OBJECTIFS DE L’ETUDE - Effets cytoprotecteurs des inhibiteurs physiologiques de la coagulation

CHAPITRE 3 : PROPRIETES CYTOPROTECTRICES ET ANTI-INFLAMMATOIRES DE L’ANTITHROMBINE

B. Effets cytoprotecteurs des inhibiteurs physiologiques de la coagulation

II. OBJECTIFS DE L’ETUDE

Comme nous l’avons rappelé dans la première partie de ce mémoire, l’HNF est le seul dérivé héparinique à possèder un antidote, le sulfate de protamine. La protamine neutralise partiellement l’activité anticoagulante des HBPMs mais est inefficace vis-à-vis du fondaparinux. Dans ces conditions, en cas de surdosage en fondaparinux et de complications hémorragiques, l’utilisation d’agents hémostatiques ne ciblant pas spécifiquement le fondaparinux (FVIIa ou CCPa) est la seule alternative. Cependant, leur utilisation induit un état prothrombotique transitoire non souhaitable chez des patients traités dans le cadre d’une thrombose (145, 304). Deux ATs inactivées, l’une obtenue par modification chimique et l’autre par production d’un variant d’AT recombinant, ont déjà été caractérisées pour leur capacité à neutraliser les dérivés hépariniques. A l’aide d’un test mesurant l’activité inhibitrice du FXa (anti-FXa) des dérivés hépariniques, notre équipe a montré que ces deux ATs sont efficaces pour neutraliser leur activité anticoagulante aussi bien in vitro qu’in vivo (305,306). Cependant, l’AT agit au sein de la cascade à plusieurs niveaux. Elle est capable d’inhiber le FXa, mais aussi plusieurs autres protéases comme la thrombine, le FIXa ou le FVIIa. Nous avons donc cherché à : (i) Evaluer l’impact de nos ATs inactivées sur la coagulation de façon globale et pas seulement au travers d’une mesure d’activité anti-FXa. Pour cela, nous avons choisi un test explorant de façon global la coagulation : le test de génération de thrombine (ii) Comparer leur effet antidote aux stratégies thérapeutiques utilisées à l’heure actuelle en cas de surdosage de chacun des différents dérivés hépariniques.

Ces ATs inactivées pourraient également être utilisées dans une autre indication, comme agent cytoprotecteur dans l’ischémie reperfusion. En effet, l’AT possède des propriétés antiinflammatoires et cytoprotectrices à des concentrations supra-physiologiques (au-delà de 500%) dans des modèles de sepsis et d’ischémie reperfusion touchant différents organes (cœur, rein, foie…) (227,276,237,307). L’utilisation de l’AT à de telles doses est cependant incompatible avec une utilisation clinique en raison de l’activité anticoagulante de la protéine et du risque hémorragique associé. Les propriétés anti-inflammatoires et cytoprotectives de l’AT semblent indépendantes de son activité anticoagulante mais l’intégrité de son site de liaison à l’héparine est indispensable. Notre AT inactivée recombinante, développée comme antidote, pourrait être utilisée à fortes concentrations comme agent protecteur dans l’ischémie reperfusion. Nous avons donc décidé d’évaluer cette AT inactivée recombinante dans un modèle d’I/R rénale murin au cours duquel l’AT plasmatique a déjà été montré comme capable d’améliorer la fonction rénale, et de réduire les cytokines pro-inflammatoires, l’infiltrat leucocytaire ainsi que les lésions tubulaires rénales (276).

