- Email: [email protected]
The Disintegrin-like Domain of the Snake Venom Metalloprotease Alternagin Inhibits α2β1 Integrin-Mediated Cell Adhesion
Archives of Biochemistry and Biophysics Vol. 384, No. 2, December 15, pp. 341–350, 2000 doi:10.1006/abbi.2000.2120, available online at http://www.idealibrary.com on
The Disintegrin-like Domain of the Snake Venom Metalloprotease Alternagin Inhibits ␣21 IntegrinMediated Cell Adhesion D. H. F. Souza,* M. R. C. Iemma,* L. L. Ferreira,* J. P. Faria,* M. L. V. Oliva,† R. B. Zingali,‡ S. Niewiarowski,§ and H. S. Selistre-de-Araujo* ,1 *Department of Cieˆncias Fisiolo´gicas Universidade Federal de Sa˜o Carlos, Sa˜o Carlos SP; †Department of Bioquı´mica, Universidade Federal de Sa˜o Paulo, Sa˜o Paulo SP; ‡Department of Bioquı´mica Me´dica, Universidade Federal do Rio de Janeiro, RJ, Brazil; and §Department of Physiology, Temple University, Philadelphia, Pennsylvania
Received June 2, 2000, and in revised form September 8, 2000
The ␣ 2 1 integrin is a major collagen receptor that plays an essential role in the adhesion of normal and tumor cells to the extracellular matrix. Here we describe the isolation of a novel metalloproteinase/disintegrin, which is a potent inhibitor of the collagen binding to ␣ 2 1 integrin. This 55-kDa protein (alternagin) and its disintegrin domain (alternagin-C) were isolated from Bothrops alternatus snake venom. Amino acid sequencing of alternagin-C revealed the disintegrin structure. Alternagin and alternagin-C inhibit collagen I-mediated adhesion of K562-␣ 2 1-transfected cells. The IC50 was 134 and 100 nM for alternagin and alternagin-C, respectively. Neither protein interfered with the adhesion of cells expressing ␣ IIb 3, ␣ 1 1, ␣ 5 1, ␣ 4 1 ␣ V 3, and ␣ 9 1 integrins to other ligands such as fibrinogen, fibronectin, and collagen IV. Alternagin and alternagin-C also mediated the adhesion of the K562-␣ 2 1-transfected cells. Our results show that the disintegrin-like domain of alternagin is responsible for its ability to inhibit collagen binding to ␣ 2 1 integrin. © 2000 Academic Press Key Words: metalloprotease; disintegrin; snake venom; cell adhesion; collagen; ␣ 2 1 in tegrin.
The term “disintegrin” was first used in 1989 to describe a group of low molecular weight (5 to 9 kDa), cysteine-rich peptides, derived from snake venoms, which interact with integrin receptors on the surface of cell (1–3). On the other hand, integrins are transmembrane proteins which connect the extracellular matrix 1 To whom correspondence and reprint requests should be addressed. Fax: 55162608327. E-mail: [email protected].
0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
(ECM) 2 components to the cell cytoskeleton (4). Cell adhesion to the ECM is partially mediated by binding of integrin to an integrin-recognition RGD motif found on some ECM components such as fibronectin, vitronectin and fibrinogen (5). Most disintegrins are very potent inhibitors of platelet aggregation by acting as antagonists of the fibrinogen binding to platelet ␣ IIb 3 (IIb/IIIa) integrin receptor due to a cell-adhesive RGD motif in their amino acid sequence (6, 7). Some disintegrins have been demonstrated to inhibit platelet plug formation in in vivo experiments (8, 9), which could be useful in the antithrombotic therapy. RGD-disintegrins are also potent inhibitors of integrin-dependent cell adhesion, and this property has been used to inhibit experimental metastasis and platelet aggregation induced by metastatic cells (10, 11). Disintegrins are derived from proteolytic processing of hemorrhagic toxins found in viper venoms. Hemorrhagic toxins isolated from snake venoms are metalloproteinases (svMP) that digest the components of ECM leading to local and systemic hemorrhage in viperid and crotalid envenoming (12). SvMP are synthesized in the venom gland as large multidomain proteins, including a proenzyme domain and a highly conserved 2 Abbreviations used: svMP, snake venom metalloproteinases; ECM, extracellular matrix; RGD, arginine-glycine-aspartic acid; HPLC, high-performance liquid chromatography; VCAM-1, vascular cell adhesion molecule 1; BSA, bovine serum albumin; CMFDA, 5-chloromethylfluorescein diacetate; PBS, phosphate-buffered saline; MDCs, metalloprotease/disintegrin/cysteine-rich proteins; DMEM, Dulbecco’s Modified Eagle Medium; FBS, fetal bovine serum; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; CHO, Chinese hamster ovary cells; DMF, N,N-dimethylformamide; PMSF, phenylmethylsulfonyl fluoride; MHD, minimum hemorrhagic dose.
341
342
SOUZA ET AL.
zinc-protease domain (13, 14). svMP were divided into four classes (PI to PIV) according to their molecular mass and domain organization (15). Class PI includes the small svMP, with molecular mass about 24,000 Da, and low or no hemorrhagic activity. Class PII includes the medium size proteins and class PIII is represented by the large hemorrhagins (about 55,000 Da), which are believed to be the most potent hemorrhagic toxins. Members of the class IV are larger proteins (95,000 Da) but with low hemorrhagic activity (15). RGD-disintegrins are released from proteolytic processing of the PII proteins (16). Members of the PIII class have a disintegrin-like domain and an additional cysteine-rich domain which is not found in the RGD-disintegrins. In the disintegrin-like domain, the RGD motif is replaced by a D/ECD motif. These PIII proteins inhibit collagen-induced but not fibrinogen-induced platelet aggregation (17, 18). A few members of the PIII class have been described including jararhagin from Bothrops jararaca venom, atrolysin A, and catrocollastatin from Crotalus atrox venom (13, 17, 19). The term “disintegrin-like” protein was introduced to differentiate this group from the RGD-disintegrins. The whole molecule, the active protease having the disintegrin/cysteine-rich domain, and the processed domains can be isolated from snake venoms (19 –22). The processed disintegrin-like domain usually has an M r of 23,000 –28,000, and inhibits collagen-induced platelet aggregation by itself (20, 21, 23). On the other hand, the protease domain has a role in shedding of integrin receptors. Jararhagin, a metalloprotease/disintegrin isolated from B. jararaca venom, was shown to cleave the platelet ␣ 2 1-integrin that contributes to the inhibition of collagen-induced platelet aggregation (24). Therefore, the disintegrin domain was suggested to mediate the binding of jararhagin to platelet ␣ 2 1integrin, followed by proteolysis of integrin  1-subunit by the protease domain, with consequent inhibition of platelet aggregation (24). Recently, it has been demonstrated that a cyclic peptide derived from the metalloprotease domain of jararhagin blocks the collagen binding to the I-domain of the ␣ 2 subunit of the ␣ 2 1 integrin (25). These results provide evidence that the metalloprotease initially binds to the ␣ 2 I-domain at a location distinct from the active site of the enzyme, thus blocking collagen binding to the ␣ 2 1 integrin. The protease would then degrade the  1 subunit of the integrin. However, the role of the disintegrin domain in this process is not known. The inhibition of collagen-induced platelet aggregation by the disintegrin-like domain was demonstrated by different groups (20, 21, 23), but the site of the interaction with the ␣ 2 1 integrin is unknown. Many studies have been performed with the RGDdisintegrins regarding their structure and biological
effects such as those on cell adhesion to extracellular matrix, platelet aggregation, tumor cell invasion, and experimental metastasis. Neither the structure of disintegrin-like proteins nor their biological effects on other integrins or on cell adhesion have been studied. Therefore, the isolation of such disintegrins from snake venoms, which are very rich sources of these proteins, will provide new tools for the study of cell adhesion to the extracellular matrix and also for the development of good models for the design of new anti-metastatic drugs. Here we describe the isolation of the disintegrin-like/ cysteine-rich domain of a novel metalloprotease from Bothrops alternatus snake venom that selectively inhibits ␣ 2 1-integrin-mediated cell adhesion. Although there are a few papers on the inhibitory effect of disintegrin-like proteins on collagen-induced platelet aggregation, the effect of these proteins on the inhibition of cell adhesion has not been described so far. MATERIALS AND METHODS The venom of B. alternatus was kindly provided by Dr. Augusto Abe (Department of Zoology, UNESP-Rio Claro, Rio Claro, SP, Brazil). Superdex-200 and Mono Q HR were from Pharmacia. Antimouse IgG, alkaline phosphatase conjugate, bovine serum albumin, and fibrinogen were from Sigma. Fibronectin, collagen IV, fetal bovine serum, and all culture reagents were purchased from GibcoBRL. Collagen I was from Chrono-Log. Casein was from Calbiochem. The intramolecularly quenched fluorogenic substrates were synthesized by Dr. Maria Juliano (Department of Biophysics, UNIFESP, Sa˜o Paulo, Brazil). Endopeptidase Asp-N and endopeptidase Lys-C were of sequencing grade, from Boehringer-Mannheim. All other chemicals were of the highest grade available.
Cell Lines Chinese hamster ovary (CHO) cells transfected with human ␣ IIb 3 (A5) and ␣ v 3 (VNRC3) integrins (26), were kindly provided by Dr. M. Ginsberg (Scripps Research Institute, La Jolla, CA). SW480 cells (derived from a human colon carcinoma) transfected with ␣ 9 1 integrin were a gift from Dr. D. Sheppard (Lung Biology Center, University of San Francisco, CA). Jurkart cells from human acute T cell leukemia and K562 cells from human erythroleukemia were purchased from ATCC (Manassas, VA). K562 cells transfected with ␣ 2 1 and ␣ 1 1 integrin were a gift from Dr. M. E. Hemler (Dana Farber, Boston, MA). Cells were stably transfected and the expression of integrins was confirmed by flow cytometry using monoclonal antibodies against the ␣ 2 integrin subunit (clone AK-2, Pharmigen, San Diego, CA). Nontransfected cells do not express ␣ 2 integrin (27).
Protein Purification Size exclusion chromatography. B. alternatus crude venom (50 mg) was fractionated by size exclusion chromatography on a Superdex-200 column (1.6 cm ⫻ 60 cm), equilibrated with 10 mM Tris– HCl, pH 8.6, plus 100 mM NaCl. Elution was carried out using the same buffer at a flow rate of 0.5 ml/min, and the fractions were tested for enzymatic and hemorrhagic activities. Selected fractions were pooled and submitted to a second chromatography on the same column and under the same conditions. Phenyl Sepharose CL-4B hydrophobic interaction chromatography. The fractions of interest were further separated on a Phenyl Sepharose column (2 ⫻ 18 cm), previously equilibrated with 10 mM Tris–
DISINTEGRIN INHIBITION OF CELL ADHESION HCl, pH 8.6, and 1 M ammonium sulfate. Elution was carried out with a reverse ammonium sulfate gradient (1 to 0 M), at a flow rate of 2 ml/min. Mono Q HR anion exchange chromatography. The active fractions were further separated on a Mono Q Hr column, previously ¨ KTA Explorer 10 equilibrated with 10 mM Tris–HCl, pH 8.6, in an A (Pharmacia) chromatography equipment, and eluted with a NaCl gradient (0 M to 1 M), at a flow rate of 1 ml/min. All purification steps were performed at 4°C.
