INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

HPRG¹ is a relatively abundant plasma glycoprotein (125 µg/ml or about 1.5 µM) (1). Its two N-terminal cystatin modules identify it as a member of the cystatin superfamily, along with kininogen and fetuin (2). Based on its interactions with various ligands, HPRG has been proposed to be a regulator of coagulation, fibrinolysis, and the immune response and to play a role in zinc transport. In vitro HPRG binds components of the coagulation and fibrinolytic systems, including heparin (3), plasminogen (4), and fibrinogen (5); cells, viz. T-cells (6), macrophages (7), and platelets (8); and small ligands, such as transition metal ions and heme (9). No hypothesis of the function of HPRG has been able to rationalize these numerous interactions or explain how the activity of HPRG may be regulated in vivo.

The most noteworthy feature of HPRG is a very high content of histidine, 13 mol %, concentrated in the central His-Pro-rich domain which contains a pentapeptide consensus sequence (GHHPH) repeated in tandem 12 times in human HPRG (10) and the sequences GHPPH and GPPPH repeated 18 times in the rabbit protein (11). Thus, the ionic charge of HPRG is exquisitely sensitive to pH in the range of pH 6-7, and an electrophoretic titration curve of rabbit HPRG showed that protonation of histidine side chains spans the physiological pH range, with the maximum change occurring between pH 6.5 and 7 (11).

Considering these observations, surprisingly little attention has been given to the effect of pH on the structure and function of HPRG. Histidine residues in the His-Pro-rich domain of HPRG mediate interactions with heparin (12), transition metals, and heme (13). Binding of HPRG to heparin and the subsequent neutralization of heparin's anticoagulant effect have obvious pharmacological interest. An analysis of proteins from normal human plasma that bound to immobilized heparin indicated that HPRG was the only protein likely to inhibit anticoagulant activity by direct competition with antithrombin (14), although several plasma proteins display this ability in purified systems. Interaction of heparin with heparin-binding proteins is largely electrostatic but requires lysine or arginine side chains in other known cases. It is thus noteworthy that histidine residues are responsible for HPRG binding to heparin, and this leads to the immediate prediction that protonation of histidine side chains at low pH or binding of transition metals such as zinc, which also bestows positive charge on the His-Pro-rich domain, will strengthen heparin binding to HPRG. Indeed, inhibition by HPRG of the heparin-catalyzed thrombin-antithrombin reaction is modulated by pH (15) or zinc (16), and the stoichiometry of the heparin-HPRG interaction is shifted from 1:1 to 1:2 at low pH or in the presence of zinc and copper (12).

Here we characterize directly and quantitatively the modulation by pH and transition metals of HPRG-glycosaminoglycan (GAG) interactions, and we propose a new paradigm of the function of HPRG based on these data. We suggest that the unique, conserved His-Pro-rich domain is a pH sensor, providing a plausible mechanism for regulating HPRG's activity and an approach to understanding its physiological role. In this model, when the local pH drops (e.g. in ischemia or hypoxia) histidine side chains of HPRG are protonated, the His-Pro-rich domain becomes positively charged, and hence HPRG binds strongly to negatively charged GAGs on cell surfaces. As a consequence ligands of HPRG such as plasminogen can be co-immobilized at the cell surface, and other GAG-binding proteins like antithrombin may be displaced.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Reagents-- Porcine intestinal mucosa heparin, dermatan sulfate, heparan sulfate, and chondroitin sulfates A and C were purchased from Sigma, and low molecular mass (5 kDa) heparin was from Calbiochem. Bovine serum albumin (BSA)-heparin conjugate (Sigma) contained an average of 3.4 mol of heparin/mol of BSA coupled to the amino groups of the protein via the reducing ends. Bovine aorta dermatan sulfate proteoglycan (DSPG) was purchased from Becton Dickinson. Other commercial reagents were the highest quality available, and all buffers and stock solutions were prepared with water treated with a Barnstead Nanopure system.

