proteolysis of extracellular matrix by invadopodiafacilitates human breast cancer cell invasion and...

Introduction

Invasive tumor cells have increased levels of avariety of extracellular matrix-degrading proteasesthat allow them to traverse complex basementmembrane and stromal matrices (reviewed in [1–3]).Generally, the matrix-degrading proteases aresecreted by tumor and stromal cells as inactivezymogens that require association with the tumorcell surface to become activated and capable ofdegrading extracellular matrix [1,2]. The cellularsites of abnormally high extracellular matrix-degrading activity can be investigated by growing

invasive cells on monolayers of fluorescently labeledextracellular matrix proteins. Fluorescence-negativeregions underneath the invasive cells are observedthat indicate the positions where matrix proteolysishas occurred [4,5]. The proteolytic activity isdiscretely focused and corresponds to areas whereplasma membrane protrusions termed ‘invadopodia’extend from cell surfaces and contact the matrix [4, 6]. Invadopodia can degrade multiple extracel-lular matrix proteins including intact fibronectin-rich matrices produced by fibroblasts, fibronectin,laminin, type IV collagen, and type I collagen [5].A variety of invasive cells exhibit invadopodiaincluding human melanoma cells [7], transformedchicken fibroblasts [4–6], and human breast cancercells [8,9]. Invadopodial proteolysis of extracellularmatrix is thought to facilitate tumor cell invasioninto extracellular matrix [3,6,10].

111112345678910111123456789201111234567893011112345678940111123456789501111234111

Clin. Exp. Metastasis, 1998, 16, 501–512

© 1998 Kluwer Academic Publishers Clinical & Experimental Metastasis Vol 16 No 6 501

Proteolysis of extracellular matrix by invadopodiafacilitates human breast cancer cell invasion and ismediated by matrix metalloproteinases

Thomas Kelly, Yan Yan, Rebecca L. Osborne, Anupama B. Athota, Tricia L. Rozypal, J. Chris Colclasure and Winston S. Chu

Department of Pathology, Arkansas Cancer Research Center, University of Arkansas for MedicalSciences, Little Rock, AR, USA

(Received 6 February 1998; accepted in revised form 13 March 1998)

Breast cancer cell lines vary in invasive behavior and one highly invasive cell line (MDA-MB-231) proteolytically degrades extracellular matrix with invadopodia (Thompson et al. 1992, J Cell Physiol, 150,534–44; Chen et al. 1994, Breast Cancer Res Treat, 31, 217–26). Invadopodial proteolysis of extracellularmatrix is thought to be necessary for invasion; however, this has not been demonstrated directly. To obtainsuch evidence, normal (HBL-100) and malignant (MCF-7, MDA-MB-231) breast cells were evaluated forinvadopodial proteolysis of extracellular matrix and invasive behavior. We report that invadopodial prote-olysis of immobilized fibronectin is positively correlated with invasion of cells into type I collagen gels.Moreover, reducing the proteolytic activity of invadopodia with the metalloproteinase inhibitor, batimastat(BB-94), also decreases invasion indicating that breast cancer cell invasion is dependent upon proteolyti-cally active invadopodia.

Keywords: (BB-94) batimastat, fibronectin, invadopodia, invasion, metastasis, MMP

Address correspondence to: T. Kelly, Department of Pathology,Slot 753, Arkansas Cancer Research Center, University ofArkansas for Medical Sciences, 4301 W. Markham St., LittleRock, AR 72205–7199, USA. Tel: (+1) 501 686 6401; Fax: (+1)501 686 6517; E-mail: [email protected].

Matrix metalloproteinases (MMPs) are importantmediators of invadopodial degradation of extracellu-lar matrix because an inhibitor of MMPs (NP-20)decreases invadopodial degradation of type I colla-gen by Rous sarcoma virus-transformed chickenembryo fibroblasts [11]. MMPs are a family of struc-turally related enzymes that together can degrade all components of the extracellular matrix and are known to be important in tumor cell invasion [1,2]. MMPs are synthesized as inactive proenzymesthat require cleavage of the pro-peptide for activa-tion of their proteolytic activities. Proteolyticallyactive MMP-2 (gelatinase A) and MT1-MMP (mem-brane type 1-matrix metalloproteinase or MMP-14)are concentrated on invadopodial membranes sug-gesting that these enzymes are important mediatorsof invadopodial extracellular matrix degradation[11,12]. MT1-MMP is activated by furin-like enzymesprior to its expression on the cell surface [13,14]where it serves as a potent activator of latent MMP-2 [15,16]. Given the wide range of extracellularmatrix substrates degraded by invadopodia, otherproteases including other MMPs are likely to have arole in proteolysis of extracellular matrix byinvadopodia.

This study was performed to determine ifinvadopodial proteolysis of extracellular matrix playsan important role in facilitating human breast cellinvasion. Previously, it was shown that the invasiveMDA-MB-231 human breast cancer cells useinvadopodia to degrade films of fluorescent extracel-lular matrix molecules covalently attached to glutaraldehyde crosslinked gelatin [8,9]. Inde-pendently, others have shown that human breast can-cer cell lines vary in their ability to invade intoMatrigel and that human breast cancer cell lines varyin local invasion through the peritoneum wheninjected into the mammary fat pads of nude mice[17]. MDA-MB-231 cells were identified as invasivein both assays [8,17]. In contrast, MCF-7 humanbreast cancer cells were moderately invasive throughMatrigel in Boyden chamber assays and, althoughthese cells formed tumors when injected into themammary fat pads of nude mice, the tumor cells didnot invade through the peritoneum [17]. Althoughinvadopodial proteolysis of extracellular matrix andinvasiveness have been assessed separately, there hasnot been a direct evaluation of the role of invadopo-dial proteolysis of extracellular matrix in invasion.Here we correlate the HBL-100, MCF-7, and MDA-MB-231 human breast cell lines for invadopodialfunction as determined by degradation of fluores-cently labeled or radiolabeled fibronectin and forinvasiveness using type I collagen gels. The MMP

inhibitor batimastat was used to inhibit the functionof invadopodia to thereby investigate the role ofinvadopodia in breast cancer cell invasion. We showthat invadopodial proteolysis of extracellular matrixfacilitates human breast cancer cell invasion and ismediated by MMPs.

