Cellular The Geodiamolide H, Derived From Brazilian Sponge ... · 0.5% Triton X-100, 1 mg/ml RNase in 2X SSC. Nuclei were counterstained with Sytox Green (Invitrogen-Molecular Probes)

ORIGINAL ARTICLE 583J o u r n a l o fJ o u r n a l o f

CellularPhysiologyCellularPhysiology
The Geodiamolide H, Derived
From Brazilian Sponge GeodiaCorticostylifera, Regulates ActinCytoskeleton, Migration andInvasion of Breast Cancer CellsCultured in Three-DimensionalEnvironment
VANESSA M. FREITAS,1* MARISA RANGEL,2 LETICIA F. BISSON,1 RUY G. JAEGER,1
AND GLAUCIA M. MACHADO-SANTELLI1

1Department of Cell and Developmental Biology, Institute of Biomedical Sciences, University of Sao Paulo, Sao Paulo, Brazil2Laboratory of Immunopathology, Butantan Institute, Sao Paulo, Brazil

We are investigating effects of the depsipeptide geodiamolide H, isolated from the Brazilian sponge Geodia corticostylifera, on cancer celllines grown in 3D environment. As shown previously geodiamolide H disrupts actin cytoskeleton in both sea urchin eggs and breast cancercell monolayers. We used a normal mammary epithelial cell line MCF 10A that in 3D assay results formation of polarized spheroids. Wealso used cell lines derived from breast tumors with different degrees of differentiation: MCF7 positive for estrogen receptor and theHs578T, negative for hormone receptors. Cells were placed on top of Matrigel. Spheroids obtained from these cultures were treated withgeodiamolide H. Control and treated samples were analyzed by light and confocal microscopy. Geodiamolide H dramatically affected thepoorly differentiated and aggressive Hs578T cell line. The peptide reverted Hs578T malignant phenotype to polarized spheroid-likestructures. MCF7 cells treated by geodiamolide H exhibited polarization compared to controls. Geodiamolide H induced strikingphenotypic modifications in Hs578T cell line and disruption of actin cytoskeleton. We investigated effects of geodiamolide H on migrationand invasion of Hs578T cells. Time-lapse microscopy showed that the peptide inhibited migration of these cells in a dose-dependentmanner. Furthermore invasion assays revealed that geodiamolide H induced a 30% decrease on invasive behavior of Hs578T cells. Ourresults suggest that geodiamolide H inhibits migration and invasion of Hs578T cells probably through modifications in actin cytoskeleton.The fact that normal cell lines were not affected by treatment with geodiamolide H stimulates new studies towards therapeutic use for thispeptide.

J. Cell. Physiol. 216: 583–594, 2008. � 2008 Wiley-Liss, Inc.

This article includes Supplementary Material available from theauthors upon request or via the Internet at http://www.interscience.wiley.com/jpages/0021-9541/suppmat.

Contract grant sponsor: The State of Sao Paulo ResearchFoundation (FAPESP);Contract grant numbers: 06/01026-0, 06/57079-4, 06/59866-3, 06/54963-0, 07/53856-9.Contract grant sponsor: Brazilian Research Council (CNPq);Contract grant numbers: 477890/2004-0, 304868/2006-0.

*Correspondence to: Vanessa M. Freitas, Departamento deBiologia Celular e do Desenvolvimento, Instituto de CienciasBiomedicas, Universidade de Sao Paulo, Av. Prof. Lineu Prestes1524 Ed Biomedicas 1, sala 307, Sao Paulo, SP 05508-900, Brazil.E-mail: [email protected]

Received 2 July 2007; Accepted 1 February 2008

DOI: 10.1002/jcp.21432

Breast tumor derived cell lines grown in three-dimensional(3D) culture conditions recapitulate essential structuralfeatures of the tumor in vivo, illustrating the contrast betweennormal and tumor cells. The 3D reconstituted basementmembrane assay allows epithelial cells to organize themselvesinto structures that resemble their in vivo architecture. Thesimilarities with their original tissue have emerged thesestructures as tractable cell-based models that allowinvestigations of new drugs and therapies (Weaver et al., 1997;Bissell et al., 2002; Wang et al., 2002; Debnath and Brugge,2005).

We have been studying the potential of the cyclicdepsipeptide geodiamolide H as an anti-cancer drug in differentsystems. The peptide was isolated from crude extracts of themarine sponge Geodia corticostylifera from the Brazilian coast.The crude extracts of this sponge exhibit neurotoxic, hemolyticand pore-forming properties in biological and artificialmembranes (Rangel et al., 2005). A previous work from ourgroup (Rangel et al., 2006) described the anti-proliferativeeffects of peptides from G. corticostylifera on sea urchin eggs andhuman breast cancer cell lines cultured in bi-dimensional (2D)assays. Furthermore, using fluorescence techniques andconfocal microscopy we observed that geodiamolide Hdisrupted the actin cytoskeleton of cancer cell lines.

To further explore our previous results obtained in 2Dcultures, we are currently investigating the effect of

� 2 0 0 8 W I L E Y - L I S S , I N C .

geodiamolide H in 3D cultures of mammary cell lines withdifferent degrees of differentiation. Three cell lines were used:(1) a normal mammary epithelial cell line MCF 10A that in 3Dassay results in formation of polarized acini-like spheroids(Debnath et al., 2003); (2) MCF7 cells, characterized by a

584 F R E I T A S E T A L .

non-invasive and rapid growth behavior (Schiemann et al., 1998);and (3) Hs578T cells, representative of invasive and metastaticphenotype of breast cancer (Kirschmann et al., 2002). Light andconfocalmicroscopy showed that this peptide induced importantmodifications in the morphology, polarity and actin cytoskeletonof Hs578T cells cultured in 3D environment. Cells treated by thispeptide showed reversionofmalignant phenotype and clear signsof actin disassembly. Geodiamolide-induced actin disassemblymay have impact on cell motility. We carried out differentexperiments with Hs578T cells, addressing the role played bygeodiamolide H in migration and invasion of this cell line.Time-lapse microscopy and Transwell invasion assaydemonstrated that the peptide decreased migration and invasionof Hs578T cells. Our results suggest that geodiamolide Hdecreases migration and invasion of Hs578T cells probablythrough modifications in the actin cytoskeleton.

