Assembly of a chondrocyte-like pericellular matrix on non-chondrogenic

cells

Role of the cell surface hyaluronan receptors in the assembly of a pericellular matrix

WARREN KNUDSON* and CHERYL B. KNUDSON

Departments of Biochemistry and Pathology, Rush-Presbyterian-St Luke's Medical Center, Chicago, Illinois 60612, USA

* Author for correspondence

Summary

In this study, we have examined the capacity ofvarious cell types, which express cell surface hyalur-onan receptors, to organize a chondrocyte-like peri-cellular matrix when given chondrocyte-derivedextracellular matrix macromolecules exogenously.The assembly of a pericellular matrix was visualizedby a particle exclusion assay. Without the addition ofexogenous macromolecular components, none of thecell types studied exhibited significant pericellularmatrices extending from their plasma membranes.However, upon the addition of high molecular weighthyaluronan in combination with aggregating carti-lage proteoglycan monomers, large pericellularmatrices were formed within two hours of incu-bation. No pericellular matrices were formed if thesemacromolecular components were added separatelyat equivalent concentrations or if the componentswere added in the presence of hyaluronan hexa-saccharide, a competitive inhibitor of hyaluronaninteraction with cell surface hyaluronan receptors.Fully assembled pericellular matrices could also bedisplaced by the subsequent addition of hyaluronan

hexasaccharides. Nonliving, glutaraldehyde-fixedcells, which retained functional hyaluronan recep-tors, maintained the capacity to assembly pericellu-lar matrices with exogenous components, in serum-containing or serum-free medium. Cells that wereincubated with exogenous matrix macromoleculesfor 24 h, followed by a chase incubation in mediumminus the exogenous macromolecules, continued tomaintain the matrix for up to 6 h on live cells andmore than 24h on glutaraldehyde-fixed cells. Celltypes that did not express hyaluronan receptorswere not capable of organizing such pericellularmatrices when incubated with these exogenouscomponents. These findings suggest that cellsexpressing hyaluronan receptors have a significantcapacity to organize their immediate extracellularenvironment via hyaluronan-hyaluronan receptorinteractions. Possible physiological functions for thistype of matrix organizing capacity are discussed.

Key words: hyaluronan, hyaluronan receptors, chondrocyte,pericellular matrix, particle exclusion assay.

Introduction

Several cell types in culture, including embryonic andadult chondrocytes, have been shown to exhibit largepericellular matrices or 'coats' extending from theirplasma membrane by as much as one cell diameter(Clarris and Fraser, 1968; Knudson and Toole, 1985a;Knudson and Kuettner, 1990; Underhill and Toole, 1982).These pericellular matrix halos are readily visualized by aparticle-exclusion assay whereby small particles (in ourstudies these particles are formalin-fixed red blood cells)are added to cells in low-density culture. Upon settling,the particles are excluded from a distinct area or zonesurrounding the cells, clearly defining the boundary of thepericellular matrix. This assay represents the most usefulmethod of visualizing the full extent of a cell's pericellularmatrix, since conventional staining of the matrix byhistochemical or immunohistochemical techniques oftenleads to significant collapse of this structure (Clarris andFraser, 1968; Knudson and Toole, 19856).

On chondrocytes this pericellular matrix is composedpredominantly of chondroitin sulfate-rich proteoglycanJournal of Cell Science 99, 227-236 (1991)Printed in Great Britain © The Company of Biologists Limited 1991

(termed 'aggrecan'), although its composite structureappears dependent on the presence of hyaluronan (Knud-son and Toole, 1985a). Treatment of these cells withStreptomyces hyaluronidase, an enzyme that specificallydegrades hyaluronan, completely removes the pericellularmatrix within 30 min. After addition of fixed red bloodcells to the hyaluronidase-treated cultures, no excludedarea around the cells is observed and the particles nowabut the plasma membrane directly (Knudson and Toole,1985a; Knudson and Kuettner, 1990). Other extracellularmatrix macromolecules are undoubtedly resident withinthis pericellular matrix but have not as yet been fullycharacterized.

These same chondrocytes that display an extensivepericellular matrix also express specific, saturable, high-affinity hyaluronan binding proteins or receptors on theircell surface (Knudson and Toole, 1987; McCarthy andToole, 1989). The chondrocyte receptors have similarproperties to hyaluronan receptors as reported on manyother cells types (Nemec et al. 1987; Underhill and Toole,1979; Orkin et al. 1985; Green et al. 1988). One importantproperty shared by all of these cell surface hyaluronan

227

receptors is the minimum size of a hyaluronan oligosac-charide required to compete effectively for the binding ofnative hyaluronan to its receptor, which is a hyaluronanhexasaccharide. This property is helpful in differentiatingthe specific binding of hyaluronan to the cell surfacereceptor from the specific aggregation of hyaluronan withother extracellular matrix macromolecules such as carti-lage proteoglycan and link protein, which require aminimum hyaluronan oligosaccharide sequence of 10-12monosaccharides for competition (Solursh et al. 1980;Kimura et al. 1979; Hascall and Heinegard, 1974). Wehave shown previously that the assembly of a pericellularmatrix on chondrocytes, utilizing endogenous macromol-ecules, can be blocked by the addition of hyaluronanhexasaccharides (Knudson and Kuettner, 1990; Knudsonet al. 1991). In the presence of the hexasaccharides, thedisplaced matrix macromolecules are recovered intact inthe culture medium of the chondrocytes (Knudson andKuettner, 1990; Knudson et al. 1991). A fully assembledchondrocyte pericellular matrix can also be displaced fromthe cell surface in 2h by the subsequent addition ofhyaluronan hexasaccharides (Knudson and Kuettner,1990). Therefore, the chondrocyte pericellular matrix notonly contains hyaluronan as an integral part of itsmeshwork but is, in fact, anchored to the cell surface viainteraction between hyaluronan and hyaluronan recep-tors. These hyaluronan receptors thus have the capacity toassemble and organize the extracellular componentswithin the pericellular matrix.

