Requirement of lysine residues outside of the proposed pentasaccharide binding region for high affinity heparin binding and activation of human antithrombin III.

Variant forms of human antithrombin III with glutamine or threonine substitutions at Lys114, Lys125, Lys133, Lys136, and Lys139 were expressed in insect cells to evaluate their roles in heparin binding and activation. Recombinant native ATIII and all of the variants had very similar second order rate constants for thrombin inhibition in the absence of heparin, ranging from 1.13 × 105 M−1min−1 to 1.66 × 105 M−1min−1. Direct binding studies using 125I-flouresceinamine-heparin yielded a Kd of 6 nM for the recombinant native ATIII and K136T, whereas K114Q and K139Q bound heparin so poorly that a Kd could not be determined. K125Q had a moderately reduced affinity. Heparin binding affinity correlated directly with heparin cofactor activity. Recombinant native ATIII was nearly identical to plasma-purified ATIII, whereas K114Q and K139Q were severely impaired in heparin cofactor activity. K125Q and K136T were only slightly impaired. Based on these data, Lys114 and Lys139, which are outside of the putative pentasaccharide binding site, play pivotal roles in the high affinity binding of heparin to ATIII and the activation of thrombin inhibitory activity.

M ؊1 min ؊1 to 1.66 ؋ 10 5 M ؊1 min ؊1 . Direct binding studies using 125 I-flouresceinamine-heparin yielded a K d of 6 nM for the recombinant native ATIII and K136T, whereas K114Q and K139Q bound heparin so poorly that a K d could not be determined. K125Q had a moderately reduced affinity. Heparin binding affinity correlated directly with heparin cofactor activity. Recombinant native ATIII was nearly identical to plasma-purified ATIII, whereas K114Q and K139Q were severely impaired in heparin cofactor activity. K125Q and K136T were only slightly impaired. Based on these data, Lys 114 and Lys 139 , which are outside of the putative pentasaccharide binding site, play pivotal roles in the high affinity binding of heparin to ATIII and the activation of thrombin inhibitory activity.
The D-helix and adjacent regions have been widely implicated to play an important role in the heparin cofactor activity of human antithrombin III (1)(2)(3)(4). Several positively charged amino acids in this region, including Lys 114 , Lys 125 , Arg 129 , Arg 132 , Lys 133 , Lys 136 , and Lys 139 , align and present a region of high positive charge density (5). This region is believed to represent the primary binding site in ATIII 1 for heparin, presumably through coordinate bonds formed with sulfate groups in the linear heparin chain. Although the binding of heparin to ATIII is a requisite step in the activation process, it is not known whether the binding of heparin to the D-helix is sufficient for activation or if other residues away from the helix may be required to achieve the conformational change that leads to the activated form of ATIII. The close proximity of the lysine and arginine residues in the D-helix and adjacent regions in the intact protein would suggest that it is not possible for all of them to play roles in heparin binding, given the distribution of sulfate groups in the heparin chain (6). The first step in under-standing the activation process is thus to define the role(s) of individual residues in the D-helix region, which represents at least the initial heparin binding site in ATIII.
Naturally occurring variants of ATIII have facilitated defining the precise role of two residues in the D-helix region; Arg 47 , which is in the A-helix but lies in close proximity to the amino end of the D-helix in the native protein (7)(8)(9)(10), and Arg 129 , which lies near the carboxyl end of the D-helix (11). In both cases the variant proteins display reduced binding to heparin-Sepharose and reduced heparin cofactor activity. Most of the information on the roles of other residues has come from chemical modification studies. Modification of Lys 125 with pyridoxal-5Ј-phosphate abolishes binding to heparin-Sepharose (12,13). The effect on heparin cofactor activity was not evaluated in those studies. Several other residues including Lys 107 , Lys 114 , Lys 133 , Lys 136 , and Lys 139 have been similarly implicated by chemical modification or protection from chemical modification in the presence of heparin. These studies are limited in interpretation, however, because two or more residues were modified simultaneously (14 -16). Recently ATIII was expressed in baby hamster kidney cells (17,18). The data from these studies confirmed that Lys 125 is an important residue in heparin binding and activation of ATIII, whereas Arg 132 and Lys 133 are at least important in heparin binding. However, several other positively charged residues, including Lys 290 , Lys 294 , and Lys 297 , which are well outside of the proposed heparin binding region, were found to be unimportant.