Materials and methods

Cell cultureHuman breast adenocarcinoma cell lines MDA-MB-231, MCF-7, and the normal breast cell line HBL-100 were obtained from American Type CultureCollection (Rockville, MD). All cell lines were main-tained in Eagle’s minimal essential medium (Gibco-BRL, Gaithersburg, MD), supplemented with 10%heat-inactivated fetal bovine serum (FBS; AtlantaBiologicals, Norcross, GA), 10 mg/ml bovine insulin(Sigma, St. Louis, MO), 10 mg/ml glutamine (Gibco-BRL, Gaithersburg, MD), 1% penicillin-strepto-mycin (Gibco-BRL) 2 × 10–8 mM b-estradiol (Sigma)and maintained at 37oC in 95% air and 5% CO2.The FBS used for growing cells was either complete(Figure 4), depleted of MMPs and tissue inhibitorsof metalloproteinases (TIMPs; Figures 1 and 3A) ordepleted of MMPs, TIMPs and plasmin/plasminogen(Table 1, Figures 2, 3B and 5) as described below.

Fluorescent fibronectin substrates for determininginvadopodial proteolysis of extracellular matrixThis assay relies on growing cells on fluoresceinisothiocyanate-labeled fibronectin that is covalentlybound to a glutaraldehyde-crosslinked gelatin filmattached to a glass coverslip. Fluorescence-negativeregions underneath cells represent areas where theextracellular matrix has been degraded [4,5]. Humanplasma fibronectin (Becton Dickinson Labware,Medford, MA) was coupled to fluorescein isothio-cyanate (Research Organics, Cleveland, OH)according to the manufacturer’s instructions. Theassays reported here were allowed to proceed for 24or 72 h at 37oC in 95% air and 5% CO2 prior to fixa-tion and preparation for fluorescence microscopy.

Fluorescence microscopyThe cells were fixed, stained with rhodamine phal-loidin (Molecular Probes, Inc., Eugene, OR) asdescribed earlier [5]. The cells were observed using × 40, × 60 and × 100 objectives of a NikonLABOPHOT microscope equipped for epifluores-cence photomicroscopy and images were recordedusing Ilford HP5 400 ASA black and white film. To quantify invadopodial matrix proteolysis, 10 micro-

111112345678910111123456789201111234567893011112345678940111123456789501111234111

T. Kelly et al.

502 Clinical & Experimental Metastasis Vol 16 No 6

scopic fields on each coverslip were randomly selectedand the fluorescence of the fibronectin film visualizedwith the × 40 objective and scored for the presence orabsence of matrix proteolysis as judged by the pres-ence of focal fluorescence negative spots in areaswhere the matrix was degraded [4,5,8]. The results arepresented as the average number of fields positive formatrix degradation per 10 fields examined.

[125I]fibronectin substrates for determining invadopo-dial proteolysis of extracellular matrixThis second invadopodial proteolysis assay utilized[125I]fibronectin as the substrate and protease-depleted FBS in the growth medium. The assayinvolves: iodination of the fibronectin, coupling the[125I]fibronectin to crosslinked gelatin films, deter-mination of [125I]fibronectin bound to the substrate,extensive washing to remove free [125I]fibronectin,and determination of invadopodial proteolysis.

Removal of interfering extracellular matrix-degrad-ing proteases from FBSGelatin Sepharose chromatography was used toremove MMPs and TIMPs from FBS [18,19]. A 10 mlgelatin-Sepharose (Pharmacia Biotech, Uppsala,Sweden) was equilibrated with the binding buffer20 mM Tris, 0.5 M NaCl, 1.0 mM CaCl2, 10% glycerol(v/v), 0.05% BRIJ-35 (v/v), 0.02% NaN3 (v/v), pH7.6. MMPs and TIMPs in FBS (heat inactivated)were bound to the column by loading the serum(50 ml) onto the column at 35 ml/h. The void volumewas discarded, and the flow-through, depleted ofMMPs and TIMPs but containing the other serumcomponents, was collected and sterile filtered.Gelatin zymography of breast cancer cell growthmedium made 10% (v/v) with respect to FBSrevealed no MMP activity (not shown).

For the batimastat inhibitor studies (batimastatkindly provided by British Biotech Pharmaceuticals,Ltd, Oxford, UK), the FBS was depleted of MMPs,TIMPs, and the broad spectrum protease, plasmin/plasminogen, that is abundant in serum. MMPs andTIMPs were removed as described and then lysineSepharose was used to remove the plasmin/plasmino-gen [20]. A 10 ml lysine Sepharose column(Pharmacia Biotech) was packed and equilibratedwith the same binding buffer as described above. TheMMP/TIMP-depleted FBS (50 ml) was loaded ontothe column at 35 ml/h. The MMP/TIMP and plas-minogen/plasmin-depleted serum in the flow-throughwas collected and sterile filtered. Immunoblot analy-sis of breast cancer cell growth medium made 10%(v/v) with respect to MMP/TIMP and plasminogen/plasmin-depleted FBS revealed no plasminogen or

plasmin immunoreactivity with a goat IgG directedagainst bovine plasminogen (American Diagnostics,Greenwich, CT) (not shown).

