Thrombin Hydrolysis of Human Osteopontin Is Dependent on Thrombin Anion-binding Exosites*

The cytokine osteopontin (OPN) can be hydrolyzed by thrombin exposing a cryptic α4β1/α9β1 integrin-binding motif (SVVYGLR), thereby acting as a potent cytokine for cells bearing these activated integrins. We show that purified milk OPN is a substrate for thrombin with a kcat/Km value of 1.14 × 105 m–1 s–1. Thrombin cleavage of OPN was inhibited by unsulfated hirugen (IC50 = 1.2 ± 0.2 μm), unfractionated heparin (IC50 = 56.6 ± 8.4 μg/ml) and low molecular weight (5 kDa) heparin (IC50 = 31.0 ± 7.9 μg/ml), indicating the involvement of both anion-binding exosite I (ABE-I) and anion-binding exosite II (ABE-II). Using a thrombin mutant library, we mapped residues important for recognition and cleavage of OPN within ABE-I and ABE-II. A peptide (OPN-(162–197)) was designed spanning the OPN thrombin cleavage site and a hirudin-like C-terminal tail domain. Thrombin cleaved OPN-(162–197) with a specificity constant of kcat/Km = 1.64 × 104 m–1 s–1. Representative ABE-I mutants (K65A, H66A, R68A, Y71A, and R73A) showed greatly impaired cleavage, whereas the ABE-II mutants were unaffected, suggesting that ABE-I interacts principally with the hirudin-like OPN domain C-terminal and contiguous to the thrombin cleavage site. Debye-Hückel slopes for milk OPN (–4.1 ± 1.0) and OPN-(162–197) (–2.4 ± 0.2) suggest that electrostatic interactions play an important role in thrombin recognition and cleavage of OPN. Thus, OPN is a bona fide substrate for thrombin, and generation of thrombin-cleaved OPN with enhanced pro-inflammatory properties provides another molecular link between coagulation and inflammation.

Purification of Human OPN from Milk-OPN was purified from human milk (Mothers Milk Bank, San Jose, CA) using a previously published procedure with modifications (21). Briefly, 1 liter of human milk pooled from several donors was allowed to separate on ice. The whey was carefully decanted from the curd and clarified by centrifugation at 23,000 ϫ g for 60 min at 4°C and then clarified further over a 0.2-m pore size filtration unit. Benzamidine (2 mM) and DTT (2 mM) were added, and the clarified whey fraction was batch-adsorbed to Q-Sepharose (GE Healthcare) for 2 h at 4°C. The resin was washed with 10 mM sodium phosphate/0.2 M NaCl, 2 mM DTT, 2 mM benzamidine, pH 7.4, and then eluted with 10 mM sodium phosphate/1.0 M NaCl, 2 mM DTT, pH 7.4. The eluant was dialyzed overnight in 10 mM sodium phosphate/4 M NaCl, 2 mM DTT, pH 7.4. The dialyzed protein was then loaded onto a 5-ml HiTrap phenyl-Sepharose HP column (GE Healthcare) at 1 ml/min and washed to base line. A linear gradient was developed from 4 to 1 M NaCl at 1 ml/min with 1-ml fractions collected. Pure fractions were pooled and dialyzed extensively against phosphate-buffered saline, pH 7.4. Partially pure fractions were diluted in 10 mM sodium phosphate/4 M NaCl, 2 mM DTT, pH 7.4, and reapplied over the HiTrap phenyl-Sepharose HP column. The concentration of purified protein was estimated at 280 nm using an extinction coefficient (1 cm path length) of 22920. (4 -31 M) was reacted with thrombin (Ϸ5-100 nM) in assay buffer containing 20 mM HEPES, 5 mM CaCl 2 , 0.1% polyethylene glycol 8000, pH 7.5, and at various salt concentrations (Ϸ20 -200 mM NaCl) at 37°C. Reactions were terminated with the addition of PPACK (5 M) followed by DTT (5 mM). HPLC-gel filtration chromatography (HPLC-GFC) was employed to quantitate the generation of the 15-kDa C-terminal fragment (CTF) liberated by thrombin cleavage. Various amounts (10 -40 l) of the reaction mixture were injected over a Shodex 8 ϫ 300 mm KW-803 column (Showa Denko America Inc, New York), and proteins were resolved in 20 mM sodium phosphate, 300 mM NaCl, 5 mM DTT, pH 7.6, at 1 ml/min. The peak area was detected at 254 nm and was converted to nmoles of product using a calibration curve.

