Histidine-rich Glycoprotein Binds Fibrin(ogen) with High Affinity and Competes with Thrombin for Binding to the γ′-Chain*

Histidine-rich glycoprotein (HRG) is an abundant protein that binds fibrinogen and other plasma proteins in a Zn2+-dependent fashion but whose function is unclear. HRG has antimicrobial activity, and its incorporation into fibrin clots facilitates bacterial entrapment and killing and promotes inflammation. Although these findings suggest that HRG contributes to innate immunity and inflammation, little is known about the HRG-fibrin(ogen) interaction. By immunoassay, HRG-fibrinogen complexes were detected in Zn2+-supplemented human plasma, a finding consistent with a high affinity interaction. Surface plasmon resonance determinations support this concept and show that in the presence of Zn2+, HRG binds the predominant γA/γA-fibrinogen and the γ-chain elongated isoform, γA/γ′-fibrinogen, with Kd values of 9 nm. Likewise, 125I-labeled HRG binds γA/γA- or γA/γ′-fibrin clots with similar Kd values when Zn2+ is present. There are multiple HRG binding sites on fibrin(ogen) because HRG binds immobilized fibrinogen fragment D or E and γ′-peptide, an analog of the COOH terminus of the γ′-chain that mediates the high affinity interaction of thrombin with γA/γ′-fibrin. Thrombin competes with HRG for γ′-peptide binding and displaces 125I-HRG from γA/γ′-fibrin clots and vice versa. Taken together, these data suggest that (a) HRG circulates in complex with fibrinogen and that the complex persists upon fibrin formation, and (b) by competing with thrombin for γA/γ′-fibrin binding, HRG may modulate coagulation. Therefore, the HRG-fibrin interaction may provide a novel link between coagulation, innate immunity, and inflammation.

Fibrinogen is the soluble precursor of fibrin, a critical component of blood clots that endows them with strength and elasticity. Fibrinogen is a glycoprotein composed of three pairs of polypeptide chains, termed A␣, B␤, and ␥, that are connected by disulfide bonds (1). Approximately 10 -15% of circulating fibrinogen has a variant ␥-chain termed the ␥Ј-chain, which results from differential processing of the ␥ A -chain mRNA transcript (2)(3)(4). The ␥Ј-chain is distinguished from the ␥ A -chain by the presence of an acidic 20-residue extension at its COOH terminus (2,5).
Thrombin catalyzes the conversion of fibrinogen to insoluble fibrin, and during this process some thrombin remains bound to the fibrin network (6). Exosites 1 and 2 are two regulatory domains that flank the active site of thrombin and mediate its binding to fibrin (7,8). Exosite 1 of thrombin interacts with the central E-domain of fibrin, whereas exosite 2 binds only to the COOH terminus of the ␥Ј-chain. Consequently, thrombin binds ␥ A /␥ A -fibrin in a univalent fashion with a K d value of 2-4 M. In contrast, both exosites are engaged when thrombin binds to ␥ A /␥Ј-fibrin, resulting in a higher affinity interaction (K d value of 0.08 -0.18 M) (5,9). Fibrin-bound thrombin remains active, and the protease is protected from inhibition by fluid-phase inhibitors, such as antithrombin and heparin cofactor II (6). Because of its bivalent interaction with ␥ A /␥Ј-fibrin, thrombin bound to ␥ A /␥Ј-fibrin is more protected from inhibition by fluid-phase inhibitors than thrombin bound to ␥ A /␥ Afibrin (10).
Like thrombin, histidine-rich glycoprotein (HRG) 3 binds to fibrinogen and is incorporated into fibrin clots (11). Although the plasma concentration of HRG ranges from 1.6 to 2 M, the concentration in platelet-rich thrombi may be higher because HRG is stored in the alpha granules of platelets and is released when platelets are activated (12,13). A 75-kDa glycoprotein, HRG, is composed of two NH 2 -terminal cystatin-like domains, a central histidine-rich region (HRR) flanked by two prolinerich regions, and a COOH-terminal domain (14). In addition to fibrinogen, HRG also binds plasminogen, heparan sulfate, and divalent cations, such as Zn 2ϩ (12,15,16). Therefore, HRG is hypothesized to be an important accessory or adapter protein that brings different ligands together under specific conditions (14).
HRG-deficient mice exhibit a shorter prothrombin time and accelerated fibrinolysis compared with wild-type mice, raising the possibility that HRG modulates coagulation and fibrinolysis (17). In addition to its potential role in hemostasis, HRG also has been implicated in innate immunity and inflammation (18). HRG exhibits antifungal and antimicrobial activity in vitro, and these activities are enhanced at low pH or in the presence of Zn 2ϩ , conditions that promote ligand binding (19,20). The antimicrobial activity of HRG has also been demonstrated in vivo and appears to be fibrin-dependent. Thus, compared with wild-type mice, HRG-deficient mice are more susceptible to the lethal effect of Streptococcus pyogenes infection and are rescued with HRG supplementation (21). This phenomenon is fibrindependent because fibrin is essential for HRG-mediated bacterial entrapment and killing, processes that prevent bacterial dissemination. In addition, the HRG-fibrin interaction modulates inflammation because HRG-deficient mice exhibit attenuated abscess formation in response to subcutaneous injection of bacteria. Based on these findings, it has been postulated that HRG plays a fibrin-dependent role in both inflammation and innate immunity (21).