Iodination of fibronectin, coupling to crosslinkedgelatin and determination of cpm [125I]fibronectinbound to the substrateFibronectin (50 mg) was iodinated with 0.5 mCi 125Iusing chloramine T (2 mg/ml) as described [21]. [125I]-fibronectin was separated from free 125I by gel filtra-tion using an extracellulose G5 column (Pierce,Rockford, IL) that had been equilibrated in 1 mg/mlbovine serum albumin (BSA) and phosphate bufferedserum (PBS). An equivalent amount of [125I]-fibronectin in 200 ml was coupled to the glutaralde-hyde crosslinked gelatin film coating 15 mm glassround coverslips as previously described for fluores-cent fibronectin [4,5,22]. The level of [125I]fibronectinused in independent experiments varied from 2 × 106

cpm to 1010 cpm. The 125I-labeled fibronectin wasallowed to bind to the coverslips for at least 12 h. Atthe end of coupling, 1 ml of the coupling fluid wascounted and the volume of coupling fluid was mea-sured to determine the amount of [125I]fibronectin thatdid and did not bind to each coverslip.

The [125I]fibronectin-coupled crosslinked gelatinfilms were subjected to a series of washes designedto remove any free or weakly bound [125I]fibronectinand to block exposed aldehyde groups. The washesincluded 3 ml of the following solutions: 70%ethanol (1 × 5 min, 22oC), PBS (3 × 5 min, 22oC),growth medium containing 10% MMP/TIMP andplasminogen/plasmin-depleted FBS (1 × 3 h, 37oC),PBS (2 × 5 min, 22oC), and serum-free growthmedium (1 × 24 h, 37oC). A sample (10 ml) was takento determine the level of [125I]fibronectin in eachwash. The total [125I]fibronectin that did not bindeach coverslip was determined by adding theradioactivity (cpm) in the unbound fraction to thesum of the radioactivity (cpm) released by all of thewashes. Then the amount of [125I]fibronectin boundto each coverslip was determined by subtracting thetotal unbound radioactivity from the radioactivityoriginally added to the coverslip.

Determination of invadopodial matrix proteolysisusing immobilized [125I]fibronectinHuman breast cells were harvested using trypsin–EDTA, diluted with growth medium containing 10%protease-depleted FBS and washed three times withsterile PBS. Coverslips were placed in the wells of a6-well culture plate. Cells (1 × 105) were placed ontoeach coverslip in 200 ml of medium and allowed toattach to the [125I]fibronectin for 1 h at 37oC. Then

111112345678910111123456789201111234567893011112345678940111123456789501111234111

Invadopodial MMPs facilitate invasive behavior

Clinical & Experimental Metastasis Vol 16 No 6 503

2.8 ml of growth medium containing 10% MMP/TIMP and plasminogen/plasmin-depleted FBS con-taining no additives, DMSO, or 10 mM batimastat inDMSO was added to achieve a final volume of 3 ml.The plates were placed back into the incubator and50 ml aliquots were taken from each well under ster-ile conditions and counted at 3, 24, 48, 72 and 96 h. Ateach time point, 50 ml of growth media containing 10 % protease-depleted FBS were put back into eachwell to maintain the 3 ml volume of the assay. Theradioactivity of the aliquots was determined with aPackard gamma counter.

The total amount of 125I released into the mediafor each time point was calculated and added to thesum of the radioactivity in the 50 ml aliquots fromprior time points. This value was divided by the totalcpm [125I]fibronectin bound to that coverslip andmultiplied by 100 to give the percentage of totalbound 125I released into the media.

Cell viabilitiesThe cytotoxicity of 72 h exposure to batimastat andDMSO was investigated by performing trypan blueexclusion assays as described by others [23].

Gelatin zymographyGelatin zymography was performed essentially asdescribed by Heussen and Dowdle [24] with 1 mg/mlgelatin co-polymerized into SDS–PAGE that was10% (w/v) with respect to acrylamide.

Conditioned growth mediumTo determine the protease activities released into themedium by human breast cancer cells and to investi-gate the inhibitory effects of batimastat on those pro-teases, MDA-MB-231 cells were grown to highdensity (7 × 105 cells/ml) in growth medium contain-ing 10% (v/v) FBS in 75 cm2 flasks. Complete growthmedium was removed and the cells washed threetimes with 10 ml sterile PBS. After removing the finalwash, the cells were incubated in serum-free minimalessential media for 48 h at 37oC in 5% CO2. The con-ditioned medium was collected, cells and debrisremoved by centrifugation (5000 g, 5 min), and con-centrated 10 times using centricon 30 filter concen-trators (Amicon, Inc., Beverly, MA). The concen-trated conditioned medium was tested for proteasesin the presence or absence of 10 mM batimastat byzymography.

Invasion assayType I collagen gel invasion assays were performedin 24-well plates as described by Liebersbach andSanderson [25] except that the assays were allowed

to proceed for 72 h. Quantification of the percentageof invading cells was achieved by first determiningthe number of non-invading cells that could beremoved from the top of the gel with trypsin-EDTAand then counting the invading cells that werereleased from within the type I collagen gel byextensive collagenase digestion as described byLiebersbach and Sanderson [25]. The leading frontof invasion was defined as the point where the twomost distantly migrating breast cancer cells weresimultaneously in focus in one field under × 200magnification. After 72 h, each gel was searched ina Z pattern and the mean distance ± standard errorof the mean was determined using the calibratedmicrometer of a Nikon inverted phase contrastmicroscope. Within experiments, duplicate assayswere performed for each cell line and each experi-ment was repeated at least two times.

Results

Invadopodial extracellular matrix degrading activi-ties of human breast cell linesThe MDA-MB-231 human breast cancer cells areproteolytically active and degrade extracellularmatrix substrates by an invadopodial-dependentmechanism [8,9]. Other work has shown that humanbreast cancer cell lines differ in invasive behavior asdetermined in vitro using a Boyden chamber inva-sion assay with Matrigel as the barrier to invasionand the invasiveness observed in vitro correlatesclosely with in vivo invasive behavior observed usingthe nude mouse model [17]. These workers identi-fied the MDA-MB-231 cell line as one of the mostinvasive. The purpose of this study was to determinethe role of invadopodial proteolysis of extracellularmatrix in the invasion process.

