Specific degradation of subendothelial matrix proteoglycans by brain-metastatic melanoma and brain endothelial cell heparanases

JOURNAL OF CELLULAR PHYSIOLOGY 172:334–342 (1997)

Specific Degradation of SubendothelialMatrix Proteoglycans by Brain-Metastatic

Melanoma and Brain EndothelialCell Heparanases

DARIO MARCHETTI*Department of Tumor Biology, The University of Texas M.D. Anderson Cancer Center,

Houston, Texas

One of the many features of the malignant phenotype, in vitro and in vivo, iselevated heparanase production and activity. Using in vitro model systems, weexamined the capacity of murine (B16B15b) and human (70W) brain-metastaticmelanoma cells to degrade the subendothelial matrix produced by endothelialcell monolayer cultures. B16B15b and 70W melanoma cells solubilized sulfatedmatrix proteoglycans at levels significantly higher than their parental lines (B16F1,MeWo). Sulfated matrix proteoglycans were rich in heparan sulfate (HSPGs), withminor amounts of chondroitin and dermatan sulfates. When matrix HSPGs weretreated with pronase and alkaline borohydride to cleave the core proteins, theresulting glycosaminoglycan chains (GAGs) had an estimated Mr of Ç2.7 1 104

Da, with a minor subpopulation possessing an Mr of Ç4.5 1 104 Da. After theirincubation with brain-metastatic melanoma cells, new HS fragments with lowerMr estimated at Ç9 1 103 Da were detected. This confirms action in these cellsof heparanase, which is capable of cleaving GAGs at specific intrachain sitesand releasing fragments of a relatively high Mr. The pattern of HSPG degradationby brain-metastatic melanoma cells differed from that of less metastatic parentalcells or cells metastatic to organs other than the brain. Moreover, supraadditivelevels of heparanase activity were found when brain endothelial cells were coin-cubated with brain-metastatic melanoma cells in equicellular amounts. Coopera-tive interactions between heparanases from tumor and endothelial sources in theinvasion process are suggested and their potential mechanisms discussed. J.Cell. Physiol. 172:334–342, 1997. q 1997 Wiley-Liss, Inc.

Several studies on the degradation of heparan sulfate sues, blood-borne metastatic tumor cells arrested incapillaries must invade both the endothelial cell layer(HS), as component of glycosaminoglycan chainsand the subendothelial extracellular matrix (ECM)(GAGs) of heparan sulfate proteoglycans (HSPGs),(Nicolson, G. L., 1989). This is thought to occur by ahave demonstrated an excellent correlation betweencombination of enzymatic and mechanical destruction,high heparanase activity and the metastatic potentialwith the former considered far more important (Jonesof various murine and human melanoma andand De Clerck, 1982; Liotta et al., 1982; Terranova etlymphoma sublines (Marchetti et al., 1993, 1996; Mar-al., 1986; Rifkin et al., 1984). Because subendothelialchetti and Nicolson, 1997; Nakajima et al., 1986; Vlo-matrix is composed of collagens, fibronectins, laminin,davsky et al., 1983; Benezra et al., 1992; Cohen et al.,and HSPGs (Iozzo and Murdoch, 1996), its destruction1994). Furthermore, incubation of brain-metastaticby invading cells must require a variety of hydrolasesmelanoma cells with members of a family of neuro-that act either simultaneously or sequentially to solubi-trophic factors, the neurotrophins (NTs), results in thelize the matrix and to allow its penetration by meta-increased release of a heparanase (Marchetti et al.,static cells. Certainly, as HSPGs’ degradative enzymes,1993, 1996) capable of degrading a subpopulation ofheparanases must play a crucial role in these processescell-surface HS molecules (Marchetti et al., 1996). Pre-

vious studies have also shown that NTs behave as para-crine growth factors regulating the invasion of meta-static melanoma cells into the brain (Marchetti et al., Contract grant sponsor: National Institutes of Health; Contract

grant number: R29-CA64178.1995; Herrmann et al., 1993; Menter et al., 1995) andthat heparanase is a key enzyme responsible for this *Correspondence to: Dr. Dario Marchetti, Department of Tumor

Biology, Box 108, The University of Texas M.D. Anderson CancerNT-mediated invasive capacity (Marchetti et al., 1993,Center, 1515 Holcombe Blvd., Houston, TX 77030.1996).

To gain access to the surrounding extravascular tis- Received 15 November 1996; Accepted 11 April 1997

q 1997 WILEY-LISS, INC.

JCP-511D/ 8925$$511d 07-21-97 21:13:30 wlcpa W Liss: JCP

ECM PROTEOGLYCANS AND CELLULAR HEPARANASES 335

(Marchetti et al., 1993, 1996; Nakajima et al., 1983, MeWo cells (parental MeWo, brain-metastatic 70W) ofless than eight passages from an original frozen stock1986; Vlodavsky et al., 1982, 1983). Melanoma he-

paranase has been shown to be an endo-b-D-glucuroni- were grown to subconfluence in a 1:1 (v/v) mixture ofDMEM/F-12 medium supplemented with 5% FBS asdase capable of degrading HS chains at specific intra-

chain sites (Nakajima et al., 1983, 1984, 1986; Bar-Ner reported previously (Marchetti et al., 1993). MBE cellswere grown in dishes coated in gelatin (0.5%) and con-et al., 1985). To study tumor-endothelial cell interac-

tions and extravasation as related to heparanase ac- taining DMEM/F-12 plus 5% FBS and 50 mg/ml endo-thelial cell growth factor (ECGF; Biomedical Technolo-tion, we have used an in vitro model consisting of endo-

thelial cell monolayers known to produce ECM (Gos- gies Inc., Stoughton, MA) reconstituted in DMEM/F-12and 1% FBS. The cells were then split with trypsinpodarowicz et al., 1980; Vlodavsky et al., 1982). Using

a metabolically 35S-labeled subendothelial matrix ob- (0.25%)-EDTA at an optimal 1:4 ratio. When cellsreached confluence, the MBE-conditioned mediumtained from cloned murine brain endothelial cells

(MBE), we have extracted and characterized [35S]HS (CM) was drawn off and replaced with DMEM/F-12 and0.5% FBS. MBE-CM was then centrifuged for 1 hr atchains from corresponding [35S]HSPGs. These purified

