e at SciVerse ScienceDirect

Toxicon 64 (2013) 96–105

Contents lists availabl

Toxicon

journal homepage: www.elsevier .com/locate/ toxicon

Identification of a2b1 integrin inhibitor VP-i withanti-platelet properties in the venom of Vipera palaestinae

Franziska T. Arlinghaus a, Tatjana Momic b, Narmeen Abu Ammar b,c, Ela Shai c, Galia Spectre c,David Varon c, Cezary Marcinkiewicz d, Heinrich Heide e, Philip Lazarovici b,Johannes A. Eble a,*

aCenter for Molecular Medicine, Department of Vascular Matrix Biology, Frankfurt University Hospital, Excellence Cluster Cardio-Pulmonary System,Theodor-Stern-Kai 7, 60590 Frankfurt, Germanyb School of Pharmacy, Institute for Drug Research, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem, IsraelcDepartment of Hematology, Coagulation Unit, Hadassah-Hebrew University Medical Center, Jerusalem, IsraeldDepartment of Biology, Temple University College of Science and Technology, Philadelphia, PA, USAeMolecular Bioenergetics, Frankfurt University Hospital, Excellence Cluster Macromolecular Complexes, Theodor-Stern-Kai 7, 60590 Frankfurt, Germany

a r t i c l e i n f o

Article history:Received 14 September 2012Received in revised form 14 December 2012Accepted 4 January 2013Available online 11 January 2013

Keywords:VP12Integrin a2b1A domainC-type lectin-related protein

* Corresponding author. Tel.: þ49 69 6301 87651;E-mail addresses: arlinghaus@med.uni-frankfurt

eble@med.uni-frankfurt.de (J.A. Eble).

0041-0101/$ – see front matter � 2013 Elsevier Ltdhttp://dx.doi.org/10.1016/j.toxicon.2013.01.001

a b s t r a c t

Integrins are receptors of the extracellular matrix (ECM), playing a vital role in patho-physiological processes. They bind to ECM ligands like collagens and can mediate woundhealing as well as tumor metastasis and thrombosis, thus being a part of cell adhesion andmigration as well as platelet aggregation. For this reason, identifying a2b1 integrin-specificantagonists can assist in the development of drugs to treat tumor progression, angio-genesis, and cardiovascular diseases. Snake venoms have been shown to contain antago-nists which target collagen-binding integrins. EMS16, rhodocetin, and VP12 are threetoxins belonging to the C-type lectin-related protein family and have been proven toinhibit the a2b1 integrin, specifically the a2 integrin A domain.To specifically isolate antagonists targeting the a2b1 integrin A domain, we developeda protocol based on affinity chromatography. Using this novel approach, the toxin VP-i wasisolated from Vipera palaestinae venom. We show that VP-i binds to the a2 integrin Adomain and that it successfully inhibits adhesion of various cells to type I collagen as wellas cell migration. Moreover, our results indicate that VP-i differs structurally from thepreviously purified VP12, although not functionally, and therefore is a further venomcompound which can be utilized for drug development.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Snake venoms contain many different proteins whichcause a variety of effects following envenomation, some ofthem contradictory, e.g. inducing aggregation/agglutina-tion or inhibiting it (Arlinghaus and Eble, 2012; Morita,

fax: þ49 6301 87656..de (F.T. Arlinghaus),

. All rights reserved.

2005; Ogawa et al., 2005). C-type lectin-related proteins(CLRPs) constitute a large fraction of snake venoms andare known to function inter alia as integrin inhibitors(Clemetson, 2010; Momic et al., 2011).

Integrins are responsible for cell functions such asadhesion and migration by binding to proteins of theextracellular matrix (ECM) and linking these to the cytos-keleton (Barczyk et al., 2010). They are heterodimeric,transmembrane proteins composed of non-covalentlyassociated a and b subunits (Hynes, 1992). 18 a and 8b subunits in different combinations yield 24 known

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105 97

integrins with distinctive binding specificities (Takadaet al., 2007). Of the four integrins recognizing collagen,one is a1b1, recognizing type IV collagen, and another a2b1,which recognizes type I collagen (Eble, 2005). The a2b1integrin receptor is located on endothelial and epithelialcells as well as on platelets (Heino, 2000; Nuyttens et al.,2011; Zutter et al., 1998), where it is the sole collagen-binding integrin. a2b1 integrin is essential in physiologi-cal and pathological processes such as wound healing,tumor metastasis, and thrombosis, all of which depend onthe binding of the integrin to collagen (Broos et al., 2012;Eckes et al., 2006; Shaw,1999). Naturally occurring collagenreceptor antagonists can be employed for drug design totreat tumor progression and angiogenesis as well as car-diovascular diseases.

It is recognized that not all CLRPs target integrins(Clemetson and Clemetson, 2008), but to our knowledge,three snake venom CLRPs were shown to block the a2b1integrin: VP12 (Staniszewska et al., 2009), rhodocetin (Ebleet al., 2002), and EMS16 (Marcinkiewicz et al., 2000). All ofthese are specific for the a2 integrin A domain (a2A), whichis the major collagen-binding domain and homologous tothe von Willebrand Factor (vWF) A domain (Emsley et al.,2000).

