Vipera lebetina Venom Contains Two Disintegrins Inhibiting Laminin-binding (cid:1) 1 Integrins*

To explain the myotoxic effects of snake venoms, we searched for inhibitors of (cid:2) 7 (cid:1) 1 integrin, the major lami- nin-binding integrin in skeletal muscle. We discovered two inhibitors in the venom of Vipera lebetina . One of them, lebein-1 (known as lebein), has already been proposed to be a disintegrin because of its RGD-containing primary sequence. The other, lebein-2, is a novel protein that also interacts firmly with (cid:2) 3 (cid:1) 1 , (cid:2) 6 (cid:1) 1 , and (cid:2) 7 (cid:1) 1 inte- grins, but not with the collagen-binding (cid:2) 1 (cid:1) 1 and (cid:2) 2 (cid:1) 1 integrins. Ligand binding of laminin-recognizing (cid:1) 1 integrins was efficiently blocked by both lebein-1 and le-bein-2. In cell attachment assays, lebein-1 and lebein-2 inhibited myoblast attachment not only to laminin, but also to fibronectin. However, neither lebein-1 nor leb-ein-2 interacted with (cid:2) 7 (cid:1) 1 integrin in an RGD-depend- ent manner, similar to the interaction of the laminin with (cid:2) duplicate

Integrins are cell adhesion molecules consisting of two genetically non-related ␣ and ␤ subunits (reviewed in Refs. 1 and 2). Different combinations of these subunits give rise to 24 known integrin receptors with distinct ligand-binding specificities. The ␤ 1 subunit-containing integrins are receptors mainly for extracellular matrix (ECM) 1 proteins such as collagen, laminin, and fibronectin and are responsible for cell anchorage and motility. Moreover, by transducing mechanical forces between cells and the ECM, integrins maintain the tension within tissue and, consequently, tissue shape and structure. At the myotendinous junction, ␤ 1 integrins, especially ␣ 7 ␤ 1 integrin, are thought to play a pivotal role (3,4). Disruption of the ␣ 7 integrin gene leads to progressive muscle dystrophy, whereas the conditional knockout of the ␤ 1 integrin subunit in cardiac muscle results in lethal cardiac failure (4). However, other integrins such as ␣ 3 ␤ 1 , ␣ 6 ␤ 1 , and ␣ 6 ␤ 4 (5) and nonintegrin receptors such as ␣-dystroglycan (6) also bind to laminins, albeit with different preferences for various laminin isoforms (7,8).
Laminins are major constituents of basement membranes (9,10), and their isoform compositions vary with the tissue. The basement membrane of the myotendinous junction is enriched in laminin-2 (11). Merely differing in its ␣ chain, laminin-2 has the ␤1 and ␥1 chains in common with laminin-1. The networks of both laminin and type IV collagen, together with smaller proteins such as nidogen and proteoglycans, form the molecular basis of the basement membrane. The laminin meshwork of the basement membrane allows integrin-mediated anchorage of muscle fibers and other cells (9,10). In addition, basement membranes separate the connective tissue from any other tissue. Only certain cells such as leukocytes and invasively growing tumor cells are able to penetrate through the basement membrane. Other processes such as wound healing require cell movement along the basement membrane. Cell migration through and along the basement membrane is mediated by laminin-binding integrins (5,12).
Some snake venoms contain components that are selectively directed against integrins. These so-called disintegrins interfere with the ability of integrins to bind to their cognate ECM ligands. Hence, cell/matrix interactions are disrupted, and tissue integrity is destroyed. However, snake venoms are known to exert several different effects on snakebite victims, among which are failure of the cardiovascular system and damages to muscle and neural tissue. In addition to disintegrins, these pleiotropic effects are caused by a variety of venom proteins. Being abundant in most snake venoms, phospholipase A 2 is mainly responsible for myotoxic effects (13,14). Various proteases of snake venoms specifically cleave blood-clotting factors or components of the ECM, thus resulting in bleeding dysfunctions and tissue necrosis (15)(16)(17).
Most disintegrins known to date contain an RGD sequence mimicking the ligands for the platelet ␣ IIb ␤ 3 and other RGDdependent integrins (18), which cause failure of ␣ IIb ␤ 3 integrinmediated platelet aggregation and blood clotting. Moreover, disintegrins are not limited to snake venoms, but are also found as domains in ADAM (a disintegrin and a metalloproteinase) family members, multidomain proteins that are abundant in a wide range of animal species (15,19). ADAM proteins play important roles in various biological processes such as fertilization and cell/cell interactions and proteolytic processing (19). Whereas few ADAM proteins interact with integrins (19 -21), snake venom disintegrins interfering with integrin/laminin interaction have not been described so far.
