|Home | About | Journals | Submit | Contact Us | Français|
Heparin allosterically activates antithrombin as an inhibitor of factors Xa and IXa by enhancing the initial Michaelis complex interaction of inhibitor with protease through exosites. Here, we investigate the mechanism of this enhancement by analyzing the effects of alanine mutations of six putative antithrombin exosite residues and three complementary protease exosite residues on antithrombin reactivity with these proteases in unactivated and heparin-activated states. Mutations of antithrombin Tyr253 and His319 exosite residues produced massive 10–200-fold losses in reactivity with factors Xa and IXa in both unactivated and heparin-activated states, indicating that these residues made critical attractive interactions with protease independent of heparin activation. By contrast, mutations of Asn233, Arg235, Glu237, and Glu255 exosite residues showed that these residues made both repulsive and attractive interactions with protease that depended on the activation state and whether the critical Tyr253/His319 residues were mutated. Mutation of factor Xa Arg143, Lys148, and Arg150 residues that interact with the exosite in the x-ray structure of the Michaelis complex confirmed the importance of all residues for heparin-activated antithrombin reactivity and Arg150 for native serpin reactivity. These results demonstrate that the exosite is a key determinant of antithrombin reactivity with factors Xa and IXa in the native as well as the heparin-activated state and support a new model of allosteric activation we recently proposed in which a balance between attractive and repulsive exosite interactions in the native state is shifted to favor the attractive interactions in the activated state through core conformational changes induced by heparin binding.
Antithrombin, a member of the serine protease inhibitor (serpin) superfamily, acts as a critical anticoagulant regulator of hemostasis through its ability to rapidly inhibit the blood coagulation proteases, thrombin, factor Xa, and factor IXa (1). The physiologic importance of this anticoagulant protein is indicated from the increased risk of thrombotic diseases associated with acquired or inherited deficiencies of the protein in humans (2) and the lethality and coagulopathy associated with complete antithrombin deficiency in mice (3) and zebrafish (4). Antithrombin circulates in the blood in a repressed state capable of slow inhibition of its target coagulation proteases and becomes activated at sites of injury by extravascular heparin and heparan sulfate glycosaminoglycans to localize and prevent the dissemination of blood clots (5, 6). Heparin activation of antithrombin involves both an allosteric component wherein a specific pentasaccharide sequence in heparin induces activating conformational changes in the serpin that promotes its interaction with protease and a bridging component in which the extended heparin chain in the antithrombin-heparin complex provides a site for the protease to bind next to the serpin to promote the serpin-protease interaction (7,–9). The allosteric mechanism activates antithrombin reactivity with the target proteases, factors Xa and IXa, but only minimally with thrombin, whereas the bridging mechanism activates antithrombin reactivity with all three proteases (10).
The mechanism of allosteric activation of antithrombin has been the subject of numerous studies. Early models based on x-ray structures of native and heparin-activated antithrombin suggested the importance of releasing the reactive center protease binding loop (RCL)2 from a constrained interaction with sheet A in the native state, first as a means of making the critical P1 Arg recognition determinant accessible to target proteases (11) and later as a means to gain access to exosites outside the RCL to augment the protease interaction (12,–15). Although a critical role for exosites in augmenting the antithrombin RCL interaction with factors Xa/IXa in the allosterically activated state is now well established, a role for RCL expulsion from sheet A in making the exosites available for interaction with protease is not supported by much available data.
In particular, the observation that mutations of a critical Tyr253 exosite residue cause large losses in antithrombin reactivity with factors Xa and IXa in both unactivated and heparin-activated states (12) has suggested that the exosite contributes to antithrombin reactivity in both native and activated states and therefore that release of the RCL from its interaction with sheet A is not required to engage the exosite. This and other data suggested an alternative model in which the antithrombin exosite is available to interact with RCL-bound proteases in the native RCL-constrained state, but the interaction is less favorable due to repulsive interactions counteracting the attractive interactions (16). Allosteric activation in this model enhances antithrombin reactivity by mitigating the repulsive interactions and augmenting the attractive interactions with protease. More recently, it was found that an antithrombin variant with mutations designed to promote the allosteric-activating changes in the heparin binding site was fully activated as an inhibitor of factors Xa and IXa without the need for heparin (17). Significantly, this activation was not accompanied by the signature tryptophan fluorescence changes that report RCL expulsion from sheet A, although such changes could be induced by heparin, consistent with the idea that RCL expulsion is not required for allosteric activation.
To provide a more rigorous test of the new model of allosteric activation and in particular to provide evidence for the proposed different modes of exosite interaction in native and allosterically activated states, we chose to mutate six antithrombin residues that comprise the critical exosite in x-ray structures of heparin-antithrombin-protease Michaelis complexes (13, 14) and assess their importance as determinants of antithrombin reactivity with factors Xa, IXa, and thrombin in unactivated and heparin-activated states of the serpin. Complementary mutations of three active-site loop residues in factor Xa that interact with the antithrombin exosite in the Michaelis complex structures (18, 19) were done to assess the dependence of the exosite interaction on these protease residues. Our results demonstrate the importance of antithrombin and protease exosite residues for reactivity in both native and heparin-activated states of the serpin and thus show that RCL expulsion is not required to engage the exosite. Significantly, they support a new model of allosteric activation in which a balance between attractive and repulsive exosite interactions in the native state designed to repress reactivity at an otherwise attractive P1 Arg residue is shifted to favor the attractive interactions in the activated state through core conformational changes induced by heparin binding.
