Heparin and its analogs, mediating their anticoagulant activity through antithrombin (AT) activation, remain largely used for the preventive and curative treatment of thrombosis. The major adverse reaction of these drugs is the bleeding risk associated with overdose. Unfractionnated heparin (UFH) can be efficiently and rapidly neutralized by protamine sulfate, but this reversal partially neutralizes low-molecular-weight heparin (LMWH) and is inefficient in reversing fondaparinux. To secure administration of AT-mediated anticoagulants and counteract bleeding disorders, we previously designed a recombinant inactive AT as an antidote to heparin derivatives.
To get around the limited production level of recombinant AT, we propose in this study an alternative strategy to produce a chemically modified inactive AT, exhibiting increased heparin affinity, as an antagonist of heparin analogs.
Plasma-derived AT was chemically modified with 2,3 butanedione, a diketone known to specifically react with the arginine side chain. The chemical reaction was conducted in the presence of heparin to preserve basic residues within the heparin binding site from modifications.
AT treated by butanedione and selected for its high heparin affinity (AT-BD) was indeed modified on reactive Arg393 and thus exhibited decreased anticoagulant activity and increased heparin affinity. AT-BD was able to neutralize anticoagulant activity of heparin derivatives in vitro and in vivo and was devoid of intrinsic anticoagulant activity, as assessed by activated partial thromboplastin time assay.
AT-BD appears to be as efficient as protamine to neutralize UFH in vivo but could be more largely used because it also reverses fondaparinux and LMWH.
Despite the emergence of new anticoagulant drugs such as direct factor Xa (FXa) or thrombin (FIIa) inhibitors, heparin or its derivatives remain widely used for the treatment and prophylaxis of venous thromboembolism (VTE). Heparin derivatives act as AT-dependent inhibitors of coagulation and include unfractionated (UFH) and low-molecular-weight heparins (LMWHs). UFH had been the antithrombotic drug of reference until LMMW, and more recently fondaparinux, became excellent alternatives to UFH. Fondaparinux is a synthetic analog of a short pentasaccharidic motif found in UFH and LMWH, whose anticoagulant activity is thus mediated through AT-dependent inactivation of FXa. Its favorable bioavailability and its plasma half-life of 17 h permit once-daily administration . Furthermore, only three controversial cases of heparin-induced thrombocytopenia (HIT) have been reported to date in patients treated with fondaparinux [2-4], and its synthetic nature avoids ‘allergic-type’ adverse effects due to contamination by over-sulfated chondroitin sulfate, as reported recently for heparin .
So far UFH is the drug of choice to ensure anticoagulation of blood in extracorporeal circulation devices during cardiac surgery or hemodialysis. Indeed, UFH has a short half-life and can further be rapidly and totally neutralized by protamine sulfate, in the case of severe spontaneous or intraoperative bleedings. In contrast, protamine only partially neutralizes the anticoagulant activity of LMWH and has no effect on fondaparinux, thereby preventing the use of fondaparinux as an anticoagulant drug, notably in cardiac surgery. Although UFH and protamine have both been reported to induce frequent adverse reactions, such as immune responses or bleeding complications [7-9], no safer and more efficacious alternative has been approved by clinical trials and the association of UFH and protamine thus remains the current standard of care for anticoagulation and further neutralization during cardiopulmonary bypass (CBP).
Recently, a recombinant variant of AT has been described as a reversal agent of all heparin derivatives . This variant is devoid of anticoagulant activity but retains a high heparin affinity, thus trapping all heparin derivatives, including fondaparinux, in an inactive complex by competition with endogenous plasma AT. The main limitation of the clinical use of such a potential antidote is the low level of production of recombinant AT in eukaryotic expression systems. These are incompatible with the large amounts required to counteract the high concentrations of endogenous AT in plasma. To circumvent this problem, an inactive AT was prepared on a large scale by chemical modification of purified plasma-derived AT with 2,3-butanedione. This α-diketone is known to specifically react with arginine side chains, including, very likely, Arg393, which is crucial for the anticoagulant activity of AT.