The extracellular matrix-degrading activity of thehuman breast cell lines was investigated by growingthe cells on fluorescently labeled fibronectin that wascovalently coupled to a glutaraldehyde-crosslinkedgelatin film. HBL-100 normal human breast cells donot degrade the matrix and the fibronectin substrateremains intact and uniformly fluorescent underneaththe cells (Figure 1A, B). MCF-7 human breastcancer cells reveal limited degradation of the fluo-rescent-fibronectin film, with one microscopic fieldin ten having the spots of decreased fluorescencethat are indicative of matrix proteolysis (not shown).The MDA-MB-231 breast cancer cell line revealsextensive degradation of the fluorescent fibronectinfilms, with five microscopic fields in ten havingevidence of matrix degradation (Figure 1C, D).

111112345678910111123456789201111234567893011112345678940111123456789501111234111

T. Kelly et al.


The fact that MMP-2 and MT1-MMP are concen-trated on invadopodia [11,12] led us to investigate theeffect of MMP inhibition on proteolysis of extracellu-lar matrix by MDA-MB-231 cells. Batimastat (Mr477) is a broad spectrum MMP inhibitor that mimicsone part of the principle MMP-cleavage site in colla-gen. Batimastat was used because it has nM IC50 val-ues for inhibition of several MMPs including: MMP-1(interstitial collagenase), MMP-3 (stromelysin),MMP-2 (gelatinase A), MMP-9 (gelatinase B) andMMP-7 (matrilysin) [26]. Batimastat also blocks theactivation of MMP-2, presumably by inhibition ofMMP-14 (MT1-MMP) [26]. Batimastat does notinhibit metalloproteinases such as angiotensin-con-verting enzyme and enkephalinase nor does it inhibitother proteases implicated in tumor cell invasion suchas plasmin, urokinase-type plasminogen activator,and cathepsins (personal communication, Peter D.Brown, British Biotech Pharmaceuticals, Ltd, Oxford,UK). Other desirable properties of batimastat includelow cytotoxicity and a minimal cytostatic effect[26,27]. Moreover, cells treated with batimastatexhibit normal migration in chemotactic assays, indi-cating that the adhesive and motile machinery of cellsis not affected by batimastat [26].

Batimastat was used to investigate the effect ofMMP inhibition on extracellular matrix proteolysisby invadopodia. Batimastat completely inhibitedinvadopodial degradation of fluorescent fibronectinfilms by MDA-MB-231 breast cancer cells (Figure2C–F). Invadopodial degradation of extracellularmatrix was apparent underneath the untreatedMDA-MB-231 cells (Figure 2A, B). Neither theDMSO vehicle nor batimastat were toxic to the cellsduring incubation periods up to 72 h or at any ofthe batimastat concentrations used in this study asjudged by trypan blue exclusion assays.

To quantify the extracellular matrix proteolysis,normal HBL-100 and malignant MDA-MB-231breast cells were grown on immobilized [125I]-fibronectin and the radioactivity released from thesubstrate by the proteolytic action of the cells wasdetermined. MDA-MB-231 cells release 11-foldmore [125I]fibronectin degradation products from thesubstrate than HBL-100 cells after 72 h (Figure 3A).The effect of the batimastat on the proteolysis ofextracellular matrix by the MDA-MB-231 breast can-cer cell line was investigated. Batimastat reduced therelease of [125I]fibronectin by MDA-MB-231 humanbreast cancer cells to below that released by the

111112345678910111123456789201111234567893011112345678940111123456789501111234111



Figure 1. Normal HBL-100human breast cells do notdegrade fibronectin butmalignant MDA-MB-231 hu-man breast cancer cells de-grade fibronectin. NormalHBL-100 cells are visualizedby rhodamine phalloidin fluo-rescence (A). The fibronectinsubstrate in the same micro-scopic field is visualized byfluorescein–fibronectin fluo-rescence respectively (B).Arrowheads point to identi-cal locations in the micro-scopic fields. MalignantMDA-MB-231 human breastcells (C) degrade fibronectinthat is covalently linked to thesubstrate (D). Arrowheadspoint to identical locations inthe same microscopic field.The observed fluorescence-negative spots where thefibronectin has been removedare characteristic of inva-dopodia-mediated proteoly-sis (arrowheads: D). Cellswere grown on the fibronectinfilms for 72 h. Bar = 10 mm.

no-cell control (Figure 3B). The release of[125I]fibronectin by MDA-MB-231 cells was onlyslightly reduced by the DMSO vehicle as compared tountreated cells (Figure 3B). Although it was not cyto-toxic, batimastat was active as an inhibitor of MMPs.Batimastat inhibited the 92 and 72 kDa gelatinaseactivities released into serum-free medium by MDA-MB-231 cells (Figure 3C). The complete inhibition ofMDA-MB-231 proteolysis of [125I]fibronectin by bati-mastat suggests that virtually all of the observedinvadopodial proteolysis of extracellular matrix byMDA-MB-231 cells is mediated by MMPs.

Invadopodial extracellular matrix degrading activitycorrelates with invasion potential of human breastcell linesInvasion was evaluated by seeding the cells on topof a native type I collagen gel and allowing them tointeract with the gel for 72 h (Figure 4). Thepercentage of cells that invaded into the type Icollagen gels after 72 h was determined (Figure 4A).

The HBL-100 normal human breast cell line is notinvasive in this assay (Figure 4A, H100). The MCF-7 human breast cancer cell line is somewhat inva-sive (Figure 4A, MCF7) and the MDA-MB-231human breast cancer cell line is the most invasive inthis assay (Figure 4A, M231).