III. Evaluation d’Antithrombines inactivées comme

antidotes des dérivés hépariniques dans un test

Deux ATs inactivées ont été développées au laboratoire dans le but de développer un antidote vis-à-vis de l’ensemble des dérivés hépariniques. Deux approches ont été utilisées pour inactiver ces ATs tout en augmentant leur affinité pour l’héparine. La première est une approche recombinante qui a consisté à insérer une proline entre les résidus P1et P1’ de la boucle réactive de l’AT. Ces résidus sont critiques pour son interaction avec les serine-protéases cibles de la coagulation. Pour augmenter la liaison à l’héparine, une substitution de l’Asparagine en position 135 par une glutamine a été réalisée. Cette mutation entraine la perte d’une séquence consensus de glycosylation et conduit donc à l’absence de la chaine glycane à cette position, sachant que la présence de cette chaine glycane entraine une diminution de l’affinité pour les dérivés hépariniques (308). Ce double variant ATN135Q-Pro394 présente une réduction de l’activité anticoagulante d’un facteur 1000 et une augmentation de l’affinité à l’héparine d’un facteur 3 (305). La seconde approche a consisté en une modification chimique d’AT plasmatique par un réactif, la 2,3-butanedione, une dicétone capable de réagir avec les groupements amines et de modifier les arginines. Cette modification est réalisée après immobilisation de l’AT sur une colonne d’héparine pour protéger les résidus arginine du site de liaison, tout en exposant l’arginine en P1 au niveau de la boucle réactive de la Serpine (306). L’AT modifiée chimiquement présente une perte d’activité anticoagulante modérée d’un facteur 10 et un gain d’affinité pour l’héparine d’un facteur 7 (306). Dans un test mesurant l’activité inhibitrice vis-à-vis du FXa, ces deux ATs présente la même efficacité à neutraliser un surdosage des trois dérivés hépariniques in vitro et in vivo. Cependant l’utilisation de ce test ne permet ni de comparer nos ATs inactivées aux agents actuellement utilisés pour reverser l’activité anticoagulante des dérivés hépariniques, ni d’apprécier leur impact sur la coagulabilité du plasma. Afin de poursuivre la caractérisation de nos ATs, nous avons choisi un test global de la coagulation, le test de génération de thrombine (TGT). Contrairement aux autres tests de la coagulation classiques qui n’évaluent que la phase d’initiation de la coagulation, le TGT permet d’évaluer aussi bien la phase d’amplification que la phase d’inhibition de la coagulation. Le TGT permet de suivre la quantité de thrombine générée en fonction du temps après activation de la coagulation par le facteur tissulaire dans un plasma pauvre ou riche en plaquettes grâce à un substrat fluorogénique même après formation du caillot. Le résultat obtenu est représenté par une courbe appelée thrombinographe à partir duquel sont exploités les paramètres suivants : le « lag-time » (le temps pour la détection de la thrombine), le « time to peak » (le temps pour l’obtention du pic de thrombine générée), le « peak » (le pic de génération de thrombine) et enfin l’ETP « endogenous thrombin potentiel » (l’aire sous la courbe, qui représente la quantité totale de

thrombine générée). Ces paramètres permettent d’apprécier l’activité anticoagulante ou procoagulante d’un plasma (309).

Inactivated antithrombins as fondaparinux antidotes: a promising

alternative to haemostatic agents as assessed in vitro in a

thrombin-generation assay

Yasmine Bourti1; Judicael Fazavana1; Marine Armand2; François Saller1; Dominique Lasne1,2; Delphine Borgel*,1,2; Elsa P Bianchini*,1 1INSERM UMR-S1176, Univ. Paris-Sud, Université Paris-Saclay, Le Kremlin-Bicêtre, France; 2APHP, Laboratoire d’Hématologie, Hôpital Universitaire Necker–Enfants Malades, Paris, France

Summary

In the absence of specific antidote to fondaparinux, two modified forms of antithrombin (AT), one recombinant inactive (ri-AT) and the other chemically inactivated (chi-AT), were designed to antagonise AT-mediated anticoagulants, e. g. heparins or fondaparinux. These inac-tive ATs were previously proven to effecinac-tively neutralise anticoagulant activity associated with heparin derivatives in vitro and in vivo, as as-sessed by direct measurement of anti-FXa activity. This study was undertaken to evaluate in vitro the effectivity of inactive ATs to re-verse anticoagulation by heparin derivatives and to compare them with non-specific fondaparinux reversal agents, like recombinant-acti-vated factor VII (rFVIIa) or actirecombinant-acti-vated prothrombin-complex concentrate (aPCC), in a thrombin-generation assay (TGA). Addition of fondapari-nux (3 µg/ml) to normal plasma inhibited thrombin generation by pro-longing lag time (LT) as much as 244 % and lowering endogenous

thrombin potential (ETP) to 17 % of their control (normal plasma) values. Fondaparinux-anticoagulant activity was reversed by ri-AT and chi-AT, as reflected by the corrections of LT up to 117 % and 114 % of its control value, and ETP recovery to 78 % and 63 %, respectively. Un-like ri-AT that had no effect on thrombin generation in normal plasma, chi-AT retained anticoagulant activity that minimises its reversal ca-pacity. However, both ATs were more effective than rFVIIa or aPCC at neutralising fondaparinux and, unlike non-specific antidotes, inactive ATs specifically reversed AT-mediated anticoagulant activities, as sug-gested by their absence of procoagulant activity in anticoagulant-free plasma.

