig. 4B), and neutrophils forming aggregates on Thr/CRP-treated thrombi incorporated PS+ platelets in the confines of the developing aggregate in vivo (Fig. 4C) and ex vivo (fig. S5, F and G).

Fig. 4. Neutrophil aggregation is selectively induced by PS+ platelets.

(A to C) Mice were administered indicated fluorescence probes (A and C) or Alexa 488–annexin V (PS+ platelets) and DyLight 647–anti-GPIbβ antibodies (platelets) (B) before needle puncture of mesenteric veins, followed by local microinjection of either ADP or Thr/CRP. (A) Representative confocal images depict neutrophil aggregate formation and detachment (red) from ADP or Thr/CRP thrombi (green) ~30 min after agonist injection. (B) Annexin V binding to thrombi (PS+ platelets) at the indicated times after agonist injection (n = 3). (C) Confocal images depicting PS+ platelets (green/yellow) within detaching neutrophil aggregates (red) from Thr/CRP-stimulated thrombi. (D and E) Confocal images depicting neutrophil aggregates (red) anchored by PS+ platelets (green/blue, cyan) in mesenteric veins, as demonstrated by channel overlays (white and cyan in magnified images, respectively) (D), and in pulmonary vasculature (dotted line) (E) after gut I/R injury. (F) Confocal images depicting PS+ platelet formation (green/blue, cyan) in intestinal microvasculature during gut I/R injury. (G and H) Percent (%) of intestinal vessels containing PS+ platelets (G) and the percent of platelets in a PS+ state in the intestinal microvasculature (H) after I/R injury [I/R, n = 4 (G); sham, n = 3 (G); I/R, n = 5 (H)]. (I) Confocal image depicting PS+ platelets [green/blue, cyan (arrows)] adherent to pulmonary vasculature (dotted line) after gut I/R injury. (J) Confocal images depict occlusive neutrophil aggregates (green) and colocalized fibrin (red) and platelets (blue) in the pulmonary vasculature 2 hours after gut I/R injury (n = 5). Scale bars, 50 μm (D), 20 μm (E), 100 μm (F), 10 μm (I), and 50 μm (J). Error bars represent means ± SEM. *P < 0.05; *P < 0.01; ***P < 0.005; ****P < 0.0001.

To confirm the relevance of these findings to neutrophil aggregation induced by intestinal I/R injury, we examined whether platelets within the confines of neutrophil aggregates were PS+. All Gr-1+ leukocyte aggregates costained positively for PS+ platelets in the mesenteric (Fig. 4D and movie S2), pulmonary (Fig. 4E), and intestinal (fig. S6A) vasculature. The PS staining was of platelet origin because the PS+ spots costained with anti-GPIbβ antibody (Fig. 4D, annexin V/platelet overlay). Analysis of the origin of PS+ platelets revealed extensive annexin V staining of platelets adherent to the intestinal microcirculation (Fig. 4F and fig. S6B), particularly in the more severely injured areas of the gut. Up to 25% of vessels (primarily postcapillary venules) contained PS+ platelets (Fig. 4G), and 28% of all platelets bound annexin V (Fig. 4H). PS+ platelet deposition was also noted throughout liver sinusoids (fig. S6, C and D) and the pulmonary vasculature (Fig. 4I) after gut I/R injury, consistent with the possibility that inflammatory mediators released from the ischemic gut induce changes in the vasculature of distant organs (1, 31). Notably, formation of neutrophil aggregates correlated with increased levels of circulating PS+ platelet-neutrophil complexes in the portal and systemic arterial and venous circulation (fig. S6, E and F). These findings indicate that gut ischemia is a potent inducer of platelet PS+ exposure on the endothelium in the intestines, liver, and lungs, leading to the formation of neutrophil aggregates.

To investigate relationships between neutrophil aggregation and fibrin generation, we performed confocal or intravital microscopy on the pulmonary circulation using a fibrin-specific antibody. Fibrin formation occurred exclusively at sites of neutrophil aggregation in the pulmonary circulation (Fig. 4J). Notably, we could not detect the presence of procoagulant neutrophil extracellular traps (NETs) within the leukocyte aggregates (fig. S7A), indicating that these structures were unlikely to play a major role in promoting fibrin formation. Furthermore, fibrin was never detected on rolling neutrophil aggregates in mesenteric veins or in nonocclusive aggregates in lungs (fig. S7, B and C), raising the possibility that fibrin was likely to form when vessels were severely blocked by neutrophil macroaggregates. Consistent with this, analysis of vessel patency using fluorescently labeled dextran revealed that vessels containing fibrin-rich neutrophil aggregates were occluded (fig. S7D). Similarly, real-time confocal microscopy in the mesenteric circulation confirmed that the vessels plugged with large neutrophil aggregates had complete cessation of blood flow (fig. S7E). These findings suggest that fibrin formation is likely initiated at sites of neutrophil aggregation, thereby consolidating vascular occlusion.

