Role of Thrombin Exosites in Inhibition by Heparin Cofactor 11*

We determined the role of specific thrombin “exo- sites” in the mechanism of inhibition by the plasma serine proteinase inhibitors heparin cofactor I1 (HC) and antithrombin (AT) in the absence and presence of a glycosaminoglycan by comparing the inhibition of a-thrombin to t- and yT-thrombin (produced by partial proteolysis of a-thrombin by elastase and trypsin, re-spectively). All of the thrombin derivatives were inhib- ited in a similar manner by AT, either in the absence or presence of heparin, which confirmed the integrity of both heparin binding abilities and serpin reactivities of C- and yT-thrombin compared to a-thrombin. Antithrombin activities of HC in the absence of a glycosa- minoglycan with a-, t-, and yT-thrombin were similar with rate constants of 3.5,2.4, and 1.2 X lo4 M” min”, respectively. Interestingly, in the presence of glycosaminoglycans the maximal inhibition rate constants by HC with heparin and dermatan sulfate, respectively, were as follows: 30.0 X lo7 and 60.5 X lo7 for a- thrombin, 14.6 X lo7 and 24.3 X lo7 for t-thrombin, and 0.017 X 10’ and 0.034 X lo7 M” min” for yT- thrombin. A hirudin carboxyl-terminal peptide, which binds to anion-binding exosite-I of a-thrombin, dra- matically reduced a-thrombin inhibition by HC in the presence of heparin but not in its absence. We analyzed our results in relation to the recently determined x-ray structure of D-Phe-Pro-Arg-chloromethyl ketone- a-thrombin (Bode, S. and Hofsteenge, J. (1989) EMBO 3467-3475). Our results suggest that the &loop region of anion-binding exosite-I in a-thrombin, which is not present in yT-thrombin, is essential for the rapid inhibition reaction by HC in the presence of a glycos- aminoglycan. Therefore, a-thrombin and its derivatives would be recognized and inhibited differently by HC and AT in the presence of a glycosaminoglycan. The three-dimensional structure of thrombin reveals The PPACK-a-thrombin structure was modeled using the SYBYL software package from Tripos Associates and drawings were made using the Tripos program NITRO on a Macin-tosh 11.

We determined the role of specific thrombin "exosites" in the mechanism of inhibition by the plasma serine proteinase inhibitors heparin cofactor I1 (HC) and antithrombin (AT) in the absence and presence of a glycosaminoglycan by comparing the inhibition of athrombin to t-and yT-thrombin (produced by partial proteolysis of a-thrombin by elastase and trypsin, respectively). All of the thrombin derivatives were inhibited in a similar manner by AT, either in the absence or presence of heparin, which confirmed the integrity of both heparin binding abilities and serpin reactivities of C-and yT-thrombin compared to a-thrombin. Antithrombin activities of HC in the absence of a glycosaminoglycan with a-, t-, and yT-thrombin were similar with rate constants of 3.5,2.4, and 1.2 X lo4 M" min", respectively. Interestingly, in the presence of glycosaminoglycans the maximal inhibition rate constants by HC with heparin and dermatan sulfate, respectively, were as follows: 30.0 X lo7 and 60.5 X lo7 for athrombin, 14.6 X lo7 and 24.3 X lo7 for t-thrombin, and 0.017 X 10' and 0.034 X lo7 M" min" for yTthrombin. A hirudin carboxyl-terminal peptide, which binds to anion-binding exosite-I of a-thrombin, dramatically reduced a-thrombin inhibition by HC in the presence of heparin but not in its absence. We analyzed our results in relation to the recently determined xray structure of D-Phe-Pro-Arg-chloromethyl ketonea-thrombin (Bode, W., Mayr, I., Baumann, U., Huber, R., Stone, S. R., and Hofsteenge, J. (1989) EMBO J. 8 , 3467-3475). Our results suggest that the &loop region of anion-binding exosite-I in a-thrombin, which is not present in yT-thrombin, is essential for the rapid inhibition reaction by HC in the presence of a glycosaminoglycan. Therefore, a-thrombin and its derivatives would be recognized and inhibited differently by HC and AT in the presence of a glycosaminoglycan. a-Thrombin is a trypsin-like serine proteinase important for hemostasis (1, 2 and references cited therein). It hydrolyzes fibrinogen to fibrin in the final step of blood coagulation and interacts with other substrates, receptors, and inhibitors (1,2). The three-dimensional structure of thrombin reveals * This work was supported in part by Research Grant HL-32656 from the National Institutes of Health and a Grant-in-Aid from the North Carolina Affiliate of the American Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The specificity of HC appears to be largely limited to thrombin and chymotrypsin (20)(21)(22) while AT inhibits most of the proteinases in the intrinsic coagulation pathway (14). An intriguing difference in the two serpins is the reactive site: Leu-Ser in HC (12,23) and the expected Arg-Ser in AT (14). The occurrence of the reactive site Leu in HC raises speculation that residues far separated in primary structure may be crucial for recognition between thrombin and HC. Recent experimental evidence using HC mutants supports a hypothesis that an additional acidic region of HC (far removed in primary structure from the reactive site) is partially responsible for the accelerated thrombin inhibition reaction in the presence of a glycosaminoglycan (24)(25)(26).