Keywords

Antithrombin, fondaparinux, heparin, reversal, thrombin-generation assay

Correspondence to:

E. P. Bianchini

UMR-S1176, 80 rue du Général Leclerc 94276 Le Kremlin-Bicêtre Cedex, France Tel.: +33 1 49595646, Fax: +33 1 46719472 E-mail: elsa.bianchini@u-psud.fr

Financial support:

This work was supported by the French National Research Agency, grant ANR-11-RPIB-0018. Y. Bourti received a grant from CORDDIM (domaine d’intérêt majeur „Cardiovasculaire – Obésité- Rein – Diabète“, Île-de-France).

Received: December 4, 2015

Accepted after major revision: May 14, 2016 Epub ahead of print: July 14, 2016

http://dx.doi.org/10.1160/TH15-12-0927 Thromb Haemost 2016; 116: ■■■ * D. Borgel and EP. Bianchini contributed equally to this work.

Coagulation and Fibrinolysis

Introduction

Heparin derivatives are widely used as treatment or prophylaxis of thrombotic diseases. Unfractionated heparin (UFH), the first coagulant approved clinically, has been a reference drug for anti-thrombotic therapy, until low-molecular-weight heparin (LMWH) and, more recently, fondaparinux were developed. The latter have more predictable anticoagulant activity (1). However, like other anticoagulants, the major heparin-associated complication is the increased risk of bleeding that may require rapid anticoagulant-ac-tivity neutralisation before emergency invasive surgery or to stop bleeding (2). Protamine easily neutralises UFH but only partially reverses LMWH anticoagulant activity and is inactive against fon-daparinux (3, 4). Because no fonfon-daparinux antidote is available, in

vitro and in vivo study results suggested using off-label

haemos-tatic agents, like recombinant-human activated-factor VII (rFVIIa) or activated prothrombin-complex concentrate (aPCC),

to counteract fondaparinux anticoagulation (5–10). Several case reports and large-scale meta-analyses (11–14) support using rFVIIa for serious bleeding complications or acute surgery required under fondaparinux, but off-label rFVIIa or aPCC use for critical bleeding is associated with more thromboembolic events (15, 16). Thus, the risk of thrombotic complications should be considered, especially in patients taking anticoagulants for a thrombosis risk.

To specifically reverse fondaparinux and other heparin deriva-tives with antithrombin (AT)-mediated anticoagulant activities, two inactive AT derivatives with reduced/abolished inhibitory ac-tivity against procoagulant enzymes were developed (17, 18); they remain able to tightly bind heparin derivatives. Recombinant inac-tive AT (ri-AT) was obtained by inserting a proline residue be-tween the arginine393 and serine394 residues within the reactive center loop, and by replacing asparagine135 with a glutamine resi-due (19). The first mutation inhibited enzymatic activity by at least

Thrombosis and Haemostasis 116.3/2016 © Schattauer 2016

2 Bourti et al. Inactive antithrombins as potent antidotes to heparins

3-log₁₀, while the second, resulting in the absence of the carbohy-drate chain at position 135, increased fondaparinux affinity three-fold compared to human plasma AT (17). Chemically inactivated AT (chi-AT) was obtained by incubating plasma-derived AT with 2,3-butanedione that reacts specifically with arginine side chains. The chemical modification was performed in the presence of UFH to protect arginine residues involved in heparin binding and spe-cifically targets arginine393, which is more solvent-exposed when AT is bound to heparin (20). This product lost ~90 % of its antico-agulant activity but binds 7-fold more tightly to fondaparinux than plasma AT (18). Both inactive ATs showed potent reversal ability against fondaparinux and other AT-mediated anticoagulants, like UFH or LMWH in vivo, in fondaparinux- or UFH-treated mice, and in vitro, in human plasma containing high fondaparinux, LMWH or UFH concentrations (17, 18).

The reversal activities of inactive ATs were previously assessed in an anti-FXa assay that measures exogenous FXa inhibition by native AT present in plasma. That assay determines only AT and AT-mediated anticoagulant activities; it is not appropriate to esti-mate the overall tendency of a plasma sample to form a clot, and thus to compare inactive ATs with other haemostatic agents, al-ready proposed as non-specific fondaparinux antidotes (5).