Remnant dying platelets induce neutrophil macroaggregation

Platelet PS exposure after potent stimulation is associated with plasma membrane instability, leading to the formation of flow-induced protrusions (FLIPRs) (41) and shed small microparticles (MPs) (Fig. 5A and fig. S8, A and B) (42). Platelet microparticles bind and activate neutrophils, increasing their prothrombotic and proinflammatory functions (43). The large remnant cell body of PS+ platelets (Fig. 5A) can also support neutrophil adhesion (44), although the functional relevance of this interaction and its contribution to the induction of neutrophil macroaggregation remains unclear. To examine this, we stimulated spread human platelets with high-dose Thr and CRP to induce microparticle release, removed the microparticles, and then perfused human neutrophils over the remnant PS+ platelet monolayers. Prominent neutrophil macroaggregates formed, with aggregates containing up to 30 to 40 neutrophils (Fig. 5B). This process did not occur on unstimulated platelet monolayers (PS) (Fig. 5C). To assess the impact of microparticles, we collected microparticles from the Thr/CRP-stimulated platelet monolayers and then coperfused them with neutrophils over PS platelet monolayers. Microparticle coperfusion led to infrequent and small neutrophil aggregates (two to four cells per aggregate), even when microparticles were concentrated 10- to 100-fold (Fig. 5D). In the presence of microparticles, only a small percentage of rolling neutrophils participated in the formation of neutrophil aggregates (3% with 1× MP and 15% with 100× MPs), whereas with remnant PS+ platelet monolayers, up to 60% of rolling neutrophils contributed to the formation of aggregates (Fig. 5D). Confocal imaging revealed the incorporation of large membrane fragments into the forming neutrophil aggregates from the surface of remnant PS+ platelets (Fig. 5E, left). Moreover, in some aggregates, up to 80% of the periphery of neutrophils were covered with PS+ platelet remnants, whereas less than 10% of the perimeter of neutrophils were coated by shed MPs after coperfusion over PS monolayers (Fig. 5E, right). These findings suggest that large membrane fragments from remnant PS+ platelets play an important role in inducing neutrophil macroaggregation.

Fig. 5. Remnants of PS+ platelets induce neutrophil macroaggregation.

(A) Representative differential interference contrast (DIC) images depicting spread human platelets pre-Thr/CRP stimulation (Thr/CRP 0′), remnant platelets post-Thr/CRP stimulation (Thr/CRP 9.3′) after adhesion to fibrinogen, and shed microparticles from Thr/CRP-stimulated platelets in suspension. Scale bar, 1 μm. (B and C) Representative phase contrast images depicting neutrophil macroaggregate formation on remnant PS+ platelets (B) or PS platelets (C) at the indicated shear rates (n = 3). (D) Quantification of the percentage of adherent neutrophils aggregated (left) and size of neutrophil aggregates formed (right) after neutrophil perfusion over remnant PS+ platelets, relative to microparticle coperfusion (1× or 10 to 100× MPs) over PS platelets. Error bars represent means ± SEM (n = 3). ****P < 0.001. (E) Representative confocal images depicting the extent of incorporation of large remnant PS+ platelet membrane fragments within aggregating neutrophils on remnant PS+ platelets or after microparticle coperfusion (1× MP) over PS platelets (n = 3 to 5). Scale bars, 10 μm.