Despite emerging evidence that the HRG-fibrin(ogen) interaction is physiologically important, little is known about the biochemical foundation of this interaction or its functional consequences. To address these gaps in knowledge, we set out to (a) quantify the binding of HRG to fibrin(ogen), (b) identify the HRG binding domains on fibrin(ogen), and (c) determine whether there is overlap between the HRG and thrombin binding domains on fibrin(ogen).

Materials
Reagents-Human thrombin, prothrombin, and plasminogen-free fibrinogen were from Enzyme Research Laboratories (South Bend, IN). Plasmin and factor XIII were from Haematologic Technologies Inc. (Essex Junction, VT). Prionex was from Pentapharm (Basel, Switzerland). A 20-amino acid analog of the COOH terminus of the ␥Ј-chain of fibrinogen, ␥Ј-peptide (VRPEHPAETEYDSLYPEDDL), was prepared by Bachem Bioscience, Inc. (King of Prussia, PA), and a sheep antibody against this peptide was from Affinity Biologicals (Ancaster, ON). The two Tyr residues within the ␥Ј-peptide were modified with phosphate groups in place of sulfate to enhance stability (22). The ␥Ј-peptide-directed IgG was subjected to affinity chromatography using immobilized ␥Ј-peptide (SulfoLink Immobilization Kit for Peptides, Thermo Scientific, Rockford, IL). D-Phe-Pro-Arg chloromethyl ketone and D-Tyr-Pro-Arg chloromethyl ketone (FPRck and YPRck, respectively) were from EMD Chemicals (Gibbstown, NJ). Fibrinogen was rendered factor XIII-free, and ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen were separated by fractionation on DEAE-Sepharose (GE Healthcare) and characterized as previously described (10,23). HRG was purified by metal-chelate chromatography, and HRG-deficient plasma was prepared as described previously (24). Non-immune sheep IgG and a human HRGdirected IgG from sheep were prepared by Affinity Biologicals, and the HRG-specific IgG fraction was isolated by affinity chromatography using immobilized HRG (24). Unless otherwise specified, other reagents were from Sigma.
Preparation of Fibrinogen Fragments-Fragment X was prepared by limited plasmin digestion of fibrinogen (23). Fragments D and E were generated by plasmin digestion of ␥ A /␥ Afibrinogen (25). Digested material was applied to a 12-ml UNO Q-12 ion exchange column (Bio-Rad) using a Bio-Rad Biologic Duoflow system at a flow rate of 5 ml/min. To elute non-specifically bound proteins, the column was washed with 30 ml of 0.02 M sodium phosphate, 0.01 M citric acid, pH 7.6. Fragment D was eluted with 100 ml of 0.1 M sodium phosphate, 0.05 M citric acid, pH 5.0, whereas fragment E was subsequently eluted with 40 ml of the same buffer at pH 4.4 (26). Protein-containing fractions were identified by absorbance at 280 nm and pooled. Purified fragments were dialyzed into 10 mM Hepes-NaOH, 150 mM NaCl, pH 7.4 (HBS) and concentrated. Final concentrations were determined at 280 nm (27). The integrity of the fragments was assessed by SDS-PAGE analysis on 4 -15% polyacrylamide gels (Ready-Gel, Bio-Rad) under reducing and non-reducing conditions. Samples were stored in aliquots at Ϫ80°C.
Preparation of ␥Ј-directed IgG Fab Fragments-For some experiments, fragment antibody binding (Fab) regions from the affinity-purified ␥Ј-peptide-directed IgG were generated by papain digestion (28), isolated with a Fab preparation kit (Pierce), assessed for purity by SDS-PAGE analysis, and then concentrated and dialyzed against HBS.

Methods
Surface Plasmon Resonance (SPR)-The interaction of HRG with immobilized ␥ A /␥ A -or ␥ A /␥Ј-fibrin(ogen), biotinylated-␥Ј-peptide, or fragments X, D, or E was assessed by SPR using a BIAcore 1000 (GE Healthcare) as previously described (24,30) but with some modifications. Briefly, proteins were covalently linked to separate flow cells of a carboxymethylated dextran (CM4) biosensor chip at a flow rate of 5 l/min using an amine coupling kit (GE Healthcare). Proteins were immobilized using 10 mM acetate buffer at varying pH values to maximize adsorption. ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen or fragments X and D were immobilized at pH 5.5 to ϳ3000 -7000 response units (RU). Fragment E was immobilized at pH 4.5 to ϳ2000 -3000 RU. For fibrin binding studies, immobilized ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen was converted to fibrin by three successive 60-min injections of 100 nM thrombin at 5 l/min (30). To prepare streptavidinconjugated CM4 flow cells, 0.4 mg/ml streptavidin (Sigma) at pH 4.5 was injected. Biotinylated ␥Ј-peptide (30) was adsorbed to the immobilized streptavidin to 200 -300 RU. Remaining reactive groups were neutralized with 1 M ethylenediamine, and non-specifically adsorbed proteins were removed by treatment with 0.5 M NaCl. An unmodified flow cell served as the control. All SPR procedures were done in HBS containing 0.005% Tween 20 and 2 mM CaCl 2 , and flow cells were regenerated with 250 mM imidazole and 2 mM EDTA between runs.