Enzymatic Assays All chromatographic steps were followed by enzymatic assays of the eluted fractions including proteolytic activity using casein and N-p-tosyl-Gly-Pro-Lys-p-nitroanilide as substrates, thrombin-like activity on purified bovine fibrinogen and clotting activity on bovine plasma (28). Fluorometric enzyme assay. Hydrolysis of fluorogenic peptides was measured following the procedure previously described (29). The assay was carried out for 10 min at 30°C in 50 mM Tris–HCl, pH 8.0 (1.5 ml), or 50 mM Tris–HCl, pH 8.0, 15 mM EDTA (1.5 ml), and the hydrolysis was monitored by measuring the fluorescence at emission of 420 nm and excitation of 320 nm, in a Hitachi F-2000 spectrofluorimeter.
Protein Characterization Protein purification was followed by SDS–PAGE (30) and Western blotting. First antibody was polyclonal and produced in mice using a recombinant metalloprotease/disintegrin from Agkistrodon contortrix laticinctus that was expressed in Escherichia coli (14). The molecular mass of the purified protein was estimated by SDS–PAGE. Protein concentration was determined using Coomassie brilliant blue G250 according to the method of Bradford (31).
N-Terminal Sequencing and Sequence Analysis Amino acid sequence analyses were performed by protein microsequencing system. Edman degradation of the reduced and alkylated (32) protein was performed in a gas-phase sequence equipment (Applied Biosystem mod 473 and a Porton/Beckman 2090 machine), using the conditions recommended by manufacturer. Cleavage of proteins were performed with endopeptidases Glu-C, Asp-N, and Lys-C. S-Pyridinethylated proteins (approximately 2 nmol) were incubated with endopeptidase Asp-N or endopeptidase Lys-C in 50 mM ammonium acetate buffer, pH 6.5 and pH 4.0, respectively, for 14 h at 37°C, with an enzyme:protein ratio of 1:20. Adjustments of the solution to pH 2.0 with formic acid ended the reaction. The amino acid sequences were compared to sequences in the GenBank, SwissProt or PIR databases using the BLAST program (33).
Minimum Hemorrhagic Dose (MHD) Hemorrhagic activity was determined in male mice as previously reported (34). Briefly, mice were injected intradermally with different doses (one dose per mouse) of the purified fraction. Two hours after injection, animals were killed, their skins removed, and the diameters of the hemorrhagic spots measured. The minimum hemorrhagic dose was defined as the amount of protein that produces a halo of 1 cm, 2 h after injection.
Adhesion Assays All cells were cultured in DMEM or RPMI media containing 10% FBS, L-glutamine, penicillin, streptomycin, and geneticin for the transfected cells at 37°C in a water-jacketed CO 2 incubator. Adhe-
343
sion of cultured cells labeled with 5-chloromethylfluorescein diacetate (CMFDA) was performed as described previously (35). Briefly, inhibitors (2.5 g/well), and ligands fibrinogen (1 g/well), fibronectin (1 g/well), collagen type I and IV (0.5 g/well), VCAM-1 (0.3 g/well), were immobilized on 96-well microtiter plates (Falcon, Pittsburgh, PA) in phosphate-buffered saline overnight at 4°C. Wells were blocked with 1% bovine serum albumin in Hank’s balanced salt solution buffer. Cells (5 ⫻ 10 6/ml) were labeled by incubation with 12.5 M 5-chloromethylfluorescein diacetate in Hank’s-balanced salt solution buffer containing 1% bovine serum albumin at 37°C for 30 min. Unbound label was removed by washing with the same buffer. Labeled cells were incubated in the presence or absence of inhibitors before being transferred to the plate (1 ⫻ 10 5 cells/well) and incubated at 37°C for 30 min. After washing to remove unbound cells, the remaining cells were lysed by the addition of 0.5% Triton X-100. In parallel, a standard curve was prepared in the same plate using known concentrations of labeled cells. The plates were read using a Cytofluor 2350 fluorescence plate reader (Millipore, Bedford, MA) with a 485-nm excitation and 530-nm emission filters.
RESULTS
Purification of Alternagin B. alternatus crude venom was separated into about five major peaks after fractionation on a Superdex 200 gel filtration column (Fig. 1a). B. alternatus has a high amount of a hemorrhagic protein of M r 55,000, which is characteristic of metalloproteinases carrying a disintegrin domain (15). Fractions were also tested for fibrinogen-clotting activity and proteolytic activity using casein as substrate (not shown). Fractions having hemorrhagic activity and with the expected size (fraction II) were pooled and applied again into the same column, and eluted in the same conditions. In this step, fraction II was resolved into two major peaks, which were named II-I and II-II (Fig. 1b). Fraction II-I was further purified in a phenyl Sepharose column (Fig. 1c). The selected fractions from this chromatography did not present any fibrinogen-clotting activity, only proteolytic and hemorrhagic activity. SDS–PAGE analysis showed that the purified hemorrhagic protein corresponds to a molecular mass of 55,000 kDa (Fig. 2a) and a minimum hemorrhagic dose of 1 g in mice. This protein was named alternagin. A series of peptide substrates displaying an o-aminobenzoyl (Abz) as fluorophore and a N-(2,4-dinitrophenyl)ethylenediamine (EDDNP) as a quencher was used to analyze the interaction with the metalloprotease. The substrates were based on the sequence of the reactive site of serine protease inhibitors, cysteine proteinase inhibitor, and the human kininogen sequence flanking the tissue kallikrein scissile bonds (29). The substrate with kininogen related quenched sequence Abz-Met-Ile-SerLeu-Met-Lys-Arg-Pro-EDDnp, which includes a partial amino sequence of bradykinin was hydrolyzed by the enzyme, with the following kinetic parameters: K m 2.12 M; k cat 0.036 s⫺1, and k cat/Km 1.7 104 M⫺1. sec⫺1. The hydrolysis was blocked in the presence of 15 mM EDTA confirming the action of the metalloproteinase on the
344
SOUZA ET AL.
FIG. 1. Purification of alternagin and its processed disintegrin domain. Elution profile of Superdex 200 gel filtration chromatography of 50 mg of B. alternatus crude venom (a) and fraction II (b). Elution buffer 10 mM Tris–HCl, pH 8.6, and 100 mM NaCl. Flow rate 0.5 ml/min. The selected fraction is indicated. (c) Phenyl Sepharose CL-4B hydrophobic interaction chromatography of fraction II-I from the gel filtration of crude venom. Fractions were eluted with a reverse gradient (1 M to 0 M) of ammonium sulfate. Flow rate, 2 ml/min; fractions were collected at 2-min intervals. (d) Mono Q anion exchange HPLC of fraction II-II from the second gel filtration. Elution buffer 10 mM Tris–HCl, pH 8.6, plus a NaCl gradient (from 0 to 1 M). Flow rate 1 ml/min. The major peak corresponds to alternagin-C.
substrate. The protease cleaves the substrate at the LeuMet bond (not shown). Fraction II-II does not have any significant proteolytic activity but very low plasma clotting activity. It was further separated into several peaks in an anion exchange chromatography (Fig. 1d). The major peak eluted in this chromatography corresponds to a protein of M r 28,000 (Fig. 2a). Both proteins of M r 55,000 and 28,000 strongly react with an antibody against a recombinant metalloproteinase/disintegrin (Fig. 2b), suggesting that the smaller protein could be the processed disintegrin domain. This protein was named alternagin-C. Usually, about 0.4 mg of purified alternagin-C were isolated from 50 mg of crude venom. The yield of the whole molecule was lower, about 0.2 mg, due to the loss caused by the processing. Purified alternagin does not have any detectable proteolytic activity on casein but the specific activity on the fluorescent substrate was about 211 U/mg. When the B. alternatus crude venom was treated with 20 mM EDTA and 2 mM PMSF before gel filtra-
tion, fraction II-II was markedly decreased (not shown), thus suggesting that this protein was generated by proteolytic processing from a precursor form present in the venom. Partial Amino Acid Sequence of Alternagin-C The N-terminus of alternagin was blocked suggesting the presence of pyroglutamic acid as the N-terminal residue, in agreement with the data in literature for other metalloproteinases from snake venoms (15). In order to better characterize the protein and to confirm the presence or absence of the RGD sequence, only the processed disintegrin domain was further analyzed by amino acid sequencing. The partial sequence of alternagin-C was established by automated Edman degradation of the protein. Peptide fragments produced by Lys-C, Glu-C, and Asp-N endoproteinase cleavage of 4-vinyl pyridine reduced protein were sequenced after isolation by reverse phase chromatography. Enough overlaps were ob-
DISINTEGRIN INHIBITION OF CELL ADHESION
FIG. 2. SDS–PAGE and Western blotting analyses of purified fractions. (A) Coomassie blue R250-stained 15% SDS–polyacrylamide gel. Samples: 1, Molecular markers; 2, B. alternatus crude venom; 3, purified metalloproteinase/disintegrin alternagin (ALT); 4, alternagin-C (ALT-C). (B) The same samples were transferred to nitrocellulose membrane and probed with antirecombinant metalloproteinase/disintegrin serum (1:1,000, B).
tained to define the majority of the primary structure (Fig. 3a). The N-terminal sequence of alternagin-C showed that this sequence has a disintegrin structure, with high identity with the disintegrin domain reported for other PIII snake venom metalloproteinases (13, 14). A sequence comparison of the N-terminus of homologous proteins shows that alternagin-C belongs to the class of disintegrin-like proteins due to the presence of highly conserved amino acid residues in the disintegrin-like protein family (Fig. 3b). The striking difference between the two disintegrin families is the pattern of disulfide bonds, although several other residues in the N-terminal region can differentiate the disintegrins from the disintegrin-like proteins, as shown in Fig. 3b. The position of cysteine residues is highly conserved in the disintegrin domain but there are at least two conserved cysteine substitutions in the RGD-disintegrins and one is located in the proposed adhesive sequence D/ECD (23). Several conserved residues in the disintegrin-like protein family such as Ile 1, Pro 4, Cys 7, Val 14, and Cys 26 (numbering as in Fig. 3b) are also found in alternagin-C. In contrast, RGD-disintegrins have different but also conserved residues in these positions (Fig. 3b). The primary structure of alternagin-C confirmed the homology with the disintegrin-like proteins and the lack of the RGD motif. Although the complete sequence of alternagin-C is not known yet, it can be clearly seen from the sequencing results that it belongs to the family of the disintegrin-like proteins, with the ECD conserved motif.
