Jararhagin Disruption of Endothelial Cell Anchorage Is Enhanced in Collagen Enriched Matrices

Toxicon 108 (2015) 240e248

Contents lists available at ScienceDirect

Toxicon

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

Jararhagin disruption of endothelial cell anchorage is enhanced in collagen enriched matrices

C. Baldo a, D.S. Lopes b, E.L. Faquim-Mauro a, J.F. Jacysyn c, S. Niland d, J.A. Eble d, * P.B. Clissa a, A.M. Moura-da-Silva a, a Laboratorio de Imunopatologia, Instituto Butantan, Av. Vital Brazil, 1500, 05503-900, Sao~ Paulo, SP, Brazil b Instituto de Genetica e Bioquímica, Universidade Federal de Uberlandia,^ Uberlandia,^ MG, Brazil c LIM62, Faculdade de Medicina da Universidade de Sao~ Paulo, Sao~ Paulo, SP, Brazil d Institute of Physiological Chemistry and Pathobiochemistry, University of Münster, 48149, Münster, Germany article info abstract

Article history: Hemorrhage is one of the most striking effects of bites by viper snakes resulting in fast bleeding and Received 21 August 2015 ischemia in affected tissues. Snake venom metalloproteinases (SVMPs) are responsible for hemorrhagic Received in revised form activity, but the mechanisms involved in SVMP-induced hemorrhage are not entirely understood and the 19 October 2015 study of such mechanisms greatly depends on in vivo experiments. In vivo, hemorrhagic SVMPs accu- Accepted 27 October 2015 mulate on basement membrane (BM) of venules and capillary vessels allowing the hydrolysis of collagen Available online 31 October 2015 IV with consequent weakness and rupture of capillary walls. These effects are not reproducible in vitro with conventional endothelial cell cultures. In this study we used two-dimension (2D) or three- Keywords: Snake venom dimension (3D) cultures of HUVECs on matrigel and observed the same characteristics as in ex vivo Metalloproteinase experiments: only the hemorrhagic toxin was able to localize on surfaces or internalize endothelial cells Endothelial cell in 2D cultures or in the surface of tubules formed on 3D cultures. The contribution of matrigel, fibro- Extracellular matrix nectin and collagen matrices in jararhagin-induced endothelial cell damage was then analyzed. Collagen Hemorrhage and matrigel substrates enhanced the endothelial cell damage induced by jararhagin allowing toxin Collagen binding to focal adhesions, disruption of stress fibers, detachment and apoptosis. The higher affinity of jararhagin to collagen than to fibronectin explains the localization of the toxin within BM. Moreover, once located in BM, interactions of jararhagin with a2b1 integrin would favor its localization on focal adhesions, as observed in our study. The accumulation of toxin in focal adhesions, observed only in cells grown in collagen matrices, would explain the enhancement of cell damage in these matrices and reflects the actual interaction among toxin, endothelial cells and BM components that occurs in vivo and results in the hemorrhagic lesions induced by viper venoms. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction hemorrhage that involves direct damage of microvessels (Baldo et al., 2010). Extracellular matrix (ECM) consists of a complex macromolec- Venom induced hemorrhage is a relevant symptom of enven- ular network responsible for tissue structural integrity and modu- omation with viper snakes and causes the massive local damage by lation of cellular activity. Collagens are the most abundant proteins disruption of capillary vessels. Consequently, blood leaks into tis- in the ECM. Collagen remodeling by matrix metalloproteinases sues and distal tissues become ischemic due to the lack of blood (MMPs) is commonly observed during normal tissue homeostasis. supply. Venom-induced hemorrhage has been attributed to the A rapid and extensive degradation of collagens is also detected in ability of snake venom metalloproteinases (SVMPs) to degrade some pathological situations, such as cancer, autoimmune diseases basement membrane (BM) components (Baramova et al., 1989; (Rosenblum et al., 2010), and on pathogenesis of venom-induced Escalante et al., 2006). However, different SVMPs are described and they vary considerably in hemorrhagic activity. PIII-class SVMPs are generally potent hemorrhagins while PI-class SVMPs present weak or undetectable hemorrhagic activity. Since both PI * Corresponding author. and PIII-SVMPs are equally able to degrade ECM components, the E-mail address: [email protected] (A.M. Moura-da-Silva). http://dx.doi.org/10.1016/j.toxicon.2015.10.016 0041-0101/© 2015 Elsevier Ltd. All rights reserved. C. Baldo et al. / Toxicon 108 (2015) 240e248 241 structural difference that could account for the hemorrhagic ac- previously described (Paine et al., 1992) with modifications tivity is the presence of additional adhesive domains, present only (Moura-da-Silva et al., 2003). BnP1 was purified from B. neuwiedi in PIII SVMPs, which could potentiate the catalytic action of the venom according to Baldo et al. (2008). The purity of toxins was toxins by binding to ECM or endothelial cells (Fox and Serrano, evaluated by SDS-PAGE 12.5%, under reducing conditions. Toxins ® 2005; Kamiguti et al., 1996; Tanjoni et al., 2010). This issue was were used in the native form or labeled with Alexa Fluor 488 investigated by our group comparing the mechanisms related to (Molecular Probes, USA) according to the manufacturer in- hemorrhagic activity of jararhagin, a well-known hemorrhagic PIII structions. BSA (Bovine serum albumin e Sigma Aldrich, USA) was SVMP isolated from Bothrops jararaca venom (Paine et al., 1992) and used as control of non-toxic protein in different sets of BnP1, a non-hemorrhagic PI SVMP isolated from Bothrops neuwiedi experiments. venom (Baldo et al., 2008). The results demonstrated that hemorrhagic activity of jararhagin is correlated to its ability to bind to 2.2. Cell culture collagen (Baldo et al., 2010; Moura-da-Silva et al., 2008). In vivo, this toxin is able to selectively degrade the fibrillar collagen and base- HUVECs were harvested by type IV collagenase (Worthington, ment membrane collagen IV, co-localizing with the basement USA) digestion as