HS chains were then used as substrates for heparan- 25,000 g and filtered through 0.22 mm filters. Controlcalf bovine aortic endothelial cells (BAE) were grown inase-containing cellular extracts of tumor and endothe-

lial origin. In the present study, we used this model to the presence of FBS and basic fibroblast growth factor(bFGF; Upstate Biotechnology Inc., Lake Placid, NY)compare the ability of highly brain-metastatic mela-

noma cells, both murine (B16B15b) and human (70W), as described previously (Gospodarowicz et al., 1976).and MBE cells to degrade ECM HSPGs when seeded

Metabolic labeling of endothelialdirectly on top of a labeled subendothelial matrix de-cell monolayersnuded of its endothelial cells. We observed that

B16B15b and 70W cells hydrolyzed the subendothelial Preconfluent cultures of cloned endothelial cells[brain (MBE) and control aortic (BAE)] were incubatedsulfated HS chains into lower Mr fragments to a much

higher extent than 1) their parental cell lines (B16F1, with Na2[35S]SO4 at 50 mCi/ml in 35-mm dishes for 72hr. This value was chosen following time-dependentMeWo), 2) cells metastatic to organs other than brain

(B16F10), and 3) brain endothelial (MBE) cells. In ev- optimization studies, because a difference in the pro-portion of sulfated material in the medium with label-ery case, the degradation appeared to be the result of

heparanase activity. Moreover, a supraadditive effect ing time was observed. At the end of the incubationperiod, the cultures were rinsed three times with pre-was found when both the tumor and the endothelial

cell types were incubated together with sulfated ECM warm medium and the subendothelial matrix was iso-lated. Cells were detached from culture dishes and col-HS chains from corresponding HSPGs.lected by centrifugation at 47C. Tumor, endothelial cellMATERIALS AND METHODSsuspensions (2 ml), or medium alone were added toeach dish and incubated at 377C, and at selected timesMaterialsrepresentative dishes, in triplicate, were taken forHeparan and chondroitin sulfates (HS and CS-A, -B,analysis. The medium was removed from each dish andand -C), D-saccharic acid 1,4-lactone (SAL), Sepharosecentrifuged at 40,000 g for 30 min at 47C. Macromolecu-molecular weight (M.W.) markers, cetylpyridiniumlar radioactivity present in the supernatant was deter-chloride (CPC), 3-(3-chloroamidopropyl)-dimethylam-mined as follows: up to 100 ml of the centrifuged me-monio-1-propane sulfonate (CHAPS), phenylmethyl-dium was spotted onto Corning GF/A glass filters (2.4sulphonylfluoride (PMSF), N-ethylmaleimide (NEM),cm) and allowed to dry. 35S-labeled GAGs were assayedNonidet-P-40 (NP-40), sodium dodecyl sulfate (SDS),using a modification of the method of Glimelius et al.bovine serum albumin (BSA), and ethylenediaminetet-(1978).raacetic acid (EDTA) were acquired from Sigma Chemi-

cal (St. Louis, MO). [35S]sulfate (43 Ci/mg) was pur- Matrix isolation and characterizationchased from ICN Biochemical (Irvine, CA). Chondroi- of proteoglycans and GAGstinases AC (Arthrobacter aurescens; EC 4.2.2.5), ABC

Endothelial cells were plated at an initial density of(Proteus vulgaris; EC 4.2.2.4), heparitinase (Flavobac-41 104 cells/35-mm dish and maintained in logarithmicterium heparinum; EC 4.2.2.8), and hyaluronidasegrowth conditions. Following exposure with Na2[35S]-(Streptomyces hyalurolyticus; EC 4.2.2.1) were fromSO4, cell monolayers were rinsed and incubated withSeikagaku America, Inc. (Ijamsville, MD). Sepharose5 mM Tris-HCl buffer, pH 7.5, at 377C for 10 min. Cellswas obtained from Pharmacia Fine Chemicals (Upp-were allowed to swell in 20 mM NH4OH for 5 min atsala, Sweden), and Bio-Gel P and A-5M polyacrylamide257C with monitoring of lysis under a microscope. Nu-gels were supplied by Bio-Rad Laboratories (Hercules,clei and cytoskeletons were removed by extensiveCA). Fetal bovine serum (FBS) and Dulbecco’s modifiedwashing with DPBS containing 5% FBS and then equil-Eagle medium (DMEM) were purchased from GIBCOibrated at 377C in medium plus 10% FBS. Proteogly-(Grand Island, NY), guanidine hydrochloride fromcans were then extracted from the subendothelial ma-GIBCO-BRL (Gaithersburg, MD), and pronase fromtrix by incubation overnight at 47C with extractionBoehringer-Mannheim (Indianapolis, IN). Suraminbuffer (4 M guanidine-HCl, 20 mM Tris, 1% CHAPS, 5was a generous gift from Dr. Motowo Nakajima (No-mM EDTA, pH 6.0, and protease inhibitors) (Hasselvartis Pharma, Takarazuka, Japan).et al., 1980). To determine the exact composition of

Cells and tissue culture [35S]HSPGs obtained from ECM, the digested materialwas extensively dialyzed against ddH20 and passedMurine melanoma B16 cells (parental F1, metastatic

F10, and brain-metastatic B15b) and human melanoma through a desalting PD-10 column to collect proteogly-


MARCHETTI336

cans eluted at Vo of the column. Sepharose CL columns(CL-2B, -4B, -6B, 1.1 1 100 cm; Pharmacia Biotech,Uppsala, Sweden) performed under dissociating condi-tions (4 M guanidine-HCl in 0.1 M Na acetate, pH 5.5,at a flow rate of 5 ml/hr) were used in order to deter-mine the approximate molecular mass (Mr) of sulfatedECM-released material. Fractions corresponding to thepeak were then combined and digested with chondroi-tinases ABC and/or AC (10 U/ml each) in a solutioncontaining 50 mM Tris, 50 mM Na citrate, and BSA(100 mg/ml), pH 7.4. Chondroitinase-resistant sampleswere separated from digested material by Sephadex G-50 column chromatography and then were character-ized directly by filtration on a Sepharose CL-6B column(0.7 1 35 cm) with 4 M guanidine-HCl, 0.1 M Na ace-tate, pH 5.5, at a flow rate of 5.5 ml/hr. The excludedvolume, Vo, was marked by blue dextran (Mr Ç2.0 1106) and the total included volume, Vt , by vitamin B12(Mr Å 1.35 1 103). The B12-marked peak coeluted with[35S]SO4

02 and accurately marked the Vt for each col-umn. Free GAGs were obtained by pronase digestion