345
and, among the cells tested, specific for cells expressing ␣ 2 1 integrin (Fig. 4). Neither protein interfered with the adhesion of cells expressing ␣ IIb 3, ␣ 1 1, ␣ 5 1, ␣ 4 1 ␣ V 3, and ␣ 9 1 integrins to other ligands such as fibrinogen, fibronectin collagen IV and VCAM-1 (Table I). Even the main collagen type IV receptor, the integrin ␣ 1 1, was not affected by alternagin and alternagin-C. We have not tested the effect of both proteins on the ␣ 3 1 integrin, which also binds to collagen. The amount of protein that inhibits 50% of the cell adhesion (IC50) was 134 and 100 nM for alternagin and alternagin-C, respectively. The values of IC50 for both proteins show that the disintegrin domain has an essential role in the integrin inhibition. The difference between IC50 values is not statistically significant, but indicates that proteolysis is likely less important because the enzymatically inactive protein alternagin-C has the same activity on cell adhesion as the whole molecule. To investigate if alternagin or alternagin-C could act as adhesion molecules by themselves, wells were coated with the two purified proteins and further incubated with ␣ 2 1-tranfected and nontransfected K562 cells. The control of nontransfected cells was made using fibronectin as the ligand since these cells do not bind to collagen. Nontransfected K562 cells did not adhere to alternagin or alternagin-C (Fig. 5A). In contrast, the two proteins induced significant adhesion of ␣ 2 1-transfected K562 cells thus demonstrating that the disintegrin domain also binds to the ␣ 2 1 integrin (Fig. 5B). These results support the idea that the disintegrin domain probably inhibits collagen binding to the ␣ 2 1 integrin by competition. We next compared the data in the literature regarding the inhibition of collagen binding to platelet ␣ 2 1 integrin induced by snake venom metalloproteinases and their disintegrin domains. In platelets the disintegrin domain seems to be less active than the whole molecule (Table II), although the data were collected from different laboratories, assayed in different conditions regarding type and concentration of collagen, and in some cases, with native and recombinant proteins, which makes the comparison more difficult. The results presented in this paper were collected in the same conditions, and therefore the comparison is more reliable allowing the conclusion that the disintegrin domain is the domain responsible for the collagen inhibition of ␣ 2 1 integrin in K562 transfected cells. Jararhagin (whole molecule) was also tested in the same cell adhesion assay, giving similar results (36). However, the authors have not tested the processed disintegrin domain in this system.
Adhesion Studies Alternagin and alternagin-C are potent inhibitors of the collagen binding to ␣ 2 1 integrin. Their ability to inhibit collagen-induced adhesion was dose-dependent
DISCUSSION
This paper describes the isolation and structural characterization of a novel metalloproteinase/disinte-
346
SOUZA ET AL.
FIG. 3. (A) Amino acid sequence of alternagin-C. The peptides obtained after fragmentation by enzyme digestion with Lys-C, Glu-C, and Asp-N endoproteases are indicated. (B) Sequence comparison of the processing region of the disintegrin domain in RGD-disintegrins and disintegrin-like proteins. References: alternagin, this work; jararhagin (19); ACLD (14); catrocollastatin (17); Echis I (49); trigramin (50); trimucrin (51); rhodostomin (52); halystatin (53), and atrolysin e (13). The first five sequences are from disintegrin-like proteins and the others are RGD-disintegrins. The amino acid residues conserved in each family are indicated by arrowheads.
347
DISINTEGRIN INHIBITION OF CELL ADHESION
FIG. 4. Effect of Alternagin and Alternagin-C on adhesion of K562␣ 2 1 cells to collagen type I. Collagen I (0.5 g/well) was immobilized overnight at 4°C on a 96-well plate in phosphate-buffered saline (PBS). After blocking with 1% BSA, the 5-chloromethyfluorescein diacetate-labeled cells incubated with different concentration of inhibitors were added to each well. The plate was then incubated at 37°C for 30 min. The adhesion experiment was performed as described under Materials and Methods. Error bars indicate S.E. from three independent duplicated experiments.
grin and the processed disintegrin domain from B. alternatus venom. The two proteins were named alternagin and alternagin-C to be consistent with the literature (20, 21). Although the inhibitory effect of this class of protein on collagen-induced platelet aggregation is well established, it is not known which domain is responsible for this effect. Data in the literature so far are mostly from independent groups using native and recombinant proteins, which could make the comparison more difficult. Here we describe the inhibitory effect of a metalloproteinase/disintegrin on a different cell system, and we compare the results obtained with the whole molecule and the processed disintegrin/cysteine-rich domain in the same system. Although the complete amino acid sequence of alternagin is not yet known, we suggest that alternagin-C is derived by limited proteolysis from alternagin or one of its isoforms, as usually found in snake venoms. The proteolytic activity of alternagin was tested in two different assays. Hemorrhagic activity is a conse-
quence of the proteolytic digestion of basal membrane components such as collagen, laminin and fibronectin (12). The MHD for alternagin is 1 g in mice, which is consistent for this class of snake venom metalloproteinase (15). In a more refined assay, alternagin was shown to cleave a synthetic peptide with a bradykinin related sequence, and the kinetic parameters were defined. Preliminary results indicated that alternagin also cleaves human high molecular weight kininogen but further investigation is necessary to identify the products of kininogen hydrolysis. Jararhagin, a homologous protein was shown to process the tumor necrosis factor (TNF)-␣ from its precursor on cell membrane in vitro (37). These results show that svMP may be very useful tools for the study of important physiological processes. Integrins ␣ 1 1 and ␣ 2 1 are the major cellular receptors for collagen, and the collagen binding domain called I domain, is located within the ␣ subunit (38). Alternagin and the processed disintegrin/cysteine-rich domain, alternagin-C, are potent inhibitors of the ␣ 2 1 integrin. Their ability to inhibit collagen-induced adhesion is dose-dependent and specific for cells expressing ␣ 2 1 integrin. Even the collagen type IV receptor, the integrin ␣ 1 1, was not affected by alternagin. The inhibitory effect of a disintegrin domain on the ␣ 2 1 integrin was only described for platelets (17, 20, 21). Here we show that other cells can be affected by this class of proteins and several types of integrins are insensitive to alternagin-C. Recently, it has been also demonstrated that jararhagin also inhibits the collagen binding to ␣ 2 1-transfected K562 cells (36). However, the inactivation of jararhagin with o-phenanthroline only slightly increased the IC50 for the collagen binding, which gives more evidence that the proteolysis does not have a significant role in this effect. It has been proposed that the metalloproteinase domain has a significant contribution to the inhibitory effect due to the presence of a ␣ 2-I domain-binding motif (RKK) in this domain (25). It was shown that this
TABLE I
Effect of Alternagin (ALT) and Alternagin-C (ALT-C) on the Adhesion of Cells Expressing Different Integrins IC 50 (nM) Cell suspension
Integrins
A5 VNRC3 K562 K562 (␣ 2) K562 (␣ 1) SW480 (␣ 9) Jurkart
␣ IIb 3 ␣v 3 ␣ 5 1 ␣ 2 1 ␣ 1 1 ␣ 9 1 ␣ 4 1
Ligand Fibrinogen Fibronectin Fibronectin Collagen I Collagen IV VCAM-1 VCAM-1
Assay CA CA CA CA CA CA CA
ALT
ALT-C
⬎5000 ⬎5000 ⬎5000 134 ⬎5000 ⬎5000 ⬎5000
⬎5000 ⬎5000 ⬎5000 100 ⬎5000 ⬎5000 ⬎5000
348
SOUZA ET AL.
FIG. 5. Adhesion of (A) non transfected and (B) transfected K562 cells with ␣ 2 1 integrin to immobilized inhibitors. Alternagin and Alternagin-C (2.5 g/well) were immobilized overnight at 4°C on a 96-well plate in phosphate-buffered saline (PBS). After blocking with 1% BSA, the 5-chloromethyfluorescein diacetate-labeled cells were added to each well and the plate was incubated at 37°C for 30 min. Error bars indicate S.E. from three independent experiments. Controls were made with immobilized fibronectin (1.0 g/well, in A) and collagen I (0.5 g/well, in B).
motif recognizes the I domain in the ␣2-subunit therefore inhibiting the binding of collagen. Crovidisin, isolated from Crotalus viridis venom, is a protein similar to catrocollastatin that also binds directly to collagen fibers thus blocking the interaction of platelets and collagen (39), although in this case it is not known which domain has this collagen-binding activity. Another important contribution is the fact that the active protease could cleave the  1-subunit therefore impairing the binding of collagen to platelets, as it has been demonstrated for jararhagin (24). In fact, inactivation of jararhagin, a homologous protein with 1,10-phenanthroline increased the IC50 from 40 to 140 (24). The cleavage of the  1-subunit results in inhibition of the collagen-stimulated phosphorylation events associated with the platelet response (40). Therefore, the metal-
loprotease domain has a significant contribution to the inhibition of collagen-induced platelet aggregation. On the other hand, the disintegrin domain would act as an antagonist of the interaction of ␣ 2 1 with collagen. Our results show that the processed domain is a potent inhibitor of this integrin and can completely inhibit cell adhesion in the absence of the catalytic domain. Disintegrins have received much attention in the recent years due to the discovery of homologous membrane-bound proteins in mammalian tissues, which seem to have important roles in several physiological processes, including fertilization, cell differentiation, and shedding of receptors (41). These proteins are named ADAMs (for a disintegrin and metalloprotease) or MDCs (for metalloprotease/disintegrin/cysteine-rich proteins). More than 23 ADAMs were described in
TABLE II
Comparison of the Inhibitory Effects of svMDC and Isolated Disintegrin/Cysteine-Rich Domains on the Collagen I Receptor Protein
Source
Cell system
Ligand
IC50 (nM)
Reference
Alternagin Alternagin-C a Jararhagin Jararhagin-I b Jararhagin Jararhagin Jararhagin-I b Jararhagin-C Catrocollastatin Catrocollastatin-C Atrolysin a Atrolysin a D/C Crovidisin
Bothrops alternatus Bothrops alternatus Bothrops jararaca Bothrops jararaca Bothrops jararaca Bothrops jararaca Bothrops jararaca Bothrops jararaca Crotalus atrox Crotalus atrox Crotalus atrox Crotalus atrox Crotalus viridis
K562 ␣21 K562 ␣21 K562 ␣21 K562 ␣21 Platelet Platelet Platelet Platelet Platelet Platelet Platelet Platelet Platelet
Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen
134 100 50 140 50 40 140 100 50 66 110 470 170
This work This work 36 36 18 24 24 21 17 20 23 23 39
a b
The symbol (-C) means the disintegrin, cysteine-rich domain. 1,10-Phenanthroline-inactivated jararhagin.
