Aprotinin Inhibits Thrombin Generation by Inhibition of the Intrinsic Pathway, but is not a Direct Thrombin Inhibitor

Background  Aprotinin is a broad-acting serine protease inhibitor that has been clinically used to prevent blood loss during major surgical procedures including cardiac surgery and liver transplantation. The prohemostatic properties of aprotinin likely are related to its antifibrinolytic effects, but other mechanisms including preservation of platelet function have been proposed. Aim  Here we assessed effects of aprotinin on various hemostatic pathways in vitro, and compared effects to tranexamic acid(TXA), which is an antifibrinolytic but not a serine protease inhibitor. Methods  We used plasma-based clot lysis assays, clotting assays in whole blood, plasma, and using purified proteins, and platelet activation assays to which aprotinin or TXA were added in pharmacological concentrations. Results  Aprotinin and TXA dose-dependently inhibited fibrinolysis in plasma. Aprotinin inhibited clot formation and thrombin generation initiated via the intrinsic pathway, but had no effect on reactions initiated by tissue factor. However, in the presence of thrombomodulin, aprotinin enhanced thrombin generation in reactions started by tissue factor. TXA had no effect on coagulation. Aprotinin did not inhibit thrombin, only weakly inhibited the TF-VIIa complex and had no effect on platelet activation and aggregation by various agonists including thrombin. Aprotinin and TXA inhibited plasmin-induced platelet activation. Conclusion  Pharmacologically relevant concentrations of aprotinin inhibit coagulation initiated via the intrinsic pathway. The antifibrinolytic activity of aprotinin likely explains the prohemostatic effects of aprotinin during surgical procedures. The anticoagulant properties may be beneficial during surgical procedures in which pathological activation of the intrinsic pathway, for example by extracorporeal circuits, occurs.


Introduction
Aprotinin is a broad acting serine protease inhibitor that has been clinically used to prevent blood loss during major surgical procedures including cardiac surgery and liver transplantation. 1,2 The drug was withdrawn from clinical use in 2007, following a randomized controlled clinical trial that suggested aprotinin use in cardiac surgery was associated with an increased risk of complications and death. 3 The comparator in these studies was a lysine analogue (either tranexamic acid or epsilon aminocaproic acid).
The effect of aprotinin in reducing perioperative blood loss is generally assumed to be related to its antifibrinolytic effect. 4 Aprotinin is a direct inhibitor of plasmin, but also inhibits FXIIa-dependent activation of fibrinolysis by inhibiting kallikrein which activates factor XII in the contact pathway. By inhibition of kallikrein, aprotinin also inhibits activation of coagulation via the contact pathway. Finally, aprotinin has anti-inflammatory effects related to inhibition of kallikrein. A combined antifibrinolytic, anticoagulant, and anti-inflammatory effect of aprotinin is clinically evident by decreased D-dimer, prothrombin fragment 1 þ 2, and bradykinin plasma levels in patients undergoing cardiac surgery receiving aprotinin compared with those who did not. [5][6][7][8][9] The antifibrinolytic effects of aprotinin may also reduce plasminmediated activation of platelets 10,11 or plasmin-mediated proteolysis of platelet receptors such as glycoprotein Ib, 12 thereby preserving platelet function.
As aprotinin is a broad-spectrum serine protease inhibitor, other targets that may explain the mode of action of aprotinin as a hemostatic drug have been proposed. It has even been questioned whether the antifibrinolytic effect is a main contributor to the prohemostatic effect of aprotinin. 13 Other mechanisms that have been proposed to contribute to the prohemostatic properties of aprotinin are plasmin-independent platelet-preserving effects and anticoagulant properties, although it is not entirely clear how anticoagulant properties could contribute to a prohemostatic effect. Conflicting reports on these alternative modes of action of aprotinin have appeared in literature. For example, whereas some studies have reported that aprotinin inhibits platelet activation by a wide variety of agonists, [14][15][16] other studies report that aprotinin inhibits platelet activation by thrombin but not by other agonists. 17,18 Specifically, studies from one laboratory have shown that aprotinin in clinically relevant concentrations inhibits thrombin-mediated platelet activation via protease activated receptor (PAR)-1, but not via 19 This selective inhibition of PAR-1 activation is not explained by direct thrombin inhibition of aprotinin, which would also block PAR-4 activation. In addition, one study has demonstrated that aprotinin is a direct thrombin inhibitor, 18 but the K i reported (61 uM or 2900 KIU/ml) is around an order of magnitude higher than peak plasma concentrations that are clinically observed (Plasma concentrations of aprotinin in patients undergoing cardiac or liver transplant surgery are between 100 and 400 KIU 20-23 ). Clinically, substantial evidence for a platelet-preserving effect of aprotinin exists, [24][25][26][27] and although these platelet-preserving effects have been suggested to be mediated by direct effects of aprotinin on platelets, 28 the effect may also be fully explained by a combination of inhibition of plasmin-mediated platelet activation 10 and cleavage of platelet receptors, 12 and a decrease of thrombin generation via the intrinsic pathway. 7 Also, it has been suggested that aprotinin is an inhibitor of the tissue factor-FVIIa complex, 29 although the aprotinin doses used in this study were supratherapeutic.
To assess effects of aprotinin in clinically relevant concentrations on the hemostatic system, we have studied the effects of aprotinin in a range of in vitro assays using the currently marketed clinical product. Results were compared with tranexamic acid, which is an antifibrinolytic drug that is thought to reduce bleeding only by reducing plasmin formation.