Neutrophils drag and rip membranes from PS+ platelets ex vivo and in vivo

A hallmark feature of potently activated PS+ platelets is the proteolytic disassembly of the actin cytoskeleton (Fig. 6A), resulting in membrane swelling and the adoption of a characteristic “balloon-like” appearance (45, 46). Consistent with this, exposing remnant PS+ spread platelets to hemodynamic shear stress resulted in membrane deformation and fragmentation (Fig. 6B). Membrane fragmentation commenced at a wall shear stress of 1.2 Pa (1800 s−1), with about 50% of the PS+ platelets exhibiting membrane failure at 4.8 Pa (7200 s−1) and 100% at 19.2 Pa (28,800 s−1) (Fig. 6, B and C). Static force balance analysis (47) indicated that membrane failure commenced at drag forces of 4.5 pN/μm (1800 s−1), with all platelet membranes disrupted at ~72 pN/μm (28,800 s−1) (Fig. 6C). As a consequence of this membrane instability, neutrophils perfused over PS+ spread platelets pulled large membrane fragments (up to 1000 nM) from the platelet surface (Fig. 6D), leaving residual platelet bodies behind (Fig. 6E). Similarly, real-time fluorescence microscopy during neutrophil perfusion over partially spread platelets demonstrated that rolling neutrophils “ripped” large membrane fragments from the PS+ remnants or “dragged” entire remnants from the matrix. Quantitatively, 53.9% of PS+ remnant platelets were either ripped (40%) or dragged (13.9%) within 2 min of neutrophil perfusion (Fig. 6, F and G, and movie S3). The ripped and dragged fragments physically wrapped around neutrophils and formed “adhesive bridges” between adjacent aggregating neutrophils (Fig. 6H). A similar membrane-pulling process was also readily apparent on PS+ thrombi, with multiple PS+ remnant platelets promoting neutrophil macroaggregation (fig. S8C). Calculation of the drag forces imposed on platelet remnant membranes by rolling leukocytes (4749) revealed a minimum of ~10-fold increase in platelet membrane tension when neutrophil adhesion is distributed evenly over the entire surface of the spread platelet remnant (~10 μm in diameter) at 150 s−1, increasing to ~100-fold when pulling localized membrane fragments from the remnant surface (~100 nm) (fig. S8D), thus providing a mechanistic explanation for the ability of rolling neutrophils to extract membrane fragments at relatively low wall shear rates (<100 s−1).

Fig. 6. Rolling neutrophils extract membranes from fragile remnant PS+ platelets.

(A) Fluorescence and DIC images depicting the level of filamentous actin (phalloidin) in remnant PS+ human platelets (annexin V). (B and C) DIC images demonstrate remnant platelet membrane deformation and detachment (B), quantification of platelet detachment, and calculated drag forces at the indicated shear rates (C) (means ± SEM) (n = 9). (D and E) Representative scanning electron microscopy (SEM) images depicting the size of membrane fragments (MFs) pulled by neutrophils from spread remnant platelets (D) and the integrity of PS+ and PS platelet membranes after neutrophil perfusion (E). (F and G) Representative fluorescence images depict the ripping and dragging of remnant PS+ platelets (green) by rolling neutrophils (*, red) from nonspread platelets (F) (ripped platelet, yellow dotted circles and white arrows; dragged platelet, yellow and white dotted circles and white arrows) over the indicated time frames and percent of total PS+ platelets being ripped or dragged (means ± SEM) (n = 3) (G). (H to M) Confocal and SEM images depicting the ripping of remnant PS+ platelet membranes by rolling neutrophils (*, red) from nonspread platelets (green) (H) over the indicated time frames (yellow and white circles and yellow arrows) and spread platelets (J to M), platelet membrane wrapping (I and J, white arrows), and bridging adjacent neutrophils (K and L, white arrows). (M) Representative fluorescence image depicting the extensive surface coating of aggregating neutrophils (hollow and unlabeled) by remnant PS+ platelet membranes (fluorescent) after perfusion over spread PS+ platelets. Scale bars, 3.8 μm (D, left), 1 μm (D, middle), 2 μm (D, right), 2.5 μm (E, left), 2 μm (E, right), 10 μm (F), 10 μm (H), 1 μm (J), 2 μm (K, left), 1 μm (K, right), and 10 μm (M). 2D, two-dimensional.

To examine the significance of neutrophil ripping and dragging of PS+ remnant platelet membranes for neutrophil aggregation in vivo, we performed intravital imaging of neutrophil-thrombus interactions in mesenteric veins at magnifications that can visualize large membrane fragments (>500 nm). Neutrophils rolling over PS+ thrombi ripped large membrane fragments from PS+ platelets or dragged entire PS+ remnants from the thrombus surface, leading to neutrophil macroaggregate formation (Fig. 7A, mesenteric, and movie S4). Similar neutrophil ripping and dragging processes were also noted in the gut microcirculation (Fig. 7B and movie S5), pulmonary vasculature (Fig. 7C), and liver sinusoids (fig. S8E) after gut IR injury. The importance of neutrophil drag forces for platelet membrane extraction was underscored by the inability of slow rolling neutrophils in areas of sluggish blood flow to drag PS+ platelets from the surface of thrombi, whereas rapidly rolling neutrophils were highly effective at extracting platelet membranes. Consistent with the requirement for neutrophil adhesion to the surface of thrombi for subsequent ripping and dragging of platelet membranes, P-selectin−/− mice, which do not support neutrophil-platelet adhesion, displayed a marked reduction in the detachment of large PS+ platelet fragments from the surface of thrombi at sites of mechanical vessel injury (Fig. 7D). Notably, 100% of neutrophil aggregates forming in vivo contained large PS+ platelet fragments (>500 nM) (Fig. 7E). The importance of large platelet membrane fragments to induce neutrophil aggregation was further underscored by the inability of small shed microparticles in plasma to induce neutrophil macroaggregation in the mesenteric or pulmonary circulations when injected into naïve mice (Fig. 7F).