From the venom of Vipera lebetina, we have purified and characterized two inhibitors that target laminin-binding integrins. One of them is the recently described lebein. Although it has been proposed to be a disintegrin for RGD-dependent integrins (22), we show in this work that lebein or lebein-1, as we refer to it, also avidly binds to the laminin-binding ␤ 1 integrins in an RGD-independent manner. In addition, we isolated a new disintegrin (lebein-2) with homology to lebein-1 and similar binding specificity. The interaction of both inhibitors with laminin-binding ␤ 1 integrins both in vitro and in in vivo cell attachment assays demonstrates their potential to manipulate integrin-dependent cell/laminin interactions.

Recombinant Production of Soluble
Integrins-For the construction of pUC-hygMT-␣7X2-Fos, the ␣ 7 integrin cDNA was amplified by reverse transcription-PCR from RNA isolated from mouse myotubes and verified by sequencing. The ␣ 7 X2 integrin ectodomain cDNA encoding amino acids 1-1036 was generated by PCR, introducing an additional SalI site at the 3Ј-end. The ␣ 3 integrin ectodomain-coding cDNA of the previously described vector pUC-hygMT-␣3-Fos (7) was replaced by the ␣ 7 X2 integrin ectodomain.
The construct pUC-hygMT-␣6X1-Fos, coding for the ␣ 6 integrin ectodomain splice variant X1 (amino acids 1-989 of the published sequence) (23), was generated by isolating the HindIII-and XbaIflanked cDNA fragment encoding amino acids 1-938 of the ␣ 6 integrin ectodomain from pRC-CMV-␣6X1 (kindly provided by Dr. A. Sonnenberg, The Netherlands Cancer Institute, Amsterdam, The Netherlands) and by synthesizing an XbaI-and SalI-flanked cDNA fragment encoding amino acids 939 -989 of the ␣ 6 integrin ectodomain by standard PCR. Taking advantage of the SalI site, the two cDNA fragments were directionally inserted into the pUC-hygMT-␣3-Fos vector (7) from which the ␣ 3 integrin ectodomain-coding cDNA had been removed.
Soluble ␣ 2 ␤ 1 and ␣ 3 ␤ 1 integrins were generated and isolated as described previously (7,24). The generation and isolation of soluble ␣ 1 ␤ 1 integrin will be described elsewhere. 2 Integrin Inhibition ELISA-To screen for inhibition of integrin-mediated laminin binding, microtiter plates were coated with 6 g/ml murine laminin-1 (kindly provided by Dr. R. Timpl, Max Planck Institute for Biochemistry, Martinsried, Germany) or 10 g/ml human laminin-2/4 (Merosin, Invitrogen, Karlsruhe, Germany) in Tris-buffered saline (TBS), pH 7.4, containing 1 mM MgCl 2 (TBS/MgCl 2 buffer) overnight at 4°C. Nonspecific protein-binding sites were blocked with 1% heat-denatured bovine serum albumin (BSA) in TBS/MgCl 2 buffer for 2 h at room temperature. Together with protease inhibitors, phenylmethylsulfonyl fluoride, and 1,10-phenanthroline (each at 1 mM) and aprotinin, leupeptin, and pepstatin (each at 3 g/ml), soluble ␣ 7 ␤ 1 integrin was added at a concentration of 15 nM in 1% heat-denatured BSA in TBS, pH 7.4, containing 2 mM MgCl 2 and 1 mM MnCl 2 , either without any supplements (positive control) or with 10 mM EDTA (nonspecific binding, negative control) or with snake venom solutions of 2 mg/ml. Lyophilized snake venoms were purchased from the Berchtes-gadner Schlangenfarm (Berchtesgaden, Germany) and Sigma. After 2 h of incubation at room temperature, wells were washed twice with 50 mM HEPES, pH 7.5, 150 mM NaCl, 2 mM MgCl 2 , and 1 mM MnCl 2 . Bound ␣ 7 ␤ 1 integrin was fixed with 2.5% glutaraldehyde in the same buffer for 10 min at room temperature and detected in an ELISA procedure using rabbit antiserum directed against the human ␤ 1 integrin subunit (1: 400; kind gift of Dr. K. Kü hn, Max Planck Institute for Biochemistry) and an alkaline phosphatase-conjugated anti-rabbit IgG antibody (1: 600; Sigma) as described previously (7). Nonspecific binding was measured in the presence of 10 mm EDTA and subtracted from all values. For the calculation of the relative inhibitory activity, binding signals in the presence of inhibitor were normalized to the non-inhibited control.
