Evren Kilinç, Department of Internal Medicine, Laboratory of Clinical Thrombosis and Haemostasis, Maastricht University, PO Box 616, UNS 50: Box 8, 6200 MD Maastricht, the Netherlands. Tel.: +31 43 3881542; fax: +31 43 3884159. E-mail: firstname.lastname@example.org
See also Mutch NJ. Emerging roles for factor XII in vivo. This issue, pp 1355–8.
Summary. Background: Particulate matter (PM) is a key component of ambient air pollution and has been associated with an increased risk of thrombotic events and mortality. The underlying mechanisms remain unclear. Objectives: To study the mechanisms of PM-driven procoagulant activity in human plasma and to investigate mainly, the coagulation driven by ultrafine particles (UFPs; < 0.1 μm) in genetically modified mice. Methods: Thrombin generation in response to PM of different sizes was assessed in normal human platelet-poor plasma, as well as in plasmas deficient in the intrinsic pathway proteases factors XII (FXII) or XI (FXI). In addition, UFPs were intratracheally instilled in wild-type (WT) and FXII-deficient (FXII−/−) mice and plasma thrombin generation was analyzed in plasma from treated mice at 4 and 20 h post-exposure. Results: In normal human plasma, thrombin generation was enhanced in the presence of PM, whereas PM-driven thrombin formation was completely abolished in FXII- and FXI-deficient plasma. UFPs induced a transient increase in tissue factor (TF)-driven thrombin formation at 4 h post-instillation in WT mice compared with saline instillation. Intratracheal instillation of UFPs resulted in a procoagulant response in WT mice plasma at 20 h, whereas it was entirely suppressed in FXII−/− mice. Conclusions: Overall, the data suggest that PM promotes its early procoagulant actions mostly through the TF-driven extrinsic pathway of coagulation, whereas PM-driven long lasting thrombogenic effects are predominantly mediated via formation of activated FXII. Hence, FXII-driven thrombin formation may be relevant to an enhanced thrombotic susceptibility upon chronic exposure to PM in humans.
Particulate matter (PM) comprises all small components of air pollution and is derived from both natural and anthropogenic sources such as wood smoke and engine exhaust . PM can be classified into three main subtypes according to particle size: coarse between 2.5 and 10 μm, fine in the range between 0.15 and 2.5 μm, and ultrafine particles (UFPs) with a diameter smaller than 0.1 μm . Epidemiological and experimental studies suggest that exposure to PM is associated with increased risk of arterial and venous thrombosis, and increased risk of cardiovascular death [1,3–6].
Exposure to PM has been associated with a hypercoagulable state in both humans and rodents , although the mechanistic basis for PM-triggered procoagulant activity is poorly understood. One possible mechanism for a PM-driven procoagulant state may be the increased expression of tissue factor (TF) in lung tissue  through inflammatory mediators such as interleukin-6 . In addition, suppressed endothelial thrombomodulin activity aggravates the procoagulant response to PM in rodents . Moreover, inhalation of airborne PM in humans has been shown to result in pulmonary inflammation and systemic inflammation, reduced release of tissue-plasminogen activator (t-PA), and increased activation and adhesiveness of platelets, resulting in increased ex vivo thrombosis generation [10–12].
Results of animal exposure studies suggest that due to their small size, UFPs can transmigrate through the lung epithelium and vascular barrier into the systemic circulation [13–19], where they may directly interact with blood cells and plasma proteins. On the other hand, whether UFPs can readily access the circulation in humans is still controversial [20,21].
In the case of translocation into the plasma compartment, UFPs may interact with the intrinsic pathway of coagulation, which is initiated upon contact of factor (F) XII with negatively charged surfaces. In addition to the traditional concept where anionic surfaces activate FXII, FXII activation in plasma occurs with a greater efficiency at hydrophilic surfaces than at an equal surface area of hydrophobic ones . Recently, physiological FXII activators such as collagen, misfolded proteins and polyphosphates were identified [23–25]. Activated FXII (FXIIa) drives the intrinsic pathway by activating its substrate FXI, which in turn activates FIX , ultimately resulting in fibrin formation. Studies using FXII-deficient (FXII−/−) mice implicate a role of the intrinsic pathway of coagulation in arterial thrombosis , cerebral ischemia reperfusion injury  and pulmonary embolism . Despite the fact that the exact role of FXII in thrombosis remains to be further elucidated , clinical evidence has shown positive association with an increased risk of acute thrombotic events [30,31].