Fig. 1. Characterization of MBE subendothelial matrix sulfatedof the proteoglycans and by alkaline cleavage (Carlson,GAGs by Sephadex G-50 filtration as undigested matrix GAG (solidD. M., 1968; Cifonelli, J. A., 1968) as described below.circles) or heparanase-treated GAG (open circles). Material obtainedThe HS content in the isolated GAGs was determined from the extracellular matrix was mostly chondroitinase resistant

by nitrous acid degradation (Shively and Conrad, (solid triangles). Presence of chondroitinase-sensitive sulfated proteo-glycans was confirmed by running CS samples in parallel and subse-1976). The Mr of intact and degraded GAG was thenquent chondroitinase digestion (open triangles; refer to Materials anddetermined by comparison to radiolabeled CS chains ofMethods). Radiolabeled ECM HSPG heparan sulfate content was de-defined molecular weight (Wasteson, 1971). termined by sensitivity to nitrous acid deamination (Shively and Con-rad, 1976). Vo and Vt of the column were at fractions 15 and 50,Solubilization of subendothelial matrix HS respectively.

components by cellular heparanasesSubendothelial matrices isolated from 35S-labeled en-

dothelial cells monolayers were incubated with eitherhuman melanoma cells of various brain-metastatic concentration was determined using the Coomassiecapabilities (parental MeWo, brain-metastatic 70W), blue protein assay protocol (Pierce, Rockford, IL). He-murine melanoma cells (parental B16F1, metastatic paranase assays involving purified [35S]HS were per-B16F10 and brain-metastatic variant B16B15b), or formed as previously described (Marchetti et al., 1996).MBE. Labeled matrices were incubated with tumor or Confirmatory of this assay were also determinationsendothelial cells for 12–72 hr at 377C. The released 35S of heparanase activity by high-speed gel permeationlabel was then analyzed (Shively and Conrad, 1976; chromatography (Marchetti et al., 1993, 1996).Gallagher et al., 1990).

RESULTSTo obtain free GAGs, the proteoglycan samples wereeither incubated overnight with pronase (10 mg/ml) or Isolation and characterization oftreated with alkaline borohydride (b-elimination) at subendothelial matrix from brain457C in the presence of 0.05 M NaOH and 1 M sodium endothelial cellsborohydride for 24 hr, followed by neutralization withacetic acid (Kjellen and Lindahl, 1991; Gallagher et al., Hypotonic treatment of MBE monolayers followed by

extraction with NP-40 detergent (Gospodarowicz et al.,1990; Marchetti et al., 1996). Precipitates were col-lected, supernatants were dialyzed with ddH20, radio- 1980; Vlodavski et al., 1980; Kramer et al., 1980, 1982)

was successfully used to isolate brain endothelial ECM.activity was determined, and specific activity was cal-culated. The medium was adjusted to 4 M guanidine- Approximately 60–70% of incorporated label (Na2[35S]-

SO4) was recovered from the culture medium. The re-HCl and chromatographed on Sepharose columns tomonitor differences between the various elution pro- mainder was associated in the cell monolayer, primar-

ily the endothelial ECM. Trypsin treatment removedfiles. The released GAG chains were then characterizedand parallel control experiments performed to monitor virtually all the existing sulfate label, suggesting that

most of the GAGs were extracellular. The culture me-the spontaneous release of [35S]HSPGs into the me-dium. dium, cell layers, and isolated matrix were subjected to

exhaustive pronase digestion followed by b-elimination.Preparation of heparanase-containing cellular According to chemical (nitrous acid deamination;

extracts and heparanase assays Shively and Conrad, 1976) and enzymatic (incubationwith chondroitinases and hyaluronidase; Jackson et al.,Subconfluent cells (2 1 108) were harvested and solu-

bilized in 50 ml of 50 mM Tris-HCl, pH 7.5, containing 1991) methods, HS was the major component of bothsecreted (70%) and ECM-associated (84%) sulfated pro-1 mM PMSF, 5 mM NEM, 0.05% sodium azide, 0.5%

Triton X-100 at 47C for 30 min. The cell extracts were teoglycans, with minor variations resulting from thetwo methodologies or endothelial cell lines employed.centrifuged at 12,000g for 30 min at 47C, and protein



Fig. 3. Solubilization of 35S-labeled subendothelial matrix compo-nents by brain endothelial and brain-metastatic cells. Sulfated ECMwas exposed either to intact brain endothelial (MBE) cell monolayers(solid triangles), brain-metastatic B16B15b cells (open triangles), ora combination of the two in equicellular amounts (solid squares). Sam-Fig. 2. Sepharose CL-6B chromatographic analysis of brain endothe-ples were incubated at 377C at indicated times, and the total amountlial and ECM [35S]HSPGs showing elution profiles of subendothelialof radioactivity released into the medium was determined. Each pointECM [35S]HSPGs before (open triangles) and after (solid squares)represents the average of three experiments, where the SD was lessalkaline borohydride hydrolysis. The HS nature of GAG chains wasthan 10% of raw data. Amounts of radioactivity spontaneously re-confirmed by parallel nitrous acid degradation (solid circles). Alsoleased from ECM were subtracted from data shown.shown are determinations for [35S]HSPGs secreted from brain endo-

thelial cell (MBE) monolayers before (solid triangles) and after (opensquares) alkaline borohydride analysis and nitrous acid degradation(open circles). Vo and Vt of the column were at fractions 18 and 53,respectively. an elution profile at Vt , indicating that they were HS.

The same was true for [35S]HSPGs secreted from brain-endothelial cell monolayers (Fig. 2). Free GAGs hadelution profiles distinct from their HSPGs (Fig. 2).Chondroitin and dermatan sulfates made up the re-

mainder in each case.Solubilization of subendothelial matrix sulfatedFigure 1 shows the matrix GAGs characterized by

proteoglycans by brain endothelial cellsSephadex G-50 filtration. Most of the radiolabeled ma-and brain-metastatic cellsterial obtained from the ECM and processed to obtain

free GAGs eluted with a broad peak starting at Vo of Subendothelial matrices isolated from prelabeled en-dothelial monolayers and incubated with brain-meta-the column; smaller components eluted at a region close

to the Vt of the column. The high-Mr component eluting static B16B15b melanoma cells produced a gradual,time-dependent, and significant solubilization of la-at Vo was by the most part chondroitinase resistant

and nitrous acid sensitive and so was identified as beled material for up to 72 hr of exposure, as notedafter samples were run on a Sepharose 6B column (Fig.mainly HS. The low-Mr component (peak at fraction

