Astrocytoma Cell Interaction WithElastin Substrates: Implications for

Astrocytoma Invasive PotentialSHIN JUNG,1 ALEKSANDER HINEK,2 ATSUSHI TSUGU,1 SHERRI LYNN HUBBARD,1

CAMERON ACKERLEY,3 LAURENCE E. BECKER,3 AND JAMES T. RUTKA1*1Sonia and Arthur Labatt Brain Tumor Research Center, Division of Neurosurgery, The Hospital

for Sick Children, Toronto, Ontario, Canada2Division of Cardiovascular Research, The Hospital for Sick Children,

Toronto, Ontario, Canada3Division of Neuropathology, The Hospital for Sick Children, Toronto, Ontario, Canada

KEY WORDS elastin; elastin binding protein; astrocytoma; invasion; organotypiccultures

ABSTRACT Elastin has been identified within the meninges and the microvascula-ture of the normal human brain. However, the role that elastin plays in either facilitatingastrocytoma cell attachment to these structures or modulating astrocytoma invasion hasnot been previously characterized. We have recently shown that astrocytoma cell linesand specimens produce tropoelastin, and express the 67 kDa elastin binding protein(EBP*). In the present report, we have established that astrocytoma cells attach toelastin as a substrate in vitro. The U87 MG astrocytoma cell line demonstrated thegreatest degree of adhesion. In addition, all astrocytoma cell lines examined werecapable of penetrating and migrating through an intact elastin membrane, and ofdegrading tritiated-elastin, a process that could be prevented by the pre-incubation ofastrocytoma cells with EDTA, but not with a1-antitrypsin. Astrocytoma cells were alsocapable of penetrating 1 mm sections of human brain tissue maintained as organotypiccultures. Interestingly, the invasive potential of cultured astrocytoma cells plated onorganotypic cultures of human brain was significantly increased after exposure to elastindegradation products (k-elastin), which interact with astrocytoma cell surface EBP. Ourdata show that astrocytoma cells express a functional 67 kDa EBP, enabling them topotentially recognize and attach to elastin as a substrate. These data also suggest thatthis elastin receptor may be involved in processes which regulate regional astrocytomainvasion. GLIA 25:179–189, 1999. r 1999 Wiley-Liss, Inc.

INTRODUCTION

Astrocytomas are the most common primary braintumors. In their malignant form, they are characterizedby rapid cellular proliferation and diffuse brain inva-sion. While both of these histopathological featurescontribute to the poor prognosis of patients with ana-plastic astrocytomas, in the present study we addressthe issue of astrocytoma invasion by analyzing theability of astrocytoma cells to respond to elastin whenpresented as an ECM macromolecule. By synthesizingand depositing a permissive ECM, tumor cells cancreate a microenvironment that facilitates their inva-

siveness. Several ECM macromolecules have now beenimmunohistochemically localized to primary humanastrocytomas (Bellon et al., 1985; McComb and Bigner,

Grant sponsor: Medical Research Council of Canada; Grant sponsor: NationalCancer Institute of Canada.

Abbreviations: DTT, Dithiothreitol; EBP, elastin binding protein; ECM, extra-cellular matrix; FCS, fetal calf serum; GAGs, glycosaminoglycans; MMP, matrixmetalloproteinase; MEM, minimal essential medium; PMSF, phenylmethylsulfuonic acid.

*Correspondence to: James T. Rutka, Division of Neurosurgery, Suite 1504,The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, CanadaM5G 1X8. E-mail: rutka@sickkids.on.ca

Received 07 April 1998; Accepted 30 June 1998

GLIA 25:179–189 (1999)

r 1999 Wiley-Liss, Inc.

1985; McComb et al., 1987; Rutka et al., 1987c). Inaddition, astrocytoma cells in vitro have been shown tosecrete ECM proteins (McKeever et al., 1985, 1986;Rutka et al., 1986a). Some of the more commonlydemonstrated ECM macromolecules within human as-trocytomas include tenascin, vitronectin, collagen typesIII, IV, and VI, fibronectin, laminin, hyaluronan, chon-droitin sulfate, and heparan sulfate proteglycans (Bel-lon et al., 1985; McComb and Bigner 1985; Bignami andDahl, 1986; Rutka et al., 1986a, 1987a,b,c, 1988; Mc-Comb et al., 1987; Bourdon and Ruoslahti, 1989; Asherand Bignami, 1992; Ventimiglia et al., 1992; Giese etal., 1995; Giese and Westphal, 1996; Gladson, 1996).

Several families of cell surface receptors have beenidentified that interact with specific domains of ECMproteins. These interactions have been shown to triggersuch diverse cellular events as cell attachment andadhesion, alterations in cell morphology, and initiationof second messenger pathways (Venstrom and Rei-chardt, 1993; Rosales et al., 1995). The 67 kDa EBP is anon-integrin cell surface receptor expressed on fibro-blasts, smooth muscle cells, chondroblasts, leukocytes,and certain cancer cells (Hinek, 1996). It recognizesseveral non-identical hydrophobic domains on elastin,laminin, and type IV collagen (Hinek, 1994).

We have previously shown that human astrocytomacells in vitro and in situ express elastin, and the 67 kDaEBP (Jung et al., 1998). The existence of these proteinsin human astrocytoma cells suggested to us that theymay potentially subserve a role in astrocytoma attach-ment to and migration along elastin-rich subtsrates.Accordingly, in the present study, we have tested thecapabilities of astrocytoma cells in vitro to attach toelastin-coated dishes, and elastin membrane prepara-tions. In addition, we have studied the capability ofastrocytoma cells to invade elastin membranes and 1mm sections of human brain tissue maintained asorganotypic cultures.

MATERIALS AND METHODSAstrocytoma Specimens, Cell Lines, Cell

Cultures, and Culture Conditions

Human malignant astrocytoma cell lines SF126,SF188, and SF539 have been previously characterized(Rutka et al., 1986a, 1987b). U87 MG, U251 MG, andU373 MG were obtained from the American TypeCulture Collection (Rockville, MD). U343-MG-A wasobtained from Dolores Dougherty, the Brain TumorResearch Center, University of California (San Fran-cisco, CA). The cell lines were grown in monolayerculture, and routinely maintained in aMEM (GibcoBRL, Gaithersburg, MD) supplemented with 10% FCSand 100 U/ml penicillin, 100 µg/ml streptomycin, and25 µg/ml amphotericin B (Gibco BRL, Gaithersburg,MD). Human astrocytoma specimens were obtained atthe time of craniotomy. Permission to utilize this tissuewas granted by the Research Ethics Board, The Hospi-tal for Sick Children.