The resulting 2,3-butanedione-modified AT (AT-BD) exhibited decreased anticoagulant activity, tightly bound to heparin derivatives, and thus efficiently reversed their anticoagulant activity in vitro and in vivo. In vivo, AT-BD was as efficient as protamine for neutralizing UFH. However, in contrast to protamine, AT exhibited no intrinsic anticoagulant properties when added at high concentrations in plasma.
Preparation of 2,3-butanedione-modified AT (AT-BD)
Twenty milliliters of purified plasma-derived AT (Aclotine®, kindly provided by LFB, France) at 3 g L−1 in 100 mM phosphate buffer, 150 mM NaCl, 10 mM EDTA, pH 7.5 (reaction buffer), were loaded onto four heparin-sepharose columns (Hi-trap HP 5 mL, GE Healthcare, Velizy-Villacoublay, France) in series and equilibrated in reaction buffer. Free AT was washed out with the same buffer before the injection of 2,3-butanedione (Sigma-Aldrich, Saint Quentin Fallavier, France) (20 mL at 6.5 g L−1 in reaction buffer) onto the columns. The columns were then sealed and incubated at 37 °C for 20 h in the dark. The columns were extensively washed with reaction buffer containing 0.8 M NaCl to remove unreacted 2,3-butanedione and AT forms with low heparin affinity. High heparin affinity AT (referred as AT-BD) was finally step-eluted with reaction buffer containing 2 M NaCl, concentrated by ultrafiltration on a 30 kDa membrane and recovered with phosphate buffer saline pH 7.4 (PBS, Invitrogen, Saint Aubin, France). AT-BD concentrations were estimated by Bradford protein assay. Control AT (cAT) was prepared from plasma AT strictly as described above, but in the absence of 2,3-butanedione.
Mass spectrometry analysis
Peptides resulting from cAT or AT-BD digestion with endoproteinase were analyzed by matrix-assisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOF MS) . Prior mass spectrometry analyses, cAT and AT-BD were either treated with trypsin (Roche, Diagnostics, Meylan, France) for 16 h at 37 °C using a trypsin to AT ratio of 1:50 (w:w) according to the manufacturer's recommendations, or treated with the endoproteinase Asp-N followed by peptides separation on a C18 reverse phase column (Uptisphere, 300Å, 5 μm, 25 cm × 3 mm, Interchim, Montluçon, France) by high pressure chromatography (RP-HPLC). AT-BD or cAT was treated with Asp-N after desalting onto a Hitrap Desalting column 5 mL (GE Healthcare); 600 μg of lyophilized ATs were diluted in 100 mM of ammonium bicarbonate buffer, pH 8, containing 8 M urea, 36 mM DTT. Samples were denatured and reduced for 30 min at 55 °C and then alkylated by addition of 63 mM iodoacetamide for 30 min in the dark at room temperature. Samples were then diluted 10-fold in water and endoproteinase digestion was initiated by addition of 4.8 μg Asp-N (Roche Diagnostics). Digestion proceeded for 16 h at 37 °C with a supplemental spike of 4.8 μg Asp-N after 8 h of incubation; 100 μg of the protein digest were analyzed by RP-HPLC.
Kinetics of protease inhibition by ATs
The second-order rate constants (kon) for inhibition of FXa and FIIa (enzyme research laboratories, Stago, Asniere sur Seine, France) were measured in PBS containing 0.1% polyethylene glycol 8000 (PEG-8000, Fisher BioReagents, Fisher Scientific, Illkirch, France) and 0.1% bovine serum albumin (Merck millipore, Molsheim, France). The kon constants were estimated in the presence of a saturating concentration of fondaparinux (Arixtra®, GSK, Marly-le-Roi, France), LMWH (Enoxaparin, Lovenox®, Sanofi-Aventis, Paris, France) or UFH (heparin sodium, Panpharma, Luitré, France) under pseudo first-order conditions, as previously described . AT-BD and cAT inhibition kinetics were initiated by addition of enzyme (2 nM) to a pre-warmed mix containing AT (20–200 nM), fondaparinux (5 μM), LMWH (5 U mL−1) or UFH (5 U mL−1), and chromogenic substrate (200 μM, S2238 for FIIa or S2765 for FXa, Instrumentation Laboratory, Le Pré Saint Gervais, France). Substrate hydrolysis was measured at 405 nm, and the kon values were obtained by fitting the data to the equation for slow-binding inhibition and corrected for the competition introduced by the substrate .