To measure the affinity of HRG for immobilized fibrinogen, fibrin, or fibrinogen fragments, aliquots of HRG (0 -1 M) in buffer containing 20 M ZnCl 2 were injected at a flow rate of 30 l/min. To quantify Zn 2ϩ dependence, binding of 200 nM HRG to immobilized ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen was monitored in the presence of varying concentrations of ZnCl 2 (0 -60 M) using dual injection mode. The binding of thrombin to immobilized fragment D or E was monitored by injection of FPRckthrombin (0 -15 M) into flow cells.
The ␥Ј-peptide-directed IgG was used to assess the contribution of the ␥Ј-chain to the interaction of HRG with immobilized ␥Ј-peptide or fibrinogen. A saturating amount of ␥Ј-peptidedirected IgG or a non-immune IgG (2 M) was first injected into flow cells containing immobilized ␥Ј-peptide or ␥ A /␥Ј-or ␥ A /␥ A -fibrinogen. The binding of HRG to fibrinogen or fibrinogen fragments was then measured as described above, except that flow cells were re-saturated with 0.5 M ␥Ј-peptide-directed IgG or control IgG before each HRG injection.
Binding of HRG to immobilized ␥Ј-peptide in the absence or presence of FPRck-thrombin was monitored using a BIAcore T200. Biotinylated ␥Ј-peptide was adsorbed to streptavidinconjugated CM4 flow cells to 100 RU in the presence of 20 M ZnCl 2 at a flow rate of 5 l/min. An unmodified flow cell served as a control. Using dual injection mode, 1 M HRG was injected at 10 l/min followed by a second injection of FPRck-thrombin or prothrombin at concentrations ranging from 0 to 8 M. Flow cells were regenerated with 250 mM imidazole, 2 mM EDTA, and 1 M NaCl. All experiments were performed at least twice.
SPR Data Analysis-K d values were determined by kinetic analysis of on-and off-rates of HRG binding to immobilized ligands using Scrubber2 version 2.0a (Bio-Logic Software Co., Campbell, Australia) as described previously (24,30). For further assessment of binding, the amount of HRG bound at the equilibrium position (Req) was determined using the Langmuir 1:1 binding model (BIAEvaluation software Version 3.2) and was plotted against the titrant concentration. Molar stoichiometries were determined as described in the BIAtechnology handbook (BIAcore 1000). The correction factor to account for the orientation of the immobilized fibrinogen and fibrin was 0.25, which corresponds to 25% of the amount of immobilized fibrin accessible to the ␥Ј-peptide-directed IgG as determined in a separate study (10). The correction factor for immobilized ␥Ј-peptide was 0.7, which corresponds to 70% correct orientation of peptide accessible to an analyte (BIAcore).
Interaction of 125 I-HRG with Fibrin Clots-In a series of microcentrifuge tubes, ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen, in concentrations ranging from 0 to 1.25 M, was clotted with 10 nM thrombin in the presence of 40 nM 125 I-HRG as previously described (30). Clots were formed in 20 mM Tris-HCl and 150 mM NaCl, pH 7.4 (TBS), containing 0.005% Tween (TBS-Tween) and 2 mM CaCl 2 and 20 M ZnCl 2 or 10 M sodium diethyldithiocarbamate trihydrate. The concentration of 125 I-HRG bound to fibrin was calculated by subtracting the concentration in the clot supernatant from the value obtained in controls prepared in the absence of fibrinogen. Plots of bound 125 I-HRG versus fibrin concentration were analyzed by nonlin-ear regression of a rectangular hyperbola to determine K d . Experiments were performed twice in duplicate.
Effect of Competitors on the Binding of 125 I-HRG to Fibrin Clots-Clots were formed in TBS-Tween containing 2 mM CaCl 2 and 20 M ZnCl 2 . Fab fragments derived from ␥Ј-peptide-directed IgG were used to assess the contribution of the ␥Ј-chain to 125 I-HRG binding to clots. After preincubation of 0.25 M ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen with ␥Ј-peptide-directed Fab fragments or control sheep IgG (0 -4 M) for 1 h at 23°C, 60 nM 125 I-HRG was added, and clots were generated with 20 nM thrombin. To determine whether HRG and thrombin share fibrin binding sites, the effect of FPRck-thrombin on the binding of 125 I-HRG was assessed. Samples containing 2 M ␥ A /␥ Aor ␥ A /␥Ј-fibrinogen, 20 nM 125 I-HRG, and FPRck-thrombin (0 -8 M) in 0.25% Prionex were clotted with 20 nM thrombin. The effect of varying concentrations of HRG on the binding of 125 I-YPRck-thrombin was assessed in a reciprocal experiment. Samples containing 2 M ␥ A /␥ A -or 0.25 M ␥ A /␥Ј-fibrinogen, 20 nM 125 I-YPRck-thrombin, and HRG (0 -2 M) were clotted with 10 nM thrombin. The fraction of 125 I-HRG or 125 I-YPRckthrombin bound was determined as described above. All experiments were performed twice in duplicate.