Materials
Aprotinin was a generous gift from Nordic Pharma (Baarn, The Netherlands), tranexamic acid was from Mylan B.V. (Amstelveen, The Netherlands), Dabigatran was purchased from Alsachim (Illkirch Graffenstaden, France), recombinant tissue factor pathway inhibitor (TFPI) was from R&D systems (Minneapolis, MI, USA). Pooled normal plasma was obtained by combining citrated plasma from >200 healthy volunteers.

Clot Lysis Assay
Fibrinolytic capacity in plasma was assessed by monitoring changes in turbidity during clot formation and lysis of a tissue factor-induced clot by exogenous tissue plasminogen activator, as described previously. 30 Clot lysis time was defined as the time from the midpoint of the clear to maximum turbid transition, representing clot formation, to the midpoint of the maximum turbid to the clear transition, representing clot lysis (a typical example of a clot lysis curve can be found here 31 ).

PT/APTT
Prothrombin time (PT) and activated partial thromboplastin time (APTT) assays were performed on an automated coagulation analyzer (STACompact 3, Stago, Breda, the Netherlands) with the use of reagents and protocols from the manufacturer.

Thrombin Generation Test
Thrombin generation assays were performed with the fluorimetric method calibrated automated thrombinography, as described previously by Hemker et al. 32 Coagulation was activated using a diluted PT reagent (Innovin, Siemens, Erlangen, Germany, 1/900 final dilution), or a diluted APTT reagent (Synthasil, Werfen, Breda, The Netherlands, 1/40 dilution).The endogenous thrombin potential was derived with software from Thrombinoscope (Maastricht, the Netherlands). In selected experiments, thrombin generation was performed in the presence of soluble thrombomodulin (10 nM final concentration, Synapse, Maastricht, The Netherlands). Mode of Action of Aprotinin Lisman et al. e364

Viscoelastic Test
Whole blood clot formation was assessed with the ClotPro analyser (Nodia, Breda, the Netherlands), a new generation viscoelastic test that is very similar to TEG and ROTEM (www. clot.pro). Clotting was initiated by commercially available reagents (Ex test and In test) for the extrinsic and intrinsic pathway.