Kinetics of FIIa-AT complex formation on SDS-PAGE
AT-BD or unmodified AT (1.25 μM) was incubated with FIIa (0.75 μM) in PBS containing 0.1% PEG-8000 for 0, 5, 10, 15 or 20 min in a final volume of 15 μL. The reaction was quenched by dilution in 4X SDS-loading buffer (125 mM Tris-HCl, 20% glycerol, 4% sodium dodecyl sulfate, 0.2‰ bromophenol blue, pH 6.8) and immediate heating at 95 °C for 2 min. Samples (10 μL per lane) were subjected to electrophoresis on a 10% polyacrylamide gel in denaturing conditions and protein bands visualized by staining with Coomassie brilliant blue R-250 were imaged and analyzed by quantitative densitometry using Scion Image software.
Affinity of ATs for fondaparinux
Dissociation equilibrium constants (KD) were estimated by monitoring AT intrinsic fluorescence enhancement upon fondaparinux binding using an LS50B spectrofluorimeter (PerkinElmer Life Sciences, Courtaboeuf, France) with excitation and emission wavelengths of 280 (slit 3) and 340 (slit 5) nm, respectively, as previously described. The reactions were performed in PBS containing 0.1% PEG-8000, in the presence of 100 nM AT and fondaparinux (0–400 nM).
Evaluation of heparin derivative reversal properties of AT-BD in plasma
The antidote potential of AT-BD for heparin derivatives was evaluated by its ability to reduce heparin-induced anti-FXa activity using an STA-Rotachrom® heparin assay (Stago) in a pool of citrated normal human plasma. Plasma was supplemented with UFH, LMWH or fondaparinux and incubated with AT-BD in a volume ratio of 1:1. After 5 min, anti-FXa activity was measured following the manufacturer's recommendations.
AT-BD antidote activity in mice
The antidote activity of AT-BD was evaluated in Swiss female albino mice (Janvier, Saint Berthevin, France). Animal experiments were approved by the local ethical committee and performed in accordance with the European guidelines for animal experimentation. Each mouse was first anesthetized with sodium pentobarbital administered intraperitoneally (i.p.) and then received fondaparinux (0.5 mg kg−1) or UFH (1200 IU kg−1) by subcutaneous (s.c.) administration. AT-BD, protamine or placebo (PBS) was injected through the retro-orbital plexus 10 min (fondaparinux model) or 60 min (UFH model) after anticoagulant administration. Five minutes later, blood was withdrawn by vena cava puncture using syringes containing trisodium citrate. Plasma was obtained by immediate centrifugation for 15 min at 2300 g, and stored at −40 °C until analyses.
The anticoagulant activity of fondaparinux or UFH in mouse plasma was evaluated by measuring anti-FXa activity using the STA-Rotachrom® heparin assay.
Activated partial thromboplastin time (aPTT) assay
The aPTT assays were measured in a pool of normal plasma supplemented or not with UFH and supplemented with AT-BD or protamine sulfate. APTT assays were performed using Pathromtin SL (Siemens Healthcare, Saint Denis, France) on an ST4 instrument (Stago), according to the manufacturer's instructions.
Preparation and characterization of AT-BD
Purified plasma-derived AT was chemically modified with 2,3-butanedione, an α-diketone known to specifically react with arginine side chains , with the aim of targeting reactive Arg393 and producing an AT form with decreased anticoagulant activity. AT-BD was prepared under experimental conditions permitting selection of an AT conformation with high heparin affinity after solid-phase chemical modification directly on heparin-sepharose beads. Low and medium heparin affinity AT forms were washed out and only high heparin affinity AT was collected.