HRG Diffusion from Preformed Fibrin Clots-The rate of 125 I-HRG dissociation from ␥ A /␥ A -or ␥ A /␥Ј-fibrin clots was determined as previously described (10,30). Briefly, fibrin clots were formed around plastic inoculation loops (Bac-Loop, Thermo-Fisher Scientific, Waltham, MA) by adding 10 nM thrombin to solutions containing 5 M ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen, 20 nM factor XIII, and 50 nM 125 I-HRG in TBS-Tween containing 2 mM CaCl 2 and 20 M ZnCl 2 . After incubation for 45 min at 23°C, clots were removed and immersed in buffer containing 2 mM CaCl 2 plus 20 M ZnCl 2 , 2 mM CaCl 2 plus 10 M diethyldithiocarbamate trihydrate, or 2 M NaCl plus 2 mM EDTA. The fraction of clot-associated 125 I-HRG remaining at various times was determined, and time courses were fit to a two-phase exponential decay curve (Table Curve, Jandel Scientific, San Rafael, CA) (10). Experiments were repeated three times.
Detection of HRG-Fibrinogen Complexes in Plasma-HRGfibrinogen complexes in plasma were detected using a previously described sandwich ELISA (24) with some modifications. Briefly, 100 l of fibrinogen-directed capture antibody (Affinity Biologicals) diluted to 20 g/ml in 50 mM NaHCO 3 , pH 9.6, was added to wells of a 96-well Immunlon 4 HBX plate (Thermo Scientific) and incubated overnight at 4°C. To block nonspecific binding, 200 l of 10 mg/ml bovine serum albumin (Sigma) was added to each well and incubated for 1 h at 23°C. Wells were washed 3 times with 150 l of phosphate-buffered saline (PBS) containing 10 M ZnCl 2 and 0.1% Tween 20. Normal and HRG-deficient plasma samples were dialyzed against TBS to remove citrate, reconstituted with 18 M ZnCl 2 , and then serially diluted up to 1600-fold with HBS containing 1% ovalbumin, 0.1% Tween 20, 2 mM CaCl 2 , 100 nM hirudin (Behring), and 18 M ZnCl 2 . To prepare a reference curve, HRG-fibrinogen complexes were generated by incubating varying concentrations of HRG (0 -1.6 g/ml) with fibrinogen (0 -0.8 g/ml) in a 2:1 molar ratio. 100 l of each of these mixtures or plasma was added to wells and incubated for 2 h at 23°C. After three sequential washes with PBS containing ZnCl 2 , HRG-fibrinogen complexes were detected in purified and plasma systems using a HRG-directed IgG-horseradish peroxidase (HRP) conjugate, prepared as specified by the manufacturer using a Lightninglink HRP conjugation kit (Cedarlane, Burlington, ON). After incubation for 1 h at 23°C, bound HRP conjugates were detected as specified by the supplier. Experiments were repeated three times.
Interaction of HRG with Fluorescein-labeled ␥Ј-Peptide in the Absence or Presence of ZnCl 2 -The binding of 1.1 M HRG to 0.05 M fluorescein-␥Ј-peptide was monitored by fluorescence in the absence or presence of Zn 2ϩ using a PerkinElmer Life Sciences LS 50B luminescence spectrometer (9). Briefly, the base-line fluorescence (I o ) was determined at excitation and emission wavelengths (slit widths) of 492 (5 nm) and 532 nm (2.5 nm), respectively, and an emission filter at 515 nm. The mixture was then titrated with aliquots of ZnCl 2 up to 20 M, and fluorescence intensity (I) was monitored after each addition. I/I o values were plotted against the concentration of ZnCl 2 , and the data were subjected to nonlinear regression analysis as previously described (9).
Statistical Analyses-Results are presented as the mean Ϯ S.D., and the significance of differences in the means was determined using t tests. For these analyses, p Ͻ 0.05 was considered statistically significant.

Interactions of HRG with ␥ A / A -or ␥ A /␥Ј-Fibrinogen-Al
though HRG has previously been shown to bind fibrinogen (11), the distinction between ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen binding has not been investigated. SPR was used to characterize the interaction between HRG and the two isoforms of fibrinogen. ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen was immobilized on separate flow cells of a CM4 sensor chip, and an unmodified flow cell served as the control. HRG did not bind either form of fibrinogen in the presence of Ca 2ϩ alone (data not shown). Because Zn 2ϩ facilitates the binding of ligands to HRG (14), we examined the effect of Zn 2ϩ on the HRG-fibrinogen interaction. The Req increased as a function of the Zn 2ϩ concentration and saturated at ϳ30 M ZnCl 2 ( Fig. 1). At each Zn 2ϩ concentration, more HRG bound to ␥ A /␥Ј-fibrinogen than to ␥ A /␥ A -fibrinogen. The apparent K d values of ZnCl 2 necessary to promote HRG binding to ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen were 4.9 Ϯ 0.2 and 1.2 Ϯ 0.8 M, respectively. The Zn 2ϩ dependence of the interaction of HRG with fibrin(ogen) is in apparent contradiction to previous work demonstrating HRG binding to fibrin without Zn 2ϩ addition (11). Because no binding was detected in the absence of Zn 2ϩ in our study, it is likely that the HRG preparation used in the previous report contained sufficient amounts of Zn 2ϩ to enable the interaction. For the remainder of the study, ZnCl 2 was used at a concentration of 20 M. This concentration was chosen because the physiological concentration of Zn 2ϩ in plasma ranges from 10 to 20 M (15, 31). As evidenced from the similarity in saturation profiles illustrated in Fig. 1, more than 80% of HRG is bound to both forms of fibrinogen at 20 M ZnCl 2 .