described by Jaffe et al. (1973) and cultured on membrane of capillaries and venules on mouse skin (Baldo et al., gelatin-coated plastic dishes in RPMI 1640 medium supplemented 2010). On the other hand, BnP1 hydrolyzed ECM components with 10% FBS, 2 mM L-glutamine, 1 mM sodium pyruvate, 100 UI/mL in vitro (Baldo et al., 2008), but in vivo the toxin did not accumulate penicillin, 100 mg/mL streptomycin, 50 uM 2-mercaptoethanol, 5 U/ nearby blood vessels nor hydrolyzed BM collagens. These evidences mL heparin and 20 ng/mL bFGF. For bidimensional cultures (2D), 96 have been recently confirmed by Herrera et al., 2015 observing or 24 wells microplates were coated with type I collagen (BD tissue localization of two hemorrhagic SVMPs, CsH1 (PIII-SVMP Bioscience, USA), type IV collagen (BD Bioscience, USA), fibronectin from Crotalus simus) and BlatH1 (PII-SVMP from Bothriechis later- (Invitrogen, USA) or matrigel (Sigma Aldrich, USA) in a final con- alis), but not of BaP1, a weakly-hemorrhagic PI-SVMP from Bothrops centration of 50 mg/mL. Cells cultured on gelatin 1% or uncoated asper. In this sense, these observations strongly suggest the relevant plates were used as a control. For three-dimensional cultures (3D), role of interactions between adhesive domains present in the 30 mL of matrigel were added to a 24 well microplates containing toxins and BM components enhancing the proteolytic effect of glass coverslips (13 mm), following incubation of 30 min, 37 C. these enzymes, allowing disruption of microvessel walls. After that, 0.5 Â 105 cells were added on the top of gelled matrigel Endothelial cells are also essential for the integrity of micro- and cultured in the HUVECs medium for approximately 18 h, 37 C. vessels and the role of endothelial cells damage on hemorrhagic After that, the capillary-like structures could be observed. activity by PIII-SVMPs is another still open question. In vitro,both hemorrhagic and non-hemorrhagic SVMPs induce apoptosis of 2.3. Immunofluorescence assays endothelial cells at comparable levels (Diaz et al., 2005; Tanjoni et al., 2005). Additionally, hemorrhage induced by jararhagin Immunofluorescence experiments were carried out using in vivo occurs much faster than the apoptosis of endothelial cells markers for actin cytoskeleton (rhodamine-phalloidin), focal ad- in vitro, and apoptotic cells were not detected on hemorrhagic le- hesions (vinculin), and basement membrane (collagen IV and sions induced by SVMP from B. asper venom (Jimenez et al., 2008), laminin) as described below. In order to investigate the interaction ruling out that apoptosis of these cells is the prevalent mechanism of toxins with HUVECs, cells were treated with 800 nM Alexa Fluor for hemorrhages observed in vivo. However, apparently discrepant 488-labeled jararhagin (Alexa488-Jar), Alexa Fluor 488-labeled results obtained in vivo and in vitro may be due to the absence of BnP1 (Alexa488-BnP1) or Alexa Fluor 488-labeled BSA (Alexa488- experimental conditions that accurately model a complex micro- BSA) for 30 min. The samples were fixed in 3.7% formaldehyde environment observed in vivo. Most of these studies were per- for 10 min at room temperature. Nonspecific staining was blocked formed using endothelial cells grown on conventional culture by incubating the coverslips for 1 h at room temperature with a bottles, which are static and rigid surfaces. Therefore, they are very solution of Phosphate Buffer Saline (PBS) containing 1% triton X- distinct from the matrix microenvironment observed in vivo.In 100, 5% normal goat serum, 1% BSA, 0.5% glycine and 0.5% fish skin agreement with that, differences concerning proliferation, migra- gelatin. For cells cultured on 2D-model, the actin cytoskeleton was tion, morphology and activation of signaling pathways were stained with rhodamine phalloidin (1:100-Invitrogen, USA) by in- observed when cells are grown in substrates such as fibronectin cubation for 1 h, at room temperature. In 3-D model, the basement and collagen, compared to cells grown in absence of ECM compo- membrane of HUVEC tube-like structures were labeled by goat nent (Berrier and Yamada, 2007; Lim et al., 2010; Soucy and Romer, anti-rabbit type IV collagen polyclonal antibody (Chemicon, USA), 2009). Moreover, the discrepancy noted on action of SVMPs in vitro or goat anti-rabbit laminin polyclonal antibody (Chemicon, USA), and in vivo could be due to inappropriate methods used to evaluate both at a 1:40 dilution, for 18 h at 4 C. After washing with PBS, the the action of SVMPs in vitro. sections were incubated with anti-rabbit IgG conjugated to Tetra- In this sense, the aim of this study was to investigate the effect of methylrhodamine (TRITC) at 1: 500 dilution (Molecular Probes, jararhagin on endothelial cells cultured on different matrices, using USA), for 90 min at room temperature. For the analysis of focal primary culture of HUVECs as model. This set of experiments re- adhesion and cytoskeleton alterations induced by jararhagin, veals that cells grown in collagen matrices are more susceptible to HUVECs were treated with 800 nM of toxins or BSA for 3 h. Cells the action of jararhagin which concentrates in focal adhesion, dis- treated with PBS were used as negative control. The samples were rupts cellematrix interaction and enhances cell susceptibility to fixed and blocked as described above and incubated overnight at apoptosis. The results showed here can bring new insights about 4 C with mouse anti-vinculin antibodies (1:100- Sigma Aldrich, the mechanisms of SVMP-induced pathology. USA) and after washing, cells were incubated with Alexa fluor 488- labeled goat anti-mouse IgG (1:1000 e Molecular Probes, USA) and 2. Material and Methods rhodamine phalloidin (1:100) for 1 h, at room temperature. Nega- tive control consisted of omitting the primary antibody step from 2.1. Toxins the protocol. The effect of extracellular matrix on morphology and spreading of HUVECs on 2-D model was evaluated in non-treated Jararhagin was isolated from B. jararaca snake venom as cells. The samples were examined with a Confocal Microscope 242 C. Baldo et al. / Toxicon 108 (2015) 240e248