To analyze AT modification by 2,3-butanedione, cAT and AT-BD were subjected to Asp-N digestion followed by RP-HPLC separation of the peptides. Superimposition of chromatograms obtained for cAT compared with AT-BD revealed very similar profiles between the two AT species (data not shown). Eluted fractions were further analyzed by MALDI-TOF MS. Identification of peptide masses indicated that 13 arginine residues out of the 22 present in the AT sequence, were unmodified by 2,3-butanedione. This approach failed to identify the nine other arginine residues, including 5 arginines located within a single 66-amino-acid peptide at the C-terminus of AT. Both cAT and AT-BD were then subjected to trypsin digestion and directly analyzed by MALDI-TOF MS. In cAT, three peptides mapping the C-terminal region of AT were identified (371–393, 404–413 and 414–425), whereas in AT-BD four peptides mapping the same C-terminal region were identified, suggesting a partial miscleavage site on Arg393 (371–393, 371–399, 404–413 and 414–425). The miscleaved peptide was associated with an additional mass of 68 Da compared with its theoretical mass, which is compatible with diacetylation of Arg393 residue accompanied with the loss of a water molecule . The four other Arginine residues located in the C-terminal part (Arg399, Arg406, Arg413 and Arg425), as well as two additional arginine residues undetectable by the first approach, were found to be unmodified by 2,3-butanedione. Taken together MS analysis confirmed the modification of Arg393, though partial, and indicated that none of the 20 other identified Arg residues were modified by 2,3-butanedione.
To confirm the effect of chemical modification on reactive Arg393, anticoagulant activity of both AT-BD and cAT were estimated by measuring the second-order rate constants (kon) for inhibition of FXa or FIIa in the presence of saturating concentrations of fondaparinux, LMWH or UFH (Fig. 1). As expected, the rate constant for FXa or FIIa inhibition by AT-BD in the presence of heparinic cofactor was significantly decreased by approximately one order of magnitude compared with inhibition by cAT. However, this loss of inhibitory activity unexpectedly appeared moderate as compared with that previously described for P1-modified AT [15, 16], and might result from partial Arg393 modification, as indicated by MS analysis.
In addition, the kinetic of FIIa-AT complex formation was monitored by electrophoresis on polyacrylamide gel under denaturing conditions (SDS-PAGE) followed by quantitative band densitometry analysis (Fig. 2). When incubated with unmodified AT (1.25 μM), inhibition of FIIa (0.75 μM) was complete within 20 min, with disappearance of the band corresponding to the free FIIa correlated with appearance of the band corresponding to the FIIa-AT complex. In contrast, when incubated with AT-BD (1.25 μM), intensity of the free FIIa band decreased slowly, with 73% of its initial intensity remaining after 20 min incubation. This corresponded to 0.2 μM of FIIa, and thus 0.2 μM of AT-BD, covalently bound. Knowing that unmodified AT would have totally reacted within 20 min, one could calculate that the part of unmodified AT in the AT-BD preparation represented less than 16% of total AT.
The measurement of dissociation equilibrium constant (KD) for fondaparinux binding of AT-BD (KD = 3.3 ± 0.5 nM) compared with cAT (KD = 23.5 ± 4.7 nM) confirmed that the chemical modification of AT preserved the integrity of its heparin binding site. Interestingly, AT-BD bound fondaparinux with a KD much lower than cAT, due to the modification of Arg393 by 2,3-butanedione, which relaxes constraints in the reactive center loop to maintain AT-BD in a high heparin affinity conformation. In contrast, cAT was in equilibrium between a native form in which the reactive center loop is partially buried and an active form in which the loop is exposed .
Evaluation of the antidote potential of AT-BD in human plasma
As AT-BD exhibited significantly decreased anticoagulant activity while binding to heparin with higher affinity than cAT, AT-BD was then tested for its ability to neutralize the anti-FXa activity of fondaparinux, LMWH or UFH in a human plasma-assay.