Isolation of EBP

Astrocytoma cells (2 3 106 cells/sample) were col-lected by trypsinization, counted, suspended in culturemedium containing 10% FCS, and incubated as asuspension for 6 h at 37°C. Cells were then washed withcold PBS, placed on ice, and lysed with 0.1 M sodiumbicarbonate buffer (pH 8.0) containing 3 M guanidine-HCl, 10 mM Hepes, 0.1 M DTT, and proteinase inhibi-tors (5 mM EDTA, 5 mM PMSF, and 5 mM benzamidine[all Sigma Chemical Co., St. Louis, MO]) for 1 h. Aftercentrifugation, the supernatants were dialyzed exhaus-tively at 4°C against 0.1 M sodium bicarbonate (pH 8.0)and then mixed with insoluble elastin (200 mg/sample)for affinity purification (2 h at 4°C). After extensivewashing with 0.1 M sodium bicarbonate buffer (pH 8.0),the elastin bound proteins were eluted from the elastinslurry with 0.5 M lactose, concentrated and assessed bywestern blotting with an anti-S-GAL monoclonal anti-body. This antibody was raised against a syntheticdecapeptide taken from the human elastin bindingdomain of EBP. This antibody recognizes human, sheep,and bovine EBP (Mecham et al., 1989; Hinek et al.,1993). Reactions were enhanced with an ECL kit(Amersham Life Science, England) and blots weresubjected to densitometric analysis.

Immunochemistry for EBP and ECMMacromolecules

Astrocytoma cells were immunostained by usingantibodies to EBP and elastin. Subconfluent cultures ofastrocytoma cells were fixed with cold 100% methanolfor 30 min, washed in PBS, blocked with 1% bovineserum albumin (BSA; Sigma Chemical Co., St. Louis,MO) and 1% normal goat serum in PBS, and incubatedeither with a polyclonal antibody to tropoelastin (5mg/ml, diluted 1:200) (Elastin Products, Pacific, MO) orwith a polyclonal anti-S-GAL antibody recognizing EBP(2 mg/ml initial concentration, diluted 1:100) (Hinek etal., 1993). The cells were then incubated with goatanti-rabbit fluorescein-labeled secondary antibodies(Zymed, San Francisco, CA) diluted 1:200. Negativecontrols included omission of the primary antibodies.Immunostains, using antibodies to laminin (SigmaChemical Co., St. Louis, MO) (1 mg/ml, diluted 1:200),and fibronectin (Sigma Chemical Co., St. Louis, MO) (1mg/ml, diluted 1:400), were also performed by using thesame fixation technique described above. Our immuno-cytochemical technique to localize these ECM macromol-ecules has been described previously (Rutka et al.,1987b). In these experiments, the astrocytoma nucleiwere counterstained for ease of recognition with prop-idium iodide (Sigma Chemical Co., St. Louis, MO),1:10,000 dilution from a 1 mg/ml stock solution.

To localize tropoelastin in surgical specimens of nor-mal human brain tissue (obtained from routine resec-tions for seizure disorders) and human malignantastrocytomas, 7 µm paraffin sections were deparaf-

180 JUNG ET AL.

finized and endogenous peroxidase was blocked with0.3% (v/v) hydrogen peroxide in 100% methanol. Toblock nonspecific binding, sections were treated with0.5% bovine serum albumin and 0.5% normal goatserum in PBS before incubation with the polyclonalantibody to tropoelastin (10 mg/ml, diluted 1:100).Peroxidase-labeled secondary antibodies (GAR-HRP,1:200 dilution; Dako, Santa Barbara, CA) were used.The reaction was then developed with 3,38-Diaminoben-zidine and 30% hydrogen peroxide, and nuclei werecounterstained with Harris-hematoxylin.

Astrocytoma Cell Attachment to Elastin

To document whether the presence of the 67 kDa EBPmight be of physiological importance in the attachmentto elastin as a substrate, astrocytoma cells were har-vested by trypsinization, sedimented by centrifugation,and washed twice in a-MEM containing 10% serum.The cells were then resuspended in Hanks balancedsalt solution (HBSS) supplemented with 20 mM HEPES,pH 7.4. Approximately 106 cells were plated in 10 cm2

bacteriological Petri dishes (Polar Plastic, Ltd., St.Laurent, Quebec), which were pre-coated with k-elastin(Elastin Products Co., Owensville, MO). To coat thedishes, k-elastin was dissolved in 0.1 M sodium bicarbon-ate buffer before it was added to the 10 cm2 dishes in afinal concentration of 500 µg/dish. The k-elastin wasthen air dried at room temperature for 18 h. Astrocy-toma cells plated on the elastin-coated dishes wereincubated at 37°C for 4, 8, and 24 h. At the end of eachtime interval, the medium containing unattached cellswas removed and the dishes were washed twice withHBSS. The attached cells were counted at 50 randomhigh-power fields under light microscopy after fixationwith 4% paraformaldehyde and expressed as mean cellattachment/high-power field. In the absence of coatingwith k-elastin, there was no cell attachment to thedishes. The specificity of attachment to k-elastin wasdemonstrated in some experiments by incubating astro-cytoma cells in anti-S-GAL polyclonal antibody prior totheir placement on the k-elastin-coated dishes.

Astrocytoma Migration Through PurifiedElastic Membranes

An in vitro assay using pure elastin membranesprepared from sheep aorta was used to evaluate astrocy-toma attachment to, and migration and active penetra-tion through, the elastin-containing structures. Elasticmembranes were prepared by boiling adult femalesheep aortas for 45 min in 0.1 N NaOH. This procedureremoves all cellular and extracellular components ofthe vessel wall except insoluble elastin. After an addi-tional 1 h extraction with 1 M NaCl, extensive washingin water, PBS, and aMEM, the membranes were cutinto round pieces to fit snugly in the bottom of the24-well culture dishes and then stored at 4°C in aMEM.