LSM 510 Meta (Zeiss) and the quantiﬁcation of cell spreading was 4. Results and discussion carried out using the LMS Image Browser (Zeiss) software based on rhodamine phalloidin staining. 4.1. Interaction of jararhagin and BnP1 with endothelial cells

2.4. Flow cytometry assays According to our previous data, jararhagin exhibits high affinity to collagen and this binding is essential for the expression of The direct interaction of jararhagin with HUVECs or a2b1 hemorrhagic activity in vivo (Baldo et al., 2010; Moura-da-Silva integrin present on cell membrane was evaluated by flow cytom- et al., 2008). Thus, our first step in this study was to evaluate if etry, using cells cultured on collagen I, fibronectin or matrigel 2D or 3D cultures with ECM components would reproduce the during 2 days. After that, the cells were harvested by trypsin and same mechanism in vitro. For this purpose, we included in our the concentration adjusted to 106 cells/30 mL PBS containing 1% experiments also BnP1, a non-hemorrhagic PI-class SVMP, which in BSA. Cells were incubated with anti-FcgRII/III monoclonal antibody previous experiments (Baldo et al., 2008) did not show any inter- (1 mg/106 cells - BD Biosciences, USA), for 15 min, at 4 C, washed action with collagens in solid phase assays. Jararhagin and BnP1 with PBS containing 1% BSA, and resuspended in culture medium. interaction with HUVECs were then evaluated in cells cultured on Cultures were incubated with 800 nM Alexa488-Jar for 30 min. Cell matrigel in 2D- and 3D-models. As shown in Fig. 1A, in the 2D- suspensions were also incubated with FITC-labeled IgG-isotype model, binding of jararhagin to HUVECs was detectable after control mAb (R&D System, USA) or Alexa488-BSA. For a2b1 integrin, 30 min of treatment as it localized preferentially at cell surface cells cultured on collagen I were incubated with 800 nM of non- regions where actin fibers were also abundant. In contrast, treat- labeled jararhagin for 30 min at 37 C. After washing, cells were ment with BnP1 did not result in any detectable fluorescence signal, incubated with anti-a2 integrin (CD49b- R&D Systems, USA), in the similarly to control cells incubated with Alexa488-BSA, indicating concentration of 1 mg/106 of cells in PBS/1% BSA, for 30 min/4 C. that non-hemorrhagic PeI SVMPs do not bind to endothelial cells The cells were centrifuged and resuspended in 0.2 mL of PBS with cultured on matrigel. Similar results were found on 3D cultures: 1% paraformaldehyde. Flow cytometry of triplicate samples (104 after 30 min treatment with jararhagin or BnP1, only jararhagin was events/data acquisition file) were performed with FACSCalibur concentrated near the tube-like structures formed by endothelial (Becton Dickinson, USA) using CellQuest Software (BD Biosciences, cells and was localized close to capillary-like structures up to 6 h of USA). The results were expressed as the percentage of positively incubation whilst no signal was detected in BnP1 or BSA treated labeled cells. cultures (Fig. 1B). Binding of jararhagin to the basement membrane of capillary-like structures was then investigated by a double- 2.5. Solid phase assays staining protocol using Alexa488-Jar and collagen IV or laminin. Treatment of cells with labeled jararhagin revealed fluorescent To evaluate the interaction between jararhagin and ECM com- staining around capillary vessels, which co-localized with base- ponents, collagen I (10 mg/mL in 0.1 M of acetic acid), fibronectin ment membrane staining of collagen IV or laminin labeled anti- (10 mg/mL in PBS), or matrigel (10 mg/mL in PBS) were coated on bodies (Fig. 2), in line with our previous in vivo experiments (Baldo microtiter plate wells and incubated overnight at 4 C. After et al., 2010). blocking with 1% of BSA in TBS, pH 7.4, different dilutions of jar- The mechanical properties and molecular composition of the arhagin were added to the plates and incubated for 2 h, at 37 C. For extracellular matrix determine the patterns of adhesion, migration a2b1 integrin binding assays, the collagen-binding site of a2b1 and morphology, deeply influencing the behavior of cells (Discher integrin, the a2A domain, was obtained as described (Calderwood et al., 2005). Also, the susceptibility of cells to deleterious treat- et al., 1997; Eble and Tuckwell, 2003). The recombinant fragment ments may be affected by the microenvironment in which cells are was coated on microtiter plate (5 mg/mL in PBS) and TBS buffer cultured. In this aspect, we could reproduce our previous ex vivo containing 2 mM MgCl2 was used for washing procedures. Bound evidences in vitro, as jararhagin interacted with BM components jararhagin was detected with rabbit polyclonal antiserum against likewise in the 3D matrigel system and in microvessels (Baldo et al., jararhagin, followed by alkaline phosphatase-coupled secondary 2010; Herrera et al., 2015; Moura-da-Silva et al., 2008). Binding of antibody and the enzyme substrate. Binding intensity was accessed PIII-class SVMPs to BM components in the in vivo model, and their by optical density at 492 nm. localization in microvessel walls were correlated with enhanced hydrolysis of collagen IV (Baldo et al., 2010; Herrera et al., 2015) 2.6. Apoptosis assays resulting in hemorrhage because of weakness and rupture of the vessel (Gutierrez et al., 2005). Interestingly, we did not detect any Apoptosis was accessed by the DNA content of cells stained with binding of Alexa488-Jar to collagen presented in matrigel, either in Hypotonic Fluorescent Solution (50 mg/mL propidium iodide, 0.1% 2D or 3D cultures (Fig. 1). Two different hypotheses could explain Triton X-100 and 0.1% sodium citrate), analyzed by FACScalibur this observation: either the arrangement of collagen included in (BectoneDickinson, USA). DNA fragmentation was quantified by matrigel is different than the one observed in solid phase assay cell cycle analysis of total DNA content as previously described coated plates, resulting in different exposition of the epitopes (Nicoletti et al., 1991). Using the FSC x SSC parameters, debris were responsible for jararhagin binding or the interaction between jar- gated out and 10,000 events were counted. Acquisition and analysis arhagin and endothelial BM components depends also on interac- were performed with CellQuest software (BectoneDickinson, USA) tion with endothelial cell surface markers. Therefore, our next step and the results represent the percentage of the population gated in was to investigate the cellular target of jararhagin on different ECM the region corresponded to sub-G1 phase (fractional DNA content). components and in endothelial cell surface markers. Samples were run in triplicates of two independents experiments. In solid phase assays, high affinity binding of jararhagin to collagen was confirmed using isolated collagen or matrigel, which 3. Ethical statements contains collagen IV. In opposition, no binding was observed between jararhagin and fibronectin (Fig. 3A). This is in agreement Human vascular endothelial cells (HUVECS) were obtained from with previous data showing that collagens are preferred targets of umbilical cords of newborns (Hospital of University of Sao~ Paulo: PIII-class SVMPs amongst the ECM molecules. Binding of PIII-class Ethical Committee for Human Protocol: 526/04). SVMPs was demonstrated in vitro to ECM proteins as collagen I C. Baldo et al. / Toxicon 108 (2015) 240e248 243