Chromogenic Assays
In a 96-well plate, thrombin (Enzyme Research Laboratories, South Bend, IN, USA, 1 U/ml) was mixed with Biophen Thrombin chromogenic substrate (Nodia, 1 mM) in Trisbuffered saline. Immediately after mixing, optical density at 405 nm was read every 10 seconds for 3 minutes in a VERSAmax reader (Molecular devices, San Jose, CA, USA) at 37°Celcius. The initial slope of the time vs optical density curve reflects the amidolytic activity of thrombin. Similarly, the activity of recombinant factor XIa (purified as described, 33 10 nM), factor XIIa (Enzyme Research Laboratories, South Bend, IN, 19,5 µM), kallikrein (Stago, 2 nM), and activated protein C (a generous gift from the late Dr. Walter Kisiel, University of Albuquerque, New Mexico, 10 nM) were measured toward S2302 (kallikrein and FXIIa) or S2366 (XIa and activated protein C) at a final concentration of 0.5 mM (Chromogenix, Molndal, Sweden). The initial slope of the reactions without inhibitory compounds present was set at 100.
Tissue Factor-factor VIIa-mediated Activation of Factor X In a 96-well plate, tissue factor (Innovin, 1/40 final dilution), was mixed with recombinant factor VIIa (Novo Nordisk, Bagsvaerd, Denmark, 20 ng/ml) in Tris-buffered saline. Factor X was added (Bioconnect, the Netherlands, 10 ug/ml) and incubated for 3 minutes at 37°Celcius. Next, Pefachrome Xa (Stago, 1 mM) was added immediately thereafter, optical density at 342 nm was read every 10 seconds for 20 minutes in a VERSAmax reader at 37°Celcius. The initial slope of the time vs optical density curve reflects the amount of factor Xa generated. The initial slope of the reactions without inhibitory compounds present was set at 100.

Fibrin Generation
In a 96-well plate, thrombin (1 U/ml) was mixed with fibrinogen (Nodia, 2 mg/ml) in Tris-buffered saline. Immediately after mixing, optical density at 340 nm was read every 10 seconds for 20 minutes in a VERSAmax reader. Clotting time was defined as the time from mixing to the midpoint of the clear to maximum turbid transition.

Platelet Aggregation
Platelet-rich plasma was prepared by a centrifugation of human whole blood at 180 g for 10 minutes. Washed platelets were prepared as described previously. 34 Platelet-rich plasma was stimulated with collagen (Stago, Breda, The Netherlands, 2 μg/ml). Washed platelets were stimulated by thrombin (1 U/ml) in the presence of GPRP peptide (Sigma-Aldrich,1 mM). Aggregation was moni-tored using a Chrono-Log Model 700 (Chronolog, Havertown, PA, USA).

Aprotinin and Tranexamic Acid Inhibit Fibrinolysis in Plasma
We first confirmed aprotinin and tranexamic acid to have potent and dose-dependent antifibrinolytic effects in vitro at clinically relevant concentrations. Aprotinin or tranexamic acid were added to pooled normal plasma which was subsequently tested in a plasma clot lysis assay. Both aprotinin and tranexamic acid dose-dependently increased clot lysis time, with no detectable lysis during the three hours the assay is run at the highest doses tested (►Fig. 1). Aprotinin plasma levels between 100 and 400 KIU/ml [20][21][22][23] and TXA plasma levels between 30 and 100 ug/ml 36,37 have previously been measured in patients undergoing cardiac surgery or liver transplantation.

Aprotinin Inhibits Coagulation Activated by the Intrinsic but not the Extrinsic Pathway
Aprotinin and tranexamic acid were added to pooled normal plasma and plasma samples were tested in PT and APTT assays. Aprotinin dose-dependently prolonged the APTT but had no effect on the PT (►Fig. 2A, B). Tranexamic acid had no effect on both PT and APTT (►Fig. 2C, D). Next, we performed thrombin generation tests in which coagulation was initiated by an APTT or a PT reagent. Aprotinin prolonged the lag time (data not shown) and decreased the ETP of a reaction initiated by an APTT reagent, but had no effect on thrombin generation when coagulation was initiated by a PT reagent (►Fig. 3A-B). Tranexamic acid had no effect on thrombin generation regardless of the initiating trigger (►Fig. 3C-D). However, when thrombin generation experiments were performed in the presence of soluble thrombomodulin, aprotinin still inhibited thrombin generation initiated by an APTT reagent, but stimulated thrombin generation initiated by a PT reagent (►Fig. 4). In a whole blood viscoelastic test, clot formation was inhibited by aprotinin only when clot formation was initiated by an activator of the intrinsic pathway (►Fig. 5). Aprotinin and tranexamic acid had no effect on maximal clot amplitude (data not shown). Aprotinin dose-dependently inhibited the activity of purified kallikrein and FXIa, but had limited effect on FXIIa. In addition, aprotinin dose-dependently inhibited activated protein C. Tranexamic acid did not inhibit kallikrein, FXIa, FXIIa, or activated protein C (►Fig. 6).