Astrocytoma cells were trypsinized as described above,extensively washed in aMEM containing 5% FCS,incubated for 3 h with the same medium, and thenresuspended in a concentration of 105 cells/ml in freshmedium. The cell suspensions were then plated on topof the elastic membranes in the tissue culture wells andincubated for 9 days. In this model, the medium waschanged every 3 days. At the end of 9 days, themembranes were fixed in 2% paraformaldehyde and 2%formaldehyde in PBS. Serial transverse histologicalsections were prepared through the elastin membranesand examined under light microscopy to determine therate and depth of astrocytoma cell penetration.

Assay for Astrocytoma Elastolytic Activity

The elastolytic activity of each of the astrocytoma celllines was assessed in a modified in vitro assay measur-ing the degradation of an insoluble [3H]elastin sub-strate (Senior et al., 1989b; Leake et al., 1983; Bandaand Werb, 1991). Purified elastin from bovine liga-mentum nuchae (Elastin Products Co., Owensville,MO) was radiolabeled as previously described using[3H]NaBH4 (Takahashi et al., 1973). [3H]Elastin wasreconstituted at 10 mg elastin/ml (specific activity2,000 cpm/µg elastin) in Tris Assay Buffer (50 mM TrisHCl, 150 mM NaCl, 10 mM CaCl2, 0.02% Brij, and0.02% NaN3, pH 8.0) and stored at 270°C until assay.Before use, [3H]elastin was boiled and washed with theAssay Buffer until the background counts became lessthan 100 cpm/100 µl supernatant.

Astrocytoma cells from all seven cell lines wereplaced onto 24-well tissue culture plates at a celldensity of 105 cells/well. Two days after plating, thecells were nearly confluent. The culture medium wasthen exchanged with 400 µl of fresh MEM containing1% FCS, followed by the addition of 100 µl of asuspension containing 50 µg insoluble [3H]elastin for afurther 24 h. At the end of the incubation period, 300 µlof culture medium was harvested from each well,transferred to an Eppendorf vial and microcentrifugedat 8,000 g for 4 min. Then, 75 µl aliquots were mixedwith 4 ml of aqueous scintillation counter fluid andcounted in triplicate for 2 min (LKB Wallac 1219Rackbeta counter, San Francisco, CA). The mean CPMvalues were calculated and normalized for the volumeof culture medium and the DNA content of each culture,which was assessed utilizing ethidium bromide. Tocontrol for any non-enzymatic degradation of the elas-tin substrate, radiolabeled elastin was incubated incell-free wells containing medium. The counts obtainedfrom these controls were subtracted as background. Inthis way, the elastolytic activity was assessed in allseven astrocytoma cell lines. In each experiment, tripli-cate cultures were maintained either in plain medium(control) or in medium containing the serine proteinaseinhibitor, a1-antitrypsin 5 mg/ml, or the metalloprotein-ase inhibitor, EDTA (10 mM).

181ASTROCYTOMA ATTACHMENT TO ELASTIN

Astrocytoma Invasion of Brain TissueMaintained as Organotypic Cultures

In an attempt to recapitulate the matrix macromol-ecule representation normally encountered by infiltrat-ing astrocytoma cells, we adapted neural organotypiccultures as previously described (Gahwiler, 1984, 1988;Stoppini et al., 1991; Tanaka et al., 1994). Specimens ofhuman brain were obtained during the course of rou-tine craniotomy for arteriovenous malformation,trauma, or epilepsy. Permission to utilize this tissuewas granted by the Research Ethics Board, The Hospi-tal for Sick Children. The brain tissue was cut intoslices, 1 mm in thickness and 8 3 8 mm2 in width andplaced onto the upper chamber of 24 mm Transwellculture dish (0.4 µm pore size, Corning Costar Coopera-tion, Cambridge, MA). The brain slices were incubatedin medium containing Eagle’s basal medium includingsupplements such as glutamine (Gibco BRL, Gaithers-burg, MD), and 10% FCS, insulin-transferrin-sple-nium-A (1003) (Gibco BRL, Gaithersburg, MD), glu-cose and 20 nM progesterone (Sigma Chemical Co., St.Louis, MO) (Tanaka et al., 1994). Approximately 105

astrocytoma cells were placed on top of the brain sliceand incubated for 7 days in the presence or absence ofk-elastin (1 mg/ml) in the medium. Midline, transversehistologic sections of the brain slices were assessed bylight microscopy to determine the viability of the tissueand the depth of normal brain invasion by the astrocy-toma cells. The pattern of astrocytoma invasion inrelationship to the leptomeninges and vascular chan-nels containing elastin was also characterized.

Analysis of Data

In all studies, mean and standard deviations werecalculated for each group and statistical analyses of theresults were carried out by ANOVA followed by theDuncan test of multiple comparisons to establish whichgroups were significantly different.

RESULTSAstrocytoma Cells Show Elastin and EBP

Immunoreactivity In Vitro and In Situ

By immunocytochemistry, the astrocytoma cell lineswere positively immunostained with antibodies to EBP,tropoelastin, laminin and fibronectin (Fig. 1). We ob-served that tropoelastin and laminin were predomi-nantly confined to the intracytoplasmic space and werenot associated with significant decoration of the ECM.U251 MG, U343 MG-A, and SF539 astrocytoma celllines demonstrated greater intensity of intracytoplas-mic staining for tropoelastin than did U87 MG, U373MG, and SF126. Positive immunostaining for EBP wasfound in all astrocytoma cell lines along the cell surfacemembrane and within the cytoplasm. Immunoreactiv-

ity for EBP was particularly intense in the U87 andU373 astrocytoma cell lines. Elastin was immunolocal-ized to the arterial walls and leptomeninges of normalhuman brain and malignant human astrocytomas insitu. Interestingly, whereas neurons and astrocyteswere non-immunoreactive with the monoclonal anti-body to tropoelastin, anaplastic astrocytoma cells werepositively identified by this antibody (Fig. 2).

Astrocytoma Cells Produce EBP

Affinity chromatography of astrocytoma cell extracts,passed over insoluble elastin, retained the 67 kDaprotein from all cell lines tested. The identity of thiselastin-bound 67 kDa species was further confirmed bywestern blotting with anti-S-GAL antibody recognizingEBP (Fig. 3). While there was variability in expressionof EBP among the different cell lines, SF188 and SF539produced the least EBP by Western blot analysis.