Fig. 1. Interaction of Alexa488-labeled SVMPs with HUVECs cultured on matrigel on 2D (A) and 3D (B) systems. HUVECs were treated with 800 nM of Alexa488-Jar or Alexa488- BnP1 for 30 min. Cells treated with Alexa488-BSA were used as control. Actin ﬁlaments were stained with rhodamin phalloidin (1:100). The samples were examined with a Confocal Microscope LSM 510 Meta (Zeiss).

Fig. 2. Binding of jararhagin to the capillary-like 3D structures. HUVECs cultured on 3D system were treated with 800 nM of Alexa488-Jar for 30 min and incubated with goat anti- rabbit type IV collagen polyclonal antibody or goat anti-rabbit laminin polyclonal antibody (1:40), followed by incubation with anti-rabbit IgG conjugated to TRITC (1:500). The samples were analyzed in confocal microscope (LSM-510-Meta, Zeiss).

(Moura-da-Silva et al., 2008; Serrano et al., 2005; Zhou et al., 1996), receptor (Fig. 3C). Therefore, a2b1 integrin may be involved in collagen IV (Moura-da-Silva et al., 2008), collagen XII and XIV and jararhagin-endothelial cells interaction. Thus, our data, presented the matrilins 1, 3 and 4 (Serrano et al., 2006). To test the interaction here, support the previous evidences that both disintegrin-like between jararhagin and endothelial cell surface markers, we used (Kamiguti et al., 1997; Moura-da-Silva et al., 2008; Tanjoni et al., the a2b1 integrin as target based on previous results showing that 2010) and cysteine rich domains (Serrano et al., 2006, 2005, 2007) jararhagin presents affinity for this receptor (De-Luca et al., 1995; are involved in the binding to ECM proteins and integrin receptors Moura-da-Silva et al., 2001; Tanjoni et al., 2010). In solid phase indicating that non-enzymatic mechanisms are also involved in the assays, jararhagin was able to bind to immobilized recombinant a2A action of hemorrhagic SVMPs. domain integrin subunit (Fig. 3B). The a2A domain is the recognized According to these data, the localization of jararhagin to the collagen-binding site within a2b1 integrin (Calderwood et al., 1997; endothelial cell- BM-zone is possibly driven by interaction with Emsley et al., 1997). In addition, several regions of jararhagin ECM components as well as endothelial cell surface markers. These molecule may act as ligands for a2A domain present in a2b1 integrin evidences prompted us to investigate the effect of jararhagin in as a cyclic RKKH peptide derived from a charged hydrophilic region endothelial cells cultured in collagen containing matrices or in of metalloproteinase domain of jararhagin (Ivaska et al., 1999; fibronectin substrate, which expressed strong and no interaction Pentikainen€ et al., 1999) or the A domain present on vWF as the with the toxin, respectively. hypervariable region present on the cysteine-rich domain of class P-III SVMPs (Serrano et al., 2006). Additionally, the binding of anti- 4.2. Influence of ECM proteins on HUVECs behavior a2 antibodies to endothelial cells was reduced approximately 50% when cultures were treated with jararhagin. This indicates the a2b1 In order to evaluate the effect of jararhagin on endothelial cells integrin-directed antibodies compete with jararhagin for the same cultured on different substrates, we first evaluated the behavior of 244 C. Baldo et al. / Toxicon 108 (2015) 240e248

Fig. 3. Binding of jararhagin to ECM components and endothelial cells. (A) Binding of jararhagin to ECM substrates and (B) to the a2A domain of a2b1 integrin by solid phase assay: Different toxin concentrations were incubated with collagen I, fibronectin, matrigel or to the a2A domain previously immobilized to a microtiter plate. Bound jararhagin was quantified by ELISA by polyclonal anti-jararhagin antibodies, as described in Material and Methods. The results are expressed as the mean of the optical density obtained in the samples run in triplicate ± SD of one from three representative experiments. The symbol (*) represents a significant difference (p < 0.05) compared with samples non-treated with jararhagin (Control). (C) Binding of jararhagin to HUVECs surface: Cells were incubated or not with 800 nM jararhagin for 30 min. After this, the cells were incubated with FITC- conjugated anti-a2 subunit of a2b1 integrin (CD49b) antibody followed by flow cytometry analysis. The results represent the mean of the percentage of a2b1 positive cells of the samples run in triplicate ± SD. Samples of cells incubated with FITC-conjugated IgG isotype control antibody were used as control. The symbol (*) represents a significant difference (p < 0.05) compared with the control group.

