Decorsin A POTENT GLYCOPROTEIN IIb-IIIa ANTAGONIST AND PLATELET AGGREGATION INHIBITOR FROM THE LEECH MACROBDELLA DECORA*

The purification, and characterization of decorsin, a protein isolated from the North American leech Macrobdella platelet glycoprotein

The discovery, purification, and characterization of decorsin, a protein isolated from the North American leech Macrobdella decora, are described. Decorsin acts as an antagonist of platelet glycoprotein 1%IIIa (GPIIb-IIIa), and is a potent inhibitor of platelet aggregation.
The protein was purified to apparent homogeneity from crude whole leech extracts by treatment with trifluoroacetic acid followed by GPIIb-IIIa affinity chromatography and C& reverse-phase high performance liquid chromatography. Decorsin was also isolated from a solution of leech ingestate by treatment with trifluoroacetic acid followed by Cl8 reversephase high performance liquid chromatography. The primary sequence of decorsin indicates that the protein is 39 amino acids long and contains 6 cysteine and 6 proline residues, as well as the sequence Arg-Gly-Asp, (RGD), a proposed recognition site of many adhesion proteins.
A molecular mass of 4379 was obtained by fast atom bombardment mass spectrometry and is consistent with the mass calculated from the observed sequence.
Evidence for an N-3 isoform, lacking the first 3 amino-terminal residues is also presented. Both decorsin and the N-3 isoform inhibit GP IIb-IIIa binding to immobilized fibrinogen with an ICso of - Platelet aggregation plays a fundamental role in hemostasis (1). It is mediated by the interaction of fibrinogen with the platelet membrane glycoprotein IIb-IIIa (GPIIb-IIIa)' (2,3), a member of the integrin family of cell adhesion receptors (4,5). This interaction appears to be the final common step of aggregation that is induced by all platelet aggregation agonists. Since arterial thrombotic disease is mediated by platelet adhesion and aggregation, inhibitors of the fibrinogen/GPIIb-IIIa interaction may prove to be very effective agents for therapeutic intervention in thrombotic disease (6). To date the most potent inhibitors of platelet aggregation acting via antagonism of the GPIIb-IIIa receptor which have been reported have been purified or derived from natural sources. These include both the Aat95-91 and Aa572-574 RGD (7,8) and ~400-411 (9) region peptides derived from the sequence of the LY and y chains of fibrinogen, respectively (lo), anti-GPIIb-IIIa monoclonal antibodies (ll-13), and several potent inhibitors recently isolated from snake venoms (14-17). The venom proteins are highly homologous to one another and constitute a family of related proteins that interact directly with GPIIb-IIIa, thereby blocking fibrinogen binding (17).
Leeches have long been known to possess agents that affect hemostasis (18,19). Several proteins have been isolated from leeches that affect hemostasis by various mechanisms. These include hirudin (20, 21), a potent thrombin inhibitor from Hirudo medicinalis which contains 65 residues, antistasin (22), a factor Xa inhibitor from Huementuriu officinalis containing 119 residues as well as a similar factor Xa inhibitor (23) from Haementuria ghilianii, and hementin (24), a fibrinolytic enzyme from Haementaria ghilianii.
We hypothesized that hematophagous leeches might also contain antithrombotic agents that act via inhibition of fibrinogen/GPIIb-IIIa binding. After screening a number of crude leech homogenates in a solid-phase ELISA' that detects inhibition of fibrinogen binding to GPIIb-IIIa, the North American leech Mucrobdella decoru was chosen for further characterization. In this paper, we report on the discovery, purification and characterization of decorsin, a potent GPIIb-IIIa protein antagonist isolated from h4. decoru.
The protein has been purified from crude whole leech homogenate, as well as from a solution of leech ingestate, reported to contain platelet aggregation inhibitors (25). We describe both the primary sequence of the protein as well as the effectiveness of decorsin as an inhibitor of platelet aggregation. To our knowledge, decorsin is the first GPIIb-IIIa antagonist isolated from leeches. A total of approximately 300 ml of each buffer was used in this cycle washing.
Finally, the coupled GPIIb-IIIa resin was washed with 100 ml of 50 mM Tris, pH 7.5, containing 2 mM CaC& and stored at 4 "C prior to use.
Extraction and Trifhoroacetic Acid Precipitation-500 g of leeches were homogenized in 400 ml of TACTS buffer in a Waring blendor. The homogenate was spun at 11,000 X g for 20 min using a Sorvall RCSB refrigerated centrifuge.
The supernatant was saved, and the pellets were extracted again with 225 ml of TACTS buffer in a Tissuemizer (Tekmar, Cincinnati, OH) tissue homogenizer. This homogenate was spun as before, and the supernatants from both extractions were pooled to form the crude extract. The crude extract was stirred magnetically at room temperature while 50% trifiuoroacetic acid was added dropwise to a final concentration of 1%. Following the addition of trifluoroacetic acid, the mixture was stirred at room temperature for 5 min and then spun at 23,000 x g for 20 min. The pellets were discarded, and the supernatant was neutralized to pH 7.6 by the dropwise addition of ammonium hydroxide.
The neutralized extract was spun at 23,000 X g for 45 min; the supernatant was then used directly or stored frozen at -80 "C prior to use.
GPIIb-IlIa Affinity Chromatography-The neutralized trifluoroacetic acid supernatant pool containing 2 mM CaCl* was loaded onto a 0.5 x 7 cm' GPIIb-IIIa affinity column prepared as described above. The column was equilibrated in 50 mM Tris, pH 7.5, containing 2 mM CaCl*. The sample was loaded at a flow rate of 0.5 ml/min, and 25ml fractions were collected. A Vydac C,, (5 Nrn, 4.6 x 250-mm) column was equilibrated in 15% acetonitrile containing 0.1% trifluoroacetic acid. Decorsin and N-3 decorsin were eluted with a linear acetonitrile gradient (15-25%, 0.3%/min) at a flow rate of 1 ml/min. The gradient was started at the arrow and is illustrated with a dashed line; the absorbance at 214 nm is represented by the solid line.