Astrocytoma Cells Adhere to Elastinas a Substrate

To assess whether human astrocytoma cells, whichexpress the 67 kDa EBP, are capable of binding elastin,we investigated the adhesion of astrocytoma cells toelastin-coated plates in a time-course fashion. Theadhesion assay revealed that all astrocytoma cells werecapable of binding to elastin as a substrate (Table 1).U87 MG astrocytoma cells demonstrated the greatestdegree of adhesion, a finding that correlated with thehigh level of EBP expression in this particular cell lineby Western blot. SF-188 astrocytoma cells demon-strated the least amount of adhesion to elastin and hadthe lowest levels of EBP of all cell lines tested. Withrespect to the other astrocytoma cell lines, a partial butimperfect correlation was found between EBP expres-sion and attachment to elastin. For all astrocytoma celllines examined except SF539 and SF126, astrocytomaadhesion to elastin increased significantly over timewhen assessed at 4, 8, and 24 h time periods. Astrocy-toma cell lines pre-treated with anti-S-GAL polyclonalantibodies showed no significant binding to k-elastin-coated dishes (data not shown).

Migration of Astrocytoma Cells Through IntactElastic Membranes

All seven astrocytoma cell lines attached to the intactelastic membrane prepared from sheep aorta. Of these,all except SF539 and SF188 were capable of penetrat-ing the three-dimensional network of the elastin mem-branes to varying levels (Fig. 4). In the cases of SF539and SF188, these were capable of attaching to thesurface of the elastin membrane, but they were unableof penetrating into its depths.

182 JUNG ET AL.

Astrocytoma Cells Possess IntrinsicElastolytic Activity

All astrocytoma cell lines degraded preparations ofinsoluble tritiated elastin. The elastolytic activity of thecell lines was significantly decreased by the addition ofeither 10 mM EDTA (P , 0.01) but not by 5 mg/mla1-antitrypsin (Fig. 5). These data imply that astrocy-toma cells possess significant elastolytic activity whichcan be inhibited by an inhibitor of MMPs.

Characterization of Astrocytoma Migration IntoHuman Brainslice Preparations

We have studied the process of astrocytoma invasionusing cultured, 1 mm slices of human brain (Fig. 6).Astrocytoma cells placed on these organotypic cultureswere capable of penetrating the tissue to various depths(Fig. 7). Astrocytoma cells were identified within thedepths of the brain slice either as single cells or as microtumor nodules. In many fields, astrocytoma cells werefound adjacent to vascular channels. Interestingly, theincubation of astrocytoma cells on brain slices in the

presence of k-elastin appeared to expedite the processof astrocytoma invasion when compared to controls atthe same time points (Fig. 7). Brain slices maintainedas described above as organotypic cultures can remainhistologically intact for 7 days without showing overtsigns of tissue or matrix destruction and can be readilyprocessed to study the characteristics of the invadingtumor cells.

DISCUSSION

In this study, we have characterized the expression oftropoelastin and the 67 kDa EBP in astrocytoma speci-mens and cell lines. By Western blot, we have shownthat the 67 kDa EBP is widely expressed amongastrocytoma cells in vitro. By immunocytochemistry,we have demonstrated that astrocytoma cells expressEBP at the cell surface. That this protein may act as afunctional receptor for the substrate elastin is inferredfrom our cell attachment assays in which astrocytomacells attached readily to elastin-coated culture platesand to intact elastin membranes. All astrocytoma celllines expressed EBP at the cell surface. Levels of EBP

Fig. 1. Immunolocalization of fibronectin (a), laminin (b), EBP (c),and tropoelastin (d) in astrocytoma cell lines using specific antibodies.Antibodies against EBP showed strong immunoreactivity on the cellsurface and the cytoplasm (3600). Antibodies against elastin and

laminin were detected within the intracytoplasmic space but not in theECM (3400). In contrast, fibronectin was distributed primarily withinthe ECM (3400). In this composite photomicrograph, nuclei have beencounterstained with propidium iodide for ease of recognition.

183ASTROCYTOMA ATTACHMENT TO ELASTIN

expressed by astrocytoma cells correlated at least inpart with astrocytoma attachment to elastin-coateddishes

The true extracellular space of the brain comprisesapproximately 20% of its total volume (Harreveld,1966; Rutka et al., 1988; Nicholson, 1995). There are atleast three different regions of the brain in whichunique aggregations of ECM macromolecules occur(Rutka et al., 1988). The first is the glial limitansexterna, which is a basal lamina, that invests the entiresurface of the brain and typically separates astrocyticfoot processes from leptomeningeal cells. The second isthe vascular basement membrane, which separates-

TABLE 1. Time course assay of astrocytoma adhesion toelastin-coated dishes

Astrocytomacell line 4 h 8 h 24 h

U 87 MG 21.48 6 1.71* 21.56 6 1.73 27.62 6 1.93U 251 MG 11.74 6 1.07 16.18 6 1.58 17.66 6 1.26U 343 MG-A 8.22 6 1.12 15.28 6 1.34 13.98 6 1.10U 373 MG 8.96 6 1.37 11.60 6 1.2 12.88 6 0.93SF-126 13.48 6 1.73 10.96 6 0.91 13.78 6 1.23SF-188 5.74 6 1.10 11.28 6 0.93 11.34 6 1.05SF-539 13.58 6 1.69 N/D 12.26 6 1.02

*Fifty high power fields were examined for each cell line for each time point.Numbers refer to mean number of cells attached to elastin coated dishes per highpower field 1/2 standard deviation. N/D 5 not done.

Fig. 2. Immunohistochemical localization of tropolelastin in non-neoplastic brain (A) and anaplastic human astrocytoma (B) speci-mens. Whereas in non-neoplastic human brain, tropoelastin expres-sion is confined to the leptomeninges at the cortical surface, in human

anaplastic astrocytoma specimens, the cytoplasm of the majority ofthe tumor cells is positively identified with the antibody to tropoelas-tin (H & E, 3250).

Fig. 3. Western blot analysis with anti-EBP antibody showed thatall astrocytoma cell lines express the 67 kDa EBP isolated by elastinaffinity chromatography. Protein bands were further assessed by

densitometry. Values are expressed as densitometric units standard-ized to a control sample (extraction buffer run over an elastin column)in which only traces of EBP are found.