Materials and MethodsCell culture

The human mammary cell lines MCF 10A, MCF7 and Hs578T wereobtained from the American Type Culture Collection (Manassas,VA). MCF7 and Hs578T cell line were maintained in Dulbecco’smodified Eagle medium (DMEM) with 10% of fetal bovine serumand penicillin/streptomycin. Culture medium for MCF7 andHs578T cells will be referred throughout the text as ‘‘DMEM.’’ TheMCF 10A was maintained in DMEM/F12 with 5% horse serum,20 ng/ml EGF, 0.5 mg/ml hydrocortisone, 10 mg/ml insulin pluspenicillin/streptomycin. Culture medium for the MCF 10A will bereferred throughout the text as ‘‘supplemented DMEM/F12.’’

Three-dimensional cultures were prepared by growing cells toconfluence as monolayers, followed by harvesting and seeding on asolidified layer of reduced growth factor (RGF) Matrigel measuringapproximately 1–2 mm in thickness. To achieve that a round 13 mmcoverslip coated with 20 ml of RGF-Matrigel (BD Biosciences, SanJose, CA, kindly provided by Dr. Matthew Hoffman, NIDCR, NIH)was placed in each well of a 24-well plate. A 20 ml cell suspensiondrop (106 cells/ml) was placed on top of Matrigel. DMEM orsupplemented DMEM/F12, depending on the cell line, was addedafter cell adhesion to this substrate. Cells were grown in a 5% CO2

humidified incubator at 378C during 10 days. After this period hadelapsed, either DMEM or supplemented DMEM/F12 were replacedby geodiamolide H-containing medium. Concentrations of thepeptide were 20, 120, and 360 nM. After 48 h in contact with thepeptide the cells were fixed and studied by light and confocalmicroscopy. The isolation of the cyclic peptide geodiamolide H wasdescribed elsewhere (Rangel et al., 2006).

Time point analysis was carried out with Hs578T cells treatedwith geodiamolide H. Cells were cultured in Matrigel 3D for 10days followed by treatment with 120 nM of geodiamolide H inculture medium for 15 min, 2, 6, 24, or 48 h. After indicated timepoints cells were fixed, stained with rhodamine-phalloidin andstudied by confocal microscopy.

Three-dimensional cultures were also treated with cytochalasinD. Cells were grown in Matrigel during 10 days as described above.After this period had elapsed, DMEM were replaced bycytochalasin-containing medium (360 nM, Sigma). After 48 h incontact with this actin-disrupting drug the cells were fixed, stainedwith rhodamine-phalloidin and studied by confocal microscopy.

Light microscopy of mammary cells grown inthree-dimensional matrices

Treated and control samples were fixed in 4% paraformaldehyde inPBS for 24 h. Samples were dehydrated and embedded in Histogel(Perk Scientific Inc., Devon, PA; Morais Freitas et al., 2007). SinceHistogel is in aqueous media, the samples were dehydrated again,paraffin-embedded and stained by hematoxylin–eosin (H&E).

JOURNAL OF CELLULAR PHYSIOLOGY

Morphology analysis of 3D cultures by rhodamine-phalloidinstaining and confocal microscopy

Treated and control samples were fixed in 4% paraformaldehyde inPBS during 1 h. To analyze the effect of geodiamolide H in actinfilaments, cells were labeled with rhodamine-phalloidin(Invitrogen-Molecular Probes, Eugene, OR) in a buffer containing0.5% Triton X-100, 1 mg/ml RNase in 2X SSC. Nuclei werecounterstained with Sytox Green (Invitrogen-Molecular Probes).The preparations were mounted on slides with Pro Long(Invitrogen-Molecular Probes). Fluorescence images wereobtained in the Zeiss LSM 510 laser scanning confocal microscope(Carl Zeiss, Jena, Germany).

Polarization and apoptosis analysis by immunofluorescence

Treated and control spheroids from all cell lines were subjected toimmunofluorescence to detect Golgin-97, a polarization marker.Golgin-97 is a trans-Golgi network protein (Kjer-Nielsen et al.,1999; Barr and Short, 2003; Short et al., 2005). The protocol wasadapted from Debnath et al., 2003. Spheroids were fixed in 2%paraformaldehyde in PBS for 20 min, permeabilized with0.5%Triton X-100 (Sigma) for 15 min and blocked with 10% goatserum (KPL, Kirkegaard & Perry Laboratories Inc., Gaithersburg,MD) for 1 h. Antibody against golgin-97 (mouse mAb, clone CDF4,Invitrogen-Molecular Probes) was diluted in 2.5% goat serum,0.01% Tween20 in PBS and incubated overnight. Reactions wererevealed by either anti-mouse Alexa Fluor 568 or Alexa Fluor 488secondary antibodies (Invitrogen-Molecular Probes). Nuclei werecounterstained with either Sytox Green or propidium iodide(Invitrogen-Molecular Probes) and coverslips were mounted withPro Long (Invitrogen-Molecular Probes). Quantitative analysis ofpolarized cells was carried out in confocal equatorial sections ofspheroids from MCF 10A, MCF7 and Hs578T cells. We countedouter cells of 20 spheroids from each group expressing golgin-97characterizing apical orientation of the Golgi apparatus toward thecentral cells. Results with polarized golgin-97 stained cells wereexpressed as a percentage of total outer cells.