42) eluting close to the Vt (fraction 50) region was chon- 2). The same was true when MBE cells in logarithmicgrowth conditions were used. However, in the latterdroitinase sensitive but nitrous acid resistant and so

was identified as chondroitin sulfate (CS; Jackson et case, the amount of radioactivity released by the endo-thelial cells was less than that obtained in the presenceal., 1991). Confirmation of these results was accom-

plished by analyzing radiolabeled CS in parallel follow- of tumor cells, whether murine (B16B15b) or human(70W). Interestingly, coincubation of [35S]ECM withing incubation with chondroitinase ABC. The elution

profile of these samples coincided with that of the low- MBE and B16B15b (1/1 ratio) resulted in a supraaddi-tive increase in soluble radioactivity (Fig. 3). IsolatedMr component (Fig. 1). Interestingly, when ECM radio-

labeled material was exposed to partially purified mu- matrix in the absence of cells (control) was spontane-ously solubilized at an almost constant rate, possiblyrine melanoma heparanase (Marchetti et al., 1997), no

change in its elution position was detected on Sephadex because the subendothelial matrix is not a stable struc-ture. The proportion of sulfate-containing macromole-chromatography.

Sulfated subendothelial ECM and endothelial-se- cules naturally released into the medium correspondedto Ç0.03–0.05%/hr at 377C. These values were essen-creted GAGs were characterized next. Both GAG popu-

lations underwent chondroitinase ABC treatment to se- tially at background levels when compared to values ofsoluble radioactivity obtained for cells incubated on toplect for GAG chains refractory to its action. When chon-

droitinase-resistant GAG chains were degraded with of sulfated subendothelial ECM. After the incubationperiod, greater than 95% of the [35S]SO02

4 label releasednitrous acid and then analyzed by Sepharose 6B col-umn chromatography, their deamination products had by all cells was precipitable with CPC (Roden et al.,


MARCHETTI338

TABLE 1. Release of 35S-labeled HSPGs from brain endothelialmatrix in the absence or presence of tumor/endothelial cells

Degraded HSPGCell line (%)1

Human MeWo, parental 6.1 { 1.92

Human 70W, brain metastatic 29.6 { 2.0Murine B16F1, parental 7.5 { 0.8Murine B16B15b, brain metastatic 32.4 { 4.5Murine MBE, brain endothelial 7.6 { 2.3Control 0.03 { 0.011The numbers represent the mean { SD of four independent experiments/celllines.2Analysis by Sepharose 6B of degraded [35S]HSPGs released by the various celllines when plated directly on the subendothelial [35S]ECM for 48 h at 377C.Percentages refer to radioactivity values of degraded HSPGs elution peaks asseparated by Sepharose 6B (see Materials and Methods) compared to total con-stant radioactivity amounts loaded for each condition tested.

TABLE 2. Sepharose 6B approximate Mr of 35SO024 -labeled GAG

chains released from subendothelial ECM

Cell line/treatment Approximate Mr1

Fig. 4. Solubilization of ECM [35S]HSPGs by brain endothelial andbrain-metastatic melanoma cells as analyzed on a Sepharose 6B col-Human 70W, brain metastatic 9 1 103

umn. Subendothelial ECM [35S]HSPGs were incubated (72 h at 377C)Murine B16B15b, brain metastatic 9 1 103

with brain endothelial MBE cells (open triangles), parental B16F1Pronase (10 mg/ml) 3.1 1 104

cells (solid triangles), highly brain-metastatic B16B15b cells (solidAlkaline borohydride (b-elimination) 2.7 1 104

squares), and a combination of MBE and B16B15b cells in equicellularNitrous acid hydrolysis 1.5 1 103

amounts (open squares). Soluble-labeled degradation products re-1Estimations made on the basis of Kav values obtained for chondroitin sulfate leased into the incubation medium were analyzed by gel filtration,chains of known molecular weight. loading a constant amount of radioactivity on a Sepharose 6B column,

and digested HSPGs’ degradation profiles were obtained as shown.Migration of spontaneously released ECM HSPGs (solid circles) andVo and Vt determinations are shown.

1972). Thus, the tumor cells degraded HS componentsof the subendothelial matrix, but the degradation wasonly partial and not to very low-Mr fragments. HS deg-

static B16B15b and 70W had an Mr of Ç9 1 103 Da.radation was maximal when cells (either tumor or en-Precipitation with a solution of 1% CPC in 0.05 M NaCldothelial) were in logarithmic growth phase. The levelsrevealed that practically all [35S]SO02

4 -labeled solubleof radioactivity released were as much as 65% lower inmaterial was precipitable, regardless of the absence orthe presence of confluent B16B15b and MBE cells alonepresence of the various cell types. 35S-labeled materialor together (data not shown).released by chondroitinases (Fig. 1) was not. Therefore,

Characterization and analysis of matrix- even in the case of brain-metastatic HS degradation,associated sulfated proteoglycans solubilized the released [35S]SO02

4 -labeled proteoglycans fragmentsby brain endothelial/brain-metastatic were large enough to be precipitated with CPC.

melanoma cellsCharacterization and specificity ofBoth murine and human melanoma cells adhered degraded glycosaminoglycanswithin 30 min to the subendothelial ECM and re-

mained attached for at least 4 days with no loss of cell Levels of heparanase activity were examined em-ploying a specific heparanase assay developed in ourviability. When parental B16F1, MeWo, and MBE

cells were allowed to interact with metabolically laboratory (Marchetti et al., 1996). Heparanase frombrain-metastatic B16B15b cell extracts was active on35SO02

4 -labeled subendothelial ECM and then their re-leased products analyzed by chromatographic and elec- HS derived from labeled ECMs of MBE cells (Fig. 5).