Isolation of Lebein-1 and Lebein-2-Lyophilized venom of V. lebetina was dissolved in 20 mM sodium phosphate, pH 6.5, 50 mM sodium chloride, and 1 mM EDTA and separated by gel filtration on a Superose 6 column (Amersham Biosciences AB, Uppsala, Sweden) in the same buffer. The fractions containing the ␣ 7 ␤ 1 integrin-inhibiting fractions were pooled; diluted with 20 mM MES, pH 6.0; and loaded onto a Mono S column (Amersham Biosciences AB). The ␣ 7 ␤ 1 integrin inhibitory activities of lebein-2 and lebein-1 were eluted in linear sodium chloride gradients at 80 and 140 mM, respectively. The eluate fractions were individually pooled and separated on a C 8 reversed-phase column (Nucleosil, Macherey Nagel) in a linear gradient from 0.1% trifluoroacetic acid in water to 80% acetonitrile in 0.08% trifluoroacetic acid/ water. Lebein-1 and lebein-2 were eluted as individual peaks, lyophilized, and dissolved in water. Protein concentration was determined by the BCA assay. Purity was proven by SDS-PAGE.
Cell Adhesion Inhibition Tests-Microtiter plates were coated with fibronectin (gift of Dr. M. Humphries, University of Manchester) and murine laminin-1 at 10 and 40 g/ml, respectively, overnight at 4°C. After blocking for 6 h with 1% BSA in water, 100 l of a C2C12 cell suspension (0.8 ϫ 10 6 cells/ml) in Dulbecco's modified Eagle's medium without fetal calf serum and containing a 1:3 serial dilution of lebein-1 and lebein-2 were added to the wells and incubated for 45 min at 37°C in a humidified 5% CO 2 incubator. After the supernatant had carefully been discarded, attached cells were washed with phosphate-buffered saline and fixed for 10 min at room temperature with 70% ethanol. After air-drying, cells were stained with 0.1% crystal violet. For 100% binding values, lebein-1 and lebein-2 were replaced by an equal volume of Dulbecco's modified Eagle's medium. Three experiments with duplicate measurements were carried out. 2 S. Niland and J. A. Eble, manuscript in preparation.

Recombinant
Production of a Soluble ␣ 7 ␤ 1 Integrin-␣ 7 X2␤ 1 integrin is the major laminin receptor in adult skeletal muscle. To establish a protein interaction assay, we first generated soluble ␣ 7 X2␤ 1 integrin, similar to the recombinant production of soluble ␣ 3 ␤ 1 and ␣ 2 ␤ 1 integrins (7,24). The ␣ 7 X2-Fos/␤ 1 -Jun integrin ectodomain heterodimer, hereafter called soluble ␣ 7 ␤ 1 integrin, consists of the ectodomains of both integrin subunits, to the C termini of which the dimerizing motifs of the transcription factors Fos and Jun have been fused. Although the cDNAs of the ␣ 7 X2 and ␤ 1 chains were from murine and human origins, respectively, the integrin heterodimer is formed and secreted by the insect cells with an expression yield (ϳ600 g/liter) comparable to that of soluble ␣ 3 ␤ 1 integrin (7). Soluble ␣ 7 ␤ 1 integrin was purified by affinity chromatography on a resin to which the cell-binding domain of invasin had been immobilized (Fig. 1A).
The biological activity of the soluble ␣ 7 ␤ 1 integrin was tested qualitatively and quantitatively on different ECM substrates. As shown in Fig. 1B, ␣ 7 ␤ 1 integrin showed a higher binding signal for laminin-1 than for laminin-2/4. Laminin-2 is the most abundant laminin isoform of basement membranes in muscle tissue. In contrast, laminin-5 was not recognized. As expected from the purification protocol, invasin, a surface protein of Yersinia bacteria, also interacted with soluble ␣ 7 ␤ 1 integrin (Fig. 1B). Like other integrins, ␣ 7 ␤ 1 integrin is activated for ligand binding by Mn 2ϩ ions. However, unlike other integrins, this activation seems sufficient, as addition of the integrin-activating antibody 9EG7, directed against the ␤ 1 integrin subunit (25), did not increase the binding further. In the presence of EDTA, binding to any ligand was completely abolished, thus confirming the dependence of ␣ 7 ␤ 1 integrin on divalent cations (Fig. 1B).