Apoptosis was assessed with the M30 CytoDEATH fluorescein-conjugated antibody (mouse monoclonal, clone M30, RocheApplied Science Co., Indianapolis, IN) diluted 1:250 and incubatedfor 2 h at room temperature. This antibody recognizes caspase-cleaved cytokeratin 18. M30 antibody was used to monitorcaspase-3 activation (Leers et al., 1999). Nuclei werecounterstained with propidium iodide (Invitrogen-MolecularProbes), and samples were mounted with Pro Long (Invitrogen-Molecular Probes). Images were obtained with a Zeiss LSM 510laser scanning confocal microscope (Carl Zeiss) and quantitativeanalysis was performed using Image J (public domain softwaredeveloped by Wayne Rasband, NIMH, NIH, http://rsb.info.nih.gov/ij/). Similar stacks of confocal sections were obtained from treatedand control spheroids (n¼ 10). The total pixels from each channel(M30 and nuclear staining) were combined from each section of thestacks. The apoptotic index is a ratio of the total nuclear pixels/totalM30 pixels expressed in arbitrary units.

Scratch wound assay

Migration of Hs578T cells was investigated through the use of astandard in vitro monolayer wound assay (Lee et al., 2000). Cellswere grown in 24-well plates to confluence, and after 24 h ofculture in DMEM the confluent monolayer was gently scraped witha pipette tip to create a cell-free area. Thereafter, the medium waswashed out and replaced by DMEM-containing peptide (10, 20, or120 nM). Cells grown in DMEM served as control. A referencepoint was created on the bottom of the plate in the field of thewound using direct microscopic visualization. This procedurepermitted photographing the identical spot each time. Woundclosure was followed after 0, 4, and 24 h.

B R A Z I L I A N S P O N G E R E G U L A T E S A C T I N C Y T O S K E L E T O N 585

Migration assay in Hs578T cells treated by geodiamolide H

Hs578T cells were harvested from the flasks and plated for2–5 days in 60 mm dishes. Cells were treated with differentconcentrations of geodiamolide H (20, 60, and 120 nM in DMEM).Control cells were grown in DMEM. Cell migration wasinvestigated by time-lapse video microscopy using a computer-assisted microscope workstation consisting of an Axiomatmicroscope (Carl Zeiss) equipped with phase contrast andNomarski (differential interference contrast, DIC) optics. Videorecording was carried out using a black and white camera (DageCCD72, Dage MTI, Michigan City, IN) connected to a computerrunning Axiovision 4.5 image acquisition software (Carl Zeiss).During the experiments Hs578T treated and control cells weremaintained at 378C in a temperature-controlled chamber (CellObserver, Carl Zeiss). The pH was kept by adding 20 mM HEPES tothe culture medium. Phase contrast or DIC images were acquired

Fig. 1. Light microscopy (H&E) analysis shows that geodiamolide H (120peptide induces no effects either in MCF 10A cells (A,B) or MCF7 cells (C,Dshaped cells resembling fibroblasts (E). Geodiamolide H changes the morpcells are treated by the peptide (F). Scale bars: 20 mm.


every 5 min during 10 h. For motility determinations, manualmarking of the centers of cell nuclei were performed. The track ofan individual cell was defined as a sequence of positions of thecenter of its nucleus at different times. Tracking measurements ofsingle cells after different times were analyzed by the MTrackJplugin (written by Erik Meijering, Biomedical Imaging GroupRotterdam, Erasmus MC University Medical Center Rotterdam,the Netherlands). This plugin is a module of Image J, and waswritten to facilitate manual tracking of moving objects in imagesequences and the measurement of traveled distances and instantvelocities. We measured length in micrometers and velocity(mm/h) of Hs578T treated and control cells.

Invasion assay

Invasion assays were carried out in a modified Boyden chamberwith filter inserts (8 mm pores) for 6-well plates (BD Biosciences).

nM) affects the morphology of Hs578T cells grown in Matrigel 3D. The). Hs578T control cells form an arboriform configuration with spindle-

hology of Hs578T cells. Compact spheroids are observed when Hs578T


Filters were coated on ice with 40 ml of Matrigel (10–13 mg/ml).Cells (2� 105) were plated into the upper chamber in 1 mlof DMEM without serum. The lower chamber was filled with 1.5 mlof DMEM. Geodiamolide H was added into the upper chamberimmediately after cell plating at the concentration of 120 nM. After48 h in culture, cells were fixed with 4% paraformaldehyde andpost-fixed with 0.2% crystal violet in 20% methanol. Cells on theupper side of the filter, including those in the Matrigel, wereremoved with a cotton swab. Invading cells, on the lower side of thefilter, were photographed and lysed in 1% of SDS. The intensity ofstaining, reflecting the number of invaded cells, was quantified by

Fig. 2. Confocal microscopy confirms the results obtained with light micoutlinescellboundaries(red),whilenucleiappeargreen.Thepeptide(120nHs578T control cells are stellate (E). Hs578T cells treated by Geodiamolid(F,M) anddotsofdisassembled actin (F,arrowhead)arepresent.Scalebars:at www.interscience.wiley.com.]


absorbance at 600 nm using a microplate reader. Each experimentwas carried out in triplicates, and repeated at least three times.

Statistical analysis

Student’s t-test was carried out to evaluate differences betweentwo groups. Differences between three or more groups wereassessed by ANOVA, followed Bonferroni’s multiple comparisonstest. The software used was Graphpad Prism (GraphPad Software,Inc., San Diego, CA).

roscopy. Rhodamine-phalloidin staining of actin cortical cytoskeletonM) inducesnoeffectseither inMCF10Acells (A,B)orMCF7cells (C,D).e H form spheroid-like structures (F). Areas devoid of actin filaments

20mm.[Colorfigurecanbeviewed intheonline issue, which isavailable


ResultsGeodiamolide H affects the morphology of Hs578T cellsgrown in Matrigel 3D

Light microscopy analysis showed spheroid-like structuresdisplayed by MCF 10A control and geodiamolide-treated cells(Fig. 1A,B). Cells presented acinar differentiation (Fig. 1A), andduct-like structures outlining luminal spaces (Fig. 1B). MCF7control and treated cells showed solid aggregates of epithelioidcells, with neither acinar nor ductal differentiation (Fig. 1C,D).Hs578T control cells formed an arboriform pattern withspindle-shaped cells resembling fibroblasts (Fig. 1E).Geodiamolide H changed the morphology of Hs578T 3D cellcultures to form compact spheroids (Fig. 1F).