Many other cell types express similar hyaluronanreceptors but do not exhibit significant pericellularmatrices (Nemec et al. 1987; Underhill and Toole, 1979;Underhill and Toole, 1982). For instance, the humanurinary bladder carcinoma cell line, HCV-29T, expressessaturable, high-affinity high-specificity cell surface hya-luronan receptors (Nemec et al. 1987), but does notelaborate an extensive pericellular matrix. The function ofthe hyaluronan receptors present on the HCV-29T cells isnot known but, unlike chondrocytes (Knudson and Toole,1987), the receptors are not occupied by endogenoushyaluronan (Nemec et al. 1987), which they synthesize atonly very low levels (Knudson and Knudson, 1990; Pauliand Knudson, 1988). Nevertheless, the hyaluronan recep-tors on these cells may have a similar capacity tochondrocytes for organizing a chondrocyte-like pericellu-lar matrix given the appropriate extracellular matrixmacromolecules exogenously. We therefore conductedexperiments to determine if the addition of exogenoushyaluronan alone, aggregating proteoglycan monomersalone, or hyaluronan in combination with aggregatingproteoglycan, to non-chondrogenic cells, which expresshyaluronan receptors but are matrix-deficient, wouldresult in the assembly of a pericellular matrix. The datadescribed in this study demonstrate that these cells havethis capacity to form such pericellular matrices and thatthis matrix is indeed anchored to the cell surface viainteraction between hyaluronan and hyaluronan recep-tors.

Materials and methods

Cell lines and culture conditionsThe human urinary bladder carcinoma cell lines, HCV-29T(Nemec et al. 1987) and HU-456 (Pauli and Knudson, 1988);simian virus-transformed mouse 3T3 fibroblasts, SV-3T3 (Under-hill et al. 1982); rat fibrosarcoma cells (Underhill et al. 1982); andB-16 mouse melanoma cells (Biswas, 1988), have been described.

The non-invasive human urinary bladder papilloma, RT-4, wasobtained from the American Type Culture Collection (Rockville,MD). Low-passage bovine aortic endothelial cells were a gift fromDr B. Pauli (Department of Pathology, Cornell University Collegeof Veterinary Medicine, Ithaca, NY), and were similar to cellspreviously described by his laboratory (Pauli and Lee, 1988).These cells were tested by us and found to be Factor VTTT positiveby immunofluorescence (data not shown). All cells were routinelygrown in 75 cm2 or 175 cm2 tissue culture flasks (Falcon) inDulbecco's modified Eagle's medium (Sigma, St Louis, MO)containing 4.5 gl"1 glucose supplemented with 10% fetal bovineserum (Hyclone, Logan, UT), 1 % penicillin/streptomycin solution(Sigma) and 2mM glutamine (Gibco BRL, Grand Island, NY), at37°C in a humidified incubator supplied with 5% CO2/95% air.Cells from near-confluent monolayers were dissociatedwith a 0.25% trypsin/0.05 M EDTA solution (Sigma) and1x10* cells ml"1 plated into 35-mm tissue culture dishes (Falcon)in complete medium for assay.

In some experiments, HCV-29T cells were plated in 35-mmdishes as described above, allowed to attach for 18 h at 37 °C,rinsed 3 times in PBS and then exposed to a solution of 1%glutaraldehyde (Electron Microscopy Sciences, Fort Washington,PA) in phosphate-buffered saline, pH 7.4 (PBS), for 5 min at roomtemperature. The fixed cells (all trypan blue inclusive) were thenrinsed twice with PBS containing 1 % bovine serum albumin(BSA, Sigma), over a 60 min time interval. These fixed HCV-29Tcells were used in similar matrix assembly experiments to thosein which the live, non-fixed cells were used.

Particle exclusion assayThe particle exclusion assay method followed a protocol describedpreviously (Knudson and Toole, 1986a). Briefly, horse red bloodcells (MG Scientific, Buffalo Grove, H) were fixed in 1.5%formaldehyde in calcium- and magnesium-free PBS overnight atroom temperature, washed extensively and stored in CMF-PBSwith antibiotics at 4°C. For assay, the culture medium of a 35-mmdish of cells to be tested was removed and replaced with a 750 /Asuspension of fixed red blood cells (108ml~1) in PBS containing0.1 % BSA. The particles were allowed to settle for approximately10 min. The cells were then observed and photographed on anOlympus IMT-2 inverted phase-contrast microscope or a NikonOptiflot inverted phase-contrast microscope equipped with Hoff-man optics.

Morphometric analysis of matrix to cell area ratioPhotographs of randomly selected cells (20-30 for each condition),printed at the same magnification, were analyzed using a SummaSketch Plus digitizing graphics tablet (Summagraphics, Fairfield,CT) interphased with an IBM-compatible microcomputer andutilizing a Sigma Scan software program (Jandel Scientific, CorteMadera, CA). A coat:cell ratio, defined as the ratio of the areadelimited by the perimeter of the pericellular matrix to the areadelimited by the plasma membrane, was determined by tracingthe matrix and cell perimeters on photomicrographs placed on thedigitizing tablet (Knudson and Toole, 1985a). If no detectablepericellular matrix was present the coatxell ratio was 1.0.