AND DISCUSSION
Purification-Decorsin was purified to apparent homogeneity from a crude homogenate extract of whole M. decoru leeches by treatment with trifluoroacetic acid followed by GPIIb-IIIa affinity chromatography, and Cl8 reverse-phase HPLC. The activity observed in the crude homogenate was found to be soluble in 1% trifluoroacetic acid. Decorsin bound to the GPIIb-IIIa affinity resin and was eluted by treatment with EDTA, which is known to disrupt the calcium dependent GPIIb-IIIa complex (34). The affinity resin could be regenerated by treatment with Ca2+, however, the column was reuseable only two to three times, losing binding capacity with each use. Presumably, this loss of binding ability is due to irreversible dissociation of the GPIIb-IIIa complex, although the presence of a GPIIb-IIIa-binding molecule in the trifluoroacetic acid supernatants that is not eluted by EDTA cannot be excluded as a possibility.
The final purification of decorsin was carried out by Cl8 reverse-phase HPLC using pooled affinity eluates. HPLC fractions containing the activity were rechromatographed using a 0.3%/min acetonitrile gradient (Fig. 1); a second peak of activity eluted earlier than decorsin and is discussed below. The final recovery of decorsin was -10 pg; this represents less than 10% of the decorsin loaded onto the affinity column." The remainder of the decorsin was found in the flow through.
We hypothesized that decorsin may prevent blood from clotting during either feeding and/or storage of ingested blood. Inhibition of platelet aggregation has been previously observed from the dilute saliva of H. medicinalis (25). Therefore, in addition to the purification of decorsin from crude leech homogenate, we also isolated the protein from a solution of leech ingestate; the ingested arginine-saline solution was treated with trifluoroacetic acid followed by Cl8 reverse-phase HPLC. The relative level of contaminating proteins in the ingestate was far less, based on SDS-polyacrylamide gel electrophoresis (data not shown), than in whole leech homogenate. This allowed the elimination of the affinity step, which, although quite useful during the purification of decorsin from crude homogenate, requires relatively large amounts of GPIIb-IIIa. Approximately 10 pg of decorsin was isolated from 45 ml of leech ingestate. Protein isolated in this manner was idena Since both the size and affinity of decorsin for GPIIb-IIIa were unknown initially, the entire trifluoroacetic acid supernatant was applied to the affinity column.
Based on a molecular weight of 4379 for decorsin (uide infru) and assuming a 1:l stoichiometry, -145 pg represents the theoretical maximum attainable from the 8 mg of GPIIb-IIIa on the resin, assuming 100% correct orientation of the receptor. tical to decorsin purified from whole leech homogenate based on molecular weight (FAB-MS), amino acid composition, and specific activity in the fibrinogen/GPIIb-IIIa ELISA (see below).

Amino
Acid Sequence, Purity, and Molecular Weight-The entire amino acid sequence of decorsin was determined from replicate runs of the reduced protein by automated sequential Edman degradation.
The protein consists of 39 amino acids (Table I) and contains the RGD sequence, a well-known recognition sequence in many adhesion proteins (3), near its C terminus. In addition decorsin is very rich in cysteine and proline (6 each). The calculated molecular weight of the reduced protein is 4384; the calculated p1 is 4.45 (35).
During the sequencing of purified decorsin, two minor additional sequences were observed (data not shown); these sequences were identical to decorsin except that they lacked the first 2 or 3 amino-terminal residues. The separation of decorsin and the N-3 isoform is shown in Fig. 1. Amino acid composition of peak B was consistent with N-3 decorsin (Table II). These isoforms may represent native structures or more likely, result from proteolytic cleavage of decorsin. A variety of leech species have been investigated for the presence of endogenous proteases, revealing similar exopeptidase activ-  ity levels in different species, and the absence of endopeptidases (36). It is possible that decorsin itself is derived from proteolytic processing of a larger protein; proof of this would require cloning and sequencing the decorsin gene. Analysis of purified decorsin by SDS-polyacrylamide gel electrophoresis in the presence of dithiothreitol revealed a single protein band at a molecular mass ~14,400 daltons (Fig.  2). The purified protein was analyzed by FAB-MS and found to contain only one major component with a molecular mass of 4379 (based on the observed M + H' of 4379.9). A doubly charged ion was also observed which confirmed this mass (M + 2H+ = 2189.9 which corresponds to M = 4377.8). The observed mass is within 1 Da of the mass calculated from the amino acid sequence, assuming that three disulfide bonds are present in the native protein.
Amino Acid Composition-The amino acid composition of both decorsin and N-3 decorsin was determined on native protein acid hydrolysates. The data matches the composition predicted by amino-terminal sequencing, with the exception of cysteines, which were partially destroyed during hydrolysis (Table II).