These sulfated HS chains migrated as a major class withtrophoretic methods, similar elution patterns were ob-tained, with MBE possessing a detectable HSPG frag- an Mr of Ç2.7 1 104 Da. A minor class with an Mr of

Ç4.5 1 104 Da was also detected. At present, it is notmentation (Table 1). In contrast, the percentage ofdegraded HSPGs by highly brain-metastatic B16B15b known whether the two subpopulations are related. It

is possible that the appearance of the minor populationand 70W cells was significantly higher, and smallerfragments were produced (Table 2) whether in the ab- is due to self-aggregation of HS molecules or the electro-

phoretic system used (Fransson et al., 1981; Gallaghersence or in the presence of dissociative conditions (4 Mguanidine hydrochloride). Interestingly, a supraaddi- et al., 1990; Ernst et al., 1995). The GAGs examined

were confirmed to be HS in nature by the following ob-tive HSPG digestion peak was observed following coin-cubation of brain endothelial with brain-metastatic servations: 1) their resistance to digestion with chon-

droitinases ABC and AC, hyaluronidase, and proteasescells (1/1 ratio; Fig. 4). Based on these elution profilesand comparison of the experimental Kav values with (Table 2); 2) their cleavage by nitrous acid (Fig. 2; Gal-

lagher et al., 1990); 3) their sensitivity to heparitinasesthose determined for radiolabeled CS fragments of me-dian known Mr (Wasteson, 1971), it was estimated that as detected by Bio-Gel P-2 and P-4 column chromatogra-

phy of digested products (data not shown; Sasisekharanthe 35S-labeled HS fragments released by brain-meta-



(Fig. 5). Incubation of endothelial ECM [35S]HS withheparanase-containing extracts and suramin (100 mM),a potent inhibitor of melanoma heparanase (Nakajimaet al., 1991), resulted in an inhibition of GAG digestion.Inclusion of SAL, an exo-b-glucuronidase inhibitor, orchondroitin-4, or -6 sulfates did not inhibit either thedigestion of endothelial ECM [35S]HS or its migration(Fig. 5). We investigated whether the same degradativeproducts (median Mr ofÇ9 1 103 Da) could be obtainedusing different melanoma cells with distinct metastaticpotentials and experimental metastasis organ prefer-ence. Cellular extracts from B16F1 [low metastatic ca-pability to lung and extrapulmonary sites (Nicolson etal., 1978; Fidler and Nicolson, 1981)] or B16F10 (highmetastatic capability to lung ú lymph nodes ú ovaryú liver; Fidler, 1973) failed to produce the low-Mr deg-radative fragments, demonstrating a heparanase speci-ficity with respect to the activity observed in melanomacells metastatic to the brain compared to melanomacells metastatic to other sites (Fig. 6). This was alsoconfirmed by testing human brain-metastatic 70W cellsand parental, nonbrain-metastatic MeWo cells (datanot shown). Thus, the observed subendothelial ECMHS degradation appeared to be a specific result of aheparanase activity in invasive brain-metastatic tumorcells.

DISCUSSION

We have examined brain-metastatic melanoma andbrain endothelial cellular properties for their ability tosolubilize sulfated proteoglycans (PGs) from subendo-thelial ECM, particularly those proteoglycans pos-sessing GAG chains of the HS type. ECM-producingaortic and brain endothelial cells were incubated withFig. 5. Effects of suramin, SAL, and GAGs on ECM [35S]HS degrada-

tion by melanoma heparanase. The interactions of melanoma he- Na2[35S]SO4, and the subendothelial matrix was iso-paranase with suramin, the potent exoglucuronidase inhibitor SAL lated and characterized for HSPG content and degrada-(20 mM; refer to Materials and Methods for nomenclature), and GAGs tion. The major (70% of secreted PGs, 84% of ECM-[chondroitin-4-sulfate (C4S) and chondroitin-6-sulfate C6S)] were ex-

extracted PGs) GAG constituent of sulfated proteogly-amined using a specific heparanase assay (Marchetti et al., 1996).Agarose gel electrophoresis and autoradiography were subsequently cans is HS, with lesser amounts of chondroitin andperformed. Distinct bands corresponding to purified ECM [35S]HS dermatan sulfates also present. Under dissociative con-components not exposed (lane 1) or exposed (lane 2) to heparanase- ditions, several chromatographic analyses have indi-containing cellular extracts were detected. Cellular extract coincuba-

cated ECM HSPGs to have an apparent Mr of Ç1 1tion with suramin (100 mM; lane 3) effectively prevented heparanase-driven HS degradation, whereas SAL (20 mM; lane 4) or C4S or C6S 106 Da consisting of GAG chains having an estimated(20 mg/assay; lanes 5,6) were not inhibitory. Mr of Ç2.7 1 104 Da, with a minor subpopulation of

Mr Ç 4.5 1 104 Da. Incubation of subendothelial matri-ces with brain-metastatic melanoma cell extracts re-sulted in HS solubilization and the appearance of aet al., 1993); and 4) their precipitation with CPC (0.05%unique new form (MrÇ 91 103 Da) derived by degrada-in 0.6 M NaCl; Roden et al., 1972). In the absence oftion from the original high-Mr proteoglycan population.tumor cells, there was a fairly constant release of labeledThese degraded HS fragments were brain metastasis-material that consisted entirely (ú97%) of high-Mr com-specific; they could not be obtained following incubationponents eluted at Vo via Sephadex G-50 (Fig. 1) or Seph-of subendothelial [35S]HS with extracts from other mel-arose 6B columns (Fig. 2). This product was partiallyanoma cells selected for metastasis to organs otherincluded on Sepharose CL-4B and showed a Kav of 0.40than the brain. Chromatographic and electrophoreticon Sepharose CL-2B. Based on the M.W. ranges for CL-analyses of solubilized sulfated HS chains from suben-2B (7 1 104 to 4 1 107) and CL-4B (6 1 104 to 2 1 107)dothelial matrix following incubation with endothelialand on the elution characteristics of Sephadex M.W.or tumor cells showed that HS digestion proceeds bymarkers run in parallel [Apoferrin (MrÇ4.4 1 105), thy-the cleavage of HS at intrachain sites, supportive of aroglobulin (Mr Ç6.7 1 105), blue dextran (Mr Ç2.0 1heparanase-driven process. When subendothelial ma-106)] and bovine nasal cartilage proteoglycan (Mr Ç 2.5trices isolated from endothelial monolayers were incu-1 106; Hascall and Sajdera, 1970), this corresponds tobated with either MBE or B16B15b cells, a gradual butan Mr Ç 1 1 106.significant solubilization of labeled material occurred.Digestion of ECM HS preparations with heparanase-A supraadditive effect was observed when MBE werecontaining metastatic melanoma cellular extracts re-

sulted in products with a median Mr of Ç9 1 103 Da coincubated with B16B15b, indicating that endothelial


MARCHETTI340

basement membrane and its formidable blood-brain-barrier (Ruoslahti, 1996; Rutka et al., 1988; Stech andNicolson, 1993). We have demonstrated that metastatictumor cells with high brain-colonization potential pos-sess specific abilities to degrade both purified cell-sur-face HS molecules and HS chains from proteoglycansin the basal lamina-like matrix of cultured endothelialcells. These observations might therefore explain howmetastatic tumor cells migrate through the vascularwall and develop metastatic lesions in distant organsites such as the brain.