VP12 was quite recently isolated from Vipera palaestinaesnake venom (Staniszewska et al., 2009). It is a heterodimerof a and b subunits which showmolecular masses of 16 and15.9 kDa, respectively, and was shown to inhibit a2b1integrin-dependent cell adhesion to type I collagen. Otherviperidae CLRPs are known to target integrins a4 and a5,such as lebecetin (Pilorget et al., 2007).

We isolated a second integrin-binding protein from V.palaestinae, VP-i, based on affinity chromatography withthe a2A domain. Our study revealed that this integrin-binding protein effectively inhibited cell adhesion on col-lagen as well as collagen-induced platelet aggregation.Furthermore, the high functional similarity, but structuraldifference, between this protein and VP12 underlines theversatility of CLRPs. Therefore, employing affinity chro-matography to identify a2b1 integrin-targeting snakevenom proteins can be of great use to identify new proteinsfor drug design.

2. Materials and methods

2.1. Materials

Snake venom C-type lectin-related protein (CLRP) VP12was purified from V. palaestinae snake venom which wascollected from live specimens kept at SIS Pharmaceuticals(Rehovot, Israel). The snakes were manually milked overone year, and the venom from several snakes was pooled.The venom was prepared under good laboratory practice(GLP) conditions, according to the requirements of theIsraeli Ministry of Health, and kindly provided to us byDr. Naftali Primor. Collagen type IV (from bovine placentavilli) was purchased from Chemicon (Temecula, CA), andcollagen type I (from rat tail) from BD Biosciences (Bedford,MA). V. palaestinae snake anti-venom was provided by theIsraeli Ministry of Health.

2.2. Cell lines

K562 cells transfected with a1 and a2 integrin subunitswere provided by Dr. M. Hemler (Dana Farber CancerInstitute; Boston, MA) and cultured in RPMI 1640 supple-mented with 10% fetal bovine serum (FBS) and 0.5 mg/ml ofG418. Human melanoma MV3 cell line was provided byDr. E. Danen (Leiden University; Leiden, The Netherlands),human melanoma HS-939T and mouse melanoma B16F10cell lines were purchased from ATCC (Manassa, VA). Mel-anoma cell lines as well as HT1080 fibrosarcoma cells werecultured in Dulbecco’s Modification of Eagle’s Medium(DMEM) containing 10% FBS.

2.3. Isolation of the a2b1 integrin-specific venom protein VP-iby affinity chromatography

V. palaestinae venom was dissolved in phosphate-buffered saline (PBS, 20 mM sodium phosphate/150 mMNaCl, pH 7.4) at a concentration of 100 mg/ml. Using therecombinantly expressed oligo-His-tagged integrin a2Adomain (Eble et al., 2009), proteins binding to the a2b1integrin were identified by affinity chromatography usinga2A immobilized onto a 1 ml HisTrap HP column (GEHealthcare, Freiburg, Germany) as resin. Eluted with animidazole gradient, integrin A domain-containing fractionswere identified and analyzed by SDS-PAGE in a 10–20%polyacrylamide gel.

2.4. Purification of VP-i by classical chromatographicprocedures

Solubilized V. palaestinae venom proteins (500 mg/ml)were separated using a Superdex 200 10/300 GL gel filtra-tion column (GE Healthcare) at 0.5ml/minwith PBS, pH 7.4.Fractions with a2b1 integrin-inhibiting activity werediluted in 20 mMMES, pH 6.5, loaded onto a MonoS HR5/5column (GE Healthcare) and eluted with a linear gradientof 0–50% 20 mM MES/1 M NaCl, pH 6.5, at a flow rateof 0.5 ml/min. BCA assay (Thermo Scientific, Dreieich,Germany) and SDS-PAGE were used to determine proteinconcentration and purity, respectively. The resulting pro-tein bands were isolated, trypsin-digested, and analyzed bymass spectrometry using electrospray ionization-Trap-MS.

2.5. Diagonal two-dimensional (2D) gel analysis

The complexity of the purified venom proteins wasdetermined by 2D electrophoresis analysis in a 10–20%polyacrylamide gel. Proteins were separated under non-reducing and reducing conditions in the first and seconddimension, respectively, before being detected by Coo-massie staining.

2.6. Inhibition of GST-a2A binding to type I collagen by VP-i

Inhibition ELISA was performed as published previously(Eble andTuckwell, 2003),with the followingmodifications.Type I collagen was immobilized onto a microtiter plateovernight at 10 mg/ml in 0.1M acetic acid. After blocking theplate with 1% BSA in TBS/MgCl2 (50 mM Tris–HCl/150 mM

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–10598

NaCl/2 mM MgCl2, pH 7.4), the glutathione-S-transferase(GST)-tagged a2A domain was permitted to bind to type Icollagen in the presence of different fractions of the proteinpurification for 2 h at room temperature. Bound GST-a2Awas fixed for 10 min with 2.5% glutaraldehyde in HEPESbuffer (50 mM HEPES/150 mM NaCl/2 mM MgCl2/1 mMMnCl2, pH 7.4). The GST-a2Awas detectedwith a polyclonalrabbit antibody against GST (Molecular Probes, Nijmegen,The Netherlands), followed by an alkaline-phosphatase-conjugated anti-rabbit antibody (Sigma, Deisenhofen, Ger-many) as primary and secondary antibody, respectively,each diluted in 1% BSA/TBS/MgCl2. The ELISAwas developedwith p-nitrophenyl phosphate (Sigma), and absorbancemeasured at 405 nm (BioTek, Bad Friedrichshall, Germany).Nonspecific binding was assessed by binding of GST-a2Ato BSA.