To mimic an overdose, fondaparinux was added to pooled human plasma at a concentration of 3 mg L−1 (approximately 3-fold higher than the curative concentration that ranges between 1.0 and 1.26 mg L−1 during treatment of VTE disease ). Fondaparinux reversal efficiency of AT-BD was then evaluated by adding AT-BD at concentrations ranging between 0 and 1.5 g L−1, corresponding to 0 to 10 times the endogenous plasma AT concentration (assuming AT = 0.15 g L−1 in the human plasma pool). Anti-FXa activity decreased dose-dependently with increasing concentrations of AT-BD (Fig. 3A). At a concentration of 0.15 g L−1, AT-BD was able to reduce anti-FXa activity 2.2-fold in fondaparinux-overdosed plasma, which allowed bringing back the fondaparinux anticoagulant activity within the therapeutic margin. By increasing AT-BD concentrations above 0.6 g L−1, anti-FXa activity in human plasma levelled off at approximately 5% of initial anti-FXa measured in fondaparinux-overdosed plasma in the absence of modified AT. This indicated that excess of reversal up to 1.5 g L−1 was devoid of any intrinsic anticoagulant effect as assessed by anti-FXa assay. Reversal efficiency of AT-BD was also evaluated in plasma containing fondaparinux at curative concentration (1.25 mg L−1). The dose of AT-BD required to reduce anticoagulant activity below the therapeutic margin was then only 0.03 g L−1. Under these conditions, fondaparinux anti-FXa activity decreased dose-dependently with increasing concentration of AT-BD until it became and remained undetectable at high doses of AT-BD (up to 1.5 g L−1, Fig. 4).
High concentrations of UFH (1.8 U mL−1) and LMWHs (3 U mL−1), each corresponding to three times the maximal curative doses, were also used to demonstrate a dose-dependent antidote effect of AT-BD in the human plasma assay (Fig. 3B,C). Addition of AT-BD up to 0.6 g L−1 decreased anti-FXa activity of LMWH- or UFH-overdosed plasma to ~10% or ~15% of its initial activity in the absence of antidote, respectively. At higher concentrations, AT-BD maintained the anti-FXa activity stable in plasma. Thus AT-BD can efficiently reverse all kinds of heparin-like anticoagulants according to their anti-FXa activity in human plasma supplemented with anticoagulants.
Evaluation of antidote potential of AT-BD in vivo
The ability of AT-BD to neutralize heparin-like anticoagulants was evaluated in vivo in mice treated with fondaparinux or UFH. Fondaparinux (0.5 mg kg−1) was administered s.c. and after 10 min, 28 mg kg−1 AT-BD (n = 10) or PBS (n = 13) were injected i.v. through the retro-orbital plexus. The dose of AT-BD was calculated to reach a plasma concentration of 0.375 g L−1, assuming total bioavailability. This dose was chosen because 0.3 g L−1 AT-BD was expected to reduce fondaparinux anticoagulant activity by more than 80% in fondaparinux-overdosed human plasma. The concentration of fondaparinux in mouse plasma was measured by anti-FXa activity and was found to be significantly reduced in the AT-BD-treated group (0.11 ± 0.01 mg L−1) as compared with the control group (0.77 ± 0.16 mg L−1) (Fig. 5A).
To further validate AT-BD as a potential antidote for heparin derivatives, its potency was compared with protamine in vivo in mice treated with UFH. High doses of UFH (1200 IU kg−1) were administered s.c. UFH reversals were injected 60 min after UFH because the time to reach maximal UFH concentration (tmax) was ~50 min and because UFH concentration in plasma decreased slowly after the tmax. In the absence of antidote (control, n = 8), the UFH concentration was 1.36 ± 0.41 IU mL−1. Protamine (105 AHU kg−1) and AT-BD (28 mg kg−1) were administered to reach plasma concentrations of 1.4 UAH mL−1 and 0.375 g L−1 for protamine and AT-BD, respectively. At these doses, AT-BD and protamine resulted in comparable reduction in anti-FXa activity by 75.7% and 76.5%, respectively (Fig. 5B). Finally, these results clearly demonstrated that, in vivo, AT-BD reversed the anticoagulant activities of fondaparinux and UFH and was as efficient as protamine in neutralizing UFH.