Binding of soluble ␣ 7 X2␤ 1 integrin to immobilized laminin-1 and laminin-2/4 was further studied by an ELISA-type titration assay, from which apparent K d values could be calculated according to Heyn and Weischet (26). As shown in Fig. 1C, specific binding of soluble ␣ 7 ␤ 1 integrin to the immobilized laminin isoforms reached saturation. In the presence of Mn 2ϩ ions, ␣ 7 ␤ 1 integrin bound more avidly to laminin-1 (K d ϭ 0.2 nM) than to laminin-2/4 (K d ϭ 2.3 nM). Addition of the activating antibody 9EG7 did not shift the titration curves to lower ␣ 7 ␤ 1 integrin concentrations and did not alter its apparent affinity constants significantly. However, Ca 2ϩ ions reduced ␣ 7 ␤ 1 integrin binding to both laminin-1 and laminin-2/4, thus increasing the apparent K d values by ϳ15-fold (Fig. 1C).
Screening Various Snake Venoms for Their Capability to Interfere with ␣ 7 ␤ 1 Integrin Binding to Laminin-1-Playing a key role in skeletal muscle, ␣ 7 ␤ 1 integrin conceivably is a target for myotoxic snake venoms. Because of its high affinity for ␣ 7 ␤ 1 integrin, we chose laminin-1 as its interaction partner to search for a snake venom inhibitor to this interaction (Table I). Crude snake venoms and soluble ␣ 7 ␤ 1 integrin were added to laminin-1-coated microtiter plates. Protease inhibitors were also supplemented to avoid proteolytic degradation of integrin or its ligand by the numerous proteases that are abundant in snake venoms. Of 33 snake venoms tested that may cause severe muscular dysfunctions, such as venoms of the Elapidae, Viperidae, and Crotalidae families, only the venom of V. lebetina showed a drastic abolition of ␣ 7 ␤ 1 integrin binding to laminin-1 ( Table I). Titration of the V. lebetina venom demonstrated that it suppressed ␣ 7 ␤ 1 integrin binding entirely to both laminin-1 and laminin-2/4 in a dose-dependent manner (Fig. 2).
Purification of the ␣ 7 ␤ 1 Integrin Inhibitory Activity of the V. lebetina Venom-To purify the ␣ 7 ␤ 1 integrin inhibitor, the venom was separated by gel filtration, ion-exchange chromatography, and reversed-phase chromatography. From the gel filtration column, the ␣ 7 ␤ 1 integrin inhibitory activity was eluted in a single peak (Fig. 3A, gray bar). However, further purification on a Mono S column separated the inhibitory activity into two different fractions, called MS-I and MS-II (Fig.  3B). Each of these two inhibitory peaks, MS-I and MS-II, showed a characteristic elution profile upon reversed-phase chromatography (Fig. 4, A and B). Fraction MS-I could not further be separated by reversed-phase chromatography and contained only one inhibitor that eluted as a single peak (Fig. 4A).
After reversed-phase chromatography, the major inhibitory peak of fraction MS-II (Fig. 4B) was identified as lebein, a recently discovered disintegrin (22), which we refer to as lebein-1. Edman degradation of lebein-1 revealed two staggered N-terminal sequences, MNGSNPXXD and NGSNPXXD, at a 2:1 ratio. Except for the initial methionine residues, this sequence is identical to the published primary sequence of lebein (Swiss-Prot accession number P83253) (22). Furthermore, mass spectrometry determined its mass to be 14,083 Da and proved its identity to lebein. Upon SDS-PAGE, the heterodimeric lebein-1 showed an apparent molecular mass of 18 kDa. After reduction, it was separated by SDS-PAGE into two highly homologous subunits, ␣ and ␤, with apparent molecular masses of ϳ14 kDa (Fig. 5A). The difference in molecular mass determined by SDS-PAGE and mass spectrometry is likely due to the fact that the molecular masses of proteins below 20 kDa show a nonlinear behavior on polyacrylamide gels. Furthermore, the nonlinearity of the apparent molecular masses of the heterodimeric protein and its subunits upon SDS-PAGE can be explained by the fact that the Stoke radius of the two closely associated subunits within the nonreduced heterodimeric disintegrin is likely to be smaller than the sum of the Stoke radii of the individual unfolded subunits after reductive cleavage of disulfide bridges.
