Heparin Cofactor I1 Is Regulated Allosterically and Not Primarily by ...

Vol. 269, No. 52, Issue of December 30, pp. 3274732751, 1994 Printed in U.S.A.

JOURNAL OF BIOISXICAL CHEMISTRY 0 1994 by The American Societyfor Biochemistry and Molecular Biology, Inc. THE

Heparin Cofactor I1 Is Regulated Allostericallyand Not Primarily by Template Effects STUDIES WITH MUTANT THROMBINS AND GLYCOSAMINOGLYCANS* (Received forpublication, July 26, 1994, and in revised form, October 18, 1994)

John P. SheehanSg, Douglas M. TollefsenSn, and J. Evan SadlerSII1 From the lwoward Hughes Medical Institute and the Departments of medicine and IBiochemistry & Molecular Biophysics, The Jewish Hospital of St. Louis, Washington University School of Medicine, St. Louis, Missouri 63110

elicits cellularresponsesthroughthe Besides its critical role in hemostasis, the serine pro- mostasis,thrombin wound tease thrombin also participates in wound healing, in- thrombin receptor that may contribute to inflammation, flammation, and atherosclerosis. Thrombinis inhibited healing, and atherosclerosis (1).Thrombin is inhibited by the by the serpins antithrombin and heparin cofactor I1 serpins antithrombin and heparin cofactor I1 (HCII)’ (2) in (HCII) in reactions that are accelerated markedly by reactions that are accelerated markedly by specific glycosamispecific glycosaminoglycans. Following vascular injury,noglycans (3,4). Following vascular injury, thrombin inhibition thrombin must be inhibited at both intravascular and by these serpins must take place both at intravascular and extravascular sites that impose differentconstraints on extravascular sites. The specific environment can modulate the the recognition of thrombinby these inhibitors. The pres- recognition of thrombin by theseinhibitors. Forexample, ent study examines the role of anion-binding exositeI1 thrombin that binds to fibrin clot or to extracellular matrix is of thrombin in the interaction with glycosaminoglycans relatively resistant to inhibition by antithrombin-heparin (5, and HCII. Acceleration of thrombin inhibition by ser6). The availability of specific sites on thrombin that interact pins in the presence of glycosaminoglycans is proposed with serpins and anticoagulant drugs such as heparin may to occur by a template mechanism, in which inhibitor and protease bind simultaneously to the same glycos- thus be governed by the local environment. Identification of aminoglycan chain, facilitating their interaction. Ac- these siteson thrombin will illuminate molecular mechanisms cording to the template model, disruption of protease of resistance to these inhibitors. This should facilitate the debinding to glycosaminoglycan should significantly re- velopment of new strategies for anticoagulant therapy. Thrombin contains two clusters of positively charged amino duce acceleration of the inhibition. Specific mutations in exosite I1 (RSSE, R245E, K248E,and K252E)disrupted acids located at opposite poles of the molecule that are potential thrombin bindingto both dermatan sulfate and heparin, binding sitesfor glycosaminoglycans, serpins and other ligands indicating that both glycosaminoglycans bindto a com- (7). Anion-binding exosite Iinteracts with amino acid residues mon site in exosite 11. The same mutations markedly of the substrate that are on the carboxyl side of the scissile decreased the rateconstant for thrombin inhibition by bond, and exosite I contains several insertion loops that are antithrombin-heparin (up to 100-fold)but had little ef- important for the recognition of substrates, cofactors, and infect on the rate constant for thrombin inhibition by hibitors (7). Anion-binding exosite 11, which is located at the HCII-heparin (7-fold maximal reduction) and no effect opposite pole of the molecule, contains a functional heparinon the rate constant for thrombin inhibition by HCII- binding site (8-10). The serpins that inhibit thrombin, HCII, dermatan sulfate. Theseresults are incompatible witha and antithrombin have similarpredicted structures except for template model for thrombin inhibition byHCII and the unique amino-terminalextension of HCII (11).This extendermatan sulfate. In the presence of glycosaminoglycan, sion contains a sequence rich innegatively charged amino acids HCII and antithrombin interact with opposing throm(acidic domain) that is similar in composition to the carboxyl bin exosites and use distinct mechanisms of glycosamiterminus of hirudin, a potent thrombin inhibitor (12). Peptide noglycan catalysis. Antithrombin employs a template competition studies suggest that this acidic domain binds to mechanism that requires heparin to interact with thrombin exosite 11,whereas HCII employsan allosteric exosite I of thrombin (13). This interaction contributes to the mechanism that requires thrombinexosite I but is relative specificity of HCII for thrombin (14, 15). In contrast, largely independent of exosite 11. These findings have antithrombin does not interact significantly with exosite I of potential implications for glycosaminoglycan therapy thrombin (16, 17). In the presence of glycosaminoglycan, antithrombinand and for the respective physiologic roles ofHCII and HCII have been proposed to inhibit thrombin by a template antithrombin. mechanism, in which both the inhibitor and thrombinbind to the same glycosaminoglycan chain (18). This mechanism apThe serine protease thrombin must be regulated t o control pears convincing for antithrombin, as mutations in exosite I1 the response to vascularinjury. Besides its critical role in he- that decrease the affinity of thrombin for heparin also markedly decrease the ability of heparin to accelerate the inhibition *This research was supported in part by National Institutes of of thrombin by antithrombin (9, 10). An allosteric model also Health Grants HL14147 and 1K08 HL 02923-01 (to J. P. S.), an Ameri- has been proposed t o explain the inhibition of thrombin by can Society of Hematology ScholarAward (to J.P. S.), and the Monsanto Company (to D. M. T.). The costs of publication of this article were HCII (12, 19). In this model, glycosaminoglycan binds to HCII and displaces the amino-terminal acidic domain from an indefrayed in part by the payment of page charges. This article must therefore be herebymarked “advertisement” in accordancewith 18 tramolecular binding site. The acidic domain then binds t o U.S.C. Section 1734 solely to indicate this fact. anion-binding exosite I of thrombin and accelerates the fonna0 To whom correspondence should be addressed: Dept. of Medicine, WashingtonUniversity School of Medicine, 660 S. EuclidAve., Box 8022, St. Louis, MO 63110. Tel.: 314-362-9029;Fax: 314-454-3012. The abbreviation used is: HCII, heparin cofactor 11.