Aprotinin is not an Inhibitor of Thrombin and only Weakly Inhibits the Tissue Factor-factor VIIa Complex
We assessed the potential inhibitory effect of aprotinin on purified thrombin. First, we added aprotinin or dabigatran to thrombin in the presence of the chromogenic substrate S2238. Whereas aprotinin had no effect on the conversion of the chromogenic substrate by thrombin, dabigatran dosedependently inhibited the activity of thrombin toward the chromogenic substrate (►Fig . 7A, B). Similarly, when thrombin was added to purified fibrinogen, aprotinin had no effect on clot formation, whereas dabigatran dose-dependently delayed clot formation (►Fig. 7C, D). When aprotinin was added to a mixture of tissue factor, factor VIIa, and factor X, a slight decrease of FXa generation was observed at high aprotinin concentrations, whereas TFPI dose-dependently inhibited Xa generation (7E/F).

Aprotinin does not Inhibit Platelet Aggregation Induced by Thrombin or Collagen
Washed platelets were stimulated with thrombin in the presence of GPRP, which prevents fibrin formation, and aggregation was monitored over time. As shown in ►Fig. 8A, B, C, aprotinin or tranexamic acid did not prevent platelet aggregation by thrombin, but dabigatran dosedependently blocked aggregation. In platelet-rich plasma, aprotinin or TXA had no effect on collagen-induced aggregation (►Fig. 9A, B).

Aprotinin Inhibits Platelet Activation by Plasmin, but has no Effect on Platelet Activation by Various Agonists
When whole blood was stimulated with TRAP, CRP, 2MeS-ADP, or thrombin, aprotinin (600 KIU/ml) or tranexamic acid (400 ug/ml) had no effect on platelet activation as assessed by platelet P-selectin expression (►Fig. 10A) or PAC-1 binding (►Fig. 10B), which reflects activation of integrin αIIbβ3. In contrast, both tranexamic acid and aprotinin fully blocked plasmin-induced platelet activation.