Several fluorescent enhancement and 1 H-NMR studies (19 -23) along with the crystal structure of a dimerized form of ATIII (24) have determined the precise binding pocket and conformational contribution by the heparin pentasaccharide in ATIII. It has been proposed that because the pentasaccharide does not activate ATIII toward thrombin to the same extent as longer chain heparins, heparin must play a bridging role by bringing the two molecules in proximity of one another (25)(26)(27)(28)(29)(30).
In the present study we show that two residues outside of the pentasaccharide binding pocket are important for heparin binding and activation of ATIII toward thrombin, possibly leading to a new model for the role of longer chain heparins. Lys 114 , Lys 125 , Lys 133 , Lys 136 , and Lys 139 were independently replaced with either glutamine or threonine (Lys 136 ), and the variant ATIII proteins were expressed, along with recombinant native ATIII, in insect cells using a baculovirus-driven expression system. In the absence of heparin, the recombinant native ATIII and all of the variants had similar second order rate constants and formed SDS-PAGE-resistant complexes with thrombin at nearly identical levels. In contrast, the variants had heparin cofactor activities that varied when compared with the recombinant native ATIII. Although the K136T and the K125Q variants differed by about a factor of five to six, the K114Q and the K139Q variants were catastrophic, having a 10 -20-fold decrease in their rate of thrombin inhibition in the presence of heparin. Direct heparin binding studies revealed that the recombinant native ATIII bound heparin with a K d of 6 nM, as did the K136T variant. However, affinity of the K125Q variant for heparin decreased 2-fold, yielding a K d of approximately 12 nM, whereas the K114Q and K139Q variants bound heparin so poorly that K d values could not be determined. These data define, for the first time, an important role for Lys 114 and Lys 139 in the heparin cofactor activity of ATIII by demonstrating that they are essential for heparin binding despite the fact that the residues are outside of the pentasaccharide binding pocket. The residues may serve to align the longer chain heparins in a proper orientation so that they align properly as a template for thrombin, or they may serve to stabilize ATIII in its active conformation when heparin is present. Generation and Expression of Recombinant ATIII-Site-directed mutagenesis was performed using polymerase chain reaction (PCR) (31). Four oligonucleotides were required to generate each mutant. Two were ATIII cassette oligos designated ATIII cassette antisense oligo and ATIII cassette sense oligo that flanked the region where nucleotide changes were to be introduced and two within this region contained the sequence with the desired base changes. Two primary PCR reactions were required. One reaction contained ATIII cassette sense oligo with the mutant antisense oligo, and the second contained ATIII cassette antisense oligo with the mutant sense oligo. The sequences of the mutant sense and antisense oligos are as follows from 5Ј to 3Ј: K114Q, GACACCATATCAGAGCAAA-CATCTGATCAT and CTGATCAGATGTTTGCTCTGATATGGTGTC; K125Q, ACTTCTTCTTTGCCCAGCTGAACTGCCGAC and GTCG-GCAGTTCAGCTGGGCAAAGAAGAAGT; K133Q, CCGACTCTATCGA-CAGGCCAACAAATCCTC and GAGGATTTGTTGGCTCTGCGATA-GAGTCGG; K136T, ATCGAAAAGCCAACCAGTCATCCAAGTTAG and CTAACTTGGATGACTGGTTGGCTTTTCGAT; and K139Q, CAAATC-CTCCCAGTTAGTATC and GATACTAACTGGGAGGATTTG, respectively. Following size verification on Tris acetate EDTA-agarose gels, the primary PCR products were extracted and purified using Geneclean. The two primary products were then combined with the ATIII cassette antisense oligo and ATIII cassette sense oligo oligos for a second round of PCR amplification. The codon changes were verified by restriction analysis of each mutated cDNA.