Hereby, we hypothesize that PM can enhance the formation of FXIIa, thereby inducing a hypercoagulable state. Our hypothesis was tested in human plasma for coarse, fine and ultrafine particles collected from different locations and, furthermore, in FXII−/− and wild-type (WT) mice, challenged with UFPs by intratracheal instillation.
Material and methods
Sampling and characterization of PM
Coarse and fine PM samples were collected at various European locations with contrasting traffic intensities (1, Hendrik Ido Ambacht roadside tunnel; 2, Dordrecht, the Netherlands; 3, München east train station; and 4, München Grosshadern, Germany) on polyurethane foam (PUF) using a high-volume cascade impactor . Additionally, coarse, fine + UFP and UFP biosamples were collected using a Versatile Aerosol Concentration Enrichment System (VACES)  equipped with Biosamplers (SKC Inc., Eighty-Four, PA, USA; coarse, fine + UFPs, and UFPs of PM) at a platform of an underground train station (Amsterdam, the Netherlands) at two different time points. The UFP biosample, which was administered in mice, was collected with the same method near a Dutch roadside tunnel (Hendrik Ido Ambacht) that is mainly used by heavy diesel trucks.
Thrombin generation in human plasma
Plasma thrombin generation was measured by means of the calibrated automated thrombogram (CAT) method (Thrombinoscope BV, Maastricht, the Netherlands) , which makes use of a low affinity fluorogenic substrate for thrombin (Z-Gly-Gly-Arg-AMC) to continuously monitor thrombin activity in clotting plasma. In order to correct for inner-filter effects and substrate consumption, each thrombin generation measurement was calibrated against the fluorescence curve obtained in the same plasma with a fixed amount of thrombin-α2-macroglobulin complex (Thrombin Calibrator; Thrombinoscope BV), as recommended by the manufacturer. Fluorescence was read in an Ascent Reader (Thermolabsystems OY, Helsinki, Finland) equipped with a 390/460 filter set, and thrombin generation curves were calculated with the Thrombinoscope software (Thrombinoscope BV) as described previously .
The FXII-mediated effect on thrombin generation in vitro of PM from different locations was first measured using coarse and fine PM collected on PUF while the effect of all three PM size categories was examined using two biosamples from an underground train station according to the following protocol: 10 μL PM (at four different concentrations) in saline was applied to 80 μL normal human pooled platelet-poor plasma, or FXII- or FXI-deficient plasma (George King Biomedical, Overland Park, KS, USA). Thrombin generation was recorded in the absence or presence of 4 μm phospholipids (Thrombinoscope BV) with PM solution at concentrations of 3.6, 1.8, 0.9, 0.45 and 0 μg mL−1. Furthermore, the reaction was started upon addition of calcium and fluorogenic substrate. Three parameters were derived from the thrombin generation curves: lag time, peak height and endogenous thrombin potential (ETP, area under the curve).
Normal human pooled platelet-poor plasma was prepared from 90 healthy volunteers not taking any medication. Venous blood was collected in 3.2% (w/v) citrate tubes using a 21-gauge needle (BD). The first 10 mL of venous blood were discarded. Platelet-poor plasma was prepared by two centrifugation steps: the first at 2000 × g for 15 min and the second at 11 000 × g for 10 min. Obtained plasmas were pooled and aliquots were snap-frozen in liquid nitrogen and stored at −80 °C until use. All samples were thawed at 37 °C for 15 min before analysis.
Active side-inhibited FVII (ASIS; kindly provided by L.C. Petersen, Health Care Discovery, Novo Nordisk A/S, Måløv and Bagsværd, Denmark) at 30 nm final concentration was added to human plasma in all in vitro thrombin generations in human plasma before analysis, in order to investigate the contribution of the intrinsic pathway of coagulation to thrombin generation in the presence of PM. Furthermore, corn trypsin inhibitor (CTI; Haematologic Technologies Inc., Essex Junction, VT, USA) with a final concentration of 40 μg mL−1 was added to inhibit FXII activity in one of the experimental settings. The thrombin generation experiments with PM were also performed in the presence of phospholipids only (no ASIS), in order to determine the contribution of activation of the extrinsic pathway of coagulation.