Fig. 7. Rolling neutrophils rip and drag PS+ platelet membranes in vivo.

(A to E) C57BL mice or mice of the indicated genotype were administered appropriate fluorescence probes and then subjected to needle puncture of mesenteric vein, followed by local Thr/CRP microinjection (A and D) or gut I/R injury (B, C, and E). (A and B) Representative confocal images depicting ripping (A, top) and dragging (A, bottom, and B) of PS+ platelets [P1–3, cyan-colored cells (yellow outline)] by rolling neutrophils [N1–3, red cells (white outline)] and PS+ platelets bridging adjacent neutrophils (A, bottom) on thrombi 30′ after agonist injection (A) or in intestinal vasculature after I/R injury (B) over the indicated time frames. (A) Top: Large platelet fragments ripped from P1 and P2 by N1. Bottom: P1 (tracked by yellow arrow) dragged by N1, P2 dragged by N2, and P3 dragged by N3, culminating in neutrophil macroaggregation by bridging of N1 and N2 via P2 and of N1 and N3 via P1 and P3. (B) P2 dragged by N1-P1 rolling complex (tracked by yellow arrow). (C) Representative confocal images showing neutrophils interacting with (left) and ripping (right) PS+ platelets from lung vasculature (dotted line) after gut I/R injury. (D) Left: The number of PS+ platelets/fragments ripped or dragged by single (gray) or aggregated (white) rolling neutrophils or detached independently of neutrophils (black) from mesenteric thrombi over 2′ and 30′ after agonist injection in P-sel+/+ and P-sel−/− mice (percent of total). (D) Right: The time remaining on thrombi for neutrophil-associated PS+ fragments in P-sel+/+ mice (white and gray) and neutrophil-free PS+ fragments in P-sel−/− mice (n = 3 and 9 thrombi). (E) Confocal images depicting large PS+ platelet membrane fragments (cyan) within neutrophil aggregates (red) in mesenteric vein after gut I/R injury. (F) All the plasma from an I/R-injured or naïve donor mouse (plasma) was administered to a naïve recipient mouse (recipient) via the portal vein and then subjected to confocal microscopy. The number of aggregated neutrophils in the mesenteric veins and pulmonary circulation of excited lungs was quantified and compared to gut I/R–injured mice (I/R) (means ± SEM; n = 3). Scale bars, 10 μm (A), 50 μm (B), 10 μm (C), and 50 μm (E).

Deficiency of platelet cyclophilin D prevents neutrophil macroaggregation and preserves lung function

Exposure of PS on the surface of platelets is induced by two distinct cell death pathways, programmed cell apoptosis and regulated cell necrosis (5052). To investigate the platelet death pathways linked to membrane fragmentation, we performed studies on mice lacking the proapoptotic molecules Bak and Bax (51, 53) or the mitochondrial transition pore protein cyclophilin D (CypD), an important mediator of regulated necrosis (52, 54). Conditional deletion of CypD from platelets (CypDPlt−/−) had no impact on platelet count and tail bleeding time (fig. S9, A and B) or other markers of platelet activation, including P-selectin expression and integrin αIIbβ3 activation (fig. S9, C and D), similar to previous findings (52). Notably, CypDPlt−/− mice exhibited an ~80% reduction in neutrophil aggregate formation in mesenteric veins after intestinal I/R injury (Fig. 8, A and B). In contrast, deletion of Bak and Bax from platelets, using conditional Bak−/−BaxPlt−/− mice, had no inhibitory effect on neutrophil aggregation in mesenteric veins (fig. S9, E and F). CypDPlt−/− mice exhibited an 80% reduction in PS exposure on the surface of spread platelets (fig. S9, G and H), consistent with previous findings (52). This reduction in PS exposure was associated with a markedly increased resistance to shear-induced deformation and fragmentation in CypDPlt−/− platelets (fig. S9I). These studies indicate an important role for CypD-dependent platelet necrosis in regulating neutrophil aggregation.

Fig. 8. CypD deficiency reduces neutrophil aggregate formation and protects lung function after gut I/R injury.