Evaluation of AT-BD in an aPTT assay
A paradoxical anticoagulant activity was reported when protamine is used in excess as compared with UFH  and could be detected by plasma coagulation assays such as aPTT. We thus tested the effect of high AT-BD and protamine concentrations in an aPTT assay, in the absence of heparin. As previously described, addition of high doses of protamine in a heparin-free pooled human plasma resulted in a dose-dependent increase in the aPTT ratio (Fig. 6, upper panel), with aPTT ratios of 1.9 ± 0.1 and 6.3 ± 0.5 for 6 and 8 AHU mL−1 protamine. Interestingly, AT-BD, even at high concentrations (up to 1.2 g L−1) did not prolong clotting times in our aPTT assay. Similar results were obtained in plasma containing 1 IU mL−1 UFH. At high UFH concentrations, aPTT ratios increased greatly (10.9 ± 0.5) and normalized by addition of protamine (1.1 ± 0.01 at 4 AHU mL−1, Fig. 6, lower panel). However, the intrinsic anticoagulant properties of protamine induced a large increase in aPTT ratios when added at high concentrations (aPTT ratio = 4.6 ± 0.4 at 8 AHU mL−1). AT-BD (0.15–1.2 g L−1) was also tested in UFH-containing plasma and none of the concentrations tested allowed total normalization of aPTT ratios, which remained slightly higher (1.6 ± 0.02 at 0.6 g L−1 AT-BD) than UFH-free plasma aPTT ratios (1.0 ± 0.01). However, even at high AT-BD concentrations, no rebound effect was observed (Fig. 6, lower panel). Taken together, these results suggested that AT-BD was a potent reversal agent for heparin derivatives that might not be associated with bleeding risks, as observed with protamine.
Hemorrhages are the major complications associated with all anticoagulant treatments , underlining the clinical interest in the development of efficient and specific antidotes. For heparin analogs, despite a lack of specificity and a partial efficacy, protamine is used to reverse UFH and LMWH, and no antidote active against fondaparinux is available for clinical use. So far, in the absence of a specific antidote, recombinant FVIIa (rFVIIa) seems the most efficient strategy for reversing fondaparinux overdosing, although rFVIIa proved limited efficacy in a small prospective study . Moreover, rFVIIa is a prothrombotic drug with an unspecific mechanism toward fondaparinux, and even if data from hemophilic patients suggest that it is well tolerated , the safety of rFVIIa use in patients receiving an anticoagulant treatment for a prothrombotic state is elusive.
We recently described an AT variant (AT-N135Q-Pro394) as the first and only reversal agent for fondaparinux . In the present study, a novel strategy for producing a specific and efficient antidote to AT-dependent anticoagulant drugs was proposed based on chemical modification of purified plasma-derived AT. As for AT-N135Q-Pro394, the principle of this antidote lies in loss of AT anticoagulant activity combined with gain in heparin affinity, allowing the trapping of heparin analogs in an inactive complex by competition with endogenous plasma AT. Alteration of anticoagulant activity was achieved by selective modification of Arg residues by 2,3-butanedione, whereas enhancement of heparin affinity originated from protection of the heparin binding site during chemical modification processes, followed by selection of high heparin affinity conformation during purification. The resulting AT-BD molecule potently reversed the anticoagulant activity of fondaparinux in vitro and in vivo, and surprisingly, the reversal effects of AT-BD in vitro were comparable to those of AT-N135Q-Pro394 . The latter combined two point mutations, resulting in a moderate 3-fold increase in heparin affinity, and an almost abolished anticoagulant activity, as compared with purified pAT. In contrast, AT-BD exhibited a moderate ~10-fold decrease in anticoagulant activity but a major ~7-fold increase in heparin affinity. Exponential decay of anti-FXa activity in fondaparinux-overloaded plasma upon addition of both modified ATs followed very superimposable curves that level off at 3.8% or 3.5% of its maximal value, with a half-maximal effective concentration of 0.114 mg L−1 and 0.117 mg L−1 for AT-BD and AT-N135Q-Pro394, respectively. Furthermore, when administered at the same dose in mice previously treated with fondaparinux both antidotes decreased anti-FXa activity in mouse plasma by 85.7% and 85.7%, respectively. The reversal properties of AT-BD and AT-N135Q-Pro394 were also comparable for UFH and LMWH in vitro.