To determine the affinity of HRG for fibrinogen, increasing concentrations of HRG were sequentially injected into flow cells containing immobilized fibrinogen in the presence of 20 M ZnCl 2 . The sensograms reveal slow association and dissociation phases for HRG binding to both isoforms of fibrinogen (Fig. 2, A and B). K d values were obtained by kinetic analysis of the on-and off-rates by globally fitting the binding data. In the presence of Zn 2ϩ , HRG binds ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen with similar affinity, K d values of 8.8 Ϯ 0.9 and 8.9 Ϯ 3.9 nM, respectively (Table 1). These values agree with the previously reported K d of 6.7 nM determined by immunoassay (11). Plots of calculated Req values for each HRG concentration revealed saturable binding and demonstrated that binding of HRG to ␥ A /␥Ј-fibrinogen was significantly (p Ͻ 0.005) higher than that for ␥ A /␥ A -fibrinogen (Fig. 3), suggesting that 2-fold more HRG is bound to the ␥ A /␥Ј-fibrinogen isoform. The molar stoichiometries for the interaction of HRG with ␥ A /␥ A -fibrinogen and ␥ A /␥Ј-fibrinogen are 1.7 Ϯ 0.3 and 3.2 Ϯ 0.5, respectively (Table  1). Because the extended COOH terminus of the ␥Ј-chain is the feature that distinguishes ␥ A /␥Ј-fibrinogen from ␥ A /␥ A -fibrinogen, the increased binding of HRG to ␥ A /␥Ј-fibrinogen suggests that the ␥Ј-chain provides the additional HRG binding site.
Effect of the ␥Ј-Peptide-directed IgG on the Binding of HRG to ␥ A /␥ A -or ␥ A /␥Ј-Fibrinogen-To confirm that HRG binds specifically to the ␥Ј-chain of ␥ A /␥Ј-fibrin(ogen), we examined the effect of an affinity-purified IgG directed against this region. As an initial control, we demonstrated specific binding of the antibody to immobilized ␥ A /␥Ј-fibrinogen but not to ␥ A /␥ A -fibrinogen (data not shown). We next examined the effect of the antibody on HRG binding to ␥ A /␥Ј-or ␥ A /␥ A -fibrinogen. The ␥Ј-chain-directed antibody reduced HRG binding to ␥ A /␥Ј-fibrinogen to that observed with ␥ A /␥ A -fibrinogen (Fig. 3). As a control, a sheep non-immune IgG was used; the control IgG had no effect on the binding of HRG to fibrinogen (data not shown). Because the interaction of HRG with ␥ A /␥ A -fibrinogen is already of high affinity, the addition of the ␥Ј-peptide-di- rected IgG did not alter the affinity of HRG for ␥ A /␥Ј-fibrinogen. However, in the presence of the antibody, the molar stoichiometry of HRG for ␥ A /␥Ј-fibrinogen was similar to that for ␥ A /␥ A -fibrinogen (1.8 Ϯ 0.5 and 1.8 Ϯ 0.6, respectively), providing further support for the concept that the ␥Ј-chain on ␥ A /␥Ј-fibrinogen represents a unique HRG binding site. These data suggest that HRG has multiple high-affinity binding sites on fibrinogen and that blocking one binding site has minimal effects on the others.
HRG Binding to the ␥Ј-Peptide-To confirm that the ␥Ј-chain affords HRG an additional binding site, binding of HRG to synthetic ␥Ј-peptide was examined. First, the interaction of HRG with fluorescein-␥Ј-peptide was examined by fluorescence. Neither HRG nor Zn 2ϩ alone altered the fluorescence intensity of the f-␥Ј-peptide, suggesting that in the absence of Zn 2ϩ , there is no interaction. However, when Zn 2ϩ was titrated in the presence of HRG, the fluorescence intensity of fluorescein-␥Ј-peptide decreased in a dose-dependent and saturable manner (Fig. 4A), suggesting that Zn 2ϩ facilitates the binding of HRG to the peptide. Nonlinear regression analysis of the data revealed that the apparent K d for ZnCl 2 required to promote the HRG-fluorescein-␥Ј-peptide interaction was 9.1 Ϯ 4.5 M. This value is similar to the apparent K d of 1.2 M for Zn 2ϩmediated promotion of HRG binding to ␥ A /␥Ј-fibrinogen (Fig. 1).
To confirm these results, the interaction of HRG with ␥Ј-peptide was examined by SPR. Biotinylated ␥Ј-peptide was adsorbed to a streptavidin-modified flow cell, and HRG binding was monitored in the presence of 20 M ZnCl 2 . HRG bound immobilized ␥Ј-peptide in a concentration-dependent and saturable manner, and binding was blocked by the ␥Ј-peptidedirected IgG (Fig. 4B). Based on kinetic analysis, HRG binds the ␥Ј-peptide with a K d value of 0.79 Ϯ 0.01 nM in the presence of Zn 2ϩ ; there is no detectable binding in the absence of Zn 2ϩ (data not shown). These data offer independent confirmation that HRG binds to the COOH terminus of the ␥Ј-chain of ␥ A /␥Ј-fibrinogen in a Zn 2ϩ -dependent fashion.