Materials-The
The mutant constructs were subcloned into a pGemATIII vector at the NcoI and SacII sites and sequenced by Sanger dideoxy sequencing using Sequenase version 2.0 (U. S. Biochemical Corp.). The cDNAs were then cloned into the pVL 1393 transfer vector at the PstI BamHI sites. The vector was then cotransfected into Sf9 cells with BaculoGold baculovirus. 5 days post infection the media were harvested, and limited dilution plaque assays were performed. After plaque formation individual plaques were harvested for further infections. Media were harvested 5 days after infection and assayed for the presence of ATIII by a thrombin inhibition assay (32). Media that tested positive for ATIII were selected and used for further rounds of viral amplification. Amplified virus was then used to infect High 5 insect cells at an multiplicity of infection of 10:1.
Purification of Recombinant ATIII-Recombinant baculovirus harboring recombinant native ATIII and the variants were used in parallel to infect High 5 insect cells. Conditioned media, containing the recombinant proteins, were harvested 4 days post infection and centrifuged at 9,000 ϫ g for 30 min. They were further clarified by filtration through 0.22-m filters. Recombinant native ATIII and the ATIII variants were affinity purified using a goat anti-human ATIII IgG coupled to Sepharose CL-4B (33). After passage of the media over the antibody affinity column, the column was washed with 0.25 M Tris-HCl, 1 M NaCl, pH 7.5, followed by elution with 0.25 M Tris-HCl containing 3.5 M MgCl 2 (11,34). The fractions were tested by dot blot analysis using an anti ATIII monoclonal antibody and 125 I-radiolabeled secondary antibody. Fractions containing ATIII were pooled and dialyzed extensively against PBS. Following concentration, the recombinant proteins were analyzed by Western blot analysis and SDS-PAGE on 10% polyacrylamide gels. Protein concentrations were determined by the method of Bradford (35) and quantitative Western blots.
Kinetic Analysis of Thrombin Inhibition-The colorimetric substrate Chromozym-Th was used to determine second order rate constants for thrombin inhibition by the recombinant native ATIII and the ATIII variants (11). Various concentrations of ATIII with or without heparin were incubated for the indicated times with a constant amount of thrombin (20 nM) at 37°C in PBS containing 1% polyethylene glycol. At each time point the reactions were subsampled into a Chromozym-Th solution containing 50 mM Tris, pH 8.3, and 227 mM NaCl. The colorimetric reactions were allowed to proceed for 5 min and then quenched by the addition of glacial acetic acid. Absorbance was read at 405 nm. Pseudo-first order rate constants were determined by plotting V/V o versus time. Second order rate constants were determined by dividing the pseudo-first order rate constants by the respective ATIII concentrations (36).
Heparin Titration Assays-Recombinant native ATIII was incubated with the indicated heparin concentrations in the same buffer described above for 30 min at 37°C. Thrombin was then added at a final concentration of 20 nM, and the reaction was allowed to proceed for 1 min. The degree of thrombin inhibition was determined using the Chromozym-Th assay described above.
Radioiodinations-Thrombin and F-heparin were radioiodinated as described previously (32,37). The specific activities were 15,000 and 70,000 cpm/ng for thrombin and heparin, respectively.
Analysis of Heparin Binding to ATIII Variants-96-well plates were coated with ATIII at 10 g/ml in PBS for 1 h at 37°C. The wells were then blocked with 3% bovine serum albumin in 50 mM NaCl, 10 mM Tris, pH 7.4, for 1 h at 37°C. The indicated concentrations of 125 I-Fheparin were then added to each well and allowed to bind for 1 h at 4°C. Following the removal of unbound ligand, the contents of the wells were removed with 10% SDS and quantitated by ␥ counting.