Our original chemical process allowed us to significantly increase the amounts of modified AT produced, providing sufficient AT-BD for comparison with protamine in vivo but also in vitro at high doses on a plasma global coagulation test. In addition to potently reversing fondaparinux, our results showed that AT-BD was an efficient antidote for UFH. Nevertheless, while protamine normalized aPTT (ratio < 1.2) ratios when used between 1 and 6 AHU mL−1, AT-BD significantly decreased aPTT ratios to ~1.6 but, in vitro, never normalized completely this coagulation test. However, in vivo, in mice treated with UFH (1200 IU kg−1), UFH reversal was identical for protamine and AT-BD. In addition to securing the clinical use of fondaparinux, AT-BD might thus be an attractive alternative to protamine in UFH overdosing for several reasons. Indeed, although catastrophic reactions to protamine remain rare, adverse effects including systemic hypotension and pulmonary hypertension occur in 2.6% of patients during cardiac surgery . These adverse effects are related to the lack of specificity of this polycationic protein for heparin binding , and therefore should not be observed with AT-BD, whose mechanism of action is highly specific. However, we cannot rule out that arginine residues modified by 2,3-butanedione in AT-BD may be immunogenic or toxic, which will have to be evaluated. A second limitation of protamine is related to a paradoxical anticoagulant effect, when used in excess, mediated by inhibitory activities on platelets [25, 26] and on several coagulation serine proteases [19, 27, 28]. This intrinsic anticoagulant effect of protamine was evidenced by various coagulation assays, including aPTT. In accordance with previous reports , we found that addition of protamine to normal plasma resulted in a significant increase in the aPTT ratio, either in the presence or absence of UFH. In comparison, supplementation of normal plasma with AT-BD, even at high doses, did not prolong aPTT coagulation times. Thus, the therapeutic range for AT-BD use may be enlarged as compared with protamine.
The antidote potency of AT-BD could be strengthened by optimization of the chemical modification process that was designed to maximize specific modification of Arg393 and minimize alteration of other arginine residues. Indeed, identification of modified arginines revealed a single modification on Arg393, thought to be partial. It would then be possible, by increasing 2,3-butanedione concentration or by lengthening incubation with AT, to increase modification on Arg393, leading to a greater loss of AT-BD anticoagulant activity. This would permit AT-BD to achieve the same heparin-derivatives reversal effect using lower doses and would permit AT-BD to neutralize completely the anticoagulant activity of heparin derivatives. Other methods allowing specific modification of Arg393, such as enzymatic modification of AT by peptidylarginine deiminase , could also be considered as an alternative strategy to produce inactive AT as an antidote to heparin-like anticoagulants.
UFH is the unique anticoagulant available for intraoperative anticoagulation in cardiovascular surgery, because it is rapidly reversible with protamine. However, besides bleedings, major complications include heparin-induced thrombocytopenia (HIT), which is frequently suspected in patients undergoing CBP . Although only a few patients will develop thromboembolic disorders, evidence of HIT antibodies considerably complicates the management of anticoagulation patients with post-CBP HIT. In this context, fondaparinux was recently proposed to treat patients with past history of HIT  and appeared as a good alternative to UFH in CBP, provided that a rapid and effective antidote is available. A variety of molecules, including hexadimethrine , platelet factor 4 [33, 34], heparinase I  or polycationic peptides [36-38], have been evaluated. All are active towards UFH, some can partially reverse LMWH [36, 37], but none is able to neutralize fondaparinux, allowing the evaluation of fondaparinux in cardiopulmonary bypass procedures. In contrast, AT-BD potently reverses all heparin derivatives, including fondaparinux, paving the way for a possible clinical use of fondaparinux during CBP.
The possible therapeutic use of AT-BD will strongly require further preclinical studies to evaluate its immunogenicity, pharmacokinetics and safety. In this context, it is noteworthy that such properties might also be modulated by the source of AT that is used, either plasma-derived or produced as a recombinant protein in the milk of transgenic goats.
D. Borgel, E. P. Bianchini and V. Picard designed and supervised the research. E. P. Bianchini, J. Fazavana and F. Saller conducted most of the experiments. J. Fazavana and C. Smadja conducted mass spectrometry experiments, whereas C. Smadja and M. Taverna supervised them. E. P. Bianchini and D. Borgel wrote the manuscript.
This work was supported by research grants from FRM (Fondation pour la Recherche Médicale), OSEO and University Paris Sud-XI. J. Fazavana received a grant from GEHT (Groupe de Travail sur l'Hémostase et la Thrombose) and SFH (Société Française d'Hématologie). We also thank C. Boursier of University Paris Sud-XI, IFR IPSIT, for her helpful comments on mass spectrometry analyses and V. Domergue of University Paris Sud-XI, IFR IPSIT, for her help in conducting animal experimentations.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interests.