In this study we explored whether or not different thrombin The sequence numbering of thrombin is the numbering system based on the topological equivalences with chymotrypsinogen as described by Bode et al. (3), with insertions indicated by use of letters (for example, Arg-77A, Ala-l49A, and Lys-149E).

3613
This is an Open Access article under the CC BY license. This is an Open Access article under the CC BY license. exosites, which are altered or missing in three of the four derivatives (human a-, t-, and yT-thrombin and bovine athrombin (with occurrence of Glu for Lys-l49E)), play a role in the mechanism of glycosaminoglycan-enhanced inhibition by HC or AT. Specifically, we were interested in the exosite termed "anion-binding exosite-I" (an electropositive surface formed by Arg-35 to Glu-39, Arg-67 to Arg-77A, and Lys-149E) which is critical for a-thrombin interactions with fibrinogen, hirudin, and thrombomodulin (1-5, 8, 27, 28). Whereas there was no requirement for these a-thrombin exosites during the AT/heparin inhibition reaction, we found that the p-loop region of anion-binding exosite-I was essential for rapid inhibition by HC in the presence of glycosaminoglycans.
All of the thrombin derivatives were greater than 95% active by active site titration. Fibrinogen clotting activity of the thrombin derivatives was performed as detailed previously using a V,,, kinetic microplate reader in the kinetic mode (Molecular Devices, Menlo Park, CA) (30). The t -and 7,-thrombin derivatives had 53% and 1% of the fibrinogen clotting activity, respectively, compared to Hua-thrombin. Na-DodS0,-polyacrylamide gel electrophoresis was performed in 15% slab gels (34) without chemical reduction. NaDodS04-polyacrylamide gel electrophoresis of the thrombin derivatives revealed three major components in ?,-thrombin and only two components in c-thrombin as reported previously (6,8).
Thrombin Inhibition Assays-Antithrombin inhibition assays in the absence or presence of glycosaminoglycans with HC, AT, and the thrombin derivatives were performed as described (22,35). In the absence of a glycosaminoglycan, the thrombin derivatives (10 nM) were incubated with HC (1 pM) or AT (500 nM) in a final volume of 100 pl (with 100 pg/ml polybrene present). In the presence of a glycosaminoglycan, the thrombin derivatives (0.5 nM) and HC (20 nM) or AT (5 nM) were incubated in a final volume of 100 pl with various concentrations of heparin or dermatan sulfate. For both assay procedures, Chromozym T H with polybrene was added a t various time intervals. Chromozym T H hydrolysis was measured in a Vmx kinetic microplate reader and compared to thrombin controls (same reaction components minus serpin). All thrombin inhibition experiments were performed a t least three times and the results averaged.
Hua-thrombin inhibition by HC alone in the presence of hiruwas performed in the HEPES-buffered saline with 0.1 mg/ml bovine serum albumin as detailed (30). Briefly, Hua-thrombin (5 nM) was incubated with HC (500 nM), with and without 750 nM hirumozym TH. This experiment was performed three times and the results averaged. Hua-thrombin inhibition by HC/heparin in the presence of hirudin'"64 or the control peptide was performed by including the active site-specific inhibitor DAPA using the reaction conditions as detailed previously (36). Hua-thrombin and DAPA were din63-64 din63-64 , and remaining thrombin activity was measured with Chromixed to final concentrations of 50 and 450 nM, respectively, in the absense and presence of 6.25 p~ h i r~d i n '~-~~ or the control peptide, HC were determined as described under "Experimental Procedures." and a solution of HC/heparin (stock of 8 p~ HC and 10 pg/ml heparin) was added to the DAPA-thrombin solution. The reaction was monitored by following the loss of the DAPA-thrombin fluorescence as described previously (36). This experiment was performed eight times and the results averaged. Molecular Modeling-The three-dimensional coordinates of D-Phe-Pro-Arg-chloromethyl ketone (PPACK)-a-thrombin (3) were a kind gift of Dr. Wolfram Bode, Max-Planck-Institut fur Biochemie,.Martinsried, Germany. The PPACK-a-thrombin structure was modeled using the SYBYL software package from Tripos Associates and drawings were made using the Tripos program NITRO on a Macintosh 11.