Herein, inactive ATs’ reversal activities against AT-mediated anticoagulants was evaluated in vitro, in a thrombin-generation assay (TGA), which is considered better than traditional tests for evaluating in vitro anticoagulation reversal (21). To do so, ri-AT or chi-AT was added to human plasma and human plasma supple-mented with fondaparinux, LMWH or UFH. This global coagu-lation assay enabled direct comparison of inactive ATs vs haemos-tatic agents for their abilities to antagonise heparin derivatives and their impacts on thrombin generation when added to plasma. Methods

Preparation of inactive-AT derivatives

ri-AT and chi-AT were prepared as previously described (17, 18). Briefly, the plasmid carrying the cDNA coding for mutated ri-AT was prepared by QuickChange site-directed mutagenesis (Agilent Technologies) and transfected into HEK-293 cells (LGC Stan-dards). A stable expression-cell line was selected and expanded. ri-AT was purified from conditioned media heparin-affinity (HiPrep Heparin FF, GE Healthcare), gel-filtration (Hiload 16/60 Super-dex™ 200, GE Healthcare) and ion-exchange (resource Q, GE Healthcare) chromatographies, as described previously (17).

To prepare chi-AT, purified plasma-derived AT (Aclotine®, kindly provided by LFB, France) was loaded onto a heparin–Sep-harose column (HiTrap heparin HP, GE Healthcare), equilibrated in reaction buffer (100 mM phosphate, 150 mM NaCl, 10 mM EDTA, pH 7.5) containing 2,3-butanedione (6.5 g/l; Sigma-Al-drich), sealed, and incubated at 37 °C for 20 hours (h) in the dark. After extensive washing with reaction buffer containing 0.8 M NaCl to remove free 2,3-butanedione and low-heparin-affinity AT, the high-heparin-affinity AT (called chi-AT) was finally step-eluted with reaction buffer containing 2 M NaCl (18).

Both products were concentrated by ultrafiltration with a 30000-Da cut-off membrane and recovered with phosphate-buf-fered saline, pH 7.4 (PBS, Invitrogen).

Plasma samples

Commercially available frozen pools of citrated normal human plasma (Cryocheck™ Pooled Normal Plasma, Cryopep, France) were used as platelet-poor plasma (PPP). Three different lots of PPP were used, so that all the experiments relative to one heparin derivative (fondaparinux, LMWH, or UFH) were conducted in a unique lot of PPP. The 1-ml aliquots were conserved at –80 °C and thawed (incubation at 37 °C for 4 minutes [min]) just before the experiment. Platelet-rich plasma (PRP) was prepared from whole blood of healthy volunteers collected in citrate-containing tubes (9:1 v/v of blood/3.2 % sodium citrate). Immediately after being drawn, blood (25 ml) was centrifuged at 122×g for 10 min at 20 °C and the resulting supernatant was aliquoted in two 6-ml fractions. Fraction 1 was centrifuged at 2500×g for 10 min at 20 °C to obtain PPP; fraction 2 platelets were counted and adjusted to 1.5×1011/l by dilution with autologous PPP. All experiments were performed within 1 h after blood was drawn with PRPs from three different donors.

TGA

Thrombin generation was measured in vitro in plasma samples prepared according to the following protocols. First, the dose-re-sponses were determined by diluting increasing concentrations of fondaparinux (0.1–3 µg/ml; Arixtra®, GlaxoSmithKline), LMWH (0.1–1.6 U/ml; enoxaparin sodium, Lovenox®, Sanofi) or UFH (0.05–0.5 U/ml; heparin sodium Choay®, Sanofi) in PPP or PRP from respective stock solutions at 12.5 g/l, 10,000 U/ml or 5,000 U/ml, such that the plasma dilution by the anticoagulant solutions was minimal and never exceeded 0.05 %. Second, activities of inac-tive ATs, protamine (protamine sulfate Choay®, Sanofi), rFVIIa (NovoSeven®, Novo Nordisk) or aPCC (FEIBA®, Baxter) were tested alone in anticoagulant-free plasma or for their abilities to re-verse anticoagulant activity in plasma containing fondaparinux (3 µg/ml), LMWH (1.6 U/ml) or UFH (0.5 U/ml). Each of the anti-dotes was prepared as a 10X-concentrated solution in PBS and tested at the following final concentrations: 75–1200 µg/ml of in-active ATs, 0.6–2.4 µg/ml of rFVIIa, 0.5–2 U/ml of aPCC and 0.25–2 anti-heparin units (AHU)/ml for protamine. Control wells were run without anticoagulant or antidote.