Preparation and addition of exogenous macromoleculesLarge aggregating chondroitin sulfate proteoglycan monomerderived from Swarm rat chondrosarcoma and proteoglycanmonomers derived from adult bovine articular or nasal cartilageswere gifts from Drs J. Kimura and E. Thonar (Department ofBiochemistry, Rush-Presbyterian-St Luke's Medical Center,Chicago, IL), respectively. The proteoglycans (A1D1 purifiedfractions) were incubated with 0.67unitml~1 Streptomyceshyaluronidase (Type EX, Sigma) for 4h at 37 °C in order to removesmall concentrations of hyaluronan found within these prep-arations. The proteoglycans were recovered following thistreatment, by overnight precipitation at 4°C with 3 volumes of95% ethanol containing 1.3% potassium acetate or carriedthrough another dissociative cesium chloride equilibrium centri-fugation. The precipitated proteoglycan pellet was air-dried andredissolved in tissue culture medium. For the centrifuged

228 W. Knudson and C. B. Knudson

proteoglycan, the bottom l/5th fraction (A1D1-D1) was collectedand dialyzed exhaustively against 50 mM Na2SO4, followed bydistilled water, followed by culture medium. The hyaluronan usedwas high molecular weight grade-I from human umbilical cord(Sigma). Hyaluronan hexasaccharides were prepared by anexhaustive digestion of high molecular weight hyaluronan withbovine testicular hyaluronidase. Briefly, a 0.5 mgml"1 solution ofhyaluronan in 0 . 1 M sodium acetate, 0.05 M sodium chloride(Knudson et al. 1984) was incubated with 150 USP/TF units mP 1

testicular hyaluronidase (Type I-S, Sigma) for 24 h at 37 °C. Thedigestion mixture was heat-inactivated by boiling for lOmin,adjusted to pH7.2 and filter sterilized. Alternatively, purifiedhyaluronan hexasaccharides were prepared following partialenzymatic digestion (Kimura et al. 1979; Hascall and Heinegard,1974). High molecular weight hyaluronan was first digested withpapain, dialyzed against 0.5 M sodium acetate, precipitated twicewith 95% ethanol, redissolved in 0 . 1 M sodium acetate, 0 .15 MNaCl, 0.001 M EDTA, pH5.0, and digested for 16 h with testicularhyaluronidase (172 USP/TFU hyaluronidase mg"1 hyaluronan).Following a lOmin heat inactivation at 100°C, the hyaluronanoligosaccharides were then separated on a 2.5cmx 118cm columnof Bio-gel P30, eluted in 0.5 M pyridinium acetate, pH 6.5 (Solurshet al. 1980). The hexasaccharide peak was pooled (oligosacchar-ides from 2—14 monosaccharides were well resolved) andlyophilized to dryness, redissolved in distilled water andlyophilized again. We determined that the exhaustive digestionmixture, before column chromatography, containing inactivatedenzyme plus hyaluronan tetra- and hexasaccharides, had aproportional capacity to inhibit pericellular matrix assembly asthe column-purified hyaluronan hexasaccharide. A control sol-ution of buffer containing heat-inactivated enzyme alone had noeffect on the cells. Therefore, the tetra-/hexasaccharide mixturewas used in most of these studies.

The concentrations of exogenous macromolecules added to cellsused in these studies were those previously determined as optimalfor pericellular matrix assembly on chondrocytes (Knudson et al.1991). Medium from cells grown in low-density culture for 24 h in35 mm dishes was removed and replaced with 2.0 ml fresh serum-supplemented medium containing either: (1) no additionalcomponents; (2) 2.0 ing ml"1 proteoglycan monomer; (3)12/igml"1 hyaluronan; (4) 2.0mgml~1 proteoglycan monomerplus 12/(gml~1 hyaluronan; or (5) 2.0mgml~ proteoglycanmonomer, 12//gml~1 hyaluronan, and lOO^gml"1 hyaluronantetra-/hexasaccharides. Cells were allowed to incubate at 37°C inthe humidified/CO2 incubator for 2-24 h. Following this incu-bation, the medium was removed and the cultures were assayedby the particle exclusion method as described above.

Enzymatic removal of rat fibrosarcoma pericellularmatrixMedium from rat fibrosarcoma cells plated at low density in 35-mm dishes was removed and replaced with 2.0 ml fresh mediumcontaining 0.5 units of Streptomyces hyaluronidase (Type DC,Sigma). Following incubation with the hyaluronidase for l h at37 °C, the cell cultures were washed twice with Hanks' balancedsalt solution followed by the addition of medium alone, or mediumcontaining hyaluronan, proteoglycan, hyaluronan plus proteogly-can or hyaluronan plus proteoglycan and hyaluronan tetra-/hexasaccharides, as described above. Following 2 h of incubationwith these solutions, the medium was removed and the cellsassayed by the particle-exclusion method.

Hyaluronan binding assaysSaturation binding assays were performed on live, intact cells asdescribed previously by Nemec et al. (1987). Briefly, cells werereleased from monolayer cultures by treatment with 10 mM EDTAin CMF-PBS, centrifuged and resuspended in PBS. Cells werealiquoted into duplicate bovine serum albumin-coated micro-centrifuge tubes containing varying concentrations of [3H]hyalur-onan or [3H]hyaluronan plus 250 ^g non-labeled hyaluronan(added to determine non-specific background binding), in a totalvolume of 0.5 ml with lxlO6 cells/tube. The [3H]hyaluronanligand was purified from the conditioned medium of ratfibrosarcoma cells labeled with [3H]acetate as described pre-

viously (Knudson and Toole, 1987; Nemec et al. 1987). The 3H-labeled hyaluronan was of high molecular weight (voided onSepharose CL-2B), free of contaminating glycosaminoglycans (byDEAE chromatography) and had a specific activity ofeiSOOctsmin'Vg"1. Cells plus [3H]hyaluronan were incubatedfor 60min at room temperature with gentle shaking. Followingthis incubation, cells were pelleted by centrifugation at 500 g,washed twice with 5xPBS, dissolved in 0.5% sodium dodecylsulfate, mixed with scintillation cocktail (Ecoscint; ICN, CostaMesa, CA) and counted. Specific binding was determined bysubtraction of background binding.