DISINTEGRIN INHIBITION OF CELL ADHESION
many different tissues from several sources including mammals, Drosophila melanogaster and Caenorrhabits elegans (42, 43). Disintegrins from snake venoms have become important tools for understanding the roles of these proteins in cell adhesion since they can be isolated from venoms with relative ease and at significant yield. The motifs in the disintegrin/cysteine-rich domain that could be involved in the integrin recognition are not known, but it has been demonstrated that synthetic peptides having the sequence RSECD in which the cysteine is constrained by a disulfide bond, inhibit collagen-induced platelet aggregation (23). The disintegrin/cysteine-rich domain could also be divided into two domains, depending on the pattern of disulfide bonds. There are few papers on the effect of the proposed cysteine-rich domain on cell adhesion. Since there is no report on the isolation of such domain, the work has been done with recombinant proteins. It has been suggested that the cysteine-rich region of human ADAM 12 mediates tumor cell adhesion (44, 45). The cysteine-rich domain of fertilin ␣ mediates the spermegg binding during fertilization (46). More recently, it was demonstrated that the recombinant cysteine-rich domain of atrolysin A from Crotalus atrox venom inhibits the binding of MG-63 cells to collagen. This osteosarcoma cell line expresses predominantly the integrin ␣ 2 1 on its cell surface (47). These data suggest that the cysteine-rich domain must have an important role in cell adhesion. Our results are in agreement with this conclusion since the cysteine-rich domain is a part of the disintegrin/cysteine-rich domain. Since the disulfide bond pattern nor the three-dimensional structure of a disintegrin-like protein are not known yet, the contribution of the overall conformation to the integrin binding is a more difficult question to solve. We have also demonstrated that the disintegrin domain may mediate cell adhesion in an in vitro assay. Similar results were found for fertilin which was demonstrated to act as an adhesion molecule in the spermegg binding (46). Therefore, the disintegrin domain must have adhesive sequences that compete with collagen in the integrin binding. In conclusion, the present study shows that the disintegrin-like domain of a svMP binds with specificity to ␣ 2 1-transfected K562 cells, therefore inhibiting collagen-mediated cell adhesion. It is also shown that this protein may act as an adhesion molecule by itself. As far as we know, this is the first study about the interaction of the disintegrin-like proteins and other integrins in a cell adhesion assay. These results may aid in the understanding of the structure–function relationship of disintegrins as well as the molecular basis of cell adhesion. Recent studies have shown that ovarian carcinoma cells adhere preferentially to type I collagen (48). Therefore, disintegrins that specifically inhibit
349
the collagen binding to its receptor are important tools for the study of this type of cancer metastasis and for the development of specific peptide-derived inhibitors of tumor cell invasion. ACKNOWLEDGMENTS We thank Dr. Charlotte L. Ownby for critical reading of the manuscript and the technical support of Mariola Marcinkiewicz in the cell culture assays. This work was supported by grants from FAPESP, Brazil (95/9300-7, 96/08863-0, and 97/05398-8), CNPq, Brazil (521542/96-0), American Heart Association (U.S.A.), and International Foundation for Science, Sweden (F/2631-1).
REFERENCES 1. Dennis, M. S., Henzel, W. J., and Pitti, R. M. (1989) Proc. Natl. Acad. Sci. USA 87, 2471–2475. 2. Gould, R. J., Polokoff, M. A., Friedman, P. A., Huang, T-F., Holt, J. C., Cook, J. J., and Niewiarowski, S. (1990) Proc. Soc. Exp. Biol. Med. 195, 168 –171. 3. Niewiarowski S., McLane M. A., Kloczewiak M., and Stewart, G. J. (1994) Semin. Hematol. 31, 289 –290. 4. Hynes, R. O. (1992) Cell 48, 549 –554. 5. Yamada, K. M. (1991) J. Biol. Chem. 266, 12809 –12812. 6. Huang, T-F., Holt, J. C., Lukasiewcz, H., and Niewiarowski, S. (1987) J. Biol. Chem. 262, 16157–16163. 7. Maruyama, K., Kawasaki, T., Sakai, Y., Taniuchi, Y., Shimizu, M., Kawashima, H., and Takenaka, T. (1997) Peptides 18, 73–78. 8. Sheu, J. R., and Huang, T-F. (1994) J. Pharm. Pharmacol. 46, 58 – 62. 9. Sheu, J.-R., Yen, M.-H., Peng, H.-C., Chang, M.-C., and Huang, T.-F. (1995) Eur. J. Pharmacol. 294, 231–238. 10. Beviglia L., Stewart G. J., and Niewiarowski, S. (1995) Oncol. Res. 7,7–20. 11. Danen, E. H. J., Marcinkiewicz, C., Cornelissen, I. M. H. A., van Kraats, A. A., Pachter, J. A., Ruiter, D. J., Niewiarowski, S., and van Muijen, G. N. P. (1998) Exp. Cell Res. 238, 188 –196. 12. Baramova, E. N., Shannon, J. D., Bjarnason, J. B., and Fox, J. W. (1989) Arch. Biochem. Biophys. 275, 63–71. 13. Hite, L., Jia, L., Bjarnason, J. B., and Fox, J. W. (1994) Arch. Biochem. Biophys. 308, 182–191. 14. Selistre de Araujo, H. S., Souza, D. H. F., and Ownby, C. L. (1997) Biochim. Biophys. Acta 1342 (2), 109 –115. 15. Bjarnason, J. B., and Fox, J. W. (1994) Pharmacol. Ther. 62, 325–372. 16. Yamada, D., Shin, Y., and Morita, T. (1999) FEBS Lett. 451, 299 –302. 17. Zhou, Q., Smith, J. B., and Grossman, M. H. (1995) Biochem. J. 307, 411– 417. 18. De Luca, M., Ward, C. M., Ohmori, K., Andrews, R. K., and Berndt, M. C. (1995) Biochem. Biophys. Res. Commun. 206, 570 –576. 19. Paine, M. J. I., Desmond, H. P., Theakston, R. D. G., and Crampton, J. M. (1992) J. Biol. Chem. 267, 22869 –22876. 20. Shimokawa, K-I., Shannon, J. D., Jia, L-G., and Fox, J. W. (1997) Arch. Biochem. Biophys. 343, 35– 43. 21. Usami, Y., Fujimura, Y., Miura, S., Shima, H., Yoshida, E., Yoshioka, A., Hirano, K., Suzuki, M., and Titani, K. (1994) Biochem. Biophys. Res. Commun. 201, 331–339. 22. Shimokawa, K-I., Jia, L-G., Shannon, J. D., and Fox, J. W. (1998) Arch. Biochem. Biophys. 354, 239 –246.
350
SOUZA ET AL.
23. Jia, L-G., Wang, X-M., Shannon, J. D., Bjarnasson, J. B., and Fox, J. W. (1997) J. Biol. Chem. 272, 13094 –13102. 24. Kamiguti, A. S., Hay, C. R. M., and Zuzel, M. (1996) Biochem. J. 320, 635– 641. 25. Ivaska, J., Ka¨pyla´, J., Pentika¨inen, O., Hoffre´n, A-M., Hermonen, J., Huttunen, P., Johnson, M. S., and Heino, J. (1999) J. Biol. Chem. 274, 3513–3521. 26. O’Toole, T. E., Loftus, J. C., Du, X., Glass, A., Ruggeri, Z. M., Shattil, S. J., Plow, E. F., and Ginsberg, M. H (1990) Cell Regul. 1, 883– 893. 27. Marcinkiewicz, C., Lobb, R. R., Marcinkiewicz, M. M., Daniel, J. L., Smith, J. B., Dangelmaier, C., Weinreb, P. H., Beacham, D. A., and Niewiarowski, S. Biochemistry, in press. 28. Selistre, H. S., and Giglio, J. R. (1987) Toxicon 25, 1135. 29. Oliva, M. L. V., Andrade, S. A., Batista, I. F. C., Sampaio, M. U., Juliano, M., Fritz, H., Auerswald, E. A., and Sampaio, C. A. M. (1999) Immunopharmacology 45, 145–149. 30. Laemmli, U. K. (1970) Nature 227, 680 – 685. 31. Bradford, M. M. (1976) Anal. Biochem. 72, 248 –254. 32. Friedman, M., Krull, L. H., and Cavins, J. F. (1970) J. Biol. Chem. 245, 3868. 33. Altschul, S. F., Madden, T. L., Scha¨ffer, A. A., Zhang, J., Zhang, Z., Miller, W., and Lipman, D. J. (1997) Nucleic Acids Res. 25, 3389 –3402. 34. Johnson, E. K., and Ownby, C. L. (1993) Int. J. Biochem. 25(2), 267–278. 35. Marcinkiewicz, C., Calvete, J. J., Vijay-Kumar, S., Marcinkiewicz, M. M., Raida, M., Shick, P., Lobb, R. R., and Niewiarowski, S. (1999) Biochemistry 38, 13302–13309. 36. Moura-da-Silva, A. M., Marcinkiewicz, C., Marcinkiewicz, M. M., and Niewiarowski, S., Thromb. Res., in press. 37. Moura-da-Silva, A. M., Laing, G. D., Paine, M. J. I., Dennison, J. M. T. J., Politi, V., Crampton, J. M., and Theakston, R. D. G. (1996) Eur. J. Immunol. 26, 2000 –2005.
38. Takada, Y., and Helmer, M. E. (1989) J. Cell Biol. 109, 397– 407. 39. Liu, C-Z., and Huang, T-F. (1997) Arch. Biochem. Biophys. 337, 291–299. 40. Kamiguti, A. S., Markland, F. S., Zhou, Q., Laing, G. D., Theakston, R. D. G., and Zuzel, M. (1997) J. Biol. Chem. 272, 32599 – 32605. 41. Killar, L., White, J., Black, R., and Peschon, J. (1999) Ann. N.Y. Acad. Sci. 878, 442– 452. 42. Rooke, J., Pan, D., Xu, T., and Rubin, G. M. (1996) Science 273, 1227–1231. 43. Wen, C., Metzstein, M. M., and Greenwald, I. (1997) Development 124, 4759 – 4767. 44. Iba, K., Albrechtsen, R., Gilpin, B. J., Loechel, F., and Wewer, U. M. (1999) Am. J. Pathol. 154, 1489 –1501. 45. Zolkiewska, A. (1999) Exp. Cell Res. 252, 423– 431. 46. Evans, J. P., Schultz, R. M., and Kopf, G. S. (1997) Dev. Biol. 187, 94 –106. 47. Jia, L-G., Wang, X-M., Shannon, J. D., Bjarnasson, J. B., and Fox, J. W. (2000) Arch. Biochem. Biophys. 373, 281–286. 48. Fishman, D. A., Kearns, A., Chilukuri, K., Bafetti, L. M., O’Toole, E. A., Georgacopoulos, J., Ravosa, M. J., and Stack, M. S. (1998) Invasion Metastasis 18, 15–26. 49. Paine, M. J. I., Moura-da-Silva, A. M., Theakston, R. D. G., and Crampton, J. M. (1994) Eur. J. Biochem. 224, 483– 488. 50. Huang, T-F., Holt, J. C., Kirby, E. P., and Niewiarowski, S. (1989) Biochemistry 28, 661– 666. 51. Tsai, I-H., Wang, Y-M., and Lee, Y-H. (1994) Biochim. Biophys. Acta 1200, 337–340. 52. Au, L-C., Huang, Y-B., Huang, T. F., Teh, G-W., Lin, H-H., and Choo, K-B. (1991) Biochem. Biophys. Res. Commun. 181, 585– 593. 53. Fujisawa, Y., Kuroda, S., Notoya, K., Konishi, H., and Terashita, Z. (1994) J. Takeda Res. Lab. 53, 39 –56.