RESULTS
Inhibition by Antithrombin-We measured the inhibition of the thrombin derivatives by AT both in the absence and presence of heparin. In the absence of heparin, we found that e-and Bva-thrombin were inhibited by AT at about the same rate (-1.5-fold less) as Hua-thrombin, while the inhibition of ydhrombin was reduced by 6.3-fold compared to Huathrombin ( Table I). All of the thrombin derivatives showed a typical bell-shaped inhibition rate curve with AT as a function of heparin concentration, with maximal inhibition rate constants of 11.5, 10.9, 4.66, and 13.8 X 10' M" min" for Hua-, e , YT-, and Bva-thrombin, respectively (data not shown). Therefore, AT inhibition of Hua-thrombin was 1.4-fold slower than that for Bva-thrombin at the optimal heparin concentration, while the rates of t-and yT-thrombin inhibition were reduced by 1.1-and 2.5-fold, respectively, compared to Hua-thrombin.
These data agree well with previously reported values (9,28,36) and the results indicate that the thrombin derivatives are inhibited similarly by AT either in the absence or the presence of heparin. Thus, it seems that neither of the regions proteolyzed in e-or ?,-thrombin nor the substitution of Glu for Lys-149E in Bva-thrombin are of major importance for AT recognition or for heparin binding in a-thrombin.
Inhibition by Heparin Cofactor 11-We examined the inhibition of the thrombin derivatives by HC in the absence and presence of the glycosaminoglycans heparin and dermatan sulfate. In the absence of a glycosaminoglycan, we found that inhibition of e-and ?,-thrombin by HC was slightly reduced by 1.5-and 2.9-fold, respectively, compared to Hua-thrombin (Table I). HC inhibition of Bva-thrombin was 3.8-fold slower than that of Hua-thrombin (Table I). These results indicate that recognition of the thrombin derivatives by HC in the absence of a glycosaminoglycan does not depend on exosite domain(s) missing or altered in these thrombins.
As shown in Fig. 1, all of the thrombin derivatives exhibited typical bell-shape inhibition rate curves by HC as a function of glycosaminoglycan concentration. The maximal HC inhibition rate constants with heparin and dermatan sulfate were 30.0 X lo7 and 60.5 X lo7 for Hua-thrombin, 14.6 X lo7 and 24.3 X lo7 for e-thrombin, and 0.017 X lo7 and 0.034 x lo7 M" min" for ?,-thrombin, respectively (Fig. 1). The maximal inhibition rate constant of Bva-thrombin by HC in the presence of heparin was 8.40 x lo7 M" min" (data not included).
Compared to Hua-thrombin, HC inhibition rates of c-and Rate constants for inhibition of thrombin derivatives by AT and Bva-thrombin in the presence of optimal concentrations of glycosaminoglycans were reduced by 2.3-and 3.6-fold, respectively. rT-Thrombin inhibition by HC in the presence of heparin and dermatan sulfate was reduced by 1765-and 1800fold, respectively, compared to Hua-thrombin. The differences found in the maximal inhibition rates between Hua-, e-, and Bva-thrombin derivatives in the presence of glycosaminoglycans are consistent with the values determined for inhibition by HC in the absence of a glycosaminoglycan. In contrast, the values found for yT-thrombin inhibition by HC in the presence of a glycosaminoglycan are greatly reduced compared to Hua-thrombin suggesting that an exosite absent or altered in 7,-thrombin is necessary for rapid proteinase inhibition. Since t-and Bva-thrombin (both altered in the autolysis loop) have essentially normal glycosaminoglycanenhanced inhibition by HC, these results suggest that the @loop of anion-binding exosite-I of a-thrombin is involved in a n additional interaction with HC when glycosaminoglycan is present.