2.7. Titration of VP-i and VP12 with GST-a2A

VP-i or VP12 was immobilized onto a microtiter plateovernight (4 �C) at 10 mg/ml in TBS/MgCl2. The plate wasfirst blocked with 1% BSA/TBS/MgCl2, after which theimmobilized protein was incubated with different con-centrations of GST-a2A for 2 h at room temperature. Thebound GST-a2A was fixed with 2.5% glutaraldehyde inHEPES buffer and detected as described in the inhibitionassay (2.6).

2.8. Cell adhesion assay

An adhesion assay was performed using the xCELLi-gence system (Roche Diagnostics, Penzberg, Germany). AnE-plate was coated with type I collagen at 5 mg/ml in 0.1 Macetic acid at 4 �C overnight. HT1080 cells were seeded at20,000 cells per well in the absence or presence of differentconcentrations of VP-i in DMEM (Dulbecco’s Modified Ea-gle’s Medium), or with 10 mM EDTA. Values were meas-ured every 2 min for 2 h, then every 5 min for another 2 h.An alternate adhesion assay was performed as described byEble et al. (2002), with slight modifications. Briefly, type Icollagen (0.2 mg/ml in 0.1 M acetic acid) was immobilizedonto a microtiter plate overnight at 4 �C. After blocking,HT1080 fibrosarcoma cells (20,000 cells/well) were seededonto the plate in both absence and presence of differentconcentrations of VP-i, or with 10 mM EDTA, a control fornonspecific cell adhesion. After 30 min incubation at 37 �C,adherent cells were fixed and stained with crystal violet.The dye was extracted with 70% ethanol for 30 min, andabsorbance was read at 560 nm. Cell adhesion signals(means � S.D.) are corrected for signals measured in thepresence of EDTA.

Further adhesion assays were performed as described inBazan-Socha et al. (2004), with minor changes. A 96-wellplate (Nunc, Roskilde, Denmark) was coated with 10 mg/ml type I collagen or 1 mg/ml type IV collagen in 20 mMacetic acid overnight at 4 �C. After blocking with 1% BSA inHank’s Balanced Salt Solution (HBSS) containing 5 mMMgCl2 for 1 h at room temperature, CMFDA-labeled cells(105 cells/well) were seeded onto the plate in both theabsence and presence of either VP-i or VP12 (30 min pre-incubation at 37 �C). After 60 min incubation at 37 �C,

unbound cells were removed by washing with 1% BSA/HBSS. The remaining, bound cells were lysed by the addi-tion of 0.5% triton X-100 and fluorescence quantified witha SPECTRAFluor Plus plate reader (Tecan Group Ltf., Swit-zerland) at lex ¼ 485 nm and lem ¼ 530 nm. To determinethe number of adhered cells, a standard curve was gen-erated by serial dilutions of CMFDA-labeled cells of knownnumbers.

2.9. Neutralization of VP-i-induced cell adhesion withanti-sera

Anti-serum against the isolated integrin-inhibitoryprotein was obtained by immunizing a Lewis rat with2 mg/g body weight of VP-i in combination with completeFreund’s adjuvant (Sigma; Steinheim, Germany). Subse-quent immunizations were performed with venom proteintogetherwith incomplete Freund’s adjuvant (Sigma). Serumcontaining polyclonal antibodies against the venomproteinwas collected after a final boost without adjuvant.

The adhesion assay was performed as described previ-ously with a2-transfected K562 cells. A constant concen-tration of the integrin-inhibitory protein (0.05 mg/ml) waspre-incubated for 30 min with different dilutions of ratserum against the purified venom protein and horse serumagainst the whole venom.

2.10. Cell migration assay

The xCELLigence system (Roche Diagnostics) wasemployed to analyze cell migration of HT1080 cells ona CIM-plate in real-time. A modified protocol as describedin de Rezende et al. (2012) was employed. CIM-plates werecoated with type I collagen at 10 mg/ml in 0.1 M acetic acidat 4 �C overnight. Different concentrations of VP-i wereadded to the lower and upper chambers, and backgroundvalues were assessed. Subsequently, suspensions of105 cells were added to each well. Impedance values weremeasured every 5 min for 6 h and corrected for the back-ground values. Data was evaluated with the RTCA software1.2 (Roche Diagnostics).