Confocal microscope confirmed light microscopy findings.Optical sections showed the spheroid-like structures displayedby MCF 10A control cells (Fig. 2A). In these structures, stainingof the actin cortical cytoskeleton outlined the cell boundaries.Cells from spheroids were polyedrical with nuclei located at

Fig. 3. Geodiamolide H does not affect polarization of MCF 10A cells in 3DControl and geodiamolide-treated MCF 10A cells were immunostained witprotein in outer cells shows polarization toward central cells (A,B, arrowsections. Quantitative analysis showed at least 75% polarized outer cells inobserved (n U 20). In MCF7 cells (D–F) the peptide induced a trend to polaanalysis (F). Geodiamolide H increased polarization of malignant Hs578T ctreated cells are mostly polarized (H, arrowheads). Polarity analysis conficomparedtocontrol (P < 0.05).Nucleiarecounterstainedwitheither Sytoxfigure can be viewed in the online issue, which is available at www.intersc


basal region, suggesting apicobasal polarization. MCF 10A cellstreated by geodiamolide H presented the same characteristics(Fig. 2B). We did not observe modifications in the morphologyof MCF7 cells induced by geodiamolide H. Control and treatedsamples showed similar features, characterized by largespheroids, not as organized as MCF 10A cells (Fig. 2C,D). Onthe other hand Hs578T control cells grown in Matrigel 3Dpresented no spheroids, with formation of structures withmesenchymal-like morphology, composed by stellate cells withlong and thin processes (Fig. 2E). Geodiamolide H dramaticallyaffected the overall architecture of the Hs578T cells (Fig. 2F).Cells treated with the peptide presented features resemblingthe spheroids observed in cell lines derived from either normalor low-malignancy tumors, such as MCF 10A and MCF7,respectively. Furthermore Hs578T cells treated bygeodiamolide H showed signs of actin cytoskeleton disruption,with areas devoid of actin filaments (Fig. 2F, �).

Qualitative and quantitative polarization analysis of MCF10A, MCF7 and Hs578T cells were carried out using a Golgi

culture, but induces polarization of both MCF7 and Hs578T cell lines.h antibodies to a Golgi protein, Golgin-97 (red in A,B). Presence of this

heads). Golgin-97 distribution was analyzed in confocal equatorialboth control and treated spheroids (C). No statistical differences wererization, as shown by Golgin-97 staining (E, arrowheads) and polarityells. Controls cells show a random distribution of Golgin-97 (G), whilerms morphological findings (I). The asterisk indicates significant dataGreen(A,B)orpropidiumiodide(D,E,G,H).Scalebars: 20mm.[Color

ience.wiley.com.]


marker, Golgin-97 (Fig. 3). Both control and treated MCF 10Acells showed a well-polarized outer layer of cells in directcontact with Matrigel, and poorly polarized inner cells lackingcontact with the reconstituted basement membrane (Fig. 3A,B).The outer layer expressed Golgin-97 illustrating the apicalorientation of the Golgi apparatus toward the central cells(Fig. 3A,B). Polarization was quantitatively assessed in controland geodiamolide-treated MCF 10A cells. No differences werefound (Fig. 3C). It is important to emphasize that these resultswere observed after 10 days in culture. By this time is widelyknown that two populations of cells within each spheroidbecome evident: a well-polarized outer layer of cells in directcontact with matrix; and poorly polarized inner cells lackingcontact with the matrix (Debnath et al., 2003). At this timepoint the centrally located, nonpolarized cells begin to die byapoptosis, to start forming a hollow lumen (Debnath et al.,2003). Complete lumen formation is observed after 20 days inculture (Debnath et al., 2003). Thus, our polarization resultswith control and treated MCF 10A cells are in completeagreement with the literature (Weaver et al., 1997; Weaver andBissell, 1999; Bissell et al., 2002; Debnath et al., 2003; Reginatoet al., 2003; Debnath and Brugge, 2005; Reginato andMuthuswamy, 2006; Sequeira et al., 2007).

Fig. 4. Apoptosis is not affected by Geodiamolide H. Staining with the MNuclei are counterstained with propidium iodide. Arrows indicate fragmecells (B,D,F). No differences in apoptosis are observed. Morphological findiApoptotic indexwastakenasdescribedinmaterial inmethods.Nodifferenc20 mm. [Color figure can be viewed in the online issue, which is available


Golgin-97 staining of MCF7 spheroids treated bygeodiamolide H showed part of outer layer cells polarizedcompared to controls (Fig. 3D,E). Polarity measurementssuggested that geodiamolide H increased polarization of thesecells (Fig. 3F). We also observed a trend in polarization ingeodiamolide-treated Hs578T cells as shown by Golgin-97staining and polarity analysis (Fig. 3G–I).

Apoptosis is a common event in normal and neoplasticmammary cells (Weaver et al., 1997; Weaver and Bissell, 1999;Bissell et al., 2002; Debnath et al., 2003; Reginato et al., 2003;Debnath and Brugge, 2005; Reginato and Muthuswamy, 2006;Sequeira et al., 2007). We decided to study this phenomenon incontrol and geodiamolide-treated cells using the anti-M30CytoDEATH antibody. This antibody specifically detectscytokeratin-18 fragments generated by caspases after inductionof apoptosis (Leers et al., 1999; Weiske and Huber, 2006).Apoptosis was assessed in MCF 10A, MCF7, and Hs578Ttreated and control cells grown in 3D environment. Few cellswith fragmentation of cytokeratin-18 were observed in controland treated samples (Fig. 4). MCF 10A and MCF7 control andtreated cells presented cytokeratin-18 fragments only in centralcells of spheroids (Fig. 4A–D). This result was expected, sincethe inner cells of spheroids undergo apoptosis to create luminal

30 CytoDEATH antibody (green) was carried out to assess apoptosis.ntation of cytokeratin-18 in control (A,C,E) and geodiamolide-treatedngs are confirmed by apoptosis assessment (apoptosis analysis panel).esarefoundintreatedandcontrolsamplesfromallcell lines.Scalebars:at www.interscience.wiley.com.]


spaces (Weaver et al., 1997; Weaver and Bissell, 1999; Bissellet al., 2002; Debnath et al., 2003; Reginato et al., 2003; Debnathand Brugge, 2005; Reginato and Muthuswamy, 2006; Sequeiraet al., 2007). Hs578T control and treated cells showedapoptosis in central regions (Fig. 4E,F). Quantitative analysis ofapoptosis showed no difference between control andgeodiamolide-treated cells (Fig. 4, apoptosis analysis panel).