Thrombin generation was measured using the Calibrated Auto-mated Thrombogram assay (Stago, Asnieres, France). All TGAs were run in 96-well plates under standard conditions. Briefly, 72 µl of plasma containing the indicated anticoagulant concentration and 8 µl of 10X antidote solution (or PBS) were mixed with 20 µl of PPP-reagent® (5 pM tissue factor and 4 µM phospholipids) or PRP-reagent® (1 pM tissue factor, Stago) in test wells, depending whether the experiments were conducted in PPP or PRP respect-ively, or with thrombin calibrator (Stago) in calibration wells. Thrombin generation was triggered by dispensing 20 µl of

Bourti et al. Inactive antithrombins as potent antidotes to heparins

escent substrate diluted in calcium-containing buffer (FluCa® Kit, Stago), and the thrombin-generation curves were monitored and analysed with Thrombinoscope software (Stago). TGA parame-ters, i. e. lag-time (LT), time-to-peak (TTP), peak height (PH) and endogenous thrombin potential (ETP), were calculated using Thrombinoscope software. Each TGA parameter measured in the presence of antidote and/or anticoagulant was compared to its in-itial value measured without anticoagulant or antidote by one-way ANOVA followed by Dunnett’s test using the GraphPad Prism software, with p<0.05 considered as statistically significant. Results

TGA dose-response of each heparin-like anticoagulant

When tested in normal plasma, all the anticoagulants induced a rightward shift and flattening of the thrombin-generation curves in a dose-dependent manner, resulting in LT and TTP prolon-gation, and lower PH and ETP. TTP always varied proportionally to LT and PH changes always followed ETP variations, hence, throughout this study, only LT and ETP were used to compare thrombin-generation–curve variations. The TGA was very sensi-tive to fondaparinux; increasing its concentrations up to 3 µg/ml prolonged LT from 3.0 ± 0.2 to 7.2 ± 0.4 min and lowered ETP from 1739 ± 240 to 297 ± 81 nM.min (

▶

Figure 1). That high fondaparinux concentration was chosen to mimic severe over-dose in subsequent experiments, given that the optimal plasma concentration during treatment of venous thromboembolic dis-ease ranges from 1.20 to 1.26 µg/ml (22). The TGA was even more sensitive to LMWH and UFH, because increasing their

re-spective concentrations up to 1.6 U/ml or 0.5 U/ml modified all TGA parameters to an anticoagulated pattern (

▶

Figure 2). For the next experiments, the maximum LMWH and UFH concen-trations, respectively, were set at 1.6 U/ml and 0.5 U/ml; the latter is the highest TGA-detectable UFH concentration. Indeed, at that concentration, the thrombin peak was barely detectable and TGA parameters were hardly measurable, thereby explaining the broad standard deviation of the LT means at high UFH con-centrations.

Inactive ATs vs other reversal agents to neutralise fondaparinux

Inactive AT abilities to counteract fondaparinux were assessed in PPP containing fondaparinux (3 µg/ml). Both ri-AT and chi-AT neutralised high-dose fondaparinux but their neutralisation pro-files differed. Indeed, increasing ri-AT concentrations almost tot-ally restored the thrombin-generation curves right-shifted and flattened by anticoagulant addition (

▶

Figure 3 A). While addition of high-dose fondaparinux to normal plasma decreased ETP to 17 % of the control value measured without antidote or anticoagu-lant; further addition of ri-AT restored ETP up to 78 % of its con-trol value (

▶

Figure 3 D). The same was true for LT which was pro-longed 244 % with fondaparinux and corrected to 117 % by ri-AT (

▶

Figure 3 C). This ri-AT effect is specifically due to direct fonda-parinux neutralisation and not to an adverse procoagulant effect, because even when added at high concentration to anticoagulant-free plasma (up to 800 % of the endogenous plasma AT concen-tration), the thrombin-generation curves remained strictly super-imposable and none of the TGA parameters were significantly modified by ri-AT addition (

▶

Figure 3 B, C and D).

Figure 1: Fondaparinux dose-response in nor-mal plasma. A

represen-tative set of thrombin-gen-eration curves was rec-orded in PPP containing increasing fondaparinux concentrations: 0 µg/ml (solid black line, control condition), 0.2 µg/ml (dashed black line), 0.6 µg/ml (dash-dot black line), 1 µg/ml (dotted black line), 1.4 µg/ml (solid gray line), 2 µg/ml (dashed gray line), or 3 µg/ml (dash-dot gray line). The inset shows LT and ETP means ± SD of at least three distinct ex-periments; *indicate valu-es significantly different (p<0.05) from the control.

4 Bourti et al. Inactive antithrombins as potent antidotes to heparins

Figure 2: LMWH and UFH dose-responses in normal plasma.

Throm-bin generation was measured in PPP contain-ing increascontain-ing concen-trations of LMWH (top) or UFH (bottom), and the LT (left) and ETP (right) means ± SD are reported

Dans le document Evaluation d'un variant d'antithrombine dans différentes indications thérapeutiques (Page 75-140)