Determination of the Michaelis-Menten Parameters for
Thrombin Hydrolysis of Milk OPN-Several concentrations of OPN (4 -31 M) were digested with thrombin (typically 5 nM) at 37°C for various times, such that there was less than 10% hydrolysis of substrate. The reactions were terminated by the addition of 5 M PPACK and 5 mM DTT. The amount of product formed (OPN CTF) was determined by HPLC-GFC as described above. The values for K m and k cat were determined by plotting the initial velocity of cleavage against the different substrate concentrations and then fitting to the Michaelis-Menten equation by nonlinear regression analysis. Experiments were performed in duplicate, and the data were pooled for analysis.
Estimation of k cat /K m for Cleavage of OPN under First Order Rate Conditions-Values for k cat /K m were determined by full reaction progress curves at a substrate concentration well below K m for WT and representative mutants of thrombin within anion-binding exosite I (ABE-I; R20A, K21A, Q24A, R62A, K65A, H66A, R68A, T69A, R70A, Y71A, R73A, K77A, K106A, K107A), ABE-II (R89A/R93A/E94A, R98A, D122A/ R123A/E124A, E169A/K174A/D175A, R178A/R180A/D183A, R245A, K248A), the 50-insertion loop (W50A), and the Na ϩ binding site (E229A). 5 M OPN (0.26 ϫ [K m ]) was preincubated in assay buffer for 10 min at 37°C. Reactions were started with the addition of thrombin (typically 50 nM) and incubated at 37°C. Aliquots were removed at several time points (0 -1800 s) and terminated with 5 M PPACK followed by 5 mM DTT. The amount of product generated (nmoles) for each time point was determined by HPLC-GFC as detailed above. Values for k cat /K m were determined by fitting data from the time course experiments to the following equation (22), where [CTF] is the concentration of the CTF at a given time, CTF f is the concentration of the CTF at full activation, and [E T ] is the total enzyme concentration of thrombin at any given time (t). Curve fitting was performed using Prism 5 (Graph-Pad Software). Determination of K m , k cat , and k cat /K m for Hydrolysis of the Peptide OPN-(162-197)-For the estimation of Michaelis-Menten parameters, 10 -300 M OPN-(162-197) peptide was incubated with WT or mutant thrombin in assay buffer containing 20 mM HEPES, 145 mM NaCl, 5 mM CaCl 2 , 0.1% polyethylene glycol 8000, pH 7.5. Enzyme concentration and incubation time were varied such that there was less than 10% substrate depletion. Reactions were terminated with the addition of 10% perchloric acid. The cleaved and uncleaved peptides were separated by reverse-phase HPLC on a Waters Symmetry C 18 (4.6 ϫ 150 mm, 5 M) column using a 0.2-35% acetonitrile/ 0.1% trifluoroacetic acid gradient over 30 min at a flow rate of 1.0 ml/min. Cleavage was specific with no secondary cleavage sites. The cleaved peptide peak OPN-(162-168) (SVVYGLR) eluted at 19.8 min and was well resolved from OPN-(162-197) (23.6 min) and the OPN-(169 -197) cleavage product (22.3 min). Identity of the cleavage products was confirmed by mass spectrometry. The area under the SVVYGLR peak was converted to nmoles of peptide using a calibration curve constructed with purified and quantitated SVVYGLR peptide. The values for K m and k cat were determined by fitting to the Michaelis-Menten equation by non-linear regression analysis as described above. The deduced value for K m was used to determine the reaction conditions for estimating the specificity constant, k cat /K m . OPN-(162-197) peptide (10 M, 0.04 ϫ K m ) was preincubated in assay buffer (20 mM HEPES, 145 mM NaCl, 5 mM CaCl 2 , 0.1% polyethylene glycol 8000, pH 7.5) for 10 min. Reactions were started with the addition of WT or mutant thrombin (100 -500 nM) with aliquots removed at several time points (0 -1800 s) and then terminated with the addition of 10% perchloric acid. The effect of NaCl and choline chloride (ChCl) on k cat /K m was determined in the presence of assay buffer with salt concentrations ranging from 20 to 200 mM. Assays with ChCl were buffered with 5 mM Tris-HCl, pH 7.5, to reduce Na ϩ ions from pH adjustment. The amount of product (SVVYGLR) formed was quantitated as described above by reverse-phase HPLC analysis, and the values for k cat /K m were determined as described above.  1A). However, significant substrate and/or product inhibition was observed at concentrations of 50 M and higher (data not shown). To account for substrate/product inhibition effects, we estimated a value for the specificity constant k cat /K m of (1.14 Ϯ 0.17) ϫ 10 5 M Ϫ1 s Ϫ1 (Fig.  1B) by full reaction progress curves using a substrate concentration estimated to be 0.26 ϫ K m . Under these conditions the estimate for k cat /K m is ϳ2.5 times better and most likely reflects the best estimate for the specificity constant.