184 JUNG ET AL.

cerebral capillary endothelial cells from the paren-chyma. Both the glial limitans externa and the vascularbasement membrane are composed of a well-character-ized ECM containing laminin, fibronectin, type IVcollagen, and heparan sulfate proteoglycan (Rutka etal., 1988). The third region of brain ECM is the amor-phous matrix found within the brain proper separatingastrocytes, neurons, oligodendrocytes, and ependymalcells (Ruoslahti, 1996). In contrast to the ECM of mostother organ systems, the ECM found within cerebralgray and white matter is composed mainly of GAGs andproteoglycans (Ruoslahti, 1996). As an ECM macromol-

ecule, elastin has been identified within the normalcerebral vasculature and the leptomeninges in situ. Wehave previously shown that tropoelastin and elastindegradation products promote the proliferation of hu-man astrocytoma cells (Jung et al., 1998). In addition toshowing that astrocytoma cells produce tropoelastinand EBP, in the present report, we have studiedastrocytoma cell interactions with the insoluble elastinand with the soluble elastin-derived peptides (k-elastin).

It has been reported that laminin, tenascin, andcollagen may serve as permissive substrates for gliomacell migration, whereas fibronectin and vitronectin aregenerally less permissive, at least when these compo-nents are presented to migrating astrocytoma cells in

Fig. 4. Migration assay through elastin membranes. Pure elastinmembranes prepared from sheep aorta were used to evaluate theability of astrocytoma cells to attach to and migrate through theelastin-containing structures. Astrocytoma cells were plated on top ofthe elastic membranes and incubated for 9 days. Transverse histologi-

cal sections of the membranes showed that all astrocytoma cell linestested were able to attach to and migrate through these preparationsto varying degrees (H&E 3200). a, SF-188; b, SF-126; and c, U 343MG-A astrocytoma cells. Arrowheads point to astrocytoma cells on thesurface or within the depths of the elastin membrane.

Fig. 5. Elastolytic activities of U 373 MG astrocytoma cells. In thisexperiment, the elastolytic activity of SF-539 astrocytoma cells couldbe significantly inhibited by EDTA but not by a1-antitrypsin. Thesedata suggest that the elastolytic activity of these cells relates more tometalloproteinase than to serine proteinase production.

Fig. 6. Organotypic culture method for studying astrocytoma inva-sion. In this model, human astrocytoma cells are placed on top of anorganotypically cultured human brain slice measuring 1 mm inthickness. This model has the distinct advantage over other in vitroinvasion models of providing astrocytoma cells with a relevant compo-sition of brain-derived ECM macromolecules. The histological preser-vation of this tissue over time enables recognition of patterns ofastrocytoma invasion with respect to vascular channels, brain paren-chyma, and the leptomeninges.

185ASTROCYTOMA ATTACHMENT TO ELASTIN

the form of purified components (Giese and Westphal,1996). With respect to laminin, there is variability inthe response of astrocytoma cells to this ECM macromol-ecule (Berens et al., 1994; Giese et al., 1994; Reith et al.,1994; Koochekpour et al., 1995; Goldbrunner et al.,1996; Tysnes et al., 1996). Laminin can have either astimulatory or an inhibitory effect on astrocytoma cellmigration following its interaction with the integrinreceptors. Interestingly, the blocking of integrins doesnot completely abolish laminin’s effects on migrationGoldbrunner et al., 1996; Tysnes et al., 1996). A poten-tial explanation for this phenomenon might be theexistence of an alternative pathway that utilizes adifferent, non-integrin laminin receptor such as EBP(Mecham and Hinek, 1995). The 67 kDa EBP is identi-cal to the enzymatically inactive spliced variant ofbeta-galactosidase (Hinek et al., 1993; Privitera et al.,1998) and is a non-integrin cell surface receptor ex-pressed on fibroblasts, smooth muscle cells, chondro-blasts, leukocytes, and certain cancer cell types(Mecham et al., 1989; Hinek, 1994, 1996; Timar et al1995). EBP binds with several non-identical hydropho-bic domains on elastin, laminin, and type IV collagen,provided they form a similar secondary conformation

(Hinek, 1994). EBP is not a transmembrane moleculebut becomes immobilized on the cell surface by anassociation with two other proteins: the 61 kDa neur-aminidase and the 55 kDa protective protein (Hinek,1996). The EBP, which also has galactolectin proper-ties, can be shed from the cell surface after interactionwith galactosugars (Hinek, 1994). Thus, galactosugar-dependent modulation of EBP expression has beenshown to alter cell–matrix interactions including tumorcell and leukocyte chemotaxis toward elastin, laminin,collagen type IV (Mecham et al., 1989; Senior et al.,1989a,b), smooth muscle cell attachment to elastin(Hinek et al., 1992), and their migration during vascu-lar thickening (Hinek, 1996).

Before it forms multiple crosslinks with tropoelastinmonomers to become insoluble, tropoelastin is verysensitive to proteolytic degradation. Increased elastindegradation occurs in the context of several pathologi-cal conditions in which potent elastinolytic proteasesare released from inflammatory cells or bacteria(Mecham, 1997). In our study, the degradation of[3H]elastin by astrocytoma cells was blocked by EDTA,but not a1-antitrypsin, suggesting that MMPs may beresponsible for the rapid proteolysis of extracellular

Fig. 7. U87 MG and U343 MG-A astrocytoma cells placed onprepared organotypic cultures of human brain were capable of penetrat-ing the tissue to various depths. a: U87 MG astrocytoma cells, 7 dayspost-implantation. The majority of astrocytoma cells reside at the siteof implantation on top of the brain slice (open arrowhead). A vascularchannel is observed (small arrow). b: U87 MG astrocytoma cells at 7days post-implantation and post-treatment with k-elastin (1 mg/ml)for 7 days. An advancing tumor front (solid arrowhead) is visualizedwhich connects to astrocytoma cells at the implantation site (openarrowhead). Several micro tumor nodules are appreciated beneath the

site of implantation (solid arrowheads). c: U343 MG-A astrocytomacells, 7 days post-implantation. A thin layer of astrocytoma cells isseen at the surface of the organotypic culture (open arrowhead). d: U343 MG-A astrocytoma cells at 7 days post-implantation and post-treatment with k-elastin (1 mg/ml) for 7 days. Multiple micro tumornodules (solid arrowheads) are observed in the brain tissue. Someastrocytoma cells are also found at the surface of the organotypicculture (open arrowhead). U343 MG-A appeared less invasive in termsof depth of penetration into the brain slice than U87 MG astrocytomacells in this system. For all panels, Scale bar 5100 µm.