Geodiamolide H affected the morphology of Hs578T cellsgrown in Matrigel 3D. Control cells exhibited stellatephenotype while geodiamolide-treated Hs578T cells formed

Fig. 5. Time point analysis shows that effects of geodiamolide H in morphtreatment. Cells were treated with geodiamolide H as described in mateindicated time points geodiamolide-induced phenotype modifications are oprocesses(B–F), torevertthestellatephenotypeofcontrolcells(A).Adramsuggestive of actin disassembly. Scale bars: 20 mm.


clusters resembling spheroids. Analysis of actin distributionshowed dramatic modification in treated cells compared tocontrols, as shown in Figure 2E,F. Since cytoskeletal effects areoften time-dependent we carried out time point analysis ofHs578T cells treated by geodiamolide H. Cells were treated bythe peptide for 15 min, 2, 6, 24, and 48 h. Results are illustratedin Figure 5. After 15 min of geodiamolide treatment, Hs578Tcells showed modifications of morphology and actindistribution. At this time point the peptide induced retractionof stellate projection and formation of clusters resembling

ology and actin cytoskeleton of Hs578T cells are observed 15 min afterrial and methods. Actin was detected by rhodamine-phalloidin. In allbserved (B–F). Treated Hs578T cells appear to pull back their cellularaticmodificationispresentafter48h(F).ArrowheadinFindicatesdots

Fig. 6. Treatment with either geodiamolide H (120 nM) orcytochalasin D (360 nM) disrupts the actin cytoskeleton of Hs578Tcells grown in Matrigel 3D. Confocal microscopy (Z-projections)of Hs578T control cells grown on Matrigel show stellate cells withthin and long cytoplasmic protrusions (A, arrowhead). Cellstreated with geodiamolide H (B) exhibit spheroid with areas devoidof actin filaments (asterisks in B). Actin is also present as dots(B, arrowhead) and fragments (B, arrow). Hs578T cells treatedwith cytochalasin D present the same morphology of cells treatedwith geodiamolide H. The architecture is spheroid-like (C) withareas devoid of actin (C, M). Dots (C, arrowhead) and fragments(C, arrow) of disassembled actin are noteworthy. Scale bars: 20 mm.[Color figure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]


spheroids (Fig. 5B). These phenotype alterations were alsoobserved after 2, 6, 24, and 48 h of treatment by geodiamolide H(Fig. 5C–F). In all time point dots and fragments of disruptedactin were observed (Fig. 5B–F). MCF 10A and MCF7 controland treated cells showed no differences in morphology andactin distribution at the same time points (Supplementary Figs. 1and 2).

Treatment with either geodiamolide H or cytochalasin Ddisrupts the actin cytoskeleton of Hs578T cells grown inMatrigel 3D

The results obtained by treating Hs578T cells withgeodiamolide H were compared to the effect of cytochalasin Din the same cell line. The rationale of this experiment was toobserve whether phenotype modifications induced bygeodiamolide H would be similar to that observed for a well-known actin-disrupting drug (Flanagan and Lin, 1980).

Optical sections obtained with confocal microscopy ofHs578T control cells grown on Matrigel showed its typicalstellate aspect, with thin and long cytoplasmic protrusions(Fig. 6A). Cells treated with geodiamolide H (Fig. 6B), exhibitedspheroid aggregates with no protrusions. These spheroidsdepicted large areas devoid of actin filaments (asterisks inFig. 6B). Hs578T cells treated with cytochalasin D presentedthe same morphology of cells treated with geodiamolide H(Fig. 6C). The overall architecture was spheroid-like, free ofprotrusions, and with areas devoid of actin (Fig. 6C, �).Fragments of F-actin were also present (Fig. 6C, arrow). Ourresults strongly suggest that geodiamolide H disrupt actincytoskeleton of Hs578T cells.

Geodiamolide H decreases migration and invasion ofHs578T cells

Peptide-induced actin disassembly may impair cell motility. Wecarried out different assays with Hs578T cells, focusing on therole played by geodiamolide H in migration and invasion of thiscell line.

To explore whether geodiamolide H affects Hs578T cellsmigration, we performed an in vitro scratch-wound closureassay. Cell monolayers were gently scraped with a pipette tip tocreate a cell-free area. Closure of this wounded cell-free areawas followed after 0 and 24 h. The cell-free wound gap ofHs578T control monolayers was nearly closed after 24 h(Fig. 7A,B). However, in the presence of geodiamolide H, theclosure of the wounded area was impaired (Fig. 7C,D). Lesscells appeared in the wounded gap, which representeddecreased migratory activity towards the wounded area.

The remaining cell-free area at 0–24 h as a percentage of theinitial wound area was taken as an index of wound healing(Fig. 7E). In control Hs578T cells, only 20% of the wound arearemained cell-free at 24 h after wounding. However, in cellstreated by different concentrations of geodiamolide H (10, 20,120 nM) the cell-free area increased to at least 75%. Similarresults were obtained in at least three experiments.

Time-lapse images showed that cells treated withgeodiamolide H decreased the velocity compared to controlcells. Figure 8A–D shows cells trajectories obtained in control(Fig. 8B) and treated (Fig. 8C,D) Hs578T cells. Treated cellspresented shorter trajectories over time compared to controls(Fig. 8B–D). The decrease was significant and dose dependent,as shown in Figure 8E. Control cells moved at 23 mm/h, whilecells treated with 120 nM of Geodiamolide H traveled at13 mm/h (Fig. 8E).