HUVECs when cultured on different ECM components. Marked with Alexa488-BSA did not shown any labeling, showing the differences in cell shape, size and actin distribution could be specificity of the test. In cells cultured on fibronectin, the fluores- observed. On collagen IV, fibronectin and matrigel, endothelial cells cence signal was concentrated inside the cell, close to the nucleus, showed an increased spreading area and a polygonal morphology indicating that jararhagin may be internalized in HUVECs cultured with a larger number of stress fibers. On contrary, cells grown in the on these conditions (Fig. 5A-Fibronectin). In contrast, in collagen absence of substrate i.e. only in plastic bottles or in the presence of matrices (collagen-I and matrigel), jararhagin was mostly detected gelatin showed an elongated morphology with actin fibers on the cell surface and co-localized with the extremity of actin distributed in the peripheral region of the cell, and a reduction of stress fibers, apparently in focal adhesion regions (Fig. 5A). The spreading area. Cells cultured in collagen I presented an interme- shape of jararhagin treated cells grown in different matrices was diate behavior with a slight increase in cell spreading, elongated also different with clear signs of retraction of actin fibers and morphology and most of actin fibers in the periphery (Fig. 4A). The detachment on cells cultured on matrigel or collagen substrates, quantification of the spreading area, showed statistically significant while the cells cultured on fibronectin presented a similar shape of differences in the spreading area of cells cultured in collagen I and control cells despite the treatment with jararhagin (Fig. 5A). Ac- IV, fibronectin and matrigel (Fig. 4B). These results showed that size cording to these observations, it is reasonable to suggest that the and cell shape of HUVECs are profoundly influenced by ECM interaction of the toxin with matrix components drives jararhagin constitution and are in agreement with the literature that already fate within the cell. High affinity binding of jararhagin to collagen showed that the constitution of the ECM substrates is able to affect approximates the toxin to integrins present in the focal adhesions the morphology, differentiation and proliferation of cells (Discher and concentrates the catalytic action of the enzyme towards pro- et al., 2005; Engler et al., 2004). Considering the differences in teins essential to cell anchorage, leading to detachment. In contrast, endothelial cells behavior in collagen I, fibronectin and matrigel in fibronectin matrix, jararhagin binds only to integrin receptor and matrices and the differences of jararhagin interaction to these ECM without an extra support due to the lack of binding to ECM, the proteins in vitro, these substrates were chosen to the next set of toxin possibly takes a preferential route for cell internalization. experiments. Specific binding of jararhagin to HUVECs was also detected by flow cytometry in cells treated for 30 min with jararhagin on different fl 4.3. Interaction of jararhagin and endothelial cells cultured on substrates. In these experiments, the uorescent signal due to jar- different matrices arhagin binding was equally detected on cells cultured on collagen, matrigel and fibronectin (Fig. 5B), since the method detects fl Binding of jararhagin to HUVECs cultured on fibronectin, orescent signals from cell surface or internalized jararhagin. collagen and matrigel substrates was next analyzed. As shown in Cell shape is controlled by the cytoskeleton that acts as a me- Fig. 5, labeled jararhagin was detected in endothelial cells grown in chanically supporting framework (Ingber, 1993; Wang et al., 1993). the three different substrates (Fig. 5A). On contrary, cells treated Alterations of the actin cytoskeleton may be a signal for apoptosis C. Baldo et al. / Toxicon 108 (2015) 240e248 245

Fig. 4. Effect of extracellular matrix on morphology and spreading of HUVECs on 2D model. Cells were cultured on 24-well plates previously coated with different ECM substrates for 18 h. Cells grown on gelatin and uncoated plates were also evaluated. A) Actin filaments were stained with rhodamin phalloidin (1:100) and cultures were examined with a Confocal Microscope LSM 510 Meta (Zeiss). B) Spreading cell area based on F-actin staining. Quantification of actin stained was carried out using the LMS Image Browser (Zeiss) software. The experiment was performed in quadruplicate and the data are expressed as mean ± s.d. of one representative experiment. Significance was calculated by Student's T and the symbol (*) represents a significant difference (p < 0.05) compared to uncoated.

Fig. 5. Interaction of Alexa488-labeled SVMPs with HUVECs cultured on matrigel, fibronectin and collagen I on 2D-model. HUVECs were treated with 800 nM of Alexa488-Jar for 30 min. Cells treated with Alexa488-BSA were used as control. A) Cultures were observed by immunofluorescence with actin filaments stained with rhodamin phalloidin (1:100). The samples were examined with a Confocal Microscope LSM 510 Meta (Zeiss). B) Interaction of Alexa488-Jar with HUVECs was quantified by flow cytometry analysis of triplicate samples with FACSCalibur (Becton Dickinson, USA) using CellQuest Software (BD Biosciences). The results express the percentage of cells expressing Alexa488-Jar. The experiment was carried out in triplicate and the significance was calculated by Student's T test. The symbol * represent significant difference (p < 0.05) compared to samples treated with the control (Alexa488-BSA).