186 JUNG ET AL.

elastin molecules such that no elastin was found depos-ited within the tumor cell matrix of the astrocytoma celllines studied. It has been established previously thatMMPs are capable of degrading a wide array of ECMsubstrates (Stetler-Stevenson et al., 1989; Rutka et al.,1995; Uhm et al., 1996, 1997; Mecham, 1997;). Of thematrix-degrading proteases, MMP-1 and MMP-2 havebeen reported to degrade elastin (Uhm et al., 1997).

The invasive potential of several cancer cell lines hasbeen commonly tested in vitro by using matrigel as abarrier to invasion (Kramer et al., 1986; Albini et al.,1987; Hendrix et al., 1987; Erkell and Schirrmacher,1988; Kanemoto et al., 1991; Rutka et al., 1994, 1995).However, matrigel does not contain significant quanti-ties of elastin and would not permit us to test specifi-cally the role of EBP in binding to elastin as a substrate.For these reasons, we utilized preparations of purifiedelastin prepared from sheep aorta on which we placedour astrocytoma cells. This particular model may holdmore biological relevance than the matrigel assay withrespect to the migration of astrocytoma cells withinperivascular spaces commonly observed in astrocytomaspecimens. Interestingly, in our study, all astrocytomacell lines were capable of attaching to the elastinmembrane, and all but two cell lines (SF539 andSF188) were able to penetrate into the depths of theelastin preparations. We would have liked to testwhether adult human astrocytes, which under normalconditions do not migrate or proliferate, attach toelastin as a substrate. However, at present there is noreliable model for studying adult human astrocytes inmonolayer culture (Rutka et al., 1986b).

Because neither the matrigel nor elastin preparationmodel systems adequately simulate the true ECM ofthe brain, we made adaptations to pre-existing neuralorganotypic culture methods to study astrocytoma inva-sion using 1 mm slices of human brain tissue (Gahwiler1984, 1988; Stoppini et al., 1991; Tanaka et al., 1994;Gates et al., 1996). Brain organotypic cultures havelong played an important role in elucidating mecha-nisms related to fiber outgrowth, myelin formation andthe development of synaptic networks (Gahwiler, 1988).The characteristic cytoarchitecture and distinct organi-zation of brain tissue can be maintained in this systemfor several weeks to months (Gahwiler, 1984; Tanaka etal., 1994). Interestingly, Tanaka et al. have shownactive proliferation of granule cells in the externalgranular layer within cerebellar organotypic culturesderived from 9-day-old rats (Tanaka et al., 1994). Gateset al. showed that despite loss of neurons from adultmice striatal explants at 1 week post culture initiation,there was retention of certain structural features,suggesting the viability of this model for studyinggraft-scar interactions. Our interest in using organo-typic cultures of human brain tissue stemmed from ourdesire to recapitulate in vitro the three-dimensionalECM normally encountered by infiltrating human astro-cytoma cells in situ. While there are several limitationsof the brain slice culture system we have described inthis report, we have observed that the cytoarchitecture

of the brain tissue remains viable and intact withoutshowing histopathological signs of obvious tissue necro-sis for 7 days.

By using organotypic cultures of brain as a model toassess astrocytoma invasion, we observed that astrocy-toma cells were capable of penetrating the brain tissuewell beyond their sites of implantation. Invading astro-cytoma cells were identified by morphological criteriaas either single cells or as micro tumor nodules withinthe depths of the brain slices. Many astrocytoma cellshad a tendency to course towards and adhere to ele-ments of the cerebral microvasculature and the lepto-meninges. As there are no specific markers of humanastrocytoma cells, one of the problems with our organo-typic culture system was the unequivocal identificationof the solitary, invading astrocytoma cell. In futurestudies, we plan to overcome this difficulty by labelinghuman astrocytoma cells with a non-perturbing andnon-secreting marker, such as green fluorescent protein(Gerdes and Kather, 1996; Gubin et al., 1997), prior totheir placement on human brain slices.

Interestingly, the migration of the astrocytoma cellscould be stimulated in our study by medium containingk-elastin. These findings suggest that exogenously ad-ministered soluble elastin degradation products canform a gradient which stimulates the migratory behav-ior of astrocytoma cells, potentially through the samemechanism involving chemotaxis of cells bearing EBPtowards elastin-derived peptides (Blood et al., 1988;Senior et al., 1989a; Yusa et al., 1989; Hinek, 1994). Insupport of this notion, we have recently shown thattropoelastin and elastin degradation products promotethe proliferation of human astrocytoma cell lines onlyin the presence of the intact and unblocked cell surfaceEBP (Jung et al., 1998). Our results are consistent withthe findings of Parsons who has suggested that elastinand its precursor peptides may act as enhancers ofinvasion for some tumor cells (Parsons, 1993).

In summary, our data show that astrocytoma cellsexpress a functional 67 kDa EBP, enabling them torecognize, attach to, and migrate along elastin as asubstrate. In addition, our results show that astrocy-toma cells possess significant elastolytic activity, whichis likely due to the MMPs. Finally, we have shown thatelastin degradation products facilitate astrocytoma cellpenetration through slices of cultured human brain.Since elastin-derived peptides evoke intracellular sig-nals by interacting with EBP, we suggest that thiselastin receptor may be involved in processes thatregulate regional astrocytoma invasion in the brain.

ACKNOWLEDGMENTS

This work was supported in part by grants from theMedical Research Council of Canada (J.T.R., A.H.), andby the National Cancer Institute of Canada (J.T.R.). AHis a Career Investigator of the Heart and StrokeFoundation of Ontario.

187ASTROCYTOMA ATTACHMENT TO ELASTIN

REFERENCES

Albini A, Iwamoto Y, Kleinman H, Martin G, Aaronson S, et al. 1987. Arapid in vitro assay for quantitating the invasive potential of tumorcells. Cancer Res 47:3239–3245.