Results

Assembly of a chondrocyte-like pericellular matrixaround human HCV-29T bladder carcinoma cellsThe HCV-29T cell line was derived from an invasivehuman urinary bladder transitional cell carcinoma. Wehave previously reported that these HCV-29T cells expresshigh numbers of unoccupied, specific binding sites or'receptors' for hyaluronan (Nemec et al. 1987), yetsynthesize and secrete only low levels of endogenoushyaluronan (Knudson and Knudson, 1990). When placedin low-density culture for 12-24 h the cells attach, spreadon the substratum, and assume a somewhat hetero-geneous morphology, from round to polygonal. As shownin Fig. 1A and B, using the particle-exclusion assay, nopericellular matrices were detectable around the cells.This feature appeared independent of the length of time inculture (data not shown). Incubation of these cells for 24 hin the presence of exogenously added hyaluronan alone(Fig. 1C and D), or large hyaluronan-aggregating chon-droitin sulfate proteoglycan alone (Fig. IE and F) did notresult in pericellular matrix formation. However, if theHCV-29T cells were incubated for 24 h in the presence ofboth hyaluronan and proteoglycan monomers, an exten-sive pericellular matrix was assembled around the cells(Fig. 1G and H). A morphometric analysis of these data isgiven in Table 1. Matrix formation in the presence of thesecomponents occurred within 2h, with maximal matrixassembly established after 18 h (data not shown). Ad-ditionally, pericellular matrix assembly was observed onHCV-29T cells in the presence of hyaluronan andproteoglycan monomers derived from bovine articular(coatxell ratio=2.75±0.40) or, bovine nasal (2.47±0.22)cartilages, similar in size to the matrices formed withhyaluronan and rat chondrosarcoma proteoglycan(Table 1). These data suggest that the formation of anextensive pericellular matrix, with the capacity to excludefixed red blood cells, can be obtained with a minimum oftwo extracellular matrix macromolecules: namely, hyalur-onan and an aggregating proteoglycan.

To determine the mode of interaction of the pericellularmatrix with the HCV-29T cell surface, hyaluronan tetra-/hexaaaccharides were introduced into the hyaluronanplus proteoglycan mixture. As shown in Fig. 2A, thepresence of the hyaluronan hexasaccharides prevented theassembly of pericellular matrices (see also Table 1). Thesedata suggest that the mechanism of pericellular matrixassembly by these cells is similar to that of cells derivedfrom embryonic tibial chondrocytes (Knudson andKuettner, 1990), i.e. the matrix is anchored to the cellsurface via specific hyaluronan receptor sites. Addition-ally, a fully assembled pericellular matrix, formed onHCV-29T cells following 24 h incubation in the presence ofexogenous hyaluronan and proteoglycan (Fig. 1G), couldbe displaced within 3h by the subsequent addition of

Pericellular matrix on non-chondrogenic cells 229

$W$mz.

Fig. 1. Assembly of a pericellular matrix with exogenous matrix macromolecules on HCV-29T cells. (A, C, E and G) Phase-contrastphotomicrographs; (B, D, F and H) corresponding cells visualized via Hoffman optics. Medium from HCV-29T cells grown in low-density culture for 24 h was removed and replaced with either: fresh medium as a control (A and B); medium containinghyaluronan (C and D); medium containing proteoglycan (E and F); or medium containing hyaluronan plus proteoglycan (G and H).Following these additions, cells were incubated for 24 h, test or control media were removed and the cells assayed by the particleexclusion method. Only cells incubated in the presence of both hyaluronan and proteoglycan exhibited significant pericellularmatrices (G and H). Bar, 50 ^m.

230 W. Knudson and C. B. Knudson

Table 1. Morphometric analysis of matrix to cell area ratios

Cell-type Control HA PG HA/PG HA/PG/HA^

HCV-29TSV-3T3HU-456RT-4B-16Endothelial cells

1.03±0.031.04±0.031.02±0.021.02±0.03104±0.041.05±0.05

1.03±0.251.02±0.361.01±0.061.01±0.011.04±0.071.03±0.05

1.28±0.111.19±0.071.19±0.081.01±0.021.02±0.021.07±0.06

3.47±0.973.96±0.392.39±0.361.01±0.021.00±0.042.69±0.45

1.04±0.021.02±0.021.01±0.031.01±0.021.04±0.081.05±0.05

Values represent the mean ratio of the area delineated by the pericellular matrix to the area delineated by the plasma membrane. Error valuesrepresent the 95% confidence range from the mean, n=20-30 cells. Control, medium alone; HA, medium containing hyaluronan; PG, mediumcontaining proteoglycan; HA/PG, medium containing added hyaluronan and proteoglycan; HA/PG/HA4_s, medium containing hyaluronan,proteoglycan and hyaluronan tetra-/hexasaccharides.

hyaluronan hexasaccharides (Fig. 2C). Medium alone for3h had no effect on the size of the assembled matrix(Fig. 2B). Therefore the pericellular matrix is initiallyassembled and subsequently maintained via anchorage tothese hyaluronan binding sites.

To exclude the possibility that addition of exogenousproteoglycan and hyaluronan induced the synthesis of an

• ' " > .

endogenous component(s) required for matrix assembly,hyaluronan and proteoglycan were added to glutaralde-hyde-fixed HCV-29T cells. We have shown previously thatthe cell surface hyaluronan binding sites are stable to mildfixation in glutaraldehyde (Nemec et al. 1987). Following24 h of incubation of the fixed HCV-29T cells with optimalconcentrations of hyaluronan and proteoglycan, a peri-cellular matrix again assembled, with a coatxell ratio of3.57±0.77. This matrix was equivalent in size to matricesassembled on live HCV-29T cells (Table 1). Pericellularmatrices on fixed cells did not form without the addition ofthe exogenous macromolecules (1.02±0.03), or whenhyaluronan and proteoglycan were added in the presenceof excess hyaluronan tetra-/hexasaccharides (1.00±0.02).Like the live cells, fully assembled matrices on fixed cellscould also be displaced by subsequent incubation withhyaluronan tetra-/hexasaccharides (1.02±0.02). Addition-ally, the fixed HCV-29T cells were able to assemble apericellular matrix (2.54±0.31) when incubated in serum-free medium containing exogenous hyaluronan and pro-teoglycan. Therefore, serum components are also notrequired for matrix assembly.