(Korichneva and Hammerling,€ 1999; Wang et al., 1993).Flusberg we next investigated the alterations induced by jararhagin on et al. (2001) showed the ability of cytoskeleton to control the cytoskeleton and focal adhesions. Fig. 6 shows the control cells with endothelial cell survival through Bcl-2 expression as well as via the typical distribution of actin fibers with vinculin spots on cells phosphorylation of protein kinase Akt. Thus, the interference on borders. After 3 h of jararhagin treatment, a diffuse distribution of endothelial cell cytoskeletal structure caused by jararhagin may vinculin throughout the cytoplasm was observed in cells grown on correlate with the initiation of apoptosis signal transduction. Thus, all tested substrates. On collagen I and matrigel, cells retracted, 246 C. Baldo et al. / Toxicon 108 (2015) 240e248 rounded up, and lacked any pattern characteristic of focal adhe- induced only 48 h after treatment with 800 nM jararhagin (Baldo sions were observed. On fibronectin, the cells showed a minor et al., 2008) and the same length of time and toxin concentration retraction of actin cytoskeleton and preserved focal adhesions. were needed to induce apoptosis in tEnd cells cultured in plastic Comparable results were obtained by Lim et al. (2010) studying the surface (Tanjoni et al., 2005). The signaling pathways observed in effects of extracellular matrix in smooth muscle cells. The authors the latter study included a reduced phosphorylation of FAK as well reported a higher coefficient of adhesion and stability of focal ad- as its association to cytoskeleton proteins, an increase in Bax levels hesions in cells grown in fibronectin than on collagen I and IV. On and the processing of pro-caspase-3. This suggested that disruption fibronectin, the cells were well anchored and less susceptible to of focal adhesion was the main mechanism involved in jararhagin- drastic changes. In another study, HUVECs showed different induced apoptosis (Tanjoni et al., 2005). The same mechanisms adherence patterns when grown on matrix derived from fibro- may be applied to jararhagin induced apoptosis in HUVECs cultured blasts, where the co-localization of focal adhesions proteins (vin- on collagen matrices since retraction of cytoskeleton and disrup- culin, pY397FAK, pY410-130Cas) with fibronectin was significantly tion of focal adhesions were associated with the treatment with the higher compared with collagen I and IV (Soucy and Romer, 2009). In toxin. However, the strength of the apoptotic signal was much our experiments, HUVECs cultured on fibronectin were more higher in cells cultured in collagen matrices and the reaction resistant to the treatment with jararhagin and preserved the focal occurred much faster with reduced concentrations of the toxin, adhesion structures as observed by the immunofluorescence for showing the relevant role of the extracellular matrix components vinculin. However, in our model, jararhagin did not co-localize with on the behavior of endothelial cells treated with jararhagin. actin fibers or focal adhesion in fibronectin cultures but showed an apparent internalization to the endothelial cell. A remarkable dif- 5. Concluding remarks ference was observed in HUVECs cultured in collagen matrices with a co-localization of the toxin with actin fibers and focal adhesions In this study we showed that 2D- or 3D-cultures of HUVECs on (Fig. 5). The higher concentration of jararhagin with these struc- matrigel are good models to exploit in vitro the intrinsic mecha- tures in initial times would enhance the disruption cell anchorage nisms related to disruption of capillary vessels and to venom- signaling to apoptosis. induced hemorrhage. As previously observed in ex vivo experi- Apoptosis was then accessed by DNA fragmentation by flow ments, only the hemorrhagic toxin was able to localize to endo- cytometry. As shown in Fig. 7, incubation of endothelial cells with thelial cells in 2D-cultures or to the surface of tubules formed on jararhagin resulted in DNA fragmentation preferentially on cells 3D-cultures. In this scenario, collagen reproduced the effects plated on matrigel and collagen. At 100 nM jararhagin, hypodiploid observed in matrigel cultures enhancing the endothelial cell dam- nucleic cells were 43.26% and 31.12% in cultures grown in matrigel age induced by jararhagin that includes binding to focal adhesions, and collagen I, respectively. At higher concentrations (800 nM) disruption of stress fibers, detachment and apoptosis. Considering jararhagin induced DNA fragmentation in a higher number of cells that the macromolecular organization and the biomechanical sta- (matrigele79.10%; collagen Ie72.20%). DNA fragmentation was also bility of BM is mainly determined by the network of type IV observed on cells plated on fibronectin, but to a minor extent, at collagen (Kühn, 1995), the affinity of jararhagin to collagen would both concentrations of jararhagin (100 nMe11.32%; drive the localization of the toxin within BM. Moreover, once 800 nMe56.28%). located in BM, interactions of jararhagin with a b integrin would In previous studies, apoptosis of HUVECs cultured on gelatin was 2 1 favor its localization on focal adhesions, as observed in our study.

Fig. 6. Focal adhesion and cytoskeleton alteration induced by jararhagin in HUVECs cultured in different components of extracellular matrix. Cells were treated with 800 nM of jararhagin for 3 h. Cells treated with PBS were used as a control. Focal adhesions were stained with mouse anti-vinculin antibodies (1:100) followed by incubation with Alexa ﬂuor 488 goat anti-mouse IgG (1:1000). The cytoskeleton was stained with rhodamine phalloidin (1:100). The small square highlights the typical distribution of actin ﬁbers with vinculin spots on control cells borders. The samples were examined with a Confocal Microscope LSM 510 Meta (Zeiss). C. Baldo et al. / Toxicon 108 (2015) 240e248 247

Fig. 7. Induction of apoptosis by jararhagin in HUVECs cultured in different components of extracellular matrix. Cells were treated with 100 or 800 nM jararhagin for 6 h. Cells treated with PBS were used as a control. Apoptosis was evaluated by ﬂow cytometry analysis of DNA content. Data are expressed as mean ± SD of one representative experiment performed in triplicate. * p < 0.05, vs. correspondent control (T test).