CypDPlt+/+ and CypDPlt−/− mice were subjected to gut I/R injury with or without the indicated fluorescent probes. (A) Representative images of confocal intravital microscopy examining PS+ platelet-neutrophil aggregates (green/blue-red) in CypDPlt−/− and CypDPlt+/+ mesenteric veins after gut I/R injury (n = 4). (B) The number of single and aggregated neutrophils was quantified in the mesenteric veins of CypDPlt+/+ and CypDPlt−/− mice 30 to 90 min after gut I/R injury (n = 4). (C) Representative H&E staining of lung sections detecting intravascular neutrophil aggregates in CypDPlt+/+ and CypDPlt−/− mice after gut I/R injury. (D) The number of aggregated neutrophils in the pulmonary vasculature of CypDPlt+/+ and CypDPlt−/− mice after gut I/R injury (without saline flush) was quantified and normalized for lung section surface area (CypDPlt+/+, n = 10; CypDPlt−/−, n = 8). (E and F) Representative confocal images assessing neutrophil aggregates (green, Gr-1 antibody) and colocalized fibrin (red) and platelets (blue) in CypDPlt+/+ and CypDPlt−/− pulmonary vasculature (E) and the number of fibrin-rich neutrophil aggregates in three lobes of each lung quantified (F) (n = 4). Scale bars, 50 μm (A and E) and 20 μm (C). Error bars represent means ± SEM. *P < 0.05. (G and H) Line graphs depict quantification of arterial blood oxygen levels (G) and survival rates (H) in CypDPlt−/− and CypDPlt+/+ mice after gut I/R injury [CypDPlt+/+, n = 9 (G); CypDPlt−/−, n = 7 (G); CypDPlt+/+, n = 12 (H); CypDPlt−/−, n = 8 (H)]. (I) Representative images of the arterial vasculature of the left lung lobe (top), or the indicated lobe vasculature (bottom), depicting the extent of Microfil perfusion in CypDPlt+/+ and CypDPlt−/− mice (left) (scale bars, 50 μm; CypDPlt+/+, n = 8; CypDPlt−/−, n = 6), and the number of nonperfused vessels with a diameter of ≥80 μm (percent of total vasculature for each genotype) (right) in the flushed lungs of gut I/R–injured mice [CypDPlt+/+, n = 8 (24 vessels/lobe, 3 lobes/mouse); CypDPlt−/−, n = 6 (24 vessels/lobe, 3 lobes/mouse)].

To investigate the impact of platelet CypD deficiency on neutrophil aggregation and thrombosis in the lung, we performed histology and confocal imaging on mouse lungs after gut I/R injury. Similar to the findings in the intestines, histological analysis of lungs from CypDPlt−/− mice demonstrated a major reduction (>90%) in numbers of neutrophil aggregates in the pulmonary circulation after gut I/R injury (Fig. 8, C and D), whereas there was no significant difference in the number of neutrophil aggregates in Bak−/−BaxPlt−/− mice (fig. S9J). Notably, confocal imaging revealed a complete absence of fibrin generation in the pulmonary circulation of CypDPlt−/− mice after gut I/R injury (Fig. 8, E and F). It was noteworthy that the antithrombotic response in the lungs of CypDPlt−/−mice after gut I/R injury was context-specific because the carotid artery thrombotic response after electrolytic injury was minimally affected by CypDPlt deficiency (fig. S9K). Moreover, although clinically relevant doses of the commonly used antiplatelet drugs, aspirin or clopidogrel, can reduce the electrolytic carotid artery thrombotic response (55), they had no impact on neutrophil aggregate formation in mesenteric veins or the thrombotic response in the lung after gut I/R injury (fig. S10, A to C). These observations are consistent with previous findings (56) that neither aspirin nor P2Y12 receptor antagonists prevent platelet death (fig. S10D). Together, our findings suggest that inhibiting CypD-dependent platelet necrosis is likely to be a more effective means of reducing I/R-associated pulmonary thrombosis compared to aspirin and clopidogrel.

To examine whether the reduction in pulmonary thrombosis in CypDPlt−/− mice resulted in improvements in lung function after gut I/R injury, we monitored oxygen saturation levels in the carotid arteries of mice throughout gut I/R. In control C57Bl/6 and CypDPlt+/+ mice, a rapid and profound impairment of gas exchange occurred during the reperfusion phase of gut I/R injury (Fig. 8G). This marked reduction in blood oxygen levels ultimately led to respiratory failure and death (Fig. 8, G and H). In direct contrast, CypDPlt−/− mice had more sustained oxygen levels and a corresponding increase in survival during gut I/R injury (Fig. 8, G and H). Consistent with these findings, pulmonary vessels of CypDPlt−/− mice contained no neutrophil macroaggregates and remained patent, unlike those of CypDPlt+/+ mice (Fig. 8I). These findings demonstrate that the CypD-regulated thrombotic response triggered by gut I/R injury plays a major role in undermining pulmonary gas exchange and mouse survival.