In order to determine the stability or persistence of thesepericellular matrices formed from exogenous components,HCV-29T carcinoma cells were incubated with optimalconcentrations of hyaluronan plus proteoglycan for 24 h,rinsed extensively and then incubated in fresh culturemedium minus the exogenous macromolecules. Aftervarious chase times, cell cultures were assayed by theparticle exclusion assay for persistence of pericellularmatrix in the absence of the exogenous components. Asshown in Table 2, the pericellular matrix on live cells,which had a mean coatxell ratio of approximately 2.7 atthe start of the chase experiment, showed no reduction incoatxell ratio at 3 h (see also Fig. 2B), a reduced but still

Fig. 2. The effect of hyaluronan hexasaccharides on theassembly and maintenance of HCV-29T cell pericellularmatrices. (A) Medium from HCV-29T cells grown in low-density culture for 24 h was removed and replaced withmedium containing hyaluronan, proteoglycan and excesshyaluronan tetra-/hexasaccharides, incubated for 24 h and thenassayed by the particle exclusion method. No pericellularmatrix assembled under these conditions. (B and C) HCV-29Tcell medium was replaced with medium containing hyaluronanplus proteoglycan and incubated for 24 h in order to establishpericellular matrices around the cells. This medium was thenremoved and replaced with either medium alone as a control(B), or medium containing an excess of hyaluronan tetra-/hexasaccharides (C). The cells were incubated for an additional3 h, and then assayed by the particle exclusion method. Thehyaluronan tetra-/hexasaccharides displaced the pericellularmatrix (C). Cells in medium alone, representing non-specificshedding into the culture medium, retained the matrix (B).Bar, 50/on.

Pericellular matrix on non-chondrogenic cells 231

Table 2. Persistence of pericellular matrices on HCV-29T cells in the absence of exogenous matrix macromolecules asdetermined by morphometric analysis of matrix to cell area ratios

Chase times (h)

12 24

Live cellsFixed cells

2.68±0.443.01±0.58

2.83±0.572 96±0.40

1.76±0.442.69±0.53

1.35±0.09n.d.

1.06±0.041.89±0.26

Live or glutaraldehyde-fixed HCV-29T cells were incubated with optimal concentrations of hyaluronan plus proteoglycan for 24 h, rinsed and thenincubated with fresh culture medium minus the exogenous macromolecules for 0, 3, 6, 12 and 24 h. Cell cultures at each time point were assayed bythe particle-excluBion method. Values represent the mean ratio of the area delineated by the pericellular matrix to the area delineated by the plasmamembrane. Error values represent the 95 % confidence range from the mean; n=26 cells/time point; n.d., not determined.

significant pericellular matrix at 6 h and total absence ofmatrix at later time points. The pericellular matrix onfixed, nonliving HCV-29T cells showed a more gradual lossof coat size (Table 2). Therefore, the HCV-29T pericellularmatrix assembled from exogenous matrix macromoleculesdoes not depend on continuous exposure to high concen-trations of the soluble macromolecules in order to bemaintained. Nevertheless, these data also support thesuggestion that these cells cannot, on their own, syn-thesize the macromolecules required for maintenance ofthe matrix and any turnover or shedding of exogenouslyadded components cannot be replaced.

Correlation of matrix assembly with expression ofhyaluronan receptors on other cell typesIn order to test further the hypothesis that cell surfacehyaluronan receptors are essential for assembly andorganization of pericellular matrices, hyaluronan andproteoglycan were added to several cell types, which wereeither positive or negative for expression of hyaluronanreceptors. Cell surface hyaluronan-specific receptors onSV-3T3 cells have been extensively reported (Underhilland Toole, 1979; Toole, 1981). In addition, Toole et al.(personal communication), at Tufts University in Boston,using a newly developed monoclonal antibody directedagainst the hyaluronan receptor present in embryonicchick brain, have described the immunohistologicallocalization of hyaluronan receptors on bovine capillaryendothelial cells. To characterize the other cell types,direct hyaluronan binding assays were performed. Asshown in Fig. 3, only the HU-456 cells (and the HCV-29Tcells used as a positive control) exhibited specific,saturable binding of 3H-labeled hyaluronan. No bindingwas detected on the RT-4 or B-16 cells.

These cells were next studied for their capacity toassemble pericellular matrices via exogenous macromol-ecules. As shown in Table 1, like the HCV-29T cells,without addition of exogenous macromolecules none of thecell cultures tested exhibited endogenous pericellularmatrices. However, upon addition of hyaluronan andproteoglycan, the SV-3T3 and HU-456 tumor cells, as wellas the normal, low-passage bovine aortic endothelial cells,assembled pericellular matrices with coatxell ratiossimilar to the HCV-29T cells. No matrices were observedon the RT-4 or B-16 cells. Also, no matrix assembly wasobserved on any of the cell cultures in the presence ofhyaluronan alone, proteoglycan alone or hyaluronan,proteoglycan and excess hyaluronan tetra-/hexasaccharides. Thus only cells that exhibited expressionof hyaluronan receptors had the capacity to assemblepericellular matrices via exogenous macromolecules.