The accumulation of toxin in focal adhesions, observed only in cells venom hemorrhagic metalloproteinases: a biochemical and immunohisto- e grown in collagen matrices, would explain the enhancement of cell chemical study. Arch. Biochem. Biophys. 455, 144 153. Flusberg, D.A., Numaguchi, Y., Ingber, D.E., 2001. Cooperative control of Akt phos- damage in these matrices and reflects the actual interaction among phorylation, bcl-2 expression, and apoptosis by cytoskeletal microfilaments and toxin, endothelial cells and BM components that occurs in vivo and microtubules in capillary endothelial cells. Mol. Biol. Cell 12, 3087e3094. results in the hemorrhagic lesions induced by viper venoms. Fox, J.W., Serrano, S.M.T., 2005. Structural considerations of the snake venom metalloproteinases, key members of the M12 reprolysin family of metalloproteinases. Toxicon 45, 969e985. Acknowledgments Gutierrez, J.M., Rucavado, A., Escalante, T., Diaz, C., 2005. Hemorrhage induced by snake venom metalloproteinases: biochemical and biophysical mechanisms involved in microvessel damage. Toxicon 45, 997e1011. This study was supported by FAPESP (2012/16277-9, 2009/ Herrera, C., Escalante, T., Voisin, M.B., Rucavado, A., Morazan, D., Macedo,^ J.K., 09834-6), CAPES (Edict 063/2010 projects 1519/2011 and 1209/ Calvete, J.J., Sanz, L., Nourshargh, S., Gutierrez, J.M., Fox, J.W., 2015. Tissue 2011, Toxinology), CNPq (304025/2014-3; 311915/2012-4) and localization and extracellular matrix degradation by PI, PII and PIII snake venom metalloproteinases: clues on the mechanisms of venom-induced hemorrhage. INCTTox Program of MCT/CNPq/FAPESP. JAE received financial PLoS Negl. Trop. Dis. 9, e0003731. support by Deutsche Forschungsgemeinschaft (grant: SFB1009 Ingber, D.E., 1993. Cellular tensegrity: defining new rules of biological design that e project A09). govern the cytoskeleton. J. Cell Sci. 104 (Pt 3), 613 627. Ivaska, J., Kapyl€ a,€ J., Pentikainen,€ O., Hoffren, A.M., Hermonen, J., Huttunen, P., Johnson, M.S., Heino, J., 1999. A peptide inhibiting the collagen binding function Transparency document of integrin alpha2I domain. J. Biol. Chem. 274, 3513e3521. Jaffe, E.A., Nachman, R.L., Becker, C.G., Minick, C.R., 1973. Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and Transparency document related to this article can be found immunologic criteria. J. Clin. Invest. 52, 2745e2756. online at http://dx.doi.org/10.1016/j.toxicon.2015.10.016. Jimenez, N., Escalante, T., Gutierrez, J.M., Rucavado, A., 2008. Skin pathology induced by snake venom metalloproteinase: acute damage, revascularization, and re-epithelization in a mouse ear model. J. Invest. Dermatol. 128, References 2421e2428. Kamiguti, A.S., Hay, C.R., Theakston, R.D., Zuzel, M., 1996. Insights into the mechanism of haemorrhage caused by snake venom metalloproteinases. Toxicon 34, Baldo, C., Jamora, C., Yamanouye, N., Zorn, T.M., Moura-da-Silva, A.M., 2010. 627e642. Mechanisms of vascular damage by hemorrhagic snake venom metal- Kamiguti, A.S., Moura-da-Silva, A.M., Laing, G., Knapp, T., Zuzel, M., Crampton, J.M., loproteinases: tissue distribution and in situ hydrolysis. Plos Negl. Trop. Dis. 4. Theakston, R.D.G., 1997. Collagen-induced secretion-dependent phase of Baldo, C., Tanjoni, I., Leon, I.R., Batista, I.F.C., Della-Casa, M.S., Clissa, P.B., Weinlich, R., platelet aggregation is inhibited by the snake venom metalloproteinase jar- Lopes-Ferreira, M., Lebrun, I., Amarante-Mendes, G., Rodrigues, V.M., Perales, J., arhagin. Biochim. Biophys. Acta-Gen. Subj. 1335, 209e217. Valente, R., Moura-da-Silva, A.M., 2008. BnP1, a novel P-I metalloproteinase Korichneva, I., Hammerling,€ U., 1999. F-actin as a functional target for retro- from Bothrops neuwiedi venom: biological effects benchmarking relatively to retinoids: a potential role in anhydroretinol-triggered cell death. J. Cell Sci. jararhagin, a P-IIISVMP. Toxicon 51, 54e65. 112 (Pt 15), 2521e2528. Baramova, E.N., Shannon, J.D., Bjarnason, J.B., Fox, J.W., 1989. Degradation of Kühn, K., 1995. Basement membrane (type IV) collagen. Matrix Biol. 14, 439e445. extracellular matrix proteins by hemorrhagic metalloproteinases. Arch. Bio- Lim, S.M., Kreipe, B.A., Trzeciakowski, J., Dangott, L., Trache, A., 2010. Extracellular chem. Biophys. 275, 63e71. matrix effect on RhoA signaling modulation in vascular smooth muscle cells. Berrier, A.L., Yamada, K.M., 2007. Cell-matrix adhesion. J. Cell Physiol. 213, 565e573. Exp. Cell Res. 316, 2833e2848. Calderwood, D.A., Tuckwell, D.S., Eble, J., Kühn, K., Humphries, M.J., 1997. The Moura-da-Silva, A.M., Della-Casa, M.S., David, A.S., Assakura, M.T., Butera, D., integrin alpha1 A-domain is a ligand binding site for collagens and laminin. Lebrun, I., Shannon, J.D., Serrano, S.M.T., Fox, J.W., 2003. Evidence for hetero- J. Biol. Chem. 272, 12311e12317. geneous forms of the snake venom metalloproteinase jararhagin: a factor De-Luca, M., Ward, C.M., Ohmori, K., Andrews, R.K., Berndt, M.C., 1995. Jararhagin contributing to snake venom variability. Arch. Biochem. Biophys. 409, 395e401. and jaracetin: novel snake venom inhibitors of the integrin collagen receptor, Moura-da-Silva, A.M., Marcinkiewicz, C., Marcinkiewicz, M., Niewiarowski, S., 2001. alpha 2 beta 1. Biochem. Biophys. Res. Commun. 206, 570e576. Selective recognition of alpha(2)beta(1) integrin by Jararhagin, a metal- Diaz, C., Valverde, L., Brenes, O., Rucavado, A., Gutierrez, J.M., 2005. Characterization loproteinase/disintegrin from Bothrops jararaca venom. Thrombosis Res. 102, of events associated with Apoptosis/Anoikis induced by snake venom metal- 153e159. loproteinase BaP1 on human endothelial cells. J. Cell. Biochem. 94, 520e528. Moura-da-Silva, A.M., Ramos, O.H.P., Baldo, C., Niland, S., Hansen, U., Ventura, J.S., Discher, D.E., Janmey, P., Wang, Y.L., 2005. Tissue cells feel and respond to the Furlan, S., Butera, D., Della-Casa, M.S., Tanjoni, I., Clissa, P.B., Fernandes, I., stiffness of their substrate. Science 310, 1139e1143. Chudzinski-Tavassi, A.M., Eble, J.A., 2008. Collagen binding is a key factor for the Eble, J.A., Tuckwell, D.S., 2003. The alpha2beta1 integrin inhibitor rhodocetin binds hemorrhagic activity of snake venom metalloproteinases. Biochimie 90, to the A-domain of the integrin alpha2 subunit proximal to the collagen- 484e492. binding site. Biochem. J. 376, 77e85. Nicoletti, I., Migliorati, G., Pagliacci, M.C., Grignani, F., Riccardi, C., 1991. A rapid and Emsley, J., King, S.L., Bergelson, J.M., Liddington, R.C., 1997. Crystal structure of the I simple method for measuring thymocyte apoptosis by propidium iodide domain from integrin alpha2beta1. J. Biol. Chem. 272, 28512e28517. staining and flow cytometry. J. Immunol. Methods 139, 271e279. Engler, A.J., Griffin, M.A., Sen, S., Bonnemann,€ C.G., Sweeney, H.L., Discher, D.E., Paine, M.J., Desmond, H.P., Theakston, R.D., Crampton, J.M., 1992. Purification, 2004. Myotubes differentiate optimally on substrates with tissue-like stiffness: cloning, and molecular characterization of a high molecular weight hemor- pathological implications for soft or stiff microenvironments. J. Cell Biol. 166, rhagic metalloprotease, jararhagin, from Bothrops jararaca venom. Insights into 877e887. the disintegrin gene family. J. Biol. Chem. 267, 22869e22876. Escalante, T., Shannon, J., Moura-da-Silva, A.M., Gutierrez, J.M., Fox, J.W., 2006. Pentikainen,€ O., Hoffren, A.M., Ivaska, J., Kapyl€ a,€ J., Nyronen,€ T., Heino, J., Novel insights into capillary vessel basement membrane damage by snake 248 C. Baldo et al. / Toxicon 108 (2015) 240e248