The distinction in brain-metastatic ability betweenthe parental murine B16F1 line and its B16B15b sub-clone and the human MeWo line and its brain-meta-static 70W subclone has been previously shown to cor-relate with differences in NT-mediated invasive pro-cesses, NT receptor content of these lines, their bindingcapacity to selected NTs, and most importantly, theaugmented regulation of heparanase activity and pro-duction by some members of NT family of neurotrophicfactors (Marchetti et al., 1993, 1995, 1996; Herrmannet al., 1993). These findings have confirmed heparanaseas a key enzyme responsible for NT-augmented inva-sion by brain-metastatic melanoma. Here we show thatthese cells differ in their ability to degrade proteogly-cans in the subendothelial ECM. Of particular interestis the capacity of brain-metastatic B16B15b and 70Wcells to release low-Mr

35SO024 -labeled GAG fragments

from the ECM that are not released by exposing theECM to murine B16F1 or B16F10 or human MeWo orother non-brain-metastatic cells. Similarly, these sul-fated subendothelial ECM preparations were suscepti-ble to degradation when exposed to various purifiedenzymes such as trypsin, hyaluronidase, or chondroi-tinases.

Because HS constitutes a major scaffolding GAG ofthe basement membrane and other ECM as well as astorage depot for active growth factors, its preferentialdegradation by highly metastatic tumor cells may, inconjunction with solubilization of other matrix constit-uents (Jones et al., 1980; Liotta et al., 1980, 1981),facilitates tumor invasion through blood vessels andwithin tissues. Additionally, we recently found that he-paranase and a synthetic peptide called heparan sulfateFig. 6. Brain specificity of heparanase on ECM [35S]HS degradation.

Autoradiography of agarose gel electrophoresis of purified ECM interacting protein (HIP) (Rohde et al., 1996; Liu et al.,[35S]HS components following exposure to cellular extracts from mela- 1996) recognize common sites on cell-surface and ECMnoma cells possessing metastatic capabilities and specific organ colo- HS (Marchetti et al., 1997), with HIP inhibiting he-nization properties. Distinct bands corresponding to purified ECM

paranase binding to HS and its degradation. In light[35S]HS not exposed (lane 1) or exposed (lane 2) to heparanase-con-taining brain-metastatic melanoma (B16B15b) cellular extracts were of these results, heparanase must be regarded not sim-detected. Conversely, coincubation of the same HS components with ply as a nonselective HS degradative enzyme but rathercellular extracts from either low-metastatic (to lung) B16F1 cells (lane as an enzyme recognizing defined ‘‘regions’’ within the3) or high-metastatic (to lungú lymph nodesú ovaryú liver) B16F10

HS chains and capable of interaction with specific HScells (lane 4) did not result in their effective degradation. Cellularextract preparations and heparanase assay conditions were as de- binding proteins (Marchetti et al., 1997).scribed in Materials and Methods. Synergistic heparanase activity was detected in

mixed (endothelial-tumor) cell populations, suggestingheterogeneity of heparanase sources and their interac-tions. Several mechanisms can be invoked to explainand tumor cell heparanases may cooperate during ma-

lignant cell invasion. these observations. Heparanases can be present in bothlatent and active forms, whose ratio may vary de-Glycosaminoglycans have been implicated in various

aspects of tumor cell interactions (Iozzo and Murdoch, pending on the cell type, similar to what is known forother basement membrane degradative enzymes such1996; Iozzo, 1988). GAGs such as HS constitute a major

fraction of the endothelial cells basement membrane, as collagenases (Liotta et al., 1980, 1981; Lazarus,1973; Stetler-Stevenson et al., 1985). Therefore, mixingwhich acts as a barrier against invasion by blood-borne

tumor cells (Murphy and Johnson, 1975). This is partic- two heparanase sources can result in aberrant levelswhen the total activity is measured. Alternatively, dis-ularly true in the case of the brain ECM, with its thick



Fidler, I.J. (1973) Selection of successive tumor lines for metastasis.criminatory levels and activity of natural heparanaseNature, 242:148–149.inhibitors present within the two cell types and/or Fransson, L.A., Havsmark, B., and Sheehan, J.K. (1981) Self-associa-

brain microenvironment can be imagined as regulatory tion of heparan sulfate. Demonstration of binding by affinity chro-matography of free chains on heparan sulfate-substituted agarosefactors following heparanase stimulation. A mixture ofgels. J. Biol. Chem., 256:13039–13043.endothelial and tumor cells would dilute the effect of

Gallagher, J.T., Turnbull, J.E., and Lyon, M. (1990) Heparan sulfatethe inhibitor resulting in supraadditive heparanase proteoglycans. Biochem. Soc. Trans., 18:207–239.levels. Both hypotheses are entirely consistent with our Glimelius, B., Norling, B., Westermark, B., and Wasteson, A. (1978)

Composition and distribution of glycosaminoglycans in culture offindings. Furthermore, we did not observe appreciablehuman normal and malignant glial cells. Biochem. J., 172:443–456.losses of heparanase activity when preparations of par-

Gospodarowicz, D., Moran, J., Braun, D., and Birdwell, C. (1976)tially purified melanoma heparanase were used in spe- Clonal growth of bovine vascular endothelial cells: Fibroblastcific heparanase assays. It is also conceivable that both growth factor as a survival agent. Proc. Natl. Acad. Sci. USA,

73:4120–4124.tumor and vicinal cells in the invading tumor zone par-Gospodarowicz, D., Delgado, D., and Vlodavsky, I. (1980) Permissiveticipate in a more efficient manner in the degradation

effect of the extracellular matrix on cell proliferation in vitro. Proc.of HSPGs, perhaps by the production and release of Natl. Acad. Sci. USA, 77:4098–4098.more than one heparanase enzyme. In addition, the Hascall, V.C., and Sajdera, S.W. (1970) Physical properties and poly-

dispersity of proteoglycan from bovine nasal cartilage. J. Biol.enhanced heparanase activity found in tumors mayChem., 245:4920–4930.arise by several different stimulatory mechanisms, in-

Hassel, J.R., Robey, P.G., Barrach, H.J., Wilczek, J., Rennard, S.I.,cluding interactions of tumor cells with host cells and and Martin, G.R. (1980) Isolation of heparan sulfate containingwith the extracellular environment. Stimulation may proteoglycan from basement membrane. Proc. Natl. Acad. Sci. USA,

77:4494–4498.depend on the specificity of cell types in the tumor andHerrmann, J.L., Menter, D.G., Hamada, J.-I., Marchetti, D., Naka-neighboring host tissues.

jima, M., and Nicolson, G.L. (1993) A functional role for the low-As the search for the regulators of synthesis, release, affinity p75 nerve growth factor receptor in human melanoma pro-and action of tissue heparanases intensifies, more in- gression to the invasive phenotype. Mol. Biol. Cell, 4:1205–1216.

