Extracellular RNA constitutes a natural procoagulant cofactor in blood coagulation

Extracellular RNA, RNase, and Thrombus Formation in Vivo.

In an established mouse model of arterial thrombosis, local administration of ferric chloride to the adventitial vessel surface induced severe damage/necrosis of the intimal part of the vessel wall (20–22). In vehicle-treated animals, complete vessel occlusion by platelet-rich thrombus occurred after 545 ± 18 sec (Fig. 1 A–C and J). For the detection of extracellular RNA in situ, before FeCl₃ exposure, RNASelect green fluorescent cell stain was injected into the left jugular vein of mice, and extracellular RNA-related staining was discernible in association with the occlusive arterial thrombus (Fig. 1A). No cell-free DNA was found associated with the thrombus material that was mainly composed of platelets in association with fibrin as shown in larger magnifications (Fig. 1 B, E, and H). Direct binding experiments in vitro revealed association of RNA with fibrillar collagens and fibrin but not with fibrinogen, indicating a specific interaction of the nucleic acid with thrombus and vessel matrix material (data not shown).

In the RNase-treatment group mice, time to arterial occlusion was significantly prolonged to 1,457 ± 326 sec, and no RNA was found in association with the largely reduced thrombus material (Fig. 1 D–F and J), whereas administration of DNase had no influence [Fig. 1J and supporting information (SI) Fig. 4]. These results indicate that an arterial thrombus can harbor extracellular RNA and that counteracting RNase can exert a strong antithrombotic effect in vivo. For comparison, administration of a factor XIIa inhibitor had an even stronger vessel protective effect, and after 60 min, no occlusive thrombus was detectable (Fig. 1 G–J). As illustrated by corresponding intravital microscopy, RNase- and factor XIIa-inhibitor-dependent prolongation of time to full occlusion was due to the formation of large emboli leading to fragmentation of the thrombus and maintenance of residual blood flow, whereas in the vehicle and DNase treated groups, a full occlusive thrombus resulted in blood stasis (Fig. 1J and SI Fig. 4). Consequently, extracellular RNA may constitute the predominant nucleic acid species in vivo with putative procoagulant activity. In addition, tRNA from yeast or artificial RNA poly(inosinic acid)·poly(cytidylic acid) [poly(I·C)] injected into healthy rabbits induced a procoagulant response (SI Figs. 5 and 6). Pretreatment of RNA by RNase, RNase alone or the RNA equivalent dose of nucleotide monophosphates had no significant influence on the basal levels of coagulation parameters. Together, these data strongly indicate that natural as well as artifical RNA, the latter being known as structural analogue for double-stranded viral RNA (23), can facilitate blood coagulation in vivo.

Extracellular RNA Promotes (Auto-)Activation of Coagulation Proteases.

Cell lysate derived from cultured Chinese hamster ovary (CHO) cells (that do not express tissue factor) significantly augmented calcium-induced blood coagulation in normal plasma (Fig. 2A). Whereas pretreatment of cell lysate with DNase or antibodies to tissue factor had no influence, RNase abrogated the procoagulant activity, indicating that cell-derived RNA serves as procoagulant cofactor. Likewise, cell lysate and isolated nucleic acids (cellular RNA or genomic DNA) from CHO cells exhibited a significant augmentation of coagulation in whole blood by shortening the lag phase before coagulation began (Fig. 2B). Pretreatment of RNA or DNA with the respective nuclease resulted in a significant reduction or loss, respectively, of procoagulant activity. The nucleic acid-dependent induction of blood coagulation did not involve platelets, because RNA did not lead to platelet activation (as measured by P-selectin expression on washed platelets), and the nucleic acid did not induce platelet shape changes or aggregation (data not shown).

In the clinical laboratory, nonbiological polyanionic material such as kaolin is routinely used in diagnostic coagulation testing of patient plasma samples (24). Because nucleic acids appear to share the polyanionic character with kaolin, the comparative analysis with isolated RNA and DNA revealed that all three substances promoted blood coagulation in a concentration-dependent fashion in recalcified plasma in a comparable way (Fig. 2C). Although on a weight basis kaolin was the far most effective procoagulant compound because of its high polyanionic surface density, extracellular nucleic acids represent the natural analogue of kaolin.

To specify the type of RNA that contains procoagulant activity, equivalent doses of RNA from different sources were compared in a turbidity coagulation assay that measures the onset of fibrin formation in recalcified plasma, and results were expressed as “kaolin-equivalent units.” Eukaryotic ribosomal RNA isolated from CHO cells, transfer-RNA from yeast, Escherichia coli RNA, or single-stranded RNA from Qβ-phage all exhibited an appreciable procoagulant activity that was (except for E. coli RNA) abrogated by pretreatment with RNase A (Fig. 2D). Also, viral RNA from hepatitis C virus or picornavirus exhibited a comparable potent procoagulant activity (data not shown).

Binding to and Activation of Contact Phase Proteins by Nucleic Acids.