The inhibitory activity of fraction MS-I (Fig. 4A) was identified as a novel protein from V. lebetina venom. Despite its high sequence identity, its N-terminal sequence (MNSANPXXDDI), especially residues alanine and isoleucine in positions 4 and 11, respectively, clearly discriminated this inhibitor from lebein-1. Furthermore, its molecular mass of 14,735 Da, as determined by mass spectrometry, differed from that of lebein-1. The different molecular mass of lebein-2 was also observed upon SDS-PAGE (Fig. 5A). Upon reduction, the novel V. lebetina inhibitor with an apparent molecular mass of 20 kDa dissociated into two subunit chains (␣ and ␤) with apparent molecular masses of 15 and 7 kDa, respectively (Fig. 5A). Edman degradation revealed that its ␣ subunit was N-terminally blocked, whereas the N-terminal amino acid sequence of the ␤ subunit was proven to be identical to the N terminus of the nonreduced protein. This amino acid sequence has not been published yet. Because of its sequence homology to lebein and similar binding affinities, we propose the name lebein-2 for it. Thus, lebein-2 is a 14,735-Da heterodimeric protein consisting of two subunits, ␣ and ␤, the latter one of which has the N-terminal amino acid sequence MNSANPXXDDI.
Lebein-1 and Lebein-2 are the ␣ 7 ␤ 1 Integrin-binding Components of V. lebetina Snake Venom-A far-Western blot was established to test the fractions of V. lebetina snake venom for ␣ 7 ␤ 1 integrin-binding components (Fig. 5B). After electrophoretic separation, snake venom components were transferred onto a nitrocellulose membrane. Soluble ␣ 7 ␤ 1 integrin was allowed to bind to the blotted snake venom proteins and detected immunologically. The strongest signal was observed with purified lebein-2 (Fig. 5B). Remarkably, an intense signal was detected not only with the lebein-2 monomer at 20 kDa, but also with a band at ϳ30 kDa, which may be the aggregate of two lebein-2 molecules. In contrast, this band with an apparent molecular mass of 30 kDa was hardly visible on the Coomassie-stained SDS-polyacrylamide (Fig. 5A). The additional band at ϳ43 kDa, which was seen only on the far-Western blot, could not be identified yet, but might be an even higher aggregate of lebein-2. The far-Western blot (Fig. 5B) provided the first evidence that lebein-2 binds ␣ 7 ␤ 1 integrin directly. This interaction seemed to be specific and depended on the disulfide bridge-stabilized quaternary and/or tertiary structure of lebein-2 because reduction of lebein-2 entirely abolished the binding signal on the far-Western blot (data not shown).
Lebein-1 also bound ␣ 7 ␤ 1 integrin in the far-Western assay only under nonreducing conditions (Fig. 5B), albeit with a weaker binding signal than lebein-2. However, far-Western blot assays can be assessed only qualitatively, as protein interaction may be weakened because of only partial renaturation of the blotted proteins after SDS treatment. The Mono S fraction MS-II, which contains lebein-1 (Fig. 5B, MS-II lane), included an additional band at an apparent molecular mass of 44 kDa,  Viperidae, and Crotalidae snake venoms Murine laminin-1 was coated at 10 g/ml onto microtiter plates. Soluble ␣ 7 ␤ 1 integrin was added at 3.9 g/ml in the presence of 2 mg/ml snake venom and protease inhibitors. Bound integrin was quantified as described under "Materials and Methods." Means Ϯ S.D. of duplicate determinations are shown. Reduction of integrin binding to Ͻ50% indicated a strong inhibition by the venom.

% of non-inhibited control
Titration of immobilized lebein-2 and lebein-1 with soluble ␣ 7 ␤ 1 and ␣ 3 ␤ 1 integrins in the presence of divalent cations yielded titration curves that showed saturation (data not shown). Furthermore, linearization of these titration curves according to Heyn and Weischet (26) provided apparent affinity constants in the nanomolar range (Table II). Identical to laminin-1, both lebein-2 and lebein-1 bound avidly to soluble ␣ 7 ␤ 1 integrin in the presence of 1 mM Mn 2ϩ ions without any marked alteration of affinities after addition of the activating antibody 9EG7. Ca 2ϩ ions increased the apparent dissociation constants significantly. In addition, the presence of Ca 2ϩ showed more clearly that lebein-1 bound more avidly to ␣ 7 ␤ 1 integrin compared with lebein-2. Although the affinities of ␣ 3 ␤ 1 integrin for both disintegrins are lower, the tendency of Mn 2ϩ and Ca 2ϩ ions to increase and to decrease, respectively, the affinity of soluble ␣ 3 ␤ 1 integrin for lebein-2 and lebein-1 resembled the effects on soluble ␣ 7 ␤ 1 integrin. In contrast to soluble ␣ 7 ␤ 1 integrin, binding of soluble ␣ 3 ␤ 1 integrin to both disintegrins could be substantially improved by the activating antibody 9EG7 (Table II).
