Serine 380 (P14) → Glutamate Mutation Activates Antithrombin as an Inhibitor of Factor Xa*

Heparin regulates the inhibitory activity of antithrombin. It has been proposed that residues P15 and P14 are expelled from β-sheet A of antithrombin by heparin binding, permitting better interaction of the reactive center loop with factor Xa. We have made a P14 antithrombin variant (S380E) to create an activated inhibitory form of antithrombin in which P14 is already expelled from β-sheet A. S380E antithrombin fluorescence is enhanced 35 ± 5% compared with control antithrombin. There is minimal further increase in antithrombin fluorescence upon heparin binding. The variant has a 5 °C lower Tm than control antithrombin. The variant is an inhibitor of proteinases and has a nearly 200-fold increased basal rate of inhibition of factor Xa, after correction for an increased stoichiometry of inhibition. This is comparable to that of antithrombin activated by high affinity heparin pentasaccharide. Full-length high affinity heparin causes only a 7-fold additional increase in rate and a large increase in stoichiometry of inhibition. In contrast, the basal rate of inhibition of thrombin is similar to that of control antithrombin but is increased 300-fold by heparin. These findings suggest that the native state of the S380E variant exists in a loop-expelled conformation that is consequently highly reactive toward factor Xa.

In the absence of heparin the serpin antithrombin is a poorly reactive inhibitor of its principal target proteinases thrombin and factor Xa. However, in the presence of heparin reaction of antithrombin with both proteinases is greatly accelerated and thus provides an endogeneous mechanism for regulation of the activity of antithrombin. This key role of heparin in accelerating the reactions of antithrombin with blood coagulation proteinases also accounts for the widespread clinical use of heparin as an anticoagulant (1). For the reactions of antithrombin with both thrombin and factor Xa, the ability of long heparin chains to bind to both inhibitor and proteinase contributes significantly to the rate acceleration, though much more so to the thrombin-antithrombin reaction (ϳ2000-fold) (2) than to the factor Xa-antithrombin reaction (ϳ90-fold in the presence of Ca 2ϩ ) (3). A second contribution to the rate acceleration of these reactions is a major conformational change in the reactive center loop induced by heparin binding to antithrombin, which alters both the conformation and flexibility of the scissile bond (P1-P1Ј) and thereby increases the substrate-like initial interactions with proteinase (4). This conformational change results in an ϳ300-fold rate increase for the factor Xa-antithrombin reaction. The thrombin-antithrombin reaction is much less sensitive to the conformation of the reactive center loop and shows only an ϳ2-fold rate increase.
Both biochemical (5,6) and x-ray diffraction studies (7-11) support a mechanism for the heparin-induced conformational change in the reactive center loop in which the P15 and P14 residues at the N-terminal end of the reactive center loop, which are inserted into ␤-sheet A as strand 4 in the low reactivity state of native antithrombin, are expelled from the ␤-sheet upon heparin binding ( Fig. 1) to give a reactive center loop that is very much more reactive toward factor Xa. As part of the heparin-induced conformational change, the strands of ␤-sheet A that flank residues P15 and P14 in the native state are thought to close up. Because the remarkable mechanism by which serpins inhibit target proteinases requires insertion of the whole of the reactive center loop (from P15 to approximately P3) into ␤-sheet A as strand 4 (12,13), ␤-sheet A of antithrombin must reopen at the critical point of the mechanism to allow re-insertion of residues P15 and P14, as well as the remainder of the reactive center loop, with the proteinase covalently bound to the P1 residue.