2.11. Platelet aggregation assay

Blood from healthy donors, who had not been medi-cated for at least 15 days, was mixed with 3.8% trisodiumcitrate (1:9 citrate/blood) and centrifuged (90 g for 15 min)to obtain platelet-rich plasma (PRP). Platelet-poor plasma(PPP) was obtained by centrifuging the remaining bloodagain (500 g, 15 min). The platelet number was adjusted to2.5*108 platelets/ml by diluting the PRP with PPP and usingit within 2 h. All the above preparations were carried outusing plasticware or siliconized glassware. Using a dual-channel Platelet Aggregation Profiler� model PAP-8E S/WVersion 1.0.8 aggregometer (Bio/Data Corp.), the turbidi-metric method of Born and Cross (Born and Cross, 1963)was followed. Briefly, 235 ml of PRP suspension was main-tained at 37 �C in a siliconized glass cuvette and pre-incubated with different concentrations of VP-i or aquadest. (control) in 10 ml PBS for 3 min. Aggregation wasinitiated by adding 2 mg/ml type I collagen and followed

Fig. 1. Identification of VP-i. (A) The V. palaestinae venom was resolved intofour peaks (continuous line) by an imidazole gradient (dotted line); non-specifically bound proteins were eluted as a large peak (AI) with twoshoulders (AII and AIII); a smaller peak (AIV) was eluted, containing anintegrin–venom protein complex. Fractions were analyzed under reducing(B) conditions by 10–20% SDS PAGE and Coomassie staining. Under reducingconditions, peak AIV contains a2A (26 kDa) and three prominent bands at13, 17 and 20 kDa as well as a protein band at 32 kDa.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105 99

with constant stirring at 900 rpm for 6 min. In each case,aggregation induced by type I collagen alone was consid-ered as 100% aggregation. All experiments were performedin quadruplicate (four donors). The IC50 was calculatedfrom the dose-dependent curve.

3. Results

3.1. Identification of an a2b1 integrin-binding protein from V.palaestinae

The a2b1 integrin is a frequent target for venom com-ponents which block collagen-induced platelet activationand cell attachment to type I collagen (Eble et al., 2002;Marcinkiewicz et al., 2000; Staniszewska et al., 2009).However, it is an arduous process to identify individualproteins which target specific receptors by first separatingthem by classical chromatographic techniques and subse-quently screening each fraction for activity. We havedeveloped an alternative, decidedly faster identificationmethod by implementing affinity chromatography usingthe immobilized A domain of the a2b1 integrin predomi-nantly responsible for collagen-binding (Emsleyet al., 2000)as bait (Arlinghaus and Eble, 2013). An oligo-His-taggedintegrin A domain was immobilized on a Ni-NTA column,and the crude V. palaestinae venom allowed to bind to theintegrin. The proteinswere resolved into four peaks (Fig.1A,continuous line), peaks AI to AIII being nonspecificallybound proteins. The 26 kDa a2A domainwas eluted in peakAIV, and venom proteins with molecular weights of 17, 20and 32 kDa were eluted together with it (Fig. 1B).

3.2. Isolation of the integrin-binding protein independently ofa2A domain

To characterize the binding and inhibitory capacities ofthe venom component, it must be free of its binding part-ner a2A. Hence, an a2A-free purification protocol had to beestablished. Classical chromatographic techniques wereemployed to purify the venom proteins in the absence ofthe a2A domain. A two-step purification procedure wasemployed with gel filtration and subsequent ion exchangechromatography.

The crude venom of V. palaestinae was separated by gelfiltration on a Superdex 200 10/300 GL column into threemain peaks (Fig. 2A, continuous line). Analysis by SDS-PAGE (Fig. 2C) reveals the extent of purity achieved inthis process. Peak GI shows a band pattern analogous to theproteins isolated in affinity chromatography (Fig. 1B, laneAIV) as well as other proteins with higher molecularweight. Peaks GII and GIII did not reveal a band patterncorresponding to the affinity chromatography, nor did theyshow any potential to block a2A binding to type I collagen(Fig. 2A, dotted line, right axis). On the other hand, eluatefractions from peak GI reduced binding of GST-tagged a2Ato collagen (Fig. 2A, dotted line). Consequently, GI containsproteins with the potential to inhibit a2A-binding to type Icollagen.

Subsequently, the eluate fractions of peak GI, whichshowed a2A integrin-inhibiting activity, were pooled andseparated by ion exchange chromatography into five peaks

(Fig. 2B, continuous line) with a sodium chloride gradient(dashed line). Fractions were analyzed by SDS-PAGE andby an inhibition assay as before. Gel analysis underreducing conditions illustrated similar protein patterningof all peaks (Fig. 2D); however, only peak MIII containedproteins with apparent molecular weights of 17 and20 kDa as well as an additional low-molecular-weightband of 13 kDa. Also, only this peak exhibited significantinhibitory activity of GST-a2A-binding to collagen (Fig. 2B,dotted line). It is important to note that the 32 kDa proteinpresent in the a2A-binding fraction of the affinity chro-matography (Fig. 1B) is not present in the classical puri-fication procedure without the A domain. In its place,a 52 kDa protein is co-eluted with the proteins of interestin MIII which, however, only shows up in the fractions ofnonspecifically bound proteins in affinity chromatography(Fig. 1B, lanes AI to AIII).