The Disintegrin-like Domain of the Snake Venom Metalloprotease Alternagin Inhibits ␣21 IntegrinMediated Cell Adhesion D. H. F. Souza,* M. R. C. Iemma,* L. L. Ferreira,* J. P. Faria,* M. L. V. Oliva,† R. B. Zingali,‡ S. Niewiarowski,§ and H. S. Selistre-de-Araujo* ,1 *Department of Cieˆncias Fisiolo´gicas Universidade Federal de Sa˜o Carlos, Sa˜o Carlos SP; †Department of Bioquı´mica, Universidade Federal de Sa˜o Paulo, Sa˜o Paulo SP; ‡Department of Bioquı´mica Me´dica, Universidade Federal do Rio de Janeiro, RJ, Brazil; and §Department of Physiology, Temple University, Philadelphia, Pennsylvania
Received June 2, 2000, and in revised form September 8, 2000
The ␣ 2 1 integrin is a major collagen receptor that plays an essential role in the adhesion of normal and tumor cells to the extracellular matrix. Here we describe the isolation of a novel metalloproteinase/disintegrin, which is a potent inhibitor of the collagen binding to ␣ 2 1 integrin. This 55-kDa protein (alternagin) and its disintegrin domain (alternagin-C) were isolated from Bothrops alternatus snake venom. Amino acid sequencing of alternagin-C revealed the disintegrin structure. Alternagin and alternagin-C inhibit collagen I-mediated adhesion of K562-␣ 2 1-transfected cells. The IC50 was 134 and 100 nM for alternagin and alternagin-C, respectively. Neither protein interfered with the adhesion of cells expressing ␣ IIb 3, ␣ 1 1, ␣ 5 1, ␣ 4 1 ␣ V 3, and ␣ 9 1 integrins to other ligands such as fibrinogen, fibronectin, and collagen IV. Alternagin and alternagin-C also mediated the adhesion of the K562-␣ 2 1-transfected cells. Our results show that the disintegrin-like domain of alternagin is responsible for its ability to inhibit collagen binding to ␣ 2 1 integrin. © 2000 Academic Press Key Words: metalloprotease; disintegrin; snake venom; cell adhesion; collagen; ␣ 2 1 in tegrin.
The term “disintegrin” was first used in 1989 to describe a group of low molecular weight (5 to 9 kDa), cysteine-rich peptides, derived from snake venoms, which interact with integrin receptors on the surface of cell (1–3). On the other hand, integrins are transmembrane proteins which connect the extracellular matrix 1 To whom correspondence and reprint requests should be addressed. Fax: 55162608327. E-mail: [email protected].
0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
(ECM) 2 components to the cell cytoskeleton (4). Cell adhesion to the ECM is partially mediated by binding of integrin to an integrin-recognition RGD motif found on some ECM components such as fibronectin, vitronectin and fibrinogen (5). Most disintegrins are very potent inhibitors of platelet aggregation by acting as antagonists of the fibrinogen binding to platelet ␣ IIb 3 (IIb/IIIa) integrin receptor due to a cell-adhesive RGD motif in their amino acid sequence (6, 7). Some disintegrins have been demonstrated to inhibit platelet plug formation in in vivo experiments (8, 9), which could be useful in the antithrombotic therapy. RGD-disintegrins are also potent inhibitors of integrin-dependent cell adhesion, and this property has been used to inhibit experimental metastasis and platelet aggregation induced by metastatic cells (10, 11). Disintegrins are derived from proteolytic processing of hemorrhagic toxins found in viper venoms. Hemorrhagic toxins isolated from snake venoms are metalloproteinases (svMP) that digest the components of ECM leading to local and systemic hemorrhage in viperid and crotalid envenoming (12). SvMP are synthesized in the venom gland as large multidomain proteins, including a proenzyme domain and a highly conserved 2 Abbreviations used: svMP, snake venom metalloproteinases; ECM, extracellular matrix; RGD, arginine-glycine-aspartic acid; HPLC, high-performance liquid chromatography; VCAM-1, vascular cell adhesion molecule 1; BSA, bovine serum albumin; CMFDA, 5-chloromethylfluorescein diacetate; PBS, phosphate-buffered saline; MDCs, metalloprotease/disintegrin/cysteine-rich proteins; DMEM, Dulbecco’s Modified Eagle Medium; FBS, fetal bovine serum; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; CHO, Chinese hamster ovary cells; DMF, N,N-dimethylformamide; PMSF, phenylmethylsulfonyl fluoride; MHD, minimum hemorrhagic dose.
341
342
SOUZA ET AL.
zinc-protease domain (13, 14). svMP were divided into four classes (PI to PIV) according to their molecular mass and domain organization (15). Class PI includes the small svMP, with molecular mass about 24,000 Da, and low or no hemorrhagic activity. Class PII includes the medium size proteins and class PIII is represented by the large hemorrhagins (about 55,000 Da), which are believed to be the most potent hemorrhagic toxins. Members of the class IV are larger proteins (95,000 Da) but with low hemorrhagic activity (15). RGD-disintegrins are released from proteolytic processing of the PII proteins (16). Members of the PIII class have a disintegrin-like domain and an additional cysteine-rich domain which is not found in the RGD-disintegrins. In the disintegrin-like domain, the RGD motif is replaced by a D/ECD motif. These PIII proteins inhibit collagen-induced but not fibrinogen-induced platelet aggregation (17, 18). A few members of the PIII class have been described including jararhagin from Bothrops jararaca venom, atrolysin A, and catrocollastatin from Crotalus atrox venom (13, 17, 19). The term “disintegrin-like” protein was introduced to differentiate this group from the RGD-disintegrins. The whole molecule, the active protease having the disintegrin/cysteine-rich domain, and the processed domains can be isolated from snake venoms (19 –22). The processed disintegrin-like domain usually has an M r of 23,000 –28,000, and inhibits collagen-induced platelet aggregation by itself (20, 21, 23). On the other hand, the protease domain has a role in shedding of integrin receptors. Jararhagin, a metalloprotease/disintegrin isolated from B. jararaca venom, was shown to cleave the platelet ␣ 2 1-integrin that contributes to the inhibition of collagen-induced platelet aggregation (24). Therefore, the disintegrin domain was suggested to mediate the binding of jararhagin to platelet ␣ 2 1integrin, followed by proteolysis of integrin  1-subunit by the protease domain, with consequent inhibition of platelet aggregation (24). Recently, it has been demonstrated that a cyclic peptide derived from the metalloprotease domain of jararhagin blocks the collagen binding to the I-domain of the ␣ 2 subunit of the ␣ 2 1 integrin (25). These results provide evidence that the metalloprotease initially binds to the ␣ 2 I-domain at a location distinct from the active site of the enzyme, thus blocking collagen binding to the ␣ 2 1 integrin. The protease would then degrade the  1 subunit of the integrin. However, the role of the disintegrin domain in this process is not known. The inhibition of collagen-induced platelet aggregation by the disintegrin-like domain was demonstrated by different groups (20, 21, 23), but the site of the interaction with the ␣ 2 1 integrin is unknown. Many studies have been performed with the RGDdisintegrins regarding their structure and biological
effects such as those on cell adhesion to extracellular matrix, platelet aggregation, tumor cell invasion, and experimental metastasis. Neither the structure of disintegrin-like proteins nor their biological effects on other integrins or on cell adhesion have been studied. Therefore, the isolation of such disintegrins from snake venoms, which are very rich sources of these proteins, will provide new tools for the study of cell adhesion to the extracellular matrix and also for the development of good models for the design of new anti-metastatic drugs. Here we describe the isolation of the disintegrin-like/ cysteine-rich domain of a novel metalloprotease from Bothrops alternatus snake venom that selectively inhibits ␣ 2 1-integrin-mediated cell adhesion. Although there are a few papers on the inhibitory effect of disintegrin-like proteins on collagen-induced platelet aggregation, the effect of these proteins on the inhibition of cell adhesion has not been described so far. MATERIALS AND METHODS The venom of B. alternatus was kindly provided by Dr. Augusto Abe (Department of Zoology, UNESP-Rio Claro, Rio Claro, SP, Brazil). Superdex-200 and Mono Q HR were from Pharmacia. Antimouse IgG, alkaline phosphatase conjugate, bovine serum albumin, and fibrinogen were from Sigma. Fibronectin, collagen IV, fetal bovine serum, and all culture reagents were purchased from GibcoBRL. Collagen I was from Chrono-Log. Casein was from Calbiochem. The intramolecularly quenched fluorogenic substrates were synthesized by Dr. Maria Juliano (Department of Biophysics, UNIFESP, Sa˜o Paulo, Brazil). Endopeptidase Asp-N and endopeptidase Lys-C were of sequencing grade, from Boehringer-Mannheim. All other chemicals were of the highest grade available.
Cell Lines Chinese hamster ovary (CHO) cells transfected with human ␣ IIb 3 (A5) and ␣ v 3 (VNRC3) integrins (26), were kindly provided by Dr. M. Ginsberg (Scripps Research Institute, La Jolla, CA). SW480 cells (derived from a human colon carcinoma) transfected with ␣ 9 1 integrin were a gift from Dr. D. Sheppard (Lung Biology Center, University of San Francisco, CA). Jurkart cells from human acute T cell leukemia and K562 cells from human erythroleukemia were purchased from ATCC (Manassas, VA). K562 cells transfected with ␣ 2 1 and ␣ 1 1 integrin were a gift from Dr. M. E. Hemler (Dana Farber, Boston, MA). Cells were stably transfected and the expression of integrins was confirmed by flow cytometry using monoclonal antibodies against the ␣ 2 integrin subunit (clone AK-2, Pharmigen, San Diego, CA). Nontransfected cells do not express ␣ 2 integrin (27).
Protein Purification Size exclusion chromatography. B. alternatus crude venom (50 mg) was fractionated by size exclusion chromatography on a Superdex-200 column (1.6 cm ⫻ 60 cm), equilibrated with 10 mM Tris– HCl, pH 8.6, plus 100 mM NaCl. Elution was carried out using the same buffer at a flow rate of 0.5 ml/min, and the fractions were tested for enzymatic and hemorrhagic activities. Selected fractions were pooled and submitted to a second chromatography on the same column and under the same conditions. Phenyl Sepharose CL-4B hydrophobic interaction chromatography. The fractions of interest were further separated on a Phenyl Sepharose column (2 ⫻ 18 cm), previously equilibrated with 10 mM Tris–
DISINTEGRIN INHIBITION OF CELL ADHESION HCl, pH 8.6, and 1 M ammonium sulfate. Elution was carried out with a reverse ammonium sulfate gradient (1 to 0 M), at a flow rate of 2 ml/min. Mono Q HR anion exchange chromatography. The active fractions were further separated on a Mono Q Hr column, previously ¨ KTA Explorer 10 equilibrated with 10 mM Tris–HCl, pH 8.6, in an A (Pharmacia) chromatography equipment, and eluted with a NaCl gradient (0 M to 1 M), at a flow rate of 1 ml/min. All purification steps were performed at 4°C.