The studies reported here define a previously unrecognized thrombotic disorder that is triggered by the ripping of large membrane fragments from dying platelets by rolling neutrophils, leading to intravascular neutrophil macroaggregation and vascular obstruction in multiple organs. This neutrophil-dependent thrombotic process is distinct from previously defined systemic thrombotic disorders in its ability to promote thrombosis not only in the microvasculature but also in medium-sized arteries and veins, with the size of neutrophil aggregates correlating with the size of vascular obstruction. A similar mixed arterial-venous thrombotic response also occurs in the lungs of patients with ARDS (16), and we have demonstrated that neutrophil macroaggregates, associated with platelets, are also prominent in these patients. The pulmonary thrombotic mechanism defined in this report is not prevented by commonly used antiplatelet agents, raising the possibility that therapeutic targeting of platelet death pathways may represent an innovative approach to reduce thrombosis in critically ill patients.

Neutrophils have long been known to play a critical role in promoting systemic inflammation and remote organ injury (32); however, they have not previously been linked to systemic thrombotic responses. The widespread deposition of thrombi in multiple organs is a major clinical problem and is associated with a high morbidity and mortality (21). Systemic thrombosis is typically linked to activation of blood coagulation with widespread deposition of fibrin within the microvasculature of multiple organs (that is, disseminated intravascular coagulation) (57). Alternatively, dysregulated platelet aggregation in the microvasculature, as occurs in thrombotic thrombocytopenia or hemolytic uremia syndrome (thrombotic microangiopathies), also leads to widespread organ ischemia (58, 59). However, neutrophil aggregates have not been demonstrated to play an important role in any of these thrombotic disorders. Central to the thrombosis mechanism described here is the widespread deposition of platelets within the microvasculature of the gut, liver, and lungs, presumably as a result of release of inflammatory mediators from the ischemic gut that perturbs endothelial cells in distant organs (1, 31). An unexpected finding from our intravital studies was the extent of platelet death within the microvasculature of the gut, liver, and lung, with up to 30% of platelets in sinusoids and postcapillary venules expressing surface PS. Platelet PS expression was prominent in regions of the gut with the most severe areas of ischemic injury, and, in general, the degree of gut injury correlated with the extent of remote organ damage. Whether similar degrees of gut injury are necessary for platelet deposition and PS+ exposure in humans with ARDS remains unknown. The factors promoting platelet death are also unclear but may be linked to high levels of free radical generation (10, 60), the presence of potent agonists, such as Thr, and the production of inflammatory mediators (61, 62). Dying platelets have unstable membranes, and when subjected to hemodynamic drag forces imposed by rolling neutrophils, large membrane fragments become incorporated onto the surface of rolling neutrophils, where they facilitate neutrophil aggregation. The ability of neutrophil macroaggregates, in combination with fibrin, to obstruct medium-sized arteries and veins, and smaller aggregates to obstruct the microvasculature, provides a mechanistic explanation of the unusual distribution of thrombi in the lungs of mice undergoing gut I/R injury and, potentially, patients with severe ARDS (see fig. S11).

Our finding that neutrophil aggregation, fibrin formation, and vaso-occlusive thrombi can be reduced in platelets that have a prominent defect in PS exposure (CypD-deficient platelets) points to a key role for mitochondria-driven cell death in this process. The role of platelet CypD in regulating thrombosis is controversial and has only been investigated in the context of localized vascular injury (52, 56). Some studies indicate that CypD deficiency is associated with a prothrombotic phenotype (56), whereas others suggest an antithrombotic effect (52). Our studies examining carotid artery thrombosis after electrolytic injury have revealed no major role for this cell death pathway in promoting thrombosis. Although it is difficult to directly compare findings of localized thrombotic responses with the widespread systemic thrombotic response reported here, it is nonetheless reasonable to conclude that the thrombotic response in CypD-deficient mice appears to be context-specific. Thus, the platelet membrane fragmentation and neutrophil aggregation that are central to the thrombotic mechanism described here may be less likely to have a major role in promoting thrombosis in more classical arterial and venous thrombosis models that are dominated by platelets and fibrin, respectively.

A key finding from our studies is the role of large PS+ platelet membrane fragments in promoting neutrophil aggregation and vascular occlusion. These procoagulant PS+ platelet membranes may also play an important role in promoting localized fibrin formation because procoagulant NETs were not detectable within neutrophil aggregates associated with fibrin. Platelet death is associated with the generation of FLIPRs (41) and microparticles that can bind neutrophils and promote their proinflammatory (63) and prothrombotic (64) functions. Microparticles can also promote neutrophil aggregation (63, 65), although as demonstrated here, these particles typically provided limited surface coating of neutrophils and were inefficient in promoting neutrophil macroaggregation. Our in vitro perfusion experiments and intravital microscopy studies demonstrate that the shear-dependent extraction of large membrane fragments from remnant dying platelets is the principal mechanism promoting neutrophil macroaggregation in vivo. Mechanistically, this makes sense because the abundance of PS+v platelets deposited on the endothelium of ischemic microvasculature in the gut, liver, and lung places these cells in continuous close proximity to rolling neutrophils. As a consequence, the probability of a neutrophil interacting with fragile platelet membranes is high, and the constant exposure of platelet membranes to elevated hemodynamic drag forces imposed by rolling neutrophils increases the likelihood of membrane fragmentation. Moreover, the membrane fragments extracted from dying platelets were typically large, allowing them to wrap around and form a physical “bridge” between adjacent rolling neutrophils.