Preparation of Immobilized GAG Surfaces-- GAGs were coupled to hydrophobic, flat amino-silane cuvettes (Affinity Sensors, Paramus, NJ). Before coupling, these cuvettes exhibited no nonspecific binding of HPRG under any experimental condition used in this paper, in contrast to virgin carboxyl-dextran cuvettes which bound HPRG nonspecifically below pH 6.3 and therefore were not used. Heparin-BSA or DSPG at 0.2 mg/ml in 15 mM sodium phosphate, pH 7.3, 150 mM sodium chloride were allowed to react for 30 min with cuvettes previously activated with polymeric glutaraldehyde as suggested by the manufacturer. Coupling using the lysine side chains of the protein moiety optimized immobilization since the GAG moiety was not affected and ligand heterogeneity was reduced. The binding capacity of GAG surfaces for HPRG varied with pH. For instance, although there was no detectable HPRG binding to heparin-BSA at pH 8, maximal binding of an estimated 1.0-1.8 molecules of HPRG per heparin chain was achieved at pH 5.3 (the lower value assumes no heparin contribution to the refractive index of the conjugate and the higher value assumes an equal contribution of heparin and albumin).

Binding Experiments-- Binding data were collected with a thermostatted IAsys resonant mirror biosensor (Affinity Sensors). All experiments were carried out at 23 °C and maximum stirring rate in Pipes-buffered saline (PiBS), 10 mM Pipes, 140 mM sodium chloride, 0.005% Tween 20. The buffer also contained 0.2 mM NTA except when other chelators were used as described in the text or figure legends. The various metal salts were added to yield the calculated free metal concentrations noted (see below). Pipes was chosen since it can be used in the pH range 5.5-7.5 and does not interact appreciably with transition metals. After each binding/dissociation cycle, the GAG surface was regenerated by a brief wash (5-15 s) with 0.1 M sodium borate, pH 9.5, 1 M sodium chloride, 0.2 mM EDTA. Further demonstrating the specificity of interaction, excess soluble heparin completely inhibited binding of HPRG to immobilized heparin. Most experiments were carried out using both human and rabbit HPRG, and no significant differences were found. The results shown are those obtained with the human protein, unless otherwise indicated.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

The Effect of pH on HPRG Binding to Immobilized Heparin-- The amount of HPRG bound at equilibrium to immobilized heparin at physiological ionic strength was markedly influenced by pH (Fig. 1A). A plot of the amount of HPRG bound at equilibrium versus pH resembles an acid-base titration curve with human HPRG having a pK_a of 6.8, near that of histidine (Fig. 1B). HPRG treated with excess diethylpyrocarbonate (DEPC) to chemically modify histidine residues (22) does not bind, supporting the importance of histidine in the interaction with heparin (12). Binding of heparin-BSA to immobilized HPRG also showed marked pH dependence with a mid-point at pH 6.5 (not shown). In contrast, binding of antithrombin to heparin was much less dependent on pH (Fig. 1B) and exhibited a lower maximum binding, probably due to the preference of antithrombin for a specific pentasaccharide sequence, whereas HPRG is more indiscriminate. To examine HPRG binding to heparin under conditions more like those found in vivo, rabbit serum (diluted 20-fold) was added to the immobilized heparin cuvette at different pH values, followed by affinity purified anti-rabbit HPRG IgG (which does not bind in the absence of serum) (Fig. 2). An increase in signal due to the specific binding of IgG to HPRG already bound to heparin is observed at pH 6.7 and to a limited extent at, but not above, pH 7.0, in good agreement with the results obtained with the purified system (Fig. 1B).

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Effect of pH on the interaction between HPRG and immobilized heparin. A, binding of 0.1 µM HPRG to immobilized heparin-BSA in PIBS-NTA at different pH values. B, pH dependence of the amount of protein bound to heparin at equilibrium for human HPRG (filled circles), rabbit HPRG (open circles), DEPC-treated rabbit HPRG (squares), and antithrombin (triangles). Binding was measured using a resonant mirror biosensor as described under "Experimental Procedures." The units refer to the change in refractive index due to binding which is recorded in arc seconds.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Binding of rabbit serum proteins to immobilized heparin at different pH values. Serum was diluted 20-fold in PiBS-NTA, and then affinity purified anti-rabbit HPRG IgG was added at the time indicated by the arrow. Binding to immobilized heparin-BSA was measured as described in Fig. 1.