It follows from the putative effects of heparin on the conformation and reactivity of the reactive center loop of antithrombin toward factor Xa and from the need for re-insertion of the P15 and P14 residues as part of the proteinase-inhibition mechanism, that it might be possible to create an antithrombin variant that is activated as an inhibitor of factor Xa by introducing a mutation at P14 that favors the loop-expelled conformation prior to reaction with factor Xa but does not block the subsequent and obligate efficient reinsertion of the loop into ␤-sheet A during the inhibition mechanism. We had previously attempted this by making a P14 Ser 3 Trp variant (5) and later a P14 Ser 3 Cys variant that could be derivatized on the cysteine (6). Neither variant satisfactorily fulfilled both requirements. Thus the P14 Ser 3 Trp variant showed only a small increase in reactivity toward factor Xa in the absence of heparin and a greatly reduced efficiency as an inhibitor, with a stoichiometry of inhibition (SI) 1 of 29 compared with 1.2 for wild-type antithrombin. The P14 Ser 3 Cys variant was very much more reactive toward factor Xa upon derivatization of the cysteine with the fluorescein moiety (6). However, the bulky fluorescein interfered with the inhibition pathway, with the result that the species was activated only as a substrate for factor Xa. We therefore sought to create a different P14 anti- 1 The abbreviations and trivial names used are: SI, stoichiometry of inhibition, defined as the number of mols of antithrombin required to inhibit one mol of target protease; P14, position 14 residues N-terminal of the scissile bond, which is defined as the bond between the P1 and P1Ј residues; N135Q, antithrombin in which asparagine 135 has been changed to glutamine, removing a potential glycosylation site; S380E, antithrombin in which residue 380 (P14) has been changed from serine to glutamate, on an N135Q background. thrombin variant that might be activated as an inhibitor of factor Xa, to test both the feasibility of this as well as the proposed role of the P15/P14 hinge in activation of antithrombin and its subsequent inhibition of proteinases. We report here a P14 serine 3 glutamate variant of antithrombin (S380E) that has such properties.

EXPERIMENTAL PROCEDURES
Production and Isolation of Antithrombin-S380E antithrombin was created using human antithrombin cDNA on an N135Q background. The N135Q background reduces heterogeneity arising from differences in glycosylation of the recombinant antithrombin by eliminating the glycosylation site at position 135 (14,15). The resulting antithrombin resembles the ␤-form of plasma antithrombin. Site-directed mutagenesis was carried out in M13mp19 as described previously (16), using coding strand template 5Ј-GTAAATGAAGAAGGCGAGGAAGCAGCT-GCAAG-3Ј (mutation site underlined). The change was confirmed by dideoxy sequencing. The expression vector, together with the selection plasmids pRMH140 and pSV2dhfr, was used to stably transfect baby hamster kidney cells, as described previously (17). Transfected cells were selected by resistance to neomycin and methotrexate. The antithrombin variant was isolated from serum-free cycles of medium collected from stably transfected baby hamster kidney cells grown to confluence in roller bottles. The level of antithrombin in the growth medium was monitored by radial immunodiffusion assay using commercially available plates containing sheep anti-human antithrombin antibody (Binding Site Ltd., Birmingham, United Kingdom).
Purification of S380E Antithrombin-The S380E antithrombin variant was purified by chromatography on heparin-Sepharose and Q-Sepharose. The protein was initially purified on heparin-Sepharose, using a 0.1-4 M salt gradient for elution. Antithrombin eluted between 2.5 and 4 M salt, at a salt concentration higher than that for control N135Q antithrombin eluted from the same column (see "Results"). The thrombin inhibitory activity was measured for each fraction by incubation of aliquots overnight with thrombin and measuring the residual thrombin activity, as described previously (18). Fractions containing anti-thrombin activity were collected and dialyzed overnight against 20 mM sodium phosphate buffer, pH 7.4, containing 0.1 mM EDTA. Further purification to remove any bound heparin was carried out by anion exchange chromatography on Q-Sepharose, using a 0.02-1 M salt gradient to elute the antithrombin. Fractions were pooled, dialyzed against 20 mM sodium phosphate buffer, pH 7.4, containing 100 mM NaCl and 0.1% polyethylene glycol 8000, concentrated, and frozen at Ϫ70°C until needed. The concentration of S380E antithrombin was determined using the extinction coefficient of wild-type antithrombin of E 280 nm 1% ϭ 6.5 (19).
SDS-Polyacrylamide Gel Electrophoresis-The ability of the S380E antithrombin to form covalent complexes with proteinases was determined from the appearance of characteristic SDS-stable high molecular weight bands on SDS-polyacrylamide gel electrophoresis. Antithrombin was incubated with proteinases at room temperature for the times and in the amounts indicated in the figure legend. The reactions were stopped by addition of 1 mM phenylmethylsulfonyl fluoride, and the products were run on 10% polyacrylamide gels.