Recombinant antithrombin variants were constructed on an N135Q background to prevent glycosylation of Asn135 and consequent heparin binding heterogeneity (20) as described in previous studies (21). Gene mutagenesis was carried out by PCR using specially designed oligonucleotides from Integrated DNA Technologies and Pfu Turbo DNA polymerase from Stratagene according to the manufacturer's instructions. All mutations were confirmed by DNA sequencing. Antithrombin was expressed in Hi5 insect cells using the baculovirus expression system from Invitrogen and purified from culture media by a combination of heparin affinity and ion exchange chromatography, as previously described (12, 21). Protein concentration was determined from absorbance at 280 nm using an extinction coefficient of 37,700 m−1 cm−1 (22). Coagulation factors IXa and Xa from human plasma were purchased from Enzyme Research Laboratories, and human plasma thrombin was purchased from U.S. Biochemical Corp. Recombinant Gla-domainless wild-type and mutant human factor Xas were expressed in mammalian cells and activated and purified as described (18). Active protease concentrations were determined from standard substrate assays previously calibrated with active site-titrated enzyme. A synthetic heparin pentasaccharide corresponding to the antithrombin binding sequence in heparin was provided by Sanofi-Aventis (Toulouse, France). A full-length heparin (50-mer) containing the pentasaccharide binding sequence was prepared by size and affinity fractionation of commercial heparin as described (23).
The purity of recombinant antithrombins as well as plasma and recombinant proteases and their ability to form antithrombin-protease covalent complexes was analyzed by SDS-PAGE electrophoresis using the Laemmli discontinuous buffer system (24).
Antithrombin-protease reaction stoichiometries and kinetics were analyzed in ionic strength (I) 0.15, pH 7.4, buffers at 25 °C. Reactions with plasma factor Xa and factor IXa were done in 100 mm Hepes, 93.5 mm NaCl, 5 mm CaCl2, 0.1% PEG 8000, whereas reactions with plasma thrombin and recombinant factor Xa were conducted in 20 mm sodium phosphate, 100 mm NaCl, 0.1 mm EDTA, 0.1% PEG 8000 because the latter were not affected by calcium.
The stoichiometries of antithrombin inhibition of proteases were determined in the absence and presence of pentasaccharide and full-length heparins by end point assays essentially as described in past studies (12). Fixed concentrations of protease and heparin (~100 nm) were incubated with increasing molar ratios of antithrombin to protease (up to 2:1) for a time sufficient to complete the reaction based on measured rate constants. Residual protease activity was then assayed by dilution of an aliquot into 100 μm chromogenic or 50 μm fluorogenic substrate solution and the initial linear absorbance (405 nm) or fluorescence (λex 380 nm, λem 440 nm) increase was measured for several minutes. Plots of residual protease activity as a function of the molar ratio of antithrombin to protease were linear and yielded the stoichiometry of inhibition from the abscissa intercept. The substrates used were S2238 for thrombin (Chromogenix), Spectrozyme FXa for factor Xa (American Diagnostica), and Pefachrome or Pefafluor for factor IXa (Centerchem). Substrates were made up in buffers consisting of 20 mm sodium phosphate, 100 mm NaCl, 0.1 mm EDTA, 0.1% PEG 8000, pH 7.4, for S2238, 100 mm Hepes, 100 mm NaCl, 1 mm EDTA, 0.1% PEG 8000, pH 7.4, for Spectrozyme FXa, and 100 mm Hepes, 100 mm NaCl, 10 mm CaCl2, 33% ethylene glycol, 0.1% PEG 8000, pH 8, for Pefachrome and Pefafluor. 50 μg/ml of Polybrene was included in all substrates to neutralize heparin when present.
The kinetics of antithrombin inhibition of proteases was studied under pseudo first-order conditions with antithrombin concentrations at least 10-fold higher than that of the protease as in previous studies (12). For reactions without heparin or in the presence of heparin pentasaccharide concentrations sufficient to saturate antithrombin (20–100% molar excess) and with measured rate constants up to 105 m−1 s−1, full reaction time courses of the loss of protease activity were measured by assaying residual protease activity in aliquots of reaction mixtures as a function of time as described in measurements of reaction stoichiometries. Progress curves were fit by a single exponential decay function with zero end point for thrombin and factor Xa reactions and nonzero end point for factor IXa reactions to account for low levels of degraded protease less susceptible to inhibition (<10%) to obtain the observed pseudo first-order rate constant (kobs). Apparent second-order association rate constants were calculated by dividing kobs by the concentration of antithrombin. For reactions in the presence of pentasaccharide or full-length heparin that were faster than 105 m−1 s−1, reactions were conducted for a fixed time in the presence of increasing concentrations of heparin at catalytic levels and residual protease activity was then assayed. The loss in protease activity was in this case fit by an exponential function with the heparin concentration instead of time as the independent variable (10). The apparent second-order rate constant was obtained by dividing the exponential rate constant by the fixed reaction time. Because antithrombin concentrations greatly exceeded dissociation constants for antithrombin-heparin interactions at I 0.15 for wild-type and variant antithrombins (>20-fold), no corrections for unbound heparin were required. Corrected second-order association rate constants reflecting reaction along the inhibitory pathway (ka) were obtained by multiplying apparent second-order rate constants by inhibition reaction stoichiometries.
Equilibrium dissociation constants for antithrombin-heparin interactions were measured by titrations of wild-type and mutant antithrombins (100 nm) with the heparin pentasaccharide monitored from the ~40% enhancement in tryptophan fluorescence that accompanies heparin binding as in previous studies (9). Titrations were performed in I 0.35 buffer containing 20 mm sodium phosphate, 0.3 m NaCl, 0.1 mm EDTA, 0.1% PEG 8000, pH 7.4, to weaken heparin affinity and allow accurate measurements of KD. KD and the maximal fluorescence change were obtained by computer fitting data with the quadratic equilibrium binding equation (9). Stoichiometric titrations of antithrombin variants with heparin pentasaccharide were performed in I 0.15 sodium phosphate buffer to ensure that recombinant proteins were fully functional in binding the saccharide.