Inhibition of Binding of ␣ 7 ␤ 1 and ␣ 3 ␤ 1 Integrins to Their Cognate Laminin Isoform Ligands-Both lebein-2 and lebein-1 bound to the laminin-binding ␤ 1 integrins with the same divalent cation dependences as to the laminin ligands ( Fig. 1C and Table II). Furthermore, both disintegrins completely and efficiently blocked binding of ␣ 7 ␤ 1 and ␣ 3 ␤ 1 integrins to their respective laminin isoform ligands in a dose-dependent manner (Fig. 7). Because of its higher affinity for both integrins, lebein-1 showed half-maximal inhibition at up to 3-fold lower concentrations than lebein-2 ( Fig. 7 and Table III).
RGD Peptides Interfere Only Weakly with ␣ 3 ␤ 1 and ␣ 7 ␤ 1 Integrin Binding to Lebein-1 and Lebein-2-Because at least FIG. 3. Purification of the ␣ 7 ␤ 1 integrin-inhibiting factor from V. lebetina venom by gel filtration (A) and cation-exchange chromatography on a Mono S cation-exchange column (B). The absorbance of the eluate was measured at 280 nm (solid lines). Eluate fractions were tested for their capacity to inhibit ␣ 7 ␤ 1 integrin binding to immobilized laminin-1 (E). The ␣ 7 ␤ 1 integrinblocking inhibitors were eluted from the gel filtration column as a single peak, marked by the gray bar in A. The two inhibitors were separated by Mono S ionexchange chromatography into two distinct peaks, marked by the gray bars labeled MS-I and MS-II in B. Fraction MS-I contained lebein-2, whereas fraction MS-II was highly enriched in lebein-1.
lebein-1 is known to contain RGD sequences in both chains, we investigated whether the inhibitory activities of lebein-1 and lebein-2 on ␣ 7 ␤ 1 integrin depend on an RGD peptide sequence. To this end, binding of soluble ␣ 7 ␤ 1 integrin to immobilized laminin-1, lebein-1, and lebein-2 was challenged by increasing concentrations of RGD peptides or their derivatives (Fig. 8). The ␣ 7 ␤ 1 integrin/laminin-1 interaction was not affected by the GRGDS peptide even at a 1.6 million-fold molar excess (8 mM) to the integrin. In contrast, the binding of ␣ 7 ␤ 1 integrin to both lebein-1 and lebein-2 decreased at GRGDS peptide concentra- FIG. 4. Purification of lebein-2 and lebein-1 from the two Mono S fractions (MS-I and MS-II) by reversedphase chromatography on a C 8 reversed-phase column. The absorbance of the eluate was monitored at 280 nm (solid lines). Eluate fractions were tested for inhibition of ␣ 7 ␤ 1 integrin binding to immobilized laminin-1 (E). A, the Mono S fraction MS-I contains one inhibitory activity, which is the novel disintegrin lebein-2. B, the Mono S fraction MS-II contains two peaks of inhibitory activities. At lower acetonitrile concentrations, lebein-1 was eluted, whereas lebein-2 required higher acetonitrile concentrations to be eluted. were separated under nonreducing (without ␤-mercaptoethanol (w/o ␤-ME)) and reducing (with ␤-mercaptoethanol) conditions. The gel was either stained with Coomassie (A) or blotted onto a nitrocellulose membrane to detect ␣ 7 ␤ 1 integrin-binding proteins on the far-Western blot (B). The molecular masses of standard proteins as well as the bands of lebein-1 and lebein-2 are indicated. The labels refer to both panels.
Lebein-1 and Lebein-2 Inhibit Laminin-binding ␤ 1 Integrins tions above 1 mM. However, this decline in binding was incomplete even at the very high peptide concentration of 8 mM. Furthermore, control peptides such as GRGES and the scrambled sequence GRDGS showed a similar, yet less pronounced decline in integrin binding to both disintegrins at high peptide concentrations above 1 mM. At RGD peptide concentrations below 1 mM, lebein-2 and lebein-1 bound to soluble ␣ 7 ␤ 1 integrin in an RGD-independent manner, similar to the natural integrin ligand. Hence, both lebein-2 and lebein-1 must mechanistically be considered RGD-independent disintegrins when interacting with laminin-binding ␤ 1 integrins.