Steady State Fluorescence Spectra-Fluorescence emission spectra of S380E antithrombin and control antithrombin and their complexes with full-length heparin were recorded at 25°C on an SLM 8000 spectrofluorometer or SPEX fluorometer, using an excitation wavelength of 280 nm and a bandpass of 4 nm. For comparison of spectra from control and S380E antithrombins, intensities were normalized based on a concentration determined from absorbance at 280 nm, assuming equivalent extinction coefficients for the two proteins of E 1% ϭ 6.5 (19). Such an assumption is likely to be in error by a maximum of ϳ5% (20). Emission spectra were recorded with a 2-nm step and bandpass of 2 nm, using an antithrombin concentration of 100 nM. A saturating amount of heparin was added for spectra of antithrombin-heparin complexes. Basal fluorescence intensities and enhancements caused by heparin are reported as the average of 5-10 separate measurements.
Determination of Rate of Inhibition of Proteinase by Antithrombin-Except for the reactions of factor Xa and thrombin with control antithrombin and of thrombin with S380E variant antithrombin, for which the second order rate constants were slow, all rate constant determinations were made from continuous progress curve assays, using the change in absorbance at 405 nm resulting from hydrolysis of chromogenic substrate. For these measurements the antithrombin:proteinase ratio was at least five times the SI to maintain pseudo first order reaction conditions. Progress curves were fitted to a single monoexponential (Equation 1) and corrected for competition from substrate according to Equation 2 (21).
Here [P] t is the product concentration at time t, k obs is the apparent second order rate constant, and k is the second order rate constant corrected for substrate competition. Trypsin reactions used 1-2 nM trypsin and 20 -100 nM control antithrombin or 100 -500 nM S380E antithrombin. Optimal heparin concentrations were determined empirically by carrying out experiments using increasing heparin concentrations until the measured rate plateaued. These optimal concentrations represent a molar excess of heparin over antithrombin, with concentrations in the range 1-5 M.
The reactions of factor Xa and thrombin with control antithrombin and of thrombin with S380E variant antithrombin were all monitored by discontinuous assay in which antithrombin at a concentration at least five times the SI was incubated with proteinase, and the loss of proteinase activity was determined at discrete time points by the removal of an aliquot of the reaction mixture and assay of enzymatic activity upon dilution into 100 M chromogenic substrate, using spectrozyme Xa for factor Xa and S2238 for thrombin. The pseudo first order rate constant k obs was determined from the slope of a plot of log of residual enzyme activity against time and then converted into the second order rate constant by dividing by the antithrombin concentra-

FIG. 1. Proposed equilibrium between the loop-inserted and the loop-expelled states of antithrombin induced by heparin binding.
The loop-inserted structure is based on the x-ray structure of native antithrombin (2ant), and the loop-expelled structure is based on the x-ray structure of antithrombin-heparin pentasaccharide complex (1azx). It should be noted that the P1 conformation of the loop-expelled antithrombin in solution is likely to differ from that shown here, because both crystal structures are of antithrombin heterodimers in which the reactive center loop near the P1 residue interacts with ␤-sheet C of the second, latent, molecule of the dimer and so is likely to be distorted from what occurs in the monomer in solution.
tion (21). Measurements by both continuous and discontinuous methods were carried out at least five times. The rate constants reported are the averages of these measurements.
Determination of Stoichiometries of Inhibition-SI (defined as the number of mols of antithrombin required to inhibit 1 mol of target proteinase) was determined by incubating antithrombin over a range of concentrations (0 -2 M) with fixed concentration of proteinase (30 -100 nM) in polyethylene glycol-coated tubes. Reactions were allowed to go to completion, and residual proteinase activity was measured from the rate of substrate hydrolysis at 405 nm. The stoichiometry of inhibition was determined by plotting residual proteinase activity against the ratio of antithrombin to proteinase. SIs determined in the absence of heparin were carried out in buffer containing 100 g/ml polybrene to avoid effects of any traces of contaminating heparin. For SIs determined in the presence of heparin, saturating amounts of heparin over antithrombin were used. SIs reported are the average Ϯ S.D. of 3-5 measurements.