Activation of FXII
Enzymatic FXIIa activity was determined from the cleavage of Pefachrome FXIIa substrate (DSM Nutritional Products Ltd, Branch Pentapharm, Basel, Switzerland) at 37 °C. Incubations contained purified FXII (95 nm), prekallikrein (PK; 30 nm), and high-molecular-weight kininogen (HMWK; 30 nm) in Tris–Imidazol buffer pH 7.9 (150 mm NaCl, 50 mm Tris Base, 50 mm Imidazol) all at final concentrations, after addition of the substrate . In order to investigate whether PM enhances FXII activation in the absence and presence of PK and HMWK, coarse PM (at 83.3, 41.6, 20.8, 10.4 and 0 μg mL−1) and purified FXII were preincubated for 20 min at 37 °C. After addition of 0.8 mm Pefachrome FXIIa substrate with and without PK and HMWK, the increase in absorption at 405 nm was determined (linear in time).
Animals and intratracheal instillation of UFPs
Eight-week-old male FXII−/− mice [27,28] and WT littermate mice were challenged with a UFP biosample (collected near a Dutch roadside tunnel, mainly used by heavy diesel trucks) suspended in sterilized saline at a concentration of 3.6 μg mL−1. In addition, sterile saline was instilled in WT mice as control. Mice were anesthetized with 350 μL Avertin (25 μg mL−1 intraperitoneally) and UFPs were instilled intratracheally in a volume of 100–120 μL per mouse. After instillation, mice were placed in the right and left lateral decubitus position for 10–15 s for each site. Four hours and 20 h after instillation, blood samples were collected as described previously . The protocol for the use of mice was approved by the Animal Care and Use Committee at Maastricht University.
Thrombin generation in mouse plasma
Thrombin generation in mouse plasma was recorded using 10 μL plasma diluted with 30 μL HEPES buffer (25 nm HEPES, 175 nm NaCl, pH 7.7) and 4 μm final concentration of phospholipids Additionally, thrombin generation in mouse plasma was assessed either in the presence of 40 μg mL−1 CTI and 30 nm ASIS in order to evaluate the contribution of FXIa to thrombin generation in plasma, only in the presence of 40 μg mL−1 CTI to measure TF, or in the presence of 30 nm ASIS to assess the contribution of FXIIa.
Results are expressed as mean ± SD. Data analysis was performed using Prism 5 for Windows, version 5.01 (GraphPad Software Inc., San Diego, CA, USA). Differences between groups were compared using the one-way anova test with Bonferroni correction and statistical significance was defined as P <0.05.
PM enhances thrombin formation by the intrinsic pathway of coagulation in vitro
To study the contribution of the intrinsic pathway of coagulation to thrombin generation in the presence of PM in different size ranges and from various locations in human plasma, we analyzed PM-induced thrombin generation by means of CAT. All PM (coarse, fine) from different locations dose-dependently enhanced thrombin generation in normal human pooled platelet-poor plasma (in the presence of ASIS) as indicated by a shortening in lag times, as well as by increased maximal (peak height) and total (ETP) thrombin formed compared with plasma thrombin generation in the absence of PM (Fig. 1, panels A and B). Results for the highest concentration of 3.6 μg mL−1 compared with normal plasma are shown in Fig. 1 (panels D, E and F). For each individual sample, the coarse fraction had a shorter lag time and higher ETP and peak height compared with the corresponding fine fraction. Thrombin generation results at the highest concentration of 3.6 μg mL−1 for PM biosamples of different size (coarse, fine + UFPs and UFPs) from one location (an underground train station) are depicted in Fig. 2 (panels A and B). The UFP fraction of PM biosamples from the first visit to the train station yielded the strongest thrombin generation in normal plasma, whereas for the second visit the response to coarse fraction was the strongest.
The thrombin generation results in the presence of only phospholipids (4 μm; no ASIS) were comparable to the results derived from experiments with ASIS (data not shown). This suggests that the extrinsic pathway of coagulation in human plasma is not affected by PM.
Furthermore, we analyzed whether FXII is activated in the presence of PM and determined thrombin formation in FXII- and FXI-deficient human plasma. All applied PM completely failed to stimulate thrombin generation in the absence of FXII or FXI (Figs 1C and 2C). Consistently, the specific FXIIa inhibitor (corn trypsin inhibitor, 40 μg mL−1) abolished thrombin generation in normal plasma (Figs 1C and 2C). Additionally, in the absence of phospholipids, no thrombin generation was observed in any experiment (data not shown).
The formation of FXIIa is increased in the presence of PM
In order to demonstrate the activation of FXII in the presence of PM, purified FXII was incubated with coarse PM at four different concentrations or buffer, which served as a negative control. After incubation, the generation of FXIIa was assessed using a chromogenic assay. In this purified system (only FXII protein and coarse PM), coarse PM clearly accelerated FXIIa formation in a dose-dependent manner (Fig. 3A). In the presence of PK and HMWK, formation of FXIIa was increased in each corresponding dose of coarse PM (Fig. 3B).