32747

32748

Allosteric Regulation of HCII

tion of a stable inhibitor-protease complex. Deletions in the amino-terminal acidic domain of HCII markedly decrease the ability of glycosaminoglycans to accelerate thrombin inhibition (12). Likewise, mutations or autocatalytic cleavages in anionbinding exosite I of thrombin decrease the rate of thrombin inhibition by HCII plus glycosaminoglycan (16, 17). Features of both the allosteric and templatemodels have been combined in a proposed "double bridge" mechanism in which exosite I of thrombin binds to the acidic domain of HCII and exosite I1 of thrombin binds to the HCII.glycosaminog1ycan complex through theglycosaminoglycan template (12, 17). However, the relative contributions of thrombin exosites I and I1 to the reaction with HCII have not been defined. We have employed mutagenesis toexamine the role of anionof thrombin with dermatan binding exosite I1 in the interaction sulfate and HCII. Single amino acid substitutions of glutamic acid for lysine or arginine were constructed within exosite I (K52E, R68E, and K154E) or exosite I1 (R89E, K174E, R245E, K248E, and K252E) of thrombin to assess the contribution of these basic surface residues to glycosaminoglycan binding and the interaction with HCII. The results demonstrate thatheparin and dermatan sulfatebind to a common site on thrombin and thatHCII and antithrombin employ fundamentally different mechanisms of glycosaminoglycan catalysis.

mg/ml bovine serum albumin (Sigma) and 0.1% (w/v) PEG-8000 (polyethylene glycol, average M , = 8000). A 1.0-ml sample containing approximately 5 pg of thrombin was applied to the column at 0.5 mumin. The column was developed at 0.5 ml/min with 5 ml of starting buffer followed by a linear 20-ml gradient of 0.05-1.0 M NaCl in starting buffer. Fractions of 0.5 ml were collected. A 50-pl sampleof each fraction was placed in a 96-well plate (Corning). Thrombin activity was detectedby adding 100 plof 500 p~ S-2238 in0.15 M NaCl, 0.02 M Tris-HC1, pH 7.4, 0.1% PEG-8000 (TSPEG buffer) and determining the rate of change in absorbance at 405 nm for 2 min on a V, Reader (Molecular Devices). Thrombin Inhibition by HCII-Assays without glycosaminoglycan were performed at room temperature in TSPEGbuffer under pseudofirst-order conditions (150 nM HCII and 2-5 n~ thrombin). Samples were removed at appropriate intervals for quantitation of residual thrombin by determining the rate of hydrolysis of S-2238. Residual thrombin activity uersus time was plotted and values for the first-order rate constant ( k ' ) were determined by fitting the data by nonlinear least-squares regression to the equation:

E J E , = e -k't

(Eq. 1)

where E, represents thrombin activityat time= 0, E, represents thrombin activity at time = t, and k' represents the pseudo-first-order rate constant. The correlation coefficient ( Rvalue) of the fit was greater than 0.96 in all cases. The second-order rate constant(K,) of inhibition was then obtained by dividing the pseudo-first-order rate constant by the inhibitor concentration [I] (16). For inhibition in the presence of heparin or dermatan sulfate, thrombins withk , 5 1 x lo7M - ~min" were assayed asabove. Thrombins with EXPERIMENTAL PROCEDURES k , > 1x lo7M - ~min" were assayedby monitoring progress curvesin the Materials-Human plasma HCII was purified by heparin-agarose presence of 75-150 p~ S-2238, 150 nM HCII, and 0.5-3 nM thrombin. chromatographyasdescribedpreviously (20). Humanplasma pro- The hydrolysisof S-2238 t o yield p-nitroaniline product(P) was monithrombin was obtainedfrom George J. Broze (Washington University). tored continuously at 405 nm over time (t), and the data were fit by Unfractionated porcine intestinal heparin was obtained from Elkin- nonlinear regression (Marquardt algorithm)t o the equation: Sinn (Cherry Hills, NJ). Bovine mucosal dermatan sulfate was obtained [PI = u,t + (u,, - u3)/k'(1 - e - h ' t ) (Eq. 2) from Dr. Lennart Roden (University of Alabama) or from Sigma. Dermatan sulfate was treated with nitrous acid to remove contaminating to obtain estimatesof initial velocity (uo), steady statevelocity ( u 8 ) ,and heparin (21). Chromogenic substrate S-2238 (H-D-Phe-pipecolyl-Arg-p- the apparent first-order rate constant (K'). No significant substrate nitroanilide) was from Kabi-Vitrum (Stockholm). depletion or product inhibition was noted over the substrate ranges and Plasmid Constructs-Plasmid pCMVPT containing a human pro- time interval monitored. In the presence of optimal concentrations of thrombin cDNA inserted in thepCMV vector and the mutant plasmids glycosaminoglycan, the apparent first-order rate constant( k ' ) was linpK52E, pR68E,pR89E, pK154E, pK174E, pR245E, pK248E, and early related to the HCII concentration, and the initial velocity (u,) was pK252E were described previously (9, 22,23).' independent of HCII concentration, consistent with slow binding inhiRecombinant Thrombins-Transfection of CV-1cell lines (WT, K52E, bition mechanism A (27) (data not shown). Thisis similar to previous R68E, K154E, K174E, and K252E) orBHK (R89E, R245E, and K248E) results obtained for the inhibition of thrombin by antithrombin and cell lines and purificationof the recombinant thrombins were describedprotease nexin I (28). Under these conditions, the value of k , is related previously (9, 22). Thrombins were quantitated by absorbance at 280 to k' by the equation: nm using an extinction coefficient ( E ~ , of~ 1.83 ~ ) (24). Electrophoresis on SDS/10% polyacrylamide gel (25) and silver staining (Integrated Sepak , = k'(1 - u J u o ) ( l + [Sl/K,)/rIl (Eq. 3) ration Systems, Natick MA) revealed a single band for each recombiwhere [SI is the concentration of substrate (S-22381, K, is the Michaelis nant thrombin. constant for substrate, and [I] is the concentrationof HCII (28). Dermatan Sulfate Affinity Chromatography-Dermatan sulfate (3040 mg) was partially N-deacetylated by treatment at 95 "C for 15 min in 190 ml of hydrazine containing380 mg of hydrazine sulfate. The RESULTS reaction mixture then was neutralized with glacial acetic acid, dialyzed Characterization of Recombinant Thrombins-All the recomagainst water, and dried by rotary evaporation. The dermatan sulfate product wascoupled to 60 of g CNBr-activated Sepharose4B (Sigma) in binant thrombins had similarMichaelis constants (K,) for the chromogenic substrate S-2238 between 1.7 and 2.8 PM (9, 22, 150 ml of 0.5 M NaCl, 0.1 M sodium bicarbonatebuffer, pH 8.3,for 2.5 h at room temperature. The beads were washed with coupling buffer and 23). Mutations in or nearexosite I (K52E, R68E, and K154E)' incubated with 0.1M Tris-HC1, pH 8.0, toblock excess reactive groups. had almost no effect on the rate of thrombin inhibition by About 240 mg of dermatan sulfate was bound to the beads, as deter- antithrombin in the presence or absence of heparin, but had mined by subtracting unbound dermatan sulfate in the pooled wash variable effects on the clotting of fibrinogen (9,221. In general, fractions from the starting material. Unbound dermatan sulfate was mutations in exosite I1 did not affect the rate of inhibition by quantitated by the carbazole assay for uronic acid (26). Tris-agarose was prepared to serveas a control for nonspecific bind- antithrombin without heparin, but reduced the rate of inhibiof the ing. CNBr-activated Sepharose 4B was incubated as above with cou- tion by antithrombinwithheparin.Themagnitude pling buffer but without dermatan sulfate, washed, and subsequently change in affinity correlated with the amount of reduction in incubated with 0.1 M Tris-HC1, pH 8.0. the heparin-dependent inhibition rate (9). Dermatan sulfate-agarose or Tris-agarose columns (1.0 ml volume) Affinity of Recombinant Thrombins for Dermatan Sulfatewere equilibratedi n 0.05 M NaCI, 0.02 M Tris-HC1, pH 7.4, containing 10