Determination of Denaturing Temperature-Thermal denaturation of antithrombin was followed by monitoring change in ellipticity at 220 nm of the CD spectrum as a function of temperature. CD measurements were made on a Jasco 710 spectropolarimeter in a 1-mm pathlengthjacketed cell. The temperature was controlled by a Neslab water bath. Antithrombin concentrations were 0.4 mg/ml in I0.15 buffer. A bandwidth of 2 nm, step resolution of 0.5°C, response time of 16 s, and rate of temperature change of 0.5°C/min were used.
Materials-Dulbecco's modified Eagle's medium and fetal bovine serum were from Life Technologies, Inc. Human factor Xa was prepared as described previously (22). ␣-Thrombin was prepared from prothrombin, isolated from outdated human plasma by the method of Miletich et al. (23), by reaction with snake venom as described (24). Full-length high affinity heparin, M r 15,000, was prepared by fractionation of heparin first by size-exclusion chromatography and then by antithrombin affinity chromatography, as described (25).

Fluorescence Emission Spectra of S380E Antithrombin in the
Absence and Presence of Heparin-An approximately 35-40% enhancement of the fluorescence emission spectrum of plasma and recombinant wild-type antithrombin, together with a 2-nm blue shift in the position of the emission maximum occur as a result of the conformational changes that occur upon heparin binding (26). The fluorescence enhancement has been shown to arise predominantly (Ͼ70%) from changes in the environment of tryptophans 225 and 307 as a result of expulsion of the hinge of the reactive center loop from ␤-sheet A (27). These fluorescence changes have been routinely used as an indicator of the occurrence of such structural changes in antithrombin. In the absence of heparin, we found that the S380E variant had a normalized fluorescence intensity that was already 35 Ϯ 5% higher than that of control antithrombin, suggesting that the conformation of the variant was similar to that of heparinactivated antithrombin, and specifically that the environments of tryptophans 225 and 307 had been changed in the manner expected from expulsion of the reactive center loop from ␤-sheet A. In keeping with this conclusion, binding of heparin caused only a 3 Ϯ 2% increase in the emission intensity of the S380E variant, together with a 2-nm blue shift, whereas heparin caused a 40% enhancement of the fluorescence intensity of the control antithrombin and a 2-nm blue shift (Fig. 2). The small increase for the S380E variant is consistent with the expected direct contact effects of heparin binding on tryptophan 49.
Affinity of S380E Antithrombin for Heparin-Because there was such a small change in the tryptophan emission spectrum of S380E antithrombin upon addition of heparin, it was not possible to use change in tryptophan fluorescence as a means of determining the heparin-antithrombin dissociation constant for the variant. Instead we compared the elution positions of control antithrombin and S380E antithrombin from heparin-Sepharose under identical conditions, as a qualitative measure of affinity. We found that the S380E variant eluted at significantly higher salt concentration (ϳ2.2 M) compared with con-trol antithrombin (ϳ1.5 M) (Fig. 3). Similar behavior has been seen previously for a S380W variant of antithrombin (5), where the Ser 3 Trp mutation led to a partial shift toward the activated, loop-expelled state. For the purposes of analysis and interpretation of the other experiments presented here, such a qualitative assessment of heparin affinity is sufficient, because we can be sure that under similar antithrombin and heparin concentrations for wild-type and S380E antithrombins, the S380E variant will be saturated to the same or greater extent.
S380E Antithrombin Has Reduced Stability-In wild-type antithrombin it is thought that the P15 and P14 residues inserted into ␤-sheet A in the native state contribute to the overall stability of the protein. It is therefore expected that, if the mutation of the P14 residue to glutamate has displaced these residues from the ␤-sheet, there might be a reduction in the stability of the variant protein. We used change in ellipticity of the CD spectrum of the protein to follow thermal denaturation and found (Fig. 4) that the S380E variant denatured at a temperature 5°lower than that of the control antithrombin (52.5°C versus 57.5°C, respectively). This is consistent with the S380E variant existing in the loop-expelled conformation in which the P15 and P14 residues do not contribute to the stability of the protein, in contrast to the loop-inserted control antithrombin.