UFPs induce early TF pathway-mediated procoagulant shift upon intratracheal instillation in mice
Given the distinct size of PM, UFPs are theoretically more likely to translocate into the blood stream than larger-sized PM fractions. Therefore, we undertook animal experiments with UFPs, collected near a Dutch roadside tunnel (3.6 μg mL−1), which were intratracheally applied to WT (vehicle, n = 7; UFPs, n = 7) and FXII−/− mice (UFPs, n = 8). Plasma was collected at 4 h post-instillation in order to determine the early effects of UFPs on coagulation in mice. In the presence of phospholipids (4 μm) and CTI (40 μg mL−1), both WT (UFPs WT; lag time, 9.5 ± 4.6 min; ETP, 185 ± 118 nm min−1; peak height 7 ± 4 nm) and FXII−/− (UFPs FXII−/−; lag time, 8.4 ± 4.5 min; ETP, 146 ± 125 nm min−1,; peak height, 6 ± 5 nm) mice showed comparable levels of thrombin formed (Fig. 4, panels A, B and C). Consistently, saline instillation in WT mice also showed limited thrombin generation (vehicle WT; lag time, 28.6 ± 21.0 min; ETP, 113 ± 72 nm min−1; peak height, 3 ± 2 nm).
This limited procoagulant effect was entirely inhibited upon the addition of ASIS (30 nm) and ASIS combined with CTI (suggesting no FXIa)  (data not shown), thus supporting the assumption of TF-mediated thrombin generation at 4 h post-instillation. There was no thrombin generation observed in WT and FXII−/− mice at 20 h upon instillation of UFPs or saline in WT mice (data not shown).
Enhanced FXII activation is critical for the late procoagulant effects of UFPs in mice
To better appreciate the procoagulant potential of PM in time, we also tested WT (n = 8) and FXII−/− mice (n = 8) at 20 h after intratracheal instillation of UFPs. Inhibition of the extrinsic coagulation route by ASIS (30 nm) revealed an intrinsic pathway-dependent thrombin generation in WT mouse plasma (lag time, 9.9 ± 1.4 min; ETP, 285 ± 147 nm min−1; peak height, 13 ± 8 nm; Fig. 5, panels D, E and F) compared with vehicle (no thrombin generation; Fig. 5, panels D, E and F). In contrast, no thrombin generation was observed in plasma from FXII−/− animals challenged with UFPs (Fig. 5A), suggesting a FXII dependency. Consistently, addition of CTI or CTI combined with ASIS to plasma of WT mice resulted in a complete diminution of thrombin generation (Fig. 5C), thus consolidating the evidence that the formation of FXIIa is enhanced in the presence of UFPs at a late time point (20 h).
Both venous and arterial thrombotic events have been associated with exposure to PM [3,5,37,38]. Experimental studies in animals, as well as in humans, have shown that thrombosis may be triggered by PM, affecting the blood coagulation system in various ways, including the induction of TF and increased platelet reactivity [10,39–43].
The present study demonstrates the effects of PM on blood coagulation both in vitro and in mice. This report is the first to investigate the role of the contact activation pathway in mediating PM procoagulant actions. We provide new data demonstrating that UFPs exert a transient TF-dependent thrombin formation, responsible for inducing an early procoagulant response upon challenge with PM, and enhanced FXII activation is essential to sustain this prothrombotic effect in mice. These findings are further consolidated by showing that PM accelerates the formation of FXIIa in human plasma. Moreover, in a purified system enhanced FXII activity was shown in the presence of PM in a dose-dependent manner and the activity was increased in the presence of PK and HMWK.
Although our data do not provide direct evidence, it may be postulated that PM acts as a surface for coagulation reactions, including assembly of contact proteins. The contribution of PM to this reaction may in part depend on its chemical composition. It was indeed demonstrable that PM collected from the same location at two different time points did not show consistent thrombin generation results in human plasma. Although the chemical analysis of PM in relation to procoagulant activity was beyond the scope of this study, there are published data to document various effects. PM consists of transition metals such as Nickel (Ni) and Ferro (Fe), sulfates, nitrates and black carbon [44,45]. Sangani et al.  recently showed that a whole blood coagulation time decreased after addition of PM to human blood compared with addition of buffer solution alone, an effect related to the presence of metal sulfates.