Agarose-The dermatan sulfate-binding site on thrombin was Plasmid names indicate thewild type aminoacid, the residue number localized by assessing theeffect of single amino acid substitumutated, and the replacement amino acid. Amino acid residues are num- tions on the affinity of thrombin for dermatan sulfate-agarose. bered from the first residue of the human thrombin B chain. The relationPlasma-derived and wild type recombinant thrombin bound to ship of selected thrombin B chain residue numbers to the alternative dermatan sulfate-agarose and were eluted at -0.24 M NaCl chymotrypsin numbering system (in parentheses) isas follows: K52(6oD, R68(73), R89(93), K154(149e), K174(169), R245(233), K248(236), and (Fig. lA)Chromatography of plasma-derived thrombin on Trisagarose demonstrated no significant binding under these conK252(240)(7).

Allosteric Regulation of HCII

A

32749

J- 1.0

0.80 0.60

o

0

IO

io

20 30 fraction #

40

io

40

io

fraction #

0.8

E

r^

-

0.6-

; z

?

0.40

0.42

0.20

0.2

0.0

0

50

fraction #

io

o:o

10

20

io

fraction #

io

io

FIG.1. A, chromatography of thrombin on dermatan sulfate-agarose. Plasma-derived thrombin (-5 pg) was applied to a 1.0-ml column of column was developed as described under “Experimental Procedures.” Thrombin activity or dermatan sulfate-agarose (O), and the Tris-agarose (0) was detected by the rate of cleavage of the substrate S-2238 (mOD/min). Plasma-derived thrombin did not bind significantly to the Tris-agarose column (flow through) but eluted from dermatan sulfate-agarose at approximately 0.24 M NaCl. Wild type recombinant thrombin eluted from dermatan sulfate-agarose (0) in the sameposition as plasma-derived thrombin. B , chromatography of anion-binding exosite I mutant thrombins on dermatan sulfate-agarose, Chromatographs for thrombin K52E (O), thrombin R68E (O), and thrombin K154E (0)are shown. C and D, chromatography of anion-binding exosite I1 mutant thrombinson dermatan sulfate-agarose. C, chromatographs for thrombin K174E (O),thrombin R245E (O), and thrombin K252E (0). D , chromatographs for thrombin K248E (0) and thrombin R89E (0).