Analysis of Complex Formation-Complex formation in the absence of heparin was performed by incubating the indicated amounts of recombinant native ATIII and K114Q variant with 125 I-thrombin for 30 min at 37°C. The reactions were resolved by SDS-PAGE on 10% polyacrylamide gels and analyzed using a Bio-Rad GS-250 Molecular Imager and autoradiography. For heparin cofactor activity the antithrombins were first incubated with heparin at the indicated concentrations for 30 min at 37°C. 125 I-Thrombin was added, and the incubations were continued for an additional minute. The reactions were stopped and analyzed as described above.

Construction and Sequencing of ATIII Variants-Following
primary and secondary PCR reactions, size verification, and restriction analysis, each variant cDNA was cloned into a pGe-mATIII vector for sequencing by the Sanger dideoxy method using Sequenase, version 2.0. Shown in Fig. 1 are the regions of the sequencing gels that include the codon that corresponds to Lys 114 . Panel A shows the native ATIII cDNA sequence with the AAA codon that corresponds to Lys 114 in the mature protein sequence (after signal sequence cleavage). Panel B shows the same codon in the K114Q cDNA with the PCR introduced change of A to C at the first position in the codon, changing the codon specificity to a Gln. The rest of the sequence was found to be identical to the native ATIII cDNA sequence, eliminating any concern that Taq polymerase introduced errors. All of the mutants were analyzed similarly to ensure the desired codon changes and the absence of any Taq polymerase errors (data not shown).
Expression and Purification of Recombinant ATIIIs-Plaque-purified recombinant baculoviruses independently harboring recombinant native ATIII and the ATIII variant cDNAs were used to infect High 5 cells. The media were harvested, clarified, and passed over an anti-ATIII IgG column. The MgCl 2 eluates that tested positive for ATIII (see "Experimental Procedures") were dialyzed and concentrated to a final volume of 3-5 ml. The purified proteins were quantitated by the method of Bradford (35) and by quantitative Western analysis using a monoclonal antibody mapped to a region outside the region where the mutations were generated (data not shown). To assess purity, the proteins were subjected to SDS-PAGE on 10% polyacrylamide gels. A typical gel is shown in Fig. 2. Lanes 1 and 3 show samples of recombinant native ATIII and the K114Q variant, respectively. In both cases, a single ATIII band accounted for over 90% of the Coomassie-stained proteins present. Lanes 2 and 4 show the same amounts of recombinant native ATIII and the K114Q variant incubated with 600 ng of thrombin for 30 min at 37°C prior to electrophoresis. In both cases, all of the visible ATIII was shifted to a high molecular weight complex with thrombin, demonstrating that a majority of both recombinant native ATIII and K114Q variant are active. To ensure the quality of material, the same procedure was followed for all variant proteins. Other investigators have demonstrated the production of a an ATIII species in insect cells, which is not glycosylated at position 135 (36). The absence of the beta species in our preparations may be due to the use of the High 5 insect cells, which a have a much higher capacity than Sf9 cells for protein production (38). Therefore the glycosylation consensus site at Asn 135 is used efficiently because the glycosylation enzymes are not over-saturated with newly synthesized proteins. This assessment is further corroborated in Fig. 4 below, where it is shown by Scatchard analysis that heparin binding to the recombinant native ATIII reveals a single affinity class of binding sites.