Effect of the ␥Ј-Peptide-directed IgG on the Binding of HRG to ␥ A /␥ A -or ␥ A /␥Ј-Fibrin-Having shown that HRG binds fibrinogen with high affinity in a Zn 2ϩ -dependent fashion, we next used SPR to determine the affinity of HRG for fibrin. To convert immobilized fibrinogen to fibrin, flow cells were treated with thrombin (30). HRG bound to both isoforms of fibrin with affinities (Table 1) and Req values similar to those for fibrinogen, suggesting that HRG binding is unaltered when fibrinogen is converted to fibrin. To complement the SPR studies, we also assessed the binding of 125 I-HRG to fibrin clots. Clots contain-

TABLE 1 Dissociation constants and stoichiometries for the binding of HRG to ␥ A /␥ A -fibrin(ogen), ␥ A /␥-fibrin(ogen), or ␥-peptide
The binding of HRG to immobilized fibrin(ogen) isoforms or ␥Ј-peptide in the presence of 20 M Zn 2ϩ was quantified using SPR. K d values were determined by kinetic analysis of the data, and stoichiometries were calculated according to the BIAtechnology handbook.  ing varying concentrations of fibrinogen were prepared, and the amount of 125 I-HRG in the supernatants of compacted clots was determined. In keeping with our SPR data, 125 I-HRG did not bind to fibrin clots in the absence of Zn 2ϩ (data not shown). With 20 M Zn 2ϩ , 125 I-HRG bound to ␥ A /␥ A -and ␥ A /␥Ј-fibrin clots with K d values of 105.9 Ϯ 21.0 and 33.3 Ϯ 6.2 nM, respectively (Fig. 5A). Similar results were obtained in the reciprocal experiment using varying concentrations of 125 I-HRG and a fixed concentration of fibrinogen (data not shown). The binding constants obtained here are comparable with the previously reported K d value of 250 nM for the interaction of 125 I-HRG with fibrin formed from unfractionated fibrinogen, which consists of both isoforms of fibrinogen (11). Taken together, these results suggest that HRG binds fibrinogen and remains bound when fibrinogen is converted to fibrin.

K d Stoichiometry
Next, we examined the effect of varying concentrations of Fab fragments derived from the ␥Ј-peptide-directed IgG on 125 I-HRG binding to ␥ A /␥ A -or ␥ A /␥Ј-fibrin clots. Fab fragments had minimal effects on HRG binding to ␥ A /␥ A -fibrin clots (Fig. 5B). In contrast, at 4 M, the Fab fragments reduced HRG binding to ␥ A /␥Ј-fibrin clots by 50%, providing further evidence that HRG binds to the ␥Ј-chain of ␥ A /␥Ј-fibrin. A nonimmune sheep IgG was used as a control and demonstrated no effect.
HRG Binding to Fibrinogen Fragments-To localize the HRG binding domains on fibrinogen, binding of HRG to immobilized fibrinogen fragments was examined by SPR (data not shown). To avoid potential contribution of the ␥Ј-chain to HRG binding, fragments X, D, and E were prepared from ␥ A /␥ A -  fibrinogen. In the presence of Zn 2ϩ , HRG bound fragment X with a K d value of 63.5 Ϯ 11.8 nM, suggesting that the ␣C-domain of fibrinogen does not represent the primary HRG binding site. HRG also bound fragments D and E with high affinity in a Zn 2ϩ -dependent manner, with K d values of 8.0 Ϯ 1.3 and 23.3 Ϯ 2.2 nM, respectively. As a negative control, we demonstrated that HRG did not bind immobilized FPRck-thrombin in the absence or presence of Zn 2ϩ . As a positive control, we showed that FPRck-thrombin bound fragment E, with a K d value of 5.0 Ϯ 0.4 M, but did not bind fragment D, findings in agreement with previously published results (25). Therefore, these data suggest that, in addition to its interaction with the ␥Ј-chain, HRG binds to other unique sites on fibrinogen.
Diffusion of HRG from Fibrin Clots-To identify differences in the binding of HRG to ␥ A /␥ A -and ␥ A /␥Ј-fibrin clots, we monitored the dissociation of 125 I-HRG from preformed fibrin clots (Fig. 6). Diffusion in the presence of 2 M NaCl and 2 mM EDTA served as the base-line control because ionic and divalent cation-dependent interactions are abrogated (10,30). In the presence of Zn 2ϩ , the rates of diffusion of 125 I-HRG from ␥ A /␥ A -and ␥ A /␥Ј-fibrin clots were significantly (p Ͻ 0.05) slowed by 3-and 11-fold, respectively, compared with those determined in the presence of diethyldithiocarbamate trihydrate, a specific Zn 2ϩ chelator (32). Consistent with the concept that ␥ A /␥Ј-fibrin affords HRG an additional binding site, the rate of 125 I-HRG diffusion from ␥ A /␥Ј-fibrin was 3-fold slower than that from ␥ A /␥ A -fibrin (p Ͻ 0.05). Collectively, our results offer independent confirmation that the HRG-fibrinogen interaction is Zn 2ϩ -dependent and that there is an additional HRG binding site on ␥ A /␥Ј-fibrin.