The importance of six putative antithrombin exosite residues that interact with factors Xa and IXa in x-ray structures of heparin-activated antithrombin-factor Xa/IXa Michaelis complexes (13, 14) (Fig. 1) for allosteric activation of the serpin was assessed by mutagenesis. Antithrombin variants with alanine mutations of Asn233, Arg235, Glu237, Tyr253, Glu255, and His319 residues of strands 2, 3, and 4 of sheet C were made alone and in combination. Although His319 was previously not implicated as part of the exosite, intramolecular interactions of this residue with Glu237 and Tyr253 exosite residues suggested a potential role in exosite function. The variant antithrombins were evaluated for their effects on antithrombin reactivity with factors Xa and IXa in the absence and presence of specific pentasaccharide and full-length ~50-saccharide heparins containing the pentasaccharide sequence. The heparin pentasaccharide is sufficient to allosterically activate the inhibitor, whereas the full-length heparin augments the allosteric activating effect by providing an additional bridging interaction site for the protease in the ternary Michaelis complex (9). Because thrombin does not engage the exosite and its reactivity with antithrombin is solely enhanced by bridging heparin activation (25, 26), the effects of the mutations on antithrombin reactivity with thrombin served as an important control for any nonspecific effects of the mutations on serpin reactivity.
All antithrombin mutants were expressed and purified similar to wild-type and yielded amounts of protein comparable with wild-type. All mutants bound the heparin pentasaccharide with binding stoichiometries close to 1:1 and inhibited thrombin, factor Xa, and factor IXa in the absence and presence of heparin with stoichiometries of 1–2 mol of inhibitor/mol of protease that were comparable with wild-type values. The effect of the mutations on antithrombin reactivity with the three proteases was assessed by measuring the apparent second-order rate constants for the reactions under pseudo first-order conditions as in past studies (12). These were corrected for the fraction of inhibitor reacting through a competing substrate pathway by multiplying by the inhibition stoichiometries to yield the second-order association rate constant (ka) for reaction along the inhibitory pathway.
Mutations of Tyr253 and His319 had profound effects on antithrombin reactivity with factors Xa and IXa in native as well as heparin-activated states (Figs. 2 and and33 and Tables 1 and and2).2). Y253A antithrombin inhibited factor Xa with a 7-fold lower ka than wild-type in the absence of heparin and more substantial 40- and 10-fold lower ka in the presence of pentasaccharide and bridging heparins, respectively. The mutant antithrombin inhibited factor IXa with an even greater 69-fold reduction in ka from wild-type in the unactivated state but lesser 32- and 14-fold reductions in ka in pentasaccharide and bridging heparin-activated states, although the latter reductions were comparable with those observed in corresponding reactions with factor Xa. Similarly, H319A antithrombin inhibited factor Xa with a 9-fold lower ka than wild-type in the absence of heparin and greater 67- and 27-fold reduced values in the presence of pentasaccharide and bridging heparins. As observed with the Y253A mutant, H319A antithrombin inhibited factor IXa with the most dramatic 230-fold reduction in ka from wild-type in the absence of heparin compared with the lesser 50- and 20-fold reductions in ka in the presence of pentasaccharide and bridging heparins, although the latter reductions were again comparable with those observed with factor Xa. Interestingly, combining the Tyr253 and His319 mutations caused losses in antithrombin reactivity with the two proteases that were similar to or only modestly greater than those caused by mutation of either residue alone in unactivated or heparin-activated states. The double Y253A/H319A antithrombin mutant thus showed 6-, 88-, and 30-fold reductions in ka for reactions with factor Xa in the absence and presence of pentasaccharide and full-length heparins, respectively, and 200-, 89-, and 72-fold reductions in ka for reactions with factor IXa in the absence and presence of the corresponding heparin activators. Mutating either residue of the Tyr253/His319 pair after the other residue has been mutated thus produces little or no further loss in antithrombin reactivity with either protease, indicating that the contributions of each of these residues to antithrombin reactivity are strongly coupled (Figs. 4 and and55).
Contrasting these results, Y253A and H319A single and double antithrombin mutants inhibited thrombin at rates modestly altered from those of wild-type, ka being reduced 1.7-fold in the absence of heparin and increased 1.3–3.5-fold in the presence of the two heparin activators (Fig. 6 and Table 3). Tyr253 and His319 are thus critically important and interdependent contributors to antithrombin reactivity with factors Xa and IXa in both unactivated and heparin-activated inhibitor states although these residues make minor contributions to reactivity with thrombin.
Mutation of Glu255 produced significant losses in antithrombin reactivity with factors Xa and IXa, again both in native and heparin-activated states, although these losses were less than those resulting from Tyr253 and His319 mutations (Figs. 2 and and33 and Tables 1 and and2).2). E255A antithrombin inhibited factor Xa with a 1.3-fold lower ka than wild-type in the absence of heparin and 4.8- and 2.2-fold lower ka in the presence of pentasaccharide and bridging heparins, whereas the mutant antithrombin inhibited factor IXa with an 8.6-fold lower ka in the absence of heparin and 5.6- and 2.4-fold lower ka in the presence of pentasaccharide and full-length heparin activators. Combining the E255A mutation with Y253A/H319A mutations revealed that the contribution of Glu255 to antithrombin reactivity depended on residues Tyr253 and His319 and the activation state of the inhibitor (Figs. 4 and and5).5). In the absence of heparin, adding the E255A mutation to Y253A/H319A mutant antithrombins enhanced rather than depressed reactivity with factor Xa by 2–3-fold and only marginally affected reactivity with factor IXa compared with that of the Y253A/H319A mutants. By contrast, in the presence of pentasaccharide or bridging heparins, the addition of the E255A mutation to Y253A/H319A antithrombin variants produced additional decreases in reactivity with factor Xa of up to 4-fold and with factor IXa of 2–8-fold over the reactivity losses of the corresponding Y253A/H319A mutants. These reactivity decreases were in most cases less than those observed when Glu255 was mutated alone in reactions with factor Xa but comparable with or somewhat greater than those of the Glu255 single mutant in reactions with factor IXa. Glu255 thus positively contributed to antithrombin reactivity in the unactivated inhibitor in the context of wild-type Tyr253 and His319 residues, but this contribution became negative or marginally positive when Tyr253 and His319 were mutated. In the heparin-activated inhibitor, Glu255 made significant positive contributions to reactivity that were reduced by Tyr253 and His319 mutations in factor Xa reactions but were less affected by Tyr253/His319 mutations in factor IXa reactions. The effects of Tyr253/His319 mutations on Glu255 contributions to reactivity were reciprocal, the positive contributions of Tyr253/His319 residues being reduced in the context of the Glu255 mutation for factor Xa reactions and less dependent on the Glu255 mutation for factor IXa reactions (Figs. 4 and and55).