We completed our analysis by using the invasion assay inMatrigel-coated Boyden chambers. Invasion rate ofgeodiamolide-treated cells was 30% smaller than control cells(Fig. 9A,B). Thus geodiamolide H inhibits invasion of Hs578Tcells.


Fig. 7. Scratch-wound closure assay shows that geodiamolide H decreases migration in Hs578T cells. Phase contrast microscopy shows that thecell-free wound gap of Hs578T control monolayers is nearly closed after 24 h (A,B). In the presence of geodiamolide H the closure of the woundedarea is impaired(C,D). Incells treatedbydifferentconcentrations of thegeodiamolideH(10,20,120nM) thecell-freearea increases toat least 75%compared to control (20%) (E). Asterisk indicates significant data compared to control (P < 0.001). Results in (E) represent mean W standard errorof three experiments carried out at least three times. Scale bar in (D): 50 mm.


Discussion

We tested the effect of depsipeptide geodiamolide H, derivedfrom a Brazilian marine sponge, in mammary cells spheroidsobtained from cells grown in a three-dimensional environmentof Matrigel. In vivo, tumor cells in the original tissue areinfluenced by different elements that control gene expressionsuch as soluble factors, extracellular matrix and complex cell–cell interactions. Cultured cells can lose some of their inherentin vivo properties when separated from these elements


(Simone et al., 2000). Three-dimensional cultures consist ofcells aggregated in spheroids that have an intermediatecomplexity between in vivo tumors and in vitro monolayers(Debnath and Brugge, 2005). It has been demonstratedpreviously that spheroids represent a more adequate model tostudy some aspects of tumor biology than monolayer cultures.Previous spheroid studies have highlighted the importance ofcell–cell interactions in tumor biology as several tumorcharacteristics such as growth, metastasis and resistance toantitumor agents are dictated by the collective properties of a

Fig. 8. Time-lapse video microscopy confirms the results obtained with scratch-wound closure assay. A DIC image with Hs578T cell trajectoriessuperimposed (white tracks) illustrates the manual marking of the same cell at different time points. Treated cells (C,D) present shortertrajectories over time compared to controls (B). The decrease in migration induced by Geodiamolide H is significant and dose dependent (E). Themean velocity of control cells is 23 mm/h while the mean velocity of cells treated with 120 nM the Geodiamolide H is 13 mm/h (E). The asteriskindicates significant data compared to control (P < 0.05). Results in (E) represent mean W standard error of three experiments carried out at leastthree times. A number of at least 10 cells was observed in each experiment. Scale bar in (A): 100 mm.


cell population rather than those of a single cell(Kunz-Schughart, 1999; Mayer et al., 2001). Spheroids, byrecapitulating some of the morphological and functionalfeatures of the original tissue, have been used in studiesinvolving drug penetration, therapeutic macromolecules,multicellular drug resistance, cell–cell interactions,angiogenesis and other aspects of tumor biology (Zietarskaet al., 2007).

Geodiamolide H did not affect overall morphology and actincytoskeleton of MCF 10A and MCF7 cell lines. Our previousresults with monolayer culture showed Geodiamolide-induceddisorganization of actin filaments in MCF7 cells (Rangel et al.,2006). Result differences are probably caused by cultureconditions (2D vs. 3D) since peptide concentration was similarin both monolayer and 3D experiments.

MCF 10A is an immortalized non-transformed humanmammary epithelial cell line (Debnath et al., 2003), while MCF7cells are representative of rapidly growing non-invasivemammary phenotype (Schiemann et al., 1998). Geodiamolide H


did not affect morphology, polarization and actin organizationof MCF 10A cells. MCF7 cells were slightly influenced by thispeptide, since we observed a significant trend in polarization ingeodiamolide-treated cells. On the other hand this peptideshowed a dramatic effect in Hs578T cells. This cell line isderived from an invasive and metastatic tumor cell that expressvimentin and lack E-cadherin (Kirschmann et al., 2002). Thus,according with the present data Geodiamolide H influencedonly neoplastic cell lines. This selectivity could be importantconsidering this peptide as a possible anticancer drug.

Geodiamolide H induced reversion of malignant phenotypeand cytoskeletal alterations in Hs578T cells. We searched foractin modifications induced by geodiamolide H in Hs578T cells.Previous work from our group showed modifications in theactin cytoskeleton induced by G. corticostylifera depsipeptides inbreast cancer cells cultured as monolayers (Rangel et al., 2006).Present results showed that geodiamolide H affected actincytoskeleton of a malignant cell line derived from mammarycancer grown in a three-dimensional environment of Matrigel.

Fig. 9. Geodiamolide H significantly decreases invasion of Hs578Tcells. Invasion assays were carried out in Matrigel-coated Boydenchambers. Invading cells from control and treated cells are stainedwith crystal violet (A). The invasion rate of Hs578T cells treated withgeodiamolide H is 30% smaller than control cells. The asterisk in Bindicates significantdata compared tocontrol (P < 0.05). Results in (B)represent mean W standard error of three experiments carried out atleast three times. Scale bar in A: 50 mm.


The results obtained with this peptide were similar to thoseobtained with cytochalasin D, a drug which disrupts actinfilaments, leaving other cytoskeletal components unaffected(Flanagan and Lin, 1980). Modifications in morphology and actinorganization may target apoptosis (Bubb et al., 1994; Bubb et al.,2000). However, we found no differences in apoptosis incontrol and treated samples from all cell lines studied.

Actin disassembly in Hs578T cells would impair cell motility.This result prompted us to study the role played bygeodiamolide H on migration and invasion of Hs578T cells. Cellmigration was evaluated by time-lapse video-microscopy and byscratch-wound assay. We observed the impact of geodiamolideH on individual cell motility. Hs578T cells treated by thispeptide significantly decreased locomotion compared tonon-treated cells. We have also investigated the effect of thegeodiamolide H in invasion. We used Matrigel-coatedTranswell filters and observed the movement of Hs578T cellsthrough Matrigel.