Iozzo, R.V. (1988) Proteoglycans and Neoplasia. Cancer Metast. Rev.,formation on HSPGs’ degradation in physiological re-7:39–50.modeling, tissue repair, and pathological conditions is

Iozzo, R.V., and Murdoch, A.D. (1996) Proteoglycans for the extracel-required. The present investigation show that cultures lular environment: Clues from the gene and protein side offer newof brain vascular endothelial cells possess a specific perspectives in molecular diversity and function. FASEB J., 10:598–

614.heparanase activity capable of cleaving GAG chainsJackson, R., Busch, S.J., and Cardin, A.D. (1991) Glycosaminoglycans:similar to that of brain-metastatic melanoma heparan-

Molecular properties, protein interactions and role in physiologicalase. Certainly these observations warrant further in- processes. Physiol. Rev., 71:481–539.vestigations concerning the precise nature, location, Jones, P.A., and De Clerck, Y.A. (1982) Extracellular matrix destruc-

tion by invasive tumor cells. Cancer Metast. Rev., 1:289–317.and quantities of active vs. latent forms of the enzymeJones, P.A., Neustein, H.B., Gonzales, F., and Bogenmann, E. (1980)and its possible inhibitors.

Invasion of artificial blood vessel wall by human fibrosarcoma cells.Cancer Res., 40:3222–3227.ACKNOWLEDGMENTS Kjellen, L., and Lindahl, U. (1991) Proteoglycans: Structure and inter-actions. Annu. Rev. Biochem., 60:443–473.I thank Dr. Garth L. Nicolson (Institute for Molecu- Kramer, R.H., Gonzalez, R., and Nicolson, G.L. (1980) Metastatic tu-lar Medicine, Irvine, CA) for providing the cell lines mor cells adhere preferentially to the extracellular matrix underly-

and some equipment used in these studies while at ing vascular endothelial cells. Int. J. Cancer, 26:639–645.Kramer, R.H., Vogel, K.G., and Nicolson, G.L. (1982) SolubilizationM. D. Anderson Cancer Center, Dr. Motowo Nakajima

and degradation of subendothelial matrix glycoproteins and proteo-(Novartis Pharma, Takarazuka, Japan) for his gift of glycans by metastatic tumor cells. J. Biol. Chem., 257:2678–2686.the heparanase inhibitor suramin, and Drs. Magnus Lazarus, G.S. (1973) Studies on the degradation of collagen by colla-Hook and David McQuillan (Center for Extracellular genases. In: Lysosomes in Biology and Pathology. J.J. Dingles, ed.

North-Holland Publishing Co., Amsterdam, Vol. 3, pp. 338–364.Matrix Biology, Texas A&M University, Houston, TX)Liotta, L.A., Lanzer, L.W., and Garbisa, S. (1981) Identification offor providing radiolabeled CS and oligosaccharide type IV collagenolytic enzyme. Biochem. Biophys. Res. Commun.,chains as Mr standards. I extend special thanks to Mr. 98:184–190.

William C. Spohn for his technical assistance. Liotta, L.A., Thorgeirsson, V.P., and Garbisa, S. (1982) Role of colla-genase in tumor cell invasion. Cancer Metast. Rev., 1:277–288.

Liotta, L.A., Tryggvason, K., Garbisa, S., Hart, I.R., Foltz, C.M., andLITERATURE CITEDShafie, S. (1980) Metastatic potential correlates with enzymaticdegradation of basement membrane collagen. Nature, 284:65–66.Bar-Ner, M., Kramer, M.D., Schirrmacher, V., Ishai-Michaeli, R.,

Liu, S., Smith, S.E., Julian, J., Rohde, L.H., Karin, N.Y., and Carson,Fuks, Z., and Vlodavski, I. (1985) Sequential degradation of heparanD.D. (1996) cDNA cloning and expression of HIP, a novel cell surfacesulfate in the subendothelial extracellular matrix by highly meta-heparan sulfate/heparin-binding protein of human uterine epithe-static lymphoma cells. Int. J. Cancer, 35:483–491.lial cells and cell lines. J. Biol. Chem., 271:11817–11823.Benezra, M., Vlodavski, I., and Bar-Shavit, R. (1992) Thrombin en-

Marchetti, D., and Nicolson, G.L. (1997) Neurotrophin stimulation ofhances degradation of heparan sulfate in the extracellular matrixhuman melanoma cell invasion: Selected enhancement of heparan-by tumor cell heparanase. Exp. Cell Res., 201:208–215.ase activity and heparanase degradation of specific heparan sulfateCarlson, D.M. (1968) Structures and immunochemical properties ofsubpopulations. In: Advances in Enzyme Regulation. G. Weber, ed.oligosaccharides isolated from pig submaxillary mucins. J. Biol.Pergamon Press, London (in press).Chem., 243:616–626.

Marchetti, D., Menter, D.G., Jin, L., Nakajima, M., and Nicolson, G.L.Cifonelli, J.A. (1968) Reaction of heparitin sulfate with nitrous acid.(1993) Nerve growth factor effects on human and mouse melanomaCarbohydr. Res., 8:233–242.cell invasion and heparanase production. Int. J. Cancer, 55:692–Cohen, I.R., Murdoch, A.D., Naso, M.F., Marchetti, D., Berd, D., and699.Iozzo, R.V. (1994) Abnormal expression of perlecan in metastatic

Marchetti, D., McCutcheon, I., Ross, M., and Nicolson, G.L. (1995)melanoma. Cancer Res., 54:5771–5774.Inverse expression of neurotrophins and neurotrophin receptors atErnst, S., Langer, R., Cooney, C.L., and Sasisekharan, R. (1995) Enzy-the invasion from of human melanoma brain metastases. Int. J.matic degradation of glycosaminoglycans. Crit. Rev. Biochem. Mol.Oncol., 7:87–94.Biol., 30:387–444.