The Effect of Transition Metals on HPRG Binding to Immobilized Heparin-- The influence of various transition metal ions on HPRG binding to heparin in the presence of iminodiacetate buffer at physiological pH is depicted in Fig. 3A. Transition metals bound by HPRG augment the affinity of the protein for heparin at physiological pH. Zinc and copper, 10-20 µM in plasma, are more effective than nickel or cobalt, whereas metals not bound by HPRG (9) like manganese, calcium, and magnesium (not shown) do not promote HPRG binding to heparin. The effects of metals on the binding of HPRG to heparin were also assessed in the presence of citrate or NTA (Fig. 3B), since the formation of a 1:1 metal-chelator complex simplifies the computation of free metal concentration and since the amount of HPRG-metal complex formed ultimately depends on the concentration of free metal and the affinity of the interaction. The importance of the free metal concentration is demonstrated by the position of the curves representing zinc-mediated binding. Although the Zn(II) concentrations yielding equal HPRG binding in the different buffers were within 0.3 log units (a factor of 2), the corresponding total zinc concentrations differed by more than 60-fold (between 1.2 and 75 µM). It is also notable that the concentration of free metal required to produce half-maximal binding of HPRG to heparin is about 1 nM for copper and 10 nM for zinc, well below the previously published affinity of HPRG for these metals, 10 µM (23). We attribute this difference to the electrostatic concentration of divalent cations in the vicinity of the negatively charged heparin surface.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Effect of transition metals on the interaction between HPRG and immobilized heparin. A, binding of 0.1 µM HPRG to immobilized heparin-BSA was measured in PiBS, pH 7.3, containing 0.2 mM iminodiacetate and divalent transition metals as shown. B, binding of 0.1 µM HPRG to immobilized heparin in the presence of divalent transition metals was measured in PiBS, pH 7.3, containing 1 mM citrate (open symbols), 0.2 mM NTA (filled symbols), or 0.2 mM NTA with 1 mM calcium chloride (dotted symbols). Binding was assessed as in Fig. 1.

Comparison between pH- and Metal-mediated HPRG Binding to Heparin-- Treatment of HPRG with excess DEPC abolishes both the pH- and zinc-dependent interactions, indicating that histidine residues are necessary for both. In fact, the two binding modes appear to be facets of the same phenomenon, as suggested by experiments carried out at various pH and zinc concentrations (Fig. 4). At pH 6.5, when most of the histidine residues are protonated, binding of HPRG to heparin is maximal, and the presence of zinc has no further effect. Conversely, at pH 7.4 most of the histidine residues are unprotonated, and binding to heparin is directly dependent on zinc concentration. At intermediate pH values, both protonation of histidine and binding of zinc each contribute to the net positive charge of HPRG, and the overall effect is a shift of the binding curves to more alkaline pH which is proportional to the metal concentration.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Effect of zinc on the pH-dependent binding of HPRG to immobilized heparin. Zinc chloride (0-50 µM) was added to a pH-independent metal buffer system (0.2 mM NTA, 1 mM calcium chloride) to produce concentrations of free zinc between 0 and 33 nM. Binding to immobilized heparin-BSA was carried out as in Fig. 1.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Kinetics and thermodynamics of the HPRG-immobilized heparin interaction. A, sensorgram of an equilibrium titration of immobilized heparin-BSA with human HPRG in PiBS-NTA, pH 6.8. Inset, amounts of HPRG bound at equilibrium at pH 6.8 (circles) or pH 7.25 in the absence (triangles) or presence (squares) of 12.6 nM free zinc. The curves represent the nonlinear, least squares fits of experimental data to a binding isotherm. B, pseudo-first order on-rates for the HPRG-heparin interaction at pH 6.8 (circles) or at pH 7.3 in the presence of 12.6 nM free zinc (squares), assuming binding to a single class of sites. Progress curves were analyzed with the IAsys FASTfit software provided by the manufacturer, assuming a single exponential association. The fit was not significantly improved by using a double exponential association model. Binding units are as in Fig. 1.