Other, non-chondrogenic cells such as fibroblasts natur-ally exhibit endogenous hyaluronan-dependent pericellu-

4 6[3H]hyaluronan (//g)

Fig. 3. Specific binding of [3H)hyaluronan to intact cells.Varying concentrations of [3H]hyaluronan were incubated withlxlO6 EDTA-released HCV-29T (O O); HU-456 (• • ) ;RT-4 (D •) ; or B-16 (• •) tumor cells as described inMaterials and methods. Specific binding was determined bysubtraction of non-specific background binding. Only the HCV-29T and HU-456 carcinoma cells displayed significantsaturable binding of hyaluronan to their cell surface. Some ofthe differences in maximal binding between HCV-29T and HU-466 cells may be due to differences in cell size. Error barsrepresent the range of four determinations.

lar matrices in culture (Clarris and Fraser, 1968; Knudsonand Toole, 1985a). In order to determine whether thesecells can also utilize exogenous macromolecules for matrixassembly, hyaluronan and proteoglycan were added to ratfibrosarcoma cells stripped of their endogenous matrix.The large pericellular matrices of rat fibrosarcoma cells(Fig. 4A) were removed via a brief treatment withStreptomyces hyaluronidase. Following the enzymaticremoval of this pericellular matrix, regrowth of anendogenous pericellular matrix required approximately8 h of incubation (data not shown). However, upon additionof exogenous hyaluronan plus proteoglycan, a pericellularmatrix assembled within 2h (Fig. 4C), at which time noendogenous matrix was present in control cultures(Fig. 4B). Additionally, the inclusion of hyaluronan tetra-/hexasaccharides along with the hyaluronan plus proteo-glycan mixture, prevented pericellular matrix assemblyvia the exogenous macromolecules (Fig. 4D). These fibro-blast-like cells can therefore assemble similar pericellularmatrices using either endogenous or exogenous macromol-ecules. The data also suggest that these matrices areanchored via hyaluronan receptors, the expression ofwhich on these cells has been controversial (Underhill,1988).

Discussion

Extensive pericellular matrices, of varying thicknesses,

232 W. Knudson and C. B. Knudson

*:.•£.. -*.:/:•?jamm.

Fig. 4. Assembly of a pericellular matrix with exogenous macromolecules on Streptomyces hyaluronidase-stripped rat fibrosarcomacells. Rat fibrosarcoma cells, grown in low-density culture for 24 h, exhibit an extensive endogenous pericellular matrix (A). Thismatrix can be stripped from the cell surface by a l h treatment at 37°C with medium containing Streptomyces hyaluronidase. Thehyaluromdase-treated cells were then washed and incubated with either fresh medium alone (B); medium containing containinghyaluronan and proteoglycan (C); or medium containing hyaluronan, proteoglycan and excess hyaluronan tetra-/hexasaccharides(D). Following these additions, cells were incubated for an additional 2 h and then assayed by the particle exclusion method. In thistime frame, no endogenous pericellular matrix was assembled (B). However, cells incubated with exogenous hyaluronan andproteoglycan for 2h assembled a pericellular matrix (C). Pericellular matrix assembly in the presence of hyaluronanoligosaccharides was blocked (D). Bar, 50 /mi.

have been visualized surrounding a wide variety ofeukaryotic cells in culture (reviewed by Toole, 1981; alsoClarris and Fraser, 1968; Knudson and Toole, 1985;Knudson and Kuettner, 1990; Underbill and Toole, 1982).The principal feature known concerning the structuralcomposition of these pericellular matrices is that they aredependent on hyaluronan. Clarris and Fraser (1968)demonstrated that the pericellular matrix surroundingsynovial fibroblasts was not sensitive to treatment byribonuclease, deoxyribonuclease, neuraminidase, EDTAor dilute trypsin, but was highly sensitive to variouspurified hyaluronidases (though the hyaluronidases usedwere not specific for hyaluronan). Subsequent studiesusing Streptomyces hyaluronidase demonstrated thathyaluronan was indeed an essential structural componentof these matrices (Toole, 1981; Knudson and Toole,1985a). Goldberg and Toole (1982) further showed thatinhibition of hyaluronan synthesis by treatment offibrosarcoma cells with monensin also inhibited formationof detectable pericellular matrices. Underhill and Toole(1982) suggested that the size of the coat may even beproportional to the content of cell surface hyaluronan. Insubsequent studies, however it was observed that cells ofembryonic limb mesenchyme could turn on, or off, theexpression of pericellular matrices without change in thelevels of their cell surface hyaluronan, suggesting thatother components are also required for matrix formation(Knudson and Toole, 1985a). Indeed, it was suggested

earlier that these matrices were too thick to be composedentirely of hyaluronan (Toole, 1981) and most likelyrepresented secondary interactions between hyaluronanand other molecules such as proteoglycans, fibronectin orproteoglycan-collagen complexes, none of which havebeen fully characterized in terms of their capacity to beorganized into a coat structure. Therefore, the firstconclusion from our present results is that the formation ofa pericellular matrix, which can exclude particles or cells,can be obtained with as few as two components, theaddition of exogenous hyaluronan and an aggegating-typeproteoglycan. Preliminary work (C. Knudson, unpub-lished data) has further shown that the proteoglycancomponent must also be structurally intact. No pericellu-lar matrix was formed when exogenous hyaluronan wasadded in combination with: (1) chondroitinase ABC-digested proteoglycan, (2) reduced and alkylated proteo-glycan, (3) the proteoglycan hyaluronan-binding domain,or (4) the chondroitin sulfate-rich domain of the proteogly-can.