Kinetics of Thrombin Inhibition by Recombinant Native ATIII and the K114Q, K125Q, K136T, and K139Q Variants-Second order rate constants for the recombinant native ATIII and ATIII variants in the absence of heparin were generated using Chromozym-Th as a chromogenic substrate (11). The time-dependent inhibition of thrombin was monitored over a broad concentration range of recombinant native ATIII as well as each variant protein. Second order rate constants were calculated by dividing the pseudo-first order rate constants by the inhibitor concentration (36). Table I  The ATIII K114Q and K139Q Variants Have Markedly Impaired Heparin Binding-Although heparin binding and activation are linked events, the amino acid residues involved in binding may not be directly involved in the activation process. Conversely, amino acid residues involved in activation may contribute little to the affinity of heparin binding. It is essential to look at both processes independently to gain a clearer understanding of the activation mechanism. To determine the effect of the substitutions at the different Lys positions, direct heparin binding studies were done using a solid phase binding assay in 96-well plates. 125 I-F-heparin was added to each well at a concentration of 42 nM along with various amounts of unlabeled F-heparin. At equilibrium, free ligand was removed, and the contents of the wells were solubilized in 10% SDS and quantitated by ␥ counting (Fig. 3). 125 I-F-heparin bound to recombinant native ATIII specifically and saturably, as demonstrated by the effective competition for binding by unlabeled F-heparin. In contrast, very little specific binding was observed with the K114Q variant, and the total binding was 7-8-fold lower than the recombinant native ATIII.
To determine the affinity of heparin for recombinant native ATIII and the variants, the binding of 125 I-F-heparin was quantitated over a broad concentration range (Figs. 4A and 5). 125 I-F-heparin binding to immobilized recombinant native ATIII was concentration-dependent and began to saturate between 50 and 100 nM. Binding to the immobilized K114Q variant was also saturable but was at least 7-fold below that of the recombinant native, demonstrating that the K114Q variant is severely impaired in the binding of heparin. A Scatchard transformation of the data in Fig. 4A is shown in Fig. 4B. A linear   (20 nM) was incubated at 37°C with recombinant native ATIII and the ATIII variant proteins at several different concentrations, ranging from 65 to 135 nM. Samples were taken at 10-min intervals for 40 min and assayed for residual thrombin activity using Chromozym-Th as a colorimetric substrate to generate pseudo-first order rate constants. Second order rate constants were generated by dividing the pseudo-first order rate constants by the respective inhibitor concentrations. Dissociation constants were measured using solid phase binding assays as described under "Experimental Procedures." The asterisks indicate an unmeasurable K d .

Inhibitor
Second  (36,39). Because of the drastically reduced affinity of heparin for the K114Q variant, it was not possible to transform the data to determine a K d . The same experiment was also performed for the K125Q, K136T, and K139Q variants (Fig. 5). The binding isotherm demonstrated that the lysine at position 136 is not important in terms of heparin binding. Lys 125 , as expected from previous studies, was involved in heparin binding. This is demonstrated by the fact that when the data from the binding isotherm for the K125Q variant was transformed for Scatchard analysis, a K d of approximately 12 nM was determined. This is a 2-fold lower affinity for heparin by the K125Q variant relative to recombinant native ATIII. Like the K114Q variant, the data from the binding isotherm for the K139Q variant could not be transformed into a clear Scatchard analysis. This indicates that the K139Q variant also bound heparin very poorly.
To ensure that the data observed in the Scatchard analysis above were not artifacts caused by a differential adsorption of recombinant native ATIII and the ATIII variants, the control experiment was done to measure the amount of ATIII physically immobilized in the wells. Recombinant native ATIII and the variant proteins were immobilized in a 96-well plate as described above. The relative amount of each protein bound was determined using monoclonal antibody 4D11/B10 (see "Experimental Procedures" and data not shown). Because the recombinant native ATIII and the variants were detected at the same levels, the difference in the binding isotherms is thus the result of a true difference in heparin binding and not an artifact resulting from unequal immobilization of the recombinant forms of ATIII to the wells.