Isolated nucleic acids were tested for activation of contact phase proteins in citrated plasma and were found to induce a significant (kaolin-like) 15- to 20-fold elevation of serine protease activity that could be prevented by preincubation with the respective nuclease (Fig. 2E). As expected, tissue factor had no effect on induction of serine protease activity. In RNA-supplemented citrated plasma, both the broad spectrum inhibitor aprotinin and specific inhibitors against factor XIIa (corn trypsin inhibitor) or synthetic inhibitors against factor XIa and kallikrein were equally effective to inhibit RNA-induced serine protease activity completely (Fig. 2F). These results were corroborated by thrombelastographic and clotting assays in whole blood or plasma, respectively (SI Fig. 7), and underline our proposition that nucleic acids substantially promote induction/elevation of serine protease activity related to enzymes of the contact phase of blood coagulation. Analogously, these proteases were activated by artificial RNA (data not shown), indicative for the highly specific nature of nucleic acid material to interact with these plasma proteins.

After binding and crosslinking of radiolabeled hepatitis C virus RNA to individual coagulation proteins, strong interactions between factors XII or XI and RNA was observed, whereas prekallikrein and kininogen showed weaker binding. Other coagulation factors, such as fibrinogen, BSA, and the vitamin K-dependent proteins IX, X, and II (prothrombin), did not bind to RNA (Fig. 3A). Likewise, RNA-complexes containing contact phase proteins were discernable after SDS-gel electrophoresis (SI Fig. 8), which dissappeared after proteinase K treatment. These results indicate that factors XII and XI, prekallikrein and kininogen, can be defined as RNA-binding plasma proteins. When radiolabeled RNA was allowed to react with contact proteins in different stoichiometric ratios followed by electrophoretic separation under nondenaturing conditions, strong interactions between, e.g., factor XI and RNA (Fig. 3B) and kininogen and RNA (data not shown) was noted as reflected by electrophoretic mobility of the complexes. In contrast, no shifting of RNA was seen upon addition of factor X or prothrombin (factor II).

In purified systems, RNA or isolated DNA induced protease activation of factors XII and XI (Fig. 3C) or factor XI and thrombin system (Fig. 3D), the latter representing the essential feedback reaction of coagulation for the production of factor XIa (13). Likewise, in the combination of factor XII, kininogen, and prekallikrein, RNA (or isolated DNA) augmented protease formation by >40-fold compared with control (Fig. 3E). As an example for the concentration-dependent augmentation of protease activation by RNA, kallikrein formation in the presence of kininogen is documented in Fig. 3F. The bell-shaped curve reflects a template mechanism of RNA function that is reminiscent of the molecular mechanism described for high molecular mass heparin in catalyzing thrombin inhibition by antithrombin (25).

Materials.

Materials used were as follows: coagulation factors and control proteins (American Diagnostica, Pfungstadt, Germany and Enzyme Research, Essen, Germany); aprotinin (BAYER, Leverkusen, Germany); factor XIIa inhibitor H-d-pro-phe-arg-chloromethylketone (Bachem, Basel, Switzerland); thrombin and corn trypsin inhibitor (Calbiochem, Schwalbach, Germany); hirudin, CJ545, and CJ626 (kind gifts from J. Stuerzebecher, Jena, Germany); recombinant tissue factor and anti-TF 9-6B4 antibody (kindly provided by W. Ruf, La Jolla, CA); chromogenic substrates S-2288, S-2366, and S-2302 (Haemochrom, Essen, Germany); RNase A and DNase I (Fermentas, St. Leon-Rot, Germany); DMEM cell culture medium (GIBCO, Karlsruhe, Germany); yeast tRNA (Sigma, Munich, Germany); coagulation factor-deficient plasmas (DADE–Behring, Liederbach, Germany). Blood samples were taken from healthy volunteers by using 0.38% sodium citrate as anticoagulant and processed within 2 h after withdrawal to obtain platelet-poor-plasma. Aliquots were snapfrozen until further use. Cellular nucleic acids (total cellular RNA or genomic DNA) were isolated from cultured Chinese hamster ovary (CHO) cells or yeast, E. coli, and Qβ-phage by using TRIzol-reagent (Invitrogen, Karlsruhe, Germany), an RNeasy kit, or a genomic DNA isolation kit (Qiagen, Hilden, Germany).

Cell Lysate.

Cell lysate from 50 × 10⁶ CHO cells was generated by exposing washed cells to three freeze-thaw cycles in 1 ml 0.1 M imidazol, pH 7.4, between liquid nitrogen and a 37°C water bath. Cell lysates were cleared by centrifugation (22,000 × g for 15 min), and their procoagulant activity was analyzed in an turbidity assay. In BSA-blocked microtiter plate wells, 30-μl cell lysate dilutions were pretreated with each 10 units per ml of RNase A or DNase I, 10 μg/ml of anti-TF9-6B4 in 0.1 M imidazol buffer, pH 7.4, for 30 min at 37°C. After the addition of 30 μl of normal human plasma, coagulation was initiated after 5 min by recalcification with 100 μl of 20 mM CaCl₂-solution.

Serine Protease Activity in Plasma.