ditions, indicating that binding to dermatan sulfate-agarose TABLE I Rate constants for inhibitionof recombinant thrombins represented the interactionof thrombin with dermatan sulfate Rate constants(k,)for the inhibition by HCII (150 m)of recombinant (Fig. lA). The exosite I mutants K52E, R68E, and K154E (Fig. lB)and wild type thrombin and mutant thrombins(0.5-5nM) were determined as described under “Experimental Procedures.” Whenpresent, theconthe exosite I1 mutant K174E (Fig. 1C) eluted at sodium chlo- centration of heparin was 10 unitdml, and theconcentration of dermaride concentrations similar to that for wild type thrombin. tan sulfate was500 pg/ml. Values are expressed as the mean f S.D.of Thrombin K174E alsoexhibited a second, slightly delayed three or more determinations. peak. Upon rechromatography under more dilute conditions, Rate constant (k,) of inhibition however, the material from this second peak eluted in thepoThrombin HCII HCII + dermatan HCII + he ann (x 109 sulfate (x lo6) ( x 104 sition of the first peak (data not shown). Thrombin K174E therefore may undergo reversible concentration-dependent agM P min” gregation. ThrombinK174E did not bind to5s-agarose (data Wild type 41 2 4 683 f 62 678 2 14 not shown). Thus, mutations located in three different insertion loops in anion-bindingexosite I (K52E, R68E,and K154E), Exosite I mutations 340 f 20 475 f 46 K154E 38 f 2 and on the periphery of anion-binding exosite I1 (K174E), did K52E 63 f69 421 10 5 13 not significantly affect thrombin affinity for dermatan sulfate. R68E 8 -c 0.2 1.1 f 0.1 1.8 f 0.4 In contrast, thrombins R89E, R245E, K248E, and K252E did not bind significantly to dermatan sulfate-agarose (Fig. 1, C Exosite I1 mutations 45 f 4 792 f 128 760 * 108 K174E and D).These results are qualitatively similar to those obK252E 39 f 1 683 106 419 * 81 tained for chromatography on heparin-agarose (9): mutations R245E 43 f 4 759 f 99 228 * 81 that reduced binding to dermatan sulfate also reduced binding R248E 42 f 1 695 f 115 119 8 R89E 87 f 1 776 f 100 98 t 24 to heparin, and mutations that did not affect binding to dermatan sulfate also did not affect binding to heparin. This correlation suggests that heparin and dermatan sulfatebind to a similar site within anion-bindingexosite 11. glycosaminoglycan, HCII interacts weakly with the so-called Inhibition of Recombinant Thrombins by HCII without “60 insertion loop” near theactive site of thrombin (containing Glycosaminoglycan-Rate constants for the inhibition of mutation K52E) and the“70-80 loop” in anion-bindingexosite thrombin by HCII were determined under pseudo-first-order I (containing mutation R68E), but not with the“149 insertion conditions. As reported previously (161, the mutations R68E loop” on the “south rim” of the active site cleft (containing and K52E in exosite I caused reductions in theobserved inhi- mutation K154E) (7). In contrast, mutations inanion-binding bition rate constants of 5- and 10-fold, respectively (Table I). exosite I1 of thrombin did not significantly decrease the rate Thrombin K154E was inhibited at a rate constant similar to constant for inhibition by HCII (Table I). The largesteffect was that for wild type thrombin(Table I), as reported previously for a 2-fold increase in the rate constant for inhibition caused by the mutation K154A (16). These results suggest that, without the mutation R89E, which is relatively close to the active site. _+

32750

.‘e E

Allosteric Regulation of HCII R89E, only -7-fold less than wild-type recombinant thrombin (Table I). In contrast, the same mutation reduced the rateconstant for inhibition by antithrombin plus heparin by -100-fold (9).These results suggest that theinteraction of exosite I1with heparin (the template mechanism) is significantly less important to the inhibition of thrombin by HCII. DISCUSSION