A second conclusion of this study is that the ability toassemble a pericellular matrix requires the expression ofcell surface hyaluronan binding sites or 'receptors'. Thepericellular matrix that is organized around the cells usedin this study (i.e. HCV-29T, SV-3T3, HU-456 and bovineaortic endothelial cells) appears, to be anchored to the cellsurface via interactions between hyaluronan and hyaluro-nan receptors. This conclusion was drawn because the

Pericellular matrix on non-chondrogenic cells 233

pericellular matrices could be prevented from assembling,or displaced from the cells once assembled, via the additionof an excess of hyaluronan hexasaccharides, which arecompetitive inhibitors of hyaluronan receptors. Similarobservations have been made concerning the assembly ofpericellular matrices on embryonic chick chondrocytesand adult chondrocytes from a rat chondrosarcoma cellline (Knudson and Kuettner, 1990; Knudson et al. 1991).Cells such as the B-16 mouse melanoma or RT-4 non-invasive bladder papilloma, which did not express hyalur-onan binding sites, did not assemble a pericellular matrixwhen presented with exogenously added hyaluronan plusproteoglycan. The capacity to assemble a pericellularmatrix, given the appropriate exogenous components,appears indicative of hyaluronan receptor expression andmay therefore serve as a simple, indirect assay for thedetection of hyaluronan receptors.

The in vivo presence or function of extensive pericellularmatrices that surround many cell types in culture is notwell established. Clearly, the expression of pericellularmatrices around chondrocytes correlates well with theirprimary cellular function: to generate and organize anextracellular matrix capable of withstanding load forces.The degree of extracellular matrix surrounding chondro-cytes in vivo also seems to correlate well with theexpression of pericellular matrices in vitro, suggestingthat these structures do in fact exist in vivo (Knudson andToole, 1985a).

The physiological function of pericellular matricessurrounding other cell types is less well understood. Earlywork in this area suggested that such matrices may serveas a protective barrier around cells. For example, humanwhite blood cells were observed to be excluded from a zonesurrounding cultured synovial fibroblasts (Clarris andFraser, 1968), in a fashion similar to the exclusion of redblood cells used in the particle exclusion assays. McBrideand Bard (1979) demonstrated that T-lymphocytes weresimilarly excluded from a zone surrounding fibrosarcomacells in culture. This exclusion apparently inhibited theirlymphocyte-mediated cytolysis, which returned afterremoval of the pericellular matrix with hyaluronidases.Thus, extensive pericellular matrices may function torepel or exclude other cells, conferring a degree ofprotection from cytotoxic cell types. It has also beenproposed that extensive pericellular matrices reduceinteraction between homologous cells, in other words,modulate cell-cell aggregation (Underhill, 1982; Knudsonand Maleski, 1987).

In considering whether these structures exist in vivo,the source of the extracellular macromolecules requiredfor the formation of these pericellular matrices must alsobe considered. Certainly chondrocyte-derived aggregatingproteoglycans are not present at high concentrations inother distant connective tissues. It is interesting toconsider that cells such as the HCV-29T carcinoma cells(Knudson and Knudson, 1990; Pauli and Knudson, 1988)and the bovine aortic endothelial cells (Knudson et al.1988) have the capacity to stimulate the synthesis ofmatrix macromolecules by neighboring cell types such asfibroblasts. It is therefore possible that the function of thisstimulation of synthesis may be to provide the matrixcomponents required for the formation of a pericellularmatrix around the tumor or endothelial cells. Canfibroblasts produce the required matrix-forming macro-molecules, that is, hyaluronan plus an aggregatingproteoglycan? In preliminary studies, we observed thatthe addition of 25 % conditioned medium isolated from rat

fibrosarcoma cells to HCV-29T cells for 24 h resulted in theassembly of a prominent pericellular matrix (coatxellratio=2.23±0.82) after 24h. The exact composition of themacromolecules present in this conditioned medium isstill under investigation. It is known that these cellssynthesize copious amounts of hyaluronan (Underhill andToole, 1982). However, our model would imply that afibroblast-derived proteoglycan, capable of aggregatingwith hyaluronan, must also be present. Other non-cartilagenous proteoglycans such as versican (Johannssonet al. 1975; Zimmerman and Rouslahti, 1989), as well asglial hyaluronan binding protein (Perides et al. 1989) andhyaluronectin (Delpech et al. 1982), have also beenreported to interact strongly with hyaluronan (or proposedto interact via sequence homology to cartilage linkproteins). It remains to be seen whether any of thesemolecules, in association with hyaluronan, can promotethe assembly of pericellular matrices around matrix-deficient cells expressing hyaluronan receptors. Nonethe-less, it is interesting to note that during the infiltration ofbasal cell carcinomas, part of the intense stromal responseat the advance of the invading tumor cells includes theelevated deposition of hyaluronectin, synthesized by thestromal fibroblasts (Delpech et al. 1982). This hyaluronec-tin, along with hyaluronan, may serve as an exogenouscomponent for pericellular matrix formation around theinfiltrating neoplastic cells or migrating endothelial cellsinvolved in neovascularization.

This work was supported in part by Public Health Service grantAR-39239 from the Institute of Arthritis and Musculoskeletal andSkin Diseases, National Institutes of Health. We thank Larissa J.Coombs for her technical assistance.

References

BISWAS, C. (1988). Heparin and heparin sulfate binding sites on B-16melanoma cells. J. cell. Physiol. 136, 147-153.

CLARHIS, B. J. AND FRASER, J. R. E. (1968). On the pericellular zone ofsome mammalian cells in vitro. Expl Cell Res. 49, 181-193.

DELPECH, A., DELPECH, B., GIRARD, N., BOULLIE, M. C. AND LAURET, P.(1982). Hyaluronectin in normal human skin and in basal cellcarcinoma. Br. J. Derm. 106, 561-568.

GOLDBERG, R. L. AND TOOLE, B. P. (1982). Monensin inhibition ofhyaluronate synthesis in rat fibrosarcoma cells. J. biol. Chem. 258,7041-7046

GREEN, S. J., TARONE, G. AND UNDERHILL, C. B. (1988). Distribution ofhyaluronate and hyaluronate receptors in the adult lung. J. Cell Sci.89, 145-156.