Enzymatic Assays All chromatographic steps were followed by enzymatic assays of the eluted fractions including proteolytic activity using casein and N-p-tosyl-Gly-Pro-Lys-p-nitroanilide as substrates, thrombin-like activity on purified bovine fibrinogen and clotting activity on bovine plasma (28). Fluorometric enzyme assay. Hydrolysis of fluorogenic peptides was measured following the procedure previously described (29). The assay was carried out for 10 min at 30°C in 50 mM Tris–HCl, pH 8.0 (1.5 ml), or 50 mM Tris–HCl, pH 8.0, 15 mM EDTA (1.5 ml), and the hydrolysis was monitored by measuring the fluorescence at emission of 420 nm and excitation of 320 nm, in a Hitachi F-2000 spectrofluorimeter.
Protein Characterization Protein purification was followed by SDS–PAGE (30) and Western blotting. First antibody was polyclonal and produced in mice using a recombinant metalloprotease/disintegrin from Agkistrodon contortrix laticinctus that was expressed in Escherichia coli (14). The molecular mass of the purified protein was estimated by SDS–PAGE. Protein concentration was determined using Coomassie brilliant blue G250 according to the method of Bradford (31).
N-Terminal Sequencing and Sequence Analysis Amino acid sequence analyses were performed by protein microsequencing system. Edman degradation of the reduced and alkylated (32) protein was performed in a gas-phase sequence equipment (Applied Biosystem mod 473 and a Porton/Beckman 2090 machine), using the conditions recommended by manufacturer. Cleavage of proteins were performed with endopeptidases Glu-C, Asp-N, and Lys-C. S-Pyridinethylated proteins (approximately 2 nmol) were incubated with endopeptidase Asp-N or endopeptidase Lys-C in 50 mM ammonium acetate buffer, pH 6.5 and pH 4.0, respectively, for 14 h at 37°C, with an enzyme:protein ratio of 1:20. Adjustments of the solution to pH 2.0 with formic acid ended the reaction. The amino acid sequences were compared to sequences in the GenBank, SwissProt or PIR databases using the BLAST program (33).
Minimum Hemorrhagic Dose (MHD) Hemorrhagic activity was determined in male mice as previously reported (34). Briefly, mice were injected intradermally with different doses (one dose per mouse) of the purified fraction. Two hours after injection, animals were killed, their skins removed, and the diameters of the hemorrhagic spots measured. The minimum hemorrhagic dose was defined as the amount of protein that produces a halo of 1 cm, 2 h after injection.
Adhesion Assays All cells were cultured in DMEM or RPMI media containing 10% FBS, L-glutamine, penicillin, streptomycin, and geneticin for the transfected cells at 37°C in a water-jacketed CO 2 incubator. Adhe-
343
sion of cultured cells labeled with 5-chloromethylfluorescein diacetate (CMFDA) was performed as described previously (35). Briefly, inhibitors (2.5 g/well), and ligands fibrinogen (1 g/well), fibronectin (1 g/well), collagen type I and IV (0.5 g/well), VCAM-1 (0.3 g/well), were immobilized on 96-well microtiter plates (Falcon, Pittsburgh, PA) in phosphate-buffered saline overnight at 4°C. Wells were blocked with 1% bovine serum albumin in Hank’s balanced salt solution buffer. Cells (5 ⫻ 10 6/ml) were labeled by incubation with 12.5 M 5-chloromethylfluorescein diacetate in Hank’s-balanced salt solution buffer containing 1% bovine serum albumin at 37°C for 30 min. Unbound label was removed by washing with the same buffer. Labeled cells were incubated in the presence or absence of inhibitors before being transferred to the plate (1 ⫻ 10 5 cells/well) and incubated at 37°C for 30 min. After washing to remove unbound cells, the remaining cells were lysed by the addition of 0.5% Triton X-100. In parallel, a standard curve was prepared in the same plate using known concentrations of labeled cells. The plates were read using a Cytofluor 2350 fluorescence plate reader (Millipore, Bedford, MA) with a 485-nm excitation and 530-nm emission filters.
RESULTS
Purification of Alternagin B. alternatus crude venom was separated into about five major peaks after fractionation on a Superdex 200 gel filtration column (Fig. 1a). B. alternatus has a high amount of a hemorrhagic protein of M r 55,000, which is characteristic of metalloproteinases carrying a disintegrin domain (15). Fractions were also tested for fibrinogen-clotting activity and proteolytic activity using casein as substrate (not shown). Fractions having hemorrhagic activity and with the expected size (fraction II) were pooled and applied again into the same column, and eluted in the same conditions. In this step, fraction II was resolved into two major peaks, which were named II-I and II-II (Fig. 1b). Fraction II-I was further purified in a phenyl Sepharose column (Fig. 1c). The selected fractions from this chromatography did not present any fibrinogen-clotting activity, only proteolytic and hemorrhagic activity. SDS–PAGE analysis showed that the purified hemorrhagic protein corresponds to a molecular mass of 55,000 kDa (Fig. 2a) and a minimum hemorrhagic dose of 1 g in mice. This protein was named alternagin. A series of peptide substrates displaying an o-aminobenzoyl (Abz) as fluorophore and a N-(2,4-dinitrophenyl)ethylenediamine (EDDNP) as a quencher was used to analyze the interaction with the metalloprotease. The substrates were based on the sequence of the reactive site of serine protease inhibitors, cysteine proteinase inhibitor, and the human kininogen sequence flanking the tissue kallikrein scissile bonds (29). The substrate with kininogen related quenched sequence Abz-Met-Ile-SerLeu-Met-Lys-Arg-Pro-EDDnp, which includes a partial amino sequence of bradykinin was hydrolyzed by the enzyme, with the following kinetic parameters: K m 2.12 M; k cat 0.036 s⫺1, and k cat/Km 1.7 104 M⫺1. sec⫺1. The hydrolysis was blocked in the presence of 15 mM EDTA confirming the action of the metalloproteinase on the
344
SOUZA ET AL.
FIG. 1. Purification of alternagin and its processed disintegrin domain. Elution profile of Superdex 200 gel filtration chromatography of 50 mg of B. alternatus crude venom (a) and fraction II (b). Elution buffer 10 mM Tris–HCl, pH 8.6, and 100 mM NaCl. Flow rate 0.5 ml/min. The selected fraction is indicated. (c) Phenyl Sepharose CL-4B hydrophobic interaction chromatography of fraction II-I from the gel filtration of crude venom. Fractions were eluted with a reverse gradient (1 M to 0 M) of ammonium sulfate. Flow rate, 2 ml/min; fractions were collected at 2-min intervals. (d) Mono Q anion exchange HPLC of fraction II-II from the second gel filtration. Elution buffer 10 mM Tris–HCl, pH 8.6, plus a NaCl gradient (from 0 to 1 M). Flow rate 1 ml/min. The major peak corresponds to alternagin-C.
substrate. The protease cleaves the substrate at the LeuMet bond (not shown). Fraction II-II does not have any significant proteolytic activity but very low plasma clotting activity. It was further separated into several peaks in an anion exchange chromatography (Fig. 1d). The major peak eluted in this chromatography corresponds to a protein of M r 28,000 (Fig. 2a). Both proteins of M r 55,000 and 28,000 strongly react with an antibody against a recombinant metalloproteinase/disintegrin (Fig. 2b), suggesting that the smaller protein could be the processed disintegrin domain. This protein was named alternagin-C. Usually, about 0.4 mg of purified alternagin-C were isolated from 50 mg of crude venom. The yield of the whole molecule was lower, about 0.2 mg, due to the loss caused by the processing. Purified alternagin does not have any detectable proteolytic activity on casein but the specific activity on the fluorescent substrate was about 211 U/mg. When the B. alternatus crude venom was treated with 20 mM EDTA and 2 mM PMSF before gel filtra-
tion, fraction II-II was markedly decreased (not shown), thus suggesting that this protein was generated by proteolytic processing from a precursor form present in the venom. Partial Amino Acid Sequence of Alternagin-C The N-terminus of alternagin was blocked suggesting the presence of pyroglutamic acid as the N-terminal residue, in agreement with the data in literature for other metalloproteinases from snake venoms (15). In order to better characterize the protein and to confirm the presence or absence of the RGD sequence, only the processed disintegrin domain was further analyzed by amino acid sequencing. The partial sequence of alternagin-C was established by automated Edman degradation of the protein. Peptide fragments produced by Lys-C, Glu-C, and Asp-N endoproteinase cleavage of 4-vinyl pyridine reduced protein were sequenced after isolation by reverse phase chromatography. Enough overlaps were ob-
DISINTEGRIN INHIBITION OF CELL ADHESION
FIG. 2. SDS–PAGE and Western blotting analyses of purified fractions. (A) Coomassie blue R250-stained 15% SDS–polyacrylamide gel. Samples: 1, Molecular markers; 2, B. alternatus crude venom; 3, purified metalloproteinase/disintegrin alternagin (ALT); 4, alternagin-C (ALT-C). (B) The same samples were transferred to nitrocellulose membrane and probed with antirecombinant metalloproteinase/disintegrin serum (1:1,000, B).
tained to define the majority of the primary structure (Fig. 3a). The N-terminal sequence of alternagin-C showed that this sequence has a disintegrin structure, with high identity with the disintegrin domain reported for other PIII snake venom metalloproteinases (13, 14). A sequence comparison of the N-terminus of homologous proteins shows that alternagin-C belongs to the class of disintegrin-like proteins due to the presence of highly conserved amino acid residues in the disintegrin-like protein family (Fig. 3b). The striking difference between the two disintegrin families is the pattern of disulfide bonds, although several other residues in the N-terminal region can differentiate the disintegrins from the disintegrin-like proteins, as shown in Fig. 3b. The position of cysteine residues is highly conserved in the disintegrin domain but there are at least two conserved cysteine substitutions in the RGD-disintegrins and one is located in the proposed adhesive sequence D/ECD (23). Several conserved residues in the disintegrin-like protein family such as Ile 1, Pro 4, Cys 7, Val 14, and Cys 26 (numbering as in Fig. 3b) are also found in alternagin-C. In contrast, RGD-disintegrins have different but also conserved residues in these positions (Fig. 3b). The primary structure of alternagin-C confirmed the homology with the disintegrin-like proteins and the lack of the RGD motif. Although the complete sequence of alternagin-C is not known yet, it can be clearly seen from the sequencing results that it belongs to the family of the disintegrin-like proteins, with the ECD conserved motif.