To determine the complexity of the integrin-inhibitorytoxin, diagonal two-dimensional gel analysis was per-formed. Under non-reducing conditions in the first dimen-sion (Fig. 3, horizontal lane), proteins were separated intofour bands with apparent molecular weights of 52, 90, 110,and 130 kDa, two of them being more prominent than theothers (90 and 110 kDa). In the second dimension, the

Fig. 3. Diagonal two-dimensional gel analysis of purified V. palaestinaevenom protein. In the first dimension, the venom protein was separatedunder non-reducing conditions (Vp/, horizontal lane). Second dimensionanalysis (vertical) occurred after reduction and visualizing the proteins byCoomassie-staining; the reduced, purified protein serves as a control (VpY).

Fig. 2. Gel filtration and ion exchange chromatography of VP-i. (A) V.palaestinae gel filtration on a superdex 200 10/300 GL (GE Healthcare)resolved the venom into three peaks (continuous line, GI – GIII). The per-centage of a2b1-inhibition after addition of the respective eluate fractions isindicated by the dotted line (right side axis). (B) Proteins from peak GI fromthe gel filtration were separated into five peaks (continuous line, MI – MV)along a sodium chloride gradient (dashed line) by ion exchange chroma-tography. The percentage of a2b1-inhibition after addition of the respectiveeluate fractions is indicated by the dotted line (right side axis). Venomproteins separated by gel filtration with superdex column (C) and subse-quent Mono S chromatography (D) were analyzed by reducing SDS-PAGE in10–20% acrylamide gels and Coomassie-staining.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105100

130 kDa protein has an apparent molecular weight ofw65 kDa after reduction; the control under non-reducingconditions does not contain a protein of this size. The130 kDa protein may nevertheless be a dimer of the 52 kDaprotein as strong disulfide crosslinking can result in theproteins running above the diagonal in a 2D gel. Massspectrometry analysis revealed these bands to be a zinc-metalloproteinase which is inactive in a zymography(data not shown). The more prominent bands were eachseparated into two low-molecular-weight bands. The110 kDa protein band consisted of 17 and 20 kDa proteins,while the 90 kDa protein band consisted of 13 and 19 kDaproteins. Mass spectrometry results for these protein bandsindicate that the 17, 19, and 20 kDa proteins are all highlyhomologous to the C-type lectin A14 (Macrovipera lebetina)(Fig. 4A, dark gray shading), which in turn is homologous tothe VP12 b chain (Fig. 4A, light gray shading); the differ-ences in molecular mass may be due to posttranslationalmodifications. The 13 kDa protein is homologous to C-typelectin-like 5 (Daboia russelii siamesis) (Fig. 4B, dark grayshading), a protein homologous to VP12 a chain (Fig. 4B,light gray shading). The results suggest the integrin-inhibitory protein, coined VP-i, to be a complex of severalproteins, which are partially connected by disulfide links,and some of which possibly being the same protein onlybearing posttranslational modifications.

3.3. VP-i binds to the a2A domain

The protein isolated from the V. palaestinae venomwas identified by its a2A-binding capability via affinity

Fig. 4. Sequence alignment. Protein samples were trypsin digested and analyzed by mass spectrometry. (A) Spots 1–3 (1: 20 kDa; 2: 19 kDa; 3: 17 kDa) arealigned with C-type lectin A14, their identical amino acid residues highlighted dark gray. Homology to VP12 b subunit is highlighted in light gray. The invertedamino acid (black highlight) indicates a likely exchange of glycine (G) for aspartate (E), which increases sequence homology. (B) Spot 4 (13 kDa) is aligned withC-type lectin-like 5, their identical amino acid residues highlighted dark gray. Homology to VP12 a subunit is highlighted in light gray.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105 101

chromatography. To determine its affinity for the integrina2A domain, titration assays with GST-a2Awere performed(Fig. 5). The purified integrin-inhibitory protein, VP-i, wascompared to VP12, an integrin-binding protein previouslypurified from V. palaestinae snake venom (Staniszewskaet al., 2009). GST-a2A binds to both proteins in a satu-rable manner and yields dissociation constants (Kd) ofapproximately 5.45 nM and 4.59 nM for VP-i and VP12,respectively.

3.4. VP-i and VP12 inhibit cell adhesion mediated by a2b1integrin

To assess whether VP-i is effective at a cellular level,adhesion assays were performed using different cell linesas well as different techniques. Using real-time impedancemeasurement, adhesion of HT1080 cells to type I collagenin the presence of VP-i was analyzed. Cell adhesion wasinhibited completely and efficiently with an IC50 value of

Fig. 5. Comparison of molecular activity of two integrin-binding proteinsfrom V. palaestinae. The integrin-inhibitory proteins VP-i (A) and VP12 (B)were immobilized and titrated with the a2A domain; BSA served as back-ground control. The integrin bound to both venom proteins in a saturablemanner yielding a Kd of 5.45 and 4.59 nM for VP-i and VP12, respectively.