In addition to demonstrating that the recombinant native ATIII produced in insect cells behaves identically to plasmapurified ATIII with respect to heparin binding, the binding isotherm data also demonstrate the presence of one predomi-nant heparin binding affinity form of ATIII. If a significant mixture of the alpha and beta isoforms were present, the Scatchard analysis should have been biphasic given the large difference in affinity of the two forms for heparin (40).
The K114Q and K139Q Variants Have Markedly Reduced Heparin Cofactor Activity-To assess the effect of the point mutations on heparin cofactor activity, two different assay methods were used for each of the variants. The first employed a Chromozym-Th assay where the time-dependent inactivation of thrombin in the presence and the absence of heparin was compared. The second was a short time frame assay where linkage formation was quantitated using 125 I-thrombin under conditions where the concentration of heparin was varied and the time was held constant. To first establish the appropriate working concentration of heparin in the Chromozym-Th assay, recombinant native ATIII (65 nM) was incubated at 37°C for 30 min with increasing concentrations of heparin, followed by the addition of thrombin (20 nM). After 1 min residual thrombin activity was measured using Chromozym-Th. From the plot it was determined that the concentration of heparin required to achieve half-maximal activation of recombinant native ATIII (EC 50 ) was 42 nM (data not shown).
The heparin cofactor activity of recombinant native ATIII and the ATIII variants were directly compared in parallel. The reaction conditions were identical to those described above in the absence of heparin or in the presence of heparin at concentrations ranging from 21 to 84 nM at 37°C for 30 min. Thrombin was then added to a final concentration of 20 nM, and the incubations were continued. At the indicated times the residual thrombin activity was measured as described above using Chromozym-Th (Fig. 6). The thrombin inhibitory activity of the recombinant native ATIII was sharply accelerated by 42 nM heparin. At 2.5 min 50% of the thrombin was inactivated in the presence of heparin compared with the heparin minus control where 50% inhibition was not achieved until 20 min. In contrast, the thrombin inhibitory activity of the K114Q variant was not significantly accelerated by heparin even when the concentration of heparin was increased to 84 nM.
The same experiment described above was performed on the K125Q, K136T, and K139Q variants in the presence of 42 nM high affinity heparin (Fig. 7). Like the previous experiment, the recombinant native ATIII achieved a 50% inhibition in about 2.5 min. The K125Q and the K136T variants did not achieve 50% inhibition until 14 and 17 min, respectively. The K139Q variant, however, was much more catastrophic than even the K114Q variant. At 20 min the K139Q variant achieved only 30% inhibition of thrombin, whereas the recombinant native ATIII achieved this level of inhibition by about 1 min. In terms of time this is a 20-fold difference where the K114Q was about 10-fold slower.
Comparison of Recombinant Native ATIII and the K114Q ATIII Variant in a Complex Formation Assay-Although several factors are involved in the physiological inactivation of thrombin by ATIII in vivo, heparin availability and the time the reactants spend in proximity are two of the most important.
Because the systemic administration of heparin is routinely used to control thrombosis, it then follows that in vivo, heparin is a rate-limiting component. To test the effects of heparin availability under conditions where time is limited, which would also occur in vivo, we chose the K114Q variant, because of its poor heparin binding capability, to use in a short time frame linkage assay in the presence of various amounts of heparin.
To first verify that the linkage assay would accurately reflect the inhibition of thrombin by ATIII as seen in the Chromozym-Th assay, nonheparin-activated linkage formation was Recombinant native ATIII (f) and the K 114Q variant (q) were used to coat the wells of a 96-well plate at a concentration of 10 g/ml. Following blocking with 3% bovine serum albumin in Tris-buffered saline, the wells were incubated with 42 nM of 125 I-F-heparin and the indicated amounts of unlabeled F-heparin at 37°C for 1 h. The wells were then washed with PBS three times and solubilized with 10% SDS. The amount of 125 I-F-heparin bound was quantitated by ␥ counting. examined. 125 I-Thrombin (20 nM) was reacted with increasing amounts of recombinant native ATIII or the K114Q variant for 30 min at 37°C. The reactions were quenched by the addition of SDS-PAGE sample buffer, and the ATIII-thrombin complexes were resolved by SDS-PAGE on 10% polyacrylamide gels (Fig. 8). In both cases, increasing the ATIII concentration resulted in more complexes formed, and the relative amount of complexes observed were indistinguishable for recombinant native ATIII (Fig. 8, lanes 2-4) and the K114Q variant (Fig. 8,  lanes 5-7).