Ability of S380E Antithrombin Variant to Form Covalent Complexes with Proteinase-The ability of the S380E variant to form the SDS stable complexes with proteinase that are the hallmark of proteinase inhibition by the serpin suicide substrate mechanism was examined for reaction with factor Xa, thrombin, and trypsin. For each proteinase there was clear evidence for the presence of lower mobility bands on SDSpolyacrylamide gel electrophoresis corresponding to a covalent complex between antithrombin and the proteinase (Fig. 5). However, from the relatively low intensity of bands corresponding to high molecular weight covalent complexes compared with the much higher intensity of bands corresponding to reactive center loop-cleaved antithrombin, it can be gauged that the SI for reaction with each proteinase must be much greater than 1. The same reactions were also examined in the presence of heparin but in each case gave only cleaved antithrombin as a detectable product (not shown), indicating that any complex formed must represent a very small fraction of the total reaction products and therefore be below the detection limit of this method.
Stoichiometries of Inhibition-To obtain more accurate values for the SIs for inhibition of factor Xa, thrombin, and trypsin than could be obtained from SDS gels, both in the absence and presence of heparin, we used kinetic assays as described under "Experimental Procedures." Plots of residual activity against antithrombin concentration for fixed proteinase concentration gave straight lines in each case, showing that the covalent complexes were stable over the time course of the experiments (Fig. 6). Whereas control antithrombin gave SI values for each proteinase close to the expected values of about 1, the S380E variant was less efficient as an inhibitor and gave elevated SI values in the absence of heparin and very much increased SI values in the presence of heparin (Table I). The S380E variant was most effective as an inhibitor of thrombin (SI ϭ 4.2) and trypsin (SI ϭ 5) and less effective as an inhibitor of factor Xa (SI ϭ 7). The effect of increasing the temperature from 25 to 37°C was examined for the antithrombin-factor Xa reaction but was found to give only a small increase in inhibition efficiency (SI ϭ 6.5). For all three proteinases, the effect of having heparin present was to greatly increase the SI so that less than 1% of the reaction intermediate followed the branch of the pathway to a stable complex compared with Ͼ99% that followed the substrate branch of the pathway (Table I). Because heparin is known to stabilize the loop-expelled antithrombin conformation and so acts to hinder the re-insertion of the reactive center loop during the inhibition reaction, the findings indicate that the S380E variant is much more sensitive to this hindrance than is control antithrombin, consistent with the higher heparin affinity of the variant. This is also in keeping with the SI values in the absence of heparin being Ͼ1, such that flux along the substrate branch of the pathway is already significant relative to the inhibitory branch.
S380E Antithrombin Is Activated Against Factor Xa-Second order rate constants were determined for the reactions of control and S380E variant antithrombins with factor Xa, thrombin, and trypsin both in the absence and presence of heparin. Because the rate constants were determined from the kinetics of loss of proteinase activity, the apparent rate constants determined in this way must be corrected for the flux along the substrate pathway to give the true second order rate constant. These corrected rate constants show (Table II)  factor Xa, thrombin, and trypsin, respectively. The multiple bands for complex probably arise from the presence of a small amount of free factor Xa that is able to cleave the factor Xa-antithrombin complex. Such cleavages by excess proteinase are well documented for many serpin-proteinase pairs. In the case of the reactions with thrombin and trypsin there is still some unreacted antithrombin at the end of the incubation time that serves to prevent any proteolysis of the complex by free proteinase. For the factor Xa-antithrombin reaction, all of the antithrombin has reacted. The gel is a 10% SDS-polyacrylamide gel run under nonreducing conditions. the absence of heparin, the S380E variant reacts with factor Xa ϳ190-fold faster than does control antithrombin. In contrast, the S380E variant reacts with thrombin only twice as fast as control and with trypsin five times as fast as control. This selective enhancement for the factor Xa reaction is in keeping with the conformation of S380E antithrombin being that of the loop-expelled, activated form, because only the reaction of factor Xa with antithrombin shows a strong dependence of the rate on the conformation of the reactive center loop (25).