Our results showed that the peptide significantly decreasedmigration and invasion of Hs578T cells probably due tomodifications in actin cytoskeleton. Geodiamolide H could berelated to other mechanisms by which malignant cells invadeand spread. To invade, cancer cells penetrate the basementmembrane with the help of matrix metalloproteinases (MMPs)that degrade collagen type IV and other elements of this matrix(McCawley and Matrisian, 2000). We predicted thatgeodiamolide may affect both actin cytoskeleton and MMPsecretion, leading to invasion impairment. However, MMPanalysis by zymography and gel densitometry demonstratedthat geodiamolide H induced no effect in protease activity ofHs578T cells compared to controls (Supplementary Fig. 3). Ourresults indicate that actin is the major target of geodiamolide Hin Hs578T cells, with evident importance in cell motility.


Actin cytoskeleton is a key component involved in cellmigration. Malignant cancer cells utilize their intrinsic migratoryability to invade adjacent tissues and the vasculature, andultimately to metastasize (Yamaguchi and Condeelis, 2007).The spread of cancer cells to distant sites in the body is themajor cause of cancer patient death (Entschladen et al., 2004). Itis a major challenge in cancer therapy to inhibit the spreading oftumor cells from primary tumor sites to other organs (Hayotet al., 2006). Cell migration occurs by membrane protrusion(either lamellipodia or filopodia), adhesion to extracellularmatrix, cell body translocation, and tail retraction. All theseprocesses imply actin dynamics, which is controlled byhundreds of proteins both inside and outside the cells (Gigantiand Friederich, 2003; Rao and Li, 2004). Migration and invasionare closely related words in cancer research. The differencebetween invasion and migration lays in the barriers that cellsneed to cross to go from one site to another (Mareel and Leroy,2003).

Various products have been shown to have anti-migratoryeffects, and some of them have also been shown to be able toprevent tumor metastasis. Most of their migration-inhibitingeffects are based either on interference with tumor adhesion toextracellular matrix components or on reduction of tumor cell-associated protease activity (Decaestecker et al., 2007). Inparticular, whereas therapeutic agents targeting actin dynamicshave been relatively little investigated in the literature, actindynamics play a number of essential roles in the spread of tumorcells and constitute a potential target of exceptional interest foranti-cancer drug development (Decaestecker et al., 2007).

In contrast to the microtubules, which have been targetedsuccessfully with anti-tumor drugs such as taxol-likecompounds and the vinca alkaloids, very few actin targetingdrugs have been characterized. To date, no actin targeting drugshave been used in clinical trials due to their severe cytotoxicity.One reason for this cytotoxicity is that drugs such as thecytochalasins and latrunculins disrupt actin microfilaments inboth non-tumor and tumor cells (Stehn et al., 2006). Ourresults show that geodiamolide H selectively interferes withactin cytoskeleton of tumor cells (Hs578T), leaving normal cells(MCF 10A) unaffected. This marine natural product has greatpotential to be developed as therapeutic agent towards cancercells.

Acknowledgments

We are grateful to Dr. Matthew P. Hoffman (NIDCR, NIH) forvaluable suggestions on the manuscript, and to Roberto CabadoModia for technical expertise with confocal analysis. V.M. Freitas is recipient of a Post Doctoral fellowship fromFAPESP (06/54963-0) and L.F. Bisson is recipient of anundergraduate fellowship from FAPESP (07/53856-9). Thiswork was supported by The State of Sao Paulo ResearchFoundation (FAPESP) grants #2006/01026-0 (to G.M.Machado-Santelli), # 2006/57079-4 (to R.G. Jaeger), # 2006/59866-3 (to V.M. Freitas), and Brazilian Research Council(CNPq) grants # 477890/2004-0 (to G.M. Machado-Santelli)and # 304868/2006-0 (to R.G. Jaeger).

Literature Cited

Barr FA, Short B. 2003. Golgins in the structure and dynamics of the Golgi apparatus. CurrOpin Cell Biol 15:405–413.

Bissell MJ, Radisky DC, Rizki A, Weaver VM, Petersen OW. 2002. The organizing principle:Microenvironmental influences in the normal and malignant breast. Differentiation70:537–546.

Bubb MR, Senderowicz AM, Sausville EA, Duncan KL, Korn ED. 1994. Jasplakinolide, acytotoxic natural product, induces actin polymerization and competitively inhibits thebinding of phalloidin to F-actin. J Biol Chem 269:14869–14871.

Bubb MR, Spector I, Beyer BB, Fosen KM. 2000. Effects of jasplakinolide on the kinetics ofactin polymerization. An explanation for certain in vivo observations. J Biol Chem275:5163–5170.

Debnath J, Brugge JS. 2005. Modelling glandular epithelial cancers in three-dimensionalcultures. Nat Rev Cancer 5:675–688.


Debnath J, Muthuswamy SK, Brugge JS. 2003. Morphogenesis and oncogenesis of MCF 10Amammary epithelial acini grown in three-dimensional basement membrane cultures.Methods 30:256–268.

Decaestecker C, Debeir O, Van Ham P, Kiss R. 2007. Can anti-migratory drugs be screened invitro? A review of 2D and 3D assays for the quantitative analysis of cell migration. Med ResRev 27:149–176.

Entschladen F, Drell TLt, Lang K, Joseph J, Zaenker KS. 2004. Tumour-cell migration, invasion,and metastasis: Navigation by neurotransmitters. Lancet Oncol 5:254–258.

Flanagan MD, Lin S. 1980. Cytochalasins block actin filament elongation by binding to highaffinity sites associated with F-actin. J Biol Chem 255:835–838.

Giganti A, Friederich E. 2003. The actin cytoskeleton as a therapeutic target: State of the artand future directions. Prog Cell Cycle Res 5:511–525.

Hayot C, Debeir O, Van Ham P, Van Damme M, Kiss R, Decaestecker C. 2006.Characterization of the activities of actin-affecting drugs on tumor cell migration. ToxicolAppl Pharmacol 211:30–40.