Glycosaminoglycan binding to thrombin can promote its interaction with thrombomodulin, antithrombin, factor XI, and possibly with extracellular matrix. Understanding of these interactions is critical to rational therapeutic intervention in he1 10 100 1000 mostatic disorders. The heparin-binding site hasbeen localized [DS] ugh1 to anion binding exosite I1 by chemical modification and muFIG.2. Effect of dermatan sulfate concentration on the rate tagenesis of thrombin (8-10). The dermatan sulfate-binding constant of thrombin inhibition by HCII. Wild type thrombin (0) site had been localized tentatively to a different site. A synand thrombin R89E (0)are shown for comparison. Inhibition assays thetic 13-amino-acid peptide derived from the sequence of the were performed as described under “Experimental Procedures.” 60 insertion loop of thrombin was reported to inhibit partially the binding of thrombin t o dermatan sulfate within the extraThese results suggest that HCII does not interact significantly cellular matrix (61, suggesting that dermatan sulfate might with exosite I1 of thrombin without glycosaminoglycan.In this bind near the active site or in anion-binding exosite I. The respect, HCII appears to be similar to antithrombin (9). mechanism of inhibition by this peptide was not determined. Inhibition of Recombinant Thrombins by HCII Plus DermaOur results do not supporta major interaction of dermatan sultan Sulfate-As reported previously, the mutation R68E de- fate with the 60 insertion loop of thrombin. Specific mutations I1(RSSE, R245E, K248E, and K252E) creased the rateconstant for inhibition by HCII and dermatan within anion-binding exosite sulfate -600-fold, and other mutations in exosite I had more abolished the binding of thrombin to dermatan sulfate-agarose. modest inhibitory effects (TableI) (16). The stimulatory activi- Similar mutations located in the60 insertion loop (K52E) or elseties of both heparin and dermatan sulfate were impaired t o where in exosite I (R68E and K154E),or on the periphery of exosite approximately the same extent by any particular mutation in I1 (K174E), did not affect the binding of thrombin to dermatan thrombin exosite I. sulfate-agarose. These results effectively localize the dermatan In contrast to the effect of mutations in exosite I, mutations sulfate binding site to anion-binding exosite11. The effects of exosite I1 mutations on binding to dermatan in exosite I1 that abolished the binding of thrombin to dermatan sulfate did not affect the rate of thrombin inhibition by sulfate-agarose correlated with their previously reported efHCII plus dermatan sulfate. In particular, the second-order fects on the binding of thrombin to heparin-agarose (9). These data indicate that both heparin and dermatansulfate bind t o a rate constants for inhibition of thrombins R89E,R245E, K248E, and K252E were indistinguishable from the value ob- similar site within anion-binding exosite 11. Exosite I1 binds tained forwild-type recombinant thrombin (Table I), even heparin predominantly through electrostatic interactions (29) though none of these mutant thrombins bound t o dermatan and binds dermatan sulfate, suggesting that this site can bind I1 sulfate agarose (Fig. 1,C andD). These results suggest that the a broad range of glycosaminoglycans. Anion-binding exosite HCII.dermatan sulfate complex does not interact with throm- may interact with the chondroitin sulfate moiety of thrombobin through exosite 11,and that a templatemechanism does not modulin (30, 31) and with glycosaminoglycansin the extracelexplain the ability of dermatan sulfate to accelerate thrombin lular matrix (61, and it may mediate the glycosaminoglycancatalyzed activation of factor XI by thrombin (32). inhibition by HCII. Both antithrombin and HCII have been proposedt o employ a A templatemechanism still might apply if the concentration of dermatan sulfate employed in thereaction (500 pg/ml) were template mechanism for the glycosaminoglycan-stimulated inhigh enough t o overcome the decrease in thrombin affinity. To hibition of thrombin (18). In thecase of HCII, glycosaminoglyaddress this possibility, the dose responses to dermatan sulfate cans also have beenproposed to promote the inhibition of were compared for wild type thrombin and mutant thrombin thrombin by an allosteric mechanism. Deletions within the (Fig. 21, indicating that the amino-terminal acidic domain of HCII do not impair reaction R89E. No differences were observed decreased affinity of thrombin R89E for dermatan sulfate does with thrombin in the absence of glycosaminoglycan,but such not affect the rate of inhibition by HCII, even at suboptimal deletions markedly reduce the rate of thrombin inhibition in the presence of either heparin or dermatan sulfate (12, 19). concentrations of dermatan sulfate. Inhibition of Recombinant Thrombins by HCII Plus Modifications of exosite I of thrombin have similar selective Heparin-In the presence of heparin, the rate constant for effects on the rate of inhibition by HCII; glycosaminoglycanthrombin inhibition by HCII was reduced -400-fold by the stimulated rates are reduced markedly, but inhibition rates or without glycosaminoglycan are affected only slightly (16, 17, mutation R68E in exosite I (Table I) (16). Other mutations in near exosite I (K52E,K154E) had no effect onthe rateenhance- 33). These results suggest that HCII interacts with anion-bindment by heparin. Mutations in exosite I1 reduced modestly the ing exosite I on thrombin, and this interaction is promoted by rate constant for inhibition by HCII in a manner that corre- glycosaminoglycans. These results are consistent with an allated with their effect on affinity for heparin. The mutation losteric model in which the acidic amino-terminal domain of K174E, which does not decrease the affinity of thrombin for HCII occupies a site on HCII that also can bind glycosaminoheparin-agarose (9), did not decrease the rate of inhibition by glycan. Binding to glycosaminoglycan displaces the amino-terHCII plus heparin. Thrombins with the mutations K252E, minal acidic domain so that it can interact with exosite I on R245E, K248E,and R89E have progressively loweraffinity for thrombin and accelerate the formation of HC1I.thrombin comheparin (9) and exhibited progressively lowerrate constants for plexes. Because thrombin also binds to heparin or dermatan inhibition by HCII plus heparin (Table I). The maximal de- sulfate, these data do not exclude a template mechanism that crease in the rate constant was observedfor the mutation requires a glycosaminoglycan-binding site on thrombin. AC-