Discussion
This study shows that aprotinin at clinically relevant concentrations has both antifibrinolytic and anticoagulant properties, and inhibits plasmin-induced platelet activation. The anticoagulant effect of aprotinin is restricted to inhibition of intrinsic activation mediated by inhibition of kallikrein and factor XIa. Under our experimental conditions, aprotinin did not inhibit thrombin, thrombin-mediated fibrinogen to fibrin conversion, or platelet activation by various agonists including thrombin, and only weakly inhibited the TF-VIIa complex. The inhibitory effect on the purified TF-VIIa complex is likely physiologically insignificant, as aprotinin did not inhibit TF-induced thrombin generation in plasma. Aprotinin increased thrombin generation via the extrinsic pathway in the presence of thrombomodulin, likely as a result of inhibition of activated protein C. 38,39 However, as aprotinin decreases thrombin generation in vivo, 5-7 it is unlikely this prohemostatic effect of aprotinin is clinically relevant. Collectively, our results suggest that plasmin and kallikrein are the only relevant hemostatic targets of aprotinin at clinically used concentrations in a plasma or whole blood environment. Tranexamic acid only had antifibrinolytic properties, and thus mechanistically, aprotinin has two theoretical advantages over tranexamic acid. First, the antiinflammatory properties via inhibition of kallikrein may be beneficial. Second, the inhibitory effects of aprotinin on intrinsic activation of coagulation may mean that undesired activation of coagulation, for example by cardiopulmonary bypass circuits, is prevented by aprotinin, whereas physiological hemostasis remains intact. Aprotinin thus may function as an anticoagulant that is antithrombotic without substantially increasing bleeding risk, similar to new-generation antithrombotic agents that target FXI or FXII. 40 In our in vitro experiments, the antifibrinolytic and anticoagulant effects of aprotinin are apparent at plasma concentrations of 75 KIU and higher. As the plasma concentrations of aprotinin in vivo (in patients undergoing cardiac or liver transplant surgery) are between 100 and 400 KIU, [20][21][22][23] it is likely that the antifibrinolytic and anticoagulant effects of aprotinin described herein are clinically relevant. Indeed, aprotinin decreases plasma levels of D-dimer and prothrombin fragment 1 þ 2 in patients undergoing cardiac surgery. [5][6][7][8][9] Our data are also in agreement with previous studies demonstrating in vivo platelet-preserving effects of aprotinin. [24][25][26][27] By inhibition of plasmin, aprotinin prevents or decreases plasmin-mediated proteolysis of platelet receptors and plasmin-mediated platelet activation. In addition, by inhibition of thrombin generation via the intrinsic pathway, aprotinin may also prevent platelet activation.
We do not confirm previous data on direct inhibitory effects of aprotinin on platelet activation, and do not find aprotinin to be a relevant thrombin inhibitor at the doses tested (up to 600 KIU/ml, which is supratherapeutic), which is in line with the K i of 61 uM or 2900 KIU/ml previously reported. Why we fail to find inhibitory effects of aprotinin on thrombin-induced platelet activation, or on platelet activation induced by other agonists is not entirely clear. We thus have no clear explanation for the divergent results with previous studies but do note that 1) the high K i for thrombin does not readily explain why aprotinin at clinically relevant concentrations would inhibit thrombin-mediated platelets activation and 2) previous studies on the inhibitory effect of   Fig. 7 Effects of aprotinin on thrombin and the tissue factor-factor VIIa complex. A and B: Aprotinin or dabigatran were added in increasing concentrations to purified thrombin after which thrombin activity was assessed using the chromogenic substrate S2238. C and D: aprotinin and dabigatran were added in increasing concentrations to purified fibrinogen, after which thrombin was added. Clot formation was monitored by turbidity measurements over time. E and F: Aprotinin or TFPI were added to the tissue factor-factor VIIa complex in the presence of factor X and factor Xa generation over time was assessed using a chromogenic substrate. Shown are mean values of 3 independent experiments. Error bars indicate standard deviations (but are not visible for the majority of data points as standard deviations were very low). aprotinin on the direct thrombin-independent platelet inhibitory properties also yielded contradictory results. [14][15][16][17]19 Taken together, our in vitro studies show that aprotinin has both antifibrinolytic and anticoagulant activity at clinically relevant concentrations. We propose that the antifibrinolytic activity is primarily responsible for the prohemostatic effects of aprotinin during surgical proce-dures. In addition, we propose that the anticoagulant properties may be beneficial in clinical situations in which substantial activation of the intrinsic pathway of coagulation occurs. As aprotinin is a selective inhibitor of intrinsic coagulation activation, it will dampen undesired coagulation activation by, for example extracorporeal circuits, without interfering with physiological hemostasis, and will therefore not contribute to perioperative bleeding. Aprotinin has now carefully been reintroduced, but may have clinical applications beyond complex cardiac surgery. For example, the anticoagulant properties of aprotinin may be exploited in clinically challenging situations in which activation of the intrinsic pathway leads to thrombotic events. Such scenarios include extracorporeal circuits used outside cardiac surgery, such as ECMO. As we have shown that aprotinin is not a direct thrombin or platelet inhibitor, aprotinin is not expected to increase bleeding risk in these scenarios.

What is Known
• Aprotinin is a clinically used antifibrinolytic with an incompletely characterized mode of action • Aprotinin is an inhibitor of plasmin but also inhibits kallikrein which leads to inhibition of intrinsically activated coagulation and of inflammation • Other proposed targets of aprotinin are the tissue factor-VIIa complex, thrombin, platelets, and protease activated receptor 1 on platelets

What this Study Adds
• We studied hemostatic effects of aprotinin in various in vitro models • Aprotinin inhibits fibrinolysis and the intrinsic pathway of coagulation, but does not inhibit platelet function, thrombin, or TF-VIIa • The anticoagulant properties of aprotinin might be beneficial during surgeries using extracorporeal circuits Author Contributions TL: Designed the study, supervised experimental work, analyzed and interpreted data, wrote the manuscript; JA: Performed, analyzed and interpreted experiments, revised the manuscript; DH: Performed, analyzed and interpreted experiments, revised the manuscript; JCMM: Designed the study, interpreted data, revised the manuscript. All authors approved the final version of the manuscript.

Funding
This study was sponsored in part by an unrestricted educational grant of Nordic Pharma to TL. In addition, this collaboration project is co-financed by the Ministry of Economic Affairs and Climate Policy of the Netherlands by means of the PPP-allowance made available by the Top Sector Life Sciences & Health to stimulate public-private partnerships.

Conflicts of Interest
This study was sponsored in part by an unrestricted educational grant of Nordic Pharma to TL. The other authors have no conflicts of interest to report.