The additional bridging effects of full-length high affinity heparin with each proteinase were also consistent with the S380E variant already being in the loop-expelled conformation. Thus the reaction with factor Xa was increased only a further 7-fold, corresponding to a small bridging contribution in the absence of Ca 2ϩ (3,25). For reaction with thrombin, where almost all of the large rate enhancement from full-length heparin arises from a bridging contribution, heparin gave rise to large enhancements for both control and S380E variants (to 1.7 ϫ 10 7 M Ϫ1 s Ϫ1 and 0.7 ϫ 10 7 M Ϫ1 s Ϫ1 , respectively). For trypsin there are only small rate enhancements from both conformational change and bridging. In keeping with this the rate of reaction of the S380E variant in the presence of heparin was only 6-fold faster than in its absence. DISCUSSION With the demonstration that the S380E variant of antithrombin is an inhibitor of proteinases by the serpin suicide substrate inhibition mechanism, through its ability to form covalent SDS stable complexes, and the finding that it reacts with factor Xa at a rate 190-fold higher than that of control antithrombin, we have established the feasibility of creating an activated inhibitory antithrombin through modification of the hinge region of the reactive center loop. The tryptophan fluorescence, heparin affinity, and thermal stability of the S380E variant are all consistent with activation resulting from the reactive center loop being fully expelled from ␤-sheet A, whereas the high sensitivity of the SI to heparin being bound is consistent with the importance of facile reinsertion of the reactive center loop into ␤-sheet A for efficient inhibition of pro-teinase. The agreement between the observed and expected properties of this variant provides strong support for the currently favored model of how antithrombin inhibits proteinases and how heparin influences both the rate of reaction of antithrombin and proteinase and the outcome of the reaction with respect to relative fluxes along the inhibitory and substrate branches of the serpin pathway.
The tryptophan emission spectra in the absence and presence of heparin provide direct structural evidence that the S380E variant is a P14-expelled form. Of the four tryptophans in antithrombin (at positions 49, 189, 225, and 307), only tryptophan 49 is within the heparin binding site. However, we have previously shown that tryptophans 225 and 307 are responsible for ϳ70 -80% of the heparin-induced fluorescence enhancement and are therefore specific reporters of the longer range conformational change effects of heparin binding. This is now understandable in molecular terms from the recently determined structure of the antithrombin-heparin pentasaccharide complex, which shows that heparin binding removes the P14 side chain from contact with tryptophan 225 and causes a 180°f lip of the side chain of tryptophan 307 (11). The ϳ35% enhanced tryptophan fluorescence of S380E antithrombin and relative insensitivity of the fluorescence to further change upon binding heparin are therefore probably due to the environments of tryptophans 225 and 307 being equivalent to those of heparin-activated antithrombin, which would require the P15 and P14 residues to have been expelled from ␤-sheet A. Heparin binding should then cause no further change with respect to these two tryptophans, even though it should still induce conformational change within the heparin binding site. The small blue shift and 3% enhancement upon binding heparin to the S380E variant are consistent with this, because we have also previously shown that direct contact of heparin with tryptophan 49 causes a small increase in fluorescence intensity, together with the overall blue shift of the tryptophan emission spectrum of antithrombin (27).
The lower thermal stability and higher heparin affinity of the S380E variant compared with control antithrombin are also consistent with such a loop-expelled conformation. Two other P14 variants of antithrombin that we have previously examined show alterations in thermal stability and heparin affinity that parallel whether the loop-inserted conformation is favored or disfavored. Thus an S380W variant, which favors loop expulsion, is less stable than control antithrombin by 5.5°C (5), whereas an S380C variant, which favors loop insertion, is more stable by 2.3°C (6). The heparin affinities show the reverse pattern, with higher affinity for the less stable S380W variant and lower affinity for the S380C variant.
The observed rate constants for reaction of the S380E variant with different proteinases also show the expected changes from control antithrombin. Thus, the expectation for reaction of factor Xa with a reactive center loop-expelled conformer of antithrombin is that the rate constant should be increased several hundred-fold. This is because the IAGR sequence of the P4-P1 residues of antithrombin, although close to the optimal one for factor Xa (IEGR), is thought to have an inappropriate fixed conformation in the loop-inserted state of antithrombin but to become more conformationally labile and hence more reactive in the loop-expelled conformation. In contrast, because the P3-P1 residues, AGR, are far from the optimal sequence for reaction with thrombin (FPR), the change in conformational flexibility arising from loop expulsion results in little increase in reactivity of antithrombin with thrombin. For trypsin, which has low specificity for residues other than P1, the rate of reaction with the loop inserted conformation of antithrombin is already quite high (ϳ10 5 M Ϫ1 s Ϫ1 ) and gives a relatively modest increase upon expulsion of the reactive center loop. The rate constants for reaction of the S380E variant are in agreement with these expectations for each of the three proteinases. In addition, the further effects of long chain high affinity heparin are also as expected if the S380E conformation is already in the loop-expelled conformation, because only additional bridging contributions would be expected, which would be modest for factor Xa (in the absence of Ca 2ϩ ) and trypsin (as seen) but very large for thrombin (again as seen).