Contrasting these results, Glu255 made a negative contribution to antithrombin reactivity with thrombin both in the absence and presence of heparin (Fig. 6 and Table 3). The E255A antithrombin mutation thus increased the ka for inhibition of thrombin 4-fold in unactivated and ~2-fold in heparin-activated states. Adding the E255A mutation to Y253A/H319A mutant antithrombins somewhat augmented or did not affect the increase in ka in the absence or presence of pentasaccharide heparin and slightly diminished the increase in the presence of the bridging heparin, consistent with the latter mutations modulating the negative contribution of Glu255.
Single mutations of Asn233, Arg235, and Glu237 residues of the putative exosite resulted in modest changes in antithrombin reactivity with factors Xa and IXa in the absence or presence of the two heparin activators (1–2.5-fold), except for the Asn233 mutation, which caused more significant 3.3–5.3-fold losses in antithrombin reactivity with factor IXa with or without heparin activation (Figs. 2 and and33 and Tables 1 and and2).2). The reactivity of these antithrombin variants with thrombin was also only modestly altered from wild-type (<3-fold) (Fig. 6 and Table 3). To assess whether these residues made any contribution to the factor Xa/IXa-specific exosite in antithrombin, we investigated the effect of mutating Asn233, Arg235, and Glu237 residues together as a block. Surprisingly, the combined N233A/R235A/E237A (NRE) mutations caused substantial enhancements of antithrombin reactivity with factors Xa and IXa of 3.6- and 8.3-fold, respectively, in the absence of heparin, whereas the mutations caused losses in reactivity of 1.5–5-fold in the presence of pentasaccharide or bridging heparin activators. That this pattern of reactivity changes was unique to factors Xa and IXa was indicated from the insignificant effect of the NRE block mutations on antithrombin reactivity with thrombin in the absence or presence of pentasaccharide heparin and modest 1.8-fold gain in reactivity in the presence of the bridging heparin (Fig. 6 and Table 3). Asn233, Arg235, and Glu237 thus together make a large negative contribution to antithrombin reactivity with factors Xa and IXa in the unactivated state and this negative contribution becomes positive upon heparin activation of the serpin. By contrast, the three residues make no or small contributions to antithrombin reactivity with thrombin in unactivated and heparin-activated states.
Coupling between the Asn233/Arg235/Glu237 block mutations and Tyr253, Glu255, and His319 exosite residues was evaluated by adding Y253A, E255A, and H319A mutations to the NRE block antithrombin mutant as single, double, or triple mutations and measuring the effects on antithrombin reactivity with the three target proteases. Notably, in the absence of heparin, where the NRE block mutations resulted in significantly enhanced reactivities with factors Xa and IXa, Tyr253, Glu255, and His319 (YEH) mutations produced equal or greater reactivity losses in NRE mutant than in wild-type backgrounds (Figs. 4 and and5).5). These augmented losses in reactivity were modest for factor Xa reactions (maximally 1.6-fold greater) but large for factor IXa reactions (5–36-fold greater) and reflected NRE suppression of the positive contributions of YEH residues. In the presence of heparin where NRE block mutations reduced antithrombin reactivity with factors Xa and IXa, YEH mutations produced smaller losses in reactivity in the NRE mutant than in wild-type backgrounds, reflecting NRE enhancement of the positive contributions of YEH residues. The contribution of YEH residues to antithrombin reactivity with factors Xa and IXa was thus significantly attenuated by the NRE triad in the unactivated inhibitor, whereas the YEH contribution was enhanced by the NRE triad with pentasasccharide and bridging heparin activation of the serpin.
The overall effect of mutating all six putative antithrombin exosite residues was to produce a minor 1.3-fold loss in antithrombin reactivity with factor Xa in the absence of heparin, but substantial 48- and 15-fold losses in reactivity in the presence of pentasaccharide and bridging heparins, respectively (Table 1). Notably, the effect of the six mutations on antithrombin reactivity with factor IXa was more pronounced, i.e. a 55-fold decrease in the absence of heparin and 680- and 120-fold decreases in the presence of pentasaccharide and bridging heparin activators, respectively (Table 2). The greater antithrombin reactivity losses in the heparin-activated than the unactivated state for both factor Xa and IXa reactions was accounted for by the negative contributions of NRE residues in the unactivated state being alleviated in the heparin-activated state and the positively contributing YEH residues in both states being enhanced by heparin activation. By contrast, the six exosite residue mutations resulted in relatively small 2–3-fold enhancements in reactivity with thrombin in the absence or presence of heparin that reflected a dominant-negative contribution of YEH residues (Table 3).