Asher R, Bignami A. 1992. Hyaluronate binding and CD44 expressionin human glioblastoma cells and astrocytes. Exp Cell Res 203:80–90.

Banda M, Werb Z. 1991. Mouse macrophage elastase. Biochem J193:589–605.

Bellon G, Caulet T, Cam Y. 1985. Immunohistolocalisation of macromol-ecules of the basement membrane and extacellular matrix of humangliomas and meningiomas. Acta Neuropathol (Berlin) 66:245–252.

Berens M, Rief M, Loo M, Giese A. 1994. The role of extracellularmatrix in human astrocytoma migration and proliferation studiedin a microliter scale assay. Clin Exp Metastasis 12:405–415.

Bignami A, Dahl D. 1986. Brain-specific hyaluronate-binding protein:A product of white matter astrocytes? J Neurocytol 15:671–679.

Blood CH, Sasse J, Brodt P, Zetter BR. 1988. Identification of a tumorcell receptor for VGVAPG, an elastin-derived chemotactic peptide. JCell Biol 107:1987–1993.

Bourdon MA, Ruoslahti E. 1989. Tenascin mediates cell attachmentthrough an RGD-dependent receptor. J Cell Biol 108:1149–1155.

Erkell L, Schirrmacher V. 1988. Quantitative in vitro assay for tumorcell invasion through extracellular matrix or into protein gels.Cancer Res 48:6933–6937.

Gahwiler BH. 1984. Slice cultures of cerebellar, hippocampal andhypothalamic tissue. Experientia 40:235–243.

Gahwiler BH. 1988. Organotypic cultures of neural tissue. TrendsNeurosci 11:484–489.

Gates MA, Fillmore H, Steindler DA. 1996. Chondroitin sulfateproteoglycan and tenascin in the wounded adult mouse neostriatumin vitro: Dopamine neuron attachment and process outgrowth. JNeurosci 16:8005–8018.

Gerdes H-H, Kather C. 1996. Green fluorescent protein: applicationsin cell biology. FEBS Lett 389:44–47.

Giese A, Loo M, Rief M, Tran N, Berens M. 1995. Substrates forastrocytoma invasion. Neurosurgery 37:294–301.

Giese A, Rief M, Loo M, Berens M. 1994. Determinants of humanastrocytoma migration. Cancer Res 54:3897–3904.

Giese A, Westphal M. 1996. Glioma invasion in the central nervoussystem. Neurosurgery 39:235–252.

Gladson CL. 1996. Expression of integrin avb3 in small blood vesselsof glioblastoma tumors. J Neuropathol Exp Neurol 55:1143–1149.

Goldbrunner R, Haugland H, Klein C, Kerkau S, Roosen K, Tonn J.1996. ECM dependent and integrin mediated tumor cell migrationof human glioma and melanoma cell lines under serum-free condi-tions. Anticancer Res 16:3679–3687.

Gubin AN, Reddy B, Njoroge JM, Miller JL. 1997. Long-term, stableexpression of green fluorescent protein in mammalian cells. Bio-chem Biophys Res Commun 236:347–350.

Harreveld AV. 1966. Extracellular space in the central nervous system.Proc Koninklijke Nederlandse Akademie Wetenschappen - Series C:Biol Med Sci 69:17–21.

Hendrix M, Seftor E, Seftor R, Fidler I. 1987. A simple quantitativeassay for studying the invasive potential of high and low humanmetastatic variants. Cancer Lett 38:137–147.

Hinek A. 1994. Nature and the multiple functions of the 67-kDelastin-/laminin binding protein. Cell Adhes Commun 2:185–193.

Hinek A. 1996. Biological roles of the non-integrin elastin/lamininreceptor. Biol Chem 377:471–480.

Hinek A, Boyle J, Rabinovitch M. 1992. Vascular smooth muscle celldetachment from elastin and migration through elastin is promotedby the chondroitin sulfate-induced shedding of the 67 kD elastinbinding protein. Exp Cell Research 203:344–353.

Hinek A, Rabinovitch M, Keeley F, Okamura-Oho J, Callahan J. 1993.The 67 kD elastin/laminin-binding protein is related to an enzymati-cally inactive, alternatively-spliced for of b-galactosidase. J ClinInvest 91:1198–1205.

Jung S, Rutka JT, Hinek A. 1998. Tropoelastin and elastin degrada-tion products promote proliferation of human astrocytoma cell lines.J Neuropathol Exp Neurol 57:439–448.

Kanemoto T, Martin G, Hamilton T, Fridman R. 1991. Effects ofsynthetic peptides and protease inhibitors on the interaction of ahuman ovarian cancer cell line (NIH:OVCAR-3) with a reconsti-tuted basement membrane (Matrigel). Invasion Metastasis 11:84–92.

Koochekpour S, Merzak A, Pilkington G. 1995. Extracellular matrixproteins inhibit proliferation, upregulate migration and inducemorphological changes in human glioma cell lines. Eur J Cancer31A:375–380.

Kramer R, Bensch K, Wong J. 1986. Invasion of reconstituted base-ment membrane matrix by metastatic human tumor cells. CancerRes 46:1980–1989.

Leake D, Hornebeck W, Brechemier D, Robert L, Peters T. 1983.Properties and subcellular localization of elastase-like activities ofarterial smooth muscle cells in culture. Biochim Biophys Acta761:41–47.

McComb RD, Bigner DD. 1985. Immunolocalization of laminin neo-plasms of the central and peripheral nervous systems. J Neuro-pathol Exp Neurol 44:242–253.

McComb RD, Moul JM, Bigner DD. 1987. Distribution of type IVcollagen in human gliomas: Comparison with fibronectin and glioma-mesenchymal matrix glycoprotein. J Neuropathol Exp Neurol 46:623–633.

McKeever PE, Fliegel SEG, Varani J. 1986. Products of cells culturedfrom gliomas IV: Extracellular matrix proteins of gliomas. Int JCancer 37:867–874.

McKeever PE, Varani J, Papadoupolos SM, Wang M, McCoy P. 1995.Products of cells from gliomas IX: evidence that two fundamentallydifferent mechanisms change extracellular matrix expression bygliomas. J Neuro-Oncol 24:267–280.