Kirschmann DA, Seftor EA, Fong SF, Nieva DR, Sullivan CM, Edwards EM, Sommer P, CsiszarK, Hendrix MJ. 2002. A molecular role for lysyl oxidase in breast cancer invasion. CancerRes 62:4478–4483.

Kjer-Nielsen L, Teasdale RD, van Vliet C, Gleeson PA. 1999. A novel Golgi-localisationdomain shared by a class of coiled-coil peripheral membrane proteins. Curr Biol 9:385–388.

Kunz-Schughart LA. 1999. Multicellular tumor spheroids: Intermediates between monolayerculture and in vivo tumor. Cell Biol Int 23:157–161.

Lee H, Goetzl EJ, An S. 2000. Lysophosphatidic acid and sphingosine 1-phosphate stimulateendothelial cell wound healing. Am J Physiol Cell Physiol 278:C612–C618.

LeersMP,KolgenW,BjorklundV,BergmanT,TribbickG,PerssonB,BjorklundP,RamaekersFC,Bjorklund B, Nap M, Jornvall H, Schutte B. 1999. Immunocytochemical detection and mappingof a cytokeratin 18 neo-epitope exposed during early apoptosis. J Pathol 187:567–572.

Mareel M, Leroy A. 2003. Clinical, cellular, and molecular aspects of cancer invasion. PhysiolRev 83:337–376.

Mayer B, Klement G, Kaneko M, Man S, Jothy S, Rak J, Kerbel RS. 2001. Multicellular gastriccancer spheroids recapitulate growth pattern and differentiation phenotype of humangastric carcinomas. Gastroenterology 121:839–852.

McCawley LJ, Matrisian LM. 2000. Matrix metalloproteinases: Multifunctional contributors totumor progression. Mol Med Today 6:149–156.

Morais Freitas V, Nogueira da Gama de Souza L, Cyreno Oliveira E, Furuse C, Cavalcanti deAraujo V, Gastaldoni Jaeger R. 2007. Malignancy-related 67 kDa laminin receptor inadenoid cystic carcinoma. Effect on migration and beta-catenin expression. Oral Oncol43:987–998.

Rangel M, Konno K, Brunaldi K, Procopio J, De Freitas JC. 2005. Neurotoxic activity inducedby a haemolytic substance in the extract of the marine sponge Geodia corticostylifera. CompBiochem Physiol C Toxicol Pharmacol 141:207–215.


Rangel M, Prado MP, Konno K, Naoki H, Freitas JC, Machado-Santelli GM. 2006.Cytoskeleton alterations induced by Geodia corticostylifera depsipeptides in breast cancercells. Peptides 27:2047–2057.

Rao J, Li N. 2004. Microfilament actin remodeling as a potential target for cancer drugdevelopment. Curr Cancer Drug Targets 4:345–354.

Reginato MJ, Muthuswamy SK. 2006. Illuminating the center: Mechanisms regulating lumenformation and maintenance in mammary morphogenesis. J Mammary Gland Biol Neoplasia11:205–211.

Reginato MJ, Mills KR, Paulus JK, Lynch DK, Sgroi DC, Debnath J, Muthuswamy SK, Brugge JS.2003. Integrins and EGFR coordinately regulate the pro-apoptotic protein Bim to preventanoikis. Nat Cell Biol 5:733–740.

Schiemann S, Schwirzke M, Brunner N, Weidle UH. 1998. Molecular analysis of twomammary carcinoma cell lines at the transcriptional level as a model system forprogression of breast cancer. Clin Exp Metastasis 16:129–139.

Sequeira SJ, Ranganathan AC, Adam AP, Iglesias BV, Farias EF, Aguirre-Ghiso JA. 2007.Inhibition of proliferation by PERK regulates mammary acinar morphogenesis and tumorformation. PLoS ONE 2:e615.

Short B, Haas A, Barr FA. 2005. Golgins and GTPases, giving identity and structure to theGolgi apparatus. Biochim Biophys Acta 1744:383–395.

Simone NL, Paweletz CP, Charboneau L, Petricoin EF III, Liotta LA. 2000. Laser capturemicrodissection: Beyond functional genomics to proteomics. Mol Diagn5:301–307.

Stehn JR, Schevzov G, O’Neill GM, Gunning PW. 2006. Specialisation of the tropomyosincomposition of actin filaments provides new potential targets for chemotherapy. CurrCancer Drug Targets 6:245–256.

Wang F, Hansen RK, Radisky D, Yoneda T, Barcellos-Hoff MH, Petersen OW,Turley EA, Bissell MJ. 2002. Phenotypic reversion or death of cancer cells by alteringsignaling pathways in three-dimensional contexts. J Natl Cancer Inst94:1494–1503.

Weaver VM, Bissell MJ. 1999. Functional culture models to study mechanisms governingapoptosis in normal and malignant mammary epithelial cells. J Mammary Gland BiolNeoplasia 4:193–201.

Weaver VM, Petersen OW, Wang F, Larabell CA, Briand P, Damsky C, Bissell MJ. 1997.Reversion of the malignant phenotype of human breast cells in three-dimensional cultureand in vivo by integrin blocking antibodies. J Cell Biol 137:231–245.

Weiske J, Huber O. 2006. The histidine triad protein Hint1 triggers apoptosis independent ofits enzymatic activity. J Biol Chem 281:27356–27366.

Yamaguchi H, Condeelis J. 2007. Regulation of the actin cytoskeleton in cancer cell migrationand invasion. Biochim Biophys Acta 1773:642–652.

Zietarska M, Maugard CM, Filali-Mouhim A, Alam-Fahmy M, Tonin PN, Provencher DM,Mes-Masson AM. 2007. Molecular description of a 3D in vitro model for the study ofepithelial ovarian cancer (EOC). Mol Carcinog 46:872–885.

Documents

Cellular The Geodiamolide H, Derived From Brazilian Sponge ... · 0.5% Triton X-100, 1 mg/ml RNase in 2X SSC. Nuclei were counterstained with Sytox Green (Invitrogen-Molecular Probes)