1

1

2

1

Inhibition of HPRG Binding to Heparin by Other GAGs-- The relative affinities of HPRG for various GAGs were assessed by competition experiments, in which HPRG was added to solutions containing increasing concentrations of GAGs in PIBS buffer at pH 6.8 (Fig. 6). The concentration of GAG at which 50% of HPRG binding to immobilized heparin is inhibited is proportional to the affinity of HPRG for the added GAG. It can be shown that the relative amount of HPRG bound is given by the following: B/B_max = (1 + [HPRG]/K_hep)/(1 + [HPRG]/K_hep + [GAG]/K_GAG). Since the concentration of HPRG used (50 nM) was approximately equal to its affinity for immobilized heparin, K_hep, 50% inhibition by a GAG was exerted at a concentration about twice the affinity of HPRG for that GAG, K_GAG. Consistent with the electrostatic nature of the HPRG-GAG interaction, heparin, a highly sulfated form of heparan sulfate, binds to HPRG more strongly than any other GAG tested. The inhibition experiments yielded an affinity of HPRG for heparin of about 60 nM, comparable with the K_d obtained directly, and the affinity for the low molecular weight heparin was somewhat lower, about 300 nM. Dermatan sulfate and heparan sulfate (with average mass 12 and 10 kDa, respectively) had affinities of 8 and 16 µM, respectively, whereas chondroitin sulfates A and C did not interact significantly with HPRG, despite having similar charge densities. The GAGs bound by HPRG (heparin, dermatan sulfate, and heparan sulfate) all contain iduronic acid, a residue thought to enhance protein binding by increasing conformational flexibility of GAGs (27).

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Inhibition of the HPRG-heparin interaction by soluble GAGs. Equilibrium binding of 50 nM HPRG to immobilized heparin-BSA was measured in PiBS-NTA, pH 6.8, in the presence of heparin (filled squares), low molecular weight heparin (filled circles), heparan sulfate (open squares), dermatan sulfate (open triangles), and chondroitin sulfates A (open circles) and C (filled triangles). Binding was assessed as in Fig. 1.

HPRG Binding to Dermatan Sulfate-- In view of the strong pH and zinc dependence of HPRG binding to heparin, how these factors affect the interaction between HPRG and GAGs which occur on cell surfaces was of interest. Dermatan sulfate, the most effective competitor of HPRG binding to heparin tested, was immobilized onto amino silane cuvettes either by reductive amination via its reducing ends or via lysine side chains in dermatan sulfate proteoglygan (DSPG). As shown in Fig. 7A, the interaction between HPRG and DSPG was also pH-dependent, with the mid-points of the titration curve at pH 6.5 for human HPRG and at pH 6.3 for rabbit HPRG both being somewhat lower than seen for the HPRG-heparin interaction (pH 6.8, Fig. 1B). Binding of both human and rabbit HPRG to immobilized dermatan sulfate exhibited a mid-point at pH 6.5, and the titration curve was somewhat broader than for DSPG (data not shown). The acidic shift of the titration curves of HPRG binding to a dermatan sulfate surface relative to a heparin surface is not surprising in view of the electrostatic nature of the interaction: since dermatan sulfate has a lower negative charge density than heparin, a larger increase in the positive charge of HPRG is needed and is achieved at lower pH.

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Binding of HPRG to immobilized dermatan sulfate proteoglycan. A, effect of zinc on the pH-dependent binding of 0.25 µM human HPRG (open symbols) to immobilized DSPG. Zinc chloride (0-100 µM) was added to a pH-independent metal buffer (0.2 mM NTA, 1 mM calcium chloride) to yield concentrations of free zinc between 0 and 100 nM. The pH dependence for the binding of rabbit HPRG (filled circles) to DSPG in the absence of zinc is shown for comparison. B, cooperative binding of rabbit HPRG (filled symbols) or human HPRG (open symbols) to immobilized DSPG (circles) or dermatan sulfate (squares) was measured in PiBS-NTA, pH 6.3. Binding units are as in Fig. 1.