To further investigate the role of anion-binding exosite-I in a-thrombin, we determined the rate of inhibition by HC (with and without heparin) in the presence of a hirudin carboxyl-terminal peptide (residues 53-64). H i r~d i n~~-~~ binds t o anion-binding exosite-I toblock fibrinogen recognition without a significant detrimental effect on the active site of a-thrombin (4,5,30,(37)(38)(39)(40). Furthermore, we and others have reported that h i r~d i n '~-~~ does not adversely influence AT inhibition of a-thrombin either in the absence or presence of heparin (30,40). In the absence of heparin, the rate of Huathrombin inhibition by HC in the absence and presence of h i r~d i n '~" j~ (150-fold molar excess of peptide to thrombin) was the same, 3.8 f 0.1 X lo4 M" min". However, we found that hirudinS3"j4 reduced the rate of Hua-thrombin inhibition by HC/heparin, as shown by the inhibition rates of 1.32 f 0.13 x lo6 in the absence of peptide and 0.78 f 0.11 X lo6 M" min" in the presence of a 125-fold molar excess of peptide to thrombin (Fig. 2). The control peptide had no effect on Huathrombin inhibition by HC/heparin under the same conditions. Thus, in the presence of h i r~d i n '~-~~ at this concentration, Hua-thrombin inhibition by HC/heparin is reduced by more than 40%. This result implies further that anion-binding exosite-I of a-thrombin is necessary for rapid inhibition by HC in the presence of a glycosaminoglycan. . Control experiments verified that hirudin""j4 (and the control peptide) had no effect on the fluorescence of thrombin-DAPA, and did not itself react with DAPA.

DISCUSSION
This study was undertaken to determine whether specific exosites of a-thrombin participate in recognition of HC and AT in either the absence or presence of a glycosaminoglycan. a-Thrombin derivatives have been previously used to probe the role of secondary binding sites during the interaction with the physiological substrate fibrinogen, the leech antithrombin protein hirudin, and the cell-surface receptor thrombomodulin (1,2,8,28,41). Heparin-binding properties of various athrombin derivatives have also been investigated (9,36,42). The usefulness of these thrombin derivatives is enhanced due to the recently determined crystal structures of PPACKthrombin and thrombin-hirudin complex (3)(4)(5). A three-dimensional view of a-thrombin is shown in Fig. 3. There is now evidence to describe two anion-binding exosites in athrombin (2). Anion-binding exosite-I (see Introduction) is important for interaction with fibrinogen, the carboxyl terminus of hirudin, and thrombomodulin (1,2). "Anion-binding exosite-11" is an electropositive surface comprised of Lys-169, Arg-175, Arg-233, Lys-236, and Lys-240 and is the putative heparin-binding site (5, 36).
Our results demonstrate that thrombin recognition of AT and HC in the absence of a glycosaminoglycan is similar for all of the thrombin derivatives. The small differences in inhibition of the thrombin derivatives by AT and HC indicate that neither the autolysis loop nor the @-loop are essential for proteinase interaction with serpin alone. The AT/heparin reaction is also not dramatically altered when the various thrombins are compared. This further suggests the importance of anion-binding exosite-11, present in all of these thrombin derivatives, during formation of the heparin bridge between AT and thrombin (Fig. 3).
An exciting finding is the difference in inhibition of the thrombin derivatives by HC in the presence of a glycosaminoglycan. Similar to AT/heparin, inhibition rates of Hua-, t -, and Bva-thrombin by HC/glycosaminoglycan are quite comparable. This indicates that the autolysis loop and Lys-149E of anion-binding exosite-I are not critical for rapid thrombin inhibition by HC in the presence of a glycosaminoglycan. However, yT-thrombin inhibition by HC/glycosaminoglycan is greatly reduced. This large difference in inhi-  (36). The thrombin derivatives had the following alterations from a-thrombin: c-thrombin is cleaved a t Ala-l49A, YT-thrOmbin is cleaved a t Lys-149E and the @-loop is not present, and Bvathrombin has a substitution of Glu for Lys-149E. bition of yT-thrombin cannot be explained by an inactive heparin-binding site, as shown by the typical heparin dependence of the inhibition rate curve by AT and by other experiments (heparin-Sepharose binding, data not shown). Additionally, which interacts predominantly with the p-loop of anion-binding exosite-I (4,5), reduces the rate of athrombin inhibition by HC/heparin but not by HC alone. Our results suggest that the p-loop region of anion-binding exosite-I in a-thrombin, which is missing in yT-thrombin, participates in the recognition of HC when glycosaminoglycan is bound to anion-binding exosite-11. Inspection of the thrombin structure implies that a homologous serine proteinase without a region similar to the p-loop might not be rapidly inhibited by HC in the presence of a glycosaminoglycan (Fig. 3). Consistent with this hypothesis, we have shown previously that chymotrypsin inhibition by HC is not enhanced by glycosaminoglycans (21). The "calcium loop" of chymotrypsin, which is analogous to the p-loop of a-thrombin, is not a highly electropositive exposed region (3).