We next compared the two under conditions of limited time and heparin availability. Recombinant native ATIII and the K114Q variant, both at a concentration of 65 nM, were incubated with heparin at the indicated concentrations for 30 min at 37°C. 125 I-Thrombin (20 nM) was added, and the reactions were continued for an additional minute. The reactions were quenched and analyzed as described above. Recombinant native ATIII was activated by heparin in a dose-dependent manner, with a half-maximal activation occurring at a heparin concentration of 2 nM (Fig. 9A). This is in excellent agreement with the K d of 6 nM determined in the Scatchard analysis. In contrast the K114Q variant, displayed no activation under these conditions, even at a 10-fold higher heparin concentration (Fig. 9B). Thus, whereas the effect of the K114Q substitution was very apparent under conditions optimized to promote thrombin inhibition, the effect is even more dramatic under conditions where heparin availability and time are limited. The latter is more likely to reflect the in vivo conditions where these two factors play an important role. DISCUSSION In the present study we have investigated independently the role(s) of several positively charged lysine residues in and around the D-helix of ATIII including Lys 114 , Lys 125 , Lys 133 , Lys 136 , and Lys 139 by changing them to either glutamine or threonine (Lys 136 ). Although other studies have implied a role for Lys 114 , Lys 136 , and Lys 139 , none has done so under conditions where these were the sole residue chemically modified (12)(13)(14)(15)(16) or protected (2). In addition, no other study has carefully evaluated the role of these lysine residues by directly measuring heparin binding or by quantitating the effect of a mutation at this position on heparin cofactor activity. The remaining positively charged residues in this region include Arg 129 , Arg 132 , Lys 133 , and Arg 47 in the A-helix, all of which have been studied in detail in naturally occurring or genetically engineered variants. Arg 47 (7)(8)(9)(10) and Arg 129 (11) appear to be important in heparin binding and activation, whereas Arg 132 and Lys 133 appear to be involved in heparin binding (18). In order to understand the mechanism by which ATIII is heparinactivated, each of these residues in the D-helix and adjacent regions have been evaluated independently at a detailed level to fully understand their role(s).
In the present studies glutamine or threonine was substituted at each residue studied because they are uncharged and similar in size to Lysine but still have polar character. To determine if the individual substitutions altered the ability of the recombinant ATIII variants to inhibit thrombin, the second order rate constants for recombinant native ATIII and all of the variant proteins were determined in parallel. The second order rate constants were found to be very similar and are in good agreement with other published values (36,39). The similar levels of thrombin inhibition in the absence of heparin strongly suggests that the overall structure of the ATIII has not been severely altered by the substitutions. The effect of the substitutions on heparin activation was initially evaluated by measuring the time-dependent inactivation of thrombin in the presence and the absence of heparin. There was no effect seen for the K133Q variant, which was not expected from a previous study (18) (data not shown). However, the previous study only considered activation of ATIII by heparin toward factor Xa, not thrombin. In contrast, at a heparin concentration of 42 nM, which gave half-maximal activation of recombinant native ATIII, little or no activation of the K114Q or K139Q variants was observed. The K136T variant was much less affected. Although the K114Q and K139Q variants were slowed at least 10 -20-fold, the K136Q was only about 5-fold slower. As expected from other studies (17) the K125Q variant was only about 6-fold slower in its inhibition of thrombin. In all cases, when the concentration of heparin was raised, no additional activation was seen.