345
and, among the cells tested, specific for cells expressing ␣ 2 1 integrin (Fig. 4). Neither protein interfered with the adhesion of cells expressing ␣ IIb 3, ␣ 1 1, ␣ 5 1, ␣ 4 1 ␣ V 3, and ␣ 9 1 integrins to other ligands such as fibrinogen, fibronectin collagen IV and VCAM-1 (Table I). Even the main collagen type IV receptor, the integrin ␣ 1 1, was not affected by alternagin and alternagin-C. We have not tested the effect of both proteins on the ␣ 3 1 integrin, which also binds to collagen. The amount of protein that inhibits 50% of the cell adhesion (IC50) was 134 and 100 nM for alternagin and alternagin-C, respectively. The values of IC50 for both proteins show that the disintegrin domain has an essential role in the integrin inhibition. The difference between IC50 values is not statistically significant, but indicates that proteolysis is likely less important because the enzymatically inactive protein alternagin-C has the same activity on cell adhesion as the whole molecule. To investigate if alternagin or alternagin-C could act as adhesion molecules by themselves, wells were coated with the two purified proteins and further incubated with ␣ 2 1-tranfected and nontransfected K562 cells. The control of nontransfected cells was made using fibronectin as the ligand since these cells do not bind to collagen. Nontransfected K562 cells did not adhere to alternagin or alternagin-C (Fig. 5A). In contrast, the two proteins induced significant adhesion of ␣ 2 1-transfected K562 cells thus demonstrating that the disintegrin domain also binds to the ␣ 2 1 integrin (Fig. 5B). These results support the idea that the disintegrin domain probably inhibits collagen binding to the ␣ 2 1 integrin by competition. We next compared the data in the literature regarding the inhibition of collagen binding to platelet ␣ 2 1 integrin induced by snake venom metalloproteinases and their disintegrin domains. In platelets the disintegrin domain seems to be less active than the whole molecule (Table II), although the data were collected from different laboratories, assayed in different conditions regarding type and concentration of collagen, and in some cases, with native and recombinant proteins, which makes the comparison more difficult. The results presented in this paper were collected in the same conditions, and therefore the comparison is more reliable allowing the conclusion that the disintegrin domain is the domain responsible for the collagen inhibition of ␣ 2 1 integrin in K562 transfected cells. Jararhagin (whole molecule) was also tested in the same cell adhesion assay, giving similar results (36). However, the authors have not tested the processed disintegrin domain in this system.
Adhesion Studies Alternagin and alternagin-C are potent inhibitors of the collagen binding to ␣ 2 1 integrin. Their ability to inhibit collagen-induced adhesion was dose-dependent
DISCUSSION
This paper describes the isolation and structural characterization of a novel metalloproteinase/disinte-
346
SOUZA ET AL.
FIG. 3. (A) Amino acid sequence of alternagin-C. The peptides obtained after fragmentation by enzyme digestion with Lys-C, Glu-C, and Asp-N endoproteases are indicated. (B) Sequence comparison of the processing region of the disintegrin domain in RGD-disintegrins and disintegrin-like proteins. References: alternagin, this work; jararhagin (19); ACLD (14); catrocollastatin (17); Echis I (49); trigramin (50); trimucrin (51); rhodostomin (52); halystatin (53), and atrolysin e (13). The first five sequences are from disintegrin-like proteins and the others are RGD-disintegrins. The amino acid residues conserved in each family are indicated by arrowheads.
347
DISINTEGRIN INHIBITION OF CELL ADHESION
FIG. 4. Effect of Alternagin and Alternagin-C on adhesion of K562␣ 2 1 cells to collagen type I. Collagen I (0.5 g/well) was immobilized overnight at 4°C on a 96-well plate in phosphate-buffered saline (PBS). After blocking with 1% BSA, the 5-chloromethyfluorescein diacetate-labeled cells incubated with different concentration of inhibitors were added to each well. The plate was then incubated at 37°C for 30 min. The adhesion experiment was performed as described under Materials and Methods. Error bars indicate S.E. from three independent duplicated experiments.
grin and the processed disintegrin domain from B. alternatus venom. The two proteins were named alternagin and alternagin-C to be consistent with the literature (20, 21). Although the inhibitory effect of this class of protein on collagen-induced platelet aggregation is well established, it is not known which domain is responsible for this effect. Data in the literature so far are mostly from independent groups using native and recombinant proteins, which could make the comparison more difficult. Here we describe the inhibitory effect of a metalloproteinase/disintegrin on a different cell system, and we compare the results obtained with the whole molecule and the processed disintegrin/cysteine-rich domain in the same system. Although the complete amino acid sequence of alternagin is not yet known, we suggest that alternagin-C is derived by limited proteolysis from alternagin or one of its isoforms, as usually found in snake venoms. The proteolytic activity of alternagin was tested in two different assays. Hemorrhagic activity is a conse-
quence of the proteolytic digestion of basal membrane components such as collagen, laminin and fibronectin (12). The MHD for alternagin is 1 g in mice, which is consistent for this class of snake venom metalloproteinase (15). In a more refined assay, alternagin was shown to cleave a synthetic peptide with a bradykinin related sequence, and the kinetic parameters were defined. Preliminary results indicated that alternagin also cleaves human high molecular weight kininogen but further investigation is necessary to identify the products of kininogen hydrolysis. Jararhagin, a homologous protein was shown to process the tumor necrosis factor (TNF)-␣ from its precursor on cell membrane in vitro (37). These results show that svMP may be very useful tools for the study of important physiological processes. Integrins ␣ 1 1 and ␣ 2 1 are the major cellular receptors for collagen, and the collagen binding domain called I domain, is located within the ␣ subunit (38). Alternagin and the processed disintegrin/cysteine-rich domain, alternagin-C, are potent inhibitors of the ␣ 2 1 integrin. Their ability to inhibit collagen-induced adhesion is dose-dependent and specific for cells expressing ␣ 2 1 integrin. Even the collagen type IV receptor, the integrin ␣ 1 1, was not affected by alternagin. The inhibitory effect of a disintegrin domain on the ␣ 2 1 integrin was only described for platelets (17, 20, 21). Here we show that other cells can be affected by this class of proteins and several types of integrins are insensitive to alternagin-C. Recently, it has been also demonstrated that jararhagin also inhibits the collagen binding to ␣ 2 1-transfected K562 cells (36). However, the inactivation of jararhagin with o-phenanthroline only slightly increased the IC50 for the collagen binding, which gives more evidence that the proteolysis does not have a significant role in this effect. It has been proposed that the metalloproteinase domain has a significant contribution to the inhibitory effect due to the presence of a ␣ 2-I domain-binding motif (RKK) in this domain (25). It was shown that this
TABLE I
Effect of Alternagin (ALT) and Alternagin-C (ALT-C) on the Adhesion of Cells Expressing Different Integrins IC 50 (nM) Cell suspension
Integrins
A5 VNRC3 K562 K562 (␣ 2) K562 (␣ 1) SW480 (␣ 9) Jurkart
␣ IIb 3 ␣v 3 ␣ 5 1 ␣ 2 1 ␣ 1 1 ␣ 9 1 ␣ 4 1
Ligand Fibrinogen Fibronectin Fibronectin Collagen I Collagen IV VCAM-1 VCAM-1
Assay CA CA CA CA CA CA CA
ALT
ALT-C
⬎5000 ⬎5000 ⬎5000 134 ⬎5000 ⬎5000 ⬎5000
⬎5000 ⬎5000 ⬎5000 100 ⬎5000 ⬎5000 ⬎5000
348
SOUZA ET AL.
FIG. 5. Adhesion of (A) non transfected and (B) transfected K562 cells with ␣ 2 1 integrin to immobilized inhibitors. Alternagin and Alternagin-C (2.5 g/well) were immobilized overnight at 4°C on a 96-well plate in phosphate-buffered saline (PBS). After blocking with 1% BSA, the 5-chloromethyfluorescein diacetate-labeled cells were added to each well and the plate was incubated at 37°C for 30 min. Error bars indicate S.E. from three independent experiments. Controls were made with immobilized fibronectin (1.0 g/well, in A) and collagen I (0.5 g/well, in B).
motif recognizes the I domain in the ␣2-subunit therefore inhibiting the binding of collagen. Crovidisin, isolated from Crotalus viridis venom, is a protein similar to catrocollastatin that also binds directly to collagen fibers thus blocking the interaction of platelets and collagen (39), although in this case it is not known which domain has this collagen-binding activity. Another important contribution is the fact that the active protease could cleave the  1-subunit therefore impairing the binding of collagen to platelets, as it has been demonstrated for jararhagin (24). In fact, inactivation of jararhagin, a homologous protein with 1,10-phenanthroline increased the IC50 from 40 to 140 (24). The cleavage of the  1-subunit results in inhibition of the collagen-stimulated phosphorylation events associated with the platelet response (40). Therefore, the metal-
loprotease domain has a significant contribution to the inhibition of collagen-induced platelet aggregation. On the other hand, the disintegrin domain would act as an antagonist of the interaction of ␣ 2 1 with collagen. Our results show that the processed domain is a potent inhibitor of this integrin and can completely inhibit cell adhesion in the absence of the catalytic domain. Disintegrins have received much attention in the recent years due to the discovery of homologous membrane-bound proteins in mammalian tissues, which seem to have important roles in several physiological processes, including fertilization, cell differentiation, and shedding of receptors (41). These proteins are named ADAMs (for a disintegrin and metalloprotease) or MDCs (for metalloprotease/disintegrin/cysteine-rich proteins). More than 23 ADAMs were described in
TABLE II
Comparison of the Inhibitory Effects of svMDC and Isolated Disintegrin/Cysteine-Rich Domains on the Collagen I Receptor Protein
Source
Cell system
Ligand
IC50 (nM)
Reference
Alternagin Alternagin-C a Jararhagin Jararhagin-I b Jararhagin Jararhagin Jararhagin-I b Jararhagin-C Catrocollastatin Catrocollastatin-C Atrolysin a Atrolysin a D/C Crovidisin
Bothrops alternatus Bothrops alternatus Bothrops jararaca Bothrops jararaca Bothrops jararaca Bothrops jararaca Bothrops jararaca Bothrops jararaca Crotalus atrox Crotalus atrox Crotalus atrox Crotalus atrox Crotalus viridis
K562 ␣21 K562 ␣21 K562 ␣21 K562 ␣21 Platelet Platelet Platelet Platelet Platelet Platelet Platelet Platelet Platelet
Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen Collagen
134 100 50 140 50 40 140 100 50 66 110 470 170
This work This work 36 36 18 24 24 21 17 20 23 23 39
a b
The symbol (-C) means the disintegrin, cysteine-rich domain. 1,10-Phenanthroline-inactivated jararhagin.