Mecham R. 1997. Elastic Fibers. In: Crystal RG, JB West, editors. TheLung. Philadelphia: Lippincott-Raven Publishers. p 729–736.

Mecham R, Hinek A, Entwistle R, Wrenn D, Griffin G, Senior R. 1989.Elastin binds to a multifunctional 67-kilodalton peripheral mem-brane protein. Biochemistry 28:3716–3722.

Mecham RP, Hinek A. 1995. Non-integrin laminin receptors. In: TimplR, editor, Laminins. New York: Academic Press. p 159–183.

Nicholson C. 1995. Extracellular space as a pathway for neuron-glialcell interactions. In: Ransom BR, Kettermann H, editors: Neuroglia.New York: Oxford University Press. p 387–397.

Parsons D. 1993. Tumor cell interactions with stromal elastin and typeI collagen: the consequences of specific adhesion and proteolysis.Tumour Biol 14:137–143.

Privitera S, Prody CA, Callahan JW, Hinek A. 1998. The 67-kDaenzymatically inactive alternatively spliced variant of beta-galactosidase is identical to the elastin/laminin binding protein. JBiol Chem 273:6319–6326.

Reith A, Bjerkvig R, Rucklidge G. 1994. laminin: a potential inhibitorof rat glioma cell invasion in vitro. Anticancer Res 14:1071–1076.

Rosales C, O’Brien V, Kornberg L, Juliano R. 1995. Signal transduc-tion by cell adhesion receptors. Biochem Biophys Acta 1242:77–98.

Ruoslahti E. 1996. Brain extracellular matrix. Glycobiology 6:489–492.

Rutka J, Aprodaca G, Stern R, Rosenblum M. 1988. The extracellularmatrix of the central and peripheral nervous systems: structure andfunction. J Neurosurg 69:155–170.

Rutka JT, Giblin JR, Apodaca G. 1987a. Inhibition of growth andinduction of differentiation in a malignant human glioma cell line bynormal leptomeningeal extracellular matrix proteins. Cancer Res47:3515–3522.

Rutka JT, Giblin JR, Dougherty DV, Liu HC, McCulloch JR, et al.1987b. Establishment and characterization of five cell lines derivedfrom human malignant gliomas. Acta Neuropathol 75:92–103.

Rutka JT, Giblin JR, Hoifodt HK, Dougherty DV, Bell CW, et al. 1986a.Establishment and characterization of a cell line from a humangliosarcoma. Cancer Res 46:5893–5902.

Rutka JT, Hubbard SL, Fukuyama K, Matsuzawa K, Dirks PB, BeckerLE. 1994. Effects of antisense glial fibrillary acidic protein comple-mentary DNA on the growth, invasion, and adhesion of humanastrocytoma cells. Cancer Res 54:3267–3272.

Rutka JT, Kleppe-Hoifodt H, Emma DA, et al. 1986b. Characterizationof normal human brain cultures: Evidence for the outgrowth ofleptomeningeal cells. Lab Invest 55:71–85.

Rutka JT, Matsuzawa K, Hubbard SL, Fukuyama K, Becker LE, et al.1995. Expression of TIMP-1, TIMP-2, 72- and 92-kDa type IVcollagenase transcripts in human astrocytoma cell lines: correlationwith astrocytoma invasiveness. Int J Oncol 6:877–884.

Rutka JT Myatt CA, Giblin JR. 1987c. Distribution of extracellularmatrix proteins in primary human brain tumours: an immunohisto-chemical analysis. Can J Neurol Sci 14:25–30.

Senior R, Hinek A, Griffin G, Crauch E, and Mecham RP. 1989a.Neutrophils are chemotactic to type IV collagen and its 7S domainand contain 67 kDa type IV collagen binding protein with galactolec-tin properties. Am J Resp Cell Molec Biol 1:479–487.

Senior R, Roman J, Hinek A, and Mecham R. 1989b. Extracellularmatrix components as mediators of inflammation. In: PM Hansonand PC Murphy, editors. Handbook of inflammation, vol. 6, media-tors of inflammatory process. New York: Elsevier Science.

Stetler-Stevenson WG, Krutzsch HC, Liotta LA. 1989. Tissue inhibitorof metalloproteinase(TIMP2): a new member of metalloproteinaseinhibitor family. J Biol Chem 264:17374–17378.

188 JUNG ET AL.

Stoppini L, Buchs P-A, Muller D. 1991. A simple method for organo-typic cultures of nervous tissue. J Neurosci Methods 37:173–182.

Takahashi S, Seifter S, Yang F. 1973. A new radioactive assay forenzymes with radioactivity using reduced tritiated elastin. BiochimBiophys Acta 327:138–145.

Tanaka M, Tomita A, Yoshida S, Yano M, Shimizu H. 1994. Observa-tion of the highly organized development of granule cells in ratcerebellar organotypic cultures. Brain Res 641:319–327.

Timar J, Diczhazi C, Ladanyi A, Raso E, Hornebeck W, et al. 1995.Interaction of tumour cells with elastin and the metastatic pheno-type. Ciba Found Symp 192:321–335.

Tysnes B, Larsen L, Ness G, Mahesparan R, Edvardsen K, et al. 1996.Stimulation of glioma-cell migration by laminin and inhibition byanti-alpha3 and anti-beta1 integrin antibodies. Int J Cancer 67:777–784.

Uhm JH, Dooley NP, Villemure J-G, Yong VW. 1996. Glioma invasionin vitro: regulation by matrix metalloprotease-2 and protein kinaseC. Clin Exp Metast 14:421–433.

Uhm JH, Dooley NP, Villemure JG. 1997. Mechanism of gliomainvasion: role of matrix-metalloroteinases. Can J Neurol Sci 24:3–15.

Venstrom K, Reichardt L. 1993. Extracellular matrix 2: role ofextracellular matrix molecules and their receptors in the nervoussystem. FASEB J 7:996–1003.

Ventimiglia JB, Wikstrand CJ, Ostrowski LE. 1992. Tenascin expres-sion in human glioma cell lines and normal tissues. J Neuroimmu-nol 36:41–55.

Yusa T, Blood CH, Zetter BR. 1989. Tumor cell interaction with elastin:implications for pulmonary metastasis. Am Rev Repir Dis 140:1458–1462

189ASTROCYTOMA ATTACHMENT TO ELASTIN