Allosteric Regulation of HCII

32751

cordingly, hybrid mechanisms have been proposed in which cular sites, possibly through exosite 11. This may explain the exosite Iof thrombin binds to acidic the amino-terminal domain observed resistance of extracellular matrix-bound thrombin to of HCII and exosite I1 binds the glycosaminoglycan moiety of inhibition by antithrombin-heparin (5, 6).Thrombin bound to extracellular matrix can cleave fibrinogen and activate platethe HCII.glycosaminog1ycan complex (12, 17). Mutagenesis andinactivation of the dermatan sulfate-bind- lets (51, suggesting that exosite I is accessible to substrates. ing site permitsa critical test of the template model for HCII. Thus, thrombin bound to extracellular matrix may have perMutant thrombins were constructed that do not bind signifi- sistent procoagulant activity (37). The availability of exosite I cantly t o dermatan sulfate in buffers of physiological ionic suggests thatmatrix-bound thrombin may remain susceptible to HCII, providing a theoretical basis for development of low strength (Fig. 1, C and D ) . These thrombins were inhibited normally by HCII and dermatan sulfate(Table I, Fig. 21, effec- molecular weight glycosaminoglycans (such as dermatan sultively excluding a template mechanism for the reaction. The fate oligosaccharides) that accelerate thrombin inhibition prisame mutations markedly impaired binding the of thrombin to marily by inducing a n allosteric conformational changein heparin (9), but decreased by only 2-7-fold the rate constant for HCII. Such agentscould be useful for the treatmentof vascular thrombin inhibition by HCII and heparin (Table I). Thus, in- injuries inwhich residual thrombin activity may cause thromteraction of thrombin exosite I1 with heparin contributesmini- bosis or stimulate cellular processes that lead to intimal mally to the rateof thrombin inhibition by HCII and heparin. hyperplasia (37). This interpretation is consistent with the effect of deleting the Acknowledgments-Among our colleagues at Washington University, acidic amino-terminal domain of HCII. For this HCII mutant, we thank George J. Broze for providing plasma prothrombin, and Philip heparin accelerates the rate of thrombin inhibition by only W. Majerus for antibody to human thrombin. -35-fold whereas dermatan sulfate has a negligible effect (12); REFERENCES for wild type HCII the reaction is accelerated >16,000-fold by 1. Coughlin, S. R., Vu, T. K. H., Hung, D. T., and Wheaton, V. I. (1992)J. Clin. either glycosaminoglycan (12) (Table I). The major effect of Invest. 89, 351-355 2. Broze, G. J., Jr., and Tollefsen, D.M. (1994)in The Molecular Basis of Blood glycosaminoglycan binding toHCII is, therefore, to promote the Diseases (Stamatoyannopoulos, G., Nienhuis, A.W.,Majerus, P. W., and inhibition of thrombin allosterically. Varmus, H., eds) pp. 629656,W. B. Saunders Company, Philadelphia 3. Jordan, R. E., Oosta. G. M., Gardner, W. T., andRosenberg, R. D. (1980)J.B i d . An allosteric mechanismis consistent with the effects of high Chem. 266, 10081-10090 concentrations of glycosaminoglycan, which were interpreted 4. Tollefsen, D. M., Pestka, C. A,, and Monafo, W.J. (1983)J. Biol. Chem. 268, previously in termsof a template mechanism. Both antithrom671.1-6716 5. Weitz, J. I., Hudoba, M., Massel, D., Maraganore, J., andHirah, J. (1990) bin and HCII achieve maximal rates of thrombin inhibition at J. Clin. Invest. 86,385391 an optimal concentration of glycosaminoglycan, above which 6. Bar-Shavit, R., Eldor, A,, and Vlodavsky, I. (1989)J. Clin. Invest. 84, 10961104 therate of inhibitiondecreases markedly. The slowing of 7. Bode W., Turk, D., and Karshikov, A. (1992)Protein Sci. 1,426471 thrombin inhibition with “excess” glycosaminoglycan occurs at 8. Church, F. C., Pratt, C.W., Noyes, C. M., Kalayanamit, T., Sherrill, G . B., Tobin, R. B., and Meade, J. B. (1989)J. Biol. Chem. 284, 16419-18425 heparin or dermatan sulfate concentrations of >1 mg/ml or >5 U. S. A. 91, 9. Sheehan, J. P., and Sadler, J. E. (1994)Proc. Nutl. Acad Sci. mg/ml, respectively (17). A template mechanism explains this 5518-5522 phenomenon inthecase of antithrombin. However, other 10. Gan, Z.