Johnson, M.S., 1999. “RKKH” peptides from the snake venom metalloproteinase speciﬁc proteolysis of von Willebrand factor and inhibition of von Willebrand of Bothrops jararaca bind near the metal ion-dependent adhesion site of the factor-mediated platelet aggregation. FEBS J. 274, 3611e3621. human integrin alpha(2) I-domain. J. Biol. Chem. 274, 31493e31505. Soucy, P.A., Romer, L.H., 2009. Endothelial cell adhesion, signaling, and morpho- Rosenblum, G., Van den Steen, P.E., Cohen, S.R., Bitler, A., Brand, D.D., genesis in ﬁbroblast-derived matrix. Matrix Biol. 28, 273e283. Opdenakker, G., Sagi, I., 2010. Direct visualization of protease action on collagen Tanjoni, I., Evangelista, K., Della-Casa, M.S., Butera, D., Magalhaes, G.S., Baldo, C., triple helical structure. PLoS One 5, e11043. Clissa, P.B., Fernandes, I., Eble, J.A., Moura-da-Silva, A.M., 2010. Different regions Serrano, S.M., Kim, J., Wang, D., Dragulev, B., Shannon, J.D., Mann, H.H., Veit, G., of the class P-III snake venom metalloproteinase jararhagin are involved in Wagener, R., Koch, M., Fox, J.W., 2006. The cysteine-rich domain of snake venom binding to alpha(2)beta(1) integrin and collagen. Toxicon 55, 1093e1099. metalloproteinases is a ligand for von Willebrand factor A domains: role in Tanjoni, I., Weinlich, R., Della-Casa, M., Clissa, P.B., Saldanha-Gama, R.F., Freitas, M.S., substrate targeting. J. Biol. Chem. 281, 39746e39756. Barja-Fidalgo, C., Amarante-Mendes, G.P., Moura-da-Silva, A.M., 2005. Jar- Serrano, S.M.T., Jia, L.G., Wang, D.Y., Shannon, J.D., Fox, J.W., 2005. Function of the arhagin, a snake venom metalloproteinase, induces a specialized form of cysteine-rich domain of the haernorrhagic metalloproteinase atrolysin A: tar- apoptosis (anoikis) selective to endothelial cells. Apoptosis 10, 851e861. geting adhesion proteins collagen I and von Willebrand factor. Biochem. J. 391, Wang, N., Butler, J.P., Ingber, D.E., 1993. Mechanotransduction across the cell surface 69e76. and through the cytoskeleton. Science 260, 1124e1127. Serrano, S.M.T., Wang, D., Shannon, J.D., Pinto, A.F., Polanowska-Grabowska, R.K., Zhou, Q., Dangelmaier, C., Smith, J.B., 1996. The hemorrhagin catrocollastatin in- Fox, J.W., 2007. Interaction of the cysteine-rich domain of snake venom met- hibits collagen-induced platelet aggregation by binding to collagen via its alloproteinases with the A1 domain of von Willebrand factor promotes site- disintegrin-like domain. Biochem. Biophysical Res. Commun. 219, 720e726.