Marchetti, D., McQuillan, D.J., Spohn, W.C., Carson, D.D., and Nicol-Fidler, I.J., and Nicolson, G. L. (1981) Immunobiology of experimentalmetastatic melanoma. Cancer Biol. Rev., 2:171–234. son, G.L. (1996) Neurotrophin stimulation of human melanoma cell


MARCHETTI342

invasion: Selected enhancement of heparanase activity and he- Roden, L., Baker, J.R., Cifonelli, J.A., and Matthews, M.B. (1972)Isolation and characterization of connective tissue polysaccharides.paranase degradation of specific heparan sulfate subpopulations.Methods Enzymol., 28:73–140.Cancer Res., 56:2856–2863.

Rohde, L.H., Julian, J., Babaknia, A., and Carson, D.D. (1996) CellMarchetti, D., Shouchun, L., Spohn, W.C., and Carson, D.D. (1997)surface expression of HIP a novel heparin/heparan sulfate-bindingHeparanase and a synthetic peptide of heparan sulfate interactingprotein of human uterine epithelial cells and cell lines. J. Biol.protein (HIP) recognize common sites on cell surface and extracellu-Chem., 271:11824–11830.lar matrix heparan sulfate. J. Biol. Chem. 272(25):15891–15897.

Ruoslahti, E. (1996) Brain extracellular matrix. Glycobiology, 6:489–Menter, D.G., Herrmann, J.L., Marchetti, D., and Nicolson, G.L.492.(1995) Involvement of neurotrophins and growth factors in brain

Rutka, J.T., Apodaca, G., Stern, R., and Rosenblum, M. (1988) Themetastasis formation. Invasion Metast., 14:372–384.extracellular matrix of the central and peripheral nervous systems:Murphy, M.E., and Johnson, P.C. (1975) Possible contribution of base-Structure and function. J. Neurosurg., 69:155–170.ment membrane to the structural rigidity of blood capillaries. Micro-

Sasisekharan, R., Bulmer, M., Moremen, K.W., Cooney, C.L., andvasc. Res., 9:242–245.Langer, R. (1993) Cloning and expression of heparinase I gene fromNakajima, M., Irimura, T., Di Ferrante, D., Di Ferrante, N., andFlavobacterium heparinum. Proc. Natl. Acad. Sci. USA, 90:3660–Nicolson, G.L. (1983) Heparan sulfate degradation: Relation to tu-3664.mor invasive and metastatic properties of mouse B16 melanoma

Shively, J.H., and Conrad, J.E. (1976) Formation of anhydrous sugarssublines. Science, 220:611–613.in the chemical depolymerization of heparin. Biochemistry,Nakajima, M., Irimura, T., Di Ferrante, N., and Nicolson, G.L. (1984)15:3932–3942.Metastatic melanoma heparanase. J. Biol. Chem., 259:2283–2290.

Steck, P.A., and Nicolson, G.L. (1993) Metastasis to the central ner-Nakajima, M., Irimura, T., and Nicolson, G.L. (1986) Heparanase and vous system. In: Molecular Genetics of Nervous System Tumors. A.tumor metastasis. J. Cell. Biochem., 36:157–167. Levine and H. Schmid, eds. Wiley, New York, pp. 371–379.Nakajima, M., De Chavigny, A., Johnson, C.E., Hamada, J., Stein, Stetler-Stevenson, W.G., Brown, P.D., Onisto, M., Levy, A.T., andC.A., and Nicolson, G.L. (1991) Suramin. J. Biol. Chem., 266:9661– Liotta, L.A. (1985) Tissue inhibitor of metalloproteases-2 (TIMP2)9666. mRNA expression in tumor cell lines and human tissues. J. Biol.Nicolson, G.L., Brunson, K.W., and Fidler, I.J. (1978) Specificity of Chem., 265:13833–13838.arrest, survival and growth of selected metastatic variant cell lines. Terranova, V.P., Hujanen, F.S., and Martin, G.R. (1986) BasementCancer Res., 38:4105–4111. membrane and the invasive activity of metastatic tumor cells. J.

Nicolson, G.L. (1989) Metastatic tumor cell interactions with endothe- Natl. Cancer Inst., 77:311–316.lium basement membrane and tissue. Curr. Opin. Cell Biol., Vlodavski, I., Lui, G.M., and Gospodarowicz, D. (1980) Morphological1:1009–1019. appearance growth behavior and migratory activity of human cells

Nicolson, G.L., Menter, D.G., Herrmann, J.L., Cavanaugh, P., Jia, L., maintained on extracellular matrix vs. plastic. Cell, 19:607–616.Hamada, J., Yun, Z., and Marchetti, D. (1994a) Tumor metastasis to Vlodavsky, I., Ariav, Y., Atzmon, R., and Fuks, Z. (1982) Tumor cellbrain: Role of endothelial cells, neurotrophin, and paracrine growth attachment to the vascular endothelium and subsequent degrada-factors. Crit. Rev. Oncogen., 5:451–471. tion of the subendothelial extracellular matrix. Exp. Cell Res.,

Nicolson, G.L., Nakajima, M., Herrmann, J.L., Menter, D.G., Cavan- 140:149–159.augh, P., Park J.S., and Marchetti, D. (1994b) Malignant melanoma Vlodavsky, I., Fuks, Z., Bar-Ner, M., Ariav, Y., and Schirrmacher, V.metastasis to brain: Role of degradative enzymes and responses to (1983) Lymphoma cell mediated degradation of sulfated proteogly-paracrine growth factors. J. Neurooncol., 18:139–149. cans in the subendothelial extracellular matrix: Relationship to tu-

Rifkin, D.B., Moscatelli, D., Gross, J., and Jaffe, E. (1984) Proteases, mor cell metastasis. Cancer Res., 43:2704–2711.Angiogenesis and Invasion. In: Cancer Invasion and Metastasis: Wasteson, A. (1971) A method for the determination of the molecularBiologic and Therapeutic Aspects. G.L. Nicolson and L. Milas, eds. weight and molecular weight distribution of chondroitin sulfate. J.

Chromatogr., 59:87–97.Raven Press, New York, pp. 187–200.


Documents

Specific degradation of subendothelial matrix proteoglycans by brain-metastatic melanoma and brain endothelial cell heparanases