The degree of invasiveness was also reflected inmeasurements of the leading front of invasion. Theleading front of invasion is defined as the deepestlevel in the gel where at least two cells are simulta-neously in focus and is determined at the end of theexperiment using the calibrated fine focus of aninverted microscope (Figure 4B). The HBL-100normal human breast cell line did not invade and theleading front of cells was 100 mm or approximatelyone cell diameter into the gel (Figure 4B, H100). Thehuman breast cancer cell lines invaded into type Icollagen with the leading front of MCF-7 cellsapproximately six cell diameters into the gels and theleading front of MDA-MB-231 cells was 12 cell diam-eters into the gels (Figure 4B, MCF7 and M231).

111112345678910111123456789201111234567893011112345678940111123456789501111234111

T. Kelly et al.


Figure 2. Inhibition ofMDA-MB-231 invadopodialmatrix degradation by bati-mastat. Cells (A, C, E) andunderlying substrates (B, D,F) in the same microscopicfields (A:B, C:D, and E:F) arevisualized for rhodaminephalloidin (cells) and fluores-cein–fibronectin (substrates)after growing for 24 h.Proteolytic degradation offluorescent fibronectin sub-strates associated with inva-dopodia is detected in controlcells (arrows: A and B) butnot in cells treated with0.1 mM (C and D) or 1.0 mM

batimastat (E and F). Thesesubstrates had less fluores-cent intensity than those usedfor Figure 1. Bar = 10 mm.

Batimastat inhibits MDA-MB-231 human breastcancer cell invasion into type I collagen gelsThe correlation between invadopodial matrix-degrading activity (Figures 1–3) and invasion intotype I collagen gels (Figure 4) suggested that inhi-bition of invadopodial matrix proteolysis woulddecrease invasiveness of the malignant cells. Tosubstantiate the observed correlation, the effect ofinhibiting invadopodial proteolysis of matrix oninvasiveness was investigated. Batimastat inhibitsinvasion of MDA-MB-231 cells into type I collagengels (Figure 5A). Batimastat inhibited invasion from2.7-fold to 10-fold relative to the percentage ofinvading cells observed in control cultures growingin the presence of DMSO (Figure 5A). Malignant

MDA-MB-231 cells growing in growth medium orthose growing in growth medium with added DMSOrevealed large invading cell populations (Figure 5A).Normal HBL-100 cells had low numbers of cellsinvading the type I collagen gels (Figure 5).

Batimastat also reduced the leading front of invasion observed for MDA-MB-231 human breastcancer cells (Figure 5B). Batimastat caused theleading front of invasion to be indistinguishable from cells at the top of the gel or located less than half as deep as that observed for cells grownwithout batimastat (Figure 5B). MDA-MB-231breast cancer cells growing in growth medium, or in growth medium with added DMSO revealedleading fronts of invasion ranging from 9 to 11 cell

111112345678910111123456789201111234567893011112345678940111123456789501111234111



Figure 3. Quantification ofmatrix proteolysis by normal andmalignant human breast cells. (A)Malignant MDA-MB-231 humanbreast cells (j) release up to 11-fold more fibronectin into themedia than the normal HBL-100human breast cells (d) over 72 h.Levels of radioactivity weredetermined in two separateexperiments. The average of thetwo determinations after back-ground subtraction is plotted ateach time point. (B) Inhibition ofMDA-MB-231 human breast can-cer invadopodial proteolysis ofmatrix by the metalloproteinaseinhibitor batimastat. Batimastat-treated (10 mM) MDA-MB-231human breast cancer cells (-u-)release less [125I]fibronectin fromthe substrate than the no-cell con-trol (background subtracted).MDA-MB-231 human breast can-cer cells degrade [125I]fibronectinin the absence ( . . . u . . . ) orpresence of the DMSO (j) vehi-cle. Results are expressed as per-centage of substrate released andplotted to the average of deter-minations made in two separateexperiments. The average of thetwo determinations after back-ground subtraction is plotted foreach time point. (C) Lane 1 is azymogram showing gelatinaseactivities at 92 and 72 kDasecreted into serum-free growthmedium conditioned by MDA-MB-231 cells. Lane 2 is a zymo-gram of the same sampleincubated with 10 mM batimastatcausing loss of the 92 and 72 kDagelatinase activities.

diameters into the gel (Figure 5B; M231 andM231 + DMSO). Normal HBL-100 cells invadedapproximately 2.5 cell diameters into the type Icollagen gels (Figure 5B, H100). Together, theseresults indicate that proteolysis of extracel-lular matrix by invadopodia facilitates human breast cancer cell invasion and is mediated byMMPs.

Discussion

Invadopodial proteolysis of extracellular matrixgreatly facilitates invasion by human breast cancercells. Two main lines of evidence support thisconclusion. First, the invasion potential of breastcancer cells is directly correlated with the extracel-lular matrix degrading activity of their invadopodia.

111112345678910111123456789201111234567893011112345678940111123456789501111234111