DISINTEGRIN INHIBITION OF CELL ADHESION
many different tissues from several sources including mammals, Drosophila melanogaster and Caenorrhabits elegans (42, 43). Disintegrins from snake venoms have become important tools for understanding the roles of these proteins in cell adhesion since they can be isolated from venoms with relative ease and at significant yield. The motifs in the disintegrin/cysteine-rich domain that could be involved in the integrin recognition are not known, but it has been demonstrated that synthetic peptides having the sequence RSECD in which the cysteine is constrained by a disulfide bond, inhibit collagen-induced platelet aggregation (23). The disintegrin/cysteine-rich domain could also be divided into two domains, depending on the pattern of disulfide bonds. There are few papers on the effect of the proposed cysteine-rich domain on cell adhesion. Since there is no report on the isolation of such domain, the work has been done with recombinant proteins. It has been suggested that the cysteine-rich region of human ADAM 12 mediates tumor cell adhesion (44, 45). The cysteine-rich domain of fertilin ␣ mediates the spermegg binding during fertilization (46). More recently, it was demonstrated that the recombinant cysteine-rich domain of atrolysin A from Crotalus atrox venom inhibits the binding of MG-63 cells to collagen. This osteosarcoma cell line expresses predominantly the integrin ␣ 2 1 on its cell surface (47). These data suggest that the cysteine-rich domain must have an important role in cell adhesion. Our results are in agreement with this conclusion since the cysteine-rich domain is a part of the disintegrin/cysteine-rich domain. Since the disulfide bond pattern nor the three-dimensional structure of a disintegrin-like protein are not known yet, the contribution of the overall conformation to the integrin binding is a more difficult question to solve. We have also demonstrated that the disintegrin domain may mediate cell adhesion in an in vitro assay. Similar results were found for fertilin which was demonstrated to act as an adhesion molecule in the spermegg binding (46). Therefore, the disintegrin domain must have adhesive sequences that compete with collagen in the integrin binding. In conclusion, the present study shows that the disintegrin-like domain of a svMP binds with specificity to ␣ 2 1-transfected K562 cells, therefore inhibiting collagen-mediated cell adhesion. It is also shown that this protein may act as an adhesion molecule by itself. As far as we know, this is the first study about the interaction of the disintegrin-like proteins and other integrins in a cell adhesion assay. These results may aid in the understanding of the structure–function relationship of disintegrins as well as the molecular basis of cell adhesion. Recent studies have shown that ovarian carcinoma cells adhere preferentially to type I collagen (48). Therefore, disintegrins that specifically inhibit
349
the collagen binding to its receptor are important tools for the study of this type of cancer metastasis and for the development of specific peptide-derived inhibitors of tumor cell invasion. ACKNOWLEDGMENTS We thank Dr. Charlotte L. Ownby for critical reading of the manuscript and the technical support of Mariola Marcinkiewicz in the cell culture assays. This work was supported by grants from FAPESP, Brazil (95/9300-7, 96/08863-0, and 97/05398-8), CNPq, Brazil (521542/96-0), American Heart Association (U.S.A.), and International Foundation for Science, Sweden (F/2631-1).
REFERENCES 1. Dennis, M. S., Henzel, W. J., and Pitti, R. M. (1989) Proc. Natl. Acad. Sci. USA 87, 2471–2475. 2. Gould, R. J., Polokoff, M. A., Friedman, P. A., Huang, T-F., Holt, J. C., Cook, J. J., and Niewiarowski, S. (1990) Proc. Soc. Exp. Biol. Med. 195, 168 –171. 3. Niewiarowski S., McLane M. A., Kloczewiak M., and Stewart, G. J. (1994) Semin. Hematol. 31, 289 –290. 4. Hynes, R. O. (1992) Cell 48, 549 –554. 5. Yamada, K. M. (1991) J. Biol. Chem. 266, 12809 –12812. 6. Huang, T-F., Holt, J. C., Lukasiewcz, H., and Niewiarowski, S. (1987) J. Biol. Chem. 262, 16157–16163. 7. Maruyama, K., Kawasaki, T., Sakai, Y., Taniuchi, Y., Shimizu, M., Kawashima, H., and Takenaka, T. (1997) Peptides 18, 73–78. 8. Sheu, J. R., and Huang, T-F. (1994) J. Pharm. Pharmacol. 46, 58 – 62. 9. Sheu, J.-R., Yen, M.-H., Peng, H.-C., Chang, M.-C., and Huang, T.-F. (1995) Eur. J. Pharmacol. 294, 231–238. 10. Beviglia L., Stewart G. J., and Niewiarowski, S. (1995) Oncol. Res. 7,7–20. 11. Danen, E. H. J., Marcinkiewicz, C., Cornelissen, I. M. H. A., van Kraats, A. A., Pachter, J. A., Ruiter, D. J., Niewiarowski, S., and van Muijen, G. N. P. (1998) Exp. Cell Res. 238, 188 –196. 12. Baramova, E. N., Shannon, J. D., Bjarnason, J. B., and Fox, J. W. (1989) Arch. Biochem. Biophys. 275, 63–71. 13. Hite, L., Jia, L., Bjarnason, J. B., and Fox, J. W. (1994) Arch. Biochem. Biophys. 308, 182–191. 14. Selistre de Araujo, H. S., Souza, D. H. F., and Ownby, C. L. (1997) Biochim. Biophys. Acta 1342 (2), 109 –115. 15. Bjarnason, J. B., and Fox, J. W. (1994) Pharmacol. Ther. 62, 325–372. 16. Yamada, D., Shin, Y., and Morita, T. (1999) FEBS Lett. 451, 299 –302. 17. Zhou, Q., Smith, J. B., and Grossman, M. H. (1995) Biochem. J. 307, 411– 417. 18. De Luca, M., Ward, C. M., Ohmori, K., Andrews, R. K., and Berndt, M. C. (1995) Biochem. Biophys. Res. Commun. 206, 570 –576. 19. Paine, M. J. I., Desmond, H. P., Theakston, R. D. G., and Crampton, J. M. (1992) J. Biol. Chem. 267, 22869 –22876. 20. Shimokawa, K-I., Shannon, J. D., Jia, L-G., and Fox, J. W. (1997) Arch. Biochem. Biophys. 343, 35– 43. 21. Usami, Y., Fujimura, Y., Miura, S., Shima, H., Yoshida, E., Yoshioka, A., Hirano, K., Suzuki, M., and Titani, K. (1994) Biochem. Biophys. Res. Commun. 201, 331–339. 22. Shimokawa, K-I., Jia, L-G., Shannon, J. D., and Fox, J. W. (1998) Arch. Biochem. Biophys. 354, 239 –246.
350
SOUZA ET AL.
23. Jia, L-G., Wang, X-M., Shannon, J. D., Bjarnasson, J. B., and Fox, J. W. (1997) J. Biol. Chem. 272, 13094 –13102. 24. Kamiguti, A. S., Hay, C. R. M., and Zuzel, M. (1996) Biochem. J. 320, 635– 641. 25. Ivaska, J., Ka¨pyla´, J., Pentika¨inen, O., Hoffre´n, A-M., Hermonen, J., Huttunen, P., Johnson, M. S., and Heino, J. (1999) J. Biol. Chem. 274, 3513–3521. 26. O’Toole, T. E., Loftus, J. C., Du, X., Glass, A., Ruggeri, Z. M., Shattil, S. J., Plow, E. F., and Ginsberg, M. H (1990) Cell Regul. 1, 883– 893. 27. Marcinkiewicz, C., Lobb, R. R., Marcinkiewicz, M. M., Daniel, J. L., Smith, J. B., Dangelmaier, C., Weinreb, P. H., Beacham, D. A., and Niewiarowski, S. Biochemistry, in press. 28. Selistre, H. S., and Giglio, J. R. (1987) Toxicon 25, 1135. 29. Oliva, M. L. V., Andrade, S. A., Batista, I. F. C., Sampaio, M. U., Juliano, M., Fritz, H., Auerswald, E. A., and Sampaio, C. A. M. (1999) Immunopharmacology 45, 145–149. 30. Laemmli, U. K. (1970) Nature 227, 680 – 685. 31. Bradford, M. M. (1976) Anal. Biochem. 72, 248 –254. 32. Friedman, M., Krull, L. H., and Cavins, J. F. (1970) J. Biol. Chem. 245, 3868. 33. Altschul, S. F., Madden, T. L., Scha¨ffer, A. A., Zhang, J., Zhang, Z., Miller, W., and Lipman, D. J. (1997) Nucleic Acids Res. 25, 3389 –3402. 34. Johnson, E. K., and Ownby, C. L. (1993) Int. J. Biochem. 25(2), 267–278. 35. Marcinkiewicz, C., Calvete, J. J., Vijay-Kumar, S., Marcinkiewicz, M. M., Raida, M., Shick, P., Lobb, R. R., and Niewiarowski, S. (1999) Biochemistry 38, 13302–13309. 36. Moura-da-Silva, A. M., Marcinkiewicz, C., Marcinkiewicz, M. M., and Niewiarowski, S., Thromb. Res., in press. 37. Moura-da-Silva, A. M., Laing, G. D., Paine, M. J. I., Dennison, J. M. T. J., Politi, V., Crampton, J. M., and Theakston, R. D. G. (1996) Eur. J. Immunol. 26, 2000 –2005.
38. Takada, Y., and Helmer, M. E. (1989) J. Cell Biol. 109, 397– 407. 39. Liu, C-Z., and Huang, T-F. (1997) Arch. Biochem. Biophys. 337, 291–299. 40. Kamiguti, A. S., Markland, F. S., Zhou, Q., Laing, G. D., Theakston, R. D. G., and Zuzel, M. (1997) J. Biol. Chem. 272, 32599 – 32605. 41. Killar, L., White, J., Black, R., and Peschon, J. (1999) Ann. N.Y. Acad. Sci. 878, 442– 452. 42. Rooke, J., Pan, D., Xu, T., and Rubin, G. M. (1996) Science 273, 1227–1231. 43. Wen, C., Metzstein, M. M., and Greenwald, I. (1997) Development 124, 4759 – 4767. 44. Iba, K., Albrechtsen, R., Gilpin, B. J., Loechel, F., and Wewer, U. M. (1999) Am. J. Pathol. 154, 1489 –1501. 45. Zolkiewska, A. (1999) Exp. Cell Res. 252, 423– 431. 46. Evans, J. P., Schultz, R. M., and Kopf, G. S. (1997) Dev. Biol. 187, 94 –106. 47. Jia, L-G., Wang, X-M., Shannon, J. D., Bjarnasson, J. B., and Fox, J. W. (2000) Arch. Biochem. Biophys. 373, 281–286. 48. Fishman, D. A., Kearns, A., Chilukuri, K., Bafetti, L. M., O’Toole, E. A., Georgacopoulos, J., Ravosa, M. J., and Stack, M. S. (1998) Invasion Metastasis 18, 15–26. 49. Paine, M. J. I., Moura-da-Silva, A. M., Theakston, R. D. G., and Crampton, J. M. (1994) Eur. J. Biochem. 224, 483– 488. 50. Huang, T-F., Holt, J. C., Kirby, E. P., and Niewiarowski, S. (1989) Biochemistry 28, 661– 666. 51. Tsai, I-H., Wang, Y-M., and Lee, Y-H. (1994) Biochim. Biophys. Acta 1200, 337–340. 52. Au, L-C., Huang, Y-B., Huang, T. F., Teh, G-W., Lin, H-H., and Choo, K-B. (1991) Biochem. Biophys. Res. Commun. 181, 585– 593. 53. Fujisawa, Y., Kuroda, S., Notoya, K., Konishi, H., and Terashita, Z. (1994) J. Takeda Res. Lab. 53, 39 –56.