After blocking microtiter plate wells with 3% BSA solution, 10 μl of normal human plasma in a total volume of 100 μl 0.1 M imidazol buffer, pH 7.4, containing 0.2 mM chromogenic substrate S-2288 was incubated in the absence or presence of 10 μg/ml of cellular total RNA or genomic DNA, hydrolyzed RNA, or DNA. As controls, 10 ng/ml of kaolin or 0.8 μg/ml of recombinant tissue factor were added instead of nucleic acids. For inhibitors studies, serial dilutions of aprotinin, corn trypsin inhibitor, hirudin, CJ545, or CJ626 were included, and after addition of 30 μg/ml of total cellular RNA, the reaction was started by the addition of citrated plasma. Absorbance at 405 nm was measured up to 60 min, and V_max was determined by KC4 software (BIO-TEK, Bad Friedrichshall, Germany).

Reconstitution of Contact Phase Pathway.

Isolated coagulation proteins dissolved in 150 mM NaCl/10 mM Tris·HCl, pH 7.4 were added alone or in combinations (see Fig. 3 legend) to preblocked (3% BSA) microtiter wells, and protease activity in the absence or presence of 1 μg/ml each of cellular RNA or genomic DNA was recorded in the presence of 0.2 mM of chromogenic substrate S-2366. Either 50 μM ZnCl₂ or 2 mM CaCl₂ was present in the reaction mixtures. To evaluate the template effect of RNA, 20 μM prekallikrein was incubated with 20 nM high molecular weight kininogen in the presence of increasing concentrations of yeast total RNA and 0.3 mM chromogenic substrate S-2302 and 50 μM ZnCl₂. Protease activity was recorded at 405 nm by using KC4 software (BIO-TEK).

FeCl₃-Induced Arterial Thrombosis.

Wild-type C57BL6/J mice were anesthetized by intraperitoneal injection with imidazolame [5 mg/kg body weight (BW); Ratiopharm, Ulm, Germany], medetomidine (0.5 mg/kg BW; Pfizer, Karlsruhe, Germany), and fentanyl (0.05 mg/kg BW; CuraMed Pharma, Munich, Germany). Polyethylene catheters (Portex, Hythe, U. K.) were implanted into the right jugular vein, and subsequently, RNase A or DNase (100 μl each of a 1 mg/ml solution containing 0.9% NaCl, heat-pretreated to destroy protease activity), factor XIIa inhibitor (10 mg/kg BW) (27), or vehicle (0.9% NaCl) were administered i.v. via the jugular vein. Platelets (60 × 10⁶ platelets) were isolated from separate donor animals and labeled with 5-carboxyfluorescein diacetate succinimidyl ester as described earlier (41) and infused i.v. Finally, the right common carotid artery was dissected free and vascular injury of the carotid artery was induced by local application of ferric chloride (FeCl₃) essentially as described in ref. 21. In brief, a filter paper (0.5 × 1.0 mm) saturated with 10% FeCl₃ was applied to the adventitial surface of the vessel. Arterial thrombosis was monitored by intravital videofluorescence microscopy of fluorescence-tagged platelets (21).

To visualize thrombus-bound RNA, a 25 μM stock solution of SYTO RNASelect green fluorescent cell stain (Molecular Probes, Poortgebouw, The Netherlands) was freshly prepared, and immediately before FeCl₃ application, 60 μl of this solution was injected into the left jugular vein. In the treatment group, 15 μg of RNase A was dissolved in 50 μl of 0.9% NaCl and administered via tail vein injection 30 min and immediately before the described thrombus induction by FeCl₃ treatment. In the control group, the same amount of 0.9% NaCl was administered. RNase A was incubated at 95°C for 15 min to inactivate any decontaminating proteases before administration. Fifteen minutes after FeCl₃ administration, arteries were clamped, and the mice were killed with an overdose of the anesthetic. Arteries were excised and embedded in Tissue Tek OCT (Miles Laboratories, Naperville, IL), snap-frozen, and stored at −80°C until use. Samples were sectioned on a Leica cryostat (6 μm) and placed on slides coated with poly(l-lysine) (Sigma) for immunohistochemical analysis. For morphometric analyses, hematoxylin and eosin staining was performed according to standard protocols. After blocking sections with 10% normal goat serum, platelets were detected by using monoclonal rat anti-mouse CD41 (glycoprotein Ib, 1:200; BD Biosciences, Heidelberg, Germany) (42), and goat anti-rat IgG conjugated to Alexa Fluor 546 (Molecular Probes). Slides were mounted in Vectashield medium H-1000, containing DAPI, 2HCl (Calbiochem), and evaluated by using an epifluorescence microscope (DMRB; Leica, Wetzlar, Germany). All procedures involving experimental animals were approved by the institutional committee for animal research of the Technical University (Munich) and the Justus-Liebig-University (Giessen) and complied with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication no. 86-23, revised 1985).

Statistical Analysis.

Unless otherwise stated, all data are presented as mean ± SEM. All statistics were performed by using SPSS software; siginificance level was set at P < 0.05.