T. Kelly et al.


Figure 4. MDA-MB-231 human breast cancer cells invade into type I collagen gels. (A) Invasion of different humanbreast cell lines into type I collagen gels. The percentage of cells that invade into type I collagen after 72 h was deter-mined for three different human breast cell lines. The HBL-100 normal human breast cell line (H100) is least invasivewith only 1.2% ± 0.1 invading cells, the MCF-7 human breast cancer cell line (MCF7) is moderately invasive with 3.58%±0.4 invading cells, and the MDA-MB-231 human breast cancer cell line (M231) is most invasive with 10.7% ± 2.7 invadingcells. In these experiments, cells were grown in medium containing 10% complete FBS. For each cell line, the bar lengthindicates the average value of 16 different determinations obtained in eight separate experiments. Error bars indicate± the standard error of the mean. (B) Depth of the leading front of invasion into type I collagen gels of different humanbreast cell lines. The depth of the leading front of cells invading into type I collagen gels was determined after 72 h forthree human breast cell lines. The HBL-100 cells (H100) invade least deeply into the type I collagen gels (100 mm, ± 5.8),MCF-7 cells (MCF7) invade to a moderate depth (667 mm, ± 72.7) and MDA-MB-231 cells (M231) invade furthest intothe type I collagen gels (1.233 mm, ± 88.2). For each cell line, the bars indicate the average value of 40 determinationsobtained in the same eight experiments reported in (A). Error bars are drawn to ± the standard error of the mean. (C)MDA-MB-231 human breast cancer cells growing on top of a type I collagen gel. Cells (5 × 104) were seeded on topof the gel and allowed to invade for 72 h. In this panel, the plane of focus is the top of the gel. (D) Three MDA-MB-231 cells (arrows) are shown invading 600 mm into the type I collagen gel. They were located by focusing down fromthe top of the gel in the microscopic field shown in (C). The scale bar (100 mm) for (C) and (D) is shown in (D).

For each cell line tested, the relative level ofinvadopodial proteolysis of extracellular matrix asmeasured using fluorescent or radiolabeled fibro-nectin substrates, positively correlated with each oftwo different measurements of invasiveness: (i) thepercentage of cells invading into type I collagen gelsand (ii) the distance traveled by cells at the leadingfront of invasion. Second, treatment of invasiveMDA-MB-231 human breast cancer cells with theMMP inhibitor, batimastat, reduced invadopodialfunction as assessed by measuring cell-mediatedproteolysis of immobilized fibronectin. Batimastatalso reduced MDA-MB-231 cell invasion into typeI collagen gels as reflected by the reduced percent-ages of invading cells and shorter distances invaded

into the gels relative to untreated cells. Batimastatinhibition of invadopodial matrix proteolysis andinvasion is apparently due to its inhibition of MMPproteolytic activities because batimastat effectivelyinhibited two gelatinase activities at 92 and 72 kDathat were secreted by the MDA-MB-231 cells.However, batimastat did not affect cell viability. Thelow cytotoxicity of batimastat and MMP inhibitionby batimastat over the range of concentrations usedin this study (0.1–10 mm) has also been observed forhuman MDA-MB-435 breast cancer cells [28]. Theresults presented here confirm separate observationsregarding either invadopodial proteolysis or invasionof human breast cancer cells reported by others [8,17]. This study strengthens the linkage of

111112345678910111123456789201111234567893011112345678940111123456789501111234111



Figure 5. Batimastat inhibits theinvasive behavior of MDA-MB-231 human breast cancer cells. Thepercentage of HBL-100 cells(H100) and MDA-MB-231 cells(M231) invading into type I colla-gen gels (A) and the leading frontof invading cells (B) observed intwo different experiments areshown (A, B: hatched bars = exper-iment 1, black bars = experiment2). (A) Batimastat inhibits invasionof MDA-MB-231 breast cancercells into type I collagen gels.Batimastat (1.0 mM) inhibited inva-sion up to 10-fold relative toDMSO control. Increased batimas-tat concentrations resulted indecreased invasion (experiment 1(hatched bars): M231 + 0.1 mM Batand M231 + 1.0 mM Bat; experi-ment 2 (black bars): M231 + 1 mM

Bat and M231 + 10 mM Bat). Thegraphs are the average of two datapoints and the bars indicate therange of the determinations. (B)Batimastat reduces the distancetraveled into the type I collagen gelby cells at the leading front of inva-sion. In experiment 1 (hatchedbars), the leading front of invasionwas reduced 2.2-fold by 1.0 mM

batimastat (M231 + 1.0 mM Bat)and 2.9-fold by 0.1 mM batimastat.In experiment 2 (black bars),invading cells were not detected (*)below the top of the type I collagengel in the batimastat-treatedgroups (M231 + 10 mM Bat* andM231 + 1.0 mM Bat*). The graphsare averages of measurements ofthe leading front of invasion at fivelocations within the gel and barsrepresent ± standard error of themean.

invadopodial proteolysis of extracellular matrix tothe invasion process because both parameters wereobserved and manipulated in the same study.

MMPs play a critical role in mediating theinvadopodial proteolysis of extracellular matrix thatfacilitates tumor cell invasion. This is evidenced bythe effectiveness of the MMP inhibitor batimastat insimultaneously reducing invadopodial proteolyticactivity and invasiveness of malignant breast cells. Anumber of MMPs have been implicated in having arole in proteolysis of extracellular matrix by humanbreast cancers including: MMP-2 (gelatinase A) [29,30], MMP-9 (gelatinase B) [2, 31, 32], MMP-11(stromelysin-3) [33], an 80 kDa MMP [34] and MMP-14 (MT1-MMP) [35]. Batimastat inhibition of MMPshas been effective in reducing the growth and spreadof mammary tumors in animal models [28, 36].Moreover, batimastat and a more bio-availablederivative of batimastat called marimastat (BB-2516)are being evaluated in clinical trials for use as an anti-tumor therapeutics [37, 38]. The development ofMMP inhibitors for use in anti-tumor therapies con-tinues to be a promising area of research [26].

Although there is mounting evidence that MMPshave a critical role in tumor cell invasion into thecomplex basement membrane and stromal matriceswithin living organisms, the evidence also suggeststhat coordinated activity of other proteases and gly-cosidases in addition to MMPs is needed to efficientlydegrade the matrix and enable invasion [1]. Thisstudy was primarily focused on the role of MMPs inbreast cancer cell invasion because the gel substrateused, triple helical type I collagen, is resistant to proteolytic cleavage by other classes of proteases [2].However, an in vitro study showed that in the pres-ence of serum, batimastat only inhibited MDA-MB-231 proteolysis of human endothelial basementmembranes by 30%. An additional 30–40% inhi-bition of basement membrane degradation wasachieved when batimastat was used in combinationwith inhibitors of urokinase-type plasminogen activa-tors [39]. There is an apparent need for other classesof proteases for tumor cells to efficiently degradecomplex extracellular matrices.