To ascertain the contribution of the complementary protease exosite residue, Arg150 of factors Xa and IXa, to allosteric activation of antithrombin (Fig. 1), we mutated this residue in factor Xa and determined the effect on reactivity with wild-type and exosite mutant antithrombins relative to wild-type factor Xa (Fig. 7 and Table 4). These studies were done with recombinant factor Xa that lacks the Gla domain. The reactivity of the recombinant Gla-domainless factor Xa with antithrombin resembles that of plasma factor Xa measured in the presence of calcium (Tables 1 and and5).5). The R150A mutation reduced the reactivity of factor Xa with wild-type antithrombin by 2.2-fold in the absence of heparin and 20- and 4.3-fold in the presence of pentasaccharide and bridging heparin activators, respectively, confirming an important positive contribution of Arg150 to reactivity that was greatest in the allosterically activated inhibitor (Fig. 7 and Table 4). Significantly, this positive reactivity contribution was abolished or transformed to a negative contribution in reactions of factor Xa with antithrombin possessing mutations of the critical Tyr253 or His319 exosite residues, as judged by the comparable or greater reactivities of the antithrombin exosite mutants with R150A factor Xa relative to wild-type factor Xa in the absence or presence of heparin. Moreover, the positive contribution of Arg150 of factor Xa was reduced by Glu255 or NRE block mutations in antithrombin, based on the observed lessened effect of the Arg150 mutation on factor Xa reactivity with these mutant antithrombins relative to the wild-type serpin. Notably, the NRE mutations retained a strong negative effect on antithrombin reactivity with R150A factor Xa in the unactivated state as with wild-type factor Xa, but in contrast to the wild-type factor Xa reaction, this negative contribution was not alleviated by heparin activation. The overall effect of mutating all six exosite residues was to enhance antithrombin reactivity with R150A factor Xa by 4-fold in the absence of heparin and marginally decrease reactivity by 1.1- and 1.6-fold in the presence of pentasacharide and bridging heparins, respectively (Table 4). These effects contrasted sharply with the neutral effect of the exosite residue mutations on reactivity with factor Xa in the absence of heparin and large 48- and 15-fold losses in reactivity in the presence of pentasaccharide and bridging heparins, respectively (Table 1). The exosite residue contributions to antithrombin reactivity with wild-type factor Xa are thus lost as a result of the R150A mutation in factor Xa.
The importance of two other factor Xa autolysis loop residues, Arg143 and Lys148, that interact with the antithrombin exosite in the x-ray structures of antithrombin-factor Xa/IXa Michaelis complexes (Fig. 1) was similarly evaluated by mutating these residues and determining the effect on factor Xa reactivity with wild-type and a subset of the exosite mutant antithrombins. R143A factor Xa was inhibited by wild-type antithrombin ~3-fold faster than wild-type factor Xa in the absence of heparin, whereas the mutant factor Xa was inhibited 3–5-fold slower than wild-type factor Xa in the presence of the two heparin activators. Similarly, K148A factor Xa was inhibited by wild-type antithrombin 2-fold faster than wild-type factor Xa in the absence of heparin, but 2–10-fold slower than wild-type factor Xa in the presence of the heparin activators (Fig. 8 and Table 5). Arg143 and Lys148 thus both make negative contributions to factor Xa reactivity with unactivated antithrombin but positive contributions with the heparin-activated serpin. Similar to wild-type antithrombin, exosite mutant antithrombins showed 2–4-fold enhanced reactivities with R143A and K148A mutant factor Xa relative to wild-type factor Xa in the absence of heparin, indicating that the negative contributions of the protease basic residues were independent of antithrombin exosite residues. In the presence of pentasaccharide or bridging heparins, by contrast, antithrombins with mutations in the critical Tyr253, Glu255, and His319 residues exhibited reactivities with R143A and K148A factor Xa mutants more comparable with those with wild-type factor Xa, implying that the positive contributions of Arg143 and Lys148 in the heparin-activated state depended on an interaction with the antithrombin exosite.
The exosite mutations not only affected antithrombin reactivity with proteases, but also had significant effects on heparin affinity. This was assessed by equilibrium binding titrations of mutant antithrombin-heparin pentasaccharide interactions monitored from the ~40% tryptophan fluorescence enhancement that accompanies the binding interaction (Table 6). Mutations of Tyr253 or His319 alone significantly weakened heparin affinity 3- and 8-fold, respectively, with the double mutant producing an intermediate 6-fold affinity loss. By contrast, mutation of Glu255 somewhat enhanced heparin affinity and combining this mutation with Tyr253 and His319 mutations attenuated the losses in heparin affinity produced by the latter residues. The block of NRE mutations enhanced heparin affinity to an even greater extent of 5-fold, largely accountable for by smaller additive affinity enhancements from each individual residue mutation. Adding Tyr253 and His319 mutations countered this affinity enhancement to approach wild-type affinities. Adding the Glu255 mutation together with Tyr253 and His319 to the NRE mutant slightly reduced the negative effect of the latter residues on affinity. Notably, the observed fluorescence enhancements reporting allosteric activation showed no clear pattern of change, yielding values ranging from 37 to 48% that only marginally differed from wild-type.
Our studies have shown that six putative antithrombin exosite residues that interact with protease in heparin-activated antithrombin-factor Xa/IXa Michaelis complex structures are significant contributors to antithrombin reactivity with factors Xa and IXa not only in the heparin-activated state but more importantly also in the unactivated state. Tyr253 and His319 in particular were found to be the most crucial determinants of the exosite, mutations to Ala causing ~10–200-fold reactivity losses in the unactivated state and 10–90-fold losses in heparin-activated states in reactions with factors Xa and IXa. Notably, the greatest reactivity losses were observed in unactivated antithrombin reactions with factor IXa. Such findings argue most strongly that Tyr253 and His319 are key determinants of antithrombin reactivity in both native and heparin-activated states. The alternative possibility that these residues only contribute to reactivity in the heparin-activated state would otherwise imply that the observed major effects of mutating these residues on unactivated antithrombin reactivity arise from a minor fraction of activated antithrombin in conformational equilibrium with native antithrombin. However, estimates of this fraction based on the reactivities and heparin affinities of antithrombin variants locked in native or heparin-activated states (27, 28) indicate an amount (~0.1%) that is insufficient to account for the major reactivity losses observed.3 Moreover, the observed 3–8-fold decreases in heparin affinity caused by the Tyr253/His319 mutations (Table 6) suggest that the activated fraction in these mutants is considerably reduced from wild-type.4 The observed 1 to 2 order of magnitude losses in wild-type antithrombin reactivity produced by Tyr253 and His319 mutations in the absence of heparin thus imply that the mutations are almost exclusively affecting an intrinsic native state reactivity and not just the reactivity of a minor equilibrium fraction of activated serpin.