Determination of the Kinetic Parameters for Hydrolysis of Purified Human Breast Milk OPN by
Unfractionated Heparin, LMW Heparin, and Unsulfated Hirugen Inhibit Thrombin Hydrolysis of OPN-Osteopontin contains two putative heparin-binding domains (23), therefore we hypothesized that heparin could act as a bridge between the heparin-binding domains and the thrombin ABE-II domain, enhancing hydrolysis. OPN can be bound to HiTrap heparin-Sepharose and eluted with 0.37 M NaCl, suggesting the presence of OPN heparin-binding domains (data not shown). However, we found that LMW heparin and unfractionated heparin caused dose-dependent inhibition of thrombin hydrolysis of OPN with IC 50 values of 31.0 Ϯ 7.9 and 56.6 Ϯ 8.4 g/ml, respectively ( Fig. 2A). Further, the inhibition curves were not bell-shaped, suggesting that heparin does not function as a macromolecular bridge binding both thrombin and OPN to enhance thrombin cleavage of OPN. Rather, both unfractionated and LMW heparins compete against OPN for binding to the thrombin ABE-II, indicating the importance of ABE-II in thrombin cleavage of OPN.
Unsulfated hirugen is a tight binding inhibitor specific only for thrombin ABE-I. Hirugen also caused dose-dependent inhibition of thrombin cleavage of OPN with an IC 50 of 1.2 Ϯ 0.2 M (Fig. 2B). The inhibition studies using LMH heparin and hirugen suggest that both exosites play a role in thrombin cleavage of OPN.
Mapping of Residues in ABE-I and ABE-II Important for the Hydrolysis of OPN-A bank of more than 56 mutant thrombins has been used to map domains important in the cleavage of FV (24), FVIII (25), FXI (26), FXIII (22), and fibrinogen (27) for thrombomodulin-dependent activation of protein C and thrombin-activable fibrinolysis inhibitor (27,28) and for heparin-accelerated inhibition of antithrombin, heparin cofactor II, and protein C inhibitor (29 -31). These studies have identified key residues within ABE-I and ABE-II important in defining the interaction interface for all these interactions. The specificity  constant was determined for 14 mutants in ABE-I and seven mutants in ABE-II along with two thrombin mutants (W50A and E229A) known to affect the thrombin active site (23-32). Ten thrombin ABE-I mutants had less than 50% WT activity, with mutants K65A (29%), H66A (14%), Y71A (16%), and R73A (20%) severely affected (Table 1), which indicates a very large interaction interface within ABE-I. Three mutants in ABE-II (total of seven alanine-substituted residues) have impaired activity relative to WT thrombin, R98A (41%), R89A/R93A/ E94A (31%), and R178A/R180A/D1778A (46%), confirming the role of this ABE from the heparin inhibition studies. The mutant thrombins W50A and E229A were also assessed for their ability to hydrolyze OPN. Trp 50 forms part of the 50-insertion loop and is important in defining the insertion loop and the aryl binding site, and Glu 229 maintains the integrity of the Na ϩ -binding loop. The crystal structure of E229A shows the collapse of the active site with almost complete occlusion (32). Alanine substitution of both of these residues has a profound effect in the ability of thrombin to hydrolyze OPN, both with 9% WT activity. These mutations clearly show that a functional thrombin active site is needed to cleave at the P 1 arginine of SVVYGLR/SKSKKFQ.