Elucidation of the mechanisms that concentrateMMPs and other extracellular matrix-degradingproteases to invadopodia may lead to new malignantcell diagnostics and strategies to inhibit their inva-sion. MMP-2 and other matrix-degrading proteasesare not restricted to invadopodia. MMP-2 is secretedas a soluble enzyme that can be found within thecytoplasm of cells and embedded in the extracellularmatrix as well as on invadopodia. The integralmembrane MMP, MT1-MMP, has been identified as

an activator of latent MMP-2 and proposed to serveas a cell surface receptor for MMP-2 [40]. In addi-tion, the avb3 integrin has also been shown to serveas a cell surface receptor for MMP-2 [41]. MT-MMPs, integrins or both of these molecules mayhave a role in concentrating MMP-2 to invadopo-dial membranes of human breast cancer cells. It hasbeen shown that MT1-MMP must be localized toinvadopodia to stimulate the local degradation ofextracellular matrix that is characteristic ofinvadopodia and that its cytoplasmic domain has arole in directing MT1-MMP to invadopodialmembranes [12]. Another mechanism for recruit-ment of active proteases to invadopodial membranescould involve directed oligomerization of thesubunits of integral membrane proteases such asseprase, fibroblast activation protein-a, dipeptidylpeptidase IV, and meprin [3]. This directedoligomerization might occur in response to signalingthrough integrins [42]. Perturbing the recruitmentand assembly of proteases on the invadopodialsurface might effectively reduce invadopodial prote-olysis of extracellular matrix and limit invasion.

Acknowledgements

We are grateful to British Biotech Pharmaceuticals,Ltd. for the gift of the batimastat used in this study.The authors thank Aubrey J. Hough, Jr, for encour-aging these studies and Drs Marie Chow, BarbaraL. Parsons, Timothy C. Chambers and Ralph D.Sanderson for critical review of the manuscript. Thiswork partially fulfills the requirements for theMaster of Science degrees of Yan Yan and WinstonS. Chu. Supported by grants from the DoD,Department of the Army #DAMD17-96-1-6097,Council for Tobacco Research SA035, and ArkansasScience & Technology Authority 96-B-26 to TK.JCC was supported by the Arkansas CancerResearch Center Partners in Research Programfunded through NIH grant R25 CA49425 to D. H.Berry. The content of this report does not reflectthe position or the policy of the US. governmentand no official endorsement should be inferred.

References1. Mignatti P and Rifkin DB, 1993, Biology and biochem-

istry of proteinases in tumor invasion. Physiol Revs,73, 161–95.

2. Stetler-Stevenson WG, Aznavoorian S and Liotta LA,1993, Tumor cell interactions with the extracellularmatrix during invasion and metastasis. Annu Rev CellBiol, 9, 541–73.

111112345678910111123456789201111234567893011112345678940111123456789501111234111

T. Kelly et al.


3. Chen WT, 1996, Proteases associated with invadopo-dia, and their role in degradation of extracellularmatrix. Enz and Prot, 49, 59–71.

4. Chen W-T, Olden K, Bernard BA and Chu F-F, 1984,Expression of transformation-associated protease(s)that degrade fibronectin at cell contact sites. J Cell Biol,98, 1546–55.

5. Kelly T, Mueller SC, Yeh Y and Chen W-T, 1994,Invadopodia promote proteolysis of a wide variety ofextracellular matrix proteins. J Cell Physiol, 158,299–308.

6. Chen W-T, 1989, Proteolytic activity of specialized sur-face protrusions formed at rosette contact sites oftransformed cells. J Exp Zool, 251, 167–85.

7. Aoyama A and Chen W-T, 1990, A 170-kDa mem-brane-bound protease is associated with the expressionof invasiveness by human malignant melanoma cells.Proc Natl Acad Sci USA, 87, 8296–300.

8. Chen W-T, Lee C-C, Goldstein L, et al. 1994,Membrane proteases as potential diagnostic and thera-peutic targets for breast malignancy. Breast Cancer ResTreat, 31, 217–26.

9. Coopman PJ, Thomas DM, Gehlsen KR and MuellerSC, 1996, Integrin a3b1 participates in the phagocyto-sis of extracellular matrix molecules by human breastcancer cells. Mol Biol Cell, 7, 1789–804.

10. Monsky WL and Chen W-T, 1993, Proteases of celladhesion proteins in cancer. Semin Cancer Biol, 4,251–8.

11. Monsky WL, Kelly T, Lin C-Y, et al. 1993, Binding andlocalization of Mr 72,000 matrix metalloproteinase atcell surface invadopodia. Cancer Res, 53, 3159–64.

12. Nakahara H, Howard L, Thompson EW, et al. 1997, Transmembrane/cytoplasmic domain-mediatedmembrane type 1-matrix metalloprotease docking toinvadopodia is required for cell invasion. Proc NatlAcad Sci USA, 94, 7959–64.

13. Pei D and Weiss SJ, 1996, Transmembrane-deletionmutants of the membrane-type matrix metallopro-teinase-1 process progelatinase A and express intrinsicmatrix-degrading activity. J Biol Chem, 271, 9135–40.

14. Imai K, Ohuchi E, Aoki T, et al. 1996, Membrane-typematrix metalloproteinase 1 is a gelatinolytic enzymeand is secreted in complex with tissue inhibitor of met-alloproteinases 2. Cancer Res, 56, 2707–10.

15. Strongin AY, Collier I, Bannikov G, et al. 1995,Mechanism of cell surface activation of 72 kDa type IVcollagenase. J Biol Chem, 270, 5331–38.