Effect of FPRck-thrombin on HRG Binding to the ␥Ј-Chain-In addition to binding HRG, the ␥Ј-chain COOH extension binds thrombin (5,9). SPR was used to determine whether the two proteins compete for binding to this region. This approach exploits the fact that the dissociation rate of HRG from immobilized ␥Ј-peptide is much slower than that of thrombin. HRG (1 M) was injected into flow cells containing immobilized biotinylated ␥Ј-peptide, and the subsequent dissociation phase was monitored in the absence or presence of FPRck-thrombin or prothrombin in concentrations up to 8.0 M. Prothrombin was used as a negative control because it does not bind to the ␥Ј-peptide (33). Whereas FPRck-thrombin displaced HRG from the ␥Ј-peptide in a concentration-dependent manner, prothrombin did not (Fig. 7). These data confirm that thrombin and HRG compete for binding to the ␥Ј-peptide.
To determine whether the same was true with fibrin clots, we next examined the effect of increasing concentrations of FPRck-thrombin on 125 I-HRG binding to ␥ A /␥ A -or ␥ A /␥Јfibrin clots. At 10 M, FPRck-thrombin reduced the amount of HRG bound to ␥ A /␥Ј-fibrin clots by 90% but only reduced HRG bound to ␥ A /␥ A -fibrin clots by 15% (Fig. 8A). Similar results were obtained in the reciprocal competition experiments using varying concentrations of HRG and a fixed concentration of 125 I-YPRck-thrombin (Fig. 8B). Collectively, these data confirm that HRG and thrombin compete for binding to the ␥Ј-chain on ␥ A /␥Ј-fibrin(ogen) in a mutually exclusive fashion.

Detection of HRG-Fibrinogen Complexes in Plasma-Be-
cause HRG binds fibrinogen with high affinity, it was of interest to determine whether HRG-fibrinogen complexes can be detected in plasma by immunoassay. Plasma was first dialyzed to remove citrate and then reconstituted with 18 M ZnCl 2 . The concentration of HRG-fibrinogen complexes detected in normal plasma was 1 M, whereas no complexes were detected in HRG-deficient plasma. Because the plasma concentration of HRG ranges from 1.6 to 2 M (12, 13, 24), our findings suggest that in the presence of Zn 2ϩ ϳ50 -60% of HRG in plasma circulates in complex with fibrinogen.

DISCUSSION
Despite increasing evidence that the interaction of HRG with fibrin(ogen) plays an important role in innate immunity, inflammation, and coagulation, little is known about this inter- action. To address this gap, we characterized the binding of HRG to ␥ A /␥ A -and ␥ A /␥Ј-forms of fibrinogen and fibrin. In the presence of physiological concentrations of Zn 2ϩ , HRG binds ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen with similar affinities (K d values of ϳ9 nM). The affinities of HRG for both isoforms of fibrin are comparable to those for fibrinogen, suggesting that the conversion of fibrinogen to fibrin does not alter the binding of HRG. In addition to its interaction with the unique COOH terminus of the ␥Ј-chain of ␥ A /␥Ј-fibrin(ogen), HRG binds to fragments D and E, suggesting that there are several HRG binding sites on fibrinogen. In support of these observations, HRG-fibrinogen complexes were detected in plasma. Taken together, our findings suggest that HRG circulates in plasma bound to fibrinogen and that the complex remains intact when fibrinogen is converted to fibrin.
The absolute requirement for Zn 2ϩ to promote the HRGfibrinogen interaction underscores the regulatory role this cation may have in hemostasis. Zn 2ϩ is important for the binding of HRG to bacteria, cells, and hemostatic factors, such as glyco-saminoglycans, plasminogen, and factor XIIa (12,19,24,34). It is likely that there are still other HRG interactions that have been overlooked because of the widespread use of citrate as an anticoagulant. The total Zn 2ϩ concentration in plasma is ϳ20 M, and the majority of Zn 2ϩ is bound to albumin (35). Although the concentration of Zn 2ϩ that is not bound to proteins is only 0.5-1 M (31, 36), the free Zn 2ϩ concentration can increase under a variety of conditions. For example, platelets can secrete Zn 2ϩ when they are activated at sites of vascular injury (37). Furthermore, when fatty acids bind to albumin, they displace Zn 2ϩ from the protein, thereby providing another mechanism whereby the concentration of free Zn 2ϩ in plasma can be augmented (35). In addition to alterations in the Zn 2ϩ concentration, the activity of HRG can also be modulated by changes in pH, with optimal binding to ligands observed under  In both experiments 2 mM CaCl 2 plus 20 M ZnCl 2 were present and clotting was initiated with 10 nM thrombin. After incubation at 23°C for 45 min, fibrin was pelleted by centrifugation, and free 125 I-HRG or 125 I-YPRck-thrombin in the supernatant was used to calculate the bound fraction. The percent of fibrin-bound 125 I-HRG or 125 I-YPRck-thrombin is plotted versus the FPRck-thrombin or HRG concentration, respectively. The symbols represent the mean Ϯ S.D. of two experiments, each performed in duplicate, whereas the lines represent nonlinear regression analyses of the data. more acidic conditions. Consequently, the decrease in pH that occurs with reduced tissue perfusion may also enhance the affinity of HRG for both Zn 2ϩ and its ligands (14). Therefore, the current data extend the concept that Zn 2ϩ serves as a dynamic switch that regulates HRG activity and directs it to various pathways involved in hemostasis (14,34).