In addition to the large changes in reaction rate and altered fluorescence, thermal stability, and heparin affinity, the other major difference between control and S380E variant antithrombins is the elevation of SI values, which is moderate in the absence of heparin and very large in the presence of heparin. To understand these effects it is useful to review the serpin inhibitory mechanism. Serine proteinases recognize serpins as substrates and initiate a normal proteolytic cleavage at the P1-P1Ј scissile bond. At the acyl enzyme intermediate stage, i.e. while the carbonyl is still covalently linked to the active site serine of the proteinase, a conformational change occurs in which the central strands of ␤-sheet A of the serpin separate and the reactive center loop inserts as a new antiparallel strand dragging the proteinase with it (28 -30). Because loop insertion leading to inhibition is in competition with deacylation and release of the proteinase and cleaved serpin, there are two possible products of reaction-covalent serpinproteinase complex and cleaved serpin whose relative proportions reflect the relative rates of the two processes, loop insertion versus deacylation. Anything that speeds loop insertion relative to deacylation will favor inhibition, whereas anything that hinders loop insertion will favor substrate cleavage, if the hindrance is great enough. The increases in SI for the S380E variant in the absence of heparin most likely result from slower loop insertion. This is not surprising, because normal loopinserted forms of serpins have the P14 side chain pointing below ␤-sheet A into the protein interior. Other things being equal, it is likely that there would be a higher activation energy for burial of a charged glutamate than a neutral serine and hence a slower initiation of loop insertion. This is also consistent with the conclusion that the native state of the S380E variant has the P14 glutamate displaced from ␤-sheet A, in contrast to the inserted wild-type serine. Similar increases in SI have been seen in other serpins when the P14 residue has been changed to a charged one, such as Thr 3 Arg in both ␣ 1 -proteinase inhibitor (31) and ␣ 1 -antichymotrypsin (32). The further large increases in SI for the S380E variant in the presence of heparin are also explicable in terms of a reduced rate of loop insertion. Thus, heparin binds much more tightly to the P14-expelled state than the P14 loop-inserted state. When heparin is bound and therefore stabilizing the loop-expelled state, it is harder to re-insert the reactive center loop into ␤-sheet A than it is in the absence of heparin. The consequent effect on SI has been well illustrated for wild-type antithrombin-thrombin complex formation, where, in the presence of heparin, the SI increases as a function of decrease in ionic strength and consequent increase in heparin affinity (33). For the thrombin-antithrombin reaction the SI was 1.1 at I0.3, but increased to 9.8 at I0.01. In terms of change in k 4 , the rate constant for the inhibitory branch of the serpin pathway, this represents an 88-fold change, i.e. of comparable magnitude to the changes seen here for the S380E variant upon binding heparin (e.g. ϳ80-fold for the reaction with factor Xa).
Finally, given that the S380E has properties that are consistent with it being a loop-expelled and therefore activated inhibitor of factor Xa, the possible basis for this shift in conformation should be considered. On the one hand the mutation might destabilize the loop-inserted form, whereas on the other it might stabilize the loop-expelled form. Both may also occur together. The result most relevant to this is the thermal stability of the variant, which is less than that of the control antithrombin. This suggests that the destabilizing effect on the loop-inserted form must dominate. Also, because the variant is still an inhibitor, the destabilizing effect of the mutation on the loop-inserted form must not be too great, otherwise inhibition, requiring loop re-insertion, might be slowed too much. This suggests that any stabilizing effect on the loop-expelled form, such as might occur from new salt bridge interactions, should be small. If this is correct, the effects of the mutation are rather straightforward in the absence of heparin and suggest that introduction of other charged residues, such as arginine, lysine, or aspartate might give variants with qualitatively similar properties.