16. Sato H, Takino T, Okada Y, et al. 1994, A matrix met-alloproteinase expressed on the surface of invasivetumor cells. Nature 370, 61–5.

17. Thompson EW, Paik S, Brunner N, et al. 1992,Association of increased basement membrane inva-siveness with absence of estrogen receptor and expres-sion of vimentin in human breast cancer cell lines. JCell Physiol, 150, 534–44.

18. Moll UM, Lane B, Zucker S, et al. 1990, Localizationof collagenase at the basal plasma membrane of ahuman pancreatic carcinoma cell line. Cancer Res, 50,6995–7002.

19. Davis GE and Martin BM, 1990, A latent Mr 94,000gelatin-degrading metalloprotease induced during dif-ferentiation of HL-60 promyelocytic leukemia cells: amember of the collagenase family of enzymes. CancerRes, 50, 1113–20.

20. Deutsch DG and Mertz ET, 1970, Plasminogen: purifi-cation from human plasma by affinity chromatography.Science, 170, 1095–6.

21. Hunter WM and Greenwood FC, 1962, Preparation ofiodine-131 labeled human growth hormone of highspecific activity. Nature, 194, 495–6.

22. Monsky WL, Lin C-Y, et al. 1994, A potential markerprotease of invasiveness, seprase, is localized toinvadopodia of human malignant melanoma cells.Cancer Res, 54, 5702–10.

23. Harlow E and Lane D, 1988, Antibodies: A LaboratoryManual. Cold Spring Harbor, NY: Cold Spring HarborLaboratory.

24. Heussen C and Dowdle EB, 1980, Electrophoreticanalysis of plasminogen activators in polyacrylamidegels containing sodium dodecyl sulfate and copoly-merized substrates. Anal Biochem, 102, 196–202.

25. Liebersbach BF and Sanderson RD, 1994, Expressionof syndecan-1 inhibits cell invasion into type I collagen.J Biol Chem, 269, 20013–19.

26. Brown PD and Giavazzi R, 1995, Matrix metallopro-teinase inhibition: a review of anti-tumour activity.Annals Oncol, 6, 967–74.

27. Davies B, Brown PD, East N, et al. 1993, A syntheticmatrix metalloproteinase inhibitor decreases tumorburden and prolongs survival of mice bearing humanovarian carcinoma xenografts. Cancer Res, 53, 2087–91.

28. Sledge GW, Jr. Qulali M, Goulet R, et al. 1995, Effectof matrix metalloproteinase inhibitor batimastat onbreast cancer regrowth and metastasis in athymic mice.J Natl Cancer Inst, 87, 1546–50.

29. Goldberg GI, Frisch SM, He C, et al. 1990, Secretedproteases. Regulation of their activity and their possi-ble role in metastasis. Ann NY Acad Sci, 580, 375–84.

30. Brown PD, Bloxidge RE, Anderson E and Howell A,1993, Expression of activated gelatinase in human inva-sive breast carcinoma. Clin Exp Met, 11, 183–9.

31. Davies B, Miles DW, Happerfield LC, et al. 1993,Activity of type IV collagenases in benign and malig-nant breast disease. Br J Cancer, 67, 1126–31.

32. Iwata H, Kobayashi S, Iwase H, et al. 1996, Productionof matrix metalloproteinases and tissue inhibitors ofmetalloproteinases in human breast carcinomas. Jap JCancer Res, 87, 602–11.

33. Basset P, Wolf C and Chambon P, 1993, Expression ofthe stromelysin-3 gene in fibroblastic cells of invasivecarcinomas of the breast and other human tissues: areview. Breast Cancer Res Treat, 24, 185–93.

34. Shi YE, Torri J, Yieh L, et al. 1993, Identification andcharacterization of a novel matrix-degrading proteasefrom hormone dependent human breast cancer cells.Cancer Res, 53, 1409–15.

35. Ueno H, Nakamura H, Inoue M, et al. 1997, Expressionand tissue localization of membrane-types 1, 2, and 3matrix metalloproteinases in human invasive breastcarcinomas. Cancer Res, 57, 2055–60.

36. Eccles SA, Box GM, Court WJ, et al. 1996, Controlof lymphatic and hematogenous metastasis of a rat mammary carcinoma by the matrix metallopro-teinase inhibitor batimastat (BB-94). Cancer Res, 56,2815–22.

37. Wojtowicz-Praga S, Low J, Marshall J, et al. 1996,Phase I trial of a novel matrix metalloproteinaseinhibitor batimastat (BB-94) in patients with advancedcancer. Invest New Drugs, 14, 193–202.

111112345678910111123456789201111234567893011112345678940111123456789501111234111



38. Malfetano J, Teng N, Moore D, et al. 1996, Marimastat,a novel matrix metalloproteinase inhibitor in patientswith advanced cancer of the ovary: a dose-finding study.Proc Am Clin Oncol, 764.

39. Stonelake PS, Jones CE, Neoptolemos JP and BakerPR, 1997, Proteinase inhibitors reduce basementmembrane degradation by human breast cancer celllines. Br J Cancer, 75, 951–9.

40. Sato H, Takino T, Kinoshita T, et al. 1996, Cell surfacebinding and activation of gelatinase A induced by

expression of membrane type-1-matrix metallo-proteinase (MT1-MMP). FEBS Lett, 385, 238–40.

41. Brooks PC, Stromblad S, Sanders LC, et al. 1996,Localization of matrix metalloproteinase-2 to thesurface of invasive cells by interaction with integrinaVb3. Cell, 85, 683–93.

42. Nakahara H, Nomizu M, Akiyama SK, et al. 1996, Amechanism for regulation of melanoma invasion. JBiol Chem, 271, 27221–4.

111112345678910111123456789201111234567893011112345678940111123456789501111234111

T. Kelly et al.


proteolysis of extracellular matrix by invadopodiafacilitates human breast cancer cell invasion and...

Documents