The finding that Tyr253 and His319 are key determinants of the exosite was unexpected based on x-ray structures showing that Asn233, Arg235, Glu237, and Tyr253 residues together comprise an exosite binding pocket that interacts with Arg150 of factors Xa and IXa in the allosterically activated state (13, 14) (Fig. 1). His319 was not previously thought to be part of the exosite because it makes no direct interaction with Arg150. However, a close examination of the Michaelis complex structures shows that the His319 aromatic ring closes a gap between Tyr253 and Glu237 residues of the exosite binding pocket by making an orthogonal interaction with the aromatic ring of Tyr253 and a potential ionic interaction with Glu237 when His319 is in the charged state (Fig. 1). The effects of mutating Tyr253 and His319 alone or together revealed that these residues function together as a cooperative unit, implying that a His319-Tyr253 interaction may function to position Tyr253 in the exosite binding pocket to interact with the protease Arg150 residue in both native and activated conformational states. The importance of Tyr253 and His319 for exosite function is supported by the ability to create a factor Xa/IXa-specific exosite in a P1 Arg variant of the related serpin, α1-protease inhibitor, in which His319 is conserved, by substituting Tyr in the homologous strand 3C 253 position (29).
Although Tyr253/His319 function together to promote the exosite binding pocket interaction in both native and activated states, the interaction must differ in the two conformational states based on our finding that the three other residues of the binding pocket, Asn233, Arg235, and Glu237, together make substantial negative contributions to antithrombin reactivity in the native state, but positive contributions in the heparin-activated state in reactions with both factors Xa and IXa. Moreover, coupling between Asn233/Arg235/Glu237 (NRE) and Tyr253/His319 mutations revealed that NRE residues suppress the positive contributions of the Tyr253/His319 pair in the unactivated state but augment Tyr253/His319 positive contributions in heparin-activated states (Figs. 4 and and5).5). NRE residues in the exosite binding pocket thus appear to modulate the Tyr253 interaction with protease in a manner dependent on heparin activation.
That the contribution of antithrombin exosite binding pocket residues to reactivity results from a specific interaction with Arg150 of the protease is supported by previous studies (18, 19, 29) together with our present finding that the exosite contribution is abolished in antithrombin reactions with an R150A mutant factor Xa in both unactivated and heparin-activated states (Table 4). Notably, the Arg150 mutation produced large reductions in factor Xa reactivity with wild-type and exosite mutant antithrombins only when the critical Tyr253 and His319 residues were present and otherwise enhanced reactivity (Fig. 7). The observation that NRE residues made a negative contribution to reactivity with Arg150 mutant factor Xa in the unactivated state that was not alleviated by heparin allosteric activation further suggests an important role for Arg150 in mitigating this negative contribution in the heparin-activated state.
The antithrombin Glu255 residue outside the exosite binding pocket made more modest positive contributions to antithrombin reactivity that were dependent on the critical Tyr253/His319 residues of the exosite binding pocket. This contribution required factor Xa Arg143 and Lys148 residues only in the heparin-activated state, consistent with Michaelis complex structures showing that Arg143 and Lys148 of factors Xa and IXa interact with Glu255 of antithrombin in the heparin-activated state and that this interaction occurs in concert with the Arg150 interaction with the exosite binding pocket (13, 14) (Fig. 1). The colocalization of Arg143, Lys148, and Arg150 on the autolysis loop of factors Xa and IXa explains why the interactions of these protease residues with the exosite in heparin-activated antithrombin are strongly linked. In unactivated antithrombin, Arg143 and Lys148 were found to make a modest negative contribution to factor Xa reactivity, implying that Glu255 does not interact with these residues in the native state and instead interacts with other protease residues. Together, these findings support different modes of interaction of the antithrombin exosite with protease in unactivated and heparin-activated states.
Interestingly, the exosite residues are strongly conserved or conservatively replaced in all vertebrate antithrombins except for Arg235 and Glu237, which are replaced by uncharged residues in fish (30). Surprisingly, Glu237 is the only exosite residue for which a natural antithrombin mutation (to Lys) that is associated with thrombosis has been reported (31). However, the thrombosis association appears to result from a reduced antigenic level of the circulating serpin rather than a reduced anticoagulant activity, in keeping with the slightly enhanced anticoagulant activity of Glu237 variants found in this and other studies (12, 32, 33). Of greater relevance is the report of a natural antithrombin variant associated with recurrent thrombosis with a mutation in Met251, a buried residue contiguous to the exosite (34). This residue is highly conserved in vertebrate antithrombins and its mutation to Ile reduces anticoagulant activity to a much greater extent than it affects the antigenic level of circulating antithrombin. Examination of antithrombin-protease Michaelis complex structures reveals that the Met251 side chain forms the bottom of the exosite binding pocket for the protease Arg150 residue, suggesting that the Met251 mutation may produce a defect in anticoagulant function by perturbing the structure of the exosite binding pocket through altered core packing interactions.