"Sandwich" Binding of Plasminogen to Heparin-associated HPRG-- An important question is whether HPRG bound to surface GAGs may in turn bind and co-immobilize other ligands. For instance, HPRG immobilized onto nickel chelate plates, but not soluble HPRG, strongly accelerates plasminogen activation by tissue plasminogen activator (28). Fig. 8A shows that plasminogen indeed specifically associates with heparin-bound HPRG in a manner that is completely and selectively inhibited by -aminocaproic acid (an inhibitor of the plasminogen-HPRG interaction). These GAG-HPRG-plasminogen "sandwiches" can form since the binding sites of HPRG for heparin (the His-Pro-rich domain) and plasminogen (the N/C domain) are different (28). Control experiments confirmed that plasminogen binding to the heparin surface is mediated by HPRG, since plasminogen did not bind to heparin in the absence of HPRG nor under conditions that precluded binding of HPRG to heparin (viz. pH above 7 with no transition metals). Plasminogen also binds to DSPG-associated HPRG (Fig. 8B) but not to HPRG with reductively methylated lysine residues. This modification abrogates the interaction of HPRG with plasminogen (28) but does not affect that with DSPG (Fig. 8B).

View larger version (22K): [in this window] [in a new window]   Fig. 8.   Interaction between plasminogen and HPRG bound to immobilized heparin. A, HPRG (0.1 µM) was allowed to bind to immobilized heparin-BSA at different pH values until equilibrium was reached. Glu-plasminogen (PLG, 1 µM) and -aminocaproic acid (-ACA, 2 mM) were sequentially added at the times indicated by the arrows. B, rabbit HPRG (upper trace) and reductively methylated HPRG (lower trace) were allowed to equilibrate on immobilized DSPG at pH 6.3, and then Glu-plasminogen aliquots (0.4 µM) were added at the times indicated by arrows. Specific dissociation of plasminogen from the sandwich was induced with 1 mM -aminocaproic acid. Binding units are as given in Fig. 1.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

An immediate consequence of HPRG's binding to heparin or GAGs is competition with other heparin-binding proteins. Neutralization by HPRG of the anticoagulant activity of heparin by competition with antithrombin has long been suggested to be the primary, if not the sole, effect of the HPRG-heparin interaction, and this interaction has been extensively studied due to its pharmacological relevance. Heparin is used therapeutically at concentrations (1-3 units/ml, 5-20 µg/ml, or about 0.3-1.5 µM) close to those of HPRG (1.5-2 µM), and the amount of free heparin available to exert its anticoagulant effect is strongly influenced both by the total HPRG concentration in plasma and by conditions that affect the HPRG-heparin interaction. The extreme dependence of the heparin-HPRG interaction on pH and to a somewhat lesser degree on the free zinc concentration creates a direct link between these factors and heparin bio-availability. High circulating levels of HPRG are associated with venous thrombosis (29-31), a condition characterized by the formation of a fibrin clot in areas of low blood flow and preventively treated with heparin. Our data strengthen this connection, since stagnant blood favors the development of local acidosis and thus the neutralization by HPRG of natural heparin-like molecules.

The pH dependence of HPRG-GAG interactions gains significance from the fact that the local pH can drop as much as one pH unit in hypoxia or ischemia (32, 33) or one-half pH unit in inflammation (34) due to lactic acidosis. Moreover, postoperative complications are often accompanied by severe metabolic acidosis (35). We propose that HPRG functions mainly as a plasma pH sensor that binds to heparin or cell-surface GAGs only when the local pH decreases or local free metal concentration increases. HPRG is in a unique position among plasma proteins to perform this function due to its high content of histidine in the His-Pro-rich domain. Of the plasma proteins, only kininogen possesses a similar domain that is His-Gly-Lys-rich and that also mediates binding to negatively charged surfaces such as kaolin and acidic phospholipids. However, the permanently charged lysine residues make the kininogen domain much less dependent on pH than HPRG. HPRG is the first plasma protein to which a pH sensor role can be ascribed, although an analogous pH transducer role has been proposed for another histidine-rich protein, hisactophilin from Dictyostelium, whose interaction with actin and the negatively charged inner plasma membrane is regulated by the intracellular pH (36).