Although we show that Zn 2ϩ is essential for the interaction of HRG with fibrin(ogen), the mechanism by which Zn 2ϩ mediates this binding is unknown. Both HRG and fibrin(ogen) bind Zn 2ϩ (37,38). Therefore, Zn 2ϩ may act as a cofactor that simultaneously binds HRG and fibrin(ogen) in a coordinated fashion (39). Alternatively, Zn 2ϩ binding to HRG may induce conformational changes that facilitate its interaction with fibrin(ogen), a concept supported by the observation that Zn 2ϩ alters the conformation of a synthetic His-Pro-rich peptide (16). Furthermore, the intrinsic fluorescence of HRG decreases upon Zn 2ϩ titration (data not shown), providing additional support for the notion that Zn 2ϩ alters HRG conformation. Binding of Zn 2ϩ to the HRR domain of HRG indirectly promotes the interaction of heparan sulfate or plasminogen with the NH 2 -terminal cystatin domains of HRG, suggesting that Zn 2ϩ binding to the HRR domain modulates other domains (15,40). These data point to a mechanism whereby Zn 2ϩ modulates the structure and function of HRG.
Because HRG binds to the ␥Ј-chain on ␥ A /␥Ј-fibrin(ogen), novel roles of HRG can be envisioned. In addition to binding HRG, the COOH terminus of the ␥Ј-chain also binds thrombin and factor XIIIa and, by so doing, may modulate coagulation and fibrinolysis (10,41). The importance of fibrin as a reservoir of thrombin is highlighted by the observation that thrombi harvested at autopsy contain abundant amounts of active thrombin (42). Fibrin-bound thrombin has been postulated to be an important mediator of thrombus expansion because of its capacity to locally activate platelets and to promote its own generation through activation of factor V and factor VIII (43,44). The procoagulant activity of thrombin bound to ␥ A /␥Јfibrin appears to be greater than that of thrombin bound to ␥ A /␥ A -fibrin because the ␥Ј-chain mediates the high affinity interaction of thrombin with fibrin, and ␥ A /␥Ј-fibrin affords bound thrombin more protection from inhibition by the antithrombin-heparin complex than ␥ A /␥ A -fibrin (10). Further support for this concept comes from epidemiological studies that suggest that higher circulating levels of ␥ A /␥Ј-fibrinogen are associated with an increased risk of cardiovascular disease (45,46). HRG competes with thrombin for binding to the ␥Ј-chain as evidenced by its capacity to displace FPRck-thrombin from ␥Ј-peptide or from ␥ A /␥Ј-fibrin clots. Consequently, by displacing fibrin-bound thrombin, HRG may have antithrombotic properties. HRG may also compete with factor XIII, which is proposed to bind the ␥Ј-chain (47), thereby attenuating fibrin cross-linking and endowing HRG with pro-fibrinolytic activity. Studies in HRG-deficient mice support the concept that HRG affects hemostasis. HRG-deficient mice have a shorter prothrombin time and a longer bleeding time than their wild-type counterparts. In addition, thrombi formed in HRGdeficient mice are more susceptible to fibrinolysis than those generated in control mice (17). The contribution of the HRG interaction with the ␥Ј-chain to the anticoagulant and antifibrinolytic activities of HRG remains to be determined.
In addition to its role in hemostasis, fibrinogen appears to be an important mediator of inflammation and innate immunity because fibrinogen and fibrin stimulate peripheral blood mononuclear cells and vascular smooth muscle cells to synthesize proinflammatory cytokines (48 -50). Furthermore, bacteria trapped within fibrin clots are protected from host defenses and the action of antibiotics (51). However, because HRG has antimicrobial properties, the HRG-fibrin interaction promotes bacterial entrapment and killing (21). The capacity of HRG to enhance bacterial killing has been localized to its NH 2 -terminal and HRR domains. Thus, in the presence of Zn 2ϩ or when the pH is low, HRG induces lysis of the bacterial cell wall (19,20). These conditions can occur at sites of injury or wound-healing where activated platelets release Zn 2ϩ and local ischemia lowers pH (13,52). Activated platelets also release HRG (13) and, by so doing, may amplify the antimicrobial effect. Our observation that HRG binds fibrinogen and fibrin provides a regulatory mechanism by which HRG may mediate bacterial killing within a clot. The importance of HRG in modulating the inflammatory response has been confirmed in mouse models. Thus, compared with wild-type mice, HRG-deficient mice given subcutaneous injections of S. pyogenes exhibit attenuated abscess formation and reduced recruitment of neutrophils and macrophages to the site of infection (21), suggesting that HRG modulates the inflammatory response. In support of this concept, a synthetic peptide analog of the HRR of HRG attenuated the secretion of interleukin-8 from lipopolysaccharide-stimulated, CD14-transfected monocytes (53). These observations raise the possibility that HRG plays a part in the inflammatory response to infection.
Although HRG is an abundant plasma protein with multiple ligands, its physiological role remains unknown (for review, see Ref. 18). HRG is hypothesized to be an important effector of hemostasis and immunity. The ability of HRG to bind fibrinogen and displace thrombin from fibrin may provide an important link between these two systems and reveals a potential mechanism by which this could occur. Additional regulation may result from variations in the local pH and/or Zn 2ϩ concentration.