GPIIb-Illa
Antagonist/Platelet Aggregation Inhibition Actiuity-Evidence that decorsin binds to GPIIb-IIIa is based on experiments with immobilized GPIIb-IIIa affinity resin. After binding, decorsin is eluted from the resin by treatment with EDTA, which dissociates the calcium-dependent heterodimeric structure of . In addition the interaction of GPIIb-IIIa with immobilized fibrinogen as measured by the solid-phase fibrinogen/GPIIb-IIIa ELISA is directly inhibited by decorsin. The specific activity (IC&) of both decorsin and the N-3 isoform was determined in the ELISA based on the dose-dependent response that is observed (Fig.  3); an IC& of -1.5 nM was calculated for either protein. This is considerably more potent than the pentapeptide GRGDV which has an I&, of -40 nM in this assay.
The dose-dependent inhibition of human platelet aggregation by decorsin was measured by inhibition of ADP-induced platelet aggregation in PRP. Complete inhibition was observed with 1 pM decorsin; an IC&, of -500 nM was calculated based on the titration curve. In comparison, the pentapeptide GRGDV has an IC& of -75 pM in the platelet aggregation assay.
The apparent loo-fold increase in potency of decorsin in the fibrinogen/GPIIb-IIIa ELISA relative to the platelet aggregation assay is similar to activity differences noted with the snake venom GPIIb-IIIa antagonists (17). This difference in apparent potency may reflect the combined effects of fibrinogen immobilization and low GPIIb-IIIa concentration in the ELISA, as well as a relatively high concentration of fibrinogen in the platelet aggregation assay; in addition, there may be differences in the affinity of purified GPIIb-IIIa relative to that of the intact platelet. These differences are more pronounced with the pentapeptide relative to either decorsin or the snake venom GPIIb-IIIa antagonists (17) and will be discussed elsewhere. ' Comparison with Other Proteins-Decorsin is similar to the snake venom family of GPIIb-IIIa protein antagonists in that they all contain the RGD sequence, have -16% of their amino acids present as cysteines and have very similar I& values in both the solid-phase fibrinogen/GPIIb-IIIa ELISA and platelet aggregation assays (14-17). A comparison of the primary sequences of decorsin with some of the snake venom proteins is presented in Fig. 4. Based on primary sequence analysis, decorsin does not appear to belong to the snake venom family of inhibitors; the only significant region of homology is in the region of the RGD sequence from residues 27-38 of the decorsin sequence. In this area 8 out of 13 of the residues can be aligned, with the addition of a gap inserted in the decorsin sequence to maximize homology. Although we have no direct evidence, we propose that it is the presentation of this epitope that is important for the high affinity of both Decorsin: a Potent GP IIb-IIIa Antagonist from Leech decorsin and the snake venom proteins for GPIIb-IIIa. It is known that simple RGD containing peptides are able to inhibit platelet aggregation, presumably via their interaction with 8); however these peptides are considerably less potent than decorsin or the venom proteins (17). The higher affinity observed for all of these proteins relative to RGD peptides is likely due to a specific conformation of the RGD epitope and/or to other binding interactions of the protein with the receptor. The lack of sequence homology outside the RGD containing region argues against a second remote linear epitope playing a major role in binding affinity.
Decorsin is similar in both size and cysteine content to a number of serine protease inhibitor proteins that have been isolated from leeches. These include the hirudins (21), which are potent thrombin inhibitors, and the bdellins (37), which are inhibitors of trypsin and plasmin. However, there is no apparent primary sequence homology between decorsin and these proteins. The high proline and cysteine content, assuming that all of the cysteines are present in disulfide bonds, consistent with the FAB-MS data, suggests that decorsin has a very rigid structure. A search of the Dayhoff protein sequence data base failed to identify any proteins with substantial homology to decorsin.
The function of decorsin in M. decora is not known, although it is likely that decorsin serves to keep the host blood flowing, or possibly to keep ingested blood from coagulating. This second putative function is important because leeches store ingested blood for long periods of time, digesting it slowly as needed (38). Although we cannot specifically define whether decorsin is present in saliva or secreted from cells in the digestive tract, its presence implies that it may have a primary role in hematophagous leech biology. If this is true then it is likely that there are proteins homologous to decorsin in other hematophagous leech species. In fact, a recent. report has indicated that prolonged bleeding occurs after a leech bite from H. medicinalis in the apparent absence of hirudin (39); future work will determine whether a GPIIb-IIIa antagonist homologous to decorsin might be responsible for this phenomenon.