Identification of an OPN Domain That Interacts with ABE-I-OPN has a hirudin-like domain, DIQYPDATDEDITSHMESEE (33), adjacent and contiguous to the thrombin cleavage site, that could potentially interact with ABE-I. To address this potential interaction, a peptide was made (OPN-(162-197)) that spanned the cleavage site and hirudin-like ABE-I-binding domain. Values for K m (296.6 Ϯ 10.9 M), k cat (3.3 Ϯ 0.1 s Ϫ1 ), and k cat /K m (1.11 ϫ 10 4 M Ϫ1 s Ϫ1 ) were initially determined over a range of substrate concentrations; the value derived for K m was used to define experimental conditions to estimate k cat /K m under first order rate conditions (Fig. 3). The specificity constant k cat /K m was determined for the cleavage of this OPN-(162-197) by WT thrombin and was found not to be a good substrate (k cat /K m 1.64 Ϯ 0.06 ϫ 10 4 M Ϫ1 s Ϫ1 ) when compared with milk OPN (k cat /K m 1.14 Ϯ 0.17 ϫ 10 5 M Ϫ1 s Ϫ1 ). Hirugen was found to inhibit thrombin cleavage of OPN-(162-197) (  Table 2). The ABE mutants K65A, H66A, R68A, Y71A, and R73A in general showed similar decreases in activity relative to  WT thrombin when compared with the cleavage of milk OPN. Heparin had no effect, excluding the participation of ABE-II ( Fig. 2A). Consistent with this finding, ABE-II mutants R98A, E160A/K174A/D175A, and R178A/R180A/D183A in general had no effect on cleavage of the peptide, suggesting that the OPN hirudin-like domain interacts principally with the thrombin ABE-I. The ABE-II R89A/R93A/E94A mutant showed 44% WT activity. The residue Glu 94 lies on the rim between ABE-II and the active site, so it is conceivable that mutation of this residue could influence cleavage within the active site.

Electrostatic Interactions Are Important for Hydrolysis of OPN by Thrombin-Studies between thrombin and hirudin
show the importance of complementary electrostatic fields in enhancing the rate of complex formation by "electrostatic steering" (34,35). To study this phenomenon for thrombin cleavage of OPN, k cat /K m was determined over a range of NaCl concentrations for milk-derived OPN (38 -203 mM NaCl). The plot of k cat /K m versus [NaCl] for milk OPN (Fig. 4A) shows that the specificity constant decreased by 1 order of magnitude from 3.18 ϫ 10 5 M Ϫ1 s Ϫ1 to 0.28 ϫ 10 5 M Ϫ1 s Ϫ1 when the ionic strength was increased from 72 to 203 mM NaCl. The lower value for k cat /K m at 38 mM NaCl could reflect transition from the fast (Na ϩ bound) to slow (Na ϩ free) form of thrombin (36). To address potential changes to the ABE-I electrostatic potential made by the fast to slow form transition (37), we examined the cleavage of  in the presence of either ChCl (20 -200 mM), which favors the slow form of thrombin, or NaCl (53-198 mM). The specificity constant for the cleavage of OPN-(162-197) decreased 3.2-fold from 0.34 ϫ 10 5 M Ϫ1 s Ϫ1 at 53 mM NaCl to 0.11 ϫ 10 5 M Ϫ1 s Ϫ1 at 198 mM NaCl (Fig. 4B). A 2.7-fold decrease in the specificity constant from 0.19 ϫ 10 5 M Ϫ1 s Ϫ1 to 0.71 ϫ 10 4 M Ϫ1 s Ϫ1 was also observed from 50 to 200 mM ChCl. As expected, the specificity constant was consistently reduced in the presence of ChCl, consistent with the presence of the slow form of thrombin. At physiological salt concentration (Ϸ150 mM) the k cat /K m for the cleavage of  in the presence of NaCl (1.64 Ϯ 0.06 ϫ 10 4 M Ϫ1 s Ϫ1 ) was 1.7-fold greater than the predominantly "slow form" of thrombin (9.75 Ϯ 1.70 ϫ 10 3 M Ϫ1 s Ϫ1 ) in ChCl.
A Debye-Hückel slope from 53 to 203 mM 1/2 NaCl for the cleavage of milk OPN gave a slope of Ϫ4.1 Ϯ 1.0, signifying a high degree of electrostatic interactions in forming the thrombin-OPN encounter complex. A better fit was obtained from 72 to 203 mM 1/2 NaCl, with a Debye-Hückel slope of Ϫ5.9 Ϯ 0.7, and at an intercept at 0 ionic strength the value for k cat /K m was 1.3 ϫ 10 7 M Ϫ1 s Ϫ1 . The Debye-Hückel slope for OPN-(162-197) from the NaCl titration was Ϫ2.4 Ϯ 0.2 with a k cat /K m at 0 ionic strength estimated to be 1.3 ϫ 10 5 M Ϫ1 s Ϫ1 . A similar slope (Ϫ1.9 Ϯ 0.1) was obtained from the ChCl titration, suggesting that OPN hydrolysis by thrombin is driven primarily by electrostatic steering, with decreased ionic strength in the solution favoring a faster and more productive complex formation between thrombin ABE-I and OPN-(162-197), which is independent of the allosteric state of thrombin.