It is likely that formation of neutrophil-rich thrombi in the lung after gut ischemia involves multiple inter-related mechanisms. Neutrophil aggregates forming in the gut are unlikely to efficiently traverse the liver, so many of the neutrophil aggregates in the lung almost certainly formed distal to the hepatic circulation and possibly de novo in the pulmonary circulation. De novo formation of neutrophil aggregates would almost certainly occur in the pulmonary venous system because rolling neutrophil aggregates in arteries cannot pass through capillaries and would invariably plug the distal arteriolar microcirculation. Consistent with this possibility was the identification of neutrophil macroaggregates in the left ventricle after gut I/R injury. Although not formally proven, several lines of experimental evidence suggest that the molecular mechanisms leading to neutrophil macroaggregate formation in the lung are likely to be similar to those operating in other organs. First, pulmonary neutrophil macroaggregates were of identical physical appearance to those in the gut and mesentery. Second, pulmonary, gut, and mesenteric neutrophil aggregates all contained large PS+ platelet membrane fragments. Third, the endothelial deposition of PS+ platelets, as well as the neutrophil ripping and dragging of PS+ platelet membranes, appeared to be widespread, occurring in the mesenteric, intestinal, hepatic, and pulmonary circulation. Functionally, intravascular neutrophil aggregates had the capacity to occlude blood vessels in multiple organs.

Endothelial injury is a universal feature of acute lung injury after gut ischemia that is mediated by inflammatory mediators released from the ischemic gut (1, 31), leading to widespread deposition of platelets and neutrophils in the pulmonary vasculature (35). Our demonstration that neutrophils rip PS+ platelet membranes in the pulmonary microcirculation after gut I/R injury is consistent with the possibility that neutrophil aggregates can develop locally in the pulmonary vasculature. However, a limitation of our study was the inability to perform intravital imaging of the pulmonary vasculature. This is a major technical challenge, given the dynamic nature of the thrombotic process and the requirement to perform imaging within the deep vasculature of the lung where neutrophil macroaggregates are present.

Widespread thrombosis throughout the pulmonary circulation is a common feature of ARDS (16, 17, 19, 20), resulting in pulmonary hypoperfusion (21), severe lung injury, and, ultimately, respiratory failure (16, 20, 22). This clinical scenario is associated with a high mortality (16, 20). Apart from general supportive medical therapy, there are no specific treatment options that reduce thromboinflammation and lung injury in patients with ARDS (19, 22, 66). Anticoagulants and fibrinolytic agents tested in ARDS patients have not improved clinical outcome (19, 22). Our findings raise the interesting possibility that this resistance to antithrombotic therapy may be, in part, explained by the widespread deposition of vaso-occlusive neutrophil aggregates. Preventing platelet death and the subsequent formation of neutrophil aggregates in the lung may therefore represent a feasible therapeutic option to reduce thrombosis and remote organ injury in critically ill patients. This is particularly attractive clinically because inhibition of regulated platelet necrosis does not increase bleeding risk (52), unlike all currently available antithrombotic approaches.


A detailed description of most materials and methods used in this study can be found in the Supplementary Materials.

Study design

To examine how local gut ischemia can trigger remote organ injury, we used a mouse model of intestinal I/R injury, in combination with intravital microscopy, to monitor the adhesive interactions between neutrophils and platelets in the ischemic gut and in the isolated lung. These investigations revealed a distinct thrombotic mechanism that dying platelets in the ischemic gut trigger neutrophil macroaggregates, leading to blood obstruction. To address the clinical relevance of this novel thrombotic mechanism, we examined human postmortem lungs from patients with ARDS, and comparisons were made with postmortem lung specimens from patients with acute pulmonary edema and biopsies from patients with emphysema. To clarify the role of platelets and neutrophils in this distinct form of thrombosis, we compared the development of this process in mice with global or platelet-selective P-selectin deficiency and in mice with platelets or neutrophils depleted. The ability of neutrophil aggregates to obstruct blood flow was examined using intravital microscopy and Microfil perfusion strategies for the gut and lung circulation, respectively. The involvement of specific platelet death pathways in dying platelet–mediated neutrophil aggregation was investigated using mice lacking either CypD or Bak/Bax specifically from the megakaryocytic linage. The functional impact of dying platelet–mediated neutrophil aggregation and consequent vascular occlusion on lung function was assessed by monitoring arterial blood oxygen levels and mouse survival. All studies and related analysis involving CypD-deficient mice were performed in a blinded fashion. All in vivo and in vitro studies were carried out at a minimum of triplicate independent experiments, and statistical analysis was performed wherever applicable. All studies involving the use of animals, animal tissues, and human specimens were performed with relevant ethics approval, as described in Supplementary Materials and Methods. Primary data are located in table S3.