Fig. 6. Effect of the isolated venom component on cell adhesion. (A) VP-i ac-tivitywasassessed in real-timebyHT1080cell adhesion to type I collagenusingthe xCELLigence system (Roche Diagnostics). Values for adhesion is correctedto background values; absence of venom proteins served as positive control;presence of 10mMEDTA corresponds to no binding. VP-i inhibits cell adhesionto collagen in a concentration-dependent manner with an IC50 of 5.01 mg/ml.(B) Effect of venom proteins on adhesion of a2- and a1-transfected K562 cells:adhesion of a2-K562 cells in the presence of VP12 (open circles) or VP-i (solidcircles); adhesion of a1-K562 cells in the presence of VP-i and VP12 (open andsolid triangles, respectively). (C) Effect ofVP-i on theadhesionof threedifferentmelanoma cell lines: human HS939T, human MV3, and mouse B15. Untreatedmelanoma cells – control (filled bars), melanoma cells treated with VP-i(open bars); * indicates the p value with p < 0.05 compared to control group.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105102

5.01 mg/ml (Fig. 6A). A different approach in a microtiterplate with crystal violet staining showed that this protein isselective for the A domain of the a2b1 integrin, as celladhesion to immobilized fibronectin and laminin-111 wasnot influenced (data not shown).

Additionally, VP-i, purified by the novel proceduredescribed in this paper, was compared with VP12 usingK562 cells transfected with either a2 or a1 integrin sub-units. Both proteins effectively inhibited cell adhesion ofa2-K562 to type I collagen with IC50 values of 0.045 and0.10 mg/ml for VP-i and VP12, respectively (Fig. 6B, circles);there was no inhibition of a1-K562 cell adhesion to type IVcollagen, an interaction requiring a1b1 integrin, by either

Fig. 8. Analysis of cell migration in the presence of VP-i. Cell migrationanalysis was performed using real-time impedance measurement with thexCELLigence system. The venom protein VP-i inhibited cell migration withan IC50 ¼ 1.90 mg/ml.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105 103

VP-i or VP12 (Fig. 6B, triangles). Moreover, VP-i effectivelyinhibits cell adhesion of the human melanoma cell linesHS-939TandMV-3 as well as themousemelanoma cell lineB-16 to type I collagen (Fig. 6C). The same effect had pre-viously been shown for VP12 (Staniszewska et al., 2009).

It can clearly be stated that VP-i and VP12 both effec-tively inhibit adhesion of different cells to type I collagenindependently of the a2b1 expression level, but have noeffect on a1b1-mediated adhesion to type IV collagen. In anattempt to neutralize the inhibitory effect of VP-i on celladhesion, anti-sera against VP-i and against whole V.palaestinae venomwere added to the adhesion assays witha2-transfected K562 cells. As illustrated in Fig. 7, both anti-sera canceled the inhibitory effect of VP-i on K562 celladhesion.

3.5. VP-i inhibits cell migration mediated by a2b1 integrin

The effect of the integrin-inhibitory protein VP-i was notonly tested on cell adhesion, but also on cell migration ofHT1080 cells. This was measured in real-time by monitor-ing the increase of impedance values over time as HT1080cells migrated along a haptotactic gradient of type I colla-gen and covered the electrodes on the bottom face of thefilter. The impedance decreased with an IC50 value of1.90 mg/ml with increasing concentrations of VP-i (Fig. 8).

3.6. Platelet aggregation studies

The integrin a2b1 plays a role not only in cell adhesionand migration, but also in platelet aggregation, most likelythe preferential target of the venom. Platelets have thea2b1 integrin as their sole collagen-binding integrin,making the a2A domain an auspicious target for drugswhich can then influence platelet function. To test whether

Fig. 7. Neutralization of the inhibitory effect of VP-i with anti-sera. Celladhesion of a2-K562 cells to type I collagen was inhibited by 0.05 mg/ml VP-i. The control sample is displayed as a black bar. 0: no anti-serum (whitebar); other samples contain 0.05 mg/ml VP-i and either 1:100 or 1:1000dilutions of the anti-sera. Horse anti-serum (dark gray bars), rat serumpolyclonal antibodies against purified VP-i (gray bars). The data are pre-sented as the mean (�SD) percentage of adhered cells, of three experimentsperformed at sixplicate; p < 0.01 for no anti-serum and 1:100 dilution andp < 0.05 for 1:1000 dilution of anti-serum compared to VP12 alone.

VP-i has any affect on platelet function, aggregation assayswith platelet-rich plasma were performed. Fig. 9 showsthat VP-i blocks collagen-induced platelet aggregation by60% at a concentration of 3 mg/ml.

4. Discussion

We demonstrated that V. palaestinae venom contains anintegrin-binding protein which is specific for the a2A

Fig. 9. Effect of VP-i on collagen-induced platelet aggregation. Platelet ag-gregation in human PRP was induced by 2 mg/ml of type I collagen. Resultsare presented as a percent of control (aqua dest.). Inset: Tracings of inhib-itory effects of 1: control, 2: 2 mg/ml, 3: 2.5 mg/ml, 4: 3 mg/ml VP-i.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105104

domain and acts antagonistically on the adhesion of variouscells. Also, this venom protein inhibits cell migration andplatelet aggregation. Moreover, the protein VP-i has highfunctional similarity to VP12, an integrin-binding toxinpreviously purified from the same V. palaestinae. Althoughboth VP-i and VP12 belong to the family of C-type lectin-related proteins (CLRP), VP-i has a seemingly more com-plex quaternary structure. This may be caused by the gen-tler purification protocol for VP-i, using either affinitychromatography or classical gel filtration and ion exchangechromatography. In contrast, VP12 was purified with a C18reverse-phase HPLC column using a trifluoroacetic acid andacetonitrile-containing buffer (Staniszewska et al., 2009),which is likely to disrupt the quaternary, if not tertiary,structure of the protein.