-R., Li, Y.,Chen, Z., Lewis, S. D., and Shafer,J. A. (1994)J. Biol. Chem. 269, 1301-1305 mechanisms are equally plausible in the case of HCII. For 11. Huber, R., and Carrell, R. W. (1989)Biochemistry 28,8951-8966 example, high concentrations of heparin or dermatan sulfate 12. Van Deerlin, V. M. D., and Tollefsen, D. M. (1991)J.Biol. Chem. 268,20223202.11 may bind t o thrombin exosite I (34) (which binds many anions 13. Hoziiy G . L., Tollefsen, D. M, and Benutto, B. M. (1989)J. B i d . Chem. 264, with varying degrees of specificity) and thereby impair the 13979-13982 critical interactions of HCII with this site. Thus, either a tem- 14. Church, F. C., Noyes, C. M., and Griffith, M. J. (1985)Proc. Nutl. Acad. Sci. U. S. A. 82,64314434 plate mechanism or an allosteric mechanism can accommodate 15. Parker, K. A., and Tollefsen, D. M. (1985)J. Biol. Chem. 260, 3501-5 the characteristic biphasic dose-response of antithrombin and 16. Sheehan, J. P., Wu, Q . Y., Tollefsen, D. M., and Sadler, J. E. (1993)J. Biol. Chem. 268,36394645 HCII to glycosaminoglycans. 17. Rogers, S. J., Pratt, C. W., Whinna, H. C., and Church, F. C. (1992)J. Biol. These results demonstrate that glycosaminoglycans stimuChem. 267,3613-3617 late antithrombin and HCII through fundamentally different 18. Olson, S. T., Bjork, I., Sheffer,R., Craig, P. A,, Shore,J. D.,and Choay, J. (1992) J. Biol. Chem. 267,12528-12538 mechanisms. The HCII-glycosaminoglycan complex is acti- 19. Ragg, H.,Ulshofer, T., and Gerewitz, J. (1990)J . B i d . Chem. 266, 22386vated allosterically and interactswith thrombin anion-binding 22391 D. M., Majerus. P. W., and Blank, M. K. (1982)J. Biol. Chem. 267, exosite I, probably through the amino-terminal acidic domain 20. Tollefsen, 2162-2169 of HCII (12, 19); this reaction is essentially independent of 21. Teien, A. N., Abildgaard, U., and Hook, M. (1976)Thromb. Res. 8,859-867 thrombin exosite 11. In contrast, the heparin moiety of the 22. Wu, Q. Y., Sheehan, J. P., TBiang, M., Lentz, S. R., Birktoft, J. J.,and Sadler, E. (19911Proc. Natl. Acad. Sci. U. S. A. 88, 67156719 antithrombin-heparin complex interacts with thrombin exosite 23. Wu.J.Q. Y., Picard, V., Aiach, M., and Sadler, J. E. (1994)J. Biol. Chem. 269, I1 (8-10) acting as a template to approximate antithrombin and 37254730 thrombin on a single glycosaminoglycan molecule (18); this 24. Kisiel, W., and Hanahan, D. J. (1973)Biochim. Biophys. Acta 304, 103- 113 25. Laemmli, U.K. (1970)Nature 227, 6 8 M 8 5 reaction is independent of thrombin exosite I. These contrast- 26. Bitter, T., and Muir, H. (1962)Anul.Biochem. 4,330-334 ing mechanisms of protease recognition are consistent with the 27. Morrison, J. F. (1982)n e n d s Biochem. Sci. 7, 102-105 G., Stone, S. R., Guidolin, A., Sommer, J., andMonard, D. (1992) distinct physiologic roles that have been proposed for anti- 28. Rovelli, Biochemistry 31, 3542-3549 thrombin andHCII as intravascular and extravascular inhibi- 29. Olson, S. T., Halvorson, H. R., and Bjork, I. (1991)J . B i d . Chem. 266, 63426352 tors, respectively. 30. Bourin; M. C., Lundgren-Ackerlund, E., and Lindahl. U.(199O)J.B i d . Chem. Following injury, thrombin is exposed to both intravascular 266. 15424-15431 and extravascular environments thatmay place different con- 31. Ye, J.. Rezaie, A. R., and Esrnon, C.T. (1994)J. B i d . Chem. 269,17965-17970 Gailani, D., and Broze, G. J.,Jr. (1991)Science 269,909-912 straints on the recognition of thrombin by inhibitors. The 32. 33. Phillips, J. E., Shirk, R. A., Whinna, H. C., Henriksen, R. A,, and Church, F. C. therapeutic efficacy of glycosaminoglycan anticoagulants (1993)J. Biol. Chem. 268, 3321-3327 therefore may be limited by these environmental factors. In- 34. Fenton, J. W., 11, Witting, J. I., Gouliott, C., and Fareed, J. (1989)Ann.N. Y Acad. Sci. 666, 158-165 travascular fluid-phase thrombin is inhibited rapidly by com- 35. De Agostini, A. I., Watkins, S. C., Slayter, H. S., Youasoufian, H., and Rosenberg, R. D.(1990)J. Cell B i d . 111, 1293-1304 (35). plexes of antithrombin and heparin or heparan sulfate Hatton, M. W. C. (1993)Blood Coag. Fibrinol. 4, 927-933 Upon vessel wall injury, however, thrombin may bind to der- 36. 37. Hatton, M. W. C., Moar, S. L., and Richardson, M. (1989)Am. J . Pathol. 136, matan sulfate (5, 36) or other glycosaminoglycans in extravas499-508