DISCUSSION
Among the various laminin-binding integrins, ␣ 7 ␤ 1 integrin plays an important role in the anchorage of muscles to their extracellular matrix, especially at the myotendinous junctions and intercalating disks of skeletal and cardiac muscle (3,4). Snake venoms may contain inhibitors directed against ␣ 7 ␤ 1 integrin, thus causing muscle dysfunction and paralysis. However, other venom components such as phospholipase A 2 , which leads to necrosis of muscle and other tissue (13,14), may also exert myotoxic and cytotoxic effects. Their presence precluded screening of whole snake venoms in cell tests. Therefore, we have established a cell-free screening assay in which we used a recombinantly expressed soluble ␣ 7 ␤ 1 integrin heterodimer and its binding to laminin-1. Within the muscle, ␣ 7 ␤ 1 is highly concentrated at the myotendinous and neuromuscular junctions, which are essential for the transduction of mechanical forces and nerve signals, respectively (12). Because ␣ 7 X2␤ 1 integrin seems to be the only isoform involved in laminininduced acetylcholine receptor recruitment into neuromuscular junctions (12) and is the predominant isoform in adult skeletal muscle, the splice variant X2 was chosen for recombinant production. A high yield expression of soluble ␣ 7 X2␤ 1 integrin was achieved by coexpression of the murine ␣ 7 X2 integrin ectodomain with the human ␤ 1 integrin ectodomain in Drosophila Schneider's cells. Similar to the results of von der Mark et al. (28), we showed that soluble ␣ 7 X2␤ 1 integrin bound to laminin-1 and laminin-2/4, but failed to interact with laminin-5. Despite being the major laminin isoform in muscle tissue (11), laminin-2/4 was bound by ␣ 7 X2␤ 1 integrin with a lower affinity than laminin-1. Compared with other integrins (7,24), soluble ␣ 7 ␤ 1 integrin binds its ligands with very high affinity, emphasizing its important role in muscle tissue. We also demonstrated that ␣ 7 ␤ 1 integrin belongs to the group of ␤ 1 integrins that are recognized by invasin, the surface protein of Yersinia pseudotuberculosis, and thus can be utilized by the pathogen to invade ␣ 7 ␤ 1 integrin-bearing host cells (29).
In this study, we have searched for an inhibitor that interferes with integrin-mediated adhesion to laminin. With the help of soluble ␣ 7 X2␤ 1 integrin and its high affinity binding to laminin-1, we could screen various snake venoms for their inhibitory capabilities in a protein interaction assay. Several venomous snakes from the Elapidae, Viperidae, and Crotalidae families are considered to exert myotoxic effects. However, in most of them, proteolytic enzymes, phospholipase activities, and other factors may account for their destructive effects on muscle tissue (13,14), as only one of 33 tested venoms entirely inhibited interaction of ␣ 7 ␤ 1 integrin with its laminin ligands. In addition to abundant phospholipase activity and numerous proteases (14,17), the venom of V. lebetina contains two disintegrins (lebein-2 and lebein-1) that inhibit ␣ 7 ␤ 1 integrin binding to its laminin ligands. In this study, we have established and optimized their purification from the crude toxin and have characterized their integrin-inhibiting function.
Lebein-1 and its primary structure have been published recently, as lebein (22). Because both subunits bear RGD sequences, lebein-1 was considered an RGD-dependent disintegrin. However, our study provides experimental evidence not only that lebein-1 interferes with the RGD-dependent cell attachment to fibronectin, but also that it additionally binds to laminin-binding ␤ 1 integrins in an RGD-independent manner. Therefore, it efficiently inhibits the interaction of lamininbinding integrins with their respective laminin isoforms.
During the purification of lebein-1, we also found a 44-kDa protein that showed proteolytic activity. This protein might be a precursor of lebetase, a Zn 2ϩ ion-containing metalloprotease that has been identified at the cDNA level (27). This precursor also contains a disintegrin domain (27), in good agreement with our observation that the 44-kDa protein bound ␣ 7 ␤ 1 integrin on the far-Western blot.