Having achieved our goal of purifying an a2b1-specific toxin, we went on to characterize VP-i and com-pare it to VP12. VP12 had earlier been purified from V.palaestinae venom and had been shown to be an a2b1integrin-inhibitory protein consisting of two subunits(Staniszewska et al., 2009). Two-dimensional gel analysis ofVP-i revealed a more complex structure for this CLRP,consisting not only of two but of four low-molecular weightproteins. The heterogeneity of VP-i compared to VP12 maypossibly be caused by glycosylation of the proteins in theVP-i sample, or other posttranslational modifications. Inaddition, we cannot rule out polymorphic variants of VP-ias it was purified from V. palaestinae venoms which hadbeen pooled from different individuals of the species. It isalso known that venom proteins are often generated bygene duplication and mutations can lead to divergence ofthe protein function (Wong and Belov, 2012). These vari-ants, alongwith differences in glycosylation, might result ina shift during purification and/or in SDS-PAGE analysis.

Comparison of the binding properties of the V. palae-stinae toxins for the integrin a2A domain revealed that bothVP-i and VP12 bind to the A domain with the same affinity.Additionally, VP-i also binds to the wildtype integrin, as haspreviously been shown for VP12: adhesion and migrationof HT1080 cells as well as adhesion of different melanomacells to type I collagenwas inhibited by VP-i. The selectivityof the venom protein for a2b1 was also proven, as it had noeffect on cell adhesion to fibronectin or laminin-111.Additional experiments with K562 cells revealed thatboth VP-i and VP12 inhibited adhesion of a2-transfectedcells to type I collagen to a similar degree, but neitherinfluenced adhesion of a1-transfected cells to type IV col-lagen. This indicates that the inhibitory effect is indepen-dent of the cell type and expression level of the a2b1integrin. We also showed that VP-i reduces collagen-induced platelet aggregation by 60%. The effect of VP-i onplatelets, however, is not linear and possibly due to theplatelet receptor GPVI, which also binds collagen and maystill activate platelets under the conditions used here.

5. Conclusion

The venom protein VP-i, purified by the proceduredescribed here, and VP12 share a similar function andpossess homologous sequences. Yet they differ in theirprimary sequences indicating that they are distinct

proteins. Both proteins belong to the family of C-typelectin-related proteins. With VP-i, we add an additionala2b1 integrin-directed inhibitor to this protein family,underlining that nature has instrumentalized CLRPs tospecifically inhibit a2b1 integrin. The purification of thissecond integrin-inhibitory protein VP-i from V. palaestinaemay be due to the alternative purification technique whichis performed under less stringent conditions, resulting ina more complex protein. This reinforces the viability ofaffinity chromatography to identify receptor-specificvenom proteins as well as the use of milder chromato-graphic techniques for their purification. However, howhomologous VP-i and VP12 truly are remains to be con-firmed, and it is not aided by the fact that as yet little isknown about V. palaestinae venom proteins.

Acknowledgments

We kindly thank the German Israeli Foundation forfunding this project (GIF grant number: 994-3.9/2008).Furthermore, we thank Dr. J.J. Calvete from the Instituto deBiomedicina de Valencia for mass spectrometry analysis.

Conflict of interest statement

The authors declare that there are no conflicts ofinterest.

References

Arlinghaus, F.T., Eble, J.A., 2012. C-type lectin-like proteins from snakevenoms. Toxicon 60, 512–519.

Arlinghaus, F.T., Eble, J.A., 2013. The collagen-binding integrin alpha2beta1is a novel interaction partner of the T. flavoviridis venom proteinflavocetin-A. Journal of Biological Chemistry 288 (2), 947–955.

Barczyk, M., Carracedo, S., Gullberg, D., 2010. Integrins. Cell. Tissue Res.339, 269–280.

Bazan-Socha, S., Kisiel, D.G., Young, B., Theakston, R.D., Calvete, J.J.,Sheppard, D., Marcinkiewicz, C., 2004. Structural requirements ofMLD-containing disintegrins for functional interaction with alpha 4beta 1 and alpha 9 beta1 integrins. Biochemistry 43, 1639–1647.

Born, G.V., Cross, M.J., 1963. Effect of adenosine diphosphate on theconcentration of platelets in circulating blood. Nature 197, 974–976.

Broos, K., De Meyer, S.F., Feys, H.B., Vanhoorelbeke, K., Deckmyn, H., 2012.Blood platelet biochemistry. Thromb. Res. 129, 245–249.