The positive and negative contributions of the six exosite residues to antithrombin reactivity in the unactivated state were found to be equally balanced in reactions with factor Xa (1.3-fold reactivity increase) but weighted toward the positive in reactions with factor IXa (55-fold reactivity increase). In allosterically activated and bridging heparin-activated states, the exosite made major positive contributions to antithrombin reactivity of 48- and 15-fold, respectively, in reactions with factor Xa and 680- and 120-fold in reactions with factor IXa (Tables 1 and and2).2). The reduced exosite contribution to antithrombin reactivity upon bridging heparin activation most likely reflects constraints on the ability of the protease to engage the antithrombin exosite when it must at the same time engage a bridging site on heparin in the heparin-antithrombin-protease Michaelis complex. Importantly, mutations of the six antithrombin exosite residues produced modest 2–3-fold enhancements in antithrombin reactivity with thrombin independent of heparin activation that reflected a dominant-negative contribution of Glu255, consistent with the fact that thrombin lacks the Arg150 residue of the autolysis loop and is unable to make a specific interaction with the antithrombin exosite (25, 26).
The differential contribution of exosite residues to antithrombin reactivity in unactivated and heparin-activated states was responsible for large rate enhancements of 360-fold with factor Xa5 and 140-fold with factor IXa upon allosteric activation by the heparin pentasaccharide. These rate enhancements were abrogated to 9- and 11-fold, respectively, when all six exosite residues were mutated (Tables 1 and and2).2). Mutation of the Tyr253/Glu255/His319 triad (YEH) alone accounted for much of this abrogation with factor Xa (to 11-fold) but only a small part with factor IXa (to 42-fold), suggesting a greater differential contribution of YEH to antithrombin reactivity in unactivated and activated states with factor Xa than factor IXa. Significantly, mutation of the NRE triad alone caused a marked reduction of the allosteric activation rate enhancement to 45-fold with factor Xa and more substantial reduction to 7-fold with factor IXa, a result of NRE residues mitigating YEH reactivity in the unactivated state and enhancing this reactivity in the activated state. This underscores the important role for NRE exosite residues in repressing antithrombin reactivity in the absence of heparin but enhancing reactivity upon heparin activation especially in the reaction with factor IXa. Although bridging heparin activation of antithrombin reduced the exosite contribution to antithrombin reactivity, this was compensated for by a large gain in reactivity due to heparin bridging interactions with protease promoting the Michaelis complex interaction. Notably, the heparin bridging rate enhancement was observed to increase for almost all mutant antithrombin reactions with factors Xa and IXa relative to the wild-type reaction (Tables 1 and and2).2). This would be consistent with the impaired exosite interaction in the mutant antithrombins allowing the protease to more effectively engage the heparin bridging site and thereby increase the bridging rate enhancement.
The different modes of protease interaction with the antithrombin exosite in native and allosterically activated states of the serpin can be understood based on available x-ray structures. Of particular note in this regard is the x-ray structure of native antithrombin, which shows an intramolecular interaction of the RCL P1 Arg with the exosite (33). This interaction stabilizes the native state based on previous observations that mutations or modification of the P1 Arg shift the native-activated state conformational equilibrium toward the activated state (35, 36). Our present findings support a P1 Arg native state stabilizing interaction with the exosite based on our finding that the exosite mutations perturb heparin affinity in a manner similar to that of the P1 Arg mutations. Interestingly, the mode of the P1 Arg interaction with the exosite is entirely different from that of the protease Arg150 residue interaction with the exosite in heparin-activated antithrombin-factor Xa/IXa Michaelis complex structures (Fig. 9). The P1 Arg-exosite interaction is promoted by Asn233/Arg235/Glu237 residues and antagonized by Tyr253/His319 residues based on the observed effects of mutating these residues on heparin affinity in this study and previous studies in which Glu237 was mutated (12, 33). The x-ray structure shows that the P1 Arg side chain interacts with Glu237. However, the Asn233/Arg235/Glu237 residues appear to function together to promote this interaction because their combined mutation is required to produce an increase in heparin affinity comparable with that caused by a P1 Arg to Ala mutation, which is expected to abolish the P1 Arg interaction (Table 6). The antagonism by Tyr253/His319 appears to result from the native structure constraining the entry of the P1 Arg into the exosite pocket so that the nonpolar stem of the Arg side chain is unable to interact with the Tyr253/His319 aromatic side chains.
Contrasting this constrained P1 Arg interaction with the exosite, the factor Xa and factor IXa Arg150 side chain interactions with the exosite in heparin-activated antithrombin Michaelis complex structures is presumably the most optimal in that interactions with all binding pocket residues are favorable in this state. Indeed, the structure shows that the nonpolar stem of Arg150 aligns with the Tyr253/His319 aromatic side chains and the charged end of Arg150 makes an ionic interaction with Glu237 and multiple backbone interactions (13, 14). Glu255 interactions with protease Arg143 and Lys148 residues presumably promote this favorable mode of interaction based on the observed linkage between these interactions. These favorable interactions are made possible by the major subdomain movements that are involved in allosteric activation (28, 32). This suggests a model for the less favorable mode of the factor Xa/IXa Arg150 interaction with the exosite in native antithrombin in which the protease Arg150 is constrained by the native subdomain structure to enter the exosite binding pocket in a manner that favors interaction of the nonpolar stem of the Arg side chain with Tyr253/His319 residues but does not allow the charged end of the side chain to neutralize the repulsive interactions of Asn233/Arg235/Glu237 residues. The altered mode of the Arg150-exosite binding pocket interaction in this state precludes the neighboring Arg143 and Lys148 protease residues from interacting with Glu255 and forces Glu255 to make alternative less favorable interactions with protease.
Together our findings support a new model for heparin allosteric activation of antithrombin that we recently proposed (16). In this model, factors Xa and IXa engage the antithrombin exosite in the Michaelis complex in the native unactivated state, but this interaction becomes more favorable in the allosterically activated state. The model is supported by our findings that in unactivated antithrombin, Tyr253, Glu255, and His319 exosite residues make attractive interactions with factors Xa and IXa in the Michaelis complex but these are attenuated by repulsive interactions of Asn233, Arg235, and Glu237 with the proteases that serve to down-regulate antithrombin reactivity. Heparin allosteric activation mitigates the repulsive interactions and makes the attractive interactions more favorable so that antithrombin reactivity with factors Xa and IXa is up-regulated.