DISCUSSION
OPN is a multifunctional protein that can act both as an extracellular matrix protein involved in bone resorption and as a cytokine (1-3). As a cytokine, OPN can interact with a number of intergrin-bearing cells through RGD-dependent interactions (1)(2)(3). It has been shown in vitro that thrombin can cleave OPN, which exposes the cryptic integrin-binding motif (SVVYGLR) for binding to the integrins ␣ 4 ␤ 1 or ␣ 9 /␤ 1 , independently of RGD or CD44 (13)(14)(15)(16)(17)(18)(19). This suggests that thrombin could also act indirectly as a mediator of inflammatory and immune responses by activating OPN. However, the in vivo significance of this is not well understood, and whether OPN is a real substrate for thrombin remains to be established.
In this report we have shown that OPN purified from human milk is a substrate for thrombin with a specificity constant of 1.14 ϫ 10 5 M Ϫ1 s Ϫ1 . The thrombin active site cleft is topologically similar to that of the archetypal serine protease trypsin, an enzyme that can indiscriminately cleave any protein after an exposed P 1 arginine (38,39). However, the specificity of thrombin is significantly more restricted because of two insertion loops that partially occlude the active site to macromolecular substrates and inhibitors (38,39). To overcome steric hindrance at the active site, the specific substrates and inhibitors of thrombin have to bind to two distal exosites, which are characterized by a high density of positively charged, solvent-exposed residues (38,39). ABE-I binds hirudin, PAR1, PAR4, thrombomodulin, actor XIII, and fibrinogen, and ABE-II binds the heparin-bound serpins (antithrombin, heparin cofactor II, and protease nexin 1), the leech inhibitor hemadin, and platelet glycoprotein Ib (40,41). Both exosites are required for the efficient activation of coagulation factors V, VIII, and XI (40,41). Hirugen, a specific and tight inhibitor of ABE-I (42), effectively blocked thrombin cleavage of OPN. Unfractionated and LMW heparin also inhibited thrombin cleavage of OPN, presumably through direct competition against OPN binding to ABE-II. The dependence of both ABE-I and ABE-II on optimal cleavage of OPN supports the notion that OPN is a bona fide substrate for thrombin.

peptide
The specificity constant k cat /K m was estimated for the hydrolysis of OPN-(162-197) peptide by WT and selected mutant thrombins under first order rate conditions in assay buffer fixed at 145 mM NaCl. Values for the mean Ϯ S.E. are based on at least two assays performed in duplicate. and ABE-II showed 3 of 5 mutant thrombins (potentially 7 of 15 residues) with impaired activity, confirming the hirugen and heparin studies. Whether this is the result of specific impairment of direct interactions or small conformational changes could not be distinguished, as there is no crystal structure of the thrombin-OPN complex. However it is quite remarkable that the residues identified are similar to those characterized previously as being involved in thrombin cleavage of FV, FVIII, FXI, and FXIII (23)(24)(25)(26). An interesting feature of the OPN polypeptide sequence is a hirudin-like ABE-I-binding domain C-terminal to the SVVYGLR cryptic motif and thrombin cleavage site. The sequence has a large number of acidic amino acids and three serine/threonine phosphorylation sites, along with two prolines and a tyrosine, a common feature of hirudin-like domains seen with thrombomodulin (residues 408 -426), factor VIII (residues 716 -731), PAR1 (residues 52-69), and heparin cofactor II (residues 56 -75), all of which utilize ABE-I. The peptide , synthesized to span the cryptic motif, thrombin cleavage site, and hirudin-like ABE-I-binding domain, was used in cleavage studies with WT and selected ABE-I and ABE-II mutant thrombins. The  peptide was a poor substrate (k cat /K m 1.64 Ϯ 0.06 ϫ 10 4 M Ϫ1 s Ϫ1 ) for WT thrombin compared with OPN purified from human milk (k cat /K m 1.14 Ϯ 0.17 ϫ 10 5 M Ϫ1 s Ϫ1 ). This result was not wholly unexpected because the peptide lacks the complete functional domains of native OPN and would not be able to present the thrombin cleavage site for hydrolysis in an optimal manner. Further, OPN contains many serine/threonine phosphorylation sites (43), three of which are within the OPN-(162-197) peptide (Thr 185 , Ser 191 , and Ser 195 ). The peptide OPN-(162-197) lacks these additional negative charges and thus has a reduced overall electrostatic potential, which may be important in electrostatic steering and ion pair interactions (34,35,44). Indeed, native hirudin with a sulfated Tyr 63 forms a very tight equimolar complex with thrombin with a K i of 20 femtomolar. Loss of this charge, as seen with recombinant hirudin, increases the K i 10-fold to 200 femtomolar (45). Studies are in progress to assess the role of phosphorylated Thr 185 , Ser 191 , and Ser 195 residues for interaction with the thrombin ABE-I domain. Interestingly, OPN can exist in different forms, based on the extent of phosphorylation, with potentially different biological functions (43, 46 -48). Therefore it is possible that in vivo thrombin can have different specificities toward the different OPN species.