HASCALL, V. C. AND HEINEGARD, D. (1974). Aggregation of cartilageproteoglycans. J biol. Chem. 249, 4242-4249.

JOHANSSON, S., HEDMAN, K., KJELLEN, L., CHRISTNER, J., VAHERI, A.AND HOOK, M. (1985). Structure and interactions of proteoglycans inthe extracellular matrix produced by cultured human fibroblaBts.Biochem. J. 232, 161-168.

KIMURA, J. H., HARDINGHAM, T E., HASCALL, V. C. AND SOLURSH, M.(1979). Biosynthesis of proteoglycans and their assembly intoaggregates in cultures of chondrocytes from the Swarm ratchondrosarcoma. J. biol. Chem. 254, 2600-2609.

KNUDSON, C. B., COOMBS, L. J. AND KUETTNER, K. E. (1991).Chondrocyte pericellular matrix assembly and displacement mediatedvia hyaluronate. Orthoped. Trans 15, (1991) (in press).

KNUDSON, C. B. AND KNUDSON, W. (1990). Similar epithehal-stromalinteractions in the regulation of hyaluronate production during limbmorphogenesis and tumor invasion. Cancer Letts. 52, 113-122.

KNUDSON, C. B., KNUDSON, W. AND KUETTNER, K. E. (1988). Cell-cellinteractions involved in the regulation of glycosaminoglycanproduction. Orthoped. Trans. 12, 435-436.

KNUDSON, C. B. AND KUETTNER, K. E. (1990). A role for hyaluronate inchondrocyte pericellular matrix assembly. Orthoped. Trans. 14, 370.

KNUDSON, C. B. AND MALESKI, M. P. (1987). Homotypic cell adhesion ofchick embryo limb mesoderm m vitro. J. Cell Biol. 105, 141a.

KNUDSON, C. B. AND TOOLE, B. P (1985a). Changes in the perieellularmatrix during differentiation of limb bud mesoderm. Devi Biol. 112,308-318.

234 W. Knudson and C. B. Knudson

KNUDSON, C. B. AND TOOLE, B. P. (19856) Fluorescent morphologicalprobe for hyaluronate. J. Cell Biol. 100, 1753-1768.

KNUDBON, C. B. AND TOOLB, B. P. (1987). Hyaluronate-cell interactionsduring differentiation of chick embryo limb mesoderm. Devi Bwl. 124,82-90.

KNUDSON, W., GUNDLACH, M. W., SCHMID, T. M. AND CONRAD, H. E(1984). Selective hydrolysis of chondroitin sulfates by hyaluromdase.Biochemistry 23, 368-375.

MCBRIDE, W. H. AND BARD, J. B. L. (1979). Hyaluronidase-Bensitivehalos around adherent cells. Their role in blocking lymphocyte-mediated cytolysis. J. exp. Med. 149, 507-515.

MCCARTHY, M. T. AND TOOLE, B. P. (1989). Membrane-associatedhyaluronate-binding activity of chondrosarcoma chondrocytes. J. cell.Physiol. 141, 191-202.

NEMEC, R. E., TOOLB, B. P. AND KNUDSON, W. (1987). The cell surfacehyaluronate binding sites of invasive human bladder carcinoma cells.Biochem. biophys. Res. Commun. 149, 249-257.

ORKIN, R. W., KNUDSON, W. AND TOOLE, B. P. (1985). Loss ofhyaluronate-dependent coat during myoblast fusion. Devi Biol. 107,527-530.

PAULI, B. U. AND KNUDSON, W. (1988). Tumor Invasion: A consequenceof destructive and compositional matrix alterations. Human Path 19,628-639.

PAULI, B. U. AND LEE, C.-L. (1988). Organ preference of metastasis. Lab.Invest. 58, 379-387.

PERIDES, G , LANE, W. S., ANDREWS, D , DAHL, D. AND BIONAMI, A

(1989) Isolation and partial characterization of a glial hyaluronate-binding protein. J. biol. Chem. 264, 5981-5987.

SOLURSH, M., HARDINGHAM, T. E., HASCALL, V. C. AND KIMURA, J. H.(1980). Separate effects of exogenous hyaluronic acid on proteoglycansynthesis and deposition in pericellular matrix by cultured chickembryo limb chondrocytes. Devi Biol. 75, 121-129.

TOOLE, B. P. (1981). Glycosaminoglycans in morphogenesis. In CellBiology of the Extracellular Matrix (ed. E. D. Hay), pp. 259-294. NewYork Plenum Press.

UNDERHILL, C B (1982). Interaction of hyaluronate with the surface ofsimian virus 40-transformed 3T3 cells: aggregation and bindingstudies. J. Cell Sci. 56, 177-189.

UNDERHILL, C. B. (1988). The interaction of hyaluronate with the cellsurface: the hyaluronate receptor and the core protein. In The Biologyof Hyaluronan, (ed. D. Evered and J. Whelan), pp. 87-106. Chichester,UK, John Wiley and Sons.

UNDERHILL, C. B. AND TOOLE, B. P. (1979). Binding of hyaluronate to thesurface of cultured cells. J. Cell Biol. 82, 475-484.

UNDERHILL, C. B. AND TOOLE, B. P. (1982) Transformation-dependentloss of the hyaluronate-containing coats of cultured cells. J. cell.Physiol. 110, 123-128.

ZIMMERMANN, D. R. AND RUOSLAHTI, E. (1989). Multiple domains of thelarge fibroblast proteoglycan, versican. EMBO J. 8, 2975-2981.

(Received 12 December 1990 - Accepted 27 February 1991)

Pericellular matrix on non-chondrogenic cells 235