Zinc also alters the heparin-HPRG interaction (see above and Refs. 12 and 16). However, the effect is more difficult to evaluate. One uncertainty is how much zinc is endogenously bound to HPRG. The normal plasma concentration of free zinc, about 0.2 nM (37), is too low to promote HPRG binding to heparin at physiological pH, since 1-2 nM free zinc would be needed to affect HPRG-heparin interaction significantly at physiological pH, and the interaction between HPRG and GAGs such as dermatan sulfate is even less influenced by zinc. Although several pathophysiological processes affect local pH, the plasma concentration of available zinc is more stable and depends on the total concentrations of zinc and its ligands and on their affinities. The local concentration of zinc might increase upon platelet stimulation, since total zinc is enriched 30-60-fold relative to plasma in platelet -granules (38). However, it seems most likely that zinc by itself does not regulate the interaction between HPRG and GAGs but rather acts synergistically with local acidosis to promote surface binding. For example, zinc shifts the pH dependence of HPRG binding to heparin toward physiological pH values so that smaller decreases in local pH would be required to achieve equal binding in the presence of zinc.

The interaction between HPRG and dermatan sulfate and the dependence on pH and zinc were also characterized quantitatively. Since heparin in vivo is normally confined inside mast cell granules, dermatan sulfate and heparan sulfate proteoglycans, natural constituents of the endothelial cell surface, are the most likely physiological binding partners for HPRG. Two studies of plasma proteins bound to cell-surface heparan sulfate proteoglycans (39) and immobilized dermatan sulfate (40) reported HPRG to be a major plasma ligand of these GAGs. However, elsewhere HPRG was found to be ineffective in inhibiting the interaction of antithrombin III with endothelial cell heparin-like compounds (41). This apparent discrepancy is readily explained by the present results showing that binding of HPRG to bovine aorta dermatan sulfate proteoglycan occurs only under acidic conditions or in the presence of nanomolar amounts of free zinc. It is possible that binding of HPRG to other cells such as macrophages, T-cells, or platelets is also mediated by cell-surface GAGs. In a recent report supporting this view, binding of HPRG to a T-cell line required the presence of zinc and was inhibited by heparin (42). Of special interest is the finding that HPRG can function as an adapter molecule and co-immobilize its ligands, e.g. plasminogen, on a heparin-coated surface. Since plasminogen activation by tissue plasminogen activator is stimulated by immobilized, but not soluble, HPRG (28), cell-surface GAGs may adsorb HPRG in vivo when the local pH is lowered, thereby stimulating plasminogen activation in the affected area. Preliminary results have indeed shown that plasminogen activation is accelerated in the presence of HPRG bound to immobilized GAGs at acidic pH.³ This reinforces the aforementioned analogy between HPRG and its cousin kininogen, which binds to negatively charged surfaces via its His-Gly-Lys-rich domain and co-localizes kallikrein and factor XI to enable efficient triggering by factor XII of the intrinsic coagulation pathway (47).

Finally, HPRG can be described as a facultative polycation since its His-Pro-rich domain acquires a positive charge upon protonation, and this may have important physiological consequences in addition to the already well described competition with antithrombin and heparin cofactor II for heparin. For example, cationic polyelectrolytes have numerous biological effects, including antimicrobial activity, enhancement of vascular permeability, modulation of leukocyte adherence, chemotaxis, and phagocytosis (43). In fact, polyhistidine itself is an opsonic agent (44) and a potent inducer of superoxide generation and respiratory burst in polymorphonuclear leukocytes (45, 46). It seems likely that many of the biological actions of HPRG are related to its polycationic nature and thus are regulated by pH and expressed only under conditions that produce local acidosis. HPRG could conceivably be used to target enzymes or other drugs to acidotic regions, e.g. ischemic tissues. In vitro insoluble complexes of polyhistidine and catalase can bind to endothelial cells and protect them against injury mediated by hydrogen peroxide or activated polymorphonuclear leukocytes (48).

In conclusion, the results presented here provide a new paradigm of the function of HPRG and form the basis for ongoing research on HPRG-GAG interactions.