Clemetson, K.J., Clemetson, J.M., 2008. Platelet GPIb complex as a target foranti-thrombotic drug development. Thromb. Haemost 99, 473–479.

Clemetson, K.J., 2010. Snaclecs (snake C-type lectins) that inhibit oractivate platelets by binding to receptors. Toxicon 56, 1236–1246.

de Rezende, F.F., Martins Lima, A., Niland, S., Wittig, I., Heide, H.,Schroder, K., Eble, J.A., 2012. Integrin alpha7beta1 is a redox-regulatedtarget of hydrogen peroxide in vascular smooth muscle cell adhesion.Free Radic. Biol. Med. 53, 521–531.

Eble, J.A., Tuckwell, D.S., 2003. The alpha2beta1 integrin inhibitor rho-docetin binds to the A-domain of the integrin alpha2 subunit proxi-mal to the collagen-binding site. Biochem. J. 376, 77–85.

Eble, J.A., Niland, S., Dennes, A., Schmidt-Hederich, A., Bruckner, P.,Brunner, G., 2002. Rhodocetin antagonizes stromal tumor invasionin vitro and other alpha2beta1 integrin-mediated cell functions.Matrix Biol. 21, 547–558.

Eble, J.A., Niland, S., Bracht, T., Mormann, M., Peter-Katalinic, J.,Pohlentz, G., Stetefeld, J., 2009. The alpha2beta1 integrin-specificantagonist rhodocetin is a cruciform, heterotetrameric molecule.Faseb. J. 23, 2917–2927.

Eble, J.A., 2005. Collagen-binding integrins as pharmaceutical targets.Curr. Pharm. Des. 11, 867–880.

Eckes, B., Zweers, M.C., Zhang, Z.G., Hallinger, R., Mauch, C., Aumailley, M.,Krieg, T., 2006. Mechanical tension and integrin alpha 2 beta 1 reg-ulate fibroblast functions. J. Investig. Dermatol. Symp. Proc. 11, 66–72.

F.T. Arlinghaus et al. / Toxicon 64 (2013) 96–105 105

Emsley, J., Knight, C.G., Farndale, R.W., Barnes, M.J., Liddington, R.C., 2000.Structural basis of collagen recognition by integrin alpha2beta1. Cell101, 47–56.

Heino, J., 2000. The collagen receptor integrins have distinct ligand rec-ognition and signaling functions. Matrix Biol. 19, 319–323.

Hynes, R.O., 1992. Integrins: versatility, modulation, and signaling in celladhesion. Cell 69, 11–25.

Marcinkiewicz, C., Lobb, R.R., Marcinkiewicz, M.M., Daniel, J.L.,Smith, J.B., Dangelmaier, C., Weinreb, P.H., Beacham, D.A.,Niewiarowski, S., 2000. Isolation and characterization of EMS16,a C-lectin type protein from Echis multisquamatus venom, a potentand selective inhibitor of the alpha2beta1 integrin. Biochemistry39, 9859–9867.

Momic, T., Arlinghaus, F.T., Arien-Zakay, H., Katzhendler, J., Eble, J.A.,Marcinkiewicz, C., Lazarovici, P., 2011. Pharmacological aspects ofvipera xantina palestinae venom. Toxins (Basel) 3, 1420–1432.

Morita, T., 2005. Structure-function relationships of C-type lectin-relatedproteins. Pathophysiol. Haemost. Thromb. 34, 156–159.

Nuyttens, B.P., Thijs, T., Deckmyn, H., Broos, K., 2011. Platelet adhesion tocollagen. Thromb. Res. 127 (Suppl. 2), S26–S29.

Ogawa, T., Chijiwa, T., Oda-Ueda, N., Ohno, M., 2005. Molecular diversityand accelerated evolution of C-type lectin-like proteins from snakevenom. Toxicon 45, 1–14.

Pilorget, A., Conesa, M., Sarray, S., Michaud-Levesque, J., Daoud, S.,Kim, K.S., Demeule, M., Marvaldi, J., El Ayeb, M., Marrakchi, N.,Beliveau, R., Luis, J., 2007. Lebectin, a Macrovipera lebetina venom-derived C-type lectin, inhibits angiogenesis both in vitro andin vivo. J. Cell. Physiol. 211, 307–315.

Shaw, L.M., 1999. Integrin function in breast carcinoma progression. J.Mammary Gland Biol. Neoplasia 4, 367–376.

Staniszewska, I., Walsh, E.M., Rothman, V.L., Gaathon, A., Tuszynski, G.P.,Calvete, J.J., Lazarovici, P., Marcinkiewicz, C., 2009. Effect of VP12 andviperistatin on inhibition of collagen-receptor-dependent melanomametastasis. Cancer Biol. Ther. 8, 1507–1516.

Takada, Y., Ye, X., Simon, S., 2007. The integrins. Genome Biol. 8, 215.Wong, E.S., Belov, K., 2012. Venom evolution through gene duplications.

Gene 496, 1–7.Zutter, M.M., Sun, H., Santoro, S.A., 1998. Altered integrin expression and

the malignant phenotype: the contribution of multiple integratedintegrin receptors. J. Mammary Gland Biol. Neoplasia 3, 191–200.