To more precisely determine the cause for the loss of heparin activation, direct binding studies were performed using 125 I-Fheparin. Recombinant native ATIII bound heparin specifically and saturably with a K d of 6 nM, which is in good agreement with published values (36,41). The K125Q variant had decreased heparin affinity with a K d of about 12 nM. Although this indicates that the residue is not as critical as was proposed earlier by another group using chemical modification (12,13), it does agree in scale with the recently published characterization of a Lys 125 M variant of ATIII produced in baby hamster kidney cells (18). In contrast, the K114Q and K139Q variants bound heparin so poorly that a K d could not be determined. The simplest interpretation of these data is that removal of the  (24), severely decreases binding and thus activation by the longer chains of heparin. Although some effect is seen in the K136Q variant, the data indicate that the residue is not absolutely critical.
Another group has identified a naturally occurring variant of ATIII with a Gln substation at position Arg 129 from a heterozygous individual with thrombotic disorder (11). Although direct heparin binding studies were not done, it was noted that the Arg 129 variant antithrombin no longer bound to heparin-Sepharose at a physiological salt concentration, and in addition, the degree of the loss of heparin activation was very similar to that observed in the present study. It is not immediately clear why an individual Lys or Arg residue that is part of a very extensive charge cluster should have such a strong impact on the binding of heparin. This cluster occurs on a face of the D-helix, and helical disruption is a possibility. However Lys 114 and Lys 139 are not actually part of the D-helix, and Arg 129 is at the end of the D-helix (24). In addition, there is no disruption of nonheparin-activated antithrombin activity, suggesting no major structural alteration in the molecule.
Olson and Shore have shown that upon heparin binding there is a shift in tryptophan fluorescence in ATIII, which is consistent with a conformational change in the protein (20). They found that the binding occurs in a two-step process, an initial weak interaction followed by a stabilization, which results in a 300-fold increase in heparin affinity and the conformational change. A possible explanation for the apparent ablation of heparin binding in the K114Q and K139Q variants may be that the initial binding of heparin is unaffected but that these lysine residues are required for transition to the stabilized state. If this were the case, then the equilibrium binding constant would shift from low nanomolar to micromolar. This would be consistent with our inability to obtain a binding constant employing a standard binding isotherm and Scatchard analysis. The results of the present study may help to resolve the differences in the template and pentasaccharide ATIII activation models. The pentasaccharide model suggests that the conformational change in ATIII brought about by heparin binding is all mediated through the pentasaccharide binding pocket in the D-helix, although it is not likely that ATIII actually encounters the pentasaccharide in vivo (19 -23).
Additional activation brought about by longer chain heparins is thought to be mediated through a template mechanism that immobilizes ATIII and thrombin in close proximity (25)(26)(27)(28)(29)(30). The present data demonstrate that longer chain heparins interact with residues outside of the pentasaccharide binding pocket, namely Lys 114 and Lys 139 , either to help facilitate and stabilize a full conformational change in ATIII caused by heparin or to position the heparin chain as a more efficient template. The demonstration that monovalent antibody fragments directed against residues 137-145 partially mimic the action of heparin would lend more support for the former than the latter interpretation (4), but this area is clearly open to further investigation. FIG. 9. Kinetics of 125 I-thrombin-ATIII complex formation as a function of heparin concentration. Recombinant native ATIII (panel A) and the K114Q variant (panel B) at 25 nM were preincubated with the indicated concentrations of heparin for 30 min at 37°C. 125 I-Thrombin (9.5 nM) was added, and the reactions were allowed to proceed for 1 min. The reactions were stopped by an equal volume of SDS sample buffer and resolved by SDS-PAGE on 10% polyacrylamide gels. the autoradiograms were prepared as described in the legend to Fig. 7. The positions of free 125 I-thrombin and the 125 I-thrombin-ATIII complexes are indicated by the arrows.