In the upper airways, local reactions in the lung, involving pulmonary inflammation, TF production and oxidative stress on macrophages and lung epithelial cells [8,12,47,48], may underlie activation of the TF-mediated blood coagulation by PM. Although the mucocilary clearance will remove all particles larger than 6 μm, it is suggested that the efficiency of this system is very low for UFPs and at the deeper lung regions, the clearance of particles will rely in part on phagocytic uptake . In contrast to larger particles, the macrophage uptake of UFPs seems to be less effective .
Using different animal species, it has been demonstrated that part of UFPs transmigrates from the lungs into the blood circulation [13–19], although the efficiency of translocation may be low, around 1–2.5% reaching the circulating blood in humans and rodents [21,49,51]. However, it should be noted that translocation remains controversial in humans because other investigators have not confirmed this mechanism to be operational [20,52], probably due to differences in techniques. In the present study, the main focus was on the question of whether UFPs activate coagulation, particularly the intrinsic pathway of coagulation, when they enter the bloodstream, assuming that translocation would occur. To test this hypothesis, we applied the intratracheal instillation method, which is a validated  and commonly used procedure in PM-related studies [9,54–56], although one should keep in mind that this is different from exposure to environmental pollutants. In selecting the dose ranges for PM application, we followed the reasoning in the paper by Stoeger and colleagues , demonstrating that 3–5 μg UFPs would be a relevant concentration to apply in mice, extrapolating urban environmental concentrations in man to the surface area exposed in mice.
Assuming translocation, another issue is the retention of UFPs in the circulation. In a very recent study, Hirn et al.  addressed the fate of gold labeled nanoparticles (GNPs) of different size and surface charge after intravenous administration in rats. They still found small but detectable amounts of the administered GNPs of all sizes in the blood at 24 h after administration. Their data suggested retention of GNPs in blood and their localization on cells or in serum to be size dependent. Similar mechanisms may explain the persistent presence of PM-related activity in blood after intratracheal exposure, probably associated with phospholipid surfaces (cells and microparticles), to increase the formation of FXIIa over time. The current data are consistent with results from experimental exposures in healthy human subjects, where diesel engine exhaust exposure significantly increased the ex vivo thrombus generation in a Badimon chamber as compared with filtered air . In this system peripheral venous blood from the subjects is directly applied to strips of endothelial denuded pig aorta at high and low flow rates, mimicking plaque rupture in large and small arteries. The increase in thrombus following inhalation of diesel engine exhaust particles was related to increased platelet-leukocyte adhesion, but this was not entirely explained by this effect . We suggest that the presently identified enhanced activation of FXII in the presence of UFPs is the major additional factor behind the enhanced tendency to thrombus formation in the Badimon chamber ex vivo model.
In conclusion, we have shown that PM enhances the formation of FXIIa in vitro and that traffic-related UFPs induce a slow but significant increment of thrombin generation in vivo, which is dependent on contact activation. While previous studies demonstrated early activation of coagulation to be TF mediated, our data extend the significance of FXII-driven coagulation activity for sustaining an increased level of thrombin generation in time. Given the potential of thrombin to act in a pro-atherogenic manner against a background of inflammation, these mechanisms may be relevant for the increased risk of thrombosis as well as atherosclerosis in humans exposed to PM in daily life .
E. Kilinç, H. M. H. Spronk, T. Renné and H. ten Cate designed the study and drafted the manuscript and R. van Oerle and C. Oschatz coordinated and performed laboratory investigations. J. I Borissof contributed to drafting the manuscript. M. E. Gerlofs-Nijland, N. A. Janssen and F. R. Cassee supplied the PM samples, and participated in the study design, data interpretation and drafting of the manuscript. T. Sandström has been the coordinator of the European HEPMEAP project that supplied the PM samples collected on PUF. All authors contributed to the writing and critically reviewed the manuscript.
This study was financially supported by the Netherlands Heart Foundation (grant number: 2006B064) and VR grant K2010-64X21462013 to T. Renné. A Marie Curie fellowship from the European Commission was granted to J. I. Borissoff (MEST-CT-2005-020706). We would like to thank D. Fens and P. Pluijmen for laboratory assistance and logistic support. PM samples collected on PUF were collected in the framework of a grant from the European Union under the 5th Framework Programme, Quality of Life and Management of Living Resources, Key Action 4: Environment and Health by the HEPMEAP project contract QLRT-1999-01582 (http://www.hepmeap.org). Samples from the underground were collected within the framework of the RAPTES project, funded by the RIVM strategic research program.
Disclosure of Conflict of interests
The authors state that they have no conflict of interest.