An important characteristic of the thrombin ABEs is a high density of basic residues, which produces a large positive electrostatic potential that protrudes into the solvent (35). The electrostatic field generated by thrombin ABE-I has been shown to be important for electrostatic steering with the complementary electrostatic field in its ligands in order to enhance productive complex formation (35,46). Indeed, studies looking at the binding of thrombin with the thrombomodulin "TM- 456" (EGF-456) domain show a very rapid and almost completely electrostatically steered interaction (49). For milk OPN cleavage by thrombin over a range of NaCl concentrations, we deduced a Debye-Hückel slope of Ϫ4.1 Ϯ 1.0, which is close to the reported slope of Ϫ6.0 Ϯ 1.0 for the interaction of thrombin and TM-456. The slow "Na ϩ -free" form of thrombin could potentially have a long range effect on interactions within ABE-I (37). The Debye-Hückel slopes were very similar between the NaCl (Ϫ2.4 Ϯ 0.2) and ChCl (Ϫ1.9 Ϯ 0.1) titrations, suggesting that the allosteric form of thrombin does not effect electrostatic steering. It is notable that both of the deduced Debye-Hückel slopes are similar to values reported for the release of fibrinopeptides A and B by thrombin, which have been shown to be dependent on electrostatic steering and independent of the fast or slow form of thrombin (50). Our results demonstrate that OPN hydrolysis by thrombin is highly dependent on electrostatic steering and that the interaction between ABE-I and the hirudin-like motif within the OPN-(162-197) sequence plays an important role.
Although electrostatic steering appears to play an important part in enhancing the formation of productive encounter complexes, the k cat /K m for OPN is considerably lower than that of the other substrates of thrombin. For example, hydrolysis of fibrinogen (1.2 ϫ 10 7 M Ϫ1 s Ϫ1 for fibrinopeptide A release and 4.2 ϫ 10 6 M Ϫ1 s Ϫ1 for fibrinopeptide B release) (51), the PAR1 receptor peptide (8.0 ϫ 10 7 M Ϫ1 .s Ϫ1 ) (52), FV (3.3 ϫ 10 6 M Ϫ1 s Ϫ1 ) (53), and cleavage of FVIII at Arg 372 (6.1 ϫ 10 6 M Ϫ1 s Ϫ1 ), Arg 740 (1.1 ϫ 10 8 M Ϫ1 s Ϫ1 ), and Arg 1689 (7.3 ϫ 10 6 M Ϫ1 s Ϫ1 ) (54) have specificity constants 30 -1000-fold higher. The specificity of thrombin for OPN could be restricted by suboptimal binding of the P 3 to P 3 Ј residues within the active site. The OPN P 2 Ј (Lys 170 ) and P 3 Ј (Ser 171 ) residues are non-optimal in their interaction with thrombin, which prefers bulky hydrophobic residues at P 2 Ј and positively charged residues at P 3 Ј. OPN also has a non-optimal leucine at P 2 and glycine at P 3 , with thrombin preferring a proline at P 2 and residues with an aryl side chain at P 3 (55). On the other hand, a moderately efficient catalysis would be sufficient in an extravascular compartment, where the high-affinity thrombin inhibitor, antithrombin, is normally not present, and in fact it may have a biological advantage in not generating excessive amounts of a potent cytokine that can exert its influence at very low levels. In contrast, it is imperative for thrombin to function at near maximal catalytic efficiency toward coagulation factors within the vascular compartment, because it needs to form a clot rapidly and to do so despite the presence of a high concentration of antithrombin. Therefore we propose that thrombin utilizes its ABEs to define its specific docking with different substrates but can modulate its ability to turn over substrates by differential binding modes within the active site. In this way, thrombin can link both coagulation and inflammation and generate levels of biologically active mediators appropriate to each process.