We isolated lebein-2 as a novel protein of the V. lebetina venom. It has a molecular mass of 14,735 Da, as determined by mass spectrometry, and consists of two subunits, ␣ and ␤. Whereas the ␣ subunit of lebein-2 was N-terminally blocked, its ␤ subunit shows an N-terminal amino acid sequence similar to, yet distinct from, those of the lebein-1 subunits, thus proving that lebein-2 is an independent gene product. Lebein-2 and lebein-1 strongly bound to the laminin-binding ␤ 1 integrins (␣ 7 ␤ 1 , ␣ 6 ␤ 1 , and ␣ 3 ␤ 1 ), but did not recognize the collagenbinding integrins (␣ 1 ␤ 1 and ␣ 2 ␤ 1 ) in a divalent cation-dependent manner. Although we have tested only the X1 splice variant of the ␣ 6 integrin subunit, it can be assumed that ␣ 6 ␤ 1 integrin is, in general, a target of lebein-2 and lebein-1, as all its splice variants have similar ligand-binding specificities (30). We also limited our experiments to the X2 splice variant of ␣ 7 integrin. However, this is the predominant ␣ 7 ␤ 1 integrin isoform in skeletal muscle and thus the potential target of the snake venom. Furthermore, because of their broad binding specificity for all laminin-binding ␤ 1 integrins, we assume that both disintegrins are also functional against the X1 splice variant of ␣ 7 ␤ 1 integrin, which is mainly expressed during myogenesis.
Having identified lebein-2 and lebein-1 in a protein interaction assay, we also proved their inhibitory activities on myoblasts in cell attachment studies. Interestingly, both lebein-2 and lebein-1 inhibited integrin-mediated cell adhesion not only to laminin, but also to fibronectin. The inhibition mechanism for fibronectin is probably caused by RGD sequences, although so far, only lebein-1 is known to be an RGD-dependent disintegrin. Our in vitro binding data support the conclusion that inhibition of cell attachment to laminin-1 is caused by a direct inhibitory interaction of lebein-2 or lebein-1 with laminin-binding ␤ 1 integrins on the cell surface that is independent of an RGD sequence.
Laminin-binding ␤ 1 integrins avidly bound to both lebein-2 and lebein-1, with K d values similar to those of the natural laminin ligands (Fig. 1C) (7). Like the binding to laminin, integrin binding affinities for the two V. lebetina disintegrins increased in the presence of Mn 2ϩ ions and decreased in the presence of Ca 2ϩ ions. Furthermore, EDTA completely abolished integrin binding to both laminin and disintegrins. Additionally, the fact that binding of laminin and that of lebein-1 or lebein-2 to integrins are mutually exclusive suggests that both disintegrins act as laminin mimetics that bind to or close to the ligand-binding site of laminin-binding integrins. The interaction of ␣ 7 ␤ 1 integrin with lebein-1 and lebein-2 did not depend on an RGD sequence, although at least lebein-1 bears two RGD sequences, nor did soluble ␣ 7 ␤ 1 integrin interact with laminin-1 in an RGD-dependent manner in our study, in agreement with earlier findings (31). However, binding of laminin-1 to its integrin receptors not only requires a yet undefined three-dimensional recognition site within the laminin G module, but also depends on the presence of the coiled-coil domain consisting of all three laminin chains, ␣1, ␤1, and ␥1 (32-35). Further mechanistic and structural studies will answer the question of how the small size disintegrins lebein-2 and lebein-1 can mimic the large laminin molecules and how they can achieve this inhibition.
Both lebein-1 and lebein-2 belong to the rare group of disintegrins that interact with laminin-binding ␤ 1 integrins in an RGD-independent manner. To our knowledge, fertilin (ADAM-2) and meltrin-␥ (ADAM-9) are the only examples of disintegrins that interact with the laminin-binding ␣ 6 ␤ 1 integrin (20,21,36). Like lebein-1 and lebein-2, ADAM-2 is functional only as a heterodimeric molecule together with ADAM-1, thus taking an essential part in sperm-oocyte fusion during fertilization (20). Lebein-1 shows some sequence similarities to the disintegrin domains of these ADAM proteins. However, neither ADAM-2 nor ADAM-9 contains an RGD peptide sequence (21,36). In contrast to ADAM-2 and ADAM-9, which are anchored in the plasmalemma, snake venom disintegrins are highly soluble and are able to inhibit cell/matrix interaction, thus leading to tissue dissipation.
In this study, we have identified and characterized two disintegrins from V. lebetina venom (lebein-1 and the novel lebein-2) that interact with laminin-binding ␤ 1 integrins in a divalent cation-dependent and RGD-independent manner both in vitro and in vivo. Through their potential to inhibit integrinmediated cell attachment to laminins, both lebein-1 and lebein-2 may be valuable tools to influence laminin-dependent cell functions. Among these are the attachment, mechanical force transduction, and migration of cells such as myoblasts (5,12) and ␣ 7 ␤ 1 integrin-bearing melanoma cells (31) along or through the laminin-rich basal membranes. These cell functions are of paramount importance in complex physiological processes such as wound healing and tumor invasion.