Statistical analysis

Power calculations were used to establish sample size for in vivo animal experiments (significance level, 0.01; statistical power set at 80%). If statistical significance was reached with fewer animals, no additional animals were used. Statistical significance between multiple treatment groups was analyzed using either one- or two-way analysis of variance (ANOVA), with Dunnett’s or Sidak’s post-testing, where indicated. Statistical significance between two treatment groups was analyzed using an unpaired Student’s t test with two-tailed P values (Prism software v6.07; GraphPad Software for Science). Data are means ± SEM, where n equals the number of independent experiments performed.


Materials and Methods

Fig. S1. Formation of leukocyte aggregates and fibrin in the pulmonary vasculature of mice after gut I/R injury.

Fig. S2. Neutrophil aggregate and fibrin formation in the lung vasculature of I/R-injured mice and ARDS patients.

Fig. S3. Neutrophil aggregate formation in the splanchnic circulation of the mouse after gut I/R injury.

Fig. S4. P-selectin–expressing platelets on gut vasculature and within neutrophil aggregates after gut I/R injury.

Fig. S5. Neutrophil aggregation requires potent platelet activation.

Fig. S6. PS+ platelets selectively support neutrophil aggregation in vivo.

Fig. S7. Occlusive neutrophil aggregate formation is associated with fibrin formation.

Fig. S8. Rolling neutrophils pull membrane fragments from PS+ platelets.

Fig. S9. Characterization of platelets deficient in CypD or Bak/Bax.

Fig. S10. Conventional antiplatelet therapies do not prevent neutrophil aggregate formation after gut I/R injury.

Fig. S11. Schematic illustration of the proposed mechanism linking gut I/R injury to pulmonary thrombosis.

Table S1. ARDS: Patient details.

Table S2. APE: Patient details.

Table S3. Primary data (Excel file).

Movie S1. Neutrophil aggregate formation on PS+ thrombi at sites of vascular injury.

Movie S2. PS+ platelet-neutrophil aggregate formation in the mesenteric veins after gut I/R injury.

Movie S3. Neutrophil ripping and dragging PS+ platelet membranes by rolling neutrophils in vitro (low magnification).

Movie S4. Neutrophil ripping PS+ platelets in vivo and in vitro.

Movie S5. Neutrophil extract PS+ platelets in the intestinal vasculature after gut I/R injury.

References (6773)


  1. Acknowledgments: We thank C. Zhu and L. (A.) Ju for their assistance with the leukocyte drag force estimation, H. Salem and C. Geczy for advice and helpful discussions, Z. Ruggeri for the antifibrin antibody, and M. Hickey, L. Toennesen, J. McLean, and M. Lebois for technical assistance. We also acknowledge the technical assistance of Monash Microimaging, the Australian Microscopy and Microanalysis Research Facility at the Electron Microscope Unit, the University of Sydney, and the Monash Centre for Electron Microscopy. Funding: This work was supported by the National Health and Medical Research Council (NHMRC) (APP1127278 and APP1079400). Z.K. and J.M. were recipients of scholarship from the Australian National Heart Foundation and the Wheaton family, respectively. E.C.J. was supported by NHMRC (APP1023029, APP1016647, and APP9000220) and was a recipient of Lorenzo and Pamela Galli Charitable Trust fellowship. Author contributions: Y.Y. designed and performed the research, analyzed and interpreted the results, and assisted in writing the manuscript. I.A., M.C.L.W., Z.K., K.A., and D.B. performed the research and analyzed the results. A.P. analyzed and interpreted the human clinical samples. J.M. analyzed and interpreted the clinical samples. S.M.S. performed the research, analyzed the results, and assisted in the preparation of the manuscript and figures. E.C.J. performed the research, analyzed the results, and provided the key reagents. B.T.K. provided the key reagents and interpreted the results. S.P.J. designed the research, analyzed and interpreted the results, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Correspondence and requests for materials should be addressed to S.P.J. (Heart Research Institute, Charles Perkins Centre, Building D17, John Hopkins Drive, University of Sydney; Ph: 61 2; shaun.jackson{at}

Stay Connected to Science Translational Medicine

Navigate This Article