Our results complement the recent work of Ragg et al. (24,25) and Van Deerlin and Tollefsen (26) using a series of recombinant HC molecules. Their results showed that an HC acidic region (residues 56-75) is required for the rapid inhibition of thrombin by HC/glycosaminoglycan and they suggest that the acidic region binds the HC glycosaminoglycanbinding site (residues 165-195) in the absence of a glycosaminoglycan (24-26). The HC acidic domain is presumably displaced when a glycosaminoglycan binds to the HC glycosaminoglycan-binding site (24-26). 3 Hortin et al. (43) have also We found previously that modification of lysyl residues in the glycosaminoglycan-binding site of HC (with the negatively charged shown that a synthetic HC peptide (residues 54-75) inhibits fibrinogen clotting activity of thrombin, and the peptide competes with h i r~d i n~"~~ for binding to thrombin. These results imply that the HC acidic domain interacts with an exposed basic region of a-thrombin (24-26,43).
Our collective results (24-26, 43) suggest a plausible mechanism for a-thrombin inhibition by HC in the presence of a glycosaminoglycan: (i) heparin/dermatan sulfate binds to the glycosaminoglycan-binding site of HC and anion-binding exosite-I1 of thrombin forming a bridge (or ternary complex) similar to that for AT/heparin/thrombin; and (ii) the displaced HC acidic domain interacts with the p-loop region of anion-binding exosite-I in thrombin, which facilitates rapid proteinase inhibition by HC. Therefore, thrombin inhibition by HC in the presence of a glycosaminoglycan is consistent with a "double-bridge" mechanism in which anion-binding exosite-I of a-thrombin binds to the acidic region of HC and anion-binding exosite-I1 (through the glycosaminoglycan) binds to the glycosaminoglycan-binding domain of HC.
a-Thrombin has an essential role in the processes of hemostasis and wound healing which is displayed through enzymatic and nonenzymatic action by the proteinase (1, 2). Proteolyzed forms of a-thrombin have been proposed to exist in uiuo and could participate in either physiological or pathological events (1, 2). Generation of a-thrombin derivatives with altered exosites (and changed biological properties) is possible through the proteolytic action of mast cell tryptase, neutrophil elastase, and cathepsin G (1,2). As has been shown here, these thrombin derivatives would react differently with HC and AT in the presence of a glycosaminoglycan. In uiuo targeting of thrombin inhibition either by HC/glycosaminoglycan or by AT/glycosaminoglycan could be regulated by the competency of a-thrombin exosites, particularly anion-binding exosite-I. 5. Rydel, T. J., Ravichandran, K. G., Tulinsky, A., Bode, W., Huber, R., Roitsch, C., and Fenton, J. W., I1 (1990) Science 249, 277-280 2361-2365 reagent pyridoxal 5"phosphate) results in increased Hucr-thrombin inhibition activity in the absence of a glycosaminoglycan (35). Thus, we measured inhibition of the Hua-thrombin derivatives with control and pyridoxal 5'-phosphate-modified HC. The inhibition rate constants (kohs = X min") for control and modified HC were 45 & 2 and 87 & 1 for Hun-thrombin, 34 f 4 and 68 & 1 for c-thrombin, and 19 k 2 and 20 & 5 for ?,-thrombin, respectively. The modified HC had increased antithrombin activity by 1.9-fold with Hun-and cthrombin compared to control HC but there was no change in the inhibition rate constant between control and modified HC with YTthrombin. This experiment implies that the modified glycosaminoglycan-binding site of HC has partially displaced the HC acidic domain, probably by charge repulsion, which slightly enhances thrombin inhibition in the absence of a glycosaminoglycan.