This new model significantly differs from an earlier model in which allosteric activation was proposed to make the antithrombin exosite available for interaction with RCL-bound factors Xa and IXa (12,–14, 28). This was thought to occur through the hallmark structural change that accompanies allosteric activation, the disruption of the native state stabilizing RCL-sheet A interaction. The RCL-sheet A interaction was believed to constrain the RCL in the native state and prevent bound proteases from engaging the exosite. A key observation supporting our new model, that factors Xa and IXa are able to engage the antithrombin exosite in the unactivated RCL-constrained state, indicates that the release of the RCL from its interaction with sheet A is not necessary for this engagement and therefore that the earlier model cannot be correct.
In keeping with this, recent findings suggest that the changes in the exosite accompanying allosteric activation do not require a disruption of the RCL-sheet A interaction. Mutations in the heparin binding site were thus found to induce antithrombin to undergo the full allosteric activating changes in reactivity with factors Xa and IXa in the absence of heparin without the signature fluorescence changes that report RCL expulsion from sheet A although with large CD and NMR perturbations (17). This implies that the activating conformational changes in the hydrophobic core of antithrombin are sufficient to mitigate the repulsive interactions and enhance the attractive interactions in the exosite without RCL expulsion from sheet A. Such changes in the hydrophobic core are evident from x-ray structures of so-called intermediate antithrombin-heparin complexes that have undergone all the allosteric activating changes except for RCL expulsion (28, 37). Although the exosite itself does not appear to undergo a structural change upon allosteric activation (28), movements of the subdomain in which the exosite resides could relax the structural constraints between the flexible RCL and serpin body in the native conformation and allow Arg150 of factors Xa and IXa to adopt a more productive mode of interaction with the exosite without the need for disrupting the RCL-sheet A interaction.
It is worth noting that heparin binding site mutations that were found to activate antithrombin without disrupting the RCL-sheet A interaction did not affect the ability of the mutant serpin to bind heparin and induce the accompanying fluorescence changes that report RCL expulsion from sheet A (17). Heparin binding to this mutant was accompanied by a small 3–4-fold additional increase in antithrombin reactivity with factors Xa and IXa. Such a reactivity enhancement resembles the enhancements observed upon allosteric activation of wild-type and exosite mutant antithrombins in reactions of the serpin with thrombin and likely reflects a lessening of unfavorable interactions between the RCL-bound protease and the serpin core when the RCL extends away from the core. RCL expulsion and extension away from the serpin surface may thus contribute, albeit in a minor way, to enhancing the exosite interaction in the allosterically activated state by reducing repulsive interactions between RCL-bound factors Xa and IXa and residues of the serpin body outside the exosite. This would explain the residual allosteric activation rate enhancement observed in reactions of the mutant antithrombin lacking all exosite residues with factors Xa and IXa. Such a role for RCL expulsion was postulated in the new model of allosteric activation we recently proposed (16).
Together, our findings provide important new insights into the role of an exosite in the mechanism of heparin allosteric activation of antithrombin. Our studies have unequivocally shown that the exosite is a crucial determinant of antithrombin reactivity with factors Xa and IXa in both unactivated and allosterically activated states and thus that allosteric activation does not involve making the exosite available to protease as previously suggested. They have further demonstrated the key importance of Tyr253 and His319 exosite residues in promoting the interaction with the protease Arg150 residue in both native and activated states and how Glu255 augments this interaction. Most importantly, they have shown that the exosite down-regulates reactivity in the native state through repulsive interactions of Asn233, Arg235, and Glu237 exosite residues and up-regulates reactivity upon allosteric activation by mitigating this repulsion, in accordance with a model we hypothesized previously. Finally, they suggest a structural basis for the different modes of exosite interaction with protease in native and activated states. Collectively, these findings provide a new detailed molecular understanding of the antithrombin allosteric activation mechanism.
We thank Salvenkat Vagvala for excellent technical assistance with some of the studies.
*This work was supported, in whole or in part, by National Institutes of Health Grants R37 HL39888 (to S. T. O.) and R01 HL62565 (to A. R. R.).
3R. Roth and S. T. Olson, unpublished observations. The 0.1% estimate is in keeping with the <0.3% activated fraction possible based on unactivated wild-type antithrombin reactivities with factors Xa and IXa that are 0.3 and 0.7% of allosterically activated reactivities. An activated fraction of 0.1% would imply intrinsic native state antithrombin reactivities with factors Xa and IXa less than 2-fold lower than those observed, i.e. ka values of ~2500 m−1 s−1 and ~150 m−1 s−1, respectively. If Tyr253/His319 mutations only affected the 0.1% activated fraction, then the observed reactivities of these exosite mutant antithrombins in the absence of heparin should not be less than ~2500 m−1 s−1 and ~150 m−1 s−1.
4Assuming the changes in heparin affinity of the exosite mutants reflect changes in the conformational equilibrium between native and activated states of antithrombin in the unactivated serpin and given the observation that the observed wild-type heparin affinity significantly exceeds that of the native state (28), the fold-change in the fraction of activated antithrombin caused by the exosite mutations can be approximated by the ratio, KD,WT/KD,mut. Assuming ~0.1% activated fraction in wild-type, this would imply an activated fraction in Tyr253/His319 mutants of 0.01–0.03%.
5The allosteric activation rate enhancements for antithrombin-factor Xa reactions reported here are larger than those reported previously because rate constants were measured in calcium buffer in the present study but in phosphate